Endothelial k-RasV12 expression induces capillary deficiency due to marked tube network expansion coupled to reduced pericytes and basement membranes

Abstract

Objective

We sought to determine how EC expression of the activating k-Ras mutation, k-RasV12, affects their ability to form lumens and tubes and interact with pericytes during capillary assembly.

Approach and Results

Using defined bioassays where human ECs undergo observable tubulogenesis, sprouting behavior, pericyte recruitment to EC-lined tubes, and pericyte-induced EC basement membrane deposition, we assessed the impact of EC k-RasV12 expression on these critical processes that are necessary for proper capillary network formation. This mutation, which is frequently seen in human ECs within brain arteriovenous malformations, was found to markedly accentuate EC lumen formation mechanisms, with strongly accelerated intracellular vacuole formation, vacuole fusion and lumen expansion, and with reduced sprouting behavior, leading to excessively widened tube networks compared to control ECs. These abnormal tubes demonstrate strong reductions in pericyte recruitment and pericyte-induced EC basement membranes compared to controls, with deficiencies in fibronectin, collagen type IV and perlecan deposition. Analyses of signaling during tube formation from these k-RasV12 ECs reveals strong enhancement of Src, Pak2, b-Raf, Erk and Akt activation and increased expression of PKCε, MT1-MMP, acetylated tubulin and CDCP1 (most are known EC lumen regulators). Pharmacologic blockade of MT1-MMP, Src, Pak, Raf, Mek kinases, Cdc42/Rac1, and Notch markedly interferes with lumen and tube formation from these ECs.

Conclusions

Overall, this novel work demonstrates that EC expression of k-RasV12 disrupts capillary assembly due to markedly excessive lumen formation coupled with strongly reduced pericyte recruitment and basement membrane deposition, which are critical pathogenic features predisposing the vasculature to develop arteriovenous malformations.

Graphical Abstract

INTRODUCTION

Considerable progress has occurred in defining the molecular basis for a class of vascular lesions termed vascular anomalies^1–9. These anomalies are complex and encompass many different types of lesions that arise from the blood or lymphatic vasculatures. Broadly speaking, the vascular anomalies are characterized by marked vascular morphogenic lesions such as those observed in arteriovenous, venous, capillary or cavernous malformations^{1–3, 7}, or they are characterized by more proliferative lesions such as observed in hemangiomas^8–11. In recent years, it has become increasingly appreciated that these vascular anomalies are caused by genetic mutations of key molecules that primarily stimulate Ras/Map kinase and/or PI3 kinase signaling pathways. Some of the key signaling molecules that drive these vascular anomalies are activating mutations in k-Ras, b-Raf, Map2k1 (Mek1), GNAQ, PIK3CA, and Tie2, and inactivating mutations in Rasa1, a RasGAP which inactivates Ras^{2, 3, 5–8, 11–26}. Very recently, it was found that activating mutations in k-Ras (such as k-RasV12) are observed at high frequency within endothelial cells (EC) in human brain arteriovenous malformations¹².

In our studies of EC lumen and tube formation, we reported a few years ago that k-Ras was observed to be an important regulator of tube formation along with Cdc42, Rac isoforms, and Rap1b, and furthermore, that siRNA suppression of Rasa1 accelerated the lumen formation process²⁷. Thus, it is possible that one of the key abnormalities in malformation syndromes may be due to excessive or altered EC lumen formation, a process that has not been examined in detail in this context. Over the past 25 years, considerable progress has been made investigating the mechanistic and molecular basis for EC lumen and tube assembly in vitro and in vivo^28–32. Critical steps during EC lumen formation include: 1) intracellular vacuole formation and coalescence^33–37; 2) vacuole/vesicle trafficking along the microtubule cytoskeleton to create and expand the apical membrane surface^{27, 38, 39}; 3) creation of both cytoskeletal and membrane polarity that leads to basally polarized F-actin, subapically polarized acetylated tubulin (to support vacuole trafficking and the developing apical membrane surface)^38–40, and apically polarized proteins such as activated Src isoforms, caveolin-1, Rasip-1, and podocalyxin^{27, 39, 41}; and 4) tube expansion and elongation in the extracellular matrix (ECM) secondary to membrane-type 1 metalloproteinase (MT1-MMP)-dependent proteolysis of the matrix leading to generation of vascular guidance tunnels which are matrix spaces where the EC tube networks are embedded^42–45. Over many years, key molecular regulators of these steps and this process have been identified and these include the small GTPases, Cdc42, Rac1, Rac2, k-Ras, and Rap1b, as well as RalA, RalB, and various Rab GTPases^{27, 34, 40, 41, 46}. Many critical downstream effectors of these GTPases have been identified to play a role including Pak2, Pak4, IQGAP-1, MRCKβ, GIT1, βPIX, and Rasip1^{27, 41, 46–48}. Major kinases that have been found to be important for EC lumen formation include protein kinase C epsilon (PKCε), Src family kinases including Src, Fyn and Yes, as well as Pak, Raf, and Erk kinases^{39, 46, 47, 49}. Furthermore, a key control step in EC lumen and tube formation is ECM proteolysis, which is mediated by MT1-MMP, and directly coordinated with the EC lumen signaling pathways and molecules mentioned above^{42, 43, 45}. Other interesting regulators of EC lumen formation include the microtubule tip complex proteins, EB1, p150glued and Clasp1³⁸, or the cerebral cavernous malformation (CCM) genes, CCM1, CCM2, CCM2L, and CCM3, which interestingly suppress the activation of RhoA, an inhibitor of lumen formation^50–53.

A fundamental requirement for the development of the blood vascular system are proper interactions between ECs and mural cells, such as pericytes⁵⁴, which are necessary to create and maintain capillary networks with their characteristic narrow lumen diameters^{44, 55–58}. When ECs form tube networks in the absence of pericytes in vitro, they continue to widen and become less and less elongated⁴². ECs without pericytes under our serum-free defined systems in either collagen or fibrin matrices are unable to form basement membranes, while with pericytes present, dramatic basement membrane deposition occurs^{44, 59}. The dramatic widening of ECs tubes without pericytes occurs due to a continuation of morphogenesis and lack of a basement membrane, which normally suppresses morphogenesis and facilitates tube maturation³⁰. Very recently, we investigated and identified the key EC-derived factors that regulate pericyte recruitment to EC tubes and thus, they also control capillary formation and basement membrane deposition. These EC-derived factors affecting pericyte behavior are PDGF-BB, PDGF-DD, endothelin-1 (ET-1), TGFβ1, and HB-EGF⁵⁷. Individual blockade of these factors or their receptors had modest, but significant inhibitory effects on pericyte recruitment, while combined blockade of these molecules or receptors much more dramatically interfered with pericyte recruitment leading to markedly widened tubes with little to no basement membrane deposition⁵⁷. In fact, under these highly inhibited conditions, the ECs formed tube networks that continued to expand in diameter and shorten in length as if the pericytes were not present.

