Dependence of proliferative vascular smooth muscle cells on CD98hc (4F2hc, SLC3A2)

To assess a potential role of CD98 in VSMC biology, we first analyzed CD98 expression in uninjured and injured arteries of wild-type mice. In uninjured carotid arteries (CAs) and aortas (n = 3), faint CD98 staining was seen among the quiescent VSMCs in the media (Fig. 1 A and not depicted). To examine VSMC CD98 expression in injured vessels, we modified a mouse model of carotid intimal hyperplasia (Fig. 1 B, illustrations; Simon et al., 2000). Medial injury was induced by pressure dilation, the endothelium was removed with a nylon fishing line, and flow was reduced by ligation of the external and internal CAs, thus leaving only the occipital artery open at the carotid bifurcation. In this model, a circumferential neointima was formed 3 wk after injury (Fig. 1 B, photomicrographs). Importantly, this model allowed a neointima to be formed in the presence of the remaining medial VSMCs as well as in the presence of a blood flow, making it suitable for SMC-specific gene knockout. In the injured CA (n = 3), there was a dramatic increase in CD98 expression in the neointima and among scattered cells in the underlying media (Fig. 1 C). We used flow cytometry to quantify CD98 expression in VSMCs and excluded a contribution from endothelial cells or leukocytes by eliminating CD31- and CD18-positive cells (Fig. S1). In uninjured vessels, we detected a single population of VSMCs that expressed modest levels of CD98. In sharp contrast, there was an additional CD98 bright population in the injured vessels (Fig. 1 D). Quantification of these data revealed an approximately sixfold increase in CD98 expression in the CD98 bright population (uninjured geometric mean fluorescence (GMF) = 40, injured CD98 dull GMF = 38, and CD98 bright GMF = 249). We also examined CD98 expression in cultured VSMCs plated on a mixed fibronectin/gelatin matrix in the presence of 0.2% fetal calf serum. These cultured VSMCs uniformly expressed CD98 at high levels (GMF = 325). Because the formation of intimal hyperplasia involves a change of extracellular matrix (e.g., a change from laminin to fibronectin) and secretion of growth factors, we examined how these factors regulated CD98 expression in cultured VSMCs. The VSMCs were first made quiescent on Matrigel for 2 d and then replated on laminin-1 or fibronectin with or without platelet-derived growth factor (PDGF)–BB for 24 h. Fibronectin significantly increased the CD98 expression compared with laminin-1, and PDGF-BB up-regulated the CD98 expression on both matrices (Fig. 1 E). In summary, quiescent VSMCs in uninjured vessels expressed CD98 at low levels, whereas vascular injury and tissue culture conditions dramatically increased VSMC CD98 expression. Furthermore, the VSMC CD98 expression was regulated by both extracellular matrix and growth factors.

To assess the importance of CD98 for VSMC function in vivo, we specifically deleted CD98hc in SMCs by crossing mice bearing a conditional null Slc3a2 allele (Slc3a2^fl/fl; Fig. 2 A; Féral et al., 2007) with SM22α-Cre transgenic mice that express Cre recombinase under an SMC-specific promoter (Lepore et al., 2005). In the Slc3a2^fl/flSM22α-Cre mice, the Slc3a2^fl/fl gene was deleted in >97% of aortic SMCs and CD98 protein was undetectable (Fig. 2, B and C). Despite almost complete loss of VSMC CD98, vascular morphology was apparently normal as judged by histological examination of the CA, the aorta, the femoral artery, the internal jugular vein, and the inferior vena cava (Fig. S2). The functional integrity of the vasculature was confirmed by the normal morphology of highly vascularized tissues such as lung and liver (Fig. S2), and the normal viability, fertility, and weight gain of Slc3a2^fl/flSM22α-Cre male and female mice (not depicted). We also examined the dorsal aorta at embryonic day 13.5 in Slc3a2^fl/flSM22α-Cre embryos (n = 7) and Slc3a2^fl/fl controls (n = 2). In control embryos, the vascular cells expressed CD98 at high levels. In the Slc3a2^fl/flSM22a-Cre embryos, the CD98 expression was significantly lower, but most vascular cells still stained positive for CD98 (Fig. S3). However, there was no obvious morphological difference between the two groups. Collectively, VSMC CD98hc is not necessary for normal vessel morphology under conditions where VSMCs manifest the quiescent phenotype.

Because cultured and neointimal VSMCs expressed CD98 at high levels, we assessed whether CD98hc is necessary for neointima formation after arterial injury. Slc3a2^fl/flSM22α-Cre mice (n = 8) and littermate controls (Slc3a2^fl/fl; n = 10) were subjected to the carotid injury described in Fig. 1 B. 3 wk after surgery, there was a dramatic decrease (>70%) in neointimal area in the Slc3a2^fl/flSM22α-Cre mice compared with littermate controls (Fig. 3, A and B). Furthermore, in Slc3a2^fl/flSM22α-Cre mice, the injured vessels (intima + media) contained some CD98-positive VSMCs (Fig. 3 C). In these flow cytometry studies, endothelial cells (CD31 positive) and leukocytes (CD18 positive) were excluded in the analyses. The presence of these CD98 bright cells suggests that there was a strong selection for CD98-positive cells during the neointima formation. Collectively, VSMC CD98hc is required for intimal hyperplasia, and the neointima that forms in the Slc3a2^fl/flSM22α-Cre mice is likely caused by selection of cells in which the Slc3a2^fl/fl gene is retained intact.

To study VSMCs ex vivo, we isolated the medial vessel wall layer of thoracic aortas and incubated these VSMC explants in DMEM with 10% serum (n = 6 aortas in each group). In a preliminary experiment, the number of VSMCs that grew out from Slc3a2^fl/flSM22α-Cre explants after 4 d was markedly reduced (>70%) compared with Slc3a2^fl/fl controls (Fig. 4 A). Within the Slc3a2^fl/flSM22α-Cre explant tissues, all VSMCs were negative for CD98 protein (Fig. 4 B, left). However, a selection for CD98-positive cells was observed in the outgrown cells; specifically, 78% of VSMCs that grew out from Slc3a2^fl/flSM22α-Cre explants were CD98 positive (Fig. 4 B, right). The strong selection against CD98-null VSMCs in injured vessels in vivo and in explant outgrowths ex vivo could have been caused by increased cell death or a lack of proliferation. To examine these possibilities, explant tissues at day 4 were processed for histological analyses of proliferation (BrdU incorporation over 2 h; n = 6 in each group) and apoptosis (Tdt-mediated dUTP-biotin nick-end labeling [TUNEL] assay; n = 6 in each group). In Slc3a2^fl/flSM22α-Cre explants, there were very few proliferating cells (0.3 ± 0.1%), in contrast to Slc3a2^fl/fl control explants that exhibited 2.9 ± 0.4% proliferating cells (Fig. 4 C, left). The Slc3a2^fl/flSM22α-Cre explants also exhibited a 1.8-fold increase in apoptosis relative to control explants (Fig. 4 C, right). Thus, CD98hc is important for VSMC proliferation and survival.

To precisely define the mechanism by which CD98hc mediates VSMC expansion, we acutely deleted the CD98hc gene in cultured VSMCs. Slc3a2^fl/fl VSMCs were isolated from the thoracic aorta and the CD98hc gene was deleted using adenovirus encoding Cre recombinase (Adeno-Cre); control Slc3a2^fl/fl VSMCs were infected with adenovirus encoding β-galactosidase (Adeno-LacZ). The efficiency of CD98hc gene deletion was judged by flow cytometry for CD98. At day 4 after Adeno-Cre infection, >95% of the cells had reduced CD98 protein expression to <10% of that seen in Adeno-LacZ–infected cells (unpublished data). To assess the importance of CD98hc for VSMC survival, we enumerated the number of live cells (propidium iodide [PI]–negative cells) at various times after adenovirus infection. The cells were cultured in limiting serum (0.2% serum). Beginning 5 d after Adeno-Cre infection, there was a reduction of 85 ± 3.1% in viable VSMCs, and by day 6, 98 ± 0.5% of the VSMCs were lost (compared with Adeno-Cre–infected VSMCs at day 3). Among Adeno-LacZ–infected VSMCs, there was <10% loss of live cells at days 5 (9.2 ± 2.4%) and 6 (8.1 ± 7.5%) compared with day 3 (Fig. 5 A). The massive cell death was most likely caused by apoptosis, because the Adeno-Cre–treated cells stained positive for TUNEL assay (Fig. 5 B) and exhibited activated caspase-1/3, as assessed by staining with FITC-zVAD-FMK (CaspACE; Fig. 5 C). Because PDGF-BB up-regulated CD98 expression in cultured VSMCs (Fig. 1 E), we also examined the effect of this growth factor on VSMC survival after CD98hc gene deletion. Stimulation with PDGF-BB or 10% serum caused Adeno-Cre–treated VSMCs to undergo accelerated cell death compared with cells in low serum conditions (Fig. 5 D). In conclusion, CD98hc is necessary for the survival of cultured VSMCs; i.e., they are physiologically dependent on their CD98hc expression and the CD98hc dependence is exacerbated by growth factor stimulation.

CD98hc can mediate both integrin signaling and amino acid transport. The integrin function is dependent on both the cytoplasmic and transmembrane domain, whereas the extracellular domain is necessary for amino acid transport. To assess which domains of CD98hc were important for VSMC survival, CD98hc-deficient VSMCs (Adeno-Cre treated) were reconstituted with full-length CD98hc, full-length CD69, and three CD98hc-CD69 chimeras in which the cytoplasmic (C69T98E98), the transmembrane (C98T69E98), or the extracellular (C98T98E69) domain had been switched to CD69 (Fig. 6 A; Fenczik et al., 2001; Féral et al., 2005). CD69 is a type II transmembrane protein, but it is not known to bind integrins or CD98 light chains. Full-length CD98hc and the chimera that supports integrin function (C98T98E69) rescued CD98hc-deficient VSMCs from apoptosis. In sharp contrast, full-length CD69 and the chimeras that support amino acid transport function (C69T98E98 and C98T69E98) did not rescue VSMCs from apoptosis (Fig. 6 B). Hence, the domains of CD98hc that associate with integrins are essential for VSMC survival.

Vaso-occlusive disorders, such as in-stent restenosis, vein-graft stenosis, and transplant vasculopathy, involve intimal hyperplasia, which is caused by an activation of VSMCs to migrate and proliferate (Motwani and Topol, 1998; Schwartz and Henry, 2002; Mitchell and Libby, 2007). In this paper, we show that CD98hc is markedly up-regulated in VSMCs stimulated by vascular injury and by in vitro cell-culturing conditions. Furthermore, using SMC-specific deletion of CD98hc, we show that loss of VSMC CD98hc is compatible with normal vessel morphology; however, CD98hc is required for formation of a neointima after arterial injury. The mechanism of this effect of VSMC CD98hc deletion was both suppression of proliferation and induction of apoptosis. CD98hc has two distinct functions, amino acid transport and interaction with integrins to enhance integrin signaling. Using reconstitution with CD98hc mutants, we show that the moiety of CD98hc that interacts with integrins and mediates integrin signaling is necessary and sufficient for the survival of VSMCs. These studies are the first to establish the role of CD98hc in VSMC proliferation and survival, and show that the ability of CD98hc to interact with integrins is essential for these functions. Furthermore, loss of CD98hc is selectively deleterious to activated VSMCs with high CD98 expression while sparing resident quiescent VSMCs, explaining its capacity to block neoinitima formation and indicating that CD98hc is a potential therapeutic target in vaso-occlusive disorders.

The present study shows that CD98hc expression is up-regulated when VSMCs are stimulated by vascular injury and growth factors. CD98hc expression was low in quiescent resting VSMCs in the aorta and CA but was markedly increased after vessel injury. Furthermore, CD98hc expression was high in VSMCs that grew out from tissue explants and in cultured VSMCs, and the CD98hc expression was up-regulated by fibronectin and PDGF-BB. Thus, in three different experimental systems (in vivo, ex vivo, and in vitro) CD98hc was up-regulated in activated VSMCs and can therefore be considered an activation marker. The first intron of the CD98hc gene contains a transcriptional enhancer element that is tightly regulated coordinately with cell proliferation, suggesting that this element is responsible for the increased CD98hc in activated VSMCs (Karpinski et al., 1989).

Strikingly, deletion of CD98hc in VSMCs did not seem to affect quiescent VSMCs in the normal vasculature. In the Slc3a2^fl/flSM22α-Cre embryo, a slow turnover of the CD98 protein after gene deletion might have enabled the vasculature to develop. Indeed, we found residual CD98 expression at embryonic day 13.5 (Fig. S3), which is 4 d after the SM22α-Cre transgene is reported to be expressed (Lepore et al., 2005). The finding that a mature vasculature can form and function without VSMC CD98hc shows that CD98hc can be inhibited without lethal effects on resting quiescent VSMCs. In contrast to quiescent VSMCs, CD98hc was required for activated VSMCs during the formation of intimal hyperplasia and for cultured VSMCs. The selective effect of CD98hc deletion on activated VSMCs suggests that VSMC activation leads to a dependence on CD98hc for their survival. This dependence is remarkably similar to the “addiction” of tumor cells to oncogenes, so-called oncogene addiction (Weinstein, 2002). For example, inhibition of Abl kinase with Gleevec leads to the death of BCR-Abl–expressing chronic myelogenous leukemia leukocytes but has no effect on the survival of normal leukocytes (Dan et al., 1998). The mechanism of oncogene addiction is uncertain, but one concept is that the presence of a given oncogene enhances proliferation/survival as well as apoptosis signals. Consequently, cancer cells may be addicted to the survival activity of the oncogene to avoid apoptosis (Weinstein and Joe, 2006). Our data show that the CD98 expression in activated VSMCs enables their survival; hence, these cells are similarly dependent on CD98hc to avoid apoptosis.

CD98hc performs two cellular functions: amino acid transport and integrin signaling. The extracellular domain of CD98hc is responsible for amino acid transport through interaction with one of several CD98 light chains (Fenczik et al., 2001; Wagner et al., 2001), and the transmembrane and cytoplasmic domains of CD98hc mediate interaction with β integrin subunits and the generation of survival signals such as activation of Akt (Fenczik et al., 1997, 2001; Féral et al., 2005). Reconstitution studies with CD98hc-CD69 chimeras showed that the integrin-binding moiety of CD98hc was essential for the survival of VSMCs. In contrast, the CD98hc chimeras that can interact with CD98 light chains to promote amino acid transport failed to rescue survival in vitro. Nevertheless, the interaction with CD98 light chains might be important in the functions of metabolically active VSMCs with a high proliferative rate and extensive protein synthesis. Indeed, others have reported that blockade of LAT1 and LAT2 function, two of the six CD98 light chains, can lead to apoptosis of PDGF-BB–stimulated VSMCs (Liu et al., 2004). This might explain the accelerated cell death among CD98-deficient VSMCs treated with PDGF-BB or serum in our study. On the other hand, PDGF-BB also reduces VSMC adhesion to fibronectin (Berrou and Bryckaert, 2001), and hence, the PDGF-BB–treated cells might have become more susceptible to apoptosis as a result of reduced attachment to the extracellular matrix. Clearly, the possible importance of CD98hc in VSMC amino acid transport function awaits additional study. Interestingly, the LAT1/2 inhibition did not have any effect on VSMC proliferation (Liu et al., 2004), suggesting that the integrin function and not the amino acid transport function of CD98hc regulates VSMC proliferation. In support of this, Féral et al. (2005) have shown in teratocarcinomas that the integrin-binding portion of CD98hc promotes cell proliferation. Collectively, the integrin-binding function of CD98hc seems to be an important regulator of both VSMC survival and proliferation.

VSMC αVβ3 and α5β1, two integrins that couple to CD98hc, (Zent et al., 2000) are key players in the formation of intimal hyperplasia. Inhibition of αVβ3 reduces neointima formation in several studies (for review see Kokubo et al., 2007), and α5β1 is responsible for neointima fibronectin assembly after vascular injury (Pickering et al., 2000). Binding of α5β1 integrin to fibronectin induces a synthetic VSMC phenotype and promotes VSMC proliferation and survival (Hedin et al., 1997; Mercurius and Morla, 1998; Freyer et al., 2001; Taylor et al., 2001). CD98hc interaction with integrins is important for pp125^FAK-dependent phosphoinositide 3-kinase activation with downstream activation of Akt and phosphorylation of p130^CAS leading to Rac activation (Féral et al., 2005), and these pathways have been implicated in the survival and proliferation of VSMCs (Bai et al., 1999; Stabile et al., 2003; Allard et al., 2008; Bond et al., 2008). Thus, suppressed proliferation and increased apoptosis in CD98hc-deleted VSMCs might be a result of impaired phosphoinositide 3-kinase–Akt signaling and Rac activation. Deletion of VSMC CD98 in vitro led to degradation of pp125^FAK before death (unpublished data), precluding study of the effects of CD98 on signaling events linked to pp125^FAK. However, our results underscore that only activated VSMCs require CD98hc, and the CD98hc domains that support integrin function are crucial for VSMC survival, suggesting that activation of VSMCs leads cells to become dependent on enhanced integrin signaling for survival.

