Trametinib as a promising therapeutic option in alleviating vascular defects in an endothelial KRAS-induced mouse model

Abstract

Somatic activating Kirsten rat sarcoma viral oncogene homologue (KRAS) mutations have been reported in patients with arteriovenous malformations. By producing LSL-Kras (G12D); Cdh5 (PAC)-CreERT2 [iEC-Kras (G12D^*)] mice, we hoped to activate KRAS within vascular endothelial cells (ECs) to generate an arteriovenous malformation mouse model. Neonatal mice were treated daily with tamoxifen from postnatal (PN) days 1–3. Mortality and phenotypes varied amongst iEC-Kras (G12D^*) pups, with only 31.5% surviving at PN14. Phenotypes (focal lesions, vessel dilations) developed in a consistent manner, although with unpredictable severity within multiple soft tissues (such as the brain, liver, heart and brain). Overall, iEC-Kras (G12D^*) pups developed significantly larger vascular lumen areas compared with control littermates, beginning at PN8. We subsequently tested whether the MEK inhibitor trametinib could effectively alleviate lesion progression. At PN16, iEC-Kras (G12D^*) pup survival improved to 76.9%, and average vessel sizes were closer to controls than in untreated and vehicle-treated mutants. In addition, trametinib treatment helped normalize iEC-Kras (G12D^*) vessel morphology in PN14 brains. Thus, trametinib could act as an effective therapy for KRAS-induced vascular malformations in patients.

Introduction

Blood vessels are formed and maintained by a tightly regulated process called angiogenesis. When disruptions occur, developmental defects known as vascular malformations can form. Vascular malformations are classified based on the affected vessel type; they can be singular entities or be part of a syndrome (1).

Arteriovenous malformations (AVMs) are vascular malformations in which an intervening capillary bed is absent between arteries and veins. As a result, the fast-flowing arterial blood shunts directly into the usually slow-flowing veins, often through a so-called ‘nidus’ of intermingling vessels; the vessels enlarge and/or become grossly disorganized and tortuous. Eventually, the vessels become fragile and susceptible to rupturing. AVMs can be found throughout the body, and symptoms can vary greatly (2). The severity of complications can range from mild (e.g. headaches) to severe (e.g. hemorrhagic). In most cases, an AVM is not discovered until symptoms arise that prompt patients to seek medical help. Hence, the incidence of AVMs in the population is reportedly rare, at around 0.2% (3). The most devastating effects of AVMs are induced when they are found within the brain. Hemorrhagic stroke is the most frequent (and lethal) sign of an AVM and is a leading cause of strokes in younger patients. The main methods to treat AVMs are surgical resection and embolization (3). However, given the fragile nature of lesions, an effective medicinal option would be optimal, but none exists.

The etiopathogenesis of AVMs is not well understood. The majority of cases occur spontaneously. However, AVMs are associated with certain rare disorders, notably as a hallmark in the autosomal dominant disorders hereditary hemorrhagic telangiectasia (HHT) and capillary malformation-AVM (CM-AVM). The identification of the genetic causes for these diseases has implicated two major signaling pathways for AVM: the transforming growth factor-β and the mitogen-activated protein kinase (MAPK) pathways. The former is due to loss-of-function (LoF) mutations in endoglin, activin receptor like kinase-1 or SMAD4; notably, the BMP9/10 ligand is essential. The latter is caused by LoF mutations in the EPHB4 receptor or the RAS GTPase p120RASGAP encoded by RASA1 (4–8). Even if the RASA1 mutations per se cause loss of implicated protein function, the underlying effect seems to be activation of the RAS-MAPK pathway in CM-AVM.

The essential contribution of unregulated endothelial intracellular MAPK signaling in AVM formation was supported by findings of somatic genetic mutations in several RAS/MAPK signaling associated genes in sporadically occurring AVMs. Activating mutations in BRAF, MAP2K1, HRAS and KRAS have been reported, predominantly in intracranial cases (9–14). Several of the hotspot mutations found in BRAF and KRAS [e.g. Braf (p.V600E), KRAS (p.G12D) and KRAS (p.G12V)] are the same ones observed in various types of cancers. Most sporadically occurring AVMs (one- to two-thirds) are attributed to KRAS changes, specifically within the endothelium. KRAS is a ubiquitously expressed GTPase that acts as a molecular switch to activate multiple signaling pathways, the most predominant being MAPK signaling. Initial in vitro studies in which KRAS4A (G12V) was overexpressed in human umbilical vein endothelial cells (HUVECs) resulted in abnormal endothelial cell (EC) morphology and disassembly of vascular endothelial cadherin junctions that were alleviated by MEK inhibition (14). This finding supports the notion that targeting the MAP/MEK signaling pathway can provide viable pharmacological options for AVM treatment. However, a more physiologically relevant model, such as a mouse, would be needed for preclinical trials.

