Defining the Functional Influence of Endothelial Cell–Expressed Oncogenic Activating Mutations on Vascular Morphogenesis and Capillary Assembly

Abstract

This study sought to define key molecules and signals controlling major steps in vascular morphogenesis, and how these signals regulate pericyte recruitment and pericyte-induced basement membrane deposition. The morphogenic impact of endothelial cell (EC) expression of activating mutants of Kirsten rat sarcoma virus (kRas), mitogen-activated protein kinase 1 (Mek1), phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA), Akt serine/threonine kinase 1 (Akt1), Ras homolog enriched in brain (Rheb) Janus kinase 2 (Jak2), or signal transducer and activator of transcription 3 (Stat3) expression versus controls was evaluated, along with EC signaling events, pharmacologic inhibitor assays, and siRNA suppression experiments. Primary stimulators of EC lumen formation included kRas, Akt1, and Mek1, whereas PIK3CA and Akt1 stimulated a specialized type of cystic lumen formation. In contrast, the key drivers of EC sprouting behavior were Jak2, Stat3, Mek1, PIK3CA, and mammalian target of rapamycin (mTor). These conclusions are further supported by pharmacologic inhibitor and siRNA suppression experiments. EC expression of active Akt1, kRas, and PIK3CA led to markedly dysregulated lumen formation coupled to strongly inhibited pericyte recruitment and basement membrane deposition. For example, activated Akt1 expression in ECs excessively stimulated lumen formation, decreased EC sprouting behavior, and showed minimal pericyte recruitment with reduced mRNA expression of platelet-derived growth factor-BB, platelet-derived growth factor-DD, and endothelin-1, critical EC-derived factors known to stimulate pericyte invasion. The study identified key signals controlling fundamental steps in capillary morphogenesis and maturation and provided mechanistic details on why EC activating mutations induced a capillary deficiency state with abnormal lumens, impaired pericyte recruitment, and basement deposition: predisposing stimuli for the development of vascular malformations.

Graphical abstract

Endothelial cells (ECs) form the lining of blood vessels, which are critical conduits for the exchange of fluids, nutrients, and oxygen.^1,2 Functionally, ECs can be further divided into two major promorphogenic cell types: EC tip cells and EC lumen-forming cells, which together coordinate the ability of ECs to undergo vascular morphogenesis where branched networks of EC-lined tubes are formed.3, 4, 5 The sprouting of developing blood vessels is mediated by specialized, highly motile, and invasive ECs found at the front of growing vessels, which are termed EC tip cells.5, 6, 7 Behind the sprouting tip cells, endothelial lumen-forming cells assemble to generate the three-dimensional (3D) luminal space.^6,8, 9, 10, 11 Considerable advancements have occurred over the past several decades in our understanding of the molecular and signaling basis of EC lumen formation in vitro and in vivo.^4,12, 13, 14, 15, 16 Some advances have also occurred in our understanding of vascular sprouting, particularly related to Notch signaling as being a negative regulator,^7,8,17, 18, 19 but there is much less clarity regarding signal transduction cascades and their impact on sprouting behavior. Past studies identified the factor system, an in vitro, defined, serum-free model in 3D extracellular matrices (Figure 1), which allows for the formation of capillary tube networks by stimulating both EC tip and lumen cell behaviors.^6,12 The formation of human capillary tube networks depends on the addition of five growth factors: stem cell factor (SCF), IL-3, stromal-derived factor-1α (SDF-1α), fibroblast growth factor-2 (FGF-2), and insulin (factors).^6,12,20 The factors also strongly drive EC sprouting responses in 3D matrices. Induction of tubulin stabilization with pharmacologic agents, such as paclitaxel and epothilone B, can markedly and selectively inhibit sprouting behavior, while not affecting EC lumen formation.^5,6 Furthermore, blockade of lumen formation using inhibitors of Src, Notch, or both together leads to marked acceleration of EC sprouting behavior.^5,6,21,22 In addition, ECs expressing active k-RasV12 excessively drive lumen formation, while at the same time demonstrating reduced tip cell sprouting behavior.²³ These findings suggest that EC tip cell sprouting and lumen formation, two critical EC steps of vessel formation, are controlled through separate and possibly opposing mechanisms. Genetic disruption of Ras-interacting protein 1 (Rasip1), which blocks EC lumen formation,²⁴ leads to increased EC sprouting,²⁵ and so does genetic disruption of cerebral cavernous malformation (CCM)-1, which also increases sprouting behavior, and decreases EC lumen formation, in a manner dependent on disrupted Notch signaling.²⁶

Figure 1.

Factor system–derived platform that is used to delineate the endothelial cell (EC) signaling basis for human capillary morphogenesis and maturation. The factor system–derived platform consists of four types of bioassay models of morphogenesis in three-dimensional (3D) matrices, which includes EC vasculogenic, sprouting, and aggregate, and EC-pericyte co-culture assays, and two types of molecular analysis involving protein signaling and gene expression. Altogether, this approach allows for an in-depth study of three critical steps of capillary formation, tip cell sprouting, lumen formation, and pericyte recruitment. The indicated active mutation EC lines were generated and subjected to the factor system–derived platform bioassays for this study and were compared with control ECs. This platform allows for a systematic, high-throughput analysis of how individual molecules control specific morphogenic outcomes and relates these findings to the functional influence of genetic mutations on EC morphogenic and signaling responses. Thus, this platform can be used to uncover the molecular basis for vascular diseases, such as the anomalies, where individual genetic mutations can be directly linked to the altered behaviors of EC tip cells, ECs forming lumens, or pericytes associating with capillary tube surfaces, which leads to capillary deficiency states. Finally, the factor system–derived platform described here is a powerful investigative tool to develop and advance therapeutic strategies. Black arrowhead indicates EC monolayer. Red tubes indicate EC lined structures. Green ovals indicate pericytes. Scale bars: 200 μm (far left vasculogenic assay image); 100 μm (all other images). Ac, active; Akt1, Akt serine/threonine kinase 1; AVM, arteriovenous malformation; CCM, cerebral cavernous malformation; FGF-2, fibroblast growth factor 2; HUVEC, human umbilical vein EC; Jak2, Janus kinase 2; kRas, Kirsten rat sarcoma virus; Mek1, mitogen-activated protein kinase 1; PIK3CA, phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; Rheb, Ras homolog enriched in brain; SCF, stem cell factor; SDF-1α, stromal-derived factor-1α; Stat3, signal transducer and activator of transcription 3.

A third important step in capillary assembly is the recruitment of pericytes, which, in conjunction with ECs, facilitate tube maturation and stabilization by codepositing basement membrane matrices.^2,15,27, 28, 29, 30, 31 ECs without pericytes under the serum-free defined systems in either collagen or fibrin matrices cannot form basement membranes, whereas ECs with pericytes have dramatic basement membrane deposition.^28,32 Failure to recruit pericytes during development leads to vascular instability and regression.33, 34, 35, 36 Furthermore, pericytes recruit to EC-lined tubes through five EC-derived pericyte factors, which include platelet-derived growth factor (PDGF)-BB, PDGF-DD, endothelin (ET)-1, transforming growth factor (TGF)-β1, and heparin-binding epidermal growth factor (HB-EGF).³⁷ The absence of pericytes or blockade of these five pericyte recruitment factors causes EC tubes to remodel, continue to undergo morphogenesis, and become wide, less branched, and without basement membranes.³⁷ This evidence suggests that key EC signaling events control not only EC morphogenesis but also pericyte recruitment, a third critical step controlling capillary formation.

More than 20 different mutations have been described in various vascular anomalies characterized by abnormally formed vessels, and many of these mutations also possess oncogenic activity, which is directly linked to the development of malignant tumors.^13,38, 39, 40, 41, 42, 43, 44 For example, the phosphatidylinositol 3-kinase (PI3K)–Akt–mammalian target of rapamycin (mTOR) pathway has been linked to the dilated cystic lesions of CCMs, which possess ECs expressing activated PIK3CA.⁴³ Venous malformations express activated TEK receptor tyrosine kinase 2 (Tie2), leading to stimulation of the PI3K-Akt-mTOR pathway, EC expression of active kRasV12 is associated with arteriovenous malformations, and an activating mutation of G protein subunit alpha q (GNAQ) is linked to hemangiomas.^39,43,45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63 Activated kRasV12-expressing ECs markedly drive lumen formation with strongly reduced tip cell sprouting, pericyte recruitment, and basement membrane assembly.²³ Overall, these abnormalities represent key pathogenic features leading to deficient assembly of intervening capillaries, which predisposes the vasculature to develop arteriovenous malformations. Thus, key abnormalities in malformation syndromes directly result from alterations in EC function, which impact the normal process of vascular wall assembly, including those regulating tip cell sprouting, lumen formation, pericyte recruitment, or basement membrane deposition.^23,38,41,64, 65, 66, 67 Due to the clear relevance to underlying disease mechanisms, it is crucial to link specific genetic mutations to the fundamental vascular biology of EC morphogenesis and vessel wall assembly. Furthermore, characterization of genetic mutation-induced changes in morphogenesis and molecular signaling pathways should reveal therapeutic intervention points to regulate and control EC tip, lumen, and pericyte behavior to prevent or repair vascular disease states.

Because of the complexity of these questions, there is a critical need to use defined in vitro 3D platforms with human cells, which allow for systematic investigation of how specific molecules, genetic mutations, or signaling pathways impact the key steps of vascular morphogenesis and capillary assembly. The present study demonstrated that our in vitro, 3D, defined factor system represents a powerful platform to address such important questions.^4,6,12,68 To address these issues, active kRas-, Mek1-, PI3KCA-, Akt1-, Rheb-, Jak2-, and Stat3-expressing EC cell lines were generated to assess their impact on the capillary assembly process. Active kRas, Mek1, and Akt1 signals are major EC lumen drivers, enhancing lumen area, lumen width, and the formation of vascular guidance tunnels. Interestingly, active PI3KCA- and Akt1-expressing ECs showed increased cystic lumen-forming ability. Signaling experiments during 3D morphogenesis revealed active Akt1- and kRas-expressing ECs significantly drove lumen formation through protein kinase C epsilon (PKCε), proto-oncogene tyrosine-protein kinase Src (Src), p21-activated kinase 2 (Pak2), b-rapidly accelerated fibrosarcoma (b-Raf), and c-Raf activation. Individual pharmacologic blockade of Src family kinases, Pak kinases, Raf kinase, Mek kinase, and PKCε signaling markedly inhibited the ability of control or active Akt1-expressing ECs to form lumens and tubes. Furthermore, active Akt1-, kRas-, and PIK3CA-expressing ECs showed strongly reduced pericyte recruitment and basement membrane deposition compared with control. Active Mek1-expressing ECs also had reduced pericyte recruitment number and basement membrane deposition compared with control. Jak2, Stat3, Mek1, PI3K, and mTOR signaling events controlled EC tip cell sprouting behavior, and EC expression of the activated versions of these kinases led to increased EC sprouting compared with controls. In contrast, EC expression of active Akt1 and kRas caused strongly reduced EC sprouting behavior. Pharmacologic blockade of Jak, Mek, PI3K, and mTOR kinases significantly reduced EC tip cell sprouting compared with control ECs and inhibited the hypersprouting phenotype of active Jak2- and Stat3-expressing ECs. Active Jak2-, Stat3-, and Rheb-expressing ECs showed reduced pericyte recruitment with overall normal (Jak2) or slightly reduced (Stat3 and Rheb) basement membrane deposition compared with control ECs. These data indicate that EC expression of oncogenic activating mutations results in varying degrees and severities of EC morphogenic abnormalities and capillary deficiency. The most severe defects were associated with those activating mutations that significantly altered EC lumen formation mechanisms (ie, Akt1, kRas, Mek1, and PIK3CA) coupled with reduced pericyte recruitment and basement membrane deposition. Overall, these studies identified new molecules and signaling mechanisms controlling EC lumen formation versus sprouting behavior, characterize their relationship to pericyte recruitment and pericyte-induced capillary tube maturation. These data offer new insights into the underlying molecular basis for the development of vascular malformations.

