The mTORC1 inhibitor Rapamycin inhibits growth of cerebral cavernous malformation in adult mice

Abstract

Background:

Cerebral cavernous malformations (CCMs) are vascular malformations that frequently cause stroke. CCMs arise due to loss-of-function (LOF) in one of the genes that encode the CCM complex, a negative regulator of MEKK3-KLF2/4 signaling in vascular endothelial cells. Gain-of-function (GOF) mutations in PIK3CA (encoding the enzymatic subunit of the PI3K pathway associated with cell growth) synergize with CCM gene LOF to generate rapidly growing lesions.

Methods:

We recently developed a model of CCM formation that closely reproduces key events in human CCM formation through inducible CCM LOF and PIK3CA GOF in mature mice. In the present study we use this model to test the ability of Rapamycin, a clinically approved inhibitor of the PI3K effector mTORC1, to treat rapidly growing CCMs.

Results:

We show that both intraperitoneal and oral administration of Rapamycin arrests CCM growth, reduces peri-lesional iron deposition, and improves vascular perfusion within CCMs.

Conclusions:

Our findings further establish this adult CCM model as a valuable preclinical model and support clinical testing of Rapamycin to treat rapidly growing human CCMs.

Brief Summary

Cerebral cavernous malformations (CCMs) are vascular lesions that cause stroke in young people. There is no medical therapy for CCMs. Li et al show that Rapamycin effectively treats CCM in mice.

Graphical abstract

Introduction

CCMs present as either single or multiple lesions within the central nervous system that may be associated with intracerebral hemorrhage (ICH), epileptic seizures, and/or progressive neurologic deficits depending on lesion location¹. Human and mouse genetic studies have demonstrated that CCM lesions develop following endothelial cell LOF mutations in the KRIT1 (Krev interaction trapped protein 1), CCM2, or PDCD10 (programmed cell death protein 10) genes that encode components of the CCM protein complex required to negatively regulate the MEKK3-KLF2/4 signaling pathway^2,3 or due to an activating point mutation in the MEKK3 MAPK⁴. CCMs may form sporadically or in a familial pattern, the latter due to germline LOF mutations that often present with multiple lesions⁵. Familial CCMs are highly heterogeneous in clinical manifestation, and the disease progression and severity are not directly correlated with the type of germline mutations of CCM genes^6–9. Recent studies have demonstrated that environmental factors, such as the gut microbiome and the gut barrier, may contribute significantly to the formation of CCMs^10–11. Additional secondary local events following CCM germline mutation are also involved to synergize the progression of CCM pathogenesis, which worsen the stable CCMs becoming symptomatic manifestation and requires clinical management^12–14. By contrast, sporadic CCMs are often associated with developmental venous anomaly (DVA) and typically present as solitary lesions^15,16. Recent studies of surgically excised sporadic CCMs suggest that their growth is typically driven by GOF mutations in PIK3CA coupled with LOF mutations in a CCM gene or a presumed GOF mutation in the MAP3K3 gene that encodes the mitogen activated protein kinase (MAPK) directly regulated by the CCM complex^17,18. Single nuclear sequencing further demonstrates that these mutations most likely arise in a single cell, while studies in mice and humans support the hypothesis that this cell is an endothelial cell within a capillary or post-capillary venule of the central nervous system¹⁹.

Our previous studies suggest that both CCM LOF and PIK3CA GOF confer increased activity of the mTORC1 pathway downstream of phosphoinositide 3-kinases (PI3K), and that inhibition of mTORC1 using Rapamycin can prevent the formation of new CCM lesions in both neonatal mice (in which CCM LOF alone is sufficient for lesion formation) and mature mice (in which both CCM LOF and PIK3CA GOF are required for lesion formation)¹⁴. During the course of these studies, we developed an inducible adult CCM mouse model in which Cre-inducible Krit1 LOF and Pik3ca GOF results in actively growing CCM lesions that closely model human sporadic CCMs. These findings suggested that this model might be used as a pre-clinical model to test the effect of drugs on pre-existing, rapidly growing CCMs like those that are most frequently diagnosed in human patients. In the present study we apply this model to test whether Rapamycin (aka Sirolimus), a US Food and Drug Administration (FDA)-approved inhibitor of mTORC1, can arrest or even reverse the growth of pre-existing, high aggressive CCM lesions. Our studies suggest that mTORC1 inhibition may be a highly effective means of medically arresting the growth of CCM lesions, e.g. in patients in whom surgical resection is not feasible or very high risk. These studies also demonstrate the utility of this new mouse genetic model for pre-clinical assessment and comparison of putative CCM medical therapies.