In general terms, many vascular malformations show reduced mural cell interactions with the malformed EC morphogenic structures⁶⁰. A few studies have implicated alterations in PDGF-BB levels as a potential reason for these differences⁶¹, but as mentioned above, this is likely to be much more complex due to the fact that under normal conditions multiple EC-derived factors work together to stimulate this pericyte recruitment and basement membrane assembly process⁵⁷. In order to investigate these types of key questions, which are necessary to both understand the underlying pathogenesis of particular syndromes, but also to develop new therapeutic strategies to prevent or fix these abnormal vascular morphogenic or proliferative structures, there is a critical need to create definable in vitro 3D malformation or hemangioma models using human vascular cells including ECs and mural cells. This approach is important to investigate in detail the contribution of individual genetic mutations and to assess their biological impact on the key steps controlling vascular cell morphogenesis, proliferation or EC-pericyte interactions. For example, evaluation of EC lumen formation, tube elongation, sprouting behavior, pericyte recruitment to the mutant EC morphogenic structures, and the degree to which vascular basement membrane assembly occurs, are particularly relevant issues to understand the biology underlying why particular mutations lead to specific malformation syndromes.

Here, we investigate the impact of a common mutation observed in ECs within human brain arteriovenous malformations (i.e. k-RasV12), on the ability of human ECs to undergo morphogenic processes and to also assess the ability of these expressing cells to interact with pericytes to affect vascular maturation events using a defined 3D model system. Our findings demonstrate that ECs expressing activated k-RasV12 show remarkably enhanced ability to form intracellular vacuoles, lumens, vascular guidance tunnels, and tubes, which are markedly widened compared to control ECs. In contrast, the k-RasV12 ECs demonstrate reduced ability to sprout compared to control ECs, which contributes to why tubes formed by these ECs are widened and less elongated. Furthermore, the widened tubes formed by k-RasV12-expressing ECs show strongly reduced pericyte recruitment and basement membrane deposition, particularly observed with fibronectin, collagen type IV and perlecan, compared to controls. Signaling experiments during 3D morphogenesis reveal marked enhancement and activation of Src, Pak2, b-Raf, Erk, and Akt, and increased expression of PKCε, MT1-MMP, acetylated tubulin, and CDCP1 by the k-RasV12-expressing ECs. Individual pharmacologic blockade of MT1-MMP, Src family kinases, Cdc42 and Rac1, Pak kinases, Raf kinase, Mek kinase, PKCε, PI3 kinase or Notch signaling strongly inhibit the ability of k-RasV12 ECs to form lumens and tubes. Overall, k-RasV12 expression by ECs results in an inability of these cells to assemble normal capillary networks due to excessive lumen formation and tube widening combined with impaired pericyte recruitment and basement membrane deposition. These abnormalities represent key pathogenic features leading to deficiencies in the assembly of intervening capillaries, a predisposing step in the development of arteriovenous malformations.

METHODS

The authors declare that all supporting data are available within the article [and its online supplementary files].

Cell culture

The cell lines used in this paper, human umbilical vein endothelial cells (HUVECs), k-RasV12 ECs, and human brain vascular pericytes (HBVP) were cultured in our own “Supermedia,” on gelatin-coated flasks. These cells were grown in incubators containing 5% CO2 at 37°C. Our “Supermedia” consists of 20% fetal bovine serum, heparin sodium salt, bovine hypothalamus extract, amphotericin B, and gentamicin in M199 media. HUVECs were used from passage 3 to 6, k-RasV12 ECs were used from passage 6 to 12, and HBVPs were used from passage 4 to 12. The k-RasV12 expressing ECs were obtained by infecting HUVECs with a recombinant lentivirus carrying k-RasV12 (k-Ras isoform b, NP_004976) and also red fluorescent protein fused with blasticidin, which was used for selection (GenTarget). The k-RasV12 mutant gene is expressed downstream of an EF1a promoter. After infection, we waited 3 days and selected with blasticidin (10 μg/ml). The new EC line was grown up from many clones of surviving cells that were propagated into a pool of ECs expressing k-RasV12. This pool of ECs was frozen down in liquid nitrogen vials, so that they could be regrown and used in future experiments.

Vasculogenic Assay

A T75 flask of HUVECs or k-RasV12 ECs were trypsinized and resuspended in a 2.5 mg/ml type I collagen matrix⁶². The gel was then added to a 96-well, half area plate (28 ul/well) and placed in an incubator for 30 mins, allowing for polymerization/equilibration. After polymerization, feeding media was added on top of gels. Our feeding media consists of M199 media plus five growth factors (RSII with insulin, FGF2, SCF, IL3, and SDF)⁶². These cultures were fixed in 3% paraformaldehyde or 3% glutaraldehyde at predetermined timepoints for fluorescent or nonfluorescent imaging. For the drug blocking assays, the pharmacologic drugs were mixed with feeding media at the desired concentrations and were added to cultures.

EC/k-RasV12 pericyte coculture

The ECs (HUVECs or k-RasV12-expressing) and GFP-labeled pericytes were cocultured in 2.5 mg/ml type I collagen matrices in a ratio of 1:5. The GFP-pericytes were generated as previously described⁵⁵. The mixture was added to the 96-well plate. After polymerization, the media with five growth factors (RSII with insulin, FGF2, SCF, IL3, and SDF) were added to these cultures. These cultures were fixed at various timepoints with paraformaldehyde and were then immunostained for CD31 to label ECs or were stained with key basement membrane matrix components. Confocal imaging was then performed.

Sprouting Assay

The 2.5 mg/ml type I collagen matrices were added to the 96-well plate. HUVECs or k-RasV12 ECs were resuspended in the five growth factor media and were added on the top of the collagen matrices after they were polymerized⁶³. These cultures were fixed with glutaraldehyde at 16 hr and stained with 0.1% toluidine blue for imaging. To quantify the tip number of HUVECs or k-RasV12, pictures were taken using a 4X objective and ImageJ was used to count the tip number. To quantify the tip number from both cell types, each picture was taken 20 μm below the monolayer.

Culture Fluorescent Immunostaining

The cultures were fixed with 3% paraformaldehyde. After fixation, the collagen gels were plucked out and washed in Tris-Glycine for 1 hour on a shaker at 4°C. If staining for intracellular proteins, such as CD31, caveolin-1, and acetylated tubulin, gels were permeabilized with Triton TX-100 for 1 hour on the shaker. If staining for basement membrane components, for example, collagen type I or various basement membrane components such as laminin, fibronectin and collagen type IV, cells were not permeabilized. Gels were subsequentially incubated with secondary antibody-specific blocking buffer for 1 hr at 4°C. Primary antibody was added directly to the blocking buffer and the gel was incubated with the primary antibody overnight at 4°C. After several washes with wash buffer— Tris-buffered saline with Tween 20 (TBST) (if staining for intracellular proteins) or Tris-buffered saline (TBS) (if stained with basement membrane components), the same blocking buffer was added to these gels and a specific Alexa Fluor secondary antibody was added to corresponding gels and incubated for 2 hr in 4°C. After 2 hr, the secondary antibody was removed, and the gels were washed several times with TBS or TBST. Hoechst was added to the wash buffer and the gels were incubated for 1 hr. Again, TBS or TBST was used to wash gels until background has been washed out. Gels were then imaged by confocal microscopy.