Materials and Methods

All supporting data are available within the article (and its Supplemental Material). Materials used in this study are compiled and listed (Tables 1 and 2 and Supplemental Table S1).

Table 1.

Materials: Antibodies

Target antigen	Vendor or source	Catalog no.	Working concentration	Molecular weight, kDa	Source
Actin (Ab-1) JLA20	Calbiochem (San Diego, CA)	CP01	WB (0.1 μg/mL)	42	Mouse mAb
Akt	Cell Signaling Technology (Danvers, MA)	9272	WB (0.5 μg/mL)	60	Rabbit
B-Raf (L12G7)	Cell Signaling Technology	9434	WB (0.5 μg/mL)	86	Mouse mAb
C-Raf	Cell Signaling Technology	9422	WB (0.5 μg/mL)	65-75	Rabbit
GSK-3β (27C10)	Cell Signaling Technology	9315	WB (0.5 μg/mL)	46	Rabbit mAb
Histone H3 (D1H2) XP	Cell Signaling Technology	4499	WB (0.5 μg/mL)	17	Rabbit mAb
p44/42 MAPK (Erk1/2)	Cell Signaling Technology	9102	WB (0.5 μg/mL)	42 and 44	Rabbit
Pak2 (C17A10)	Cell Signaling Technology	2615	WB (0.5 μg/mL)	61	Rabbit mAb
Pak4	Cell Signaling Technology	3242	WB (0.5 μg/mL)	72	Rabbit
PKCε	BD Transduction Laboratories (Franklin Lakes, NJ)	610085	WB (0.5 μg/mL)	90	Mouse
Src (32G6)	Cell Signaling Technology	2123	WB (0.5 μg/mL)	60	Rabbit mAb
Tubulin (Lys40)-acetylated	Millipore (Burlington, MA)	ABT241	WB (0.5 μg/mL)	50	Rabbit
Tubulin-detyrosinated	Abcam (Waltham, MA)	ab48389	WB (0.5 μg/mL)	51	Rabbit
Mek1/2	Cell Signaling Technology	9122	WB (0.5 μg/mL)	45	Rabbit
Stat3 (79D7)	Cell Signaling Technology	4904	WB (0.5 μg/mL)	79 and 86	Rabbit
p38 MAPK (D13E1) XP	Cell Signaling Technology	8690	WB (0.5 μg/mL)	40	Rabbit mAb
Erm	Cell Signaling Technology	3142	WB (0.5 μg/mL)	75	Rabbit
Rho	Epitomics (Burlingame, CA)	1667-1	WB (0.5 μg/mL)	22	Rabbit
Rasip1	Sigma Prestige (St. Louis, MO)	HPA077251	WB (0.5 μg/mL)	IHC (115)	Rabbit
CDCP1	Cell Signaling Technology	13794	WB (0.5 μg/mL)	230	Rabbit
VEGFR2	Cell Signaling Technology	5168	WB (0.5 μg/mL)	140	Rabbit
Phosphorylated Akt (Thr308) (D25E6) XP	Cell Signaling Technology	13038	WB (0.5 μg/mL)	60	Rabbit mAb
Phosphorylated B-Raf (Ser445)	Cell Signaling Technology	2696	WB (0.5 μg/mL)	86	Rabbit
Phosphorylated C-Raf (Ser338) (56A6)	Cell Signaling Technology	9427	WB (0.5 μg/mL)	74	Rabbit mAb
Phosphorylated GSK-3β (Ser9)	Cell Signaling Technology	9336	WB (0.5 μg/mL)	46	Rabbit
Phosphorylated histone H3 (Ser10) (6G3)	Cell Signaling Technology	3377	WB (0.5 μg/mL)	17	Mouse mAb
Phosphorylated p44/42 MAPK (Thr202/Tyr204)	Cell Signaling Technology	4370	WB (0.5 μg/mL)	42 and 44	Rabbit mAb
Phosphorylated Pak2 (Ser20)	Cell Signaling Technology	2607	WB (0.5 μg/mL)	61-67	Rabbit
Phosphorylated Pak4/5/6 (S474/S602/S560)	Cell Signaling Technology	3241	WB (0.5 μg/mL)	72, 82, and 90	Rabbit
Phosphorylated PKCε (Ser729)	Millipore	06-821-I	WB (0.5 μg/mL)	87	Rabbit
Phosphorylated Src family (Tyr416)	Cell Signaling Technology	2101	WB (0.5 μg/mL)	60	Rabbit
Phosphorylated p38 MAPK (Thr180/Tyr182)	Cell Signaling Technology	9211	WB (0.5 μg/mL)	43	Rabbit
Phosphorylated ezrin (Thr567)/radixin (Thr564)/moesin (Thr558)	Cell Signaling Technology	3141	WB (0.5 μg/mL)	75 (Moesin)	Rabbit
Phosphorylated CDCP1	Cell Signaling Technology	14965	WB (0.5 μg/mL)	230	Rabbit
Rabbit anti-goat HRP	Dako-Agilent (Santa Clara, CA)	P0449	WB (0.7 μg/mL)
Goat anti-rabbit HRP	Dako-Agilent	P0448	WB (0.7 μg/mL)
Rabbit anti-mouse HRP	Dako-Agilent	P0260	WB (0.7 μg/mL)
CD31 (PECAM-1) (D8V9E) XP	Cell Signaling Technology	77699	IF (1:200)		Rabbit
Collagen I	Sigma (St. Louis, MO)	C2456	IF (1:120)		Mouse
Collagen IV	Developmental Studies (Iowa City, IA)	M3F7 hybridoma	IF (1:200)		Mouse
Fibronectin	Sigma	F0791	IF (1 μg/mL)		Mouse
Nidogen 1	R&D Systems (Minneapolis, MN)	AF2570	IF (5 μg/mL)		Goat
Nidogen 2	R&D Systems	AF3385	IF (5 μg/mL)		Goat
Laminin	Sigma	L9393	IF (1:200)		Rabbit
Perlecan	Invitrogen (Waltham, MA)	134400	IF (2.5 μg/mL)		Mouse
Alexa Fluor 488 Goat Anti-Mouse IgG	Invitrogen	2140660	IF (10 μg/mL)
Alexa Fluor 594 Goat Anti-Rabbit IgG	Invitrogen	2268327	IF (10 μg/mL)
Alexa Fluor 488 Goat Anti-Rabbit IgG	Invitrogen	2284594	IF (10 μg/mL)
Alexa Fluor 594 Goat Anti-Mouse IgG	Invitrogen	2192307	IF (10 μg/mL)

B-Raf, B-rapidly accelerated fibrosarcoma; CDCP1, CUB domain-containing protein 1; C-Raf, C-rapidly accelerated fibrosarcoma; Erk, extracellular signal-regulated kinase; Erm, ezrin/radixin/moesin; GSK, glycogen synthase kinase; HRP, horseradish peroxidase; IF, immunofluorescence; mAb, monoclonal antibody; MAPK, mitogen-activated protein kinase; Mek, mitogen-activated protein kinase; Pak, p21-activated kinase; PECAM, platelet endothelial cell adhesion molecule; PKC, protein kinase C; Rasip1, Ras-interacting protein 1; VEGFR2, vascular endothelial growth factor receptor 2; WB, Western blot analysis.

Table 2.

Materials: PCR Primers

Real-time PCR primers and reagents
Gene name/reagents	Vendor	Catalog no.
GAPDH	Thermo Fisher (Waltham, MA)	Hs02786624_g1
PDGFB	Thermo Fisher	Hs00966522_m1
PDGFD	Thermo Fisher	Hs00228671_m1
EDN1	Thermo Fisher	Hs00174961_m1
HBEGF	Thermo Fisher	Hs00181813_m1
TGFB1	Thermo Fisher	Hs00998133_m1
TaqMan Fast Advanced Master Mix	Thermo Fisher	4444557

PCR primers
Gene name	Forward primer	Reverse primer
GAPDH	5′-AAGGTGAAGGTCGGAGTCAACG-3′	3′-CATGAGGTCCACCACCCTGTTG-5′
KRAS	5′-AGCTCGAGCTATGACTGAATATAAACTTGTGGTAG-3′	3′-AGGGATCCTTACATAATTACACACTTTGTCTTTG-5′
MAP2K1	5′-GACATCTGGAGCATGGGACT-3′	3′-CCCAACCTGCAAAATCCACTT-5′
PIK3CA	5′-GTCAATCGGTGACTGTGTGG-3′	3′-GTCAAAACAAATGGCACACG-5′
AKT	5′-GAAGCTGGAGAACCTCATGC-3′	3′-CATGATCTCCTTGGATCCT-5′
RHEB	5′-GGCCAATTTGTGGACTCCTA-3′	3′-TCGAAAAACATCCACAGCAG-5′
JAK2	5′-TTGTGCACGGATGGATAAAA-3′	3′-TCACCTGAAGGACCACTTCC-5′
STAT3	5′-GGAGGAGTTGCAGCAAAAAG-3′	3′-TGTGTTTGTGCCCAGAATGT-5′

Cell Culture

Human umbilical vein ECs were purchased from Lonza (Hayward, CA) and used from passages 3 to 6, as previously described.³⁷ Human brain vascular pericytes were also purchased from Lonza and were passaged from passages 4 to 12. Both cell types were cultured on gelatin-coated flasks and grown in the authors’ own supermedia, with M199 as a base, 20% fetal bovine serum, bovine hypothalamus extract, heparin sodium salt, gentamicin, and amphotericin B, as described.³⁷ Cells were grown in incubators set at 37°C and 5% CO₂.

Development of EC Active Mutation Cell Lines

A set of vectors containing cDNAs of oncogenic signaling pathways was obtained from Addgene (Watertown, MA) (kit number 1000000072).⁶⁹ kRas (G12V)-pcw107, MEK1 (S218D and S222D)-pcw107, myr-FLAG-PIK3CA-pcw107, myr-FLAG-AKT1-pcw107, FLAG-Rheb (Q64L)-pcw107, JAK2 (V617F)-pcw107-V5, and Stat3 (A662C, N664C, and V667L)-pcw107-V5 lentiviral vectors were each a gift from David Sabatini and Kris Wood (Supplemental Table S2). This lentiviral system uses the human phosphoglycerate kinase promoter, which induces modest increases in the levels of the genes produced (Supplemental Figure S1). The lentiviruses were produced using the 293FT lentiviral packaging cell after transfection with Lenti-X single shots (ie, packaging vectors) along with the above individual lentiviral vectors carrying the activating mutant molecules. Lentiviral supernatants were collected according to the manufacturer's protocol and as previously described.⁷⁰ Confluent human umbilical vein ECs grown in T75 flasks were infected with lentivirus at the following ratio: 6 mL of cell culture media, 4 mL lentiviral supernatant, and 10 μL of polybrene (hexadimethrine bromide) at a final concentration of 10 μg/mL. Cells were infected for 5 hours in this media mixture. Afterward, the lentiviral supernatant mixture was aspirated, and 14 mL of supermedia was added to the flask. Cells were incubated for 72 hours before the aspiration of media and initiation of stable cell line selection using puromycin (Invitrogen, Waltham, MA) at a final concentration of 2.14 μg/mL. Cells were cultured in the presence of puromycin for 10 to 14 days to ensure stable cell line selection, in which multiple clones of ECs, for each construct, were cultured together to obtain the ECs expressing the activating mutations. These EC lines were frozen in liquid nitrogen with multiple vials at the lowest passages possible. The active mutation EC lines were used in the bioassays from passages 3 to 8.

