Knockin expression of human ADAMTS5 impairs cardiovascular development and aggravates cerebral cavernous malformations in mice

ABSTRACT

ADAMTS5 is an enzyme that cleaves chondroitin sulfate proteoglycans such as versican (VCAN) and aggrecan (ACAN). The Adamts5^−/− mouse model exhibits aortic anomalies, and increased expression of Adamts4 and Adamts5 leads to excessive versican degradation and reduced cardiac jelly. In zebrafish, knockdown of adamts5 rescues the cardiac phenotype conferred by ccm1 deficiency in zebrafish embryos. Here, we generated an ADAMTS5 knockin mouse model (ADAMTS5^KI) to characterize the effect of induced expression of human ADAMTS5 on the cardiovascular system in mice. Sustained expression of ADAMTS5 in the endothelium diminished cardiac jelly formation and proteoglycan deposition in the atrioventricular cushion, and led to cardiac development arrest. Induced expression of ADAMTS5 in the endothelium of postnatal mice impaired cardiac valve patterning. Expression of ADAMTS5 in brain endothelial cells did not confer an obvious vascular defect. However, expression of ADAMTS5 in brain endothelial cells of Ccm2-deficient mice aggravated cerebral cavernous malformation (CCM) lesion burden and shortened the life span of Ccm2-deficient mice. These findings suggest that tight regulation of ADAMTS5 in the endothelium is essential for cardiovascular development and structural integrity, and that ADAMTS5 interacts with CCM signaling, contributing to CCM disease progression.

Summary: ADAMTS5 expression in the endothelium is essential for cardiovascular development and structural integrity, and ADAMTS5 interacts with cerebral cavernous malformation (CCM) signaling, contributing to CCM disease progression.

INTRODUCTION

The extracellular matrix (ECM) constitutes a dynamic network that not only provides essential structural support for tissues but also orchestrates cellular behavior and participates in tissue repair and regeneration. Proteoglycans are critical ECM components and play an active role in maintaining structural and functional integrity (Zhao et al., 2025; Lockhart et al., 2011). The degradation of proteoglycans is tightly controlled by specific enzymes, the most well known among them being the ‘a disintegrin and metalloproteinase with thrombospondin motifs’ (ADAMTS) family (Stanton et al., 2011). ADAMTS5, a prominent member of the ADAMTS family, selectively cleaves chondroitin sulfate proteoglycans, such as versican (VCAN) and aggrecan (ACAN), exerting critical regulatory roles in cardiovascular development and disease progression (Kelwick et al., 2015; Barallobre-Barreiro et al., 2021).

Recent studies have indicated that ADAMTS5 is a crucial regulator of cardiovascular development and tissue homeostasis. In calcified human aortic valve tissues, the expression level of ADAMTS5 is markedly reduced (Li et al., 2017). In mice, genetic ablation of Adamts5 leads to cardiac valve malformations, structural abnormalities in the aortic wall and heightened vascular calcification (Li et al., 2017; Zhu et al., 2025; Dupuis et al., 2019; Fava et al., 2018). Cardiac jelly, a critical structure of embryonic heart development, is essential for ventricular trabeculation, atrioventricular cushion morphogenesis and valve formation (Lockhart et al., 2011; Männer and Yelbuz, 2019). In the mouse heart, Adamts5 expression was first detected in the myocardium under the outflow tract cushion at embryonic day (E) 11.5, and in the endocardium from E12.5 (Dupuis et al., 2011). This timing is consistent with the gradual disappearance of cardiac jelly from E11.5. Enhanced activities of ADAMTS4 and ADAMTS5 lead to excessive degradation of multifunctional proteoglycans, thereby disrupting normal cardiac jelly formation (Zhou et al., 2015). The specific role of ADAMTS5 expression in cardiovascular development, particularly in valvular morphogenesis, requires further investigation.

The cerebral cavernous malformation (CCM) pathway, mediated by the CCM1–CCM2–CCM3 protein complex, regulates vascular endothelial stability via downstream MEKK3–KLF2/4–ADAMTS signaling (Zhou et al., 2015, 2016; Hong et al., 2020). ADAMTS5, a key downstream effector of the CCM signaling pathway, plays a significant role in maintaining vascular homeostasis via its role in regulating proteoglycan degradation in the ECM environment (Fava et al., 2018; Dupuis et al., 2019; Barallobre-Barreiro et al., 2021). In zebrafish embryos, adamts5 deletion mitigates cardiac abnormalities caused by ccm1 deficiency (Zhou et al., 2015). In neonatal mice, endothelial-specific deletion of Adamts5 reduces CCM lesion formation in white matter while increasing versican accumulation (Hong et al., 2020). Conversely, tetracycline-inducible Adamts5 overexpression synergizes with Ccm1 deficiency to exacerbate early CCM lesion formation (Hong et al., 2020). Reducing versican, an ADAMTS5 substrate, also reduces CCM lesion burden. This suggests that the cleavage product, versikine, could contribute to the severity of CCM lesion burden (Hong et al., 2020). However, the precise mechanistic role of ADAMTS5 and versican in CCM lesion progression across all CCM-deficient models has yet to be elucidated.

Here, we generated a human ADAMTS5 knockin mouse model enabling conditional expression of ADAMTS5 in endothelial cells. The expression of ADAMTS5 in pan-endothelial cells or endocardial cells prevented cardiac jelly accumulation between the endocardium and the myocardium, led to proteoglycan disruption in endocardial cushions, arrested cardiac development, and ultimately led to embryonic lethality. ADAMTS5 overexpression restricted to brain endothelial cells did not cause overt vascular defects but significantly aggravated CCM lesions and reduced survival in Ccm2-deficient mice. These findings demonstrate that precisely regulated ADAMTS5 expression is essential for proper cardiovascular development, and ADAMTS5 functionally interacts with CCM signaling to promote CCM progression. This study provides new insights into the role of ADAMTS5 in cardiovascular biology and disease.

RESULTS

Sustained expression of ADAMTS5 in endothelial cells results in the arrest of heart development associated with defective vascular circulation and reduced cardiac jelly

It has been reported that endocardial-specific loss of Ccm1 increased expression of Adamts4 and Adamts5, leading to premature degradation of cardiac jelly and embryonic lethality by the mid-gestation stage (Zhou et al., 2015). Deletion of Adamts5, but not Adamts4, alleviated CCM lesion burden in the Ccm2-deficient CCM mouse model (Hong et al., 2020). To further investigate the role of ADAMTS5 in the endothelium, we generated ADAMTS5 knockin mice by inserting a floxed cassette of stop sequences followed by sequences encoding human ADAMTS5 and EGFP in the Rosa26 gene locus, hereafter termed ADAMTS5^KI (Fig. S1A). To test the effectiveness of this knockin allele in the endothelial lineage, we crossed ADAMTS5^KI mice with Tie2-Cre mice to produce Tie2-Cre;ADAMTS5^KI/+ mice. Quantitative PCR (qPCR) analysis of the whole embryo confirmed a significant elevation of ADAMTS5 expression in Tie2-Cre;ADAMTS5^KI/+ mice compared with that in littermate controls (Fig. S1B). GFP staining demonstrated that EGFP is expressed in the endothelium, including the developing vessels and endocardium (Fig. S1C).