Microscopy and Imaging

Images for quantification of lumen area/vacuoles number of HUVECs or k-RasV12 ECs were taken using an Olympus CKX41 microscope and imaging software (DPController/DPManager). Quantification of lumen area/vacuoles number was made by Fiji (ImageJ) and results were analyzed using Microsoft Excel. Quantitation of EC lumen and tube area was performed as described using Metamorph software (MetaMorph 7.8)⁶⁴. For the time-lapse movies, pictures were taken every 10 minutes using Leica Microsystems (DMI6000B microscope plus environmental chamber) and imaging software (MetaMorph 7.8) in 72 hours period. All the pictures were taken using a monochromatic Hamamatsu ORCA-ER C4742-80 camera and 10X PH1 Leica 506507 lens. To create a movie, these pictures were stack together into a movie using MetaMorph software and stabilized using Adobe After Effects. Confocal pictures were taken using Leica SP8 3X STED Laser Confocal Microscope and image software (LAS X), all pictures were obtained with 3 different lenses (20X HC PL APO, 0.75NA, WD 0.62 mm; 40X HC PL APO, water immersion, 1.1NA, WD 0.65 mm; 63X HC PL APO, Oil immersion, 1.4NA, WD 0.14 mm). The raw pictures were made and reconstructed by Fiji (ImageJ).

Western blots

The cultures (gels) were plucked out and dissolved in a 1.5% sample buffer with 5% beta-mercaptoethanol. These samples were then loaded into a heat block (100°C) for 5 minutes. After the samples cooled down, 15 μl of sample was pipetted into a Biorad gel and ran until the ladder reached the bottom. The gel was transferred with a PVDF membrane and the membrane was blocked with 3% BSA. Specific primary antibodies were added directly into blocking buffer and incubate overnight at 4°C. After several washes, a corresponding HRP antibody in 3% milk was used to detect the protein. The results were imaged on x-ray film after developing.

Reverse transcription polymerase chain reaction (RT-PCR)

Direct-zol RNA miniprep kit was used to extract EC RNA (HUVECs or k-RasV12 ECs) based on the manufacturers’ instructions. AccuScript first strand cDNA kit was used to synthesis cDNA with 500 ng of RNA. Primer sequences that were used for targeted amplification are shown in the Major Resources Table. PCR products were run on a 1% agarose gel, and the results were visualized using a Fotodyne system (Fotodyne Inc).

Statistics

Statistical data analyses were performed using Microsoft Excel (Microsoft) or Prism 8 (Graphpad). Data were analyzed for normality and equal variances were obtained and student t-tests were used to compare means between two conditions. Statistical significance was set at a minimum of p<.05.

RESULTS

ECs expressing k-RasV12 show markedly enhanced ability to form lumen and tube networks compared to control ECs

Previously, our laboratory demonstrated that k-Ras in conjunction with other small GTPases such as Cdc42, Rac isoforms, and Rap1b control the process of EC lumen and tube formation²⁷. Here, we examined whether EC expression of activated k-Ras (i.e. k-RasV12), a key activating mutation found in ECs within human arteriovenous malformations, would affect lumen and tube development. We directly compared lumen and tube formation using control ECs vs. ECs expressing k-RasV12 and using assays that mimic vasculogenesis. Cultures were fixed and stained with toluidine blue at different time points, and lumen/tube area and ECs with multiple intracellular vacuoles were quantified (Figures 1A and 2A,B). Co-immunostaining of fixed cultures with anti-CD31 antibodies to label ECs and anti-collagen type I antibodies to label the extracellular matrix (ECM) reveals strong acceleration of lumen formation and tube expansion from ECs carrying k-RasV12 (Figure 1B). EC lumen formation is accompanied by the creation of vascular guidance tunnels which are physical spaces in the ECM created by MT1-MMP-dependent proteolysis during this process⁴². ECs expressing k-RasV12 show marked expansion of EC lumens and tubes, but also vascular guidance tunnel spaces within the collagen matrices compared to control ECs (Figure 1B). In support of these conclusions, quantitative measurements of lumen area showed that ECs expressing k-RasV12 formed significantly larger lumens with markedly enhanced tube area after 24, 48, or 72 hr (Figure 3A). Detailed quantitative analysis of EC lumen formation (i.e. evaluated at 3 hr intervals) using real-time movies over 72 hr revealed the marked acceleration of lumen and tube formation that results from EC expression of k-RasV12 (Figure 3B).

Figure 1. ECs expressing activated k-RasV12 demonstrate markedly accelerated lumen and tube formation, compared to control ECs.

ECs, as well as ECs carrying k-RasV12, were cultured in 3D collagen matrices. Cultures were fixed either with glutaraldehyde or paraformaldehyde at 24, 48, or 72 hr and were stained and photographed. (A) Cultures were fixed at 24, 48, and 72 hr with glutaraldehyde and stained with toluidine blue, pictures were taken using a 10X objective and representative pictures are shown. (B) Cultures were fixed with paraformaldehyde and immunostained with antibodies to collagen type I or CD31, and nuclei were labelled with Hoechst dye. Results were analyzed by confocal microscopy and representative images are shown. Bar equals 100 μm. L indicates lumen space. White arrows indicate the borders of vascular guidance tunnels with the collagen matrices.

Figure 2. ECs carrying k-RasV12 show accelerated intracellular vacuole formation during early phases of lumen formation compared to control ECs.

ECs, as well as ECs carrying k-RasV12, were cultured in 3D collagen matrices and these cultures were fixed at the indicated timepoints. (A) ECs and k-RasV12 vasculogenic cultures were fixed with glutaraldehyde at 3, 6, and 9 hr and stained with toluidine blue, and representative images are shown. Bar equals 200 μm. (B) Cultures from (A) were photographed and quantitated for the number of ECs containing multiple intracellular vacuoles. Data are reported as average number of ECs with vacuoles per field ±standard deviation (n=12; p values are indicated).

Figure 3. ECs carrying k-RasV12 are induced to form markedly widened lumens and tubes compared to control ECs.

ECs, as well as ECs carrying k-RasV12, were cultured in 3D collagen matrices and these cultures were fixed at the indicated timepoints. (A) Cultures were fixed with glutaraldehyde at 24, 48, and 72 hr and were stained, photographed, and quantitated for EC lumen and tube area. Data are reported as average lumen area per high-power field (HPF) ±standard deviation (n = 10; p values are indicated). (B) Time-lapse movies were performed for ECs and EC k-RasV12 cultures. For each movie, images from the indicated time points were quantified for lumen area using MetaMorph software. Data are reported as average lumen area per high-power field (HPF) ±standard deviation (n=30; p values are indicated).