Methods

The data that support the findings of this study are available from the corresponding authors upon reasonable request. See Supplemental Material for detailed information on methods.

Collection of human CCMs

Human CCM tissue specimens were freshly obtained from surgical resection at the Hospital of the University of Pennsylvania after institution’s Institutional Review Board (IRB) approval. Surgery for CCM patients were based on the discretion of the treating physician and indication for surgery was unrelated to this study. CCM was confirmed by pathological assessment.

Mice

The Krit1^fl/fl; R26-LSL-Pik3ca^H1047R transgenic mice were bred and housed in a pathogen-free environment in a vivarium approved by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) at the University of Pennsylvania. All animal protocols and experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Pennsylvania and were performed in accordance with relevant guidelines and regulations. This preclinical study conformed to the updated ARRIVE 2.0 guidelines³³.

Adeno-associated virus (AAV) injection

Recombinant AAV-BR1-Cre vectors were donated by J. Körbelin’s laboratory³⁴. Immediately following craniotomy, AAV-BR1 vectors carrying Cre recombinase under the control of the CAG promoter were injected into the right somatosensory cortex centered on +1.5 mm lateral, −1.0 mm posterior to bregma at a depth of 0.5 mm using a beveled glass micropipette. A total of 5 × 10⁷ virus particles in 100 nl sterile PBS were injected over the course of 5 min for each mouse.

Rapamycin administration

Rapamycin was purchased from MedChemExpress (HY-10219). Mice were randomly divided into two groups and received either Rapamycin (100 μg) or vehicle (5% DMSO) via intraperitoneal injection (i.p.). Injections were performed daily within a two-hour window. Rapamycin diet was microencapsulated with Eudragit S100 polymer (Rapamycin Holdings) and prepared in irradiated pelleted chow (5LG6 formulation, LabDiet) at a final concentration of 14 mg microencapsulated Rapamycin per kilogram of pelleted chow. For control diet (placebo), equivalent amount of Eudragit S100 polymer were prepared in the irradiated pelleted chow using the same 5LG6 formulation.

In vivo high-frequency transcranioplasty ultrasound (TCUS) imaging and volumetric measurement

Echo with ultrasounds of frequencies ranging from 50 to 60 MHz was performed using Visual Sonic Vevo 2100 system equipped with MS700 transducer (VisualSonics). Real-time B-mode cine-loops clips with the transducer index mark orientation were recorded. Images were acquired and analyzed using Vevo Lab software (VisualSonics). CCM lesion volume was calculated using the formula V=4/3 × π × L/2 × W/2 × H/2³⁵.

In vivo cerebrovascular perfusion and live-animal imaging

Mice were anesthetized with isoflurane (3.0% induction, 1.0–2.0% maintenance in 1.0 l/min oxygen) and placed on a heating pad covered with a cotton pad to maintain a body temperature of 37.5 °C. A total of 100 μl of 50 mg/ml of FITC-Dextran (70 kDa; Sigma) was injected through tail veins. Time-lapse images were acquired using Olympus SZX16 fluorescent microscope and cellSens software (Olympus).

Contrast-enhanced microcomputed tomography (microCT) and volumetric quantification

Brains were harvested and stored in 4% (w/v) PFA/PBS (pH 7.4) and stained with non-destructive iodine contrast, and subsequent microCT imaging was performed as described³⁶. Tissue processing, imaging and volume quantification were performed in a blinded manner by investigators at the University of Chicago without any knowledge of genotype and experimental details.

Histology and immunohistochemistry

Human and mouse tissue were harvested and fixed in 4% PFA/PBS overnight, dehydrated in 100% ethanol, and embedded in paraffin. Tissue sections of 6 μm underwent dewaxing and rehydration through xylene and ethanol and were subsequently used for hematoxylin and eosin (H&E), Prussian Blue, and immunohistochemistry staining.

Statistics

All experimental and control animals were littermates, and none was excluded from analyses at the time of harvest. All data were analyzed with GraphPad Prism software and presented as means ± SEM. P values were calculated using an unpaired two-tailed Welch’s t-test, or repeated-measures one-way analysis of variance (ANOVA) followed by Dunnett post-hoc test for multiple comparisons as indicated in the figure legends. P values of less than 0.05 were considered statistically significant and are denoted as follows: *P < 0.05; **P < 0.01; ***P < 0.001.

Results

Generation of an inducible adult CCM mouse model that closely mimics aggressive human CCM disease

Prior studies of CCM formation using mouse models have primarily utilized a neonatal model in which CCM LOF is induced genetically in brain endothelial cells very shortly after birth. While this model has accurately revealed the requirement for CCM function in endothelial cells to prevent lesion formation, it has several important short-comings. First, CCM formation in the neonate after endothelial loss of CCM function is limited to the hindbrain and retina, sites of active angiogenesis. In contrast, human CCM lesions arise throughout the central nervous system. Second, in the neonatal model loss of CCM function is sufficient for lesion formation. In contrast, recent sequencing studies of human CCMs and genetic studies of adult mice demonstrate that a majority of sporadic CCMs require both CCM LOF (or MEKK3 GOF) and PIK3CA GOF to arise. Finally, broad genetic perturbation of all brain endothelial cells results in diffuse lesion formation that can only be assessed after sacrifice of the animal. In contrast, most human lesions appear as isolated lesions that are followed over time using live imaging such as MRI.

To address these limitations and generate a small animal model that more faithfully reflects the genetic and clinical presentation of human CCM we developed an inducible adult CCM mouse model using Krit1^fl/fl; R26-LSL-Pik3ca^H1047R (Krit1^fl/fl; iPik3ca^H1047R) transgenic mice in which Cre expression drives both CCM LOF and PIK3CA GOF (Figure 1A). To induce formation of a single CCM lesion we injected adeno-associated virus (AAV) vector encoding Cre recombinase (AAV-Cre) into the brains of adult Krit1^fl/fl; iPik3ca^H1047R mice following craniotomy and creation of a cranial window that permits direct visualization of subsequent lesion formation (Figure 1B). CCM formation was detected through a transparent polymethyl methacrylate (PMMA) cranial window using microscopy and ultrasound visualization as early as post-operative day (POD) 7. Control Krit1^fl/fl; iPik3ca^H1047R mice that did not receive AAV-Cre did not develop CCM lesions (Figure 1C). CCM lesions generated using this model exhibited a classic “mulberry” or “popcorn” appearance characterized by engorged vascular clusters with a dark purple color like that of isolated human CCM lesions (Figure 1D)²⁰. Human CCM lesions often exhibit low levels of blood flow²¹, and we also observed low vascular perfusion of isolated CCMs in this model following FITC-Dextran injection (Figure 1E). Histologic analysis of CCM lesions generated in Krit1^fl/fl; iPik3ca^H1047R mice showed multiple dilated, thin-walled capillaries and hemosiderin deposition (Figure 1Fb), features also prominent in human CCM lesions (Figure 1Fa). The presence of hemosiderin immediately surrounding the abnormal vasculature in both human and mouse samples is consistent with peri-lesional hemorrhage (Figure 1Fa’ and 1Fb’). Immunohistological studies showed that the CCM lesions in both human and mouse samples contained large fibrin-rich thrombi in the lumen of the affected vessels associated with the presence of the procoagulant Von Willebrand factor (vWF) (Figure 1Fa”–a”’ and 1Fb”–b”’). Following CCM lesion formation, reactive astrocytes and microglia were detected at the peri-lesional border, with aggregation of infiltrated CD45-positive leukocytes within the lesion core (Figure S1). These findings are consistent with a neuroinflammatory response during the development of CCM lesions, as previously observed^22,23.

Figure 1. Characterization of an inducible adult CCM mouse model.

(A) Schematic representation of mouse genetics of the inducible adult CCM mouse model. (B) A representative microscopic image showing craniotomy and stereotaxic injection of AAV-Cre injection into the right somatosensory cortex of a mouse brain. (C) Representative microscopic images showing mouse brains through a glass cranial window following sterile PBS injection (left) and AAV-Cre injection (right). (D) Representative microscopic images showing a single supratentorial CCM presented on the occipital lobe (human, left) and somatosensory cortex (mouse, right). (E) Representative microscopic and fluorescent images showing poor blood flow within the CCM lesion by fluorescein angiography (human, left) and FITC-Dextran cerebral vascular perfusion (mouse, right). (F) Representative images showing histology exams of human (a) and mouse (b) CCM in the cerebral cortex. H&E staining revealed an overall “popcorn” appearance with enlarged vasculature and hemosiderin hallmarks (white arrows) around the perilesional sites. Adjacent human CCM histological sections were immunostained with (a’) CD31 (red) and CD235 (green); (a”) CD31 (red) and Fibrin (green); (a”’) CD31 (red), vWF (green). Adjacent mouse CCM histological sections were immunostained with (b’) Emcn (red) and Ter119 (green); (b”) Emcn (red) and Fibrin (green); (b”’) Emcn (red) and vWF (green). Both human and mouse sections were counterstained with DAPI (Blue). The boxed regions in (a”, a”’, b”, b”’) are shown in (i-iv), respectively. Scale bar: 100 μm.

These studies demonstrate that AAV-Cre induced CCM lesions in adult Krit1^fl/fl; iPik3ca^H1047R mice exhibit most of the cardinal features associated with human CCM lesions, suggesting that this model is a faithful reproduction of aggressive, clinically relevant human disease.

Rapamycin arrests growth of pre-existing CCM lesions

Prior studies have demonstrated that Rapamycin can prevent CCM lesion formation if administered prior to genetic loss of CCM function and gain of PIK3CA function¹⁴. However, since human patients present with pre-existing lesions these studies failed to address whether Rapamycin might also be able to arrest or even reverse lesion growth. The generation of an adult inducible mouse model of CCM formation that permits direct visualization of lesion growth enabled us to address this question. CCM lesion growth was variable in our model, as it is in human patients, with lesion appearance first noted between POD 7 and POD 14 after AAV-Cre administration (Figure 1C). The transduction efficiency of intraparenchymal delivery of AAV-Cre to the brain endothelial cells was measured to be between 41.69 ± 7.15% and 69.15 ± 2.76% (Figure S2), likely explaining the fairly large variability in the onset day of lesion formation and the rate of lesion growth. To control for this variable, we restricted our studies to lesions that were between 0.5 mm and 1.0 mm in diameter, measured through microscopic imaging, on POD 11 (Figure 2A–B). Animals with CCM lesions between 0.5 mm and 1.0 mm in diameter were randomized into two groups to receive either daily Rapamycin (3.3 mg/kg) or vehicle (5% DMSO/PBS) treatment via intraperitoneal (i.p.) injection (volume of 50 μl), respectively (Figure 2A). In the vehicle-treated mice, CCM lesions grew rapidly and typically exceeded the observational cranial window within 10 days after the beginning of the experimental period. In contrast, CCM lesions in mice treated with Rapamycin remained virtually unchanged in size throughout the entire time period (i.e. POD 11 to POD 28) (Figure 2B, Figure S3). Visual and microcomputed tomography (microCT) analysis of post-mortem brains on POD 28 was used to measure actual lesion size since much of the lesion in the control group grew outside the narrow area viewable through the cranial window. These analyses confirmed a dramatic reduction in CCM lesion size in Rapamycin-treated animals compared with mice in the vehicle-treated group (P less than 0.001; Figure 2C–D, Figure S4). Immunostaining of phosphorylated ribosomal protein S6 (p-S6) in brains collected on POD 28 demonstrated strong suppression of mTOC signaling following a 17-day consecutive treatment of Rapamycin (Figure 2E). These studies reveal a strong effect of Rapamycin on lesion growth, with treatment able to achieve almost complete cessation growth in pre-existing CCM lesions.

Figure 2. Rapamycin arrests CCM lesion growth in adult CCM mouse mod.

(A) Schematic representation of adult Krit1^fl/fl; iPik3ca^H1047R mice between age 8–10 weeks undergo craniotomy and receive AAV-Cre injection to induce Krit1 deletion and Pik3ca^H1047R over expression. Mice with CCM lesion size between 0.5 mm and 1.0 mm on POD 11 were enrolled in the study and received either rapamycin (3.3 mg/kg) or vehicle (5% DMSO/PBS) every day via intraperitoneal (i.p.) injection (volume of 50 μl), respectively, from POD 11 to POD 28. (B) A serial of microscopic images through the cranial windows showing CCM growth in either vehicle- or rapamycin-treated mouse brains from POD 11 to POD 28. Representative images of two mice in each group are shown. (C) Representative visual images and (D) microCT quantification of CCM lesion volumes of vehicle- and rapamycin-treated mouse brains collected on POD 28. Scale bar: 1 mm. (E) Representative images showing immunostaining for Emcn (red), p-S6 (green) and DAPI (blue) are shown. Boxes in the images denote the area of the magnified images (a) and (b). White arrows indicate p-S6-positive endothelial cells. Scale bar: 100 μm. Vehicle, n = 6; Rapamycin, n = 8. Data shown are means ± SEM. ***P=0.0007 by unpaired two-tailed Welch’s t-test.

Serial ultrasound imaging demonstrates CCM lesion growth arrest with Rapamycin

To better define the temporal effect of Rapamycin on CCM lesion growth we next performed serial live imaging using high-frequency transcranioplasty ultrasound (TCUS). TCUS was enabled by the use of a sonolucent PMMA window placed at the time of craniotomy. To compare the effect of Rapamycin with vehicle-treated control animals using TCUS we restricted our analysis to animals with a solitary CCM lesion (Figure 3A–B, Figure S5D) as the presence of multiple small CCM lesions would preclude accurate ultrasonographic measurement of lesion volume. TCUS was used to measure CCM lesion volume every 3–4 days between POD 11 and POD 28 in both Rapamycin-treated and vehicle-treated animals (Figure 3A). Similar to the echogenic appearance of human CCM lesions during intraoperative ultrasound (Figure S5B), CCMs generated using this model appeared as contrast-enhanced hyperechoic lesions distinct from adjacent healthy brain tissues (Figure S5A–B). Rapamycin treated mice exhibited stable CCM lesion volumes throughout the study period when measured using TCUS. Lesion TCUS volumes made using B-mode ultrasound imaging exhibited tight correlation with 2-dimensional visual measurements made using microscopic imaging (Figure S5C). In vehicle-treated mice, CCM lesions exceeded the cranial window over time and CCM lesion volume was not able to be accurately measured at later time points with TCUS. These findings demonstrate (i) that CCM lesion volume can be reliably measured using high-frequency ultrasound in vivo, and (ii) that Rapamycin is a potent inhibitor of CCM growth. They also suggest that Rapamycin treatment can inhibit CCM growth but may not regress pre-existing lesions (discussed further below).

Figure 3. Rapamycin stabilizes CCM lesion volume.

(A) A serial of ultrasonic images through the PMMA cranial windows showing CCM growth in either vehicle- or rapamycin-treated mouse brains from POD 11 to POD 28. CCM lesions are outlined with red dotted contours at the bottom panels to represent the lesion appearance using ultrasound imaging. (B) Lesion volumes of CCM lesions were calculated using the formula V=4/3 x π x L/2 x W/2 x H/2. All lesion volumes between POD 11 and POD 28 were normalized to the lesion volume calculated on POD11. N = 3 of solitary CCM lesions in both groups were analyzed. The boxed region outlined with red dotted contour indicates the lesion volumes between POD 21 and POD 28 of the vehicle-treated mice cannot be accurately measured because the lesions have grown beyond the PMMA cranial window. Data shown are means ± SEM. P=0.2667 by repeated-measures one-way analysis of variance (ANOVA) in Rapamycin-treated mice.

