T-cadherin Is Essential for Adiponectin-mediated Revascularization*

Background: Adiponectin has vascular protective actions and is bound by T-cadherin. Results: T-cadherin-deficient mice lack skeletal muscle tissue-resident adiponectin and display impaired revascularization that is not improved by treatment with exogenous adiponectin. Conclusion: Expression of T-cadherin is critical for revascularization actions of adiponectin in vitro and in vivo. Significance: The T-cadherin/adiponectin interaction is important for vascular homeostasis. Adipose tissue secretes protein factors that have systemic actions on cardiovascular tissues. Previous studies have shown that ablation of the adipocyte-secreted protein adiponectin leads to endothelial dysfunction, whereas its overexpression promotes wound healing. However, the receptor(s) mediating the protective effects of adiponectin on the vasculature is not known. Here we examined the role of membrane protein T-cadherin, which localizes adiponectin to the vascular endothelium, in the revascularization response to chronic ischemia. T-cadherin-deficient mice were analyzed in a model of hind limb ischemia where blood flow is surgically disrupted in one limb and recovery is monitored over 28 days by laser Doppler perfusion imaging. In this model, T-cadherin-deficient mice phenocopy adiponectin-deficient mice such that both strains display an impaired blood flow recovery compared with wild-type controls. Delivery of exogenous adiponectin rescued the impaired revascularization phenotype in adiponectin-deficient mice but not in T-cadherin-deficient mice. In cultured endothelial cells, T-cadherin deficiency by siRNA knockdown prevented the ability of adiponectin to promote cellular migration and proliferation. These data highlight a previously unrecognized role for T-cadherin in limb revascularization and show that it is essential for mediating the vascular actions of adiponectin.


Adipose tissue secretes protein factors that have systemic actions on cardiovascular tissues. Previous studies have shown
that ablation of the adipocyte-secreted protein adiponectin leads to endothelial dysfunction, whereas its overexpression promotes wound healing. However, the receptor(s) mediating the protective effects of adiponectin on the vasculature is not known. Here we examined the role of membrane protein T-cadherin, which localizes adiponectin to the vascular endothelium, in the revascularization response to chronic ischemia. T-cadherin-deficient mice were analyzed in a model of hind limb ischemia where blood flow is surgically disrupted in one limb and recovery is monitored over 28 days by laser Doppler perfusion imaging. In this model, T-cadherin-deficient mice phenocopy adiponectin-deficient mice such that both strains display an impaired blood flow recovery compared with wildtype controls. Delivery of exogenous adiponectin rescued the impaired revascularization phenotype in adiponectin-deficient mice but not in T-cadherin-deficient mice. In cultured endothelial cells, T-cadherin deficiency by siRNA knockdown prevented the ability of adiponectin to promote cellular migration and proliferation. These data highlight a previously unrecognized role for T-cadherin in limb revascularization and show that it is essential for mediating the vascular actions of adiponectin.
Above normal body mass is associated with increased mortality due in large part to the increased prevalence of vascular diseases including ischemic heart disease and stroke (1). It is now well appreciated that adipose tissue functions as an endocrine organ secreting both pro-and anti-inflammatory factors, referred to as adipokines, that affect the functions of cardiovascular tissues (2,3). Under conditions of obesity, proinflamma-tory cytokine secretion predominates, and the secretion of antiinflammatory factors is attenuated. Adiponectin is an antiinflammatory adipokine that is expressed as an abundant, multimeric protein. Circulating adiponectin levels display a strong inverse correlation with body mass index (3). Adiponectin has a well described role in maintaining metabolic homeostasis (2). Adiponectin also functions to protect the heart and vasculature from stress in a variety of models (3)(4)(5)(6)(7). Our laboratories and others have shown previously that adiponectin promotes vascular function and angiogenesis using endothelial cell culture systems (8,9), mouse models of tumor growth (10,11), and a mouse model of ischemic hind limb revascularization (12,13).
Although it is clear that adiponectin has vascular protective actions, the receptor(s) that mediates these effects is unknown. Recent studies have identified AdipoR1 and AdipoR2 as the receptors responsible for mediating the metabolic effects of adiponectin in both skeletal muscle and liver, although the reported metabolic phenotypes of AdipoR1-and AdipoR2-deficient strains are somewhat contradictory (14 -20). AdipoR1 and AdipoR2 are widely expressed seven-transmembrane receptors with a topology opposite that of G-protein-coupled receptors. AdipoR1 preferentially binds globular adiponectin, whereas AdipoR2 has equal affinities for globular and fulllength adiponectin (14). A third candidate receptor is T-cadherin, a glycosylphosphatidylinositol-anchored protein that binds the hexameric and high molecular weight isoforms of adiponectin (21).
T-cadherin was first identified over two decades ago as an axon guidance molecule (22) and modulator of neural crest cell migration (23). It is now appreciated that T-cadherin has functions that extend beyond the typical cadherin behavior of cellto-cell adhesion. T-cadherin is highly expressed in the vasculature including endothelial cells (24), smooth muscle cells (25), and pericytes (26). It has been implicated in modulation of angiogenic activities in cultured endothelial cells; however, some studies report that it promotes angiogenesis (27)(28)(29)(30)(31), whereas others report its attenuation (32,33). Importantly, T-cadherin has been shown to localize adiponectin to tumor vasculature (31) and heart tissue (47). In human subjects, single nucleotide polymorphisms in CDH13, the gene encoding T-cadherin, are associated with circulating levels of adiponectin (34 -39). Genome wide association studies have also identified links between CDH13 and cancer (40), blood pressure (41), blood lipid levels (42), metabolic syndrome (43,44), type II diabetes, and ischemic stroke (44). Given its high level of expression on vascular cells and its role in modulating angiogenic behavior, we hypothesized that the T-cadherin/adiponectin interaction would be important for the revascularization activity of this adipokine in ischemic tissue.
The aim of this study was to determine whether expression of T-cadherin is required for the revascularization activity of adiponectin in the ischemic hind limbs of mice, a model of peripheral artery disease (45). The first objective was to determine whether T-cadherin deficiency in mice phenocopies the impairment in revascularization that is observed in adiponectin-deficient mice. Next, we assessed whether T-cadherin is essential for the provascularization effects of adiponectin by comparing the responses of adiponectin-deficient mice and T-cadherin-deficient mice to the administration of exogenous adiponectin. Finally, we examined whether cultured T-cadherin-deficient endothelial cells are defective in their proangiogenic responses to exogenously administered adiponectin in vitro.