Increased intracellular vacuole formation from ECs expressing k-RasV12 leading to enhanced lumen formation and apical membrane targeting of caveolin-1

Previous work from our laboratory has demonstrated a key relationship between the ability of ECs to form intracellular vacuoles and for them to rapidly form lumens and tube networks in 3D extracellular matrix environments^{33, 35, 36, 39, 40}. Here, we addressed whether the enhanced lumen and tube formation seen following EC expression of k-RasV12 was related to their ability to form intracellular vacuoles during the lumen formation process. In fact, the ECs carrying k-RasV12 do show strong increases in the accumulation of EC intracellular vacuoles compared to control ECs during the early phases of lumen formation (Figure 2A,B). We have previously reported that caveolin-1 labels intracellular vacuoles^{33, 40} and can accumulate at the developing apical membrane surface as apically targeted vacuoles fuse with this surface during EC lumen formation⁴⁰. Immunostaining of control vs. k-RasV12 expressing ECs with an anti-caveolin-1 antibody (and imaged using confocal microscopy) reveals strong labeling of both intracellular vacuoles, but also the developing EC apical membrane, which is markedly accentuated by the expression of k-RasV12 in ECs (Figures 4A and Suppl. Figure I). Intracellular vacuoles are decorated with caveolin-1, which is observed at 3, 6, or 9 hr of culture, and particularly evident using the k-RasV12 ECs (Figures 4A and Suppl. Figure I). The same anti-caveolin-1 immunostaining of cultures at 24, 48, and 72 hr, reveals the marked expansion of the lumen and tube spaces that occurs when ECs expressing k-RasV12 undergo morphogenesis as compared to control ECs (Figure 4B).

Figure 4. Marked acceleration of lumen formation and tube expansion from k-RasV12 expressing ECs compared to control ECs with enhanced targeting of caveolin-1 to intracellular vacuoles and the apical membrane surface.

ECs, as well as ECs carrying k-RasV12, were cultured in 3D collagen matrices and these cultures were fixed at the indicated timepoints. (A,B) Cultures were fixed with paraformaldehyde at 3, 6, 9, 24, 48, and 72 hr and were stained with antibodies directed to caveolin-1 and acetylated tubulin and labeled with Hoescht dye to stain nuclei. Representative confocal images of these stained cultures at early times (3, 6, 9 hr) (A) and at later times (24, 48, 72 hr) (B) are shown. For (A), Bar equals 20 μm; and for (B), Bar equals 50 μm. L indicates lumen spaces, while v indicates intracellular vacuoles. White arrowheads indicate selective apical membrane staining of caveolin-1, while the black arrowheads demonstrate caveolin-1 staining of intracellular vacuole membranes.

Real-time movies demonstrate marked accentuation of EC intracellular vacuolation as well as lumen and tube expansion from ECs expressing k-RasV12 compared to control ECs

To further evaluate the above observations during these EC lumen and tube formative events, we performed real-time movie analysis comparing control ECs (Suppl. Movies I–III) to k-RasV12 expressing ECs (Suppl. Movies IV–VII). By visualizing real-time movies over 72 hr, we observe strong increases in EC vacuolation within the first 24 hr and marked lumen formation and tube expansion in the k-RasV12 ECs compared to control ECs from 24–72 hr of culture. We selected multiple independent fields for this analysis and representative movies are shown to demonstrate these findings (Suppl. Movies I–VII). The results observed in these movies strongly support the conclusions that ECs expressing k-RasV12 markedly accelerate intracellular vacuolation, vacuole fusion, lumen formation and tube width expansion during vascular morphogenesis compared to control ECs (Figures 1–4).

ECs carrying k-RasV12 show reduced sprouting behavior compared to control ECs

Previous work from our laboratory has shown that there is an inverse relationship between EC sprouting behavior and lumen formation, whereby treatments that lead to reduced lumen formation increase EC sprouting activity and vice-versa^{39, 63}. For example, blockade of Src family kinases, Notch signaling or both result in strong inhibition of EC lumen formation, while they concomitantly increase EC sprouting behavior. Here, we assessed the ability of k-RasV12 expressing ECs vs. control ECs to sprout in 3D matrices (Figure 5B,C). Cultures were fixed, stained, and quantified after 16 hr, using an optimized EC sprouting and invasion assay system⁶³. A statistically significant reduction in EC tip numbers was observed for k-RasV12 ECs compared to control ECs, suggesting a strongly reduced sprouting ability for these ECs (Figure 5A). In top-down images of these cultures, control ECs displayed multiple invading EC tip cells, while k-RasV12 ECs showed many fewer invading cells (Figure 5B). Similarly, images from these cultures viewed from the side revealed the same result and conclusion. EC controls are shown to be invading (black arrowheads), whereas k-RasV12 ECs showed little invasive behavior (Figure 5C). Thus, these results are very consistent with our previous work in terms of the inverse relationship of EC sprouting behavior with the ability to form EC lumens. The marked increase in EC lumen formation by k-RasV12 expressing ECs is accompanied by strong reductions in their ability to invade as single cells to initiate EC sprouting behavior.

Figure 5. ECs expressing k-RasV12 demonstrate reduced sprouting behavior compared to control ECs.

(A-C) EC sprouting assays were performed with control vs. k-RasV12 ECs and these were fixed after 16 hr and stained with toluidine blue. Invading EC tip cells were quantified (n=30, p values are indicated) (A). Representative photographs are shown from an upright view (B) (Bar equals 200 μm) or from a cross-sectional view (C) (Bar equals 500 μm). Black arrowheads indicate the monolayer surface, while the black arrows indicate invading EC tip cells.

Identification of key signaling pathways that are activated during the marked enhancement of lumen formation and tube expansion from ECs expressing k-RasV12

In order to assess signaling pathways and identify important regulators that control k-RasV12 tubulogenesis, western blots were performed to compare both control ECs and k-RasV12 expressing ECs during 3D tube morphogenic events over time (Figure 6). Particularly striking differences between these two conditions were observed with increased phosphorylation of Src, Pak2, b-Raf, Erk, Akt, CDCP1, and GSK3β (which inactivates GSK3β) from k-Ras expressing ECs compared to controls (Figure 6A,B). Lesser increases were observed with increased phosphorylation of PKC epsilon (PKCε), and c-Raf from the k-RasV12 ECs (Figure 6A). Key previous work has demonstrated critical roles for PKCε, Src family kinases, as well as Pak, Raf, and Erk kinases on EC lumen formation using our models of EC tubulogenesis and from others work^{46, 47, 49, 65}, while other studies demonstrated a role for GSK3β inactivation during Cdc42-dependent EC lumen formation⁶⁶. In addition, increased expression of MT1-MMP, α2 integrin (a collagen-binding integrin), and acetylated tubulin were observed in the k-RasV12 expressing ECs compared to controls, and these have been shown to be important for EC lumen and tube formation^{33, 42, 45}. Other molecules such as phospho-Pak4, Dusp6, phospho-cofilin, phospho-histone H3, and detyrosinated tubulin showed similar patterns of expression during this time course using both cell types (Figure 6). The marked downregulation of Dusp6, a key phosphatase which inactivates Erk occurred from both cell types during tube formation. In addition, the strong downregulation of phospho-histone H3 (a key marker of cell proliferation), indicates that the differences observed between the k-RasV12 and control ECs were not the result of cell proliferation (Figure 6B). Also, the real-time movies that we performed (Supplemental Movies I–VII) did not reveal any evidence of EC proliferation. Furthermore, our previous studies as well as others shows that vascular morphogenic events are accompanied by downregulation of EC proliferation^{58, 67, 68}, and interestingly, that is also occurring with the k-RasV12 expressing ECs during their excessive lumen and tube formation (Figure 6B). Overall, these results are consistent with the conclusion that k-RasV12 expression in ECs strongly stimulates EC lumen through acceleration of intracellular vacuole formation and lumen expansion mechanisms which are driven by key signaling pathways and molecules that have been previously described by our laboratory and others.