Transfection of ECs with siRNA

siRNA suppression was performed as previously described using Silencer Select siRNAs.¹⁰ In some cases, protein lysates were collected and subjected to Western blot analysis to confirm functional knockdown of the genes of interest.

Pharmacologic Inhibitors and Concentrations

For vasculogenic assays, ruxolitinib was added at 7.5 μmol/L, PD98059 was added at 10 μmol/L, U0126 was added at 2.5 μmol/L, and PLX8394 was added at 10 μmol/L. For aggregate assays, ruxolitinib was added at 15 μmol/L, SD1008 was added at 5 μmol/L, rapamycin was added at 0.5 μmol/L, pictilisib was added at 0.25 μmol/L, MK2206 was added at 0.5 μmol/L, and everolimus was added at 1 μmol/L, For sprouting assays, ruxolitinib was added at 10 μmol/L, SD1008 was added at 5 μmol/L, PD98059 was added at 10 μmol/L, U0126 was added at 2.5 μmol/L, pictilisib was added at 0.25 μmol/L, and everolimus was added at 1 μmol/L. For vasculogenesis experiments with active Akt1-, Stat3-, and JAK2-expressing EC lines, ruxolitinib was added at 10 μmol/L, Frax486 was added at 5 μmol/L, Frax597 was added at 5 μmol/L, SD1008 was added at 5 μmol/L, PD98059 was added at 15 μmol/L, PP2 was added at 5 μmol/L, U0126 was added at 5 μmol/L, rapamycin was added at 5 μmol/L, pictilisib was added at 5 μmol/L, PLX8394 was added 5 μmol/L, everolimus was added at 5 μmol/L, and Go6983 was added at 10 μmol/L. All drug concentrations were tested for optimal dosing.

3D Collagen Assays

ECs were suspended in 2.5 mg/mL collagen type I matrices, and assays were performed as described.⁷¹ With the exception that the culture media contained reduced serum supplement, ascorbic acid, FGF-2, SCF at 40 ng/mL, and IL-3 were added at 40 ng/mL. SDF-1α was added at 200 ng/mL into collagen type I matrices.⁶ Three types of 3D collagen assays were used: vasculogenic, angiogenic sprouting, and aggregate (Figure 1). In the vasculogenic assay, ECs were resuspended and added to the gel as single cells before gel polymerization. The authors used this assay model to perform real-time videos using the active mutation ECs versus control ECs. In the angiogenic sprouting assay, ECs were seeded as a monolayer on top of the polymerized gel. In the aggregate assay, ECs were aggregated and suspended in the gel as aggregates before gel polymerization. For all assays, medium is added on top of the gel after it has polymerized for 30 minutes in the incubator at 37°C, as previously described.²¹ Cultures were incubated at 37°C in serum-free defined media and allowed to assemble over time.²¹ Cultures were fixed at specific time points with 3% glutaraldehyde before staining with 0.1% toluidine blue in 30% methanol for nonfluorescent visualization or fixed in 3% paraformaldehyde for immunofluorescence imaging.³⁷

EC-Pericyte Co-Culture

The ECs (human umbilical vein ECs or active mutation ECs) and green fluorescent protein–labeled pericytes were co-cultured in 2.5 mg/mL type I collagen matrices in a ratio of 1:5. ECs and pericytes were suspended as single cells in the collagen gel (Figure 1). After gel polymerization, the media with five growth factors (reduced serum supplement with insulin, FGF-2, SCF, IL3, and SDF) were added to the EC-pericyte co-cultures.³⁷ Co-cultures were fixed with paraformaldehyde and were then immunostained for CD31, key basement membrane matrix components, and collagen type I. Confocal imaging was then performed.

Western Blot Analysis

The cultures (collagen gels) were plucked and dissolved in a 1.5% sample buffer with 5% β-mercaptoethanol. A total of 15 μL of the sample was pipetted into a Biorad (Hercules, CA) gel and run until the ladder reached the bottom. The proteins were transferred onto a polyvinylidene difluoride membrane. After blocking, specific primary antibodies (1:1000 dilution) were added directly into blocking buffer and incubated overnight at 4°C. After washing with Tris-buffered saline with 0.1% Tween 20, the corresponding horseradish peroxidase antibody was used to detect the protein. The results were acquired by X-ray films. Additional protein samples of select EC lines were also collected at 72 hours and subjected to Western blot analyses for protein quantification analysis. Following the same blotting protocol mentioned above, the results were collected using a Biorad Gel Doc Imager XR and quantified using ImageJ software version 1.45f (NIH, Bethesda, MD; https://imagej.nih.gov/ij), n ≥ 3.

Gene Expression Analysis of Control and ECs Expressing Activating Mutations

Total RNA was extracted from endothelial cells using Direct-zol RNA MiniPrep Kit (Zymo, Irvine, CA; 2050) and reverse transcribed into cDNA using ProtoScript First Strand cDNA Synthesis Kit (New England Biolabs, Ipswich, MA; E6300S). To quantify levels of transcript expression, real-time PCR was performed using QuantStudio 3 Real-Time PCR System (Applied Biosystems, Waltham, MA) and TaqMan Real-Time PCR Assays (Thermo Fisher Scientific, Waltham, MA). Relative gene expression was normalized to glyceraldehyde-3-phosphate dehydrogenase levels and calculated using the ΔΔC_T method. Active mutation EC line validation was performed using RT-PCR. The PCR products were subjected to agarose gel electrophoresis and imaged using a Biorad Gel Doc Imager XR.

Microscopy and Imaging

Toluidine blue nonfluorescent images were taken on an Olympus CKX41 microscope (Olympus, Center Valley, PA) with 10× objective. Lumen area and tip cell number were quantified from toluidine blue staining images using Metamorph software version 7.8 (Molecular Devices, San Jose, CA), as previously described.⁷² Confocal images were obtained using a Leica SP8 LIGHTNING White light laser confocal scanning microscope. Confocal images were generated using LAS X (Leica Microsystems, Deerfield, IL) and Fiji (ImageJ version 1.45f). EC-only collagen type I confocal images were imaged at 40× magnification. EC-pericyte co-culture staining of CD31, basement membrane matrix proteins, and collagen type I was imaged at 10× magnification. For the immunostaining, staining of all cell lines for each target set (ie, laminin, collagen IV, fibronectin, perlecan, nidogen 1, and nidogen 2) was performed at the same time and abided to the same staining protocol. During the imaging of the basement membrane components, microscope settings were kept constant, and all stained cultures within each target group were imaged at the same time to ensure uniformity among the images and conditions.

Statistical Analysis

Individual data points were obtained from triplicate wells and a minimum of 12 independent fields from these wells in both vasculogenic and aggregate assays, whereas an n ≥ 3 from triplicate wells was acquired for angiogenic tip cell sprouting counts. Also, all bioassays were performed three or more times, and the results shown are typically derived from a representative experiment. For the EC tube branching point analysis, tube intersections (where two tubes meet) per field were quantified using Metamorph software version 7.8. Statistical significance was set at minimum, with P < 0.05. t-Tests were used when analyzing two groups within a single experiment. Cross comparison of means between multiple conditions within a given experiment was analyzed using analysis of variance with follow-up post hoc Tukey tests in Prism 8 (GraphPad, Boston, MA).

Results

ECs Expressing Activated kRas, Mek1, and Akt1 Demonstrate Accelerated Lumen and Tube Formation

A 3D serum-free–defined factor system developed in house contains the growth factors, SCF, IL-3, SDF-1α, FGF-2, and insulin, termed the factors.^4,6,12,68 The factors promote both EC sprouting behavior and lumen formation, two critical EC steps in vascular morphogenesis.^21,22 Factor addition also stimulates the ability of ECs to attract pericytes so that capillary tube networks can be established and mature, as evidenced by basement membrane matrix deposition.^6,12,28 The current study focused on three major downstream signaling pathways [ie, Jak-Stat, Ras–mitogen-activated protein kinase (Mapk), and PI3K-Akt-mTOR] that appear to regulate and affect the factor system. In addition, these pathways have known oncogenic activity and can induce the development of tumors. In a few cases, when expressed in ECs, they can cause the development of various vascular anomalies.^44,64 In this latter case, vascular anomalies can be associated with abnormal morphogenic responses, thereby affecting the normal assembly of the vasculature.

Thus, the bioassay models and platform (Figure 1) were used to assess the functional role of activating mutations within these pathways on the key steps in capillary assembly. Stable EC lines expressing active kRas, Mek1, PIK3CA (a catalytic subunit of PI3K), Akt1, Rheb (a small GTPase that activates mTOR), Jak2, and Stat3 were generated using lentiviral vectors (Supplemental Table S2). Each of the active mutation genes showed elevated expression within ECs as confirmed using RT-PCR (Supplemental Figure S1). Subsequently, control and all active mutation EC lines were subjected to various functional assays derived from the factor system (Figure 1). This experimental strategy grants the ability to systematically pair specific molecules and signals to a given EC morphogenic outcome, such as lumen formation, sprouting behavior, pericyte recruitment, or pericyte-induced basement membrane deposition.

To evaluate the impact of activating mutations on EC lumen formation, all cell lines were subjected to vasculogenic assays. Lumen area analysis demonstrate that active kRas-, Mek1-, and Akt1-expressing EC lines form significantly elevated lumen area, whereas active Jak2, Rheb, and Stat3 show similar lumen formation ability compared with control, and the active PIK3CA-expressing ECs demonstrated significantly reduced lumen area (Figure 2, A and B). Real-time videos of the active mutation expressing ECs over 72 hours were made and quantitated to aid in this analysis (Figure 3 and Supplemental Videos S1–S11). Accelerated and increased lumen formation was observed using the active kRas, Mek1, and Akt1 EC lines over time, compared with control ECs, whereas the active PIK3CA ECs exhibited significantly reduced lumen area (Figure 3 and Supplemental Videos S1–S8). Toluidine blue–stained images of the 72-hour vasculogenic assay revealed that activated kRas and Akt EC lumens were wider than control (Figure 2A). A detailed lumen width analysis showed that lumens from ECs expressing either active kRas or Akt1 were almost twice as wide compared with control, whereas they were more narrow from ECs expressing either active Stat3 or PIK3CA (Figure 2, A and C). EC tube branching behavior among these different lines and morphogenic responses was evaluated (Figure 2D). The markedly excessive lumen formation ability of active kRas and Akt1 mutation ECs resulted in significantly less branched tube networks compared with control ECs (Figure 2D). The active PIK3CA-expressing ECs, which formed cysts, also showed significantly fewer tube branch points compared with control ECs (Figure 2D). These overall conclusions were also supported by immunostaining of the collagen matrices from these 72-hour cultures (Figure 2E). These gels were stained with anti-collagen type I antibodies (red) to label the matrix used in the factor system, and anti-CD31 antibodies (green) to label the ECs. Confocal images revealed the presence of vascular guidance tunnels, which represent physical spaces carved out within the collagen matrix by the ECs during lumen and tube formation (Figure 2E). Control ECs generated narrow, elongated, and branched tunnel spaces within the collagen matrix. In contrast, ECs carrying either active Akt1 or kRas expression showed much wider, less elongated, and less branched tunnel spaces, which reflect the markedly increased lumen-forming ability of these ECs.

Figure 2.

Endothelial cells (ECs) expressing activated (Ac) Kirsten rat sarcoma virus (kRas), mitogen-activated protein kinase 1 (Mek1), and Akt serine/threonine kinase 1 (Akt1) demonstrate markedly accelerated lumen and tube formation, compared with control ECs. Control ECs and ECs carrying active Janus kinase 2 (Jak2), signal transducer and activator of transcription 3 (Stat3), kRas, Mek, phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA), Akt1, and Ras homolog enriched in brain (Rheb) mutations were cultured in 3D vasculogenesis assays. Cultures were fixed with glutaraldehyde at 72 hours and were subsequently stained with toluidine blue and imaged. A: Representative images for each cell line are shown. Black arrowheads indicate EC tube structures. B: Lumen area of the vasculogenic assay was quantified, and t-test was used to determine significance compared with control. C: Lumen width of vasculogenic assay was quantified, and t-test was used to determine significance compared with control. D: EC tube branching points in the vasculogenic assay were quantified, and a t-test was used to determine significance compared with control. E: Cultures of control, active kRas-, and Akt1-expressing ECs were fixed with paraformaldehyde at 72 hours and were immunostained with antibodies to CD31 (green) to label ECs and collagen type I (red) to visualize the vascular guidance tunnels generated during the tube morphogenic process. Representative images are shown, and white arrows indicate the borders of vascular guidance tunnels. n = 15 (B–D). ∗∗P ≤ 0.01, ∗∗∗P ≤ 0.001 compared with control. Scale bars: 200 μm (A); 50 μm (E).

Figure 3.

Activated (Ac) Kirsten rat sarcoma virus (kRas), mitogen-activated protein kinase 1 (Mek1), and Akt serine/threonine kinase 1 (Akt1) expression in endothelial cells (ECs) significantly promotes lumen formation over time, whereas active phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) expression reduces lumen formation compared with control ECs, as analyzed from real-time videos. Control ECs and all active mutation cell lines were subjected to vasculogenic assays that were allowed to assemble over 72 hours, and individual fields were captured every 10 minutes. Real-time images were processed into videos. A–G: Still shots representing the time points of 0, 10, 20, 30, 40, 50, 60, and 72 hours of the indicated conditions were quantified for EC lumen area and compared with control ECs, and statistical significance was evaluated using a t-test. n = 10 for each time point (A–G). ∗∗∗P ≤ 0.001 compared with control. Jak2, Janus kinase 2; Rheb, Ras homolog enriched in brain; Stat3, signal transducer and activator of transcription 3.

Activating Mutations in kRas, Mek1 and Akt1 Demonstrate Accelerated Lumen Formation

The ability of the different EC cell lines to form lumens using a novel assay was addressed next by inhibiting EC sprouting behavior using the microtubule-stabilizing agent, epothilone B (Figure 4). Microtubule-stabilizing agents, such as paclitaxel, docetaxel, and epothilone B, completely inhibit EC sprouting ability without impacting the ability of ECs to form lumens.⁵ Therefore, 10 nmol/L epothilone B was used as an agent to block EC sprouting behavior in vasculogenic assays using the control versus activating mutation EC lines, which provides a novel approach to singularly evaluate lumen formation ability in the absence of sprouting behavior (Figure 4C). This led to an increased cystic lumen formation in the active kRas, Mek1, and Akt1 EC lines compared with that in control ECs (Figure 4, A and B). Interestingly, the active Mek1 line demonstrated the greatest stimulatory effect, which might occur because active Mek1 also stimulates EC sprouting behavior, which can oppose and balance lumen formation. In contrast, ECs expressing active PIK3CA, Rheb, Jak2, and Stat3 did not show significant changes in lumen formation ability following epothilone B treatment compared with control ECs (Figure 4, A and B). These findings provide further support for the ability of activated kRas, Mek1, and Akt1 to markedly stimulate the EC lumen formation process.

Figure 4.

Endothelial cell (EC) expression of active (Ac) Kirsten rat sarcoma virus (kRas), mitogen-activated protein kinase 1 (Mek1), and Akt serine/threonine kinase 1 (Akt1) promotes EC cystic lumen formation, compared with EC control, following treatment with the microtubule-stabilizing agent, epothilone B. Control ECs and all active mutation cell lines were subjected to vasculogenic assays and were allowed to assemble in the presence of epothilone B at 10 nmol/L. Cultures were fixed with glutaraldehyde at 24 hours and were subsequently stained with toluidine blue and imaged. A: Representative images for each cell line are shown. Arrowheads indicate cystic EC lumens. B: Cystic lumen area of the vasculogenic assay was quantified, and a t-test was used to determine significance compared with control. C: Schematic demonstrating the novel cystic lumen formation assay. Epothilone B at 10 nmol/L selectively blocks EC sprouting behavior, but not lumen formation, through microtubule stabilization. This assay allows for the specific and individual assessment of EC lumen formation. n = 15 (B). ∗∗∗P ≤ 0.001 compared with control. Scale bar = 200 μm (A). Jak2, Janus kinase 2; PIK3CA, phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; Rheb, Ras homolog enriched in brain; Stat3, signal transducer and activator of transcription 3.

Identification of Key Signaling Pathways that Regulate the Accentuated Lumen and Tube Formation Response of ECs Expressing Activated kRas and Akt1

Active kRas- and Akt-expressing ECs exhibited strongly accelerated lumen formation ability, whereas active Stat3-expressing ECs showed similar lumen formation ability compared with control ECs. Therefore, these ECs were compared during a time course of lumen and tube formation. To assess signaling pathways and identify important regulators that contribute to EC tubulogenesis, cell lysates were prepared over time, and Western blot analyses were performed to compare EC signals during these 3D tube morphogenic events (Figure 5). Active kRas and Akt1 ECs showed increased and more prolonged phosphorylation of PKCε, Src, Pak2, b-Raf, and c-Raf during tube formation compared with EC controls, which are known EC lumen formation regulators (Figure 5A).^10,73, 74, 75 Increased phosphorylation and total protein for CUB domain-containing protein 1 (CDCP1), a known activator of Src family kinases,^76,77 was observed in both active kRas and Akt ECs, compared with EC control and ECs carrying active Stat3 (Figure 5B). Western blot quantification of CDCP1 levels (total and phosphorylated) at 72 hours of culture indicated significantly elevated levels in the active k-Ras and Akt1 cultures compared with the control and active Stat3 EC cultures. Levels of vascular endothelial growth factor receptor 2 were equivalent in these different cultures (Figure 5B). Increased levels of CDCP1 correlated with the ability of ECs carrying activated kRas and Akt1 to form lumens in an accelerated manner. In addition, activation of extracellular signal-regulated kinase 1/2 and Mek1/2 was higher in active kRas-expressing ECs compared with the active Akt1-expressing ECs. As expected, phosphorylated Akt was much higher in the active Akt1 ECs compared with the active kRas-expressing ECs. In addition, the active Akt1 ECs showed increased phosphorylation of glycogen synthase kinase 3 beta (GSK3β) (phosphorylation inactivates GSK3β), which mirrored studies demonstrating a role for GSK3β inactivation during EC lumen formation.⁷⁸ Both types of ECs also showed increased levels of Rasip1, a known positive regulator of EC lumen formation.^24,79 Additional Western blot analysis data at 72 hours showed increased levels of phosphorylated PKCԑ, phosphorylated GSK3β, and phosphorylated Akt, and decreased levels of phosphorylated p38 Mapk, which was observed in the active Akt1 EC cultures compared with the other cultures (Supplemental Figure S2). Overall, these data suggest that the ECs carrying activating mutations of kRas and Akt1 appeared to drive accelerated lumen formation in overlapping, but also distinct ways. In contrast, the active Stat3 ECs showed reduced or similar levels of many of these lumen regulators, such as phosphorylated PKCε, phosphorylated b-Raf, and phosphorylated extracellular signal-regulated kinase. They also showed elevated levels of phosphorylated c-Raf compared with control ECs. Overall, the expression of activated Stat3 in ECs did not stimulate lumen formation compared with control ECs, suggesting that these signaling differences may be the causative reason for the morphologic differences between control and active kRas- and Akt1-expressing cell lines. The strong down-regulation of phosphorylated histone H3 (a key marker of cell proliferation) in all cases indicates that the morphologic differences observed between active kRas and Akt1 ECs compared with either control or active Stat3-expressing ECs were not the result of EC proliferation (Figure 5A). Furthermore, the real-time videos (Supplemental Videos S1–S11) did not show any evidence for proliferation from the different ECs during the EC tubulogenic process, which is consistent with previous findings showing that proliferation ceases during EC tube network assembly.^20,80,81

Figure 5.

Identification of key signaling pathways that regulate the accentuated lumen and tube formation response of endothelial cells (ECs) expressing active (Ac) Kirsten rat sarcoma virus (kRas) and serine/threonine kinase 1 (Akt1) versus an equivalent tube formation response of ECs expressing active Stat3 compared with control ECs. A: Vasculogenesis assays in three-dimensional collagen matrices were performed using control ECs, and ECs expressing active signal transducer and activator of transcription 3 (Stat3), kRas, and Akt1. Lysates were prepared from collagen gels from these cultures at the indicated time points (0, 6, 12, 24, 48, and 72 hours). Western blot analyses were performed to assess the expression of indicated molecules over time. B: Lysates were prepared from collagen gels from these cultures at 72 hours. Western blot analyses were performed, and results were captured using a Biorad imaging system. ImageJ version 1.45f was used to quantify protein expression. Representative Western blot analysis images for total (t-) and phosphorylated (p-) CUB domain-containing protein 1 (CDCP1), vascular endothelial growth factor receptor 2 (VEGFR2), and actin protein expression are shown. Statistical quantification was performed (t-test) for total and phosphorylated CDCP1 as well as VEGFR2 protein expression (normalized to actin). n = 3 (graphs in B). ∗P ≤ 0.05, ∗∗P ≤ 0.01, and ∗∗∗P ≤ 0.001 compared with control. Acetyl, acetylated; bRaf, b-rapidly accelerated fibrosarcoma; cRaf, c-rapidly accelerated fibrosarcoma; Detyr, detyrosinated; Erk, extracellular-signal related kinase; Erm, ezrin/radixin/moesin; GSK, glucagon synthase kinase; Mek, mitogen-activated protein kinase; Pak, p21-activated kinase; PKC, protein kinase C; Rasip1, Ras-interacting protein 1; Tub, tubulin.

Blocking PI3K-Akt-mTOR and Ras-MAPK Signaling Reduces, whereas Enhancing PI3-Akt Signaling Increases EC Lumen Formation

On the basis of the new findings showing that activated kRas, Akt1, and Mek1 are lumen-driving signals in ECs (Figure 2), a panel of pharmacologic inhibitors was selected to test the effect of these kinases (Supplemental Table S3) and pathways on normal EC lumen formation (Supplemental Figure S3). PI3K and mTOR were also selected as luminal targets in the screen because others have reported their relation to cystic lumen formation in CCMs.^39,43 Vasculogenic and EC aggregate assays were performed to evaluate these results. Results from vasculogenic assays confirmed that pharmacologic blockers targeting the Ras-MAPK pathway, including PLX8394 (bRaf and cRaf inhibitor) as well as U0126 and PD98059 (Mek inhibitors) significantly reduced EC lumen area compared with controls (Supplemental Figure S3, A and B). Results from aggregate assays showed that pharmacologic blockers targeting the PI3K-Akt-mTOR pathway, including pictilisib (PI3K inhibitor), MK-2206 (Akt inhibitor), and rapamycin and everolimus (mTOR inhibitors), significantly reduced EC lumen area compared with control conditions (Supplemental Figure S3, C and D). Furthermore, siRNA knockdown of PTEN and PP2A, two critical phosphatases that negatively regulate the PI3K-Akt signaling pathway, resulted in increased lumen formation compared with control (Supplemental Figure S3, E and F). siRNA knockdown of these targets was confirmed using Western blot analyses (Supplemental Figure S3G). Overall, these pharmacologic inhibitor and siRNA suppression data, coupled with the EC-activating mutation findings, provide strong support for the involvement of both the Ras/Mapk and the PI3K/Akt/mTOR pathways in EC lumen formation, whereas the Jak/Stat pathway does not appear to be a major regulator of this process.

Inhibition of PKCε, Src Family Kinase, Pak, Raf, Mek, and mTOR Kinases Blocks Lumen and Tube Formation in Active Akt1-Expressing ECs

Involvement of the PI3K/Akt/mTOR pathway in EC lumen formation was assessed next. Vasculogenic lumen-forming assays were used to pharmacologically block the ECs expressing active Akt1. Western blot analysis indicated key molecules and signaling pathways altered by the expression of active Akt1 in ECs, which markedly accentuated their ability to form lumen structures (Figure 5, A and B, and Supplemental Figure S2). Subsequently, the study used these active Akt1 ECs and the various pharmacologic inhibitors to address the functional role of these identified targets as well as known regulators of lumen formation (Supplemental Table S3). Previous studies have reported the effects of these drugs on the formation and stability of lumens and tubes formed by ECs expressing active kRas.²³ Here, the addition of PP2 (Src family kinase inhibitor), Frax486 and Frax597 (Pak inhibitors), PLX8394 (Raf kinase inhibitor), U0126 and PD98059 (Mek kinase inhibitor), and Go6983 (an inhibitor of PKCε) blocked lumen formation of ECs expressing active Akt1 (Supplemental Figure S4, A and B). These results support past results showing that PKCε, Src family kinases, Raf, Mek, and extracellular signal-regulated kinase are critical regulators of lumen formation.^{4,6,11,22,23,73,79} In addition, everolimus and rapamycin also significantly reduced lumen and tube formation of these ECs (Supplemental Figure S4, A and B). The three most potent blockers of lumen formation from active Akt1-expressing ECs were directed to PKCε (Go6983) or mTOR (everolimus and rapamycin).

EC Expression of Active PIK3CA and Akt1 Leads to the Appearance of Cystic Luminal Structures during Vascular Morphogenesis

Screening of the lumen-forming ability of the different EC lines indicated cystic lumen structures exclusively seen in active PIK3CA- or Akt1-expressing ECs and not in other ECs (Figure 6A). Quantitation of EC lumen cyst number per field is shown to further demonstrate this result (Figure 6B). Interestingly, growth of CCMs, which represent cystic vascular lesions, requires increased signaling through the PI3K-Akt-mTOR pathway and can involve activating mutations in PIK3CA.^39,43,82, 83, 84 Real-time videos of the active PIK3CA-expressing ECs undergoing morphogenesis demonstrated their ability to form cystic lumen structures (Supplemental Videos S5 and S6).

Figure 6.

Endothelial cell (EC) expression of active (Ac) phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) or Akt serine/threonine kinase 1 (Akt1) leads to appearance of cystic luminal structures during vascular morphogenesis. Control ECs and ECs carrying active Kirsten rat sarcoma virus (kRas) or PIK3CA mutation were cultured in vasculogenesis assays. Cultures were fixed with glutaraldehyde at 72 hours and were subsequently stained with toluidine blue and imaged. A: Representative images for each cell line are shown. Black arrowheads indicate luminal cysts, whereas the black arrows indicate narrow tube and luminal structures. B: Cyst number per field was quantified, and a t-test was used to determine significance compared with control. n = 15 (B). ∗∗∗P ≤ 0.001 compared with control. Scale bars: 200 μm (left panels); 20 μm (three right panels). Original magnifications: ×20 (left panels); ×60 (three right panels). Jak2, Janus kinase 2; Mek1, mitogen-activated protein kinase 1; Rheb, Ras homolog enriched in brain; Stat3, signal transducer and activator of transcription 3.

Pharmacologic inhibitor screen with the active Akt1-expressing ECs (Supplemental Figure S4) was used to quantitate the number of cystic lumens per field (Supplemental Figure S4C). The two mTOR inhibitors (rapamycin and everolimus), and the PKCε inhibitor (Go6983) markedly blocked cyst formation. Interestingly, several inhibitors enhanced cyst formation, including two Mek inhibitors (U0126 and PD98059) and a Pak2 inhibitor (Frax597), which may be because of their ability to block sprouting behavior to a greater extent than the ability of lumen formation.

Activated Jak2, Stat3, PIK3CA, Rheb, and Mek1 Increase, whereas Activated kRas and Akt1 Reduce Sprouting Responses

The sprouting assay model systems were used to evaluate the effects of activated signals of ECs on sprouting behavior (Figure 1). Active Jak2, Stat3, PIK3CA, Rheb, and Mek1 expression in ECs significantly promoted sprouting, whereas active kRas and Akt1 expression in ECs significantly reduced sprouting (Figure 7, A and B). Activated Akt1 expression had the greatest inhibitory effect on EC sprouting behavior (Figure 7B). These latter results with active kRas and Akt1 ECs raise the possibility that strong stimulation of one morphogenic outcome (ie, lumen formation) may lead to a similarly strong reduction in the opposing morphogenic outcome (ie, sprouting behavior).

Figure 7.

Active (Ac) Janus kinase 2 (Jak2)-, signal transducer and activator of transcription 3 (Stat3)-, phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA)-, Ras homolog enriched in brain (Rheb)-, and mitogen-activated protein kinase 1 (Mek1)-expressing endothelial cells (ECs) promote tip cell sprouting, whereas active Kirsten rat sarcoma virus (kRas) and Akt serine/threonine kinase 1 (Akt1) inhibit tip cell sprouting. A: All active mutation EC lines, as well as control ECs, were subjected to downward sprouting assays in three-dimensional collagen matrices. Assays were fixed at 16 hours and stained. Representative images are shown. Arrowheads indicate the position of the EC monolayer surface. B: Tip cell number per field in A was quantified, and a t-test was used to determine significance compared with control. C: Schematic demonstrating a novel assay for assessing EC sprouting ability by accelerating this response with PP2, an Src family kinase inhibitor that promotes EC sprouting. This assay allows for the identification of cell lines with inherent EC sprouting defects that cannot be rescued by the addition of PP2. D: PP2 (5 μmol/L) was added at 0 hours in a sprouting assay containing control ECs. Tip cell number per field of was quantified after 16 hours, and a t-test was used to determine significance compared with control. E: All active mutation EC lines, as well as control ECs, were subjected to downward sprouting assay in the presence of PP2 at 5 μmol/L. Assays were allowed to form for 16 hours before fixing and staining. EC sprouting number per field was quantified, and a t-test was used to determine significance compared with control. n = 12 (B); n = 4 (D and E). ∗P ≤ 0.05, ∗∗∗P ≤ 0.001 compared with control. Scale bars = 100 μm (A).

Blockade of Src family kinases enhanced EC sprouting behavior in conjunction with inhibition of EC lumen formation (Figure 7, C and D), as indicated previously.²² This was used to perform an additional assay to further assess the sprouting ability of each active mutation EC line versus control following addition of PP2, an Src family kinase inhibitor (Figure 7, C and D). Interestingly, the hypersprouting EC lines, including active Jak2, Stat3, PIK3CA, Rheb, and Mek1, no longer showed increased sprouting behavior compared with control ECs under these novel conditions (Figure 7E). In contrast, the activated kRas and Akt1 ECs demonstrated strongly decreased EC sprouting behavior (Figure 7B), even in the presence of PP2 (Figure 7E). This result further underlines the finding that EC expression of active Akt1 and kRas markedly inhibited EC sprouting behavior, whereas many other activating mutations did not.

Critical Role of Jak/Stat Signaling in EC Sprouting Behavior

To further address the signaling requirements for EC sprouting behavior, sprouting assays were performed using a series of pharmacologic inhibitors with control ECs (Figure 8, A–D). In addition, control ECs were treated with siRNAs directed to key components of the Jak/Stat pathway (Figure 8, E and F). First, ruxolitinib (a Jak inhibitor) was added to a vasculogenic assay with control ECs, which caused short, widened tubes with blunt ends compared with thin, elongated tubes in the untreated control (Figure 8A). Furthermore, ruxolitinib or SD1008 (Jak inhibitors) addition to the EC aggregate assay model led to strong blockade of sprouting (Figure 8, B and C). These two Jak inhibitors along with U0126 and PD98059 (Mek inhibitors) and everolimus (an mTOR inhibitor) also blocked EC sprouting behavior from a monolayer surface, whereas pictilisib (a PI3K inhibitor) did not (Figure 8D). siRNA suppression experiments were conducted to further evaluate the Jak-Stat pathway as an important driver of EC sprouting behavior. siRNA suppression of Jak1 and Jak2 individually and combined showed marked blockade of EC sprouting compared with control siRNA (Figure 8E). These results showed that Jak2 was the dominant isoform controlling the response (Figure 8E). siRNA suppression of Stat3 and Stat5A led to blockade of sprouting, whereas siRNAs directed to Stat1, Stat4, and Stat5B did not have an effect (Figure 8E). Confirmation of protein knockdown with the different siRNAs was determined using Western blot analyses (Figure 8F).

Figure 8.

Jak/Stat signaling is a critical regulator of endothelial cell (EC) sprouting behavior. A: Control (Con) ECs were seeded in vasculogenic assays in the presence or absence of ruxolitinib, a Janus kinase (Jak) inhibitor, for 48 hours before fixing and staining. Representative images are shown. White arrowheads indicate tip cell ends, whereas black arrowheads indicate blunt ends. B: Control ECs were seeded as aggregates in the aggregate assay in the presence or absence of two different Jak inhibitors for 24 hours before fixing and staining. Representative images are shown. C: Tip cell number and lumen area of the aggregate assay in B were quantified and, a t-test was used to determine significance compared with control. D: Control ECs were subjected to a downward sprouting assay in the presence or absence of the indicated pharmacologic agents for 16 hours before fixing, staining, and quantifying tip cell per field. Pictilisib is a phosphatidylinositol 3-kinase inhibitor, PD98059 and U0126 are mitogen-activated protein kinase (Mek) inhibitors, ruxolitinib and SD1008 are Jak inhibitors, and everolimus is a mammalian target of rapamycin inhibitor. A t-test was used to determine significance compared with control. E: Control ECs were subjected to siRNA knockdown (KD) of the indicated signaling targets. ECs were seeded in a sprouting assay after siRNA knockdown and were allowed to sprout for 16 hours before fixing, staining, imaging, and quantification. F: siRNA knockdown of each gene target was confirmed by Western blot analysis. n = 12 (C); n = 3 (D); n = 6 (E). ∗P ≤ 0.05, ∗∗P ≤ 0.01, and ∗∗∗P ≤ 0.001 compared with control. Scale bars = 200 μm (A and B). si, small interfering; Stat, signal transducer and activator of transcription.

