Introduction

Cerebral cavernous malformations (CCMs), accounting for 5-15% of cerebrovascular abnormalities, are characterized by blood-filled endothelial-lined cavities, and can cause seizures, headaches, neurological deficits, and recurrent stroke risk¹. Familial CCMs are due to heterozygous loss of function mutations in the KRIT1, CCM2, or PDCD10 genes with a second mutation inactivating the normal allele in random brain endothelial cells². We previously developed a zebrafish CCM model using CRISPR-Cas9 to inactivate ccm2 in a manner that replicates the mosaic genetic background of the human disease³. This zebrafish model exhibits two phenotypic phases: lethal embryonic caudal venous plexus (CVP) cavernomas at 2 days post-fertilization (dpf) in ∼30% of embryos and histologically-typical CNS CCMs in ∼100% of surviving 8 week old fish³. Both phases of this model, like their murine counterpart, depend on Krüppel-like factor 2 (KLF2), confirming shared transcriptional pathways. Furthermore, both phases of the zebrafish model exhibit similar pharmacological sensitivities (Supplementary Table 1) to murine and human lesions. Thus, this zebrafish model offers a powerful tool for genetic and pharmacological analysis of the mechanisms of CCM formation.

Anecdotal case reports^4–6 and a recent phase 2 clinical trial have suggested that propranolol a non-selective β adrenergic receptor antagonist benefit patients with symptomatic CCMs⁷. In previous studies, we observed this effect of propranolol in both zebrafish and murine models of CCM⁸. Importantly, propranolol is a racemic mixture of R and S enantiomers and elegant work from the Bischoff lab has implicated the R enantiomer, which lacks β adrenergic antagonism, as the component that inhibits SOX18 thereby suppressing infantile hemangiomas⁹. We previously found that the anti-adrenergic S enantiomer rather than the R enantiomer inhibited development of embryonic CVP cavernomas in ccm2 CRISPR zebrafish⁸. Furthermore, morpholino silencing of the gene that encodes the β1 (adrb1) but not β2 (adrb2) receptor also prevents CVP cavernomas⁸, suggesting that the β1 adrenergic receptor (β1AR), which primarily impacts hemodynamics¹⁰, contributes to the pathogenesis of CCM. Here we have inactivated the adrb1 gene to eliminate the β1AR and observed the expected reduction of heart rate and contractility which resulted in reduced blood flow through the CVP. These adrb1^-/- zebrafish were protected from embryonic CVP cavernomas and, importantly, also exhibited much reduced number and volume of adult brain CCM. Furthermore, a β1-selective antagonist, metoprolol, also inhibited adult CCM suggesting that β1AR-specific antagonists may be useful for CCM treatment and will carry less potential for β2 AR-related side effects such as bronchospasm¹¹.

Results

β1AR is important in for development of CVP cavernomas

We previously reported that morpholino silencing of adrb1 rescued CVP cavernomas in zebrafish embryos⁸. Because of morpholinos potential for off target effects^{12, 13}, we sought to confirm the β1AR’s potential involvement in CCM pathogenesis by inactivating adrb1. We used CRISPR-Cas9 to generate an 8-bp deletion resulting in a premature stop codon at 57 bp (Figure 1A). To exclude potential off-target effects, the top 20 potential off-target sites predicted by Cas-OFFinder (www.rgenome.net/cas-offinder) (Supplementary Table 2) were sequenced and were not mutated (data not shown). We intercrossed adrb1^+/- F1 offspring resulting in adrb1^-/- zebrafish embryos that displayed reduced cardiac contractility (Supplementary Video 1) and decreased heart rate (Supplementary Figure 1A). Similar to Adrb1^-/- mice¹⁴, adrb1^-/- zebrafish exhibited a blunted chronotropic response to a β1AR agonist, isoprenaline hydrochloride (Figure 1B). Adrb1^-/- embryos had no obvious defects in vascular development and they survived to adulthood and were fertile.

Figure 1.

We intercrossed adrb^+/- individuals and injected one-cell stage embryos with ccm2 CRISPR and blindly scored the presence of CVP cavernomas at 48hpf. As expected we observed CVP cavernomas in 28% of adrb1^+/+ embryos. In sharp contrast only 3% of adrb1^-/- embryos exhibited cavernomas (Fig. 1C and D) indicating that loss of β1AR prevents CVP cavernomas. these observations demonstrate that the β1AR is required for embryonic CVP cavernoma formation.

β1AR mediates formation of adult CCM in the brain

Adult ccm2 CRISPR zebrafish display highly penetrant CCMs throughout the central nervous system³. Brains from adult ccm2 CRISPR fish on adrb1^-/- (12 brains) or wild type (13 brains) background were treated with CUBIC (clear, unobstructed brain/body imaging cocktails and computational analysis)¹⁵, and these transparent brains were then scanned with light-sheet microscopy and lesions were enumerated and volumes were estimated with NIH ImageJ. While the typical multi-cavern CCMs were observed in brains on wild type background appearing as blood filled dilated vessels (Figure 2A through C), ccm2 CRISPR adrb1^-/- fish exhibited an 87% reduction in lesion volume (Figure 2D through G). Thus, genetic inactivation of β1AR prevented CCMs.

Figure 2.

A selective β1AR antagonist prevents CCMs

The non-selective β blocker propranolol reduces lesion volume in murine CCM models^{8, 16}. To ascertain whether propranolol had a similar effect in the zebrafish model, the chemical treatment was started from larval stage at concentration which allows the fish to develop to two months for CCM inspection. We added 12.5µM propranolol to or vehicle control to fish water of ccm2 CRISPR zebrafish starting at 3 weeks of age. The water was refreshed daily with drug or vehicle until fish were sacrificed and brains were examined as described (Figure 3A). Quantification based on light-sheet scanning of zebrafish brains (Figure 3B) showed that compared to vehicle-treated controls (Figure 3C, D, and E), propranolol-treated groups displayed a 94% reduction in CCM lesion volume (Figure 3F, G, and H). Similarly, administration of 50µM racemic metoprolol a β1-selctive antagonist produced a similar (98%) reduction in lesion volume (Figure 3B, I, J, and K). Importantly, neither drug at the doses administered reduced the growth of the fish or the volume of their brains. In sum, both genetic and pharmacological loss of β1 adrenergic receptor signaling markedly reduces the lesion burden in the zebrafish ccm2 CRISPR model of CCM.

Figure 3.

Loss of β1AR does not prevent increased klf2a expression in ccm2 null embryos

Inactivation of CCM genes leads to increased endothelial KLF2 expression^{17, 18}, a transcription factor important in cavernoma formation^{3, 19}. Nevertheless, silencing of adrb1 did not prevent the expected increased endothelial klf2a¹⁷ expression in ccm2 morphant Tg(klf2a:H2b-EGFP) fish in which the nuclear EGFP expression is driven by the klf2a promoter (Figure 4A and B). We previously reported that mosaic expression of KLF2a occurs in ccm2 CRISPR fish and is sufficient to drive cavernoma formation when blood is flowing³. Nevertheless, the adrb1 morphant (Figure 4C) and control morphant (Figure 4D) ccm2 CRISPR Tg(klf2a:H2b-EGFP) embryos both displayed a mosaic increase of klf2a reporter (Figure 4E). Similarly, merely slowing the heart and reducing contractility with 2,3-butanedione monoxime (BDM)²⁰ also prevented CVP cavernomas (Fig 4F, and Supplementary Figure 2). Thus, silencing β1AR does not prevent the generalized increase in endothelial KLF2a in ccm2 morphants nor does it prevent the mosaic KLF2a increase in ccm2 CRISPR zebrafish.

Figure 4.

Discussion

In this study, we provide genetic and pharmacological evidence to implicate the β1 adrenergic receptor (β1AR) in the pathogenesis of CCM. Our data suggest that inactivating β1AR via gene inactivation or pharmacological inhibition can significantly reduce the volume of CCM lesions in a zebrafish model.

Case reports^{4–6, 21} and our previous study⁸ revealed the potential benefit of the non-selective β-adrenergic receptor blocker propranolol in reducing CCMs in patients and mouse models, respectively. However, propranolol is a racemic mixture, and its R enantiomer which lacks β-adrenergic antagonism was reported to show therapeutic effect for infantile hemangiomas in an animal study⁹. Notably, our study demonstrates that adrb1^-/- zebrafish are significantly protected against the formation of CCMs, suggesting that propranolol’s therapeutic effect on CCM is through β1AR antagonism. Together with the observed significant reduction of CCM lesion volume upon metoprolol treatment, a selective β1AR antagonist, supports the therapeutic potential of β1AR antagonism in CCMs. β1 selective blockers offer the advantage of causing fewer mechanism-based side effects compared to propranolol, such as bronchospasm²², and are already in clinical use (including agents like atenolol, metoprolol, nebivolol, and bisoprolol). Thus, further studies are warranted to evaluate the potential value of β1 selective antagonists in this disease.

Consistent with the Adrb1KO mice¹⁴, the adrb1^-/- zebrafish embryos displayed the decreased chronotropic response to a beta-adrenergic agonist, isoprenaline, and decreased basal heart rate compared to that of adrb1^+/+ brood mates; however, CVP morphology of adrb1^-/- embryo was not perturbed (Supplementary Figure 3). adrb1^-/- embryo also displayed a marked decrease of blood flow in CVP (Supplementary Video 2, and Supplementary Figure 1B), A similar reduction of heart rate and contractility by 2,3-BDM, a cardiac myosin ATPase inhibitor, prevented CVP dilation (Figure.4F). We previously found that arresting blood flow prevents aberrant intussusceptive angiogenesis and the resulted CVP cavernomas in ccm2 CRISPR embryos³. Taken together, these data suggest that reduced blood flow secondary to reduced cardiac function accounts for the protective effect of loss of β1AR signaling on CVP lesions and on CCM. Importantly, although β1AR are highly expressed in cardiomyocytes and contribute to increased cardiac output²³, β1ARs are also expressed in other tissues^{24, 25} including endothelial cells of vascular anomalies in patients²⁶. Thus, future studies will be required to define the tissue specificity of the role of the β1AR in CCM pathogenesis.

Experimental procedures

Zebrafish lines and handling

Zebrafish were maintained and with approval of Institutional Animal Care and Use Committee of the University of California, San Diego. The following mutant and transgenic lines were maintained under standard conditions: Tg(fli1:EGFP)^{y1 27}, Tg(klf2a:H2b-EGFP) ²⁸, and Tg(kdrl:mcherry)^{is5 29}. adrb1^-/- zebrafish was obtained by coninjection of Cas9 protein (EnGen Spy Cas9 NLS, M0646, NEB) with gRNA targeting adrb1. Genotyping of adrb1^-/- was performed with forward primer (5’-AGAGCAGAGCGCGGATGGAA-3’) and reverse primer (5’-GATCCATACATCCAGGCT-3’).

Plasmids and morpholino

pCS2-nls-zCas9-nls (47929) and pT7-gRNA (46759) were bought from Addgene. The CRISPR RNA (crRNA) sequences used in this study are as follow: ccm2-1 5’- GGTGTTTCTGAAAGGGGAGA-3’, ccm2-2 5’- GGAGAAGGGTAGGGATAAGA-3’, ccm2- 3 5’-GGGTAGGGATAAGAAGGCTC-3’, ccm2-4 5’-GGACAGCTGACCTCAGTTCC-3’, adrb1 5’-GACTCTAAACGCGCCACGG-3’. Target gRNA constructs were generated as described before³⁰. Morpholino sequence used in this study are as follow: adrb1 (5’- ACGGTAGCCCGTCTCCCATGATTTG-3’), ccm2 (5’-GAAGCTGAGTAATACCTTAACTT CC-3’), tnnt2a (5’-CATGTTTGCTCTGATCTGACACGCA-3’), control (5’- CCTCTTACCTC AGTTACAATTTATA-3’).

RNA synthesis

The pCS2-nls-zCas9-nls plasmid containing Cas9 mRNA was digested with NotI enzyme, followed by purification using a Macherey-Nagel column, serving as the template. The capped nls-zCas9-nls RNA was synthesized using the mMESSAGE mMACHINE SP6 Transcription Kit from ThermoFisher Scientific. The resulting RNA was purified through lithium chloride precipitation, as per the instructions provided in the kit. For gRNA synthesis, the gRNA constructs were linearized using BamHI enzyme and purified using a Macherey-Nagel column. The gRNA was synthesized via in vitro transcription using the MEGAshortscript T7 Transcription Kit from ThermoFisher Scientific. After synthesis, the gRNA was purified by alcohol precipitation, as instructed in the same kit. The concentration of the nls-zCas9-nls RNA and gRNA was measured using a NanoDrop 1000 Spectrophotometer from ThermoFisher Scientific, and their quality was confirmed through electrophoresis on a 1% (wt/vol) agarose gel.

Microinjection

All injections were performed at 1-cell stage with a volume of 0.5nl. The final injection concentrations are as follow: Cas9 protein (10µM), Cas9 mRNA (750ng/μl), gRNA 120ng/μl, adrb1 MO (4ng/µl), ccm2 MO (4ng/µl), tnnt2a MO (5.3 ng/µl), control MO (4ng/µl).

Chemical treatment

Propranolol (P0995, TCI) (12.5µM) and metoprolol (M1174, Spectrum) (50µM) were used to treat the zebrafish larva from Day 21. Beginning from Day 35, adjusted concentration of propranolol (25µM) or metoprolol (100µM) were used to treat the juvenile fish. The chemicals were dissolved in fish water, and fish water containing chemicals were refreshed daily. Egg water without above chemicals was refreshed daily for fish used as negative control. 2,3-butanedione monoxime (BDM) (6mM) was added to egg water of the ccm2 CRISPR embryos at 25hpf, and CVP cavernoma was observed at 2dpf.

Airyscan imaging and fluorescence intensity analysis

To prepare the embryos for imaging, they were first anesthetized using egg water containing 0.016% tricaine (3-amino benzoic acid ethyl ester) from Sigma-Aldrich. Subsequently, the anesthetized embryos were embedded in 1% low melting point agarose obtained from Invitrogen (product number 16520050). The imaging process was carried out using a Zeiss 880 Airyscan confocal microscope, utilizing the standard Airyscan mode, with a 20x/NA 0.8 objective. The intensity of nuclear EGFP (enhanced green fluorescent protein) was quantified using ImageJ software.

Heartbeat and blood flow recording

Heartbeat and blood flow were recorded using Olympus MVX10. The embryos were treated with 0.004% tricaine which does not have effect on heartbeat^{31, 32}.

Zebrafish brain dissection, CUBIC treatment and lightsheet imaging

The dissection of zebrafish brains followed the methodology described in a previous study by Gupta et al.³³. The CUBIC method was optimized based on the findings from a previous report by Susaki et al.¹⁵. The brains were fixed in 4% paraformaldehyde (PFA) with a pH of 7.5 for 24 hours and subsequently washed with PBS (phosphate-buffered saline) for an additional 24 hours. Following the PBS wash, the brains underwent CUBICR1 treatment at 37°C in a water bath for 42 hours. For imaging, the samples were placed in CUBICR2 as the imaging medium and imaged using a ZEISS Lightsheet Z.1 microscope. Scanning was carried out utilizing 5X dual illumination optics in combination with a 5X objective.

Statistical analysis

The statistical analysis was conducted using GraphPad Prism software. P-values were calculated using an unpaired two-tailed Student’s t-test, unless otherwise specified. The bar graphs display the mean values along with their corresponding standard deviations (SD).

Acknowledgements

We gratefully acknowledge Mark H. Ginsberg, Douglas A Marchuk, and Issam A. Awad for their invaluable advice on both this study and the manuscript. This work was supported by NIH grants P01 NS092521 and P01 HL151433. We also acknowledge Jennifer Santini and Marcy Erb for microscopy technical assistance and resources provided by the UCSD School of Medicine Microscopy Core (NINDS P30 NS047101).