Figure 6. Identification of key signaling pathways that regulate the accentuated lumen and tube formation response of ECs expressing k-RasV12 compared to control ECs.

Tubulogenesis assays in 3D collagen matrices were performed using control ECs vs. ECs expressing k-RasV12. Lysates were prepared from collagen gels from these cultures at the indicated time points (0, 3, 6, 9, 24, 48, and 72 hr). Western blots were performed to assess the expression of the indicated molecules over time (A,B).

Pharmacologic inhibitors of MT1-MMP, Cdc42/Rac1, PKCε, Src family kinase, Notch, Raf, Mek, and PI3 kinase signaling strongly inhibit the ability of k-RasV12 expressing ECs to form lumens and tubes

To address the functional role of many of the molecules and signaling pathways identified through Western blot analysis (Figure 6), we added pharmacologic inhibitors from the beginning of culture and quantified their impact on the lumen formation from the k-RasV12 expressing ECs after 72 hr (Figure 7). The marked expansion of lumens and tubes from these k-RasV12 ECs also leads to expansion of vascular guidance tunnels (Figure 1B), matrix spaces which are known to be generated by MT1-MMP-dependent proteolytic processes. Addition of the matrix metalloproteinase (MMP) inhibitor, GM6001, which inhibits the function of MT1-MMP, completely blocked the ability of these ECs to form lumens (Figure 7A,B). Addition of the Src family kinase inhibitor, PP2, markedly blocks lumen formation from k-RasV12 expressing ECs, while the Notch inhibitor, DBZ, also blocks, but to a lesser extent (Figure 7C). The combination of PP2 and DBZ, also strongly interferes with lumen formation. Interestingly, past work revealed that increased expression of activated Notch4 in ECs led to arteriovenous malformations in mice⁶⁹. The Cdc42 and Rac inhibitor, MBQ167, as well as the Pak inhibitors, Frax486, and Frax597, interfere with lumen formation from these cells (Figure 7C). Our past work demonstrates the critical role of Cdc42 and Rac isoforms as well as their downstream effectors, including Pak2 and Pak4 during the EC lumen formation process. These critical molecules also play a major role in EC intracellular vacuole formation, an important regulator of EC lumen formation, and a process which is markedly accentuated by the EC expression of k-RasV12 (see Figures 2A and B, Figures 4A and Suppl. Figure I). Inhibitors of Raf kinase (PLX8394), Mek kinase (U0126) and PI3 kinase (Pictilisib) also significantly interfered with EC lumen formation from these cells, while the broad spectrum PKC inhibitor, Go6983 (an inhibitor of PKCε), also strongly blocked, while the PKCα selective inhibitor, Go6976 did not (Figure 7C,D). Importantly, both Raf kinases and PI3 kinase are direct downstream effectors of activated k-RasV12, and our findings suggest that both play roles in the ability of the k-RasV12 expressing ECs to form lumens and tubes. Other inhibitors directed to Rho kinase (Y27632), TGFβ signaling and the TGFβ receptor, Alk-5 (SB431542) had no effect on lumen formation from the k-RasV12 ECs, while the BMP signaling and BMP receptor antagonist directed to Alk-2 and Alk-3 receptors (DMH1) significantly increased lumen formation from these cells (Figure 7D). To further illustrate the effects of these agents, we show a few representative images demonstrating the strong blocking effect of GM6001, PP2, MBQ167, and U0126, while DMH1 enhances lumen formation from ECs expressing k-RasV12 (Suppl. Figure II). With these blocking agents, the ECs assumed a different morphologic or morphogenic appearance, with single cells (and no lumens) apparent following addition of GM6001, spindle-shaped cells (and no lumens) with PP2, small EC lumen cyst-like structures with MBQ167, and very small lumen structures with some branching with U0126 (Suppl. Figure II). For comparison, we added the same set of pharmacologic inhibitors to control ECs, and there were strong blocking effects of the different drugs on EC lumen formation except for Y27632, SB431542, and DMH1, which showed much less inhibitory effect (Suppl. Figure III). Overall, control ECs appear to be more sensitive to the morphogenesis-blocking effects of these signal transduction inhibitors compared to the ECs expressing k-RasV12.

Figure 7. Pharmacologic blockade of lumen and tube formation from k-RasV12 expressing ECs using inhibitors directed to MT1-MMP, PKCε, Src family kinase, Notch, Cdc42/Rac1, and Pak, Raf, Mek, and PI3 kinases.

ECs expressing k-RasV12 were seeded in 3D collagen matrices in the presence or absence of the indicated pharmacologic agents for 72 hr. (A,B) The MT1-MMP inhibitor, GM6001, was added at 20 or 10 μM vs. control conditions and cultures were fixed, stained with toluidine blue, photographed (A) (GM6001 at 10 μM), and quantitated for lumen area (B). (C,D) Other pharmacologic agents added were; the Src inhibitor, PP2 (10 μM); the Notch γ-secretase inhibitor, DBZ (5 μM); the Cdc42/Rac inhibitor, MBQ167 (2.5 μM); the Pak inhibitors, Frax486 (5 μM) and Frax597 (5 μM); the Raf inhibitor, PLX8394 (10 μM); the Mek inhibitor, U0126 (5 μM); the PI3 kinase inhibitor, Pictilisib (5 μM); the PKCα inhibitor, Go6976 (10 μM); the Rho kinase inhibitor, Y27632 (10 μM); the TGFβ signaling inhibitor, SB431542 (10 μM); the BMP signaling inhibitor, DMH1 (10 μM); and the general PKC inhibitor which blocks PKCε, Go6983 (10 μM). (B,C,D) Data are reported as average lumen area per high-power field ±standard deviation (n = 6; p values are indicated).