To assess the effects of continuous endothelial ADAMTS5 expression on cardiovascular development, we observed the morphology and quantified the number of embryos. No live Tie2-Cre;ADAMTS5^KI/+ mice were detected among 48 postnatal offspring from this cross (Table 1). Timed mating revealed that live Tie2-Cre;ADAMTS5^KI/+ embryos displayed modest developmental delay (∼2 somites) but comparable gross morphology to littermate controls at E9.5. By E10.5, all Tie2-Cre;ADAMTS5^KI/+ embryos were dead (Table 1, Fig. 1A). Thus, continuously increased endothelial expression of ADAMTS5 results in embryonic lethality during early gestation.

Table 1.

Offspring of Tie2-Cre and ADAMTS5^KI/+ mating at various developmental stages

Genotype	Age
	E9.5	E10.5	E12.5	E15.5	P1-P21
Wild type	28	22	7	9	13
Tie2-Cre	26	13	4	6	17
ADAMTS5KI/+	17	18	12	6	18
Tie2-Cre;ADAMTS5KI/+	16 (4 dead)	17 (17 dead)a	6 (6 dead)a	0a	0a
Total	87	70	29	21	48

Viability of Tie2-Cre;ADAMTS5^KI/+ mice at different developmental stages is given in parentheses. Bold values indicate the numbers of mutant embryos. ^aP<0.001, χ² test.

Fig. 1.

Increased expression of ADAMTS5 in the endothelium results in vascular and heart defects during embryonic development. (A) Representative stereomicroscopic images of Tie2-Cre;ADAMTS5^KI/+ embryos and littermate controls at embryonic day (E) 9.5 and E10.5. Scale bars: 1 mm. (B) Hematoxylin and Eosin (H&E) staining of transverse sections of E9.5 ADAMTS5^KI/+ and Tie2-Cre;ADAMTS5^KI/+ embryos shows the presence of normally lumenized dorsal aortas in the ADAMTS5^KI/+ embryos but not in Tie2-Cre;ADAMTS5^KI/+ embryos. Green arrows indicate dorsal aortas. Scale bars: 50 μm. (C) Co-immunostaining of CD31 with smooth muscle actin (SMA, top) and collagen IV (Col IV, bottom) on transverse sections of E9.5 ADAMTS5^KI/+ and Tie2-Cre;ADAMTS5^KI/+ embryos. Panels on the right show higher-magnification views. White arrows indicate the dorsal aortas. Scale bars: 100 μm. (D) H&E and Alcian Blue staining display reduced cardiac jelly in Tie2-Cre;ADAMTS5^KI/+ embryos. Panels on the right show higher-magnification views. Blue arrows indicate the reduced cardiac jelly between endocardium and the myocardium. Scale bars: 50 μm. (E) Quantification of the cardiac jelly area of E9.5 hearts. n=6, six sections analyzed for each group. Data are presented as mean±s.e.m., and statistical significance was determined using an unpaired two-tailed Student's t-test. ****P<0.0001. (F) Co-immunostaining for endomucin (EMCN) and versican on transverse sections of E9.5 ADAMTS5^KI/+ and Tie2-Cre;ADAMTS5^KI/+ embryos. Panels on the right show higher-magnification views. Scale bars: 10 μm. In C,F, nuclear staining is shown in blue. Images are representative of n=5-6 for each group.

Histological analysis of E9.5 Tie2-Cre;ADAMTS5^KI/+ and littermate control embryos revealed that Tie2-Cre;ADAMTS5^KI/+ embryos failed to form a patent dorsal aorta (Fig. 1B). Immunostaining for CD31 (also known as PECAM1), smooth muscle actin (SMA, also known as ACTA2) and collagen IV demonstrated that endothelial cells, vascular smooth muscle cells and basement membrane components were appropriately differentiated and spatially localized; however, these analyses confirmed a markedly constricted dorsal aorta lumen in Tie2-Cre; ADAMTS5^KI/+embryos (Fig. 1C). Impaired lumenization of dorsal aorta likely prevented the establishment of a functional circulation system. In contrast, the vessel patterning of the yolk sac appeared grossly normal (Fig. S1D).

Hematoxylin and Eosin (H&E) staining of the developing heart suggested a reduction in cardiac jelly in the space between the endocardium and the myocardium in hearts of Tie2-Cre; ADAMTS5^KI/+embryos (Fig. 1D). Alcian Blue staining revealed the loss of cardiac jelly wrapping around the trabeculae in the hearts of Tie2-Cre;ADAMTS5^KI/+embryos (Fig. 1D). Quantitative analyses demonstrated a significant decrease in cardiac jelly area in Tie2-Cre;ADAMTS5^KI/+ embryos compared with that in littermate controls (Fig. 1E). Immunostaining for endomucin (EMCN), an endothelial marker, and versican, a major proteoglycan component of cardiac jelly, showed that versican was nearly undetectable in Tie2-Cre;ADAMTS5^KI/+ hearts (Fig. 1F). Thus, endothelial expression of ADAMTS5 results in early-gestational heart failure associated with impaired blood circulation and loss of cardiac jelly integrity.

Sustained expression of ADAMTS5 in endocardial cells results in cardiac development arrest associated with reduced cardiac jelly

ADAMTS5 cleaves the proteoglycan versican, which is a crucial component of cardiac jelly (Kelwick et al., 2015). To determine whether the increased expression of ADAMTS5 in endocardial cells would be sufficient to induce cardiac jelly degradation and affect heart development, we crossed ADAMTS5^KI mice with Nfatc1-Cre mice to express ADAMTS5 in the endocardial lineage. qPCR analysis of E11.5 hearts showed a significant increase in ADAMTS5 expression (Fig. S2A). Immunostaining for GFP showed the specific knockin expression of ADAMTS5 in the endocardium (Fig. S2B). Timed mating demonstrated that there were no differences in gross morphology between live Nfact1-Cre;ADAMTS5^KI/+ embryos and littermate control embryos before E11.5, even though Nfatc1-Cre;ADAMTS5^KI/+ embryos started dying at E10.5, and all embryos were dead by E12.5 (Table 2). Thus, endocardial expression of ADAMTS5 results in embryonic lethality during mid-gestation.

Table 2.

Offspring of Nfatc10Cre and ADAMTS5^KI/+ mating at various developmental stages

Genotype	Age
	E9.5	E10.5	E11.5	E12.5	E15.5	P1-P21
Wild type	13	26	16	5	20	15
Nfatc1-Cre	10	21	11	7	10	15
ADAMTS5KI/+	10	16	6	7	15	10
Nfatc1-Cre;ADAMTS5KI/+	10	8 (2 dead)	7 (3 dead)	5 (5 dead)a	0a	0a
Total	43	71	40	24	45	40

Viability of Nfatc1-Cre;ADAMTS5^KI/+ mice at different developmental stages stages is given in parentheses. Bold values indicate the numbers of mutant embryos. ^aP<0.001, χ² test.

H&E staining revealed diminished cardiac jelly, fewer myocardial trabeculae and thinner myocardium in Nfatc1-Cre;ADAMTS5^KI/+ hearts at E9.5, in comparison to well-formed trabeculae and compacted myocardium in littermate controls (Fig. 2A). These phenotypes become more pronounced in Nfatc1-Cre;ADAMTS5^KI/+ hearts at E10.5 and E11.5 (Fig. 2B,C). Alcian Blue staining showed a lack of cardiac jelly in the hearts of Nfatc1-Cre;ADAMTS5^KI/+ embryos at each time point. The space between the myocardium and endocardium was filled with proteoglycan-cardiac jelly in control hearts, but this was missing in Nfatc1-Cre;ADAMTS5^KI/+ hearts. We also observed ventricular chamber dilation in the histology results for Nfatc1-Cre;ADAMTS5^KI/+ embryos at E11.5, but the phenotype was not marked before E11.5 (Fig. 2A-C).

Fig. 2.

Endocardial expression of ADAMTS5 impairs heart development. (A-C) Representative stereomicroscopic images, H&E and Alcian Blue staining of embryos and cross-sections of embryos at E9.5 (A), E10.5 (B) and E11.5 (C) showing defective heart development of Nfact1-Cre;ADAMTS5^KI/+ mice as gestation progresses. H&E staining shows a thin trabeculae myocardium (TM, orange bars) and compacted myocardium layer (CM, green bars) of Nfact1-Cre;ADAMTS5^KI/+ hearts. Alcian Blue staining displays reduced cardiac jelly at the TM (indicated by black arrows) and atrioventricular cushion (indicated by asterisks) in Nfact1-Cre;ADAMTS5^KI/+ mice. Panels on the right show higher-magnification views. Scale bars: 100 μm. (D) Quantitation plots of jelly area, and TM and CM thickness from H&E- and Alcian Blue-stained sections (E9.5, n=6; E10.5, n=6; E11.5, n=6). Data are presented as mean±s.e.m., and statistical significance was determined using an unpaired two-tailed Student's t-test. ns, not significant; **P<0.01; ****P<0.0001. (E,F) Co-immunostaining of endomucin (EMCN) and MF20 (E) shows a reduced endocardial-myocardial space in E10.5 and E11.5 Nfact1-Cre;ADAMTS5^KI/+ hearts. White arrows in E indicate the endocardial-myocardial space. Co-immunostaining of EMCN and versican (F) shows the reduction of versican in the endocardial-myocardial space in E10.5 and E11.5 Nfact1-Cre;ADAMTS5^KI/+ hearts. Nuclear staining is shown in blue. Panels on the right show higher-magnification views. Scale bars: 10 μm. Images are representative of n=5-6 for each group.

Previous studies indicate that Adamts5 is expressed during cardiac valve maturation in regions where versican cleavage fragments are detected (Cikach et al., 2018). Adamts5-null embryos had enlarged and malformed pulmonary valves, and adult Adamts5^−/− mice have myxomatous pulmonary, aortic and mitral valves, which present significant thickening of pulmonary valve cusps, and aortic and mitral valves (Dupuis et al., 2011). Consistent with these observations, Alcian Blue staining also confirmed the loss of jelly glycosaminoglycans surrounding the atrioventricular cushion in the Nfatc1-Cre;ADAMTS5^KI/+ hearts (Fig. 2A-C; Fig. S2C). Quantitation of the cardiac jelly area around the trabeculae of E9.5 hearts revealed >70% reduction, and analyses of E10.5 and E11.5 hearts showed even greater reductions (>90% and >95%, respectively) in Nfatc1-Cre;ADAMTS5^KI/+ mice compared to littermate controls (Fig. 2D). Reduction of trabeculae and myocardium thickness demonstrated impaired heart development in Nfatc1-Cre;ADAMTS5^KI/+ mice, despite the myocardium thickness still increasing during embryonic development (Fig. 2D). Co-immunostaining for endomucin and myocardium marker MF20 antibody also showed the reduction of the endocardial-myocardial space, and endomucin and versican co-immunostaining showed the lacking of versican in Nfatc1-Cre;ADAMTS5^KI/+ hearts (Fig. 2E,F). Thus, sustained endocardial expression of ADAMTS5 results in mid-gestation heart failure associated with reduced cardiac jelly and impaired myocardium growth.

Induced expression of ADAMTS5 in the endothelium of postnatal mice results in abnormal cardiac valve maturation

Due to the lethality of embryos with sustained ADAMTS5 expression in the endothelium, we crossed ADAMTS5^KI mice with Cdh5-CreER^T2 mice to generate Cdh5-CreER^T2;ADAMTS5^KI/+ mice, enabling postnatal induction of ADAMTS5 expression. qPCR and western blot analysis confirmed a significant increase of ADAMTS5 and GFP expression in 4-hydroxytamoxifen-treated Cdh5-CreER^T2;ADAMTS5^KI/+ mice compared with that in littermate controls (Fig. S3A,B). GFP immunostaining of heart and brain sections demonstrated efficient endothelial-specific recombination of the ADAMTS5 knockin allele following induction (Fig. S3C,D). Consistent with the reduced proteoglycan phenotype observed in the atrioventricular cushion of Nfact1-Cre;ADAMTS5^KI/+ embryos (Fig. 2A-C), postnatal endothelial induction of ADAMTS5 expression resulted in marked thinning of the pulmonary, aortic, mitral and tricuspid valves (Fig. 3A). Alcian Blue staining revealed decreased proteoglycan content in the aortic and mitral valves (Fig. 3B), suggesting that excessive ADAMTS5 activity leads to proteoglycan depletion and consequent structural attenuation of cardiac valves.

Fig. 3.

Induced expression of ADAMTS5 in endothelial cells impairs cardiac valve development. (A) Pups were intragastrically injected with 4-hydroxytamoxifen at postnatal day (P) 1, and the heart tissues were harvested at P8. H&E staining of heart tissues reveals significant thinning of pulmonary valves (blue arrows), tricuspid valves (orange arrows), mitral valves (red arrows) and aortic valves (green arrows) in Cdh5-CreER^T2;ADAMTS5^KI/+ mice compared with those in ADAMTS5^KI/+ mice. Scale bars: 200 μm. Middle panels show higher-magnification views. Scale bars: 100 μm. (B) Alcian Blue staining indicates the reduced extracellular matrix (proteoglycan) in aortic and mitral valves of Cdh5-CreER^T2;ADAMTS5^KI/+ hearts. Scale bars: 200 μm. (C) Representative micrographs, H&E staining and immunostaining of brain sections at P8. Stereomicroscopic images and H&E staining display no bleed or cerebral cavernous malformation (CCM) lesion in Cdh5-CreER^T2;ADAMTS5^KI/+ mice. Isolectin B4 (IB4) staining shows no significant change in vascular density of induced Cdh5-CreER^T2;ADAMTS5^KI/+ mice. Immunostaining of versican is reduced in induced Cdh5-CreER^T2;ADAMTS5^KI/+ mice brains compared with that in ADAMTS5^KI/+ control mice brains. Dotted lines mark white matter. Panels on the right show higher-magnification views. Scale bars: 2 mm (micrographs); 200 μm (H&E and immunostained images). (D) Postnatal survival curve of induced Cdh5-CreER^T2;ADAMTS5^KI/+ (n=13) and ADAMTS5^KI/+(n=10) mice. Statistical analysis was performed using the Mantel–Cox test. ****P<0.0001. (E) H&E staining of lung tissues showed reduced alveolar number in induced Cdh5-CreER^T2;ADAMTS5^KI/+ mice compared with ADAMTS5^KI/+ mice. Panels on the right show higher-magnification views. Representative images from at least three or more independent experiments are shown. Scale bars: 100 μm.

Although Adamts5 has been implicated as a downstream effector of CCM signaling, whether it is causal for CCM has not been clearly established. Gross images, H&E staining and isolectin B4 staining of brains revealed no evidence of hemorrhage or CCM lesions in either the cerebrum or cerebellum of induced Cdh5-CreER^T2;ADAMTS5^KI/+ mice (Fig. 3C). Immunostaining demonstrated reduced versican levels in 4-hydroxytamoxifen-treated Cdh5-CreER^T2; ADAMTS5^KI/+ mice compared to ADAMTS5^KI/+ controls (Fig. 3C).

Survival curves showed that the induction of ADAMTS5 from postnatal day (P) 1 caused premature death of Cdh5-CreER^T2;ADAMTS5^KI/+ mice within 2 weeks after birth (Fig. 3D). Major organs were harvested at P8 to assess the potential cause of lethality. Histological analysis demonstrated marked alveolar dilation in the lungs of induced Cdh5-CreER^T2;ADAMTS5^KI/+ mice (Fig. 3E). Immunostaining further revealed a reduction of type II alveolar epithelial cells and proliferative cells compared with those in littermate controls (Fig. S4A), suggesting impaired pulmonary development or maintenance. The gross images and H&E staining of Cdh5-CreER^T2;ADAMTS5^KI/+ mice showed no histological defect in the kidney, but there were bleeding lesions and increased inflammatory cell infiltration in the spleen and liver, respectively (Fig. S4B-D).

Induced expression of ADAMTS5 in brain endothelial cells does not cause vascular abnormality

Adamts5 is a downstream target of CCM signaling (Zhou et al., 2016; Hong et al., 2020), but it remains unclear whether increased expression of ADAMTS5 is sufficient to cause brain vascular malformation. We used the Tie2-Dre and Mfsd2a-CrexER (Dre-Cre) duo recombination system to drive ADAMTS5 expression specifically in endothelial cells of the central nervous system without the requirement of tamoxifen induction (Yang et al., 2022). We generated Tie2-Dre;Mfsd2a-CrexER;ADAMTS5^KI/+ (thereafter ADAMTS5^BECKI, where BEC refers to brain endothelial cells) mice (Fig. 4A). The dual-recombinase system effectively drove the recombination of the ADAMTS5 knockin allele in brain endothelial cells as indicated by GFP stainings (Fig. S5). ADAMTS5^BECKI mice appeared grossly normal and with normal body weight gain up to 5 months of age (Fig. 4B). Micrographs of P8 and P15 ADAMTS5^BECKI brains and micro-computed tomography (CT) imaging of 3-month-old ADAMTS5^BECKI brains did not detect any bleeding or malformed vascular lesion in the cerebrum and cerebellum, consistent with the isolectin B4, CD31 and H&E staining results (Fig. 4C-E). Versican immunostaining was reduced in ADAMTS5^BECKI mice compared to that in ADAMTS5^KI/+ mice (Fig. 4E). These results indicate that increased expression of ADAMTS5 in brain endothelial cells does not impair normal vessel development.

Fig. 4.

Knockin expression of ADAMTS5 in brain endothelial cells does not cause vascular abnormality. (A) Schematic of the genetic cross to generate ADAMTS5^BECKI mice with sustained ADAMTS5 expression in brain endothelial cells. (B) Body weight gain of ADAMTS5^BECKI mice and littermate controls (4-6 mice/group). Data are presented as mean±s.e.m. and statistical significance was determined using an unpaired two-tailed Student's t-test. ns, not significant. (C) Representative micrographs revealed normal vessel development in ADAMTS5^BECKI and ADAMTS5^KI/+ control mice at P8. Isolectin B4 (IB4) staining of vibratome sections indicates normal vessel density of ADAMTS5^BECKI mice compared with that of littermate controls. (D) Representative micrographs and CD31 immunostaining showing no CCM lesion and normal vessel density in an ADAMTS5^BECKI brain at P15. (E) Representative brain micrographs, micro-CT images and H&E staining of brain sections of 3-month-old ADAMTS5^BECKI and ADAMTS5^KI/+ mice showed no malformed vascular lesions. Versican immunostaining was reduced in ADAMTS5^BECKI mice compared with that in littermate controls. Dotted lines mark white matter. Panels on the right show higher-magnification views. Representative images from at least three or more independent experiments are shown. Micrographs and CT Scale bars: 2 mm (micrographs and micro-CT images); 200 μm (H&E and immunostained images).

Expression of ADAMTS5 synergizes with loss of Ccm2 to induce CCM

We have previously demonstrated that the BEC Dre-Cre system can be used to generate the chronic CCM mouse model by driving efficient Ccm gene deletion, with Ccm2 deletion conferring modest lesion burden among the three Ccm genes (Yang et al., 2022). MEKK3–KLF2/4 signaling has been reported to function downstream of CCM signaling for CCM lesion formation (Zhou et al., 2016). Gene expression analysis demonstrated that the expression of Klf2 and Klf4 was robustly increased, but the expression of Adamts5 was modestly increased in cerebellum endothelial cells of P6 Ccm2^BECKO (Tie2-Dre;Mfsd2a-CrexER;Ccm2^fl/fl) mice (Fig. 5A).

Fig. 5.

Expression of ADAMTS5 in brain endothelial cells synergizes with Ccm2 deficiency to induce CCM lesion formation. (A) Relative mRNA expression level of CCM downstream genes, Klf2, Klf4, Adamts1, Adamts4, Adamts5 and Adamts9, in endothelial cells isolated from control and Ccm2^BECKO mice at P6. n≥3 for each group was used. Data are presented as mean±s.e.m. and statistical significance was determined using an unpaired two-tailed Student's t-test. ns, not significant; **P<0.01; ***P<0.001. (B) Representative micrographs and micro-CT images of brains from Ccm2^BECKO and Ccm2^BECKO;ADAMTS5^KI/+ mice at P18 and 2 months. Scale bars: 2 mm. (C) Quantification of CCM lesion volumes, total lesion counts (left of the gray dashed line) and the counts of lesions of different size groups (right of the gray dashed line) in the cerebrum and cerebellum of Ccm2^BECKO (n=6) and Ccm2^BECKO;ADAMTS5^KI/+ (n=6) mice at P18. Small lesions, volume≤10⁶ μm³; medium lesions, volume=10⁶-10⁷ μm³; large lesions, volume≥10⁷ μm³. Data are presented as mean±s.e.m. and statistical significance was determined using an unpaired two-tailed Student's t-test. ns, not significant; **P<0.01; ***P<0.001; ****P<0.0001. (D) H&E staining of brain sections of Ccm2^BECKO and Ccm2^BECKO;ADAMTS5^KI/+ mice at P18. Panels on the right show higher-magnification views. Scale bars: 200 μm. (E) Postnatal survive curve of ADAMTS5^KI/+ (n=9), Ccm2^BECKO (n=10) and Ccm2^BECKO;ADAMTS5^KI/+ mice (n=16). Statistical analysis was performed using the Mantel–Cox test. ****P<0.0001. (F) Co-immunostaiinng of p-S6 and CD31 showing increased p-S6 in CCM lesions of Ccm2^BECKO and Ccm2^BECKO;ADAMTS5^KI/+ mice compared to that in Ccm2^fl/fl control at P18. White arrows indicate increased p-S6 in CCM lesions in Ccm2^BECKO and Ccm2^BECKO;ADAMTS5^KI/+ mice. Scale bars: 100 μm. Images are representative of n=5-6 for each group.

To determine whether further increased expression of ADAMTS5 contributes to CCM formation in a Ccm2-deficient background, we generated Ccm2^BECKO;ADAMTS5^KI/+ mice. Histological and micro-CT analysis revealed that the Ccm2^BECKO;ADAMTS5^KI/+ mice had markedly increased hemorrhages and CCM lesion burden in comparison to littermate Ccm2^BECKO mice observed visually at P18 and 2 months (Fig. 5B). Micro-CT analysis of CCM lesion burden at P18 showed the significant increase in lesion volume in the cerebrum (10.7-fold) and cerebellum (38.2-fold) (Fig. 5C). The total counts of CCM lesions in Ccm2^BECKO;ADAMTS5^KI/+ mice increased by 5.0- and 5.8-fold in the cerebrum and cerebellum, respectively, in comparison to those of littermate Ccm2^BECKO mice. We divided the CCM lesions into three groups according to the volume (small, ≤10⁶ μm³; medium, 10⁶-10⁷ μm³; large, ≥10⁷ μm³). Quantification of the lesion groups showed that counts of each group were significantly increased (for the cerebrum, small lesions increased from 9.33±9.52 to 52.67±17.81, medium lesions from 3.50±2.25 to 9.50±2.73; and large lesions from 0.0 to 1.50±1.97; for the cerebellum, small lesions increased from 23.67±9.09 to 109.00±16.99, medium lesions from 4.67±1.63 to 47.33±13.71, and large lesions from 0.50±0.83 to 13.30±3.39; values represent mean±s.e.m.). Examination of H&E-stained sections confirmed a significant increase in the count and size of vascular lesions in Ccm2^BECKO;ADAMTS5^KI/+ mice compared with those in littermate Ccm2^BECKO mice (Fig. 5D). Finally, brain endothelial gain of ADAMTS5 combined with Ccm2 deficiency resulted in the death of 50% of animals by P21, with complete lethality by 3 months of age; in comparison, Ccm2^BECKO mice were all viable up to 7 months of age (Fig. 5E). These results indicate the strong synergy between Ccm2 loss of function and ADAMTS5 gain of function to amplify CCM formation.

Gain-of-function variants in phosphatidylinositol 3-kinase (PI3K) have been shown to increase CCM burden in both human patients and mouse models (Weng et al., 2021; Ren et al., 2021; Peyre et al., 2021). CCM and PI3K signaling converge on mTOR activation and phosphorylation of ribosomal protein S6 (p-S6, encoded by Rps6), a downstream effector of mTOR signaling (Ren et al., 2021). To determine whether ADAMTS5 expression influences PI3K downstream signaling, we assessed p-S6 levels in cerebellar lysates and isolated endothelial cells. No increase in p-S6 was detected in Cdh5-CreER^T2;ADAMTS5^KI/+ samples compared with controls (Fig. S6A). Consistent with previous reports (Ren et al., 2021), immunostaining was able to detect increased level of p-S6 in vessels of Ccm2^BECKO brains, but no further enhancement of p-S6 level was observed in Ccm2^BECKO;ADAMTS5^KI/+ brains (Fig. 5F). These findings suggest that ADAMTS5-mediated exacerbation of CCM lesion burden occurs independently of PI3K–mTOR signaling.

Interactions between brain endothelial cells and surrounding astrocytes and inflammatory cells have been shown to influence CCM lesion severity (Lopez-Ramirez et al., 2021; Lai et al., 2022; Onyeogaziri et al., 2024). To evaluate whether ADAMTS5 expression alters the neurovascular microenvironment, we examined ECM composition and mural cell coverage. In the induced Cdh5-CreER^T2;ADAMTS5^KI/+ and control brains, co-immnunostaining of collagen IV and CD31 revealed no differences in vascular integrity, with both groups showing effective colocalization (Fig. S6B). Similarly, no significant difference were observed between ADAMTS5^BECKI and littermate control mice (Fig. S6C) or between Ccm2^BECKO;ADAMTS5^KI/+ and littermate control mice (Fig. S6D). In contrast, GFAP immunostaining revealed a disorganized astrocytic pattern in regions containing CCM lesions in Ccm2^BECKO;ADAMTS5^KI/+ brains. The GFAP signal was largely absent adjacent to endothelial cells within large lesions and showed reduced overlap with CD31 in smaller lesions. No aberrant GFAP patterning was observed in non-lesion regions of Ccm2^BECKO;ADAMTS5^KI/+ compared with Ccm2^BECKO brains (Fig. 6). These data suggest that ADAMTS5 expression exacerbates CCM lesions by altering endothelial interactions with other cell types in the neurovascular unit or by modifying ECM composition within the lesion environment.

Fig. 6.

Abnormal astrocyte arrangement at the CCM lesion region in Ccm2^BECKO and Ccm2^BECKO;ADAMTS5^KI/+ mice. (A) Co-immunostaining of GFAP and CD31 showing disorganized astrocytes in the brain region with small CCM lesions (panels e,f,j, yellow solid arrows) in Ccm2^BECKO and Ccm2^BECKO;ADAMTS5^KI/+ mice, and reduced GFAP and CD31 signal overlap in vessels of small lesions and absence of astrocyte overlap with the endothelium of large CCM lesions (panels g,h; yellow dashed arrows) in Ccm2^BECKO;ADAMTS5^KI/+ mice. White asterisks indicate disordered astrocytes in the cerebellum. Normal parallel GFAP (astrocytes) arrangement was observed in both Ccm2^fl/fl (panels a,b) and non-CCM lesion brain regions (panels c,d) in Ccm2^BECKO;ADAMTS5^KI/+ mice. Nuclear staining is shown in blue. Scale bars: 200 μm. (B) H&E staining of an adjacent section indicates the orientation of the immunostained region, lesion size and location in the cerebellum. The black arrow indicates small CCM lesions, and black asterisks indicate large CCM lesions, the locations of which were consistent with GFAP/CD31 staining (panels e-h). Horizontal lines denote the regions corresponding to the areas where panels e,f, and g,h are located. Scale bar: 200 μm. Images are representative of n=3-4 for each group.

DISCUSSION

In this study, we systematically elucidated the regulatory roles of ADAMTS5 in cardiovascular development and the pathogenesis of CCM through a human ADAMTS5 knockin mouse model. The results demonstrated that sustained endothelial ADAMTS5 expression leads to premature degradation of cardiac jelly and aberrant breakdown of proteoglycans in endocardial cushions, ultimately resulting in embryonic lethality. Postnatal induction of ADAMTS5 overexpression in endothelial cells resulted in valvular developmental defects. Although brain endothelial-specific sustained ADAMTS5 expression did not induce obvious vascular malformations, it markedly exacerbated CCM lesions and shortened survival in Ccm2-deficient mice. These findings establish that precise regulation of ADAMTS5 expression is essential for cardiovascular development and homeostasis, and highlight its synergistic pathogenic effect with the CCM pathway.

As a critical structure in embryonic heart development, cardiac jelly sustains homeostasis via balanced proteoglycan synthesis and degradation (Hinton et al., 2006; Lockhart et al., 2011). Our results showed that ADAMTS5 overexpression disrupts the structural integrity of cardiac jelly through excessive degradation, thereby impairing ventricular trabeculation and atrioventricular cushion remodeling. This supports previous findings that elevated ADAMTS4 and ADAMTS5 activity may impair normal cardiac jelly formation (Zhou et al., 2015). Interestingly, although ADAMTS5 deficiency leads to aortic valve calcification and vascular wall structural abnormalities, our data revealed that its overexpression also induces severe valvular malformation. Together, these findings emphasize that the dynamic balance in ADAMTS5 enzymatic activity is crucial for normal cardiovascular development (Santamaria and de Groot, 2020; Barallobre-Barreiro et al., 2021). Physiological levels of ADAMTS5 support tissue remodeling, but abnormal expression results in an ECM imbalance. This bidirectional regulation provides new insights into the pathogenesis of congenital cardiac valve diseases. Comparison with other ADAMTS family members further underscores this point: ADAMTS9 deficiency leads to aortic valve abnormalities and versican accumulation (Kern et al., 2010), whereas ADAMTS16 deficiency results in bicuspid aortic valve formation (Lin et al., 2024). The studies show that ADAMTS5 regulates a broader spectrum of processes, including cardiac jelly homeostasis, valvulogenesis and vascular integrity (Zhou et al., 2015; Li et al., 2017; Zhu et al., 2025).

The CCM pathway maintains vascular stability through the CCM1–CCM2–CCM3 complex, which regulates the MEKK3–KLF2/4–ADAMTS axis (Zhou et al., 2016). We found that ADAMTS5 overexpression strongly synergizes with Ccm2 deficiency to enhance the aggressiveness of CCM lesions, potentially through the influence of versican degradation products on vascular endothelial stability. This aligns with previous work showing that Adamts5 deficiency suppresses CCM lesion formation in Ccm1-deficient models, confirming that Adamts5 acts as a downstream effector of the CCM pathway (Hong et al., 2020). Importantly, induction of ADAMTS5 expression alone in brain endothelial cells was insufficient to trigger CCM, indicating that its pathogenic effects depend on dysfunction of the CCM complex, providing a theoretical basis for selective therapeutic targeting. Despite the well-established role of gain-of-function variants in PI3K in elevating CCM burden, our findings indicate that ADAMTS5-mediated exacerbation of CCM lesion burden is independent of PI3K–mTOR signaling at the level of p-S6. This suggests that ADAMTS5 acts through alternative pathways to influence CCM development. Despite the overexpression of ADAMTS5, our results show no distribution and density changes between endothelial cells and their basement membrane. On the contrary, the disordered astrocytic pattern demonstrated by GFAP immunostaining indicates that ADAMTS5 exerts its effect by interfering with the communication between endothelial cells and astrocytes. These two cell types are vital constituents of the neurovascular unit.

As a major aggrecan-degrading enzyme, ADAMTS5 is well known for driving cartilage destruction in osteoarthritis (Wang et al., 2024; Glasson et al., 2005; Stanton et al., 2005). The regulatory mechanism of ADAMTS5 on matrix proteoglycans identified in this study offers clues for understanding the pathogenesis of osteoarthritis and cardiovascular diseases. The ability of ADAMTS5 to both degrade versican in the cardiovascular system and cleave aggrecan in joints depends on its protease activity, suggesting that ADAMTS5-targeted inhibitors could simultaneously modulate matrix homeostasis in multiple tissues (Clement-Lacroix et al., 2022). The development of ADAMTS5 inhibitors for osteoarthritis treatment needs careful consideration of potential cardiovascular effects, including the risk of vascular calcification or valvular abnormalities (Larkin et al., 2015). The humanized ADAMTS5 mouse model generated in this study provides a platform to evaluate both the efficiency and systemic safety of ADAMTS5 modulators. ADAMTS5-mediated proteoglycan degradation products may serve as shared biomarkers for monitoring the progression of cardiovascular diseases and osteoarthritis.

Previous studies have largely focused on Adamts5 deficiency phenotypes in the cardiovascular system. Our study demonstrated the pathological consequences of ADAMTS5 overexpression in a humanized model, underscoring the necessity for strict regulation of its activity. In the context of CCM, this study revealed the synergistic role of ADAMTS5 expression and Ccm2 deficiency in CCM pathogenesis, enhancing our understanding of the molecular mechanism of CCM. Nonetheless, the contribution of ADAMTS5 to other proteoglycan targets and its role in adult cardiovascular homeostasis and osteoarthritis progression remain to be clarified. Potential species-specific interactions between human ADAMTS5 and the mouse proteoglycan environment should also be considered.

Taken together, our findings identify ADAMTS5 as a potential therapeutic target for CCM progression. Strategies targeting ADAMTS5 in CCM or osteoarthritis must be carefully designed to achieve cell-type specificity. The humanized ADAMTS5 knockin model described here provides a valuable tool for screening modulators with both osteoarthritis efficacy and cardiovascular safety.

MATERIALS AND METHODS

Mice

Tie2-Cre, Nfatc1-Cre, Cdh5-CreER^T2, Tie2-Dre, Mfsd2a-CrexER and Ccm2^fl/fl animals have been described previously (Zhou et al., 2015; Yang et al., 2022, 2023). The Ccm2^BECKO (Tie2-Dre;Mfsd2a-CrexER;Ccm2^fl/fl) mice were described in Yang et al. (2022). ADAMTS5^KI mice were generated via embryonic stem cell-based gene targeting by Shanghai Biomodel Organism Science and Technology Development. Experimental animals were maintained on a 129/C57BL/6J mixed genetic background. Both male and female mice were used in the experiments. Littermates were used as controls for all experiments.

The primers used to genotype ADAMTS5^KI mice were as follows: ADAMTS5KI-Forward, 5ʹ-AAGGGGGAGGATTGGGAAGACA-3ʹ, and ADAMTS5KI-Reverse, 5ʹ-CAGAAGGAGCGGGAGAAATGGATA-3ʹ.

Study approval

All animal ethics and protocols were approved by the Institutional Animal Care and Use Committee of Tianjin Medical University.

Mouse embryo processing

All E9.5-E12.5 embryos were dissected and then imaged under a stereomicroscope (Nikon, SMZ18). The embryos were fixed in 4% polyformaldehyde (PFA) at 4°C overnight and subjected to the next protocols.

Induction of ADAMTS5 expression in vivo

4-hydroxytamoxifen (Sigma-Aldrich, H7904) was dissolved in 1% ethanol-corn oil (0.5 mg/ml, C8267, Sigma-Aldrich) and used to induce Cdh5-CreER^T2;ADAMTS5^KI/+ mice for sustained expression of the ADAMTS5 gene. P1 pups were intragastrically injected with a single dose of 75 μl (37.5 μg) of 4-hydroxytamoxifen once.

Histological analysis

Fresh mouse embryos and all tissues were fixed in 4% PFA overnight and paraffin-embedded after gradual dehydration in 100% ethanol. Paraffin sections (7 μm) were obtained using a microtome (RM2245, Leica) were stained with H&E and Alcian Blue using standard protocols. Immunostaining and immunohistochemistry protocols have been previously described (Yang et al., 2022). Briefly, deparaffinized sections underwent successive processes consisting of rehydration with xylene, antigen retrieval, blocking in normal donkey serum with BSA and PBS containing 0.1% Tween-20 (Solarbio), and incubation with primary antibodies overnight at 4°C. The sections were then incubated with the secondary antibody for 2 h at room temperature after washing, and then imaged using the Nikon microscope (Ni-U). Trabeculae myocardium and compacted myocardium thickness and cardiac jelly area were quantified by ImageJ. Quantitative data of trabeculae myocardium and compacted myocardium thickness, and cardiac jelly area are from three sections for each point in E9.5, E10.5 and E11.5 embryos. Representative images from at least three or more independent experiments are shown.

Vibratome sections and staining

Fresh brain tissue was fixed in 4% PFA for 2-4 h and embedded in low-melting agarose. The coronal vibratome section was sliced in units of 100 μm using a Leica VT1200s vibratome. After the penetration of PBS solution containing 0.3% Triton X-100 and BSA, the sections were washed with PBS and stained with the antibody. Images were acquired using an Axio-Imager LSM-800 confocal microscope (Carl Zeiss).

Isolation of cerebellum endothelial cells

The cerebellum was removed from mice after they were anesthetized with 0.2% avertin (Sigma-Aldrich) and perfused with cold PBS. The cerebellum was digested with digestion mix [Dulbecco's modified Eagle medium (DMEM; Corning) containing 10% fetal bovine serum (FBS, Corning), 1 mg/ml collagenase/dispase (Roche, 10269638001) and benzonase (Millipore, E1014, 1:2000 dilution)] at 37°C for 15 min until complete digestion. The single-cell suspension was incubated with anti-CD31-conjugated magnetic beads (Miltenyi Biotec, 130-097-418) at 4°C for 15 min after they were passed through a 70 μm strainer. We collected microbead-bound cells using MACS MS columns (MiltenyiBiotec) for subsequent qPCR analysis.

Western blot analysis

Whole tissue was added to the lysis buffer (150 mM NaCl, 1% NP-40 and 50 mM Tris-HCl, pH 8.0) with protease and phosphatase inhibitors. Protein supernatants were harvested after 12,000 g centrifugation for 20 min at 4°C. The BCA kit (Thermo Fisher Scientific, 23252) was used to quantify protein concentration. 20 μg protein samples were separated by SDS-PAGE within the running buffer and transferred to nitrocellulose membranes within membrane transfer solution. Membranes were blocked with 5% skim milk at room temperature and incubated with primary antibodies at 4°C overnight. The next day, membranes were washed four times with PBS containing 0.1% Tween-20 (Solarbio) and then incubated with the secondary antibody at room temperature for 1 h. Membranes were prepared for exposure to acquire images using ECL chemiluminescence solution (Thermo Fisher Scientific) incubation.

Antibodies and reagents

Immunofluorescence

Primary antibodies and reagents used for immunofluorescence were as follows: Dylight 594 isolectin B4 (Vector Laboratories, DL1207, RRID:AB_2336415, 1:200 dilution), anti-endomucin (Abcam, ab106100, RRID:AB_10859306, 1:200 dilution), MF20 antibody (Developmental Studies Hybridoma Bank, AB2147781, RRID:AB_2147781, 1:200 dilution), rat anti-CD31 (Dianova, DIA-310, RRID:AB_2631039, 1:300 dilution), mouse anti-SMA (Sigma-Aldrich, A2547, RRID:AB_476701, 1:500 dilution), rabbit anti-GFAP (Cell Signaling Technology, 80788, RRID:AB_2799963, 1:200 dilution), rabbit anti-collagen IV (Abcam, ab6586, RRID:AB_305584, 1:200 dilution), anti-versican (Abcam, ab270445, RRID:AB_3714692, 1:200 dilution), anti-GFP (Cell Signaling Technology, 2555s, RRID:AB_10692764, 1:500 dilution), anti-p-S6 (Ser 235/236) ribosomal protein (Cell Signaling Technology, 4858T, RRID:AB_916156, 1:200 dilution), anti-prosurfactant protein C (SPC) (Abcam, ab211326, RRID:AB_2927746, 1:200 dilution) and anti-PCNA (Cell Signaling Technology, 2586s, RRID:AB_2160343, 1:200 dilution).

The secondary antibodies used were as follows: ImmPRESS (peroxidase) secondary antibody (goat anti-rat, Vector Laboratories, MP-7444, RRID:AB_2336530) and TSA fluorescence system working solution (PerkinElmer, TSA-plus tetramethylrhodamine System, NEL742001KT) for CD31 immunostaining, and anti-mouse IgG (H+L), F(ab’)2 Fragment Alexa Fluor 488 conjugate (Cell Signaling Technology, 4408, RRID:AB_10694704, 1:500 dilution), anti–mouse IgG (H+L) Alexa Fluor 594 conjugate (Cell Signaling Technology, 8890, RRID:AB_2714182, 1:500 dilution), anti-rabbit IgG (H+L) Alexa Fluor 594 conjugate (Cell Signaling Technology, 8889, RRID:AB_2716249, 1:500 dilution) and anti-rat IgG (H+L) Alexa Fluor 488 conjugate (Cell Signaling Technology, 4416, RRID:AB_10693769, 1:500 dilution).

Immunohistochemistry

The antibodies and reagents used for immunohistochemistry were anti-versican (Abcam, ab270445, RRID:AB_3714692, 1:200 dilution), HRP-conjugated goat anti-rabbit IgG (H+L) (Proteintech, SA00001-2, RRID:AB_2722564, 1:2000 dilution) and a rabbit-specific IHC polymer detection kit HRP/DAB (Abcam, ab209101).

Vibratome staining

Dylight 594 Isolectin B4 (Vector Laboratories, DL1207, RRID:AB_2336415, 1:200 dilution) was used for staining vibratome sections.

Western blotting

The antibodies used for western blotting were anti-GFP (Cell Signaling Technology, 2555s, RRID:AB_10692764, 1:500 dilution), anti-p-S6 (Ser 235/236) ribosomal protein (Cell Signaling Technology, 4858T, RRID:AB_916156, 1:200 dilution) and anti-ACTB (Abclonal, AC026, RRID:AB_2768234, 1:50,000 dilution).

Real-time PCR analysis

TRIzol reagents (Thermo Fisher Scientific, 15596018) were used to extract total RNA. The StarScript II First-strand cDNA Synthesis Kit (Yeason, 11141ES) was used to produce cDNA. ChamQ Universal SYBR qPCR Master Mix (Vazyme Biotech Co, Q711-03) was added to the system for real-time qPCR experiments. Real-time PCR was performed on QuantStudio 5 systems (Applied Biosystem, A28139).

The following are the primers used in this study: h-GAPDH-Forward, 5ʹ-GAGTCAACGGATTTGGTCGT-3ʹ; h-GAPDH-Reverse, 5ʹ-GATCTCGCTCCTGGAAGATG-3ʹ; h-ADAMTS5-Forward, 5ʹ-CTGTGCTGTGATTGAAGACGA-3ʹ; h-ADAMTS5-Reverse, 5ʹ-CAGGATCTGCTTTCGTGGTAG-3ʹ; m-Gapdh-Forward, 5ʹ-GTCCCGTAGACAAAATGGTGA-3ʹ; m-Gapdh-Reverse, 5ʹ-TTTGATGTTAGTGGGGTCTCG-3ʹ; m-Klf2-Forward, 5ʹ-CGCCTCGGGTTCATTTC-3ʹ; m-Klf2-Reverse, 5ʹ-AGCCTATCTTGCCGTCCTTT-3ʹ; m-Klf4-Forward, 5ʹ-GTGCCCCGACTAACCGTTG-3ʹ; m-Klf4-Reverse, 5ʹ-GTCGTTGAACTCCTCGGTCT-3ʹ; m-Adamts1-Forward, 5ʹ-CCTTACGGCAGCAGACACA-3ʹ; m-Adamts1-Reverse, 5ʹ-AATCTGCTGTCAGTGGCCC-3ʹ; m-Adamts4-Forward, 5ʹ-CAGTGCCCGATTCATCACT-3ʹ; m-Adamts4-Reverse, 5ʹ-GAGTCAGGACCGAAGGTCAG-3ʹ; m-Adamts5-Forward, 5ʹ-CGACCCTCAAGAACTTTTGC-3ʹ; m-Adamts5-Reverse, 5ʹ-CGTCATGAGAAAGGCCAAGT-3ʹ; m-Adamts9-Forward, 5ʹ-AGCGGAAAATCAGAATGCGAAAA-3ʹ; and m-Adamts9-Reverse, 5ʹ-TGAAGGTTTGCTCCGTGGTATAA-3ʹ.

Contrast-enhanced X-ray micro-CT and analysis

The brain was harvested and fixed in 4% PFA after the mice were anesthetized and perfused with 4% PFA/PBS solution. The brains were immersed in Lugol solution (Sigma-Aldrich, L6146) for 48 h and scanned using the micro-CT system (MicroXCT-400, Xradia, CA, USA). All samples were randomly scanned by a masked operator. The images were acquired with the following parameters: 50 kV, 10 W, 721 projections, 3 s integration per 180° rotation.

Post-lesion labeling of micro-CT scanning results was done by a masked investigator who was unaware of experimental details. We used Avizo 3D software (Thermo Fisher Scientific) to analyze the data and the specific protocols as previously described (Yang et al., 2023).

Statistical analysis

All tables were analyzed with χ² analysis, and survival curves were analyzed with the Mantel–Cox test. All other statistical analyses in the study were performed with an unpaired two-tailed t-test using GraphPad Prism 8.0 statistical software. Statistical significance was considered when P<0.05.