To confirm that the MAPK/ERK signaling pathway plays an important role in regulating angiogenesis in ECs in vivo, we generated a mouse model in which KRAS activation was directed within ECs. We observed the development of vascular lesions and consequently tested whether an MEK inhibitor was able to alleviate these vascular defects. The promising results support MEK inhibition as a potential therapeutic option for AVM patients.

Results

Postnatal induction of endothelial KRAS in mice results in significant early mortality and formation of vascular defects

Mortality and phenotypes varied amongst iEC-Kras (G12D^*) pups; more than 50% of iEC-Kras (G12D^*) mice died by postnatal (PN) 12 (Fig. 1B). The iEC-Kras (G12D^*) mice that were sick and/or predicted to be close to death exhibited weight loss or no weight gain over 2 days, excessive trembling, lethargy and occasional labored breathing. Otherwise, iEC-Kras (G12D^*) mice were indistinguishable from control littermates. The maximum age to which surviving mice were evaluated was 45 days, as it was observed that iEC-Kras (G12D^*) developed lung tumors, which could confound the present study’s focus on vasculature. Development of cancerous tumors was expected since KRAS (G12D) is a well-known hotspot mutation for several types of cancers.

Figure 1

Endothelial overactivation of Kras in neonatal mice reduced proper development and life expectancy. (A) Schematic for generation of iEC-Kras (G12D^*). Neonates are given three daily intragastric injections of tamoxifen from PN days 1–3. (B) Kaplan–Meier survival curve of revealed that more than 50% of iEC-Kras (G12D^*) pups died by PN12. N = 35 controls, n = 43 mutants (P < 0.001). A chi-squared was used to confirm significance at PN12. (C) Representative macroscopic images of vascular phenotypes seen in visceral organs at different time points. Instead of uniform tubes, affected vessels could be of variable diameters as seen a PN11 intestine (arrows). Also, in a PN11 intestine, a cluster of disorganized and tortuous micro-vessels (x). In a PN13 heart, a vascular bed included dilated and irregularly shaped vessels. In a PN23 liver, a large lesion is visible. The maximum age of observance was PN45. Besides the formation of vascular defects in several visceral organs, iEC-Kras (12GD^*) mice had grew lung tumors (^*). Although not all sickly iEC-Kras (G12D) mice were affected, chylothorax was observed (chest cavity image).

Generally, normal superficial vessels appear as uniform, tubular structures with smaller vessels branching from larger ones on organs, such as the intestine and heart; on other organs (e.g. the liver and brain), the vessels are not obvious (Fig. 1A, controls). Necropsy revealed gross vascular defects in various organs, such as the brain, lungs, heart, liver and intestine of ill iEC-Kras (G12D^*) mice collected at different stages (Fig. 1C). However, the location and severity of vascular defects varied and were unpredictable. Hemorrhaging was rarely observed in any iEC-Kras (G12D^*) mice (2/53 mice). In older surviving iEC-Kras (G12D^*) mice, phenotypes were more pronounced (Fig. 1C). Chylothorax was seen in 11/43 (~25%) iEC-Kras (G12D^*) neonates. However, since the focus of the study was on the blood vasculature, and not the lymphatic system, we did not investigate this observation further. We never observed chylous ascites in this study.

Phenotypes in iEC-Kras (G12D^*) pups develop by PN8

We further characterized the vascular defects within the iEC-Kras (G12D^*) mice at PN8 and PN16 (Fig. 2A). Mild gross focal vascular defects were visible in iEC-Kras (G12D^*) pups as early as PN8. Though the severity was variable, phenotypes included noticeably dilated vessels, irregular vascular tubes and additional branching of vessels from major vessels. In all, no abnormal vessels were observed in control mice (n = 10). For iEC-Kras (G12D^*) mice (n = 10), three exhibited dilated or abnormal vessels within the lung, four in the intestine and two in the liver (Fig. 2B). A lesion, in which a single vessel developed a bulge (indicative of a weakened vessel wall), was variably observed on the surface of one brain (Fig. 1B). Two of the 10 iEC-Kras (G12D^*) mice did not display any overt defects. The number of retinal branch points was not significantly different between iEC-Kras (G12D^*) neonates and control littermates, according to the retinal angiogenic assay (Fig. 2C).