Rapamycin treatment is associated with reduced lesional hemorrhage and increased lesional blood flow

CCM lesion hemorrhage is considered a major adverse clinical outcome that is linked to stroke and neurologic deficit²⁴. Previous studies suggested that intralesional clot formation and outflow obstruction may lead to peri-lesional hemorrhage or oozing in CCM²⁵. Thus, agents that reduce the likelihood of these events are predicted to confer clinical protection against some of the most important complications of CCM disease. ²⁹To determine whether Rapamycin treatment is associated with changes in these events we performed post-mortem histologic analyses to assess lesional hemorrhage and in vivo time-lapse fluorescent angiography imaging to assess intra-lesional blood flow.

Post-mortem analyses of vehicle-treated mouse brains collected on POD 28 revealed significant iron deposition in the perilesional regions, consistent with hemorrhage and hemosiderosis during CCM growth (Figure 4A). Lesions treated with Rapamycin exhibited significantly reduced the perilesional iron deposition (Figure 4B), suggesting that reduced lesion growth is associated with reduced lesion hemorrhage.

Figure 4. Rapamycin reduces hemorrhagic event.

(A) Representative microscopic images showing H&E and Prussian Blue stained vehicle- or rapamycin-treated mouse brains collected on POD 28. Two mice from each group were shown as representative results. Scale bar: 250 μm. (B) Quantification of iron deposition of vehicle- and Rapamycin-treated mouse brains collected on POD 28. Vehicle, n = 6; Rapamycin, n = 8. Data shown are means ± SEM. **P=0.0035 by unpaired two-tailed Welch’s t-test.

FITC-dextran injected into the tail vein can be visually identified in the cerebral vessels through the cranial window immediately following injection and appeared to be distributed evenly throughout the vasculature within 10 minutes (Figure 5A). In healthy mouse brains, baseline angiography performed using tail vein FITC-dextran injection demonstrated normal blood circulation within the superficial cerebral vasculature visible through the cranial window. In the vessels of nascent POD11 CCM lesions, FITC-dextran detection in the lesion core was weak and often entirely absent 10 minutes after injection (Figure 5B, C). Following 17-days of Rapamycin treatment, FITC-dextran detection in the lesion core appeared compared to both the same lesion at the onset of the experiment and to CCM lesions treated with vehicle only (Figure 5B). Quantification of the FITC-dextran-negative lesion area within the lesion core 10 minutes after injection revealed a reduced no-flow lesion area in Rapamycin-treated animals compared with mice in the vehicle-treated group (P less than 0.05; Figure 5C–E).

Figure 5. Rapamycin reduces clotted area at the CCM lesion core.

(A) Representative in vivo time-lapse fluorescent microscopic images showing healthy cerebral vasculature labeled by FITC-Dextran through tail vein injection. (B) Representative in vivo time-lapse fluorescent microscopic images showing vehicle- or rapamycin-treated mouse brains on POD 11 (prior to treatment) and POD 28 (after treatment). (C) Representative in vivo time-lapse fluorescent microscopic images showing the overall lesional area (red dotted contour) and no-flow area (yellow dotted contour) at t=10 min following FITC-Dextran administration. White arrows indicate FITC-Dextran-negative area within the lesions. (D, E) Quantification of the percentage of no-flow area within the CCM lesions in (D) vehicle- and (E) rapamycin-treated mouse brains following a 17-day treatment. Vehicle, n = 6; Rapamycin, n = 8. Data shown are means ± SEM. *P=0.0285 by unpaired two-tailed Welch’s t-test.

Low-dose oral rapamycin slows CCM lesion growth

The findings presented above demonstrate that when administered intra-peritoneally Rapamycin is able to prevent the growth of aggressive CCM lesions in mice. Clinical use of Rapamycin would entail oral administration at doses that could be tolerated for long periods of time. To more thoroughly test the value of Rapamycin for clinical treatment of CCMs, we next assessed oral Rapamycin treatment using the inducible mouse model. Following injection with AAV-Cre, animals with a CCM lesion size between 0.5 mm and 1.0 mm in diameter were randomly assigned to receive either Rapamycin (2.3 mg/kg/day) or vehicle (the equivalent amount of drug-coating polymers) in added to their chow after an intraperitoneal loading dose (Rapamycin 3.3 mg/kg or vehicle) (Figure 6A). In the vehicle treated control group CCM lesions grew rapidly and exceeded the observational cranial window within 10 days. Of note, 2/7 mice in the control group died within a week in association with rapid CCM lesion growth, and one was euthanized in the first week due to severe neurological deficits (Figure S6A–B). In contrast, CCM lesion sizes in mice receiving the Rapamycin diet remained small (Figure 6B, Figure S6A), despite the observation that the average of daily Rapamycin chow intake dropped slightly over time (Figure S6C). Visual and microCT analysis of post-mortem brains on post-treatment day 17 revealed significantly reduced the CCM lesion size in animals that received a low-dose Rapamycin diet compared with mice in the group receiving a placebo diet (Figure 6C–D, Figure S7). These findings support those reported above and suggest that treatment with oral Rapamycin at doses known to be tolerated for prolonged periods of time is likely to decrease lesion growth. Although our findings did not show statistically significant reductions in lesion volumes, we did observe a significant reduction in death and severe neurological deficits due to CCM progression (Figure S6B).

Figure 6. Oral rapamycin arrests CCM lesion growth in adult CCM mouse model.

(A) Schematic representation of adult Krit1^fl/fl; iPik3ca^H1047R mice between age 8–10 weeks were undergo craniotomy and receive AAV-Cre injection to induce Krit1 deletion and Pik3ca^H1047R over expression. Mice with CCM lesion size between 0.5 mm and 1.0 mm at any given PODs were enrolled in the study. Enrolled mice were received either rapamycin (3.3 mg/kg) or vehicle (5% DMSO/PBS) on the first day (PTD 0) as a loading dose. Thereafter, mice were feed with rapamycin diet or placebo for 17 continuous days. (B) A serial of microscopic images through the cranial windows showing CCM growth in either vehicle- or rapamycin-treated mouse brains from PTD 0 to PTD 17. (C) Representative visual images and (D) microCT quantification of CCM lesion volumes of placebo- and rapamycin-diet mouse brains collected on PTD 17. Placebo, n = 6; Rapamycin, n = 8. Data shown are means ± SEM. P=0.1249 by unpaired two-tailed Welch’s t-test.

Discussion

The only existing therapeutic approach to patients with symptomatic CCM disease is neurosurgical resection²⁶. While surgery is often curative, there remain large numbers of patients who are not surgical candidates due to the location of their lesions or the presence of numerous lesions. Recent insights into the molecular and genetic basis of CCM disease have revealed a critical role for increased PI3K-mTOR signaling^14,16, consistent with a pathogenesis similar to that of other venous and lymphatic malformations²⁷. Our prior studies in mice and early clinical trials for other venous and lymphatic malformations suggest that Rapamycin could be an effective medical treatment for CCM disease^14,28,29. In the present study we take advantage of a new mouse model that more closely reproduces both the genetic and morphologic features of human disease to test this hypothesis. Our findings demonstrate that Rapamycin, an FDA-approved agent with significant clinical experience can virtually arrest the growth of pre-existing, aggressive CCM lesions and reduce the extent of lesional hemorrhage. These studies lay the foundation for the use of Rapamycin and similar agents to medically treat human CCMs.

The most important finding in this study is that Rapamycin treatment is able to arrest the growth of CCMs that exhibit the characteristics of aggressive human sporadic lesions when administered orally at doses similar to those currently used for treatment of other human diseases, such as arteriovenous malformations (AVM), lymphangioleiomyomatosis, and complex vascular malformations³⁰. We believe that these findings are likely to be directly translated to treatment of human CCM disease for several reasons. First, our adult inducible model generates CCM lesions that genetically and morphologically mirror aggressive sporadic human CCMs to a greater extent than prior mouse and fish models that rely exclusively upon CCM LOF. Second, the rationale for Rapamycin use to arrest CCM lesion growth arises directly from recent discoveries that demonstrate a two-hit mechanism for the disease involving both CCM LOF and PIK3CA GOF. Whether Rapamycin is similarly effective for CCM lesions that lack PIK3CA GOF mutations is not yet clear, but previously published studies using the neonatal mouse model in which the only mutation is loss of CCM function suggests that this is likely¹⁴. Finally, the effect of Rapamycin observed in this model is large, consistent with impact on a core mechanistic pathway. While numerous other medical therapies to treat CCM disease have been proposed, including statins and beta blockers^31,32, these do not carry a similarly high level of evidence supporting efficacy. Future studies using this recently developed adult mouse model of CCM disease may directly compare the efficacy of the existing therapeutic options.

There exist several significant limitations of our study. First, while our Rapamycin treatment studies clearly demonstrated CCM lesion growth arrest they failed to demonstrate lesion regression. This negative result must be considered in light of (i) the relatively short duration of our studies, and (ii) the fact that we observed poor perfusion of nascent CCM lesions, a condition that would likely reduce drug delivery to endothelial cells already present in pre-existing lesions. Longer duration of treatment and greater opportunity for drug delivery, both in the mouse model and in human patients, might have effects not observed here. Second, while we observe reduced peri-vascular iron and improved perfusion of CCM lesions with Rapamycin treatment these changes are coincident with a dramatic reduction in lesion growth. Thus effects on vascular leak and reduced blood flow may simply reflect the effect on lesion growth and not translate to human lesions that are already sizable at the time of discovery. Finally, our mouse model generates rapidly growing lesions that can only be quantitatively measured when they are in an early stage of lesion formation. Whether Rapamycin will be equally effective for later stage lesions, especially those in which blood flow may already be compromised, is not known. These limitations are not unexpected for a mouse model that must balance accurate recapitulation of the disease with rapid onset and high throughput to enable testing under multiple conditions. In our mouse model large numbers of brain endothelial cells simultaneously express Cre after injection and exhibit both CCM LOF and PIK3CA GOF. In contrast, in human disease it is likely that an individual brain endothelial cell with CCM LOF and PIK3CA clonally expands in a slower disease time course. Clinical studies assessing Rapamycin effects in human patients are expected to yield more definitive answers to the questions of lesion regression and hemorrhage.

A final question raised by these studies is how to proceed with translation to human CCM treatment. We tested Rapamycin in mice orally at a low dosage of 2.3 mg/kg/day, which is projected to achieve a target trough level of 5–15 ng/mL in the mice³⁰. The oral Rapamycin dosage in our study is predicted to be equivalent to human trials with low daily oral doses of <2 mg/day with a blood concentration of <15 ng/mL. Since Rapamycin is already widely clinically used and the doses used in this study are in the range of those administered to patients, it is reasonable to initiate a clinical trial of Rapamycin treatment for rapidly growing sporadic lesions likely to have both CCM LOF/MAP3K3 GOF and PIK3CA GOF that are not amenable to surgical excision.

Alternatively, since clinical trials are already underway for statin (ClinicalTrials.gov Identifier: NCT02603328) and beta blocker (ClinicalTrials.gov Identifier: NCT03474614, NCT03589014) treatment of CCM disease, one could use the adult inducible mouse model to compare the efficacy of these agents in a pre-clinical model. In addition, new PIK3CA inhibitors have been developed that are already being used to treat similar venous malformations (ClinicalTrials.gov Identifier: NCT05577754, NCT03941782). Such agents may offer more complete blockade of the downstream PI3K pathway and therefore be more effective than Rapamycin, but at the present time there is less clinical experience with these agents and their costs are much higher than that of Rapamycin. Future studies that include both further studies in the mouse pre-clinical model and human trials are likely to be needed to define the most impactful and cost-effective treatment for non-resectable CCM.

Non-standard Abbreviations and Acronyms

AAV

adeno-associated virus

CCM

cerebral cavernous malformation

DAPI

4′,6-diamidino-2-phenylindole

DMSO

dimethylsulfoxide

EMCN

endomucin-1

ERG

ETS-related gene

FITC

fluorescein isothiocyanate

GFAP

Glial fibrillary acidic protein

GOF

gain-of-function

Iba-1

ionized calcium binding adapter molecule-1

IP

intraperitoneal injection

LOF

loss-of-function

PMMA

polymethyl methacrylate

POD

post-operative day

PTD

post-treatment day

SEM

standard error of the mean

TCUS

transcranioplasty ultrasound

vWF

Von Willebrand factor