Two additional experiments were performed with the k-RasV12 expressing ECs to address the effect of the pharmacologic inhibitors on the initial steps in lumen formation which was to assess their influence on vacuole formation (Suppl. Figure IV) or to assess their ability to affect pre-existing EC tubes by adding them after 48 hr of culture (Suppl. Figures V and VI). The Src family kinase inhibitor, PP2, and the Cdc42/Rac inhibitor, MBQ132, had the strongest inhibitory activity on vacuole formation from the k-RasV12 ECs, while the other drugs had less effect (Suppl. Figure IV). This is consistent with the known role of Src family kinases as well as Cdc42 and Rac in the vacuole formation process^{28, 34, 39, 46}. When the different pharmacologic agents were added to cultures after they had formed tubes for 48 hr, strong blocking effects were observed following addition of the combination of PP2 and DBZ (blocking both Src and Notch) or MBQ132 (blocking Cdc42/Rac), while addition of DMH1 (blocking Alk2 and Alk3) led to increased lumen area (Suppl. Figure V). Representative photographs from these cultures reveal tube collapse responses seen after PP2/DBZ (collapsed lumens and appearance of spindle-shaped ECs) and MBQ132 addition, which showed many cyst-like lumens, while DMH1 treatment caused expanded lumen structures (Suppl. Figure VI). Lesser, but significant blocking effects were observed with the drugs interfering with Pak, Raf, Mek, PKCε and PI3 kinases (Suppl. Figure V). These data indicate that multiple stages of EC lumen/tube formation and maintenance are affected by these signaling molecules and pathways in the context of the abnormally accelerated lumen forming ability of ECs expressing k-RasV12.

Expression of k-RasV12 in ECs leads to deficiencies in capillary assembly due to strong reductions in pericyte recruitment and pericyte-induced EC basement membrane matrix deposition

Capillaries are composed of EC tube networks with associated pericytes, and they critically intervene between arteries and veins to keep them spatially and functionally separated. If capillary networks become deficient, this spatial and functional separation disappears leading to direct connections between arteries and veins creating arteriovenous malformations. Since EC expression of k-RasV12 is known to occur within human arteriovenous malformations, we tested whether EC expression of this activated k-Ras would affect the ability of ECs to form functional networks of capillaries with associated pericytes. Control ECs were compared to k-RasV12 expressing ECs and EC-pericyte co-cultures were established and evaluated over a 120 hr period. Cultures were fixed at 120 hr, and were immunostained with anti-CD31 antibodies to visualize the EC tube networks and determine if GFP-labeled pericytes were found to be associated with the control EC or k-RasV12 EC-lined tubes (Figure 8A). The k-RasV12 EC tubes were observed to be abnormally wide compared to control EC tubes and many fewer pericytes were associated with these tube networks compared to control ECs (Figure 8A, Suppl. Figure VII). Quantitation of these events reveals significantly fewer total pericytes, fewer pericytes elongated on tubes, and a much lower percentage of pericytes associated with ECs tubes expressing k-RasV12 (Figure 8B,C). Immunostaining of control gels with anti-collagen type I antibodies reveals narrow vascular guidance tunnels where pericytes are observed to recruit to EC tubes within the tunnel spaces (Figure 8A). In contrast, ECs with k-RasV12 show much wider and less branched tunnel spaces. Note that many of the pericytes in this instance are observed outside the tunnel spaces, and thus, did not recruit to the widened tubes (Figure 8A).

Figure 8. ECs expressing k-RasV12 exhibit marked pericyte recruitment defects with accompanying widened and less branched tube networks leading to an inability to assemble proper capillary tube networks.

Control ECs and ECs expressing k-RasV12 were seeded with GFP-labeled pericytes in 3D collagen matrices, and the co-cultures were allowed to develop to evaluate the ability of the ECs to support the establishment of capillary networks. (A) Cultures were fixed after 120 hr, and then immunostained with antibodies to CD31 to label ECs and collagen type I to label the collagen matrix and to visualize the vascular guidance tunnels that are generated during the morphogenic process. Confocal microscopy was performed, and representative photographs are shown. Bar equals 200 μm. White arrows indicate the borders of vascular guidance tunnels. (B,C) After 120 hr, these co-cultures were evaluated for total pericyte number, the number of pericytes elongated on EC tubes, and the percentage of pericytes associated with tubes (Pericytes On vs. Pericytes Off Tubes) (n=6; p values are indicated).

Recently, we defined key EC-derived factors that are responsible for pericyte recruitment and basement membrane deposition during capillary formation⁵⁷. These EC factors are PDGF-BB, PDGF-DD, ET-1, HB-EGF, and the TGFβ isoforms, TGFβ1 and TGFβ2 (although this isoform is primarily expressed by pericytes). Blockade of these factors or their receptors in combination markedly interfered with pericyte recruitment and basement membrane deposition. Because of the deficiencies observed in pericyte recruitment to EC tubes expressing k-RasV12, we performed RT-PCR analysis to determine if there were any major observable differences between k-RasV12 ECs and control ECs (Figure 9B). The expression of k-RasV12 led to a clear upregulation of k-Ras in these ECs, while they showed very similar expression of major EC selective genes including VE-cadherin, VEGFR2, von Willebrand factor, claudin-5, Alk-1, and Erg (Figure 9B). Of the known EC-derived factors controlling pericyte recruitment, we observed decreased expression of PDGF-D and TGFβ2 by the k-RasV12 ECs compared to control ECs (Figure 9B), while the other key factors showed similar expression, so it is possible that these deficiencies could explain in part the defective pericyte recruitment to these widened tubes. However, because of the inherent complexity of this issue and process, we feel that a separate future study is necessary to address this question in detail.

Figure 9. ECs expressing k-RasV12 demonstrate markedly reduced basement membrane matrix deposition in response to EC-pericyte tube co-assembly, compared to control ECs, leading to a failure to properly regulate capillary tube maturation events.

Control ECs and ECs expressing k-RasV12 were seeded with GFP-labeled pericytes in 3D collagen matrices, and the co-cultures were allowed to develop to evaluate the ability of the ECs to support the establishment of capillary networks and capillary maturation events such as basement membrane deposition. (A) The co-cultures were fixed after 120 hr and were immunostained without detergent permeabilization to evaluate extracellular deposition of the basement membrane matrix proteins, laminin, collagen type IV, fibronectin, perlecan, nidogen 1 and nidogen 2. Confocal microscopy was performed to evaluate the immunostaining and representative images are shown. Bar equals 200 μm. (B) RT-PCR analysis of the expression of EC-derived factors controlling pericyte recruitment (PDGFs, ET-1, HB-EGF, TGFβs), genes selectively expressed by ECs, and basement membrane components. We prepared total RNA from control ECs versus the k-RasV12 expressing ECs and analyzed the expression of the indicated genes compared to the housekeeping gene, GAPDH. (C) Schematic diagram illustrating our findings that k-RasV12-expressing ECs induce excessive and widened EC tubes with markedly reduced pericyte recruitment and basement membrane deposition. These represent key reasons why they fail to produce properly formed and branched capillary networks (i.e. capillary deficiency), a fundamental abnormality underlying the pathogenic development of arteriovenous malformations.