Finally, a pharmacologic inhibitor panel was tested on active Jak2- and Stat3-expressing ECs to assess whether inhibition of these sprouting signals affected the hypersprouting phenotype of these ECs. Ruxolitinib, SD1008, U1026, and PD98059 significantly reduced sprouting behavior of active Jak2- or Stat3-expressing ECs (Supplemental Figure S5). Everolimus and rapamycin blocked tip cell sprouting from the active Jak2 but, interestingly, not from the active Stat3-expressing cell line. Pictilisib did the opposite and blocked sprouting from the Stat3 line, but not the Jak2 line (Supplemental Figure S5). Overall, these data indicate that Jak2, Stat3, Mek1, PIK3CA, and mTor are stimulatory regulators of EC sprouting responses, whereas active kRas and Akt1 suppress sprouting behavior.

EC Expression of Activating Mutations Causes Deficiencies in Capillary Assembly

Pericyte recruitment is a third critical step after EC tip cell sprouting and lumen formation in vascular morphogenesis, which promotes capillary assembly, maturation, and stability.^2,85,86 Pericyte loss is reported in important microvascular diseases, such as diabetic retinopathy. Lack of pericyte recruitment or poor association of pericytes with capillaries causes vascular destabilization and hemorrhages. Furthermore, reduced pericyte association with the vasculature plays a role in the pathogenic development of vascular malformations and malignant cancers.^29,87, 88, 89, 90 Because pericyte recruitment is necessary for capillary formation and maturation, whether EC expression of activated kRas, Mek1, PIK3CA, Akt1, Rheb, Jak2, and Stat3, in comparison to control ECs, affected their ability to communicate and associate with pericytes to form capillary tube networks was tested next (Figure 9). EC-pericyte co-cultures were fixed after 120 hours, and immunostained with anti-CD31 antibodies to visualize the EC tube networks and determine whether green fluorescent protein–labeled pericytes were associated with the control ECs or ECs carrying the different activated mutations. In addition, the study evaluated whether these different activated mutation EC-pericyte co-cultures were able to deposit basement membrane matrices compared with controls (Figure 10).

Figure 9.

Expression of activating mutations in endothelial cells (ECs) leads to unique capillary assembly deficiencies due to altered EC morphogenesis coupled with reduced pericyte recruitment. Control and all active (Ac) mutation ECs were seeded with green fluorescent protein–labeled pericytes in three-dimensional collagen matrices, and the co-cultures were fixed at 120 hours and immunostained with antibodies to CD31 (red). A: Confocal microscopy was performed, and representative images are shown. B: Co-cultures of all cell lines were quantitated for the percentage of pericytes associated with tubes, and significance was determined using analysis of variance. C: Real-time quantitative PCR analysis of the expression of EC-derived factors controlling pericyte recruitment (PDGFB, PDGFD, EDN1, HBEGF, and TGFB1) was performed from the control and activating mutation EC lines. Expression of the indicated genes was normalized to the housekeeping gene, GAPDH. A t-test was used to determine significance compared with control. n = 20 (B); n = 3 (C). ∗P ≤ 0.05, ∗∗P ≤ 0.01, and ∗∗∗P ≤ 0.001 compared with control; ^§§§P ≤ 0.001 compared with control's percentage of pericytes on tubes; ^†††P ≤ 0.001 compared with control's percentage of pericytes off tubes; ^###P ≤ 0.001 compared with active phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA)-expressing ECs' percentage of pericytes on tubes; ^¶¶¶P ≤ 0.001 compared with active kRas-expressing ECs' percentage of pericytes on tubes. Scale bar = 50 μm (A). Akt1, Akt serine/threonine kinase 1; Jak2, Janus kinase 2; kRas, Kirsten rat sarcoma virus; Mek1, mitogen-activated protein kinase 1; Rheb, Ras homolog enriched in brain; Stat3, signal transducer and activator of transcription 3.

Figure 10.

Endothelial cells (ECs) expressing activating mutations demonstrate reduced basement membrane matrix deposition in response to EC-pericyte tube co-assembly, leading to a failure in proper capillary tube assembly and maturation. Control ECs and active mutation-expressing ECs were seeded with green fluorescent protein–labeled pericytes in three-dimensional collagen matrices, and the co-cultures were allowed to establish capillary networks and undergo capillary maturation events, such as basement membrane deposition for 120 hours. Co-cultures were fixed and immunostained for the basement membrane matrix proteins, laminin, collagen type IV, fibronectin, perlecan, nidogen 1, and nidogen 2 (red). Confocal microscopy was performed to evaluate the immunostaining of these individual basement membrane components, and representative images are shown. Scale bars = 50 μm. Ac, active. Akt1, Akt serine/threonine kinase 1; Jak2, Janus kinase; kRas, Kirsten rat sarcoma virus 2; Mek1, mitogen-activated protein kinase 1; PIK3CA, phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; Rheb, Ras homolog enriched in brain; Stat3, signal transducer and activator of transcription 3.

Strikingly, active kRas-, Mek1-, and Akt1-expressing EC tube networks were abnormally wide compared with control EC tubes in these EC-pericyte co-cultures (Figure 9A). These three activating mutations led to accelerated lumen formation in EC-only cultures (Figures 2 and 3). This indicated that the presence of pericytes in these co-cultures did not rescue this accelerated lumen and tube formation phenotype. The same conclusion was reached by immunostaining the co-cultures with anti-collagen type I antibodies to visualize vascular guidance tunnels⁹¹ (Supplemental Figure S6). Pericytes are actively recruited within EC-generated vascular guidance tunnel spaces during EC-pericyte tube co-assembly events to promote tube maturation and stabilization.⁹¹ Widening of the tunnel spaces was observed with the active kRas-, Mek1-, and Akt1-expressing EC-pericyte co-cultures compared with controls. Reduced tunnel spaces were observed with the active PIK3CA-expressing EC-pericyte co-cultures (Supplemental Figure S6), which was also consistent with the results seen with EC-only cultures with reduced EC lumen area (Figures 2 and 3).

Quantitation of pericyte responses to the active mutation EC tubes in co-culture assays revealed marked defects in pericyte recruitment, particularly observed with active kRas, Akt1, and PIK3CA mutations, where approximately half of the pericytes were not associated with EC lined tubes (Figure 9B). Notably, of the top three pericyte recruitment deficient candidates, active Akt1-expressing ECs had significantly reduced pericyte recruitment compared with ECs expressing active kRas and PIK3CA. These data, represented as pericyte on/off EC tube ratios, showed that EC expression of each of these activated genes leads to significantly reduced pericyte recruitment, with the greatest reductions observed with active kRas-, Akt1-, and PIK3CA-expressing ECs (Figure 9B). Key EC-derived factors responsible for pericyte recruitment and basement membrane deposition during capillary formation were defined recently.³⁷ These EC factors are PDGF-BB, PDGF-DD, ET-1, HB-EGF, and TGF-β1. Blockade of these factors or their receptors in combination markedly interferes with pericyte recruitment and basement membrane deposition.³⁷ Because pericyte recruitment deficiencies are observed with each of the active mutation EC lines, quantitative PCR analysis was performed to detect differences in the expression of these factors in the active mutation carrying ECs compared with control ECs (Figure 9C). Of the known EC-derived factors controlling pericyte recruitment, active Akt1-expressing ECs showed substantially decreased mRNA expression of PDGFB, PDGFD, EDN1, and TGFB1 compared with control ECs. In line with this finding, the active Akt1-expressing ECs demonstrated the greatest pericyte recruitment defect in comparison to all other conditions (Figure 9B). They showed the greatest reductions in gene expression of EDN1 and PDGFB, which are critically important regulators of pericyte invasion and proliferation along with PDGFD (Figure 9C). The active PIK3CA-expressing ECs also demonstrated decreased PDGFB, PDGFD, and TGFB1 mRNA expression, whereas the active kRas-expressing ECs showed decreased expression of PDGFD and TGFB1. Furthermore, the active Mek1-, Rheb-, Jak2-, and Stat3-expressing ECs all showed decreased PDGFB and PDGFD expression compared with EC control (Figure 9C). Although these mRNA expression changes might account for the pericyte recruitment defects, there may be other reasons, such as degradation, increased uptake, or sequestration of the recruitment factors by the ECs expressing activating mutations.

EC Expression of Activating Mutations Leads to Decreased Pericyte-Induced EC Basement Membrane Matrix Deposition

A key step in pericyte-induced EC tube maturation and capillary network assembly is the deposition of the capillary basement membrane matrix, including laminins, collagen type IV, fibronectin, perlecan, nidogen 1 and nidogen 2.^{2,15,27,28,37,71,}92, 93, 94, 95 The ability of control versus active mutation-expressing EC tubes was compared with deposit basement membranes in EC-pericyte co-cultures (Figure 10). Control EC tube networks recruited pericytes and markedly assembled the capillary basement membrane matrix, including laminins, collagen type IV, fibronectin, perlecan, nidogen 1, and nidogen 2 (Figure 10 and Supplemental Figure S7). In contrast, the three EC types expressing activating mutations, which show the least pericyte recruitment (ie, kRas, PIK3CA, and Akt1) (Figure 9), also demonstrated the greatest reductions in basement membrane deposition (Figure 10 and Supplemental Figures S7 and S8). The active kRas-expressing EC tubes showed strong reductions in laminin, collagen type IV, and perlecan, as well as observable decreases in fibronectin, nidogen 1, and nidogen 2 deposition (Figure 10). Strong decreases in the deposition of collagen type IV, fibronectin, and perlecan were observed with the active PIK3CA ECs, whereas similar decreases in collagen type IV, fibronectin, perlecan, nidogen 1, and nidogen 2 deposition were seen with the active Akt1 ECs (Figure 10). In addition, the active Mek1-expressing EC tubes showed strong reductions in collagen type IV and perlecan deposition (Figure 10), whereas the active Rheb EC tubes showed reductions in the deposition of collagen type IV, fibronectin, and perlecan. Interestingly, the active Jak2 EC tubes showed mostly normal basement membrane deposition, except for a decrease in perlecan deposition, whereas the active Stat3 EC tubes showed decreased deposition of laminin, collagen type IV, fibronectin, perlecan, and nidogen 1. Overall, these findings demonstrated that EC expression of a wide variety of activating mutations results in deficiencies in capillary network assembly due to a combination of abnormal EC tube formation, reduced pericyte recruitment, and reduced basement membrane matrix deposition. The influence of each of the EC-expressed activating mutations along with their morphogenic consequences as well as their impact on EC-pericyte interactions and capillary basement membrane matrix assembly is summarized (Figure 11 and Table 3). The study demonstrates how their factor-derived bioassay platform can be used as a powerful investigative tool to link genes and signaling to vascular morphogenesis and downstream molecular effectors. These novel findings also provide key mechanistic information linking specific molecules and signaling pathways that control the processes of EC lumen formation, EC tip cell sprouting, pericyte recruitment to EC-lined tubes, and pericyte-induced capillary basement membrane deposition (Figure 11 and Table 3). Activating mutations in ECs and their effect on capillary morphogenesis and maturation provides novel mechanistic insights into the underlying reasons for the morphogenic and functional abnormalities observed in various types of vascular anomalies.

Figure 11.

Defining key signaling molecules that control endothelial cell (EC) lumen formation or EC tip cell sprouting behavior by assessing the functional influence of EC-expressed oncogenic activating mutations. A and B: Schematic diagrams illustrating major findings and conclusions of this work. A: a. EC expression of active Kirsten rat sarcoma virus (kRas) and Akt serine/threonine kinase 1 (Akt1) markedly accentuate lumen formation and block cell sprouting behavior, whereas active Akt1 also induces occasional cysts to form. b. EC expression of active mitogen-activated protein kinase 1 (Mek1) induces both lumen formation and EC sprouting behavior. c. EC expression of active Janus kinase 2 (Jak2), signal transducer and activator of transcription 3 (Stat3), and Ras homolog enriched in brain (Rheb) induces a selective increase in EC sprouting behavior without affecting lumen formation. d. EC expression of active phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) leads to increased EC sprouting behavior with decreased lumen formation, but also induced the formation of EC luminal cysts. B: Schematic diagram illustrating the findings in A with the added information showing that pericyte (green) recruitment is blocked following the expression of activating mutations that particularly affect EC lumen formation (red), including Akt1, kRas, PIK3CA, and Mek1, which leads to capillary deficiency. The spectrum of EC lumen formation abnormalities that were observed include excessively widened and less branched tubes (Akt1, kRas, and Mek1) or EC luminal cysts (PIK3CA and Akt1). These activating mutations are associated with vascular malformations in humans, including arteriovenous malformations (kRas and Mek1) and cerebral cavernous malformation (PIK3CA). mTOR, mammalian target of rapamycin; PI3K, phosphatidylinositol 3-kinase.

Table 3.

Summary of the Biological Effects of EC-Expressed Activating Mutations on Capillary Morphogenesis and Maturation

Activating mutation in ECs	Lumen area/lumen width	Sprouting behavior	Pericyte recruitment	Vascular guidance tunnel formation	Basement membrane formation	Formation of cystic lumen structures
Ac-kRas	↑/↑	↓	↓↓	↑↑	↓↓	ꟷ
Ac-Mek1	↑/ꟷ	↑	↓	↑	↓↓	ꟷ
Ac-PIK3CA	↓/↓	↑	↓↓	↓	↓↓	↑
Ac-Akt1	↑/↑	↓	↓↓	↑↑	↓↓	↑
Ac-Rheb	ꟷ/ꟷ	↑	↓	ꟷ	↓↓	ꟷ
Ac-Jak2	ꟷ/ꟷ	↑	↓	↑	ꟷ	ꟷ
Ac-Stat3	ꟷ/↓	↑	↓	ꟷ	↓↓	ꟷ

↑, Increase; ↑↑, strong increase; ↓, decrease; ↓↓, strong decrease; ꟷ, no effect; Ac, active; Akt1, Akt serine/threonine kinase 1; EC, endothelial cell; Jak2, Janus kinase 2; kRas, Kirsten rat sarcoma virus; Mek1, mitogen-activated protein kinase 1; PIK3CA, phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; Rheb, Ras homolog enriched in brain; Stat3, signal transducer and activator of transcription 3.

Discussion

This study sought to define the factor-induced EC signaling requirements that regulate each of the major steps in capillary assembly, including EC lumen and tube formation, sprouting behavior, pericyte recruitment, and pericyte-regulated capillary basement membrane deposition. The current study and previous work show that under serum-free–defined conditions, the addition of SCF, IL-3, SDF-1α, FGF-2, and insulin (factor system) is necessary to stimulate these critical capillary formative steps involving human ECs and pericytes.^4,6,28 These questions were addressed with an experimental approach using multiple strategies and bioassay models, which included expression of EC-activating mutations, siRNA suppression, pharmacologic inhibitors, mRNA expression analysis, and Western blot analysis to assess how EC signaling molecules and pathways controlled vascular morphogenesis and pericyte-induced capillary maturation (Figure 1). Specifically, the influence of these mutations and signaling molecules was investigated on the major steps in capillary assembly, including EC lumen and tube formation, EC sprouting behavior, EC-directed pericyte recruitment, and EC-pericyte interactions controlling capillary basement membrane deposition.

One of the key experimental approaches was to generate seven EC lines carrying active kRas, Mek1, PIK3CA, Akt1, Rheb, Jak2, and Stat3 and test each of those ECs versus controls in different assay model systems, including real-time video analysis. These results are summarized (Figure 11 and Table 3). Major findings reveal that ECs expressing active Akt1, kRas, or Mek1 formed excessively widened lumen and tube networks as well as vascular guidance tunnels. In support of these conclusions a novel assay was used to selectively suppress EC sprouting behavior by adding the microtubule-stabilizing agent, epothilone B. ECs expressing active kRas, Mek1, and Akt1, all demonstrated increased EC lumen formation compared with control and the other EC lines. In contrast, ECs expressing active Mek1, PIK3CA, Rheb, Jak2, and Stat3 demonstrated increased sprouting behavior compared with control ECs, whereas active Akt1 and kRas ECs showed reduced sprouting. The active Mek1 ECs were unique in their ability to stimulate both EC lumen formation and sprouting behavior compared with control ECs and the other activating mutation ECs. ECs expressing active PIK3CA showed significantly reduced overall lumen formation and vascular guidance tunnel spaces. However, these ECs as well as ECs expressing active Akt1 formed increased numbers of EC luminal cysts. In contrast, control ECs and the other EC lines expressing active mutations showed no ability to form cystic luminal structures. The recent finding that active PIK3CA is associated with cavernous malformations in CCM brain lesions is consistent with these observations. However, it is not sufficient to explain why expression of this activating mutation also led to increased sprouting behavior (Figure 7B), a phenomenon not observed in CCM lesions. Perhaps the loss of the CCM genes is also required in conjunction with active PIK3CA to control this phenotype. Previous studies have indicated that CCM gene knockdown on its own leads to hypersprouting phenotypes.²⁶ Because CCM genes also regulate EC lumen formation,^96,97 this hypersprouting phenotype is similar to that observed by blocking Src kinase, Notch signaling, or suppression of Rasip1 expression (all EC lumen regulators), which can all lead to increased EC sprouting behavior.^21,22,25 These questions need to be investigated further to be able to understand the reasons for these CCM disease phenotypes with cystic lumens and a lack of sprouting behavior. In the current study, ECs expressing Akt1 or kRas, but not the others that were examined, strongly suppressed EC sprouting behavior (Figure 7B). Even by imposing treatments on the ECs that stimulate sprouting behavior (ie, Src family kinase inhibition with PP2), the ECs expressing active Akt1 or kRas still demonstrated markedly inhibited sprouting phenotypes, compared with controls and the other activating mutations (Figure 7E). Because Akt1 activation is downstream of PIK3CA signaling, it may explain why active PIK3CA could enhance EC cystogenesis.

To date, there has been modest progress in delineating the signaling basis for EC sprouting behavior. The current study considerably advanced this understanding by identifying six positive regulators of EC sprouting behavior, which are Jak2, Jak1, Stat3, Stat5A, Mek1, and mTOR. The addition of Jak inhibitors, such as ruxolitinib and SD1008, led to strong suppression of EC sprouting. siRNA suppression of Jak2 and Jak1 markedly blocked EC sprouting behavior along with siRNAs directed to Stat3 and Stat5A. Furthermore, pharmacologic inhibition of Mek1 and mTOR (with U0126 and everolimus, respectively) also blocked EC sprouting. These pharmacologic inhibitors blocked the hypersprouting effects of active Jak2 and Stat3 expression in ECs. Overall, these findings reveal new signaling regulators of EC sprouting behavior, a fundamental step necessary for vascular morphogenesis.

The three EC lines expressing active PIK3CA, Akt1, or kRas showed the least ability to recruit pericytes compared with control ECs (Figure 9). Consistent with this finding was the strongly reduced basement membrane deposition observed in the EC-pericyte co-cultures with these ECs (Figure 10). ECs expressing active Mek1, Rheb, Jak2, and Stat3 showed modest, but significant, decreases in pericyte recruitment, and the ECs carrying active Mek1, Rheb, and Stat3 also showed decreases in basement membrane deposition compared with control ECs. The active Jak2-expressing ECs showed comparable basement membrane deposition compared with control ECs. Overall, EC expression of key oncogenic activating mutations led to discernible defects in vascular morphogenesis through alterations in both lumen formation and sprouting behavior coupled to defects in pericyte recruitment and basement membrane deposition. Thus, these activating mutations, to varying degrees, induced deficiencies in capillary assembly, a predisposing stimulus for the development of malformations.

Finally, four (active Akt1, kRas, PIK3CA, and Mek1) of the seven activating mutations expressed in human ECs demonstrated the greatest morphogenic alterations manifested as lumen formation abnormalities in combination with pericyte recruitment and basement membrane deposition deficits. Together, these vascular morphogenic and capillary assembly abnormalities lead to a capillary deficiency state. In the blood vasculature, each of these molecules, in their mutated and activated forms, induce vascular malformations in humans,^41,44,64 with Akt1 in Proteus syndrome,⁹⁸ kRas in arteriovenous malformations,^57,66 Mek1 in arteriovenous malformations,⁹⁹ and PIK3CA in cavernous malformations.⁴³ The current model system identified each of these mutations as causing significant and unique alterations and defects in capillary assembly compared with control ECs. The other three activating mutations examined (Jak2, Stat3, and Rheb), which have not yet been implicated in vascular lesions, had fewer morphogenic defects compared with these others. The underlying reasons why ECs carrying specific activating mutations such as Akt1, k-Ras, PIK3CA, and Mek1, demonstrated greater defects in pericyte recruitment and basement membrane deposition, are likely to be complex. EC mRNA expression levels of EC-derived factors known to control pericyte recruitment to tubes, including PDGFBB, PDGFDD, EDN1, TGFB1, and HBEGF, were variably decreased (as revealed with quantitative PCR analysis) (Figure 9C) among the activating mutation EC lines. The most severely affected EC line carried active Akt1, and these ECs showed strongly reduced PDGFB, EDN1, PDGFD, and TGFB1 mRNA expression. However, the active kRas carrying ECs showed reductions in PDGFD and TGFB1 expression, whereas the other factors appear to be expressed at similar levels with control ECs. The kRas EC tubes demonstrate markedly reduced pericyte recruitment. Thus, there must be other reasons for these recruitment defects. Possibilities could include excessive degradation, rapid clearance (perhaps through accelerated pinocytic uptake), or sequestration of these pericyte recruitment factors by the ECs carrying the activating mutations. For example, cells expressing activated k-Ras are known to show accelerated macropinocytosis,¹⁰⁰ which might facilitate more rapid clearance of these pericyte recruitment factors.

Overall, the experimental approach employed in this study represents a highly effective and efficient strategy to analyze the morphogenic and molecular impact of specific mutations observed in ECs within vascular anomalies. In addition, the work demonstrated how a model system such as this can be employed to elucidate the underlying molecular basis for the observed vascular lesions. Furthermore, these systems can be used to identify potential therapeutic agents and new therapeutic strategies to prevent or reverse the abnormalities, as shown in two recent studies addressing the molecular basis for blood capillary or lymphatic capillary regression.^101,102 Both studies identified pharmacologic drug combinations that prevent and, thus, rescue proinflammatory mediator-induced capillary regression events from both types of EC tube networks.^101,102

Author Contributions

P.K.L. and Z.S. performed experiments; P.K.L., Z.S., and G.E.D. designed experiments, analyzed data, and generated figures; and P.K.L., Z.S., and G.E.D. wrote the article. All authors reviewed the article. G.E.D. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Disclosure Statement

None declared.

Supplemental Data

Supplemental Figure S1.

RT-PCR analysis of endothelial cells (ECs) expressing activating mutations. After selection and establishment of seven stable EC lines expressing activating mutations, total RNA was prepared from these and control ECs. mRNA expression of the indicated genes was compared with the housekeeping gene, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), from each of the EC lines. RT-PCR was performed to confirm elevated expression of the activating mutation genes compared with the expression of GAPDH. Ac, active; Akt1,Akt serine/threonine kinase 1; Jak2, Janus kinase 2; kRas, Kirsten rat sarcoma virus; Mek1, mitogen-activated protein kinase 1; PIK3CA, phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; Rheb, ras homolog enriched in brain; Stat3, signal transducer and activator of transcription 3.

Supplemental Figure S2.

Quantification of protein expression quantification for key lumen formation regulatory signaling pathways using Western blot. Vasculogenic assays in three-dimensional collagen matrices were performed using control endothelial cells (ECs), and ECs expressing active signal transducer and activator of transcription 3 (Stat3), Kirsten rat sarcoma virus (kRas), and Akt serine/threonine kinase 1 (Akt1). Lysates were prepared from collagen gels from these cultures at 72 hours. Western blot analyses were performed, and results were captured using a Biorad imaging system. ImageJ, version 1.45f, was used to quantify protein expression. Statistical quantification (t-test) for protein kinase C epsilon (PKCε), glycogen synthase kinase 3 beta (GSK3β), p38, and Akt phosphorylated (phospho) protein expression (normalized to total protein expression) was performed. n ≥ 3. ∗P ≤ 0.05, ∗∗P ≤ 0.01 compared with control. Mapk, mitogen-activated protein kinase.

Supplemental Figure S3.

Pharmacologic inhibitors of phosphatidylinositol 3-kinase (PI3K)–Akt–mammalian target of rapamycin (mTOR) and Ras mitogen-activated protein kinase signaling pathways lead to blockade of endothelial cell (EC) lumen formation. A: Control (Con) ECs were subjected to the vasculogenic assay, with or without pharmacologic blockers added at 0 hours. Assays were subsequently fixed and stained after 48 hours. Representative images are shown. B: Lumen area from the vasculogenic assay was quantified, and a t-test was used to determine significance compared with control. U0126 and PD98059 are mitogen-activated protein kinase (Mek) inhibitors; PLX8394 is a B-rapidly accelerated fibrosarcoma (BRAF)/C-rapidly accelerated fibrosarcoma (CRAF) inhibitor. C: Control ECs were aggregated and subjected to the aggregate assay, with or without pharmacologic blockers added at 0 hours. Cultures were fixed after 24 hours and subsequently stained with toluidine blue and imaged. Representative images are shown. D: Lumen area of the aggregates was quantified and, and a t-test was used to determine significance compared with control. Pictilisib is a PI3K inhibitor; MK2206 is an Akt inhibitor; rapamycin and everolimus are mTOR inhibitors. E: Control ECs were seeded in a vasculogenic assay after siRNA knockdown (KD) of the indicated targets and formed for 72 hours before fixing, staining, and images. F: Lumen area of the 72-hour vasculogenic assay was quantified. G: siRNA knockdown of phosphatase and tensin homolog (PTEN) or protein phosphatase 2A (PP2A) (si-PTEN or si-PP2A, respectively) was confirmed by Western blot analysis. n = 12 (B and D); n = 24 (F). ∗∗P ≤ 0.01, ∗∗∗P ≤ 0.001 compared with control. Scale bars = 200 μm (A, C, and E).

Supplemental Figure S4.

Pharmacologic blockade of lumen and tube formation from Akt1-expressing endothelial cells (ECs) using inhibitors directed to protein kinase C epsilon (PKCε), Src family kinase, and mammalian target of rapamycin (mTOR), p21-activated kinase (Pak), rapidly accelerated fibrosarcoma (Raf), and mitogen-activated protein kinase (Mek) kinases. ECs expressing active (Ac) Akt1 were seeded in vasculogenic assays in the presence or absence of the indicated pharmacologic inhibitors for 72 hours. Cultures were subsequently fixed, stained, and imaged. Pharmacologic inhibitors added include PP2 (Src family kinase inhibitor), PD98059 (Mek inhibitor), Frax597 (Pak inhibitor), Frax486 (Pak inhibitor), U0126 (Mek inhibitor), PLX8394 (b-Raf and c-Raf inhibitor), everolimus (mTOR inhibitor), Go6983 (PKCε inhibitor), and rapamycin (mTOR inhibitor). A: Representative images of the 72-hour vasculogenic assay are shown. Black arrowheads indicate cystic lumen structures. B: Lumen area from the vasculogenic assay was quantified, and a t-test was used to determine significance compared with control. C: Cyst number from the vasculogenic assay was quantified, and a t-test was used to determine significance compared with control. n = 12 (B); n = 15 (C). ∗P ≤ 0.05, ∗∗P ≤ 0.01, and ∗∗∗P ≤ 0.001 compared with control. Scale bar = 200 μm (A).

Supplemental Figure S5.

Pharmacologic blockade of tip cell sprouting from active (Ac) Janus kinase 2 (Jak2)- and signal transducer and activator of transcription 3 (Stat3)-expressing endothelial cells (ECs) using inhibitors directed to phosphatidylinositol 3-kinase (PI3K), Jak, mitogen-activated protein kinase (Mek), and mammalian target of rapamycin (mTOR) kinases. ECs expressing activated Jak2 or Stat3 genes were seeded in downward sprouting assays in the presence or absence of the indicated pharmacologic agents for 16 hours. Subsequently, cultures were fixed, stained, imaged, and quantified for tip cell number per field. Pictilisib is a PI3K inhibitor; U0126 and PD98059 are Mek inhibitors; everolimus and rapamycin are mTOR inhibitors; and SD1008 and ruxolitinib are Jak inhibitors. A: Quantification of EC tip cell sprouting from active Jak2-expressing ECs treated with indicated pharmacologic blockers. B: Quantification of EC tip cell from active Stat3-expressing ECs treated with indicated pharmacologic blockers. n = 3 (A and B). ∗P ≤ 0.05, ∗∗P ≤ 0.01, and ∗∗∗P ≤ 0.001 compared with control.

Supplemental Figure S6.

Compared with control, active (Ac) Kirsten rat sarcoma virus (kRas)-, mitogen-activated protein kinase 1 (Mek1)-, and Akt serine/threonine kinase 1 (Akt1)-expressing endothelial cells (ECs) exhibit widened and less branched vascular guidance tunnels with reduced pericyte accumulation around tubes within tunnel spaces. In EC-pericyte co-culture assays, control ECs and active mutation-expressing ECs were seeded with green fluorescent protein–expressing pericytes in three-dimensional collagen matrices. EC-pericyte tube co-assembly was allowed to proceed for 120 hours before fixation with paraformaldehyde and immunostaining with collagen type I (red) to label the collagen matrix and visualize the vascular guidance tunnels generated during the morphogenic process. Confocal microscopy was performed, and representative images are shown. White arrows indicate the border of tunnel spaces. Scale bar = 50 μm. Jak2, Janus kinase 2; PIK3CA, phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; Rheb1, ras homolog enriched in brain 1; Stat3, signal transducer and activator of transcription 3.

Supplemental Figure S7.

Endothelial cells (ECs) expressing activating mutations demonstrate reduced basement membrane matrix deposition in response to EC-pericyte tube co-assembly, leading to a failure in proper capillary tube assembly and maturation. In EC-pericyte co-culture assays, control ECs and the indicated active (Ac) mutation-expressing ECs were seeded with green fluorescent protein (GFP)–expressing pericytes. Co-culture assays were allowed to undergo morphogenesis for 120 hours before fixing and staining with basement membrane components deposited during the morphogenic and maturation processes (red), and the images were overlaid with images of the GFP-stained pericytes (green). Representative images are shown. Scale bar = 50 μm. kRas, Kirsten rat sarcoma virus; Mek1, mitogen-activated protein kinase 1; PIK3CA, phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha.

Supplemental Figure S8.

Endothelial cells (ECs) expressing activating mutations demonstrate reduced basement membrane matrix deposition in response to EC-pericyte tube co-assembly, leading to a failure in proper capillary tube assembly and maturation. In EC-pericyte co-culture assays, the indicated active (Ac) mutation-expressing ECs were seeded with green fluorescent protein (GFP)–expressing pericytes. Co-culture assays were allowed to undergo morphogenesis for 120 hours before fixing and staining with basement membrane components deposited during the morphogenic and maturation processes (red), and the images were overlaid with images of the GFP-stained pericytes (green). Representative images are shown. Scale bar = 50 μm. Akt1, Akt serine/threonine kinase 1; Jak2, Janus kinase 2; Rheb1, ras homolog enriched in brain 1; Stat3, signal transducer and activator of transcription 3.

Supplemental Video S1

Control endothelial cell (EC) tube formation over a 72-hour period in vasculogenic assays. Control EC cultures were established and allowed to assemble into tubes for 72 hours. The video is shown at 12 frames/second.

Supplemental Video S2

Control endothelial cell (EC) tube formation over a 72-hour period in vasculogenic assays. A second representative field of control EC cultures is shown, which were established and allowed to assemble into tubes for 72 hours. The video is shown at 12 frames/second.

Supplemental Video S3

Active Kirsten rat sarcoma virus (kRas)-expressing endothelial cells (ECs) forming tubes over a 72-hour period in vasculogenic assays. Active kRas-expressing EC cultures were established and allowed to assemble into tubes for 72 hours. The video is shown at 12 frames/second.

Supplemental Video S4

Active mitogen-activated protein kinase 1 (Mek1)-expressing endothelial cells (ECs) forming tubes over a 72-hour period in vasculogenic assays. Active Mek1-expressing EC cultures were established and allowed to assemble into tubes for 72 hours. The video is shown at 12 frames/second.

Supplemental Video S5

Active phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA)-expressing endothelial cells (ECs) forming cystic lumen structures and blunted tubes over a 72-hour period in vasculogenic assays. Active PIK3CA-expressing EC cultures were established and allowed to assemble into cystic lumen structures and occasional blunted tubes for 72 hours. The video is shown at 12 frames/second.

Supplemental Video S6

Active phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA)-expressing endothelial cells (ECs) forming cystic lumen structures and blunted tubes over a 72-hour period in vasculogenic assays. A second representative field of active PIK3CA-expressing EC cultures was established and allowed to assemble into cystic lumen structures and occasional blunted tubes for 72 hours. The video is shown at 12 frames/second.

Supplemental Video S7

Active Akt serine/threonine kinase 1 (Akt1)-expressing endothelial cells (ECs) forming tubes over a 72-hour period in vasculogenic assays. Active Akt1-expressing EC cultures were established and allowed to assemble into tubes for 72 hours. The video is shown at 12 frames/second.

Supplemental Video S8

Active Akt serine/threonine kinase 1 (Akt1)-expressing endothelial cells (ECs) forming tubes over a 72-hour period in vasculogenic assays. A second representative field of active Akt1-expressing EC cultures was established and allowed to assemble for 72 hours. The video is shown at 12 frames/second.

Supplemental Video S9

Active ras homolog enriched in brain (Rheb)-expressing endothelial cells (ECs) forming tubes over a 72-hour period in vasculogenic assays. Active Rheb-expressing EC cultures were established and allowed to assemble into tubes for 72 hours. The video is shown at 12 frames/second.

Supplemental Video S10

Active Janus kinase 2 (Jak2)-expressing endothelial cells (ECs) forming tubes over a 72-hour period in vasculogenic assays. Active Jak2-expressing EC cultures were established and allowed to assemble into tubes for 72 hours. The video is shown at 12 frames/second.

Supplemental Video S11

Active Stat3-expressing endothelial cells (ECs) forming tubes over a 72-hour period in vasculogenic assays. Active Stat3-expressing EC cultures were established and allowed to assemble into tubes for 72 hours. The video is shown at 12 frames/second. Stat3, signal transducer and activator of transcription 3.