CIB1 Regulates Endothelial Cells and Ischemia-Induced Pathological and Adaptive Angiogenesis

Reagents, Cells, and Short Hairpin RNA Construction

All reagents, cell types, blotting techniques, and short hairpin (sh)RNA construction are described in the online data supplement at http://circres.ahajournals.org.

Boyden Chamber Haptotactic, Matrigel Tubule Formation, and ELISA Assays

Haptotactic migration and tubule formation were performed as described previously with minor modifications.¹⁴ These assays and ELISA proliferation assays are described in the online data supplement.

PAK1 Kinase Activation Assay

Lysates of fibronectin-adherent ECs or ischemic and nonischemic gastrocnemius muscles were prepared by addition of cold lysis buffer, supplemented with 10% glycerol and 0.2 mmol/L pervanadate. Kinase assays were performed as described previously.^14,18 For each experiments, n=3.

Gelatin Zymography

The activity of MMP-2 in conditioned media was determined by gelatin zymography. Briefly, ECs were serum starved for at least 24 hours, followed by addition of phenol-free basal medium supplemented with 100 ng/mL of growth factor. After 18 hours, conditioned media were collected, filtered through 0.45-μm Whatman filters, and concentrated using a Vivaspin 10-kDa concentrator (Sartorius, Edgewood, NY). Equal protein concentrations were loaded under nonreducing conditions onto premade 10% zymograms and developed according to the instructions of the manufacture (Bio-Rad, Hercules, Calif). For each experiment, n=3.

Ex Vivo Tissue Cultures

Tissue culturing was performed as described previously with minor modifications.^19,20 Wild-type (WT) and CIB1^−/− aortic rings were embedded between 2 layers of 250 μL of growth factor–reduced Matrigel in a 24-well culture format and cultured in 300 μL of growth media or basal media supplemented with 50 ng/mL basic fibroblast growth factor (bFGF) or vascular endothelial growth factor (VEGF). At 2, 5, and 8 days, images of aortic rings were collected and stitched together using PhotoFit Premium software (Tekmate, Tokyo, Japan). At each time point, the total number of sprouting microvessels were counted using ImageJ as described previously¹⁹ (n=8 per group). Identical procedures were used for the tibialis anterior muscle cultures and were analyzed in a similar manner after 8 days of culturing.

In Vivo Matrigel Plug Assay

Matrigel plug assays were performed as described previously,²¹ with minor modifications. WT and CIB1^−/− mice were subcutaneously injected with ≈300 μL of growth factor–reduced Matrigel mixed with 60 U/mL heparin or with heparin plus 250 ng/mL bFGF or VEGF. After 14 days, plugs were isolated along with adjacent skin and peritoneal muscle, fixed, and sectioned at interrupted intervals. Blood vessel infiltration was analyzed in 4 random hematotoxin/eosin (H&E)-stained interrupted sections using a Nikon inverted microscope and expressed as mean values of infiltrating blood vessels per plug (n=3 per group).

Oxygen-Induced Retinal Angiogenesis Assay

Postnatal day 7 (P7) mice were placed in 75% oxygen for 5 days. Following hyperoxic incubation, P12 CIB1^−/− and WT neonatal mice were housed for 5 additional days under normoxic conditions, after which P17 mouse eyeballs were fixed in 2% paraformaldehyde and either microdissected to obtain retinal flat mounts or cryopreserved in OCT (IMEB, San Marcos, Calif). Fluorescent images of flat mounts stained with Alexa594-conjugated Griffonia simplicifolia isolectin B4 (isolectin GSL-1-B4; 1:100, Invitrogen) were analyzed with a Nikon inverted microscope (PhotoFit, TekMate, Anchorage, Alaska; ImageJ; n=8 mice per group). Interrupted 6-μm sections at 50-μm intervals were obtained for cryopreserved eyeballs and H&E stained. Mean capillary area of vessels either directly under or invading the retinal inner limiting membrane (ILM) were measured from digitized images obtained from 2 randomly selected interrupted eyeball sections (n=4 mice per group).