A key step in pericyte-induced EC tube maturation and the development of capillary networks is the assembly of the capillary basement membrane matrix. Here, we compared the ability of control vs. k-RasV12 expressing EC tubes to deposit basement membranes in EC-pericyte co-cultures (Figure 9A, Suppl. Figure VII). Control EC-lined tubes recruit pericytes and markedly assemble the capillary basement membrane including laminins, collagen type IV, fibronectin, perlecan, nidogen 1 and nidogen 2. In contrast, strong reductions in the deposition of fibronectin, collagen type IV, and perlecan are observed from the k-RasV12 expressing EC tubes (Figure 9A, Suppl. Figure VII). In addition, reduced deposition of laminins, nidogen 1 and nidogen 2 are also observed compared to control ECs. To address if there is any apparent deficiency in the mRNA expression of these basement membrane components by the k-RasV12 ECs compared to control ECs, we performed RT-PCR (Figure 9B). Our results suggest that they express each of the key basement membrane components compared to control ECs and, also show increased expression of laminin α5, laminin β2, and nidogen 1 at higher levels compared to control ECs (Figure 9B). Thus, the deficiencies in basement membrane deposition appear to be due to failures in EC-pericyte interactions, but also may be related to the excessive lumen formation process driven by MT1-MMP-dependent proteolysis of ECM proteins, most notably by degrading collagens such as type IV collagen and the interstitial matrix protein, type I collagen. An important point is that ECM integrity is necessary to anchor the basement membrane matrix to the interstitial matrix during its assembly, which is needed to stabilize the capillary networks and their underlying basement membrane.

Another possibility is that k-RasV12 ECs might show changes in basement membrane protein secretion compared to control ECs. To address this possibility, we collected conditioned medium after 3 or 5 days from ECs forming lumens and tubes in 3D collagen matrices vs. those seeded onto two-dimensional (2D) collagen-coated plastic surfaces and we performed Western blots to measure the levels of the different basement membrane proteins. The k-RasV12 ECs show reduced secretion of fibronectin, perlecan and nidogen-1 compared to control ECs, while collagen type IV, laminin, and nidogen-2 appeared to be similar to control (Suppl. Figure VIII). In contrast, we examined other EC secreted proteins such as pro-MMP-1 and TIMP-1, which were present at higher amounts from the k-RasV12 ECs, while PAI-1 was comparable to EC controls. The reduced secretion of the key three basement membrane components by k-RasV12 expressing ECs (i.e. fibronectin, perlecan, nidogen-1) was observed from both sets of media from 3D gels or 2D cultures (Suppl. Figure VIII). This reduced secretion is consistent with the very strongly decreased deposition of fibronectin and perlecan in the k-RasV12 EC-pericyte co-cultures compared to control co-cultures (Figure 9A). However, the strong decrease in pericyte recruitment to these k-RasV12-expressing EC tubes (Figure 8) is also a major contributor to this reduced deposition of fibronectin and perlecan. The reduced collagen type IV deposition that is seen in the k-RasV12 EC co-cultures may also be due to reduced pericyte recruitment or the observed decrease in fibronectin deposition. Previously, we reported that blockade of fibronectin matrix assembly in EC-pericyte co-cultures strongly diminished collagen type IV deposition around EC tubes^{31, 44}.

The reduced pericyte recruitment, the decreased pericyte responsiveness to the k-RasV12 EC tubes, and the strongly reduced pericyte-induced basement membrane deposition, leads to the conclusion that k-RasV12 expression in ECs results in an inability of these cells to properly communicate with pericytes, leading to capillary insufficiency (Figure 9C). These results coupled with the marked enhancement of EC lumen and tube formation mechanisms driven by k-RasV12 expression, further leads to enlarged and abnormal vessels with poor mural cell coverage (Figure 9C). Thus, our novel findings presented here provide key mechanistic reasons for the underlying pathogenesis of human arteriovenous malformations and their direct association with the k-RasV12 activating mutation.

DISCUSSION

In this study, we sought to elucidate the biological influence of activated k-RasV12 on the ability of ECs to undergo vascular morphogenesis and interact with mural cells such as pericytes compared to control ECs. Activating mutations in k-Ras are among the most common mutations observed within ECs in arteriovenous malformations. Our key findings demonstrate a dramatic increase in the ability of k-RasV12-expressing ECs to form lumens, leading to widened tube networks with reduced sprouting behavior as well as strongly reduced pericyte recruitment and deposition of major basement membrane components including collagen type IV, fibronectin and perlecan. Thus, ECs expressing an activating k-RasV12 mutation are unable to create functional capillary tube networks due to marked accentuation of the lumen and tube formation process coupled with reduced pericyte interactions and basement membrane assembly. Overall, the k-RasV12-expressing ECs appear to induce a state of capillary deficiency, a predisposing condition that could lead to arteriovenous malformations due to the loss of intervening capillary networks which keeps arteries and veins physically and functionally separated.

Our new studies reveal that the expression of k-RasV12 in ECs leads to marked enhancement of the EC lumen formation pathway, a process that has been investigated in detail over the past 25 years. Key steps in EC lumen formation are dramatically accelerated including intracellular vacuole formation, vacuole fusion events with the developing apical membrane surface, and lumen expansion mechanisms which involve MT1-MMP-dependent proteolysis and generation of vascular guidance tunnel spaces within the extracellular matrix. Each of these major steps in the EC lumen formation cascade are strongly accelerated by the expression of k-RasV12 compared to control ECs. The marked enhancement of lumen formation comes at the expense of EC sprouting behavior, which is reduced in these cells, so the tube networks are excessively wide and are not highly branched. The real-time movies that we show here demonstrates these marked differences in the vascular morphogenic response of control vs. k-RasV12-expressing ECs (Supplemental Movies I–VII).

Our past work has revealed critical roles for the small GTPases, Cdc42, Rac isoforms, k-Ras, and Rap1b as well as key downstream effectors including Pak2, Pak4, and Rasip1 in the lumen formation process. We have also identified critical functional roles for other key molecules including the kinases, PKCε, Src family kinases, Raf, Mek and Erk in these events as well as the cell surface membrane metalloproteinase, MT1-MMP. Additional regulators of EC lumen formation include molecules that target both intracellular vacuole membranes and the apical membrane surface including activated Src family kinases, Rac1, and caveolin-1 as well as the microtubule cytoskeleton (particularly acetylated tubulin which is subapically polarized) which directs and targets vacuoles to the apical membrane during lumen formation. Our studies here demonstrate that EC expression of k-RasV12 markedly enhances the activity or function of these known EC lumen formation regulators. For example, we observed strong increases in phosphorylation of Src, Pak, Raf, and Erk kinases from the k-RasV12 ECs compared to control ECs during lumen and tube formation, and pharmacologic inhibitors of Src family kinases, Pak, Raf and Mek kinases had strong blocking effects. Increased expression of PKCε (which activates Src kinases), and MT1-MMP were observed in these cells and pharmacologic inhibitors of these molecules also blocked lumen formation. Pharmacologic inhibitors of Cdc42 and Rac, which activate Pak kinases, gamma-secretase, which is necessary for Notch activation, and PI3 kinase also interfere with lumen formation from these ECs. Interestingly, the k-RasV12-expressing ECs were also observed to show increased expression of CDCP1, a Src phosphorylation target, a molecule upregulated in cells expressing activated Ras, and a molecule implicated in malignant cancer progression^70–73. Finally, we observed robust targeting of caveolin-1 to intracellular vacuoles and the developing apical membrane surface during the accelerated lumen formation process from k-RasV12-expressing ECs. Overall, the accelerated lumen formation process from these cells appears to be occurring via well-described pathways, and it is quite straightforward to inhibit the formation of these lumens and tubes using pharmacologic agents.