EXPERIMENTAL PROCEDURES
Mice-Adiponectin-deficient (APN-KO) 2 mice were provided by N. Maeda, T. Funahashi, and Y. Matsuzawa (Osaka University, Osaka, Japan). T-cadherin-deficient (Tcad-KO) mice were described previously (31). AdipoR1-KO and AdipoR2-KO mice were originally created by Deltagen (San Mateo, CA) and ordered from Mutant Mouse Regional Resource Centers and The Jackson Laboratory (Bar Harbor, ME), respectively. T-cadherin-deficient, adiponectin-deficient double KO (TA-dKO) mice were produced for this study through cross-breeding. Wild-type mice were from Charles River Laboratories (Wilmington, MA). All mice were bred in a C57BL/6 background. The Institutional Animal Care and Use Committee of Boston University approved all study procedures. Mice were maintained on a 12-h light/dark schedule and given food and water ad libitum.
Hind Limb Ischemia Model-Ketamine (100 mg/kg) and xylazine (10 mg/kg) were used to anesthetize 9 -11-week-old male mice from the different experimental groups. A small incision was made in the upper portion of the hind limb. Using a 6-0 silk suture, the femoral artery, vein, and nerve of one limb were permanently ligated. Downstream collateral branches were severed, and a small section of the ligated femoral artery, vein, and nerve was excised. The wound was closed using surgical staples that remained in place for 10 -14 days after surgery. Limb perfusion was monitored before surgery, after surgery, and at the following time points: day 3, day 7, day 14, day 21, and day 28 by laser Doppler perfusion imaging (LDPI). Data were analyzed as a blood flow ratio (ischemic/ non-ischemic) to control for body temperature and environmental conditions.
Hydrodynamic Plasmid Delivery-In some experiments, pLEV113-mADIPO-hFc (LakePharma, Belmont, CA), an adiponectin-Fc fusion construct (adiponectin plasmid), or pLEV113-MCS (control plasmid) was injected into mice 7 days prior to hind limb ischemia surgery. Adiponectin or control plasmid (24 g) was injected via the tail vein into restrained mice in 2.0 -2.5 ml of total saline volume (10% of body weight) in 5 s according to the protocol published previously (46).
ELISA-Blood was harvested from mice either by tail vein bleed or cardiac puncture. After clotting at room temperature, samples were centrifuged at 7,000 rpm, and serum was isolated. Serum adiponectin concentration was determined by ELISA (B-Bridge, Cupertino, CA) according to the manufacturer's instructions.
Treadmill-To assess limb function at base line and 10 days post-hind limb ischemia surgery, maximum treadmill running distance was determined. To acclimate to the equipment, mice were trained for 10 min prior to the start of the experiment. The trial began at a speed of 5 m/min. The belt speed was increased by 2 m/min every 5 min. The trial was stopped when mice were exposed to the stimulus at the rear of the treadmill for 30 s.
Western Blotting-Gastrocnemius muscle and epididymal adipose tissue were harvested from 10-week-old male mice and snap frozen. Tissue was homogenized using radioimmune precipitation assay buffer containing protease inhibitors and 100 mM PMSF. Samples were centrifuged at 14,000 rpm for 10 min. The protein concentration of the supernatant was assessed using the Pierce (Thermo Fisher Pierce Protein Biology Products, Rockford, IL) bicinchoninic acid (BCA) assay. Serum or tissue lysate was resolved by SDS-PAGE and transferred to a PVDF membrane. Native Western blots of the adiponectin isoforms were performed using non-reduced, non-denatured protein samples. Membranes were blocked with 5% nonfat milk in PBS ϩ 0.1% Tween 20 for 1 h at room temperature. Primary antibody (1:1,000) was added to the membrane overnight at 4°C in blocking buffer. Adiponectin (R&D Systems, Minneapolis, MN), T-cadherin (R&D Systems), tubulin (Santa Cruz Biotechnology, Inc., Dallas, TX), and ␤-actin (Cell Signaling Technology, Inc., Danvers, MA) antibodies were used. The membrane was then incubated with secondary HRP-conjugated antibodies (1:5,000) for 1 h at room temperature and developed using SuperSignal West Pico Chemiