Identification of the adhesion GPCR ADGRL4/ELTD1 as a novel potential prognostic biomarker for cerebral cavernous malformation disease Authors: Andrea Perrelli, Ashraqat Ahmed, Anna Onisiforou, Gwenola Boulday, Juan Zalvide, Celia M Pombo, José E Martínez, Alessio Paolini, Salim Abdelilah-Seyfried, Ioannis Petrakakis, Wolfram S Kunz, Robert Geffers, Christian Hartmann, Walter Bini, Gerasimos Baltsavias, Amir Samii, Helmut Bertalanffy, Saverio F Retta, Angela J Glading, Souvik Kar Published: 2026 ABSTRACT AbstractBackgroundCerebral cavernous malformations (CCM) are angiographically occult vascular anomalies of the brain, characterized by dilated capillaries, increased vascular permeability, and loss of endothelial junctional protein complexes. Loss-of-function mutations in one of the three genes, namely KRIT1/CCM1, CCM2, and PDCD10/CCM3, have been associated with the disease pathogenesis, although the contribution of other genetic determinants besides CCM genes has been recently identified. Despite recent advances in understanding the molecular mechanism of the disease, the current lack of therapies and its unpredictable clinical behavior represent a significant challenge in the identification of diagnostic biomarkers. ADGRL4/ETLD1 (epidermal growth factor, latrophilin and seven transmembrane domain-containing protein 1), a G-protein coupled receptor (GPCR) protein is a known biomarker of angiogenesis and inflammation, and it has been suggested to be a key therapeutic target for stroke and high-grade gliomas. However, the relevance of ELTD1 in CCM pathogenesis remains unexplored.MethodsThrough different analyses, including whole RNA transcriptome, immunohistochemistry, and real-time PCR, conducted in human CCM patients, cellular and animal models, herein we show the association between CCM disease and ELTD1.ResultsThe whole RNA transcriptome approaches demonstrated ELTD1 is differentially expressed, and gene expression analysis revealed it is significantly upregulated in surgically resected tissue and plasma samples from CCMs, and clinical correlation analysis showed that increased ELTD1 associates with the presence of Focal Neurological Deficits in patients. Immuno-expression showed a strong ELTD1 immunoreactivity in endothelial cells lining affected lesions. According to these elevated levels of ELTD1 in human patients, we also showed a robust increase in the expression level of ELTD1 in cellular and animal models of CCM disease.ConclusionsTaken together, our results demonstrate for the first time ELTD1 involvement in CCM pathogenesis, and its tight link with CCM genes. Thanks to this putative new prognostic biomarker, future clinical studies in larger patient cohorts will aim at improving the CCM disease management and risk stratification in patients, as well as placing the basis for targeted therapeutic strategies in CCM.Supplementary InformationThe online version contains supplementary material available at 10.1186/s10020-026-01495-4.Keywords: ADGRL4/ELTD1, Angiogenesis, Biomarkers, Cerebral Cavernous Malformation (CCM), Diagnostic biomarkers, Prognostic biomarkers, RNA transcriptome Introduction Cerebral cavernous malformations (CCM) are vascular lesions of the brain characterized by clusters of dilated blood vessels (caverns), affecting 0.5% of the human population (Hage et al. 2025; Kar et al. 2017; Arauz et al. 2017). The genetic cause for CCM disease has been found mainly in the loss-of-function mutations in one of the three known CCM genes: Krev interaction trapped 1 (CCM1/KRIT1), cerebral cavernous malformations 2 (CCM2), and programmed cell death 10 (PDCD10/CCM3) (Draheim et al. 2014; Baranoski et al. 2016). Recent studies have revealed the role of an additional gene, PIK3CA, whose somatic gain-of-function mutation, combined to a CCM loss-of-function mutation, contributes to CCM lesion growth, suggesting that CCM arises from a cancer-like paradigm (Ren et al. 2021). In addition, the incomplete penetrance and variable expressivity of CCM disease has been attributed to the involvement of genetic modifiers of disease onset and severity (Trapani and Retta 2015; Choquet et al. 2016; Perrelli and Retta 2021). Discovered in 2001, the epidermal growth factor, latrophilin, and seven transmembrane domain-containing protein 1 (ELTD1), also known as adhesion G protein-coupled receptor L4 (ADGRL4), belongs to the member of the G-protein coupled receptor (GPCR) superfamily (Nechiporuk et al 2001). The GPCR superfamily includes 5 major families: glutamate (22 members), rhodopsin (701 members), adhesion (33 members), frizzled (11 members), and secretin (15 members), with ELTD1 belonging to the adhesion family (Serban et al. 2015). ELTD1 gene is located on chromosome 1p31.1, and it encodes for a 690 amino acids long protein containing an extracellular domain, an EGF-like domain, and a GPCR proteolysis domain. ELTD1 is reported to be an important player in brain angiogenesis and is regulated by the Vascular Endothelial Growth Factor (VEGF) and DLL4/Notch signaling pathways (Zalles et al. 2022), which are also known to play an important role in CCM pathogenesis (Goitre et al. 2010; DiStefano et al. 2014; Goitre et al. 2017). ELTD1 is strongly expressed in brain tissue, and a few studies reported its functional role in gliomas (Dai et al. 2015; Ziegler et al 2017). Towner et al. showed that ELTD1 can act as a putative biomarker in high-grade gliomas (Towner et al 2013), and a recent study by Zalles demonstrated that therapies exerting anti-ELTD1 activity and OKN-007, a nitrone anti-cancer agent, are effective for the treatment of patients affected with glioblastoma and recurrent gliomas (Zalles et al. 2022). Due to the role of ELTD1 in angiogenesis and brain, as well as cardiovascular disorders, we addressed the potential association of ELTD1 with CCM disease. We first analyzed the RNA-sequencing data from our previous published study of gene expression profiles in surgically resected human CCM lesions versus healthy brain as control (Subhash et al. 2019). Using an absolute log-fold change cutoff of 1.5 and a corrected p-value (p < 0.01), we identified ELTD1/ADGRL4 to be significantly increased in human CCM versus healthy brain controls, thus revealing a strong association between ELTD1 expression levels and CCM pathogenesis, and symptoms. Indeed, bioinformatic analysis revealed a significant correlation between ELTD1 expression level in plasma and the manifestation of Focal Neurological Deficits (FNDs) in CCM patients. Next, the gene expression level of ELTD1 was examined both in surgical samples of human CCM lesions, and in cellular and animal models of CCM disease, showing for the first time that ELTD1 is significantly upregulated compared to controls. Moreover, we found that ELTD1 is significantly upregulated in plasma samples from CCM patients and contributes to major molecular and phenotypical hallmarks like the ones induced by CCM genes loss, thus suggesting its role as a promising circulating biomarker for future studies. While confirming the potential role of ELTD1 in CCM disease pathogenesis and severity, these findings provide new insights into the identification of novel prognostic biomarkers and therapeutic targets that may contribute to the development of safe and effective precision medicine strategies for selective clinical management and target treatment of CCM patients. Methods Ethical statements The study protocol involving human subjects was strictly performed in accordance with the Declaration of Helsinki and its later amendments and approved by the local ethical committee at the Hannover Medical School, Germany (Approval Number 6960) and the Ethics Committee of University of Bonn Medical Center (Approval Number 076/08 and 042/08). All patients included in this study provided written informed consent. The animal care and experimental protocols followed the guidelines of the European Directive 2010/63/EU on the protection of animals used for scientific purposes and was approved by the local Animal Use and Care Committee of the University of Rochester (NY, USA). Patient tissue collection and preparation Only samples with confirmed neuroradiological and histopathological diagnosis for CCM were included in this study. Intraoperatively surgically resected lesions from patients suffering of cavernous malformations (n = 14) were collected on dry ice. Subsequently 4 surgical tissue samples from patients who underwent temporal lobe epilepsy (TLE) at the University Clinic of Bonn were used as healthy controls. The clinicopathological features of both the CCM and TLE patients are summarized in Tables 1 and 2. One part of the tissue sample was immediately stored at −80 °C for gene expression analysis. For immunohistochemical studies, tissues were fixed overnight in 4% buffered Formalin, embedded in paraffin, sectioned at a thickness of 4—5 µm and mounted on poly-L-lysine-coated slides. The patient clinical diagnosis and their histopathological assessment were based as described previously (Bertalanffy et al. 2002). Each patient manifested at least one or more clinical symptoms, including seizures, intracerebral hemorrhage (ICH), and focal neurological deficits (FNDs). A typical cavernoma located in the brainstem region of a 52 year-old female patient diagnosed through 1.5 Tesla MRI scan (Fig. 1). Table 1Clinicopathological characterization of CCM patientsPatient IDSex/AgeLesion LocationLesion size (volume cm3)Multiple LesionsMorphologySHEDVAOedemaSeizuresCCM 1M/66ypontine CCM1,01nobilobular with hematomasingleyesyesnoCCM 2F/22ymidbrain CCM8,49nocomplex/multi-lobular with cystsmultipleyesnonoCCM 3F/18yfrontal subcortical CCM14,52yescomplex/multi-lobular with cystssingleyesnoyesCCM 4F/33ymidbrain CCM4,53nobilobular with hematomamultipleyesyesnoCCM 5F/22ymidbrain CCM7,88nocomplex/multi-lobularmultipleyesnonoCCM 6F/29ybulbomedullary CCM0,18nosingle-lobularmultiplenononoCCM 7F/46ypontine CCM4,07nobilobular with hematomamultiplenoyesnoCCM 8F/52ypontine CCM3,33nobilobular with hematomamultiplenoyesnoCCM 9F/36ysubinsular CCM8,73nocomplex/multi-lobular with cystsmultipleyesnonoCCM 10F/39ymidbrain CCM0,46nobilobular with hematomasingleyesnonoCCM 11M/46ypontine CCM2,02nocomplex/multilobular with cystssingleyesnonoCCM 12F/29ypontine CCM0,61nobilobular with hematomasinglenononoCCM 13F/42yfrontal subcortical CCM0,37nosingle-lobularsinglenonoyesCCM 14F/36yfrontal subcortical CCM1,08nobilobular with hematomasinglenononoFND focal neurological deficit, ES epileptic seizures, y yes, n no, cerebral—TL cerebral temporal lobe, cerebellum-V vermis, SC single/few compartments—cysts, MC multiple compartments—cysts Clinicopathological characterization of CCM patients FND focal neurological deficit, ES epileptic seizures, y yes, n no, cerebral—TL cerebral temporal lobe, cerebellum-V vermis, SC single/few compartments—cysts, MC multiple compartments—cysts Table 2Clinicopathological characterization of patients with TLETLE Patient DataPatient NoAge/SexLocationHistological DiagnosisClinical diagnosisCon118/FAmygdalaLesionepilepsy with partial seizures after middle cerebral artery infarctionCon221/MAmygdalaLesionepilepsy with complex focal seizuresCon330/MAmygdalaLesionepilepsy associated with porencephaly in white matter and insula region rightCon424/MAmygdalaLesionMRI negative focal epilepsyTLEtemporal lobe epilepsy Clinicopathological characterization of patients with TLE TLEtemporal lobe epilepsy MRI image of an intra-axial brainstem cavernoma. Axial T2-weighted MRI image of a 52-year-old female with a large cavernoma located in the upper Pons (yellow arrow). The patient suffered from severe dizziness, numbness on the left side of the face and hypoesthesia on the left body side. Images and clinical data were kindly provided by Prof. Helmut Bertalanffy (Vascular Neurosurgery, INI, Hannover) Total RNA isolation and RNA quality from human CCM lesions Surgically resected tissue sections from CCM patients were stored in at least 10 volumes of RNAlater-ICE (Ambion, Austin, TX, USA) at −80 °C before RNA extraction. For RNA extraction, brain tissue specimens were sliced and further slit to smaller pieces. These pieces were transferred into vials containing ceramic beads (Precellys system, CK14L; 15,907,612, Fisher Scientific, Pittsburgh, PA, USA) and 700 µl of RLT lysis buffer from RNeasy Micro Kit (74004, Qiagen, Hilden, Germany) including 1% of beta-mercaptoethanol. Homogenization was performed by using the Precellys 24 Homogenizer (Peqlab, Erlangen, Germany) with 2—4 pulses of 5 s at 6000 rpm. Total RNA was extracted with RNeasy Micro Kit according to the company’s recommendations including an on-column DNase-I digestion step (5 min at RT). Total RNA was finally eluted in 14 µl of RNase free water. RNA samples were quantified, and purity was determined with the Nanodrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). RNA integrity was determined with an Agilent Bioanalyzer 2100 using the RNA 6000-Nano assay (5067—1511, Agilent Technologies, Santa Clara, CA, USA). RNA Integrity Numbers (RINs) of samples used for RNA-Sequencing experiments ranged from 7.0 to 8.4 with an average RIN of 7.6 for the cavernoma samples and 7.5 for the control samples. RNA library preparation, sequencing, raw data processing and quality control RNA library preparation was performed as described previously (Subhash et al. 2019). The RNA sequencing library was generated from 200 ng rRNA depleted total RNA with NEBNext® rRNA Depletion Kit (Human/Mouse/Rat), 96 reactions (New England Biolabs, Ipswich, MA, USA) followed by stranded cDNA library generation using NEBNext® Ultra Directional RNA Library Prep Kit for Illumina (New England Biolabs). Barcoded cDNA libraries were indexed using NEBNext Multiplex Oligos for Illumina – Set 1 (Index Primer 2, 3, 4, 5, 6, 7, and 12), and quantification was performed with the Qubit® dsDNA HS Assay Kit (Thermo Fisher Scientific, Darmstadt, Germany). For the sequencing run, 1.3 ml of the denatured pool was loaded on an Illumina NextSeq 550 sequencer using a High Output Flowcell for 75 bp single read sequencing (FC-404—2005, Illumina, San Diego, CA, USA). Each FASTQ file obtained a quality report using the FastQC tool. Before aligning to the reference genome, the raw FASTQ files were adapted and quality trimmed using Trim Galore (version 0.4.1) applying default settings as described in the User Guide, except for the setting of the quality cutoff (-q/–quality) which was set to a Phred score of 15. Differential expression analysis of protein-coding genes (PCGs) Differential expression (DE) analysis was conducted using Bioconductor package DESeq2 (Love et al. 2014) between healthy controls and CCM patient cohorts. Differentially expressed genes (DEG) and Protein-coding genes (PCG) have been corrected for adjusted p < 0.01 (FDR, Benjamini–Hochberg correction) and absolute log-fold change > 1.5 were considered to be significantly differentially expressed (DE). The unsupervised hierarchical clustering map of the significantly expressed PCGs was plotted using the Qlucore Omics Explorer® 3.6 (www.qlucore.com) software. Real-time PCR from human CCM tissue samples Briefly, 40 ng of total RNA was reverse transcribed to cDNA according to the manufacturer’s recommendations using the High-Capacity cDNA Reverse Transcription kit (4368814, Applied Biosystems, Waltham, MA, USA). Gene expression analysis was performed using the StepOne™ Real-Time PCR. Fold change was calculated using 2ˉΔΔCt method to determine the difference in expression level of ELTD1 between CCM lesions and control tissues. Changes in the mRNA expression levels were calculated following normalization to the housekeeping gene GAPDH. The TaqMan® assay IDs of the probes analyzed are the following: ELTD1 (Hs00381165_m1, Probe Context Sequence: > NM_080625.3:201—225 – AAGAGGTCAAGTTCCTGGGCCACCT –) and GAPDH (Hs02786624_g1, Probe Context Sequence: > NM_001256799.2:858—882 – CGCTGCCAAGGCTGTGGGCAAGGTC –). All assays were performed in triplicate. Real-time PCR from human CCM plasma samples Total RNA was extracted from 1 mL plasma samples using the Plasma/Serum Circulating and Exosomal RNA Purification Mini Kit (51000-NB, Norgen Biotek, Thorold, Canada) according to the manufacturer's protocol. Subsequently, cDNA was synthesized from 10 ng of total RNA using the High-Capacity cDNA Reverse Transcription Kit (4368814, Thermo Fisher Scientific), following the manufacturer's instructions. For quantitative PCR analysis, 2 µL of the synthesized cDNA was used to amplify the ELTD1 gene (TaqMan Assay ID: Hs00223377_m1, Thermo Fisher Scientific) and the housekeeping gene GAPDH (TaqMan Assay ID: Hs99999905_m1, Thermo Fisher Scientific) employing the TaqMan Fast Advanced Master Mix (4444963, Thermo Fisher Scientific) on a StepOne Real-Time PCR System (4444963, Thermo Fisher Scientific) with a total reaction volume of 20 µL, in accordance with the manufacturer's protocols. Each PCR reaction was performed in triplicate to ensure accuracy and reproducibility. Clinical correlation analysis Plasma ELTD1 levels were analyzed for associations with demographic, radiological, and clinical variables in patients with cerebral cavernous malformations. Continuous variables, including age and lesion size, were assessed using Pearson correlation analysis after confirming approximate normality of ELTD1 distribution. Binary clinical variables, including lesion location, DVA status, presence of oedema, hemorrhagic episode frequency, seizure status, sex, and presence of cranial nerve or focal neurological deficits, were evaluated using two-sample t-tests or non-parametric equivalents when appropriate. Given the limited sample size, analyses were performed in a univariate manner to avoid model overfitting. Stratified analyses were conducted to assess the robustness of observed associations within relevant clinical subgroups. All statistical tests were two-sided, and p < 0.05 was considered statistically significant. Immunohistochemistry Immunohistochemistry was carried out on paraffin-embedded tissue blocks from 3 surgically resected CCM specimens, and 3 adult normal brain without any clinical history of brain vascular malformations that are commercially available, have been used as control (HuFPT0CS, Amsbio, Abingdon, UK). Paraffin-embedded 5 µm thick sections were deparaffinized in two changes of xylenes (10 min each) and then rehydrated in graded series of ethanol dilutions (100%—2 times, 70%—2 times, and 50%—1 time). Antigen retrieval was performed using