The marked stimulation of EC lumen formation and the formation of the highly widened, but less branched tube networks from the k-RasV12 expressing ECs is not accompanied by increased pericyte recruitment or pericyte-induced tube maturation events. In fact, pericyte responsiveness to the accelerated EC lumens and tubes is strongly suppressed as reflected in significantly reduced pericyte numbers as well as reduced pericyte recruitment and elongation on tube surfaces. Since pericyte recruitment to and migration along the abluminal surface of EC tube networks is necessary to form capillary basement membranes, we demonstrate here that basement membrane deposition is strongly reduced, compared to controls, using the ECs carrying k-RasV12. Our previous work also demonstrated that narrow capillary tube width was controlled by pericyte recruitment and basement membrane formation, so this represents another reason why the k-RasV12-expressing EC tubes are abnormally wide in the presence of pericytes, due to the lack of pericyte recruitment and basement membrane assembly. However, the lumen formation process is markedly enhanced, including MT1-MMP-dependent lumen formation, and this metalloproteinase is known to have strong basement membrane degrading activities (e.g. a major target is collagen type IV). Thus, the lack of basement membrane deposition is likely the result of a combination of factors, including increased proteolysis and destruction of basement membranes as they are being deposited. Another important point relates to our finding of a dramatic lack of fibronectin deposition around the k-RasV12-expressing EC-lined tubes. Our past work showed the important role of EC-pericyte interactions and mechanical forces in controlling fibronectin matrix assembly^{31, 44}, a process known to be regulated by RhoA-dependent mechanotransduction. Since Rho-dependent GTPases have inhibitory activity during EC lumen formation^{34, 74}, the marked acceleration of lumen formation by the k-RasV12 may be associated with reduced Rho-dependent activities and contractile cell behavior, and thus, they may be less able to deposit fibronectin matrices due to an inability to exert mechanical forces during these processes. This possibility needs to be investigated in future work.

Based on the new mechanistic information obtained in this study, an important future direction is to try and repair the defects exhibited by the k-RasV12-expressing ECs, which would restore their ability to create functional capillary networks. For example, a potential approach would be to reduce EC lumen formation and at same time enhance pericyte recruitment and basement membrane deposition. The detailed model system that we describe would allow us to screen for pharmacologic agents, growth factors/peptides, or other factors/agents that might repair these capillary formative defects. A different strategy would be to induce regression of these abnormal vascular structures without affecting the adjacent normal vasculature. Recent work has been focused on the molecular basis for capillary regression both in terms of identifying key growth factors that induce regression, but also on ways to block regression by antagonizing specific pro-regressive factors, and pharmacologic agents that antagonize the underlying pro-regressive signaling pathways⁷⁵. One important question is whether the k-RasV12-expressing EC tube structures are susceptible to pro-regressive factors like normal capillary tube networks. Finally, one of the important issues clinically with arteriovenous malformations is their tendency to bleed which can cause major problems, such as severe hemorrhage or stroke (e.g. when the lesions are present in the brain). The reasons why these abnormal vessels are prone to hemorrhage is not well understood, although the widened vessels with poor mural cell coverage and reduced basement membrane would be expected to be less stable than their normal counterparts. This is clearly a very important issue that needs to be investigated in detail in future studies. We think the model system presented here should be very useful to investigate such key questions underlying the pathogenesis and clinical issues that occur in patients with arteriovenous malformations.

HIGHLIGHTS.

1. Expression of the activating k-Ras mutation, k-RasV12, in human ECs leads to marked stimulation of the lumen formation process.

2. The accelerated lumen formation is caused by dramatic increases in EC intracellular vacuole formation and fusion as well as lumen expansion mechanisms mediated by MT1-MMP.

3. The k-RasV12 ECs also exhibit enhanced intracellular signaling pathways that favor lumen formation including marked increases in Src, Pak, Raf, Erk, and Akt phosphorylation, and tubulin acetylation.

4. The excessive lumen and tube formation is accompanied by strong decreases in pericyte recruitment and basement membrane matrix deposition.

5. EC expression of k-RasV12 leads to a capillary deficiency state that predisposes to the development of arteriovenous malformations.

ABBREVIATIONS

Non-standard Abbreviations and Acronyms

HUVECs

Human umbilical vein endothelial cells

HBVP

Human brain vascular pericytes

VEGF

Vascular endothelial growth factor

FGF

Fibroblast growth factor

SCF

Stem cell factor

IL-3

Interleukin-3

SDF-1α

Stromal derived factor-1 alpha

PDGF

Platelet derived growth factor

HB-EGF

Heparin binding epidermal growth factor

ECM

Extracellular matrix

CCM

Cerebral cavernous malformation

ET-1

Endothelin-1

RT-PCR

Reverse transcription polymerase chain reaction

3D

3-Dimensional

MT1-MMP

Membrane-type 1 matrix metalloproteinase

TBST

Tris-buffered saline with Tween 20

TBS

Tris-buffered saline

GSK3β

Glycogen synthase kinase 3 beta

Dusp6

Dual specificity phosphatase 6

CDCP1

CUB domain containing protein 1

Cdc42

Cell division cycle 42

Rac1

Rac family small GTPase 1

Pak2

P21 (RAC1) activated kinase 2

Mek kinase

Mitogen-activated protein kinase kinase

GNAQ

G protein subunit alpha q

Rasa1

RAS P21 protein activator 1

RalA

RAS Like Proto-Oncogene A

IQGAP-1

IQ motif containing GTPase activating protein 1

MRCKβ

Myotonic dystrophy kinase-related Cdc42-binding kinases

GIT1

GIT ArfGAP 1

βPIX

Rho guanine nucleotide exchange factor 7

Rasip1

Ras interacting protein 1

PKCε

Protein kinase C epsilon

ALK-5

Activin receptor-like kinase 5

EB-1

End-binding protein 1

Clasp-1

Cytoplasmic linker protein 1

CCM2L

CCM2-like protein

Erk

Extracellular-signal related kinase

Tie2

Tek receptor tyrosine kinase

PIK3CA

Phosphatidylinositol-4,5 bisphosphate 3-kinase catalytic subunit alpha

b-Raf

v-raf murine sarcoma viral oncogene homolog B1

TIMP-1

Tissue inhibitor of metalloproteinases-1

PAI-1

Plasminogen activator inhibitor-1

Pro-MMP-1

Pro-Matrix Metalloproteinase-1