Figure 2

Phenotypes were observed starting at PN8 and worsened over time in surviving PN16 iEC-Kras (G12D^*) pups. (A) Schematic for experimental approach. Pups were collected and analyzed at PN8 and PN16. (B) Representative images of mild malformations found in iEC-Kras (G12D^*) pups [aged PN8], such a superficial focal lesion in the brain, dilated vessels in the intestine and hemorrhaging in the liver. (C) Retina angiogenic assay on PN8 pups revealed that there were slightly less branch points in iEC-Kras (G12D^*) mice, compared with control littermates. However, no direct connections between large vessels (A, arteries; V, veins) were seen (n.s., not significant), determined by Student t-test. Scale bar = 200 μm. (D) Macroscopic vascular defects were more pronounced in surviving iEC-Kras (G12D^*) pups, at PN16. To note, iEC-Kras (G12D^*) organs that did not develop gross vascular defects appeared similar to control organs.

Vascular lesions appear to worsen over time in surviving iEC-Kras (G12D^*) pups

To track the progression of lesions, phenotypes were also evaluated at a later stage, at PN16 (Fig. 2A). There were no obvious vascular defects in any of the 10 PN16 control mice. However, the formed vascular defects appeared to be aggravated in iEC-Kras (G12D^*) mice, as demonstrated in various collected soft tissues (Fig. 2E). Gross focal lesions remained unpredictable; grossly dilated vessels resembling AVMs were consistently visible in all iEC-Kras (G12D^*) mice, but the most severely affected visceral organ (e.g. brain, heart, lung, intestine and liver) was variable. Otherwise, unaffected iEC-Kras (G12D^*) organs appeared normal. Of the 10 PN16 iEC-Kras (G12D^*) observed, all had at least one defect (from noticeably dilated vessels to clusters of unusual vessels to malformed lesions) visible for superficial phenotyping analyses. Lesions were found in one brain, two hearts, five intestines, four livers and seven lungs. When iEC-Kras (G12D^*) mice were compared with control littermates, histological examination of vessels in the brain, heart, intestine and liver revealed that many vessels were dilated (Fig. 3A and B). At PN8 and PN16, the vessels of iEC-Kras (G12D^*) mice were significantly larger than those of control littermates, as measured by the average vessel lumen area within each section (Fig. 3C and D).

Figure 3

Overall vascular lumen areas were larger in all organs evaluated in iEC-Kras (G12D^*) mice. EC stained with IsoB4 within the brain, heart, intestine and liver from PN8 (A) and PN16 (B) pups. The overall average vascular channels were significantly larger in iEC-Kras (G12D^*) mice in all organs at both PN8 and PN16, compared with control littermates. PN8: control, n = 10; iEC-Kras (G12D^*), n = 10. PN16: control, n = 6–10; iEC-Kras (G12D^*), n = 10. Scale bar = 100 μm. For statistical significance: ^*, P ≤ 0.05; ^**, P ≤ 0.01; ^***, P ≤ 0.001; this was determined by Student t-test.

ECs express phosphorylated-ERK

To confirm that KRAS was activated within ECs, immunofluorescent staining of various organs for phosphorylated (p)-ERK expression was performed. The pERK expression was very low to negative in control littermates at PN8 (Fig. 4). Conversely, strong expression was seen in iEC-Kras (G12D^*) pups, with the expression pattern corresponding to that of IsoB4, indicating that pERK expression was specific to ECs. Similar findings were observed in tissue from PN16 mice, suggesting that the enhanced pERK expression persists within the ECs. Thus, the formation of vascular defects in this mouse model seems to be due to excessive endothelial KRAS/MAPK signaling. Quantification of pERK expression levels within PN12 brains, livers and intestines confirmed that sustained pERK expression was significantly higher within all organs of iEC-Kras (G12D^*) mice compared with control littermates.

Figure 4

Immunofluorescence staining for pERK revealed highly sustained pERK expression predominantly in ECs. Costaining of IsoB4 (orange) and pERK (green) of brain, liver and intestine from PN8 and PN16 pups. In control pups, pERK expression is very low in and largely negative in all organs at both PN8 and PN16. In iEC-Kras (G12D^*) pups, pERK expression is high, particularly within ECs at PN8. The high pERK expression appears to be sustained in PN16 iEC-Kras (G12D^*) mice. B, bile duct; A, artery^*, middle of large vascular channel. Scale bar = 20 μm. Quantification of pERK within PN12 brains, liver and intestine. N = 8 for each control and mutant groups and each organ. For statistical significance: ^**, P ≤ 0.01; this was determined by a Wilcoxon test for the brain and Student t-tests for the intestine and liver.

Treatment of iEC-Kras (G12D^*) pups with trametinib improved survival and helped normalize defective vessels

To test whether an MEK inhibitor could effectively treat established vascular defects associated with hyperactive MAPK signaling, we treated the iEC-Kras (G12D^*) pups with trametinib (Mekinist, GSK1120212). More specifically, we were interested in whether trametinib would be capable of alleviating the progression of established vascular lesions (as when treating already established lesions in patients), rather than blocking lesion formation. As vascular defects were observed by PN8, trametinib treatment was started at this time point (Fig. 5A). All control mice, regardless of treatment group, (non-, vehicle- or trametinib-treatment), survived the period of observation (up to PN45), with no distinct phenotypes. Thus, they were grouped together for further analyses. The survival rate (21.5%) of vehicle-treated iEC-Kras (G12D^*) pups was like that of the untreated mice (31.5%) by PN14 (Fig. 5B). However, trametinib treatment improved the survival of iEC-Kras (G12D^*) mice to 76.9% at the same time point (Fig. 5B).

Figure 5

Trametinib-treatment of iEC-Kras (G12D^*) mice improved survival and helped control lesion progression. (A) Experimental schematic of trametinib treatment regimen. Trametinib was delivered to pups via trans-mammary transfer. Lactating mothers were given 1 daily dose of vehicle or 2 mg kg⁻¹ trametinib via oral gavage, from aged PN8–12. Pups were then collected and evaluated at PN16. (B) When left until PN45, trametinib treatment improved survival of iEC-Kras (G12D^*) mice, compared with untreated and vehicle-treated iEC-Kras (G12D^*) mice. The data for the untreated iEC-Kras (G12D^*) mice were the same as presented in Figure 1B. Control mice data consist of all controls observed from all treatment groups since all mice survived. A chi-squared comparison between the trametinib-treated and each of the untreated and vehicle-treated iEC-Kras (G12D^*) mice were done to confirm significance at PN12. (C) Representative macroscopic images of various organs from un- and treated PN16 control and iEC-Kras (G12D^*) pups. Vehicle-treated iEC-Kras (G12D^*) pups developed defective vessels akin to untreated iEC-Kras (G12D^*) pups. Though vascular defects were visible in trametinib-treated iEC-Kras (G12D^*) mice, these were less severe than those in the other treatment groups. All vehicle-treated and trametinib-treated control and iEC-Kras (G12D^*) groups, n = 6 each. Arrow and ^*, abnormal vessels clusters in the heart and intestine, respectively; +, dilated vessels in the intestine.

In PN16 control [and unaffected iEC-Kras (G12D^*) vascular beds] brain, heart and intestine, normal large, superficial vessels appear as uniform tubes that gradually branch off into smaller tubes. However, in the liver and lung, vessels on the distal portions of the lobes are microvascular and not visible; thus, the appearance of any blood-filled vessels in these areas denotes a malformation, as represented in the iEC-Kras (G12D^*) mice (Fig. 5C). A comparison of various soft tissues of mice collected at PN16 showed that the vascular defects that developed in vehicle-treated iEC-Kras (G12D^*) mice were similar to those of the untreated mutant mice in the brain, heart, intestine and liver. All six vehicle-treated iEC-Kras (G12D^*) mice observed exhibited at least one vascular defect, such as visibly dilated vessels and abnormal vessel clusters, in the same soft tissue as the untreated group: lesions were found in one brain, four hearts, four intestines, five livers and two lungs. Hemorrhaging was noted in one lung. However, the vascular defects in iEC-Kras (G12D^*) mice treated with trametinib were milder, albeit still visible (Fig. 5C). Lesions were found in three out of six trametinib-treated iEC-Kras (G12D^*) mice: in one lung, two intestines and three livers. Quantification of the average vessel lumen area within IsoB4-stained histological sections revealed that all iEC-Kras (G12D^*) treatment groups had a larger average vessel size than the control mice, although, trametinib-treated samples were not significantly so within the brain, heart and intestine (Fig. 6). Within the liver, there was no difference in the average vascular lumen area in iEC-Kras (G12D^*)-treated pups. Thus, trametinib efficacy in this organ is not clear (Fig. 6).

Figure 6

Average vessel lumen areas of trametinib-treated iEC-Kras (G12D^*) pups were significantly smaller than un- and vehicle-treated iEC-Kras (G12D^*) pups at PN16. ECs of PN16 brain, heart, intestine and liver were stained with IsoB4 and overall vessel sizes were compared between the control and iEC-Kras (G12D^*) mice subjected to the three treatments. No difference in sizes was measured among control mice in response to treatment, and thus these have been combined as one group. For all groups, n = 6. Scale bar = 100 μm. For statistical analysis, an analysis of variance was carried out for the brain and intestine samples and a Kruskal–Wallis test for the heart and liver samples. Subsequently, a Tukey test was done. Adjusted P-values shown; n.s., no significance.

