Deletion of tetraspanin Cd151 results in decreased pathologic angiogenesis in vivo and in vitro

Abstract

Tetraspanin protein CD151 is abundant on endothelial cells. To determine whether CD151 affects angiogenesis, Cd151-null mice were prepared. Cd151-null mice showed no vascular defects during normal development or during neonatal oxygen-induced retinopathy. However, Cd151-null mice showed impaired pathologic angiogenesis in other in vivo assays (Matrigel plug, corneal micropocket, tumor implantation) and in the ex vivo aortic ring assay. Cd151-null mouse lung endothelial cells (MLECs) showed normal adhesion and proliferation, but marked alterations in vitro, in assays relevant to angiogenesis (migration, spreading, invasion, Matrigel contraction, tube and cable formation, spheroid sprouting). Consistent with these functional impairments, and with the close, preferential association of CD151 with laminin-binding integrins, Cd151-null MLECs also showed selective signaling defects, particularly on laminin substrate. Adhesion-dependent activation of PKB/c-Akt, e-NOS, Rac, and Cdc42 was diminished, but Raf, ERK, p38 MAP kinase, FAK, and Src were unaltered. In Cd151-null MLECs, connections were disrupted between laminin-binding integrins and at least 5 other proteins. In conclusion, CD151 modulates molecular organization of laminin-binding integrins, thereby supporting secondary (ie, after cell adhesion) functions of endothelial cells, which are needed for some types of pathologic angiogenesis in vivo. Selective effects of CD151 on pathologic angiogenesis make it a potentially useful target for anticancer therapy.

Figure 1

CD151-null mice show diminished angiogenesis, ex vivo and in vivo. (A) Thoracic aortic rings were embedded in Matrigel and incubated for 4 days. Shown are phase-contrast images of branching vessel-like structures in WT and CD151-null explants. Lengths of vessel-like structures were quantitated (n = 6; *P < .005). (B) After implantation for 7 days, sections of Matrigel plugs were stained with hematoxylin and eosin (H&E) and for endothelial marker CD34. Percentage of vascular area stained with CD34 was calculated, in 5 fields, using Scion Image software. Results are shown as the mean ± SEM. (*P < .01). Hemoglobin content was also determined (n = 5; *P < .05). Data are representative of 3 independent experiments with similar results. Scale bars represent 1 mm (A) and 100 μm (B). Bar graphs in panels A-B, and in every other figure, show mean ± SEM.

Figure 2

Impaired tumor angiogenesis in Cd151-null mice. (A) LLC tumor growth was assessed in 10 age- and sex-matched WT and knockout (KO) mice. Data are representative of 5 independent experiments. From tumors such as shown in the photo, tumor volume and tumor weight were calculated (n = 10; *P < .05). (B) Cryosections of LLC tumors were stained for endothelial marker CD31 and the number of CD31⁺ blood vessels was counted (*P < .01). Scale bars represent 1 cm (A) and 100 μm (B).

Figure 3

Corneal angiogenesis is deficient in Cd151-null mice. (A) Five days after implantation of pellets containing 20 ng bFGF, vessels were stained with PE-conjugated anti-CD31. Clock-hour length of sprouting vessels was quantitated (**P < .005). (B) Corneal pellets contained 40 ng bFGF. After 5 days, vessels were stained with anti-CD31, and vascular area was quantitated (*P < .05). Scale bars represent 100 μm in panels A-B.

Figure 4

CD151 affects endothelial cell spreading, chemotaxis, invasion, and random migration. (A) To assess spreading, MLECs were seeded onto plates coated with fibronectin, gelatin, or Matrigel. After 30 minutes, spread cells were counted in 3 high-power fields in duplicate wells (*P < .01). Data are representative of 3 independent experiments with similar results. (B) For chemotactic migration, MLECs (7 × 10⁴ cells in 200 μL DMEM; 5% FCS) were plated in upper Transwell chambers, coated with fibronectin, gelatin, or Matrigel (thin layer). Bottom chambers contained 600 μL DMEM, 5% FCS, and 20 ng/mL bFGF. Cells migrating through the filter were counted (*P < .05; **P < .01). (C) To assess invasion, cells were counted after migrating for 6 hours through a Matrigel-coated Transwell (n = 3; *P < .005). (Di) To measure random migration, cells were plated on coverslips coated with Matrigel or gelatin. (ii) Cell movements were recorded by time-lapse video microscopy every minute for 30 minutes and quantified using Scion Image tools (n = 12). Top panels show randomly selected individual migration tracks copied and combined (bars = 100 μm). Bottom panels show “T” and “D/T” where T is total distance migrated and D is distance between starting and ending point (*P < .05). High D/T ratios (approaching 1.0) indicate directional persistence. Data are representative of 3 independent experiments with similar results.

Figure 5

Impaired 3-dimensional functions on Matrigel and collagen. (A) MLECs were seeded on Matrigel and photographed at indicated times (bar = 1 mm). Total length of cellular cables was quantitated after 24 hours (*P < .001). (B) Representative photographs of Matrigel contraction data are shown for WT and Cd151-null MLECs (bar = 5 mm). Percent contraction (after 18 hours) is the average gel diameter/well diameter × 100 (*P < .05). Data are representative of 3 independent experiments with similar results. (C) MLECs were incubated within collagen gel for 48 hours, then fixed, stained with toluidine blue, and photographed (bar = 1 mm). Total tube length is quantitated (n = 2; *P < .05). (D) Representative images are shown of capillary-like structures sprouting from MLEC spheroids after 48 hours (bar = 100 μm). Total sprouting length is quantitated (*P < .005). Data are representative of 3 independent experiments with similar results. Note that the initial size of spheroids, when first transferred to collagen gel, was similar for WT and Cd151-null cells (136 ± 12 μm and 145 ± 13 μm, respectively).

Figure 6

Adhesion-dependent MLEC signaling. (A) Serum-starved MLECs were detached, suspended for 1 hour, and plated on Matrigel-coated plates for the indicated times. Equal amounts of lysate were probed with antibodies to activated Akt (p-Akt), activated Raf (p-Raf), and activated eNOS (p-eNOS). Numbers represent p-eNOS (from densitometry) normalized for eNOS amounts. (B) Serum-starved MLECs were plated on Matrigel and analyzed for Rac and Cdc42 activation at the indicated times. Western blots in this figure are each representative of 3 experiments with different cell preparations.

Figure 7

CD151 causes substantial modification of integrin complexes. (A) MLECs were labeled with [³H]-palmitate and lysed in 1% Briji-96 buffer. Immunoprecipitations were carried out using mAbs HMβ1-1 + 9EG7 (for β1 integrins) and mAb KMC8 (for CD9). Several integrin-associated proteins, including CD151 itself, are substantially decreased in amount when CD151 is absent (lane 2). These are marked with white arrows and (except for CD151 itself) are labeled as d1-d5. Proteins that are relatively unchanged are marked with black arrows. Note that the integrin β1 subunit does not appear because it does not undergo palmitoylation. Immunoprecipitation of mouse α3 complexes yielded very similar results—proteins similar in size to d1-d5 were again mostly lost from α3β1 complexes when CD151 was absent (not shown). (B) The schematic diagram emphasizes that CD151 plays a key role in linking α3β1 and α6β1 integrins to several other unknown and known proteins, including other tetraspanins CD9 and CD81.