Unilateral Femoral Artery Ligation, Laser Doppler Perfusion Imaging, and Appearance and Use Score Assessments

Ligation, imaging, and use scores were performed as described previously.^22,23 Appearance scores were based on the scale; 0=no change; 1 to 5=the number of discolored/lost nails; and 6 to 10=the number of discolored/lost digits (n=11 mice per group).

Atrophy and Capillary Density in Gastrocnemius Muscles

Twenty-one days after femoral artery ligation, gastrocnemius muscles were fixed and paraffin embedded as described previously.²² For each sample, 5 sections (50 μm apart) were obtained and stained with H&E or Masson’s trichrome or labeled with isolectin GSL-1-B4 (1:100). Three 20× images were collected for each condition from 2 randomly selected sections. Using ImageJ, muscle atrophy and capillary density were estimated as described previously (n=5–9 mice per group).²²

Animal Regulations

All housing, breeding, and experimental procedures with mice were in accordance with national guidelines and regulations, and were approved by the University of North Carolina at Chapel Hill Institutional Animal Care and Use Committee.

Statistical Analysis

Continuous variables were compared with either the Student’s t test or nonparametric Mann–Whitney U test. We considered P<0.05 to be significant. For animal studies, male and female mice demonstrated no experimental differences; thus they were combined.

Loss of CIB1 Disrupts Normal EC Function

CIB1 is expressed in highly vascularized adult mouse organs, embryonic vascular structures, and ECs (Figure I in the online data supplement). Because CIB1 associates with angiogenic kinases such as PAK1¹⁴ and FAK,¹³ we hypothesized that CIB1 may contribute to EC function. We first tested the role of CIB1 in EC migration using a conventional Boyden chamber haptotactic migration assay.¹⁴ Knockdown of CIB1 with shRNA (supplemental Figures II and III) decreased haptotaxis of immortalized mouse embryonic heart ECs (MECs) on fibronectin (38% relative to control MECs; Figure 1A). Similar results were also obtained in the presence of actinomycin D, demonstrating that the haptotactic assay was not significantly influenced by EC proliferation during the relatively short period of this assay (data not shown). Similarly, mouse lung ECs (MLECs) isolated from CIB1^−/− mice demonstrated a similar decrease in migration (24% decrease relative to WT and CIB1^+/− MLECs; Figure 1B). Although angiogenic growth factors bFGF and VEGF increased haptotaxis in untransduced MECs (data not shown) and control transduced MECs, they did not rescue haptotaxis in CIB1 knockdown MECs (Figure 1A). Interestingly, reintroduction of CIB1 in CIB1^−/− MLECs restored haptotactic migration up to, but not beyond, normal levels (Figure 1B). To confirm these findings, we also tested the role of CIB1 in EC migration using a monolayer wound-healing assay. Monolayer wounds in both CIB1-depleted MECs and CIB1^−/− mouse heart ECs (MHECs) resolved less rapidly, relative to control transduced MECs and WT MHECs, respectively (supplemental Figure IV). Therefore, these data demonstrate that CIB1 expression is necessary for normal EC migration.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Loss of CIB1 decreases EC function. A, Boyden chamber migration assays of MECs transduced with CIB1 shRNA or control shRNA toward fibronectin in the presence or absence of bFGF or VEGF. B, Haptotactic migration of MLECs either untransduced or transduced with control cDNA or CIB1 cDNA vectors. C, Quantification of BrdUrd incorporation as the fold difference in absorbance (optical density [OD]₃₇₀) of MECs cultured either in growth media or basal media supplemented with bFGF or VEGF. D, Fold change in BrdUrd incorporation in CIB1^−/− and WT MHECs and MLECs cultured in growth media. E, Differential interference contrast (DIC) and green fluorescent protein fluorescent images of CIB1 shRNA-transduced MECs and control shRNA-transduced MECs following treatment with growth media, bFGF, or VEGF. Green fluorescent protein images confirm MEC transduction and expression of vectors. Scale bars, 100 μm. F, Quantification of tubule formation in CIB1^−/− and WT MHECs following treatment with growth media, bFGF, or VEGF. Error bars represent SEM (n=at least 3 for all assays and conditions). *P<0.05, **P<0.001.

Proliferation of ECs is also fundamental to angiogenesis.⁶ Relative to control, MECs depleted of CIB1 by shRNA and cultured in either growth media or basal media supplemented with either bFGF or VEGF demonstrated modest yet significantly decreased 5-bromodeoxyuridine (BrdUrd) incorporation (Figure 1C). No difference in proliferation between untransduced and control transduced MECs was observed (data not shown). Consistent with this, CIB1^−/− MHECs and MLECs cultured in growth media also demonstrated a 61% and 35% decrease, respectively, in BrdUrd incorporation compared with WT ECs (Figure 1D). Direct cell counts of cultured CIB1-depleted MECs and CIB1^−/− MHECs further confirmed a significant decrease in proliferation relative to controls (data not shown). Thus, in addition to migration, CIB1 is also necessary for normal rates of EC proliferation.

To determine whether loss of CIB1 affects other EC functions, we also assessed tubule formation on growth factor–reduced Matrigel. In the presence of growth media, CIB1-depleted MECs and CIB1^−/− MHECs showed a 26% and 44% reduction in nascent tubule formation compared with their respective controls (Figure 1E and 1F). Treatment of CIB1-depleted MECs and CIB1^−/− MHECs with bFGF or VEGF did not stimulate tubule formation to the same extent as control cells (Figure 1E and 1F). Hence, in addition to EC migration and proliferation, CIB1 appears to be required for efficient EC sprouting and invasion, events required for angiogenesis.

Loss of CIB1 in ECs Disrupts PAK1 and ERK1/2 Activation

We previously determined that CIB1 regulates PAK1 activation in fibroblasts and transformed epithelial cells.¹⁴ Naik and Naik also have reported that CIB1 can regulate FAK activation.¹³ Because several studies have underscored the importance of PAK1, FAK, and their downstream mitogen-activated protein kinase (MAPK) effectors, ERK1/2 and p38, in EC angiogenic signaling,^15,16,24 we asked whether loss of CIB1 affects the activation of these signaling molecules. Although FAK phosphorylation was not significantly altered in CIB1^−/− ECs (data not shown), in vitro kinase assays revealed significantly less active PAK1 in CIB1^−/− MLECs relative to control cells, 20 and 45 minutes following adhesion to fibronectin (Figure 2A). Decreased PAK1 activation was also observed in CIB1-depleted MECs (data not shown). Similarly, decreased ERK1/2 phosphorylation was observed in CIB1^−/− MLECs, without changes in total ERK1/2 protein levels, 20 and 45 minutes following adhesion (Figure 2B). Control ECs also showed differences in the relative peak activation times of PAK1 and ERK1/2, suggesting the contribution of additional signaling pathways leading to ERK1/2 activation. However, despite these differences, we consistently found decreased activation of both PAK1 and ERK1/2 in CIB1^−/− ECs. Importantly, no differences in protein kinase C and p38 MAPK phosphorylation were detected in WT and CIB1^−/− MLECs following adhesion (Figure 2B and data not shown). Thus, these data demonstrate that loss of CIB1 can disrupt EC signaling, potentially via regulation of PAK1 and ERK1/2 activation in ECs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Loss of CIB1 disrupts PAK1 activation and MMP-2 expression. A, PAK1 activity in WT and CIB1^−/− MLECs held in suspension or adhered to fibronectin for the indicated times. PAK1 in vitro kinase assays were performed using myelin basic protein as substrate ([³²P]MBP). Western blots of total PAK1 protein and Coomassie stain of myelin basic protein serve as loading controls. B, Thr514 protein kinase C (PKC) and ERK1/2 phosphorylation in WT and CIB1^−/− MLECs held in suspension or adhered to fibronectin. Western blots of total ERK1/2 and Rac serve as loading controls. C, Conditioned media of growth factor–treated and untreated WT and CIB1^−/− MLECs were subjected to gelatin zymography. Clear bands in destained gel represent MMP-2. D, Western blot of MMP-2 in conditioned media of growth factor–treated and untreated WT and CIB1^−/− MLECs. For C and D, Coomassie-stained nonspecific bands serve as loading controls. E, Western blot of total MMP-2 and CIB1 protein in duplicate WT and CIB1^−/− MHEC lysates. Blot for GAPDH serves as a loading control. For each experiment, n=3.

Loss of CIB1 in ECs Decreases MMP-2 Expression

Numerous signaling cascades can be activated downstream of PAK1.²⁵ Among them are MAPKs that are implicated in the production and activation of MMPs.²⁶ Because loss of CIB1 in ECs resulted in decreased responsiveness to angiogenic growth factors and decreased PAK1 and ERK1/2 activation on adhesion to fibronectin, we asked whether MMP expression was also affected. Enzymatic activity of MMPs was assessed by gelatin zymography, which revealed increased lytic zones only at the molecular mass of MMP-2 (68 kDa) in the conditioned media of growth factor–treated WT MLECs but not CIB1^−/− MLECs (Figure 2C). Western blotting also demonstrated less growth factor–induced MMP-2 expression in CIB1^−/− MLEC conditioned media (Figure 2D), and decreased MMP-2 protein expression in lysates derived from CIB1^−/− MHECs (Figure 2E). However, MMP-9 secretion was undetectable in WT and CIB1^−/− ECs (data not shown). Furthermore, WT and CIB1^−/− ECs showed no differences in the expression of the physiological inhibitor (tissue inhibitor of metalloproteinases [TIMP]-2²⁷) and activator of MMP-2, (MMP-14 [MT1-MMP]²⁷) (data not shown).

CIB1 Deficiency Diminishes bFGF and VEGF-Induced Microvessel Growth

The decreased response to bFGF and VEGF and attenuated MMP-2 expression by CIB1^−/− ECs led us to ask whether angiogenesis is decreased ex vivo, in cultured microsections of aortas and tibialis anterior muscles from CIB1^−/− mice. CIB1^−/− aortic rings and tibialis anterior muscles cultured for 8 days in growth media produced ≈50% fewer microvessels, relative to control (Figure 3A through 3C). In addition, CIB1^−/− aortic rings also produced ≈60% fewer microvessels in response to bFGF and VEGF (Figure 3A and 3B). We further assessed growth factor–induced angiogenesis in vivo by implanting WT and CIB1^−/− mice with Matrigel plugs impregnated with bFGF or VEGF. Microvessel infiltration into the plugs was reduced by ≈80% in CIB1^−/− mice compared with WT littermates (Figure 3D and 3E). Thus, growth factor–induced angiogenesis is strongly inhibited in CIB1^−/− mice.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Reduced growth factor–induced microvessel sprouting in CIB1^−/− tissue. A, Low-magnification DIC images of WT and CIB1^−/− ex vivo aortic ring cultures show newly formed microvessels 8 days following treatment with growth media, bFGF, or VEGF. Scale bar, 0.2 mm. B, Quantification of microvessel sprouts of WT and CIB1^−/− aortic rings cultured in growth media or growth factor–supplemented basal media. Error bars are ±SEM (replicates of at least 4; n=6 to 8 aortas per genotype). C, High-magnification DIC images of microvessel sprouts in WT and CIB1^−/− tibialis anterior muscle sections cultured in growth media for 8 days (performed in at least duplicates, n=3 per genotype). Scale bar, 100 μm. D, Representative images of growth factor-reconstituted Matrigel plugs subcutaneously injected in WT and CIB1^−/− mice. Newly formed microvessels are indicated by arrowheads. Scale bar, 25 μm. E, Quantification of microvessels infiltrating untreated or growth factor-reconstituted Matrigel plugs from WT and CIB1^−/− mice. Error bars represent SEM (n=3 to 7 plugs per group). *P<0.05, **P<0.001.

CIB1 Deficiency Leads to Decreased Oxygen-Induced Retinal Angiogenesis

Mice lacking CIB1 develop and reach adult reproductive age normally.¹⁷ Therefore, not surprisingly, we observed that developmental retinal angiogenesis in CIB1^−/− mice proceeds normally (supplemental Figure V). However, because CIB1 appeared to be a critical regulator of EC function and signaling, we hypothesized that CIB1^−/− mice may still have a defect in pathological retinal angiogenesis. To test this, we examined CIB1^−/− mice using an oxygen-induced retinal angiogenesis assay, wherein in vivo exposure to an elevated inspired oxygen concentration obliterates the central retinal capillaries, leading to hypoxia in the mouse neonatal retina. Subsequent return to normoxia induces exuberant angiogenesis, resulting in formation of aberrant capillary neovessel buds that can be visualized in isolectin GSL-1-B4–stained retinal flat mounts.^20,28

Well-defined buds were detected in WT P17 flat mounted retinas. In CIB1^−/− retinas, fewer and less clearly defined buds were evident (Figure 4A and 4B). Because these buds are composed of clusters of tortuous vessels that protrude through the retinal ILM,²⁹ retinal cross-sections were obtained to assess the size and morphology of these neovessels. WT retinal cross-sections displayed large protruding microvessels that occasionally ruptured through the ILM, whereas CIB1^−/− retinal cross-sections showed blood vessels directly under the ILM that were smaller in size and did not protrude through as frequently (Figure 4C and 4D). In CIB1^−/− retinas, the ILM was also almost always intact (Figure 4C). Therefore, these data provide the first evidence demonstrating that CIB1^−/− mice have impaired pathological angiogenesis in response to ischemia.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. Oxygen-induced retinal neovascularization is decreased in CIB1^−/− mice. A, Isolectin GSL1–B4–stained flat mounted retinas from WT and CIB1^−/− neonatal mice exposed to hyperoxic conditions followed by normoxia (left images). Vascular buds formed in WT and CIB1^−/− ischemic retinas are indicated by arrowheads in magnified retinal areas (right images). Scale bar, 25 μm. B, Quantification of vascular buds per retina in WT and CIB1^−/− mice (n=8 per mouse genotype). C, Representative H&E-stained cross-sections from WT and CIB1^−/− retinas after exposure to hyperoxia, followed by normoxia. The WT retina shows microvessel protrusion (☆) and rupture (arrow) through the ILM. Scale bar, 50 μm. D, Quantification of mean capillary area of microvessels directly under the ILM in WT and CIB1^−/− retinas following ischemic injury (n=4 mice for each mouse genotype). V indicates vitreous; GCL, ganglion cell layer; IPL, inner plexiform layer. Error bars represent SEM. *P<0.05.

CIB1 Deficiency Delays Postischemic Recovery and Neovascularization

A common mouse model for peripheral vascular disease is experimentally induced hindlimb ischemia through unilateral femoral artery ligation.²² In this model, blood flow to the hindlimb is effectively eliminated, which rapidly induces ischemia and postischemic recovery mechanisms (ie, adaptive angiogenesis and arteriogenesis).³⁰ To extend our observations from the oxygen-induced retinal angiogenesis assay, we performed this assay to determine whether CIB1 also plays a role in adaptive ischemia-induced angiogenesis.

Noninvasive, laser-scanning Doppler velocimetry²² demonstrated that by 14 days postfemoral artery ligation, 90% of hindpaw plantar region perfusion was restored in WT mice but only 60% in CIB1^−/− mice (Figure 5A and 5B). Furthermore, ischemic tissue injury was more severe in CIB1^−/− mice, as assessed by appearance and use scores.²³ Incidence of nail and toe discoloration, toe loss, hindpaw cyanosis, as well as impaired hindlimb function were several fold higher in CIB1^−/− mice relative to WT littermates (Figure 5C and 5D). Similarly, increased atrophy and fibrosis were observed in the gastrocnemius muscles of CIB1^−/− hindlimbs, which are known to experience substantial ischemia immediately after femoral artery ligation (Figure 6A).²² Average muscle fiber size in CIB1^−/− ischemic gastrocnemius muscles was reduced by 23%, compared with only 6% in WT muscles (Figure 6B), indicating that CIB1^−/− tissue experiences more profound tissue damage after ischemia.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5. Reduced perfusion and recovery in CIB1^−/− mouse ischemic hindpaws and hindlimbs following femoral artery ligation. A, Representative laser-scanning Doppler perfusion images of the ventral hindpaw surfaces of the same ischemic hindlimbs of a single WT and CIB1^−/− mouse, performed in a blinded manner. Relative flow velocity is indicated by 6-hue pseudocolor, in which gray represents 0 and white represents maximal velocity (range of velocity values, 0 to 5000 perfusion units [PU], arbitrary units). Preligation relative blood flow velocity is high in both groups, as indicated by white and red. One day after ligation, blood flow is largely eliminated to the hindpaw, as indicated by blue. Evidence of toe gangrene is observed in CIB1^−/− mouse ischemic hindpaws by 7 days. B, Quantification of perfusion at 7, 14, and 21 days after femoral artery ligation in WT and CIB1^−/− mice (n=11 per genotype; *P<0.05, **P<0.001). C and D, Hindpaw appearance and hindlimb use scores were determined to assess rates of recovery in WT and CIB1^−/− mice (n=11 per genotype). Mann–Whitney U test: *P<0.05, **P<0.01. Error bars represent ±SEM.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6. Tissue atrophy and reduced angiogenesis in CIB1^−/− mouse ischemic gastrocnemius muscle. A, Contrast-enhanced H&E and Masson’s trichrome (MT)-stained sections of WT and CIB1^−/− ischemic and nonischemic hindlimb gastrocnemius muscles. B, Mean muscle fiber size (μm²) per muscle fascicle in ischemic and nonischemic gastrocnemius muscles from WT and CIB1^−/− mice. C, PAK1 activity, ERK1/2, and p38 phosphorylation levels in WT and CIB1^−/− ischemic (I) and nonischemic (NI) gastrocnemius muscle tissue lysates isolated from the same mouse 4 days after unilateral femoral artery ligation. Western blots of total PAK1, total ERK1/2, p38, Rac, and Coomassie stain of myelin basic protein (MBP) serve as loading controls. D, Isolectin GSL1–B4–staining pattern of capillary EC in WT and CIB1^−/−, ischemic and nonischemic gastrocnemius muscles. Scale bars, 50 μm. E, Quantification of microvessel density, expressed as the ratio of capillary number-to-muscle fiber number per fascicle in WT and CIB1^−/− mice. Error bars represent SEM (n=5 to 9 per genotype). *P<0.05.

Interestingly, tissue homogenates of CIB1^−/− ischemic gastrocnemius muscles, isolated 4 days after femoral artery ligation showed a significant decrease in PAK1 activation and ERK1/2 phosphorylation (Figure 6C). ERK1/2 phosphorylation in CIB1^−/− nonischemic gastrocnemius muscle tissue was also significantly reduced. However, no consistent differences were detected in p38 phosphorylation and MMP-2 expression between WT and CIB1^−/− ischemic and nonischemic gastrocnemius muscles (Figure 6C). Thus, these data correlate with our findings in vitro and demonstrate that CIB1^−/− mice have a specific defect in PAK1 and ERK1/2 signaling, which likely contributes to the decrease in recovery in CIB1^−/− ischemic tissue.

Attenuated PAK1 activation may also lead to decreased angiogenesis, which could also explain the observed increase in tissue damage in CIB1^−/− mice following femoral artery ligation.²⁴ To test this, we determined capillary densities in gastrocnemius muscles 21 days after femoral artery ligation in WT and CIB1^−/− mice. Isolectin GSL-1-B4 staining revealed an increase in capillary density only in WT ischemic gastrocnemius muscles (Figure 6D and 6E), thus providing further evidence of reduced ischemia-induced angiogenesis in CIB1^−/− mice. Because postmortem x-ray angiography revealed unaltered arteriogenesis in CIB1^−/− mice (unpublished data, 2007), this defect in ischemia-induced angiogenesis appears to be specific.