To visualize the effects that trametinib could have on the morphology of cerebral vessels, whole mount staining of 200-μm PN14 brain slices with isolectinB4 was performed and a three-dimensional (3D) reconstruction of vessels was made from acquired z-stacks. PN14 brains were the oldest age evaluated because of the low survival rate of untreated iEC-Kras (G12D^*) mice. In the control brain, vessels were uniform tubular structures and clear connections between branches could be seen. In untreated iEC-Kras (G12D^*) brains, dilated vessels were visible and vessels branching from the enlarged vessels were of variable size and appeared jagged. A side view of the vessels confirmed that the vessels were enlarged. In trametinib-treated mice, cerebral vessels appeared to be normalized. A side view of the slice exhibited clear vascular tubes, as seen in control mice. There was still a detectable variability in shape along some vessels (Fig. 7).

Figure 7

Whole mount staining of cerebral vessels reveals that trametinib was able to normalize iEC-Kras (G12D^*) vessels. Representative 3D reconstructions of two equivalent areas comparing untreated (UNT) and trametinib-treated control and iEC-Kras (G12D^*) IsoB4-positive cerebral vessels cerebral vessels at PN14. Morphology of vessels in iEC-Kras (G12D^*) mice appear enlarged and irregular (arrowhead). Although some vessels in trametinib-treated iEC-Kras (G12D^*) mice were still irregular, most vessels were more uniform, as similarly seen in control littermates. Most images are of the top view of the reconstruction; the third row of images is a side view of the slice indicated by the white box in the first-row images. Scale bar = 100 μm. N = 6 pups per treatment group, 1–2 brain slices each; 4–6 different comparable areas were observed in each slice. For statistical analysis, Kruskal–Wallis test was done; significance: n.s., P > 0.05; ^*, P ≤ 0.05; ^**, P ≤ 0.01.

Discussion

Most AVM patients possessing a somatic activating mutation in KRAS have a sporadically occurring cerebral AVM (11,12,14–17). However, somatic KRAS, BRAF and MAP2K1 mutations have also been identified in peripheral AVMs (9,10,12,17). Moreover, AVMs are a prominent feature in the hereditary disorder CM-AVM, in which LoF of either of the two negative regulators of RAS/MAPK signaling, RASA1 or EPHB4, leads to the development of fast-flow lesions (4,6). Global ablation of Rasa1 or EphrinB4 in mice results in embryonic lethality due to cardiac and angiogenic defects (18–20); in addition, conditional ablation results in defects in the blood and lymphatic vessel systems (21–23). As lesions can be found throughout the body, an important generalized role for endothelial MAPK signaling is indicated.

We presented a murine model in which Kras activation was directed within all vascular ECs using a Cre recombinase driven by the cadherin5 (Cdh5) promoter. Contrary to several recent endothelial (K/H) ras-induced mouse models that reported only brain AVM and hemorrhaging (24–26), our iEC-Kras (G12D^*) pups developed vascular defects (such as dilation and focal lesion formation) within soft tissues throughout the murine body as well. Fish and coworkers induced endothelial Kras activation in PN mice using the same mouse lines (24). Our approach differs in the administration of tamoxifen (a single subcutaneous injection at PN1 vs three daily intragastric injections from PN1–3). This likely leads to variable amounts of circulating tamoxifen and different levels of tissular mosaicism for the activation of the mutant Kras transgene. In addition, we performed a comprehensive analysis of various organs, instead of focusing on the cerebrovasculature. In agreement with Fish et al.’s Kras model, a significant percentage of iEC-Kras (G12D^*) mice died by PN12. However, not many details were provided, as the focus of their study was elsewhere. One major noted phenotypic difference reported in that study was an incomplete penetrance of intracranial hemorrhaging in surviving PN21 mutant mice (24); we never observed hemorrhaging. In our mouse model, lesions appeared consistently, but their localization was unpredictable. The main organ affected, the severity of defects and mortality varied amongst our Kras mutant mice. Overall, our results confirm that endothelial KRAS/MAPK signaling plays an important role in angiogenesis throughout the murine vascular system in vivo.

AVMs have a reported prevalence of about 1:100 000, with more than 50% of cases found within the head and neck (27,28). The high frequency of focal vascular defects observed in the iEC-Kras (G12D^*) mice implies that small, sporadic RAS/MAPK-associated visceral AVMs may occur much more frequently within the population than believed. This may not be too surprising as extracranial AVMs are commonly found in HHT (29). The frequency of KRAS-associated AVMs in patients likely also depends on the frequency of somatic activating mutations to occur in KRAS pathway genes (which may not be that high) and the eventual limited ‘window of opportunity’ during the development for those mutations to cause lesions. A significant percentage of our iEC-Kras (G12D^*) mice with PN induction of Kras (G12D) expression died by PN12. This was similar to mice in which Kras (G12D) was induced specifically in brain ECs using the Cre-driver line Slc1o1c-CreER (24), which developed vascular lesions similar to our model. Although we do not know the reason for death in our mice, the comparable time point in the two lines suggests an analogous mechanism (24,25).

Genetic studies on patients with sporadically occurring vascular malformations suggest that fast-flow malformations (such as AVM) are associated with enhanced RAS–RAF-MAPK signaling, while slow-flow malformations are associated with increased PI3K-AKT–mTOR signaling (30). Therefore, testing of drugs that have been developed for oncology to specifically inhibit one of the proteins activated in these two pathways, has become a new area of intense research. As a proof of principle, the general mTOR inhibitor rapamycin has proven to be effective on slow-flow malformations (31,32).

Current treatment options for fast-flow lesions are largely invasive, including surgical resection or destruction by embolization. However, due to the highly active nature of the ECs, AVMs often regrow. Additionally, given the fragile and random nature of AVMs, interventional options may even worsen patients’ situations, as suggested by the ARUBA trial (33,34). There have been no widely accepted medicinal options. A few antiangiogenic drugs have been used to treat AVM by restricting vessel growth, particularly in HHT. Bevacizumab is a monoclonal antibody that targets VEGF; when given intravenously, it has been shown to effectively reduce nosebleeds and gastrointestinal bleeding (35). Thalidomide has also been reported to alleviate hemorrhaging in both HHT patients and in different HHT mouse models; its mode of action appears to promote vascular maturation by stimulating mural cell coverage of nascent vessels (36,37). A recent report on 18 cases suggested that thalidomide could also be used for sporadic AVMs (38). However, both of these drugs have known adverse side effects, so their use is limited to severe cases in which patients are not responsive to alternative therapeutic approaches. As patients with AVM typically needed long-term/life-long treatment, the safety of such a treatment regime for both drugs needs to be assessed (38,39).

Rapamycin (in combination with nintedanib) and PI3K inhibitors have had only mild success in the treatment of AVMs and HHT, in patients or in mouse models (40,41). Moreover, PI3K inhibitor treatment of mice in which Hras was conditionally turned on within ECs, worsened hemorrhaging and vascular malformations, despite in vitro data suggesting that it should alleviate the phenotype (25). This further confirms that fast-flow lesions are affected more specifically by MAPK signaling. This was corroborated by gene expression analysis of HUVECs transfected with KRAS (G12V); when treated with MEK or PI3K inhibitors, MAPK, but not PI3K, signaling contributed to phenotypic and molecular alterations (21). In our iEC-Kras (G12D^*) mouse model, pERK expression was clearly increased within ECs; however, immunofluorescence staining of pAKT remained negative (data not shown). Thus, it is crucial to select a proper inhibitor based on the affected pathway when projecting to treat a vascular anomaly by a medicinal approach.

Trametinib is a potent dual-kinase inhibitor with a long half-life (of about 5 days) that targets MEK1 and MEK2. Thus, it may be used to dampen overt RAS-MAPK–ERK signaling. Clinically, it is used to treat metastatic melanoma and non-small cell lung cancer in patients with a BRAF (V600) mutation, in combination with the BRAF inhibitor dabrafenib. Trametinib treatment of our iEC-Kras (G12D^*) neonatal mice was efficacious. It resulted in smaller vessel lumen sizes compared to non- and vehicle-treated groups, with the average vessel lumen area comparable to control mice. Our studies are in agreement with a recent study in which prolonged trametinib treatment in a KRAS (G12V)-associated brain AVM mouse model successfully reduced the numbered of large lesions (26). Together, Park et al.’s preventative and our curative in vivo studies suggest that trametinib could potentially be used to treat AVMs in human patients, underscoring the need for clinical trials.

Trametinib had variable efficacy as treatment of lesions in our iEC-Kras (G12D^*) model. This may have various explanations. In the liver, iEC-Kras (G12D^*) vascular lumen area sizes differed greatly regardless of treatment, resulting in no significant differences seen amongst all groups. Thus, the effect of trametinib is uncertain in the liver. Moreover, since trametinib was delivered via trans-mammary transfer (from lactating dames) the exact amount of drug received by each pup cannot be calculated and is likely inconsistent. Even if the vascular defects were not completely resolved, the enhanced life span shows that there was enough trametinib transferred to have an effect.

This study suggests that we can confidently hone in on MEK inhibitors to treat AVM patients, especially those in whom a causative mutation in RAS/MAPK signaling pathway-associated genes have been found. Other studies support the effectiveness of this approach. In two BRAF-associated AVM zebrafish models, treatment with the BRAF inhibitor vemurafenib (PLX4032) was able to restore defected blood flow (9). In addition, MEK inhibition in a KRAS (G12V)-induced AVM zebrafish model led to a partial reduction in brain hemorrhaging and resolution of AV shunts, while PI3K inhibition did not (24). Moreover, recent case reports have shown promising results of trametinib treating extracranial AVMs in patients with known mutations, such as a child with a MAP2K1 in-frame deletion and a Cobb syndrome patient with a KRAS in-frame tandem duplication (42,43). A third teenage CM-AVM2 patient (with a confirmed pathogenic EPHB4 mutation) also responded positively to trametinib for a lesion in her thigh (44). Thus, trametinib, and likely other MEK inhibitors, have the potential in being an effective therapy for fast-flow vascular malformations, particularly extracranial ones, akin to the findings of our mouse model. Two clinical trials have been initiated in the US [ClinicalTrials.gov: NCT04258046], and by our group (on 10 AVM patients) in Europe [TRAMAV in July 2019 (EudraCT: 2019–003573-26)].

Materials and Methods

Generation of mice with endothelial-directed hyperactive KRAS

All animal procedures were performed according to protocols approved by the Ethical Committee for Animal Experimentation at the health care sciences sector of UCLouvain (Protocol 2018/UCL/MD/02; Brussels, Belgium). Two well-characterized transgenic mouse lines, on a mixed background, were used. Neonatal mutant mice were produced by breeding LoxP-STOP-LoxP (LSL)-Kras (G12D)-floxed mice with Cdh5 (PAC)-CreERT2, to produce bigenic LSL-Kras (G12D); Cdh5 (PAC)-CreERT2 offspring (45,46). Females were separated from males in the morning when a copulation plug was seen and allowed to deliver the litter to term. Both male and female experimental pups were used. Mutant mice were produced by treating LSL-Kras (G12D); Cdh5 (PAC)-CreERT2 [noted as iEC-Kras (G12D^*)] by daily gastric injections of 50 μg tamoxifen (dissolved in peanut oil) at PN days 1–3 (PN1–3) (Fig. 1A). Tamoxifen-injected cre-negative and LSL-Kras (G12D)-floxed littermates were used as controls. Pups were monitored daily and weighed every 2 days. Macroscopic images of neonates were taken with a Nikon Coolpix 4500 4MP digital camera. Genotype polymerase chain reaction was performed on all mice using the following primers: Kras, F1: GTCTTTCCCCAGCACAGTGC; Kras (G12D), F2: AGCTAGCCACCATGGCTTGAGTAAGTCTGCA and Kras, R: CTCTTGCCTACGCCACCAGCTC. For the Kras WT band, primers F1 + R were used to detect a 622-bp band. To detect the LSL cassette, primers F2 + R were used to generate a band size of 500 bp. For cre, primers: F1, GATCTCCGGTATTGAAACTCCAGC and R1, GCTAAACATGCTTCATCGTCGG, band size 650-bp were used. For the Kaplan–Meier curve, mice were humanely culled when exhibiting specific morbid phenotypes (no weight gain after 2–4 days, lethargy, excessive trembling, difficulty in breathing). To note, all n numbers mentioned in the figures correspond to each independent experiment described.

Administration of trametinib

For minimal handling and stress, administration of trametinib to neonates was done through trans-mammary transfer. Lactating dames were given 1 daily dose of 2 mg kg⁻¹ trametinib (LC laboratories, Boston, MS, USA) or vehicle (0.5% hydroxypropyl methylcellulose, 0.2% Tween-80 in H₂O, pH 8.0) via oral gavage for 5 days, beginning when the pups were aged PN8. Pups were allowed to freely suckle from the dame as usual, with minimal handling until the point of collection (PN8, PN14 and PN16). Treatment and handling of neonates were done at the same time each day (from 11 a.m.). To note, the average weights of all control (n = 12, ~5.6 g) and iEC-Kras (G12D^*) (n = 12, 5.2 g) pups to be treated with either vehicle or trametinib were not significantly different at PN8.

Whole mount brain and retinal immunohistochemical staining

The retinal angiogenic assay is a commonly used technique and performed as previously described on PN8 retinas (47).

For whole mount brain staining, pups were euthanized by isoflurane overdose, and whole brains were removed at PN14. Brains were fixed overnight in 4% paraformaldehyde (PFA) (Sigma-Aldrich), washed in 1X phosphate-buffered solution (PBS), and embedded in 4% agarose (low gelling temperature, ThermoFisher Scientific, Belgium). Slices (200 μm) were cut with a vibratome, washed and blocked overnight at 4°C in a 1% bovine serum albumin, 0.3% Triton-X100 in 1X PBS solution. Sections were incubated for at least 3 days in IsolectinB4 antibody (biotinylated, 1:300; Cat# B-1205-.5, Labconsult, Belgium), then in AlexFluor488-conjugated streptavidin antibody (1:250; Cat# S11223, ThermoFisher Scientific, Belgium), before being post-fixed in 4% PFA for 15 min. Stained slices were mounted with fluromount G (ThermoFisher Scientific, Belgium) and coverslips sealed with nail polish. Using a cell observer spinning disk confocal microscope (Zeiss, Germany), a series of up to 200 z-stack images were acquired for equivalent areas of the brain; Z-stack images were processed using the AriVis4D software (AriVis AG, Germany) to produce 3D vessel constructions. The diameters of unique vessels were quantified using the Vessel Analysis—ImageJ plugin.

Histological and immunohistological staining

Pups were euthanized and the brain, heart, lung, liver and intestines were removed and fixed in 4% PFA. The samples were processed for paraffin embedding and sectioned at 5 μm. Deparaffinized and rehydrated sections were subjected to heat-induced epitope retrieval, then incubated with the primary antibodies IsoB4 (biotinylated, 1:300, Labconsult, Belgium) and phospho-p44/42 MAPK (Thr202-Tyr204; Cat#9101, Bioke, Netherlands); fluorescent secondary antibodies used were against streptavidin (AlexFluor488-conjugated, 1:250, ThermoFisher Scientific, Belgium; or CY5-conjugated, 1:250; Cat#SA-1500-1, Labconsult, Belgium) and rabbit IgG (AlexaFluor488-conjugated, 1:250). Immunofluorescence-stained sections were treated with Vector Labs autofluorescence quenching solution (Labconsult, Belgium) and coverslip mounted with VectaShield vibrance antifade mounting medium with DAPI (Labconsult, Belgium). For chromogenic staining, tissues were blocked with 3% H₂O₂ (Sigma-Aldrich, Belgium) and an horseradish peroxidase-conjugated streptavidin secondary antibody (Cat#RPN1231-2ML, Cytiva, Belgium) was used. Tissue revelation was done by DAB solution (Labconsult, Belgium). Sections were counterstained with hematoxylin (Labconsult, Belgium), dehydrated and coverslip mounted with VectaMount permanent mounting medium (Labconsult, Belgium).

Microscopic image acquisition and data analysis

Whole slides were scanned using the Mirax Midi (Zeiss, Germany) or Pannoramic P250 Flash III (3DHISTECH, Ltd., Hungary) slide scanners. For isoB4-stained slides, at least six nonoverlapping snapshots from each organ (brain, heart, liver and intestine) of each mouse (n = 6–10) were acquired from the scans at 20X magnification. The luminal vessel areas of at least six isolectinB4-positive vessels were traced and measured via ImageJ for each snapshot, then averaged for each mouse (n = 6–10), appropriate organ and treatment regimen. The quantification of pERK expression in the brain, liver and intestine from 4 to 7 nonoverlapping snapshots at 40X magnification of each organ from each mouse was done using ImageJ software (NIH, USA). After appropriately setting the scale bar, pERK staining was converted to gray scale and isolated by determining the threshold; the mean gray value of the threshold was then measured.

Statistical analysis

Statistical analyses and graph generation were performed using the JMP Pro software (v. 14.3, SAS Institute, USA). Comparisons between groups were done using Student t-test for two groups or a one-way analysis of variance for >2 groups. A Kruskal–Wallis test was done when a nonparametric comparison was needed. Pairwise comparisons to detect differences between treatment groups were performed using Wilcoxon’s rank-sum test, followed by a Tukey’s Honest Significant Difference test for correction. The significance level was set at P < 0.05.

Acknowledgements

Two of the authors of this publication are members of the Vascular Anomalies Working Group (VASCA WG) of the European Reference Network for Rare Multisystemic Vascular Diseases (VASCERN)—Project ID: 769036. The authors thank Mourad El Kaddouri for mouse care and Delphine Nolf for technical assistance.

Conflict of Interest statement. None declared.

Funding

Fonds de la Recherche Scientifique—FNRS (T.0026.14 and T.0247.19 to M.V. and T.0146.16 and P.C013.20 to L.B.); Fund Generet managed by the King Baudouin Foundation (to M.V.); Pierre M. fellowship (to H-L.N.).