ready-to-use DAKO®-Target Retrieval solution (pH-6.0) (S236984—2, Agilent) in a decloaking chamber™ NxGen (DC2012—220 V, BioCare Medical, Pacheco, CA, USA) for 5 min at 110 °C. The sections were then treated with ready-to-use Dako® Peroxidase Blocking Reagent (S202386—2, Agilent) containing H2O2 for 5 min at RT to block the endogenous peroxidase activity (Dako). Then, the slides were probed with polyclonal anti-human ELTD1/ADGRL4 antibody (ab150489, dilution 1:400, Abcam, Cambridge, UK), and with polyclonal anti-human CD31/PECAM-1 antibody (NB100—2284, dilution 1:400, Novus Biologicals, Centennial, CO, USA). The primary antibodies were diluted in Dako REAL™ Antibody Diluent solution (S202230—2, Agilent). Control sections were incubated only with Dako REAL™ Antibody Diluent. All incubation steps were performed overnight at 4 °C inside a humidified chamber. Slides were then washed and incubated with ready-to-use Dako REAL™ EnVision™ HRP Rabbit/Mouse secondary antibody (K500711—2, Agilent) for 30 min at RT. All washing steps were performed in 0.01 M phosphate buffer (PBS, pH 7.4). Color development and labeling was visualized by a 5—10 min incubation with Dako REAL™ 3,3’-diaminobenzidine tetrahydrochloride (DAB, K500711—2, Agilent) chromogen and hydrogen peroxide as a substrate buffer which produces a brown color precipitate at the site of target antigen. The slides were then subsequently counterstained with Dako REAL™ hematoxylin (CS70030—2, Agilent) for 5 min, washed and dehydrated in graded series of ethanol (50%—1 time, 70%—2 times, and 100%—2 times), xylenes and mounted with Entellan (1079610500, Merck, Darmstadt, Germany). For visualization, images were captured with a Brightfield Leica DM 3000 microscope at a 40 × magnification. Semi-quantitative analysis was performed using ImageJ/Fiji to quantify the proportion of ELTD1-positive endothelial cells. Brightfield IHC staining for ELTD1 and PECAM1 was performed on parallel sections. Images were acquired at 40 × magnification using identical acquisition/exposure settings across all samples. For each sample, five representative fields were captured, and a manual ROI was drawn around the cavernous regions to exclude non-lesional tissue. Background subtraction was applied prior to thresholding. Endothelial cells were identified based on PECAM1 positivity. DAB-positive signal was segmented using the Otsu auto-threshold method, and objects were quantified using Analyze Particles with a size filter of ≥ 40 µm2 and circularity 0.00–1.00. The percentage of ELTD1-positive endothelial cells was calculated as: % ELTD1 = (ELTD1/PECAM1 cells *100)/Total PECAM1 cells. Values from the five fields were averaged to obtain one value per sample. Data are presented as mean ± SEM (controls n = 4, CCM n = 10). CCM-silenced HUVEC Human Umbilical Vein Endothelial Cells (HUVECs) (Thermo Fisher Scientific) were cultured in Dulbecco’s Modified Eagle’s Medium DMEM/F-12 (1:1 ratio), supplemented with 5% fetal bovine serum (FBS), 1% antibiotic–antimycotic solution (Gibco/Thermo Scientific, Waltham, MA, USA), 1% endothelial cell growth supplement (ECGS; ScienCell, Carlsbad, CA, USA), and 50 µM heparin (Calbiochem, La Jolla, CA, USA). Only cell passages 2–6 were used for experiments. All cells were cultured at 37 °C with 5% CO2. For either KRIT1, CCM2, or CCM3 silencing, cells were transduced with lentiviral particles containing specific shRNA against either KRIT1, CCM2, or CCM3, and compared with a shRNA targeting a scramble sequence (Scr) as negative control. Lentivirus production Lentiviral particles were produced by transfecting HEK293T cells with packaging plasmid pMDLg/pRRE, envelope plasmid pMD2.G, transcription factor plasmid pRSV-Rev, and transfer vector pLKO.1 containing either scramble shRNA, KRIT1 shRNA TRCN0000072879, CCM2 shRNA TRCN0000083234, or CCM3 shRNA TRCN0000144228, using Lipofectamine 2000 (11668019, Invitrogen, Waltham, MA, USA). Lentiviral particles were precipitated and then collected in serum-free DMEM/F-12 (1:1). HUVEC were transduced with lentivirus in complete medium containing 8 µg/ml hexadimethrine bromide (Polybrene, Sigma-Aldrich, Burlington, MA, USA). RNA isolation, reverse transcription, and RT-PCR in HUVECs The gene expression analysis was performed by real-time quantitative reverse transcription PCR (RT-PCR). The RNA was isolated using TRIzol Reagent (15,596,026, Life Technologies, Carlsbad, CA, USA), and the Direct-zol RNA MicroPrep Kit (R2060, Zymo Research, Irvine, CA, USA) following manufacturer’s instructions. RNA was eluted in ultrapure H2O and immediately used to prepare cDNA using the LunaScript® RT SuperMix Kit (E3010L, New England BioLabs), and the DNA used for RT-PCR. RT-PCR was performed with 2 µl of 1 st cDNA strand from each sample using the iTaq™ Universal SYBR® Green Supermix (1725121, Bio-Rad, Hercules, CA, USA) and the appropriate SYBR Green Primer Set for ELTD1: Frw – AGCTGCTTTTGCATGGATGTGC –, Rev – TCGGTGCTAAGCCAACATACTTTGG –. SYBR Green Primer Set for GAPDH was used as internal control for normalization: Frw – TGGTATCGTGGAAGGACTCATGAC –, Rev – ATGCCAGTGAGCTTCCCGTTCAGC –. Using the 2ˉΔΔCt method (Pfaffl 2001) amount of target RNA was calculated relative to negative control (i.e. scramble shRNA). All analyses were performed using three biological replicates, and results from RT-PCR are presented as mean values (± Standard Deviation, SD) of the normalized mRNA. KRIT1-KO mouse models A mouse model containing both conditional floxed allele for Krit1 (Krit1flox/flox), and the endothelial inducible Cre recombinase (PDGFB‐iCreERT2), as previously described (DiStefano and Glading 2020), was used (Krit1flox/flox;PDGFB‐iCreERT2). Cre-recombinase activity was induced by administration of a single intragastric injection of tamoxifen (40 mg) at P1, to induce endothelial knockout mice (Krit1ieKO). Tamoxifen‐treated Cre‐negative Krit1flox/flox (Krit1WT) mice were used as littermate control. Mice were bred and maintained under standard conditions in the University of Rochester animal facilities, which are accredited by the American Association for Accreditation of Laboratory Animal Care. All protocols were approved by the University Committee on Animal Resources. RNA isolation, reverse transcription, and RT-PCR of mouse brain tissue Adult mice (8 weeks old) were sacrificed using CO2 asphyxiation, and the brain collected into a sterile tissue culture dish. Brain tissues were homogenized using TRIzol Reagent (Life Technologies), and the RNA was isolated using Direct-zol RNA MicroPrep Kit (Zymo Research) following manufacturer’s instructions. RNA was eluted in ultrapure H2O and immediately used to prepare cDNA using the LunaScript® RT SuperMix Kit (New Enlgand BioLabs), and the DNA used for Taqman Assay RT-PCR. RT-PCR was performed with 2 µl of cDNA from each sample using the FastStart™ Universal Probe Master (Rox) (Roche, Basil, Switzerland) and the appropriate TaqMan® Gene Expression Assays (ELTD1 (Mm00500151_m1) and GAPDH (Mm99999915_m1, Thermo Fisher Scientific). GAPDH was used as internal control for normalization. Using the 2ˉΔΔCt method (Pfaffl 2001) amount of target RNA was calculated relative to wild-type (KritWT). A total of four brains per genotype were tested, and results from RT-PCR are presented as mean values (± Standard Deviation, SD) of the normalized mRNA. CCM2 flox Slco1c1-CreERT2 -/+ mouse line All animal experiments were conducted in compliance with ARRIVE guidelines and the European Directive 2010/63/UE for animal welfare, with protocols approved by the Lariboisiere-Villemin Ethics Committee (APAFIS no. 12084—201710261854702v3). Mice were housed in a conventional facility with standard environmental conditions and monitored daily for welfare. The study used C57BL/6 background mice, with conditional Ccm2 floxed mice crossed with the Slco1c1-CreERT2 line to generate a tamoxifen-inducible brain endothelial cell-specific CCM2 knockout (CCM2KO) model. Tamoxifen was administered via intragastric injection at postnatal day 1 (20 µg/g). Post-induction, animals were monitored weekly, and no exclusions were used. Analysis was performed at 4 months of age. Real-time PCR from CCM2 knockout mice Total RNA was extracted by using RNEasy Mini Kit (74104, Thermo Fisher Scientific) according to the manufacturer's protocol. Then, 15 ng of total RNA was converted into cDNA following the manufacturer's instructions using the High-Capacity cDNA Reverse Transcription Kit (4368814, Applied Biosystems). Gene expression levels were quantified using the StepOne™ Real-Time PCR system. The 2ˉΔΔCt method was applied to calculate fold changes in Eltd1 expression between CCM2KO and wild-type mice (CCM2fl/fl), with results normalized against the housekeeping gene Gapdh. The TaqMan® assay primers used for this analysis are the following: Eltd1 (Mm00500151_m1) and Gapdh (Mm99999915_g1). Each assay was conducted in triplicate. Quantification of mRNA expression by RT-PCR from krit1ty219c mutant zebrafish For RT-PCR experiments, 25 zebrafish embryos were pooled for each condition (three biological replicates). Total RNA was extracted with TRIzol (Sigma Aldrich) and Phase Lock Gel Heavy tubes (1.5 mL, Prime 5) and the corresponding cDNA was synthesized from total RNA with the RevertAid H Minus First Strand cDNA Synthesis kit (Thermo Fisher Scientific). RT-PCR experiments were performed as described (Renz et al. 2015) using 9 ng of cDNA per technical replicate and the KAPA Sybr Fast qPCR kit (K4600, Sigma-Aldrich) on a qTOWER3 Real-time Thermal Cycler (Analytik Jena, Jena, Germany). Cycle threshold (Ct) values were determined by qPCRsoft auto software (Analytik Jena). The quantitative expression levels of housekeeping gene eif1b were used for normalizing adgrl4/eltd1 levels. The following primers were used for RT-PCR: adgrl4: Frw – CACATTTGAACCCC GCCAAG –, Rev – ACGGATGTACTGCCATTGTTG –; eif1b: Frw –CAGAACCTCCAGTCCTTTGATC –, Rev –GCAGGCAAATTTCTTTTTGAAGGC –. Results were analyzed using the comparative threshold cycle method (2ˉΔΔCt) to compare gene expression levels between samples as previously described (Livak and Schmittgen 2001). As an internal reference, we used zebrafish eif1b (Renz et al. 2015). Control sample values were normalized to 1. Shown is the mean of fold changes and the corresponding ± SEM for each biological replicate after normalization. As every single biological replicate represents an independent experiment from an independent clutch of embryos, ratio paired t-tests were done with Prism 10 (GraphPad). Cell culture HEK293T cells (a gift from Gregory Tall, University of Michigan) were cultured in DMEM containing high glucose, supplemented with 5% FBS, 1% penicillin/streptomycin/L-glutamine (Thermo Fisher Scientific), and 1% nonessential amino acids (Thermo Fisher Scientific). Western blotting Lysates were prepared by washing cells with 1 × Tris-buffered saline (TBS; pH 7.4), then scraping cells into lysis buffer (50 mM Tris pH 7.4, 150 mM NaCl, 5 mM MgCl2, 0.5% NP-40) containing cOmplete mini EDTA-free protease inhibitor tablet (Roche, Basel, Switzerland) and phosphatase inhibitor cocktail (PPI; final concentration 100 μM Na4P2O7, 1 mM NaF, 1 mM Na3VO4, 300 μM β-glycerophosphate, 2 mM imidazole, 4 mM sodium tartrate). Lysates were incubated at 4 °C for 30 min with constant agitation, then clarified by centrifugation at 18,000 g for 30 min. Protein content was quantified using a bicinchoninic acid (BCA) assay kit (23227, Thermo Fisher Scientific) according to the manufacturer’s instructions. Appropriate volumes of lysate were mixed with 5 × Laemmli sample buffer. Samples have not been heated at 95 °C for 5 min to avoid the denaturation and aggregation of GPCR protein, including ELTD1. Samples were resolved by SDS-PAGE on precast 4–20% Tris–HCl gels (BioRad, Hercules, CA), transferred to a 0.45-μm nitrocellulose membrane, then blocked in 5% (w/v) nonfat dry milk (NFDM) in 1 × TBS plus 0.3% Tween 20 (TBST). Primary antibody was diluted in TBST, and the membrane was incubated in primary antibody overnight at 4 °C while rocking. The membrane was washed three times for 5 min with TBST and incubated in secondary antibody (diluted in TBST) for 1 h at room temperature. The membrane was then washed three times for 5 min with TBST to remove excess secondary antibody. Next, either SuperSignal West Femto or Pico enhanced chemiluminescent substrate (34095, Thermo Fisher Scientific) was applied to the membrane, which was then imaged on a BioRad ChemiDoc imaging system. Densitometry quantification was performed using ImageJ software (Fiji). When blotting for multiple proteins was necessary, blots were cut and probed separately. Results were normalized with respect to densitometric values of either GAPDH bands detected with a specific antibody and used as the internal loading control. Control sample values were normalized to 1, and the other samples were compared with this value. Results were analyzed using one-way ANOVA with Tukey post-hoc testing. Antibodies For western blotting, primary antibodies used in this study were the following: Rabbit Polyclonal Anti-ADGRL4/ELTD (BS-13111R, 1:500 dilution, Bioss Antibodies, Woburn, MA, USA); Rabbit Polyclonal Anti-COX2 (PA5—16,817, 1:1000 dilution, Thermo Fisher Scientific); Mouse Monoclonal Anti-NOX4 [Clone 10A8.1] (MABC616, 1:800 dilution, Sigma-Aldrich, St. Louis, MI, USA); Rabbit Polyclonal Anti-NOX2/gp91 (ABIN750690, 1:500 dilution, Antibodies Online, Pottstown, PA, USA). Primary antibodies were diluted in 1 × TBS. Secondary antibodies used in this study were the following: HRP-conjugated anti-Rabbit IgG (7074S, 1:5000 dilution, CST, Danvers, MA, USA); HRP-conjugated anti-Mouse IgG (7076, 1:5000 dilution, CST). Secondary antibodies were diluted in 1 × TBS containing 0.3% Tween (TBST). Cell migration assay An in vitro wound healing migration assay was performed as previously described (DiStefano et al. 2014). HEK293T cells were seeded on 6 well plates at 60% confluence. The following day, cells were either transduced with the appropriate lentiviral shRNA against CCM3 for 24 h, or transfected with 2 µg of ELTD1-Tango plasmid (66275, Addgene, Watertown, MA, USA) in Lipofectamine™ 2000 transfection reagent (Thermo Fisher Scientific), following manufacturer’s instruction, for 6 h. Once cells reached 100% confluence an area was denuded by a rubber policeman at the center of each well. Images were taken at the time of wounding (0 h) and after 24 h, and the relative distance traveled by the cells at the acellular front was determined using ImageJ. Data were analyzed using one-way ANOVA with Tukey’s post-hoc testing. Cell proliferation assay Thiazolyl blue tetrazolium bromide (MTT) was used to assess cell proliferation. HEK293T cells were seeded on 10 cm plates at 60% confluence. The following day, cells were either transduced with the appropriate lentiviral shRNA against CCM3 for 24 h, or transfected with 2 µg of ELTD1-Tango plasmid (Addgene) in Lipofectamine™ 2000 transfection reagent (Thermo Fisher Scientific), following manufacturer’s instruction, for 6 h. The following day, 20.000 cells/well were seeded on a 96 well plate and incubated at 37 °C and 5% CO2. After 24 h, the culture media was removed and replaced with 110 µl/well serum-free and phenol red-free media, containing 10% of MTT (475989, Sigma-Aldrich) dissolve into 1 × sterile PBS. Cells were then incubated for 3 h at 37 °C and 5% CO2, to allow the reduction of MTT to MTT-formazan. At the end of incubation period, 50 µl/well of DMSO (Sigma-Aldrich) was added, and plate was shaken for 10 s. Then cells were incubated again for 10 min at 37 °C, to induce the complete release of MTT-formazan from cells. Spectrophotometric measurement of MTT-formazan was performed with absorbance at 570 nm using a Bio-Rad Benchmark Plus plate reader (Bio-Rad). Wells treated with 10% MTT in 1 × PBS only, and not containing cells, were used to measure the background for MTT, and absorbance at 570 nm was subtracted to initial absorbance of cell samples. The results were normalized to the background, and the proliferation rate compared to HEK cells Scr silenced. Data were analyzed using one-way ANOVA with Tukey post-hoc testing. Cell permeability assay HEK293T cells were seeded on to Transwell inserts (3 μm pore size, 2 × 106 pores/cm2, Corning Life Sciences, Corning, NY, USA) coated with 10 μg/ml FN. The following day, cells were either transduced with the appropriate lentiviral shRNA against CCM3 for 24 h, or transfected with 2 µg of ELTD1-Tango plasmid (Addgene) in Lipofectamine™ 2000 transfection reagent (Thermo Fisher Scientific), following manufacturer’s instruction, for 6 h. To verify that the cells reached confluence on the opaque membrane, HEK cells were seeded in parallel on FN-coated wells in a 24-well dish and transduced or transfected the same way as above. Given that the monitor wells demonstrated a confluent monolayer, 20 μg/ml FITC-conjugated dextran (40 kDa; FD40, Sigma-Aldrich) diluted in assay medium (0.5% FBS, 0.1% ECGS, without Phenol Red) was then added to the upper chamber of the Transwell and allowed to diffuse through the cell layer for 4 h at 37 °C. Medium from the bottom chamber was then collected and loaded on a black 96-well plate (Corning Life Sciences) in triplicate. FITC fluorescence was measured at excitation 485 nm/emission 528 nm using a BioTek Gen5 Synergy LX multimode plate reader (Software Version 3.12, Agilent, Santa Clara, CA, USA) and compared to a standard curve of FITC-dextran diluted in assay medium. Data were analyzed using one-way ANOVA with Tukey post-hoc testing. Statistical analysis The significance of differential expression levels of ELTD1 between CCM models and corresponding controls was determined with unpaired t-test, and ordinary One-way ANOVA. All results from RT-PCR and western blot analyses are presented as mean values (± Standard Error of the Mean -SEM-, and ± Standard Deviation -SD-) of the normalized mRNA and protein levels, respectively. A p-value ≤ 0.05 was considered statistically significant. Results RNA transcriptome profile identifies ELTD1 to be significantly upregulated in CCM patients A growing body of evidence demonstrates that ELTD1 is aberrantly overexpressed in high grade brain tumors, such as glioblastoma, and in cardiovascular disorders (Choquet H et al. 2016; Goitre L et al. 2017; Hage SF et al. 2025; Huang H et al. 2022). Conversely, the role of ELTD1 in cerebrovascular diseases, including CCM pathogenesis, has not been clearly investigated. Therefore, we sought to determine whether the pro-angiogenic marker ELTD1 could be detected in CCM disease. The fast development of next-generation sequencing (NGS) technology in recent years has significantly contributed to the identification of low-frequency transcripts, allowing higher sensitivity and specificity in the analyses. To identify whether ELTD1 was involved in the pathogenesis of CCM disease, we assessed the published RNA-sequencing data performed on the sporadic forms of human CCM lesions (Subhash et al. 2019). Differential expressions for both up- and down- regulated protein coding genes (PCGs) were analyzed by comparing the expression patterns between CCM patients (n = 10) and healthy brain controls (n = 4). A total of 4,928 PCGs were identified to be significantly differentially expressed between CCM and control groups. Among them, there were 2,290 upregulated and 2,638 down-regulated PCGs, respectively. An absolute log-fold change greater than 1.5 and a corrected p-value (p < 0.01) were used to determine the DE transcripts between the CCM and the healthy control groups. Within the 2,290 upregulated PCGs, we identified ADGRL4/ELTD1 to be significantly increased (log fold change of 4.1) in CCM compared to matched controls (p < 0.001) (Fig. 2A). This finding suggested that ELTD1 may be a biomarker of CCM. Therefore, to assess whether the upregulation of ELTD1 was evolutionarily conserved, we further validate the expression levels of ELTD1 in distinct cellular and animal models of CCM disease, including human endothelial cells and mouse and zebrafish models. ELTD1 is upregulated in human CCM lesions. A Unsupervised hierarchical clustering map showing transcriptome profiles of 4928 protein-coding genes (PCGs) significantly differentially expressed between human CCM (n = 10) and healthy (n = 4) brain tissues (GEO accession number: GSE137596). Upregulated and downregulated PCGs are indicated in yellow and blue colors, respectively. ELTD1/ADGLR4 was identified among upregulated PCGs in human CCM tissues. A log-fold change cutoff of ±1.5 and a p < 0.01 were applied. B ELTD1 mRNA is increased in surgically resected specimen of CCMs. Bar diagram representing the fold change mean ±SEM in ELTD1 mRNA levels in human brain CCM tissue samples (CCMs; n = 10) vs paired controls (Control; n = 3), by RT-PCR gene expression assays. ***p < 0.001. C ELTD1 mRNA expression level is upregulated in plasma samples from CCM patients. Bar diagram displays the fold change mean ± SEM in ELTD1 mRNA levels in CCM plasma samples (CCM; n = 14) compared to controls (Control; n = 8), by RT-PCR assays. **p < 0.001. D ELTD1 shows high immunoreactivity in CCM patients. 1) Negative immunohistochemical control, incubated with non-immune serum, showing no ELTD1 staining. 2) Serial sections stained with PECAM showing endothelial-cell specific staining were used for comparison (red arrows). 3) Immunohistochemical analysis indicating that ELTD1 is highly expressed in endothelial cells lining human CCM lesions (red arrows). 4) CCM tissue samples with infiltrating lymphocytes (yellow arrows) around a blood vessel. 5) CCM lesion showing a macrophage (black arrow) positive for ELTD1. 6) Representative hematoxylin and eosin (H&E) staining of a histological section from a surgically resected human CCM lesion showing a cluster of abnormally enlarged capillaries (caverns) filled with blood cells and lined by a thin endothelium layer (asterisks). Scale bar: 100 µm. 7-8) Representative serial healthy brain tissue sections stained without and with ELTD1 antibody shows endothelial cells positive staining within a blood vessel (red arrows). E Semi-quantitative analysis of ELTD1in paraffin embedded CCM lesions, showing the percentage of endothelial cells positive for ELTD1 within a region of interest in CCM tissues (n = 10). Using ImageJ software we observed an increased percentage of ELTD1-positive endothelial cells compared to healthy brain tissues (CT). Bars show mean ± SEM. **p < 0.001 Increased expression of ELTD1 mRNA levels in CCM tissue and plasma patients We then analyzed the expression of ELTD1 mRNA level in a separate CCM patient cohort, through quantitative real-time PCR (RT-PCR) analysis of tissue samples from surgically removed cavernomas (n = 10). Interestingly, our results showed a robust upregulation of the endogenous ELTD1 (> 73 fold higher) in tissue from CCM patients compared to control brain tissues (p < 0.001, Fig. 2B). In addition, we assessed the expression level of ELTD1 mRNA in plasma samples from CCM patients (n = 14). Notably, our serum findings aligned with both CCM tissue and RNA-seq data, revealing a significant upregulation of endogenous ELTD1 mRNA in the sporadic form of disease compared to healthy brains (8.5 log-fold change; p < 0.001) (Fig. 2C). ELTD1 immunoreactivity in endothelial cells lining CCM caverns and in infiltrating lymphocytes To assess the localization of ELTD1 expression in human CCM lesions, histological sections were prepared from paraffin-embedded surgically resected CCM specimens and analyzed by immunohistochemistry (IHC) (Fig. 2D, panels 1—5) and hematoxylin and eosin (H&E) staining (Fig. 2D, panel 6). H&E staining showed the presence of multiple CCM caverns lined by a thin endothelium layer (Fig. 2D, panel 6, asterisks), whereas IHC with a specific ELTD1 antibody showed a positive staining both in endothelial cells lining CCM lesions (Fig. 2D, panels 3—5) and in associated infiltrating lymphocytes and macrophages (Fig. 2D, panels 3—5, arrows). Serial histological sections incubated with non-immune serum instead of primary antibody showed no staining, serving as negative controls (Fig. 2D, panel 1). Serial sections from healthy brain tissues showed both endothelial-specific ELTD1 staining lining healthy blood vessels and neurons (Fig. 2D, panels 7 and 8). No positive immunostaining was observed in neurons from CCM lesions, as they are mostly excluded from the lesion area during the progression of disease, and because of the low expression of ELTD1 in neuronal cells. The semi-quantitative analysis of IHC images revealed that the percentage of ELTD1-positive endothelial cells is significantly higher (p < 0.05) in CCM tissues compared with healthy brain tissues (Fig. 2E). These findings strongly support the conclusion that endothelial cells represent the predominant cell type expressing ELTD1 protein in CCM disease. Remarkably, our finding that ELTD1 is significantly expressed in such infiltrating lymphocytes and macrophages within the lesion raises the possibility that it may contribute to long-term neuroinflammatory responses associated with CCM disease progression and severity. ELTD1 increase correlates with focal neurological deficits The robust upregulation of ELTD1 in human CCMs suggests its potential involvement in disease pathogenesis, and a potential use as prognostic and diagnostic marker. Therefore, to test whether the enhanced ELTD was specifically associated with any symptom or patient condition, we systematically evaluated the association between plasma levels of ELTD1 and multiple demographic, radiological, and clinical parameters, including patient age and lesion size (Fig. 3 A,B), lesion location, presence of developmental venous anomalies (DVA), oedema, seizure, hemorrhagic episodes, sex, lesion morphology, and focal neurological deficits (Fig. 4 A-H). Intriguingly, ELTD1 level did not correlate with patient age or lesion size. ELTD1 level does not correlate with age or lesion size. A, B The boxplots represent the correlation between the plasma level of ELTD1 (x axis) and the age of patients (y axis, A), or the size of the cavernomas (y axis, B). Results show no significant correlation between those parameters and ELTD1 expression level ELTD1 is increased in patients with Focal Neurological Deficits. A-H The boxplots represent the correlation between the plasma level of ELTD1 (y axis) and either demographic or clinical parameters, which most commonly represent symptoms or pathological conditions in CCM patients (x axis). Results show that plasma ELTD1 expression level correlates with only the manifestation of Focal Neurological Deficits (FNDs) in patients (H). All other parameters do not show a significant correlation with ELTD1 levels (A-G) Similarly, ELTD1 did not correlate with any clinical parameter among lesion location, DVA, oedema, seizure, hemorrhagic episode frequency, sex, or lesion morphology (Fig. 4 A-G). However, our results reveal a positive correlation between ELTD1 expression and focal neurological deficits (FNDs), a common and clinically significant manifestation in CCM patients (Fig. 4H). These findings strongly support the clinical relevance of ELTD1 as a potential predictive biomarker in CCM. Notably, plasma ELTD1 levels were significantly higher in patients presenting cranial nerve or focal neurological deficits compared with those without such deficits (p = 0.0437), suggesting an association with clinically relevant neurological impairment rather than anatomical lesion characteristics alone (Figs. 3A, and 4 A, G). Additional stratified analyses did not reveal further significant associations within clinical subgroups, indicating that ELTD1 primarily reflects the presence of neurological deficits rather than disease heterogeneity within patients. ELTD1 is upregulated by depletion of CCM genes Next, to gain further insight into the regulation of ELTD1 in different CCM models, we measured the expression level of ELTD1 in human endothelial cells in which the three CCM genes were individually silenced using lentiviral shRNA. Effective silencing of each CCM gene was validated by RT-PCR, which demonstrated a reduction in mRNA of between 60 to 85% as compared to cells transduced with scramble control (Data not shown). RT-PCR analysis of silenced cells showed that ELTD1 mRNA expression levels were significantly enhanced in KRIT1 (40% increase), and CCM2 (60% increase), and CCM3 (50% increase) knockdown HUVECs as compared to cells transduced with the scramble shRNA (Fig. 5). These data suggests that ELTD1 is upregulated in response to loss of expression of any CCM gene. ELTD1 mRNA is upregulated in in vitro models of CCM disease. Bar diagram representing the mean fold change ± SD of ELTD1 mRNA levels in Human Umbilical Vein Endothelial Cells (HUVEC) 48 h after knock-down using short interfering RNAs against CCM genes (shKRIT1, shCCM2, shCCM3), compared with HUVEC treated with a scramble shRNA as control (shScr). n = 3 biological replicates. **p < 0.005; *p < 0.05 vs. control (Scr) Eltd1 is upregulated in Krit1ieKO mice and in Ccm2 deleted mice Since data from in vitro analysis strongly suggested the idea that ELTD1 might be involved in the development of CCM, we thought to determine the expression levels of ELTD1 in vivo. To test this directly, we utilized Krit1ieKO mice, which develop lesions similar to those seen in human CCM patients (DiStefano and Glading 2020). Tamoxifen injection of Krit1flox/flox; PDGFB‐iCreERT2 mice at P1 stimulates endothelial‐specific Cre‐mediated recombination and loss of Krit1 protein in > 90% of the endothelium (DiStefano and Glading 2020). We sacrificed 8 weeks old mice and collected brain tissue. Then we measured the expression level of Eltd1 in homogenized tissue from both wild-type (Krit1WT) and knockout mice (Krit1ieKO). RT-PCR analysis of brain tissues showed that the expression level of Eltd1 mRNA enhanced in Krit1ieKO brain tissue (> twofold increase), as compared to Krit1WT (Fig. 6A). These data suggest a potential relationship between Krit1 deficiency and molecular mechanisms underlying the control of ELTD1 expression, which deserves further investigation. We then extended our findings to understand whether ELTD1 is similarly upregulated in other in vivo model of CCM, and we analyzed the brain tissue from mice with a conditional knockout of the Ccm2 gene using the inducible brain endothelial specific Slco1c1-CreERT2 mouse line. Specifically, we examined Eltd1 mRNA level in n = 6 brain hemispheres of 4 month-old Ccm2-KO and in n = 6 control mice (CCM2fl/fl). RT-PCR analysis revealed a significant upregulation of Eltd1 in the Ccm2-KO mice compared to controls, with a 1.7-fold increased expression (Fig. 6B). This upregulation further provides evidence that ELTD1 may contribute to pathogenic mechanisms underlying CCM disease onset, progression, and severity. ELTD1 expression is elevated in in vivo models of CCM disease. A Conditional knock-out of KRIT1 increases ELTD1 expression in mouse brain. ELTD1 mRNA levels in mouse brain tissues isolated from 8-week-old wild-type (Krit1WT) or Krit1ieKO mice. Data shown are mean fold change ± SD. n = 4, ****p < 0.0001. B Upregulation of ELTD1 expression in CCM2 knockout mice. The bar diagram illustrates the mean ± SEM fold change of ELTD1 mRNA levels in CCM2-KO compared to wild-type as measured by RT-PCR. Statistical analysis using Student’s t-test revealed a highly significant difference across three independent experiments, using two animals each (n = 6), indicating a marked upregulation of ELTD1 in CCM KO mice. **p < 0.001. C The expression of adgrl4/eltd1 was upregulated in zebrafish expressing mutated krit1 at 54 hpf. Bar diagram represents quantitative RT-PCR on whole krit1ty219c mutant embryos in comparison to WT zebrafish embryos. Eif1b served as the internal reference gene to normalize adgrl4/eltd1 mRNA expression. Data shown are mean fold change ± SEM. n = 3, *p = 0.014 ELTD1 is overexpressed in krit1ty219c mutant zebrafish embryos Next, we tested whether this gene was upregulated in a zebrafish model of CCM. Zebrafish mutants represent an excellent model to study the development of CCM disease in real time, thanks to a clearly visible development of cardiac and cerebral vasculature (Walcott and Peterson 2014). Null mutations in the ccm1 (santa, san) or ccm2 (valentine, vtn) induce the formation of enlarged heart and a dilated subintestinal vessels in Zebrafish (Cullere et al. 2015). Also, the evolutionary conservation of elevated adgrl4/eltd1 mRNA expression may lend further significance to findings from human and murine samples, since cardiac tissue contains approx. 35% of endothelial cells and resulted in the discovery of many genes with a differential expression in CCMs, such as klf2a (Renz et al. 2015). We selected the krit1ty219c mutant form and performed gene expression analysis using the zebrafish homolog of ELTD1 from extracted cardiac tissues of krit1ty219c and wild-type (WT) zebrafish embryos (Fig. 6C). This analysis revealed that mRNA levels of adgrl4/eltd1 were significantly higher in krit1ty219c mutant than in WT embryos at 54 h post-fertilization. Overexpression of ELTD1 replicates pathological phenotypes associated with CCM genes loss-of-function To test whether the increased level of ELTD1 was also associated with main pathological and molecular phenotypes, similar to those characterizing the CCM disease, and it was not only a random event due to CCM genes loss, we induced the overexpression of ELTD1 in Human Embryonic Kidney (HEK293T) cells. HEK293T cells were plated and cultured as described above, and either transduced with lentiviral particles with specific target CCM3 (shCCM3) or transfected with 2 µg of a Tango-flagged ELTD1 plasmid (ELTD1), and the results compared with cells transduced with a Scramble shRNA (shScr). We decided to utilize a well characterized in vitro wound-healing assay to examine whether the overexpression of ELTD1 could increase cell migration of HEK293T cells, thus resuming one of the main pathological phenotypes associated with the depletion of CCM genes. The silencing of CCM3 and the expression of ELTD1 were induced 48—72 h before wounding. Then, wound images were taken at time of wounding (0 h) and after 24 h. Over the 24 h span, ELTD1-expressing cells and CCM3-depleted cells migrated to cover ~ 50% of the wound space, while Scramble shRNA expressing (shScr) cells migrated to cover 30% of the original wound space (Fig. 7A). As CCM genes loss is known to increase cell proliferation, we performed a proliferation assay exerting the reduction of MTT into MTT-formazan. Expression of ELTD1 and silencing of CCM3 was induced as described above, and cells were exposed to MTT after 48 h. Over the 48 h span, cells overexpressing ELTD1 were proliferating 20% more than cells transduced with Scramble shRNA. Also, proliferation was 40% higher in cells silenced for CCM3 (Fig. 7B). ELTD1 overexpression replicates pathological phenotypes associated with CCM disease, and CCM genes silencing. A Human Embryonic Kidney Cells (HEK293T) were either transduced with lentiviral scramble (shScr) or anti-CCM3 (shCCM3) shRNA or transfected with ELTD1 plasmid to induce its overexpression. Percent wound closure of HEK293T cells was calculated with the distance traveled by cells between 0 and 24 h. Data shown are the mean ± SD across n = 6 biological replicates, in technical triplicate. ****p < 0.0001 vs scramble shRNA alone (shScr). Overall p < 0.0001. B Percent cell proliferation of HEK293T cells overexpressing ELTD1, or transduced with lentiviral scramble (shScr) or anti-CCM3 (shCCM3) shRNA. Data shown are the mean ± SD across n = 11 biological replicates. ****p < 0.0001 vs scramble shRNA alone (shScr). Overall p < 0.0001. C Permeability of HEK293T cell monolayers to 40 kDa FITC-dextran. ELTD1 overexpression was induced through plasmid transfection, or cells were transduced with lentiviral scramble (Scr) or anti-CCM3 (shCCM3) shRNA. Data shows the mean permeability ± SD from n = 8 biological replicates. ***p = 0.0001 and **p < 0.005 vs scramble shRNA alone (shScr). Overall p = 0.0001 Then, we performed a barrier function assay to determine the ability of ELTD1 overexpression to induce cell permeability when compared to physiological condition and CCM3 silencing. HEK expressing ELTD1, or anti-CCM3 shRNA (shCCM3) or control shRNA (shScr) were plated on 3-μm pore size polyester Transwell™ inserts and allowed to grow until a confluent monolayer. After that, barrier function was measured by tracking the flux of 40-kDa FITC-dextran through the monolayer (Fig. 7C). It is well known that CCM genes silencing resulted in loss of barrier function, as indicated by 24% increase in dextran permeability when compared to scramble shRNA. The induced expression of ELTD1 also increased the monolayer permeability by 30% compared to the control. Once we established the role of ELTD1 overexpression in some of the key phenotypes and cell behaviors that characterize CCM disease, we also wanted to determine whether it could play a role in the protein expression regulation of major players involved in CCM pathogenesis. In particular, we selected NADPH oxidase 2 and 4 (NOX2/4) and Cyclooxygenase 2 (COX2). Indeed, the upregulation of NOX2/4 and COX2 has been associated with increased inflammation, oxidative stress, and altered cell signaling and homeostasis, mostly induced by CCM genes loss in CCM models (Goitre et al. 2010; Goitre et al. 2014; Finetti et al. 2020). Therefore, we decided to analyze the protein expression level of these markers in HEK293T expressing ELTD1 or silenced for CCM3. Using the Western blot analysis and relative quantification method of the detected protein bands, we observed an increase in NOX2 (1.2 fold increase) and NOX4 (1.5 fold increase), and COX2 (1.3 fold increase) in both ELTD1 overexpressing cells and CCM3-silenced cells compared to scramble shRNA control (Fig. 8), consistent with above observations in IHC analysis. ELTD1 overexpression induces the expression level of COX2 and NOX4. A Human Embryonic Kidney Cells (HEK293T) were either transduced with lentiviral scramble (shScr) or anti-CCM3 (shCCM3) shRNA or transfected with ELTD1 plasmid to induce its overexpression. Total protein fraction was isolated and blotted for ELTD1, or NOX2, or NOX4, or COX2. Blots are representative of n = 5 biological replicates. B Densitometry quantification of COX2 and NOX4 in A. Data show mean band densities ± SD normalized to GAPDH, n = 5 biological replicates. *p < 0.05; **p < 0.005; ***p = 0.0001; ****p < 0.0001 vs scramble shRNA alone (shScr). Overall p < 0.05 Discussion Since its discovery in 2001, ELTD1, an orphan member of the GPCR superfamily, has been known to be associated with cardiac and renal malfunction, high-grade gliomas and colorectal cancers (Zalles et al. 2022; Lu et al. 2017; Sun et al. 2021; Xiao et al. 2012). ELTD1 is highly expressed in endothelial cells of the brain, where it regulates angiogenesis in both physiological and tumor conditions. ELTD1 expression is known to be induced by VEGF Receptor (VEGFR) and the Transforming growth factor beta (TGF-β), while it is repressed by Notch ligand Delta-like 4 (DLL4). The deregulation of ELTD1 expression alters endothelial sprouting in vitro and causes vascular abnormalities incompatible with life in zebrafish embryos (Masiero et al. 2013; Favara et al. 2019). Higher expression of ELDT1 is linked with poor patient survival, and worse outcomes in high-grade gliomas (Towner et al 2013; Huang et al. 2022). Innovative treatments using targeted antibodies against ELTD1 yielded promising results in increasing survival, reducing tumor volumes, and normalizing tumor vasculature in mouse models of glioma (Zalles et al. 2022). While several studies reported the diagnostic and prognostic value of ELTD1 in cancers, as well as its role as key angiogenic factor, an involvement of ELTD1 in cerebrovascular diseases, including Cerebral Cavernous Malformations (CCM), has not been investigated yet. Here, we demonstrate for the first time that ELTD1 is upregulated in CCM models, thus representing an interesting biomarker for future clinical studies. Our data revealed a robust increase of ELTD1 at the transcriptome and mRNA levels in tissue, as well plasma samples, from CCM patient cohorts, as showed in Fig. 2A-C. Patient cohort had a significant increased expression of ELTD1 in high-grade gliomas, gastric cancer, colorectal cancer, breast cancer, and renal cancer (Zalles et al. 2022; Towner et al 2013; Zalles et al 2021; Sun et al. 2021; Xiao et al. 2012; Sun and Zhong 2021; Sheldon et al. 2021). Intriguingly, a recent study by Li et al. revealed an increased expression level of ELTD1 in glioma patient cohorts compared to normal brains (Li et al. 2019). The authors found that ELTD1 expression was consistently enhanced in high grade gliomas (III and IV) and positively correlated with the Karnofsky Performance Scale (KPS), tumor recurrence, poor prognosis, and reduced overall survival (Li et al. 2019). Conversely, the knockdown of ELDT1 in two cell lines with epithelial morphology isolated from glioblastoma patients, U-87MG and U-138G, significantly suppressed cancer cell proliferation, migration and invasion (Li et al. 2019). These previous findings revealed that ELTD1 is overexpressed in glioma tumor tissues, and the targeted knockdown of ELTD1, prevented tumor growth in vivo. In this context, the identification of increased levels of ELTD1 mRNA in tissue and plasma samples from patients affected by the sporadic form of CCM disease (sCCM), representing about 80% of all cases (Bacigaluppi et al. 2013; Choquet et al. 2015), is a crucial step aimed at characterizing ELTD1 as a biomarker of prognostic and predictive value for the identification of clinically symptomatic cavernomas. Notably, the positive correlation between ELTD1 increased plasma level and patients with FNDs (Fig. 4H), a common and clinically significant symptom of CCMs, it will help in better and more effective patient stratification, as it happens with different grade of gliomas, and it strongly supports the clinical relevance of ELTD1 as a potential predictive biomarker in CCM. To determine whether the expression level of ELTD1 was indeed associated not only with the sporadic form of the disease, but also with the loss-of-function mutation of CCM genes, we performed gene expression analysis of ELTD1 in specific cellular models, and preclinical animal models of CCM, including mice and zebrafish. Surprisingly, the expression level of ELTD1 mRNA was significantly increased in HUVECs individually silenced for KRIT1, CCM2, and CCM3 compared to the scramble shRNA (Fig. 5). Taking advantage of well-established murine models of CCM disease, we performed gene expression analysis to determine whether ELTD1 was overexpressed in vivo models of CCM. Our results showed a significant upregulation of ELTD1 in Krit1ieKO, CCM2 knockout mice, and krit1ty219c zebrafish compared to their wild-type littermates, respectively. This data strongly indicates the involvement of ELTD1 not only in the sporadic form of the disease but also in the familial form of CCM (fCCM), at least those driven by loss of KRIT1 or CCM2. However, similar expression levels could not be observed in brain endothelial cells isolated from Pdcd10-konckout mice as shown in Supplemental Fig. 1 (Czupalla CJ et al. 2018; Fusco C et al. 2022; Tang AT et al. 2019; Bravi et al. 2015; Assmann JC et al. 2017). It is possible that, as we used whole brain tissue to study the expression level of ELTD1 in Krit1ieKO and CCM2 knockout mice, the observed changes are due to altered expression of ELTD1 in multiple cell types, which are absent from our PDCD10 knockout samples. It is possible that other cell types present in the neurovascular unit (NVU), such as pericytes or lymphocytes, could contribute to a higher expression of ELTD1 in Krit1ieKO and CCM2 mice, in addition to brain endothelial cells, as we could observe in lesion from CCM patients (Fig. 2D, panels 4 and 5). Indeed, as shown by Orsenigo et al. (Orsenigo et al. 2020), ELTD1 might be significantly overexpressed only in endothelial tip cells following the stimulation of angiogenic factors, such as VEGF. Accordingly, ELTD1 is not upregulated in KRIT1-knockout cultured brain microvascular ECs derived from KRIT1fl/fl mice after inactivating KRIT1 (Lopez-Ramirez et al. 2017). However, ELTD1 expression was increased by KRIT1, CCM2, and PDCD10 depletion in HUVEC, and in human tissues, ELTD1 was aberrantly increased in endothelial cells lining the human CCM caverns. Indeed, ELTD1 is included in plasma protein class, and can spread in the blood flow inside extracellular vesicles from endothelial cells (Sheldon et al. 2022). Our histological and immunohistochemical analyses revealed a significant increase of ELTD1 immunoreactivity in the vasculature of CCM lesions. These data are consistent with previous studies showing brain endothelial-specific expression of ELTD1 in high-grade gliomas of the central nervous system and brain tumors (Huang et al. 2022). The increased immunoreactivity of ELTD1 in endothelial cells lining large cavernomas may suggest that it contributes to the sustained angiogenesis characterizing CCM capillaries and is directly involved in CCM lesion formation. Interestingly, the immunohistochemistry detection of a strong ELTD1 immunoreactivity in both CCM lesion endothelial cells and infiltrating lymphocytes and macrophages (Fig. 2), raises the possibility that it may contribute to long-term neuroinflammatory responses associated with CCM disease progression and severity. Indeed, the loss of mural cells in CCM, including pericytes and smooth muscle cells, affects vascular stability and endothelial–mural cell crosstalk. This altered microenvironment may also contribute to the elevated proportion of ELTD1-positive endothelial cells observed in CCM lesions. However, despite endothelial cells show most of the ELTD1 staining in CCM lesions, whether other cell types within the NVU, and in what percentage, also play a role in regulating the expression of ELTD1 is worthy of further investigation. Loss-of-function mutations in the three CCM genes increase the production of angiogenic and triggers the downregulation of anti-angiogenic factors in affected endothelial cells (Perrelli and Retta 2021; Wustehube et al. 2010; Boulday et al. 2009; You et al. 2013; Perrelli et al. 2022). This shift in molecular regulatory mechanisms from quiescent to proliferative endothelial cell phenotype results in CCM formation. Moreover, our in vitro model overexpressing ELTD1 demonstrated that this receptor might play a crucial role in vascular remodeling, blood vessel formation and growth, and endothelial cell proliferation during CCM disease, and endothelial cell permeability, as observed in HEK293T human cell line (Fig. 7). These functions of ELTD1 could then result in severe clinical outcomes, including ICH and stroke, or FDNs. Indeed, recent studies demonstrated that ELTD1 plays a major role as an angiogenic regulator, and its activation in endothelial cells results in the activation of various signaling pathways inducing major vascular remodeling (Masiero et al. 2013; Buzatu et al. 2023; Serban et al. 2017). We also observed that the induced overexpression of ELTD1 in HEK cells resume some of the major molecular hallmarks of CCM disease, including the activation of pro-oxidative and pro-inflammatory responses. This happens through the upregulation of enzymes involved in the production of reactive oxygen species (ROS), or inflammatory prostaglandins (Fig. 8), that are usually correlated with the loss-of-function of CCM genes (Goitre et al. 2010; Goitre et al. 2014; Antognelli et al. 2018; Cianfruglia et al. 2019). These results, together with the positive ELTD1 staining in lymphocytes and macrophages in proximity of CCM lesions, confirm the role of ELTD1 as a major player in inflammatory responses. Moreover, this altered cell homeostasis and inflammatory condition might result in further pathological phenotypes of CCMs, including Endothelial-to-Mesenchymal Transition (EndMT) or impaired angiogenesis, as suggested by the function of ELTD1 in other diseases (Sheldon et al. 2021; Sheldon et al. 2022). How ELTD1 could be upregulated by loss of KRIT1, CCM2 or PDCD10, and how it could trigger inflammation, and cell growth and proliferation, is unclear. However, KRIT1, CCM2 and CCM3 all stimulate the activation of VEGFR2 (DiStefano et al. 2014; Goitre et al. 2017), which has been shown to increase ELTD1 expression. Further, it has been demonstrated that DLL4 downregulates ELTD1 (Favara et al. 2019; Favara et al. 2021), and DLL4 is suppressed in absence of KRIT1 (Fig. 9) (Schulz et al. 2015; Kar et al. 2016). However, whether CCM2 and PDCD10 loss alters DLL4 signaling in unknown. Nevertheless, ELTD1 could be an important target of CCM signaling, and contribute to endothelial cell proliferation, altered homeostasis, and reduced stability during CCM lesion formation, through its angiogenic activity. Schematic representation of ELTD1 regulation in brain angiogenesis. A Normal condition (CCM genes functioning properly): 1) KRIT1, CCM2, and CCM3 inhibit the activation and release of Vascular Endothelial Growth Factor (VEGF); 2) KRIT1 alone induces Notch ligand Delta-like 4 (DLL4) function, in order to maintain physiological function on cell membrane (3) and expression levels of ELTD1 (4); 5) the inhibition of ELTD1 reduces inflammation and angiogenesis in the brain. B Loss of CCM genes (disease condition): 1) CCM genes loss induces VEGF activation and release; VEGF activates ELTD1 (2) and VEGF Receptor 2 (VEGFR2) (3) functions on cell membrane; 4) VEGFR2 activation and CCM genes loss both induce the expression level of ELTD1; 5) KRIT1 loss in particular, and possibly all CCM genes loss, also prevents the activation of DLL4; the inhibition of DLL4 induces both the activation (6) and the expression (7) of ELTD1; 8) ELTD1 expression and activation stimulates brain angiogenesis and inflammation in high grade gliomas and vascular diseases other than CCM Conclusions and future perspectives Overall, our study demonstrates for the first time that ELTD1, an established angiogenic marker in high-grade gliomas, is significantly upregulated in human CCM patients and experimental models of CCM disease, and sheds new light on possible functions of ELTD1 as a key regulator of endothelial proliferation and angiogenesis leading to the development, progression, and severity of CCM disease. ELTD1 identification from clinical plasma samples suggests its role as a valuable prognostic and predictive biomarker in patients. Indeed, despite some limitations, such as the small number of clinical samples, our current findings provide a new avenue for future clinical studies in larger patient cohorts, and show for the first time a clinical correlation with the upregulation of a molecular marker with specific symptoms in CCM patients. Lastly, future studies will help to elucidate the mechanistic role of ELTD1 in CCM disease onset and clinical outcomes aimed at improving the disease management and risk stratification in patients, as well as placing the basis for targeted therapeutic strategies in CCM. Supplementary Information Additional Material 1. Additional Material 2. Acknowledgements The authors are grateful to CCM Italia, the Italian Research Network for Cerebral Cavernous Malformation (https://www.ccmitalia.unito.it), and the Associazione Italiana Angiomi Cavernosi (AIAC) Onlus (https://www.aiac.unito.it). We thank Jaden Reinicke for assisting with data analysis and schematic figure preparation. Abbreviations Adhesion G protein-coupled receptor L4 / Epidermal growth factor, Latrophilin, and Seven Transmembrane domain-containing protein 1 Blood-brain barrier Cerebral Cavernous Malformation Cerebral Cavernous Malformation 2 Delta-like 4 G-protein coupled receptor Intracerebral hemorrhage Karnofsky Performance Scale Krev interaction trapped 1 / Cerebral Cavernous Malformation 1 Neurovascular unit Programmed cell death 10 / Cerebral Cavernous Malformation 3 Phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha Temporal lobe epilepsy Transforming Growth Factor beta Vascular Endothelial Growth Factor VEGF Receptor Authors’ contributions Study conception and planning: S.K.; Experimental design: S.K., A.J.G.; Methodology: S.K., S.F.R., A.J.G.; Experiments conducted on human CCM tissue and blood samples: S.K., A.A.; Experiments conducted on cellular and animal models: A.P., A.A., C.M.P., J.E.M., A.P.; Data curation and interpretation: S.K., and A.J.G.; Writing original draft: S.K., A.P.; Providing experimental tools and materials (patient clinical data, mice models, zebrafish models, CCM tissue samples, control TLE tissues): A.O., I.P., C.H., W.B., G.B., J.Z., S.A.S., G.B., H.B., A.S.,W.S.K.; Transcriptome data: R.G., Supervision: S.K., A.J.G. All authors approved the manuscript in its current form. Funding This work was supported by the German Federal Ministry of Education and Research (BMBF), through ERA-NET NEURON, “MECACCM—Preclinical study targeting mechanosensitive Ca2 + channels for Cerebral Cavernous Malformations therapy and early diagnosis.” (grant number 01EW2306 to S.K.); NIH R01 (grant number 1HL141131 to A.J.G.); the American Heart Association (grant number 24POST1191892 https://doi.org/10.58275/AHA.24POST1191892.pc.gr.190829 to A.P.); the FEDER/Ministerio de Ciencia, Innovación y Universidades-Agencia Estatal de Investigación of Spain (grant numbers PID2021—123365OB-I00 and SFB958 to J.Z. and C-M.P.); the Deutsche Forschungsgemeinschaft (DFG) (grant numbers SE2016/7—3, SE2016/10—1, SE2016/13—1 to S.A.-S.), and the Leducq Transatlantic Network of Excellence (grant number 21CVD03—ReVAMP to S.A.-S.). Data availability The RNA-seq data used in this study has been deposited at the GEO repository with the accession number GSE137596 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE137596) (Subhash et al. 2019). Declarations Ethics approval and consent to participate CCM Patients: The study protocol was strictly performed in accordance with the Declaration of Helsinki and with permission by the local ethical committee of the Hannover Medical School, Germany (Approval Number 6960) and the Ethics Committee of University of Bonn Medical Center (Approval Number 076/08 and 042/08). Written informed consent was collected from all the patients involved in this study. Animal Study: All experimental animal procedures were conducted in accordance with the ARRIVE guidelines and the European Directive 2010/63/EU on the protection of animals used for scientific purposes. The study was approved by the relevant ethics committee and authorized by the French Ministry of Research, and University of Rochester. Animals were housed under standard conditions, and all procedures were performed with attention to minimizing distress and ensuring the well-being of the animals throughout the study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Angela J. Glading, Email: angela_glading@urmc.rochester.edu Souvik Kar, Email: kar@ini-hannover.de. References Associated Data Supplementary Materials Additional Material 1. Additional Material 2. Data Availability Statement The RNA-seq data used in this study has been deposited at the GEO repository with the accession number GSE137596 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE137596) (Subhash et al. 2019).