Our data suggest that inhibition of VSMC CD98hc function could be therapeutically useful. In particular, we show that activated VSMCs with high CD98 expression require CD98hc for survival and proliferation, whereas quiescent normal VSMCs with low CD98 expression do not. Most vascular interventions enable local postangioplastic treatments applied in a perivascular gel or attached to an endoluminal stent. Local application of small interfering RNA, antisense oligonucleotides, or lentiviral-delivered short hairpin RNA against CD98hc is a potential therapeutic strategy (Crooke, 2004; Banno et al., 2006; Akhtar and Benter, 2007). It may also be possible to develop blocking antibodies against CD98hc that could be locally administered in connection with vascular surgery (Hashimoto et al., 1983). Finally, our data suggest that the integrin–CD98hc interaction is a potential target for a small molecular inhibitor that could selectively kill activated VSMCs. The interaction sites for CD98hc on the β1 and β3 integrin tails have been mapped (Prager et al., 2007), and recent studies have established the feasibility of developing cell-permeable inhibitors of integrin cytoplasmic domain interactions (Ambroise et al., 2002; Su et al., 2008). RGT, a synthetic peptide corresponding to the β3 integrin cytoplasmic C-terminal sequence, selectively inhibits outside-in signaling in human platelets by disrupting the interaction of αIIbβ3 integrin with Src kinase (Su et al., 2008). Thus, our studies establish the selective importance of CD98hc for the survival and proliferation of activated VSMCs, show that the connection to integrins is important for this activity, and identify CD98hc as a potential therapeutic target in vaso-occlusive disorders.

All surgical procedures were approved by the Institutional Animal Care and Use Committee at the University of California, San Diego. CD98hc conditional null mice (Slc3a2^fl/fl; Féral et al., 2007) were crossed with SM22α-Cre transgenic mice (Lepore et al., 2005) to delete the CD98hc gene (Slc3a2) in SMCs. For all in vivo and ex vivo experiments, Slc3a2^fl/flSM22α-Cre mice and sex-matched Slc3a2^fl/fl littermate controls aged 7–9 wk were used. The carotid injury model (Fig. 1 B) was based on a previously published study (Simon et al., 2000). The CA was dilated at 91 Pa for 20 s, and the endothelium was denuded with a nylon wire (0.19 mm in diameter). To facilitate neointima formation, the postangioplastic blood flow was reduced by leaving only the occipital artery open at the carotid bifurcation (Korshunov and Berk, 2003). Injured arteries were analyzed 450–550 µm proximal to the carotid bifurcation. Neointimal area was quantified (pixels) using the National Institutes of Health (NIH) Image J software (available at http://rsbweb.nih.gov/ij/) on cross sections visualized by autofluorescence (excitation wavelength = 488 nm; 530 nm band-pass filter; Fig. 3 A).

Tissue samples were processed and prepared for standard histological and immunohistochemical procedures (Fogelstrand et al., 2005). Formalin-fixed tissues were pressure perfused at 100 mmHg. Formalin-fixed paraffin sections were used for all histology except for CD98 immunostaining, which was performed on frozen sections. The following antibodies and kits were used: rabbit anti–human SM22α (Proteintech Group, Inc.), horseradish peroxidase–conjugated goat anti–rabbit antibody (SouthernBiotech), rat anti–mouse CD98 (clone RL388; eBioscience), rat IgG2a isotype control (eBioscience), Alexa Fluor 488–conjugated goat anti–rat (Invitrogen), the In Situ Cell Death Detection Kit (Roche), and the BrdU Labeling and Detection Kit I (Roche).

Immunolabeled cells were analyzed on a flow cytometer (FACSCalibur; BD) with CellQuest software (BD). For all vessels, the adventitia was removed after a 5-min treatment with 10 mg/ml collagenase at room temperature, and the remaining intima + media were digested to a single-cell suspension at 37°C in a 10 mg/ml collagenase and 2.5 mg/ml elastase solution (Worthington Biochemical Corp.). Endothelial cells (CD31 positive) and leukocytes (CD18 positive) were excluded by gating during the data analyses. The following antibodies were used: PE-conjugated anti–mouse CD98 (clone RL388; eBioscience), PE-conjugated rat IgG2a control antibody (eBioscience), FITC-conjugated anti–mouse CD31 (clone MEC 13.3; BD), and FITC-conjugated anti–mouse CD18 antibodies (clone M18/2; eBioscience). For cell counting, cells were trypsinized and resuspended in a fixed volume of PBS with 5 µg/ml PI (Invitrogen). The relative number of live cells (PI negative) was determined by collecting resuspended cells for 60 s on the flow cytometer.

Approximately 1 × 1 mm media explants were prepared from the thoracic aorta (Choi et al., 2004). Endothelial cells were killed by drying the lumen with 2 × 20 ml of air from a syringe. To facilitate the removal of the adventitia, the isolated vessels were treated for 5 min with 10 mg/ml collagenase at 37°C. The isolated media explants were cultured luminal side down in DMEM with 10% serum, under the weight of a culture plate insert. At day 4, all outgrown cells from each explant were counted manually under a phase contrast inverted tissue culture microscope (model CK2; Olympus). For flow cytometry analyses, all outgrown cells at day 4 were collected by trypsinization, and the remaining explant tissues were digested to a single-cell suspension. For proliferation analyses, BrdU was added to the culture media (10-µM final concentration) 2 h before harvest.

VSMCs were isolated from the thoracic aorta from Slc3a2^fl/fl mice. The cultured VSMCs were characterized by morphological (Fig. S4 A) and immunological criteria (flow cytometry quantification of smooth muscle α-actin; Fig. S4 B). For all in vitro experiments, VSMCs were used at passages 5–7. All culture plates were coated with a mix of 1 µg/ml fibronectin and 0.02% gelatin for 30 min at room temperature. The CD98hc gene was deleted using Adeno-Cre (600 particles/cell). Adeno-LacZ was used as control (Vector Laboratories).

700,000 Slc3a2^fl/fl VSMCs were plated in a 10-cm petri dish coated with 200 µl Matrigel (BD) in low-glucose DMEM supplemented with 1% serum and 2% Matrigel. The medium was changed on day 1 to low-glucose DMEM supplemented with 0.2% serum and 2% Matrigel. On day 2, the cells were detached with dispase (BD) and replated in a 12-well plate (20,000 cells/well) coated with either 20 µg/ml of mouse laminin-1 (BD) or 10 µg/ml of bovine fibronectin (Sigma-Aldrich). 10 µg/ml laminin-1 or 10 µg/ml fibronectin was also added to the media. 20 ng/ml PDGF-BB was added to half of the cells. On day 3, the VSMC CD98 expression was analyzed using a flow cytometer.

DNA fragmentation on Adeno-Cre–treated VSMCs was analyzed with In Situ Cell Death Detection Kit, and caspase-1/3 activity was demonstrated with flow cytometry using FITC-zVAD-FMK (CaspACE; Promega) according to manufacturer’s instructions.

Reconstituted cells were generated by infecting Slc3a2^fl/fl VSMCs with MSCV-IRES-GFP retrovirus (Cherry et al., 2000) encoding human CD98hc, human CD69, or CD98hc-CD69 chimeras (Fig. 6 A) and GFP (Fenczik et al., 2001). Viruses were generated in the Phoenix EcoPack mouse ecotropic packaging cell line (Imgenex) by transfection (Lipofectamine Plus; Invitrogen) of the MSCV-IRES-GFP constructs together with a pCL-Eco mouse ecotropic packaging plasmid (Imgenex). On days 2 and 3 after transfection, the virus-containing cell-culture media were filtered (0.45-µm SFCA filter; Corning), supplemented with 6 µg/ml polybrene (Sigma-Aldrich), and added to the VSMCs. 2 d later, the VSMCs were replated in 12-well plates and infected with Adeno-Cre to delete the endogenous mouse CD98hc gene. At days 3, 4, 5, and 6 after Adeno-Cre treatment, the number of PI-negative/GFP-positive cells was counted using a flow cytometer.

Data were analyzed with a two-tailed Student’s t test and a two-way analysis of variance with the Bonferonni posthoc test using Prism 5 software (GraphPad Software, Inc.).

Fig. S1 shows flow cytometry data analyses on injured CAs. Fig. S2 shows the histology of different vessels and organs from Slc3a2^fl/flSM22α-Cre and Slc3a2^fl/fl mice. Fig. S3 shows CD98 immunostaining of the dorsal aorta from Slc3a2^fl/flSM22α-Cre and Slc3a2^fl/fl embryos at embryonic day 13.5. Fig. S4 shows morphological and immunological characterization of isolated VSMCs.

We gratefully acknowledge R. Weichert for assistance in genotyping the mice.

This study was supported by grants from the NIH (HL 31950, AR27214, and HL 078784). P. Fogelstrand held a postdoctoral fellowship from the Swedish Research Council, and C.C. Féral held a postdoctoral fellowship from the Arthritis Foundation.

The authors have no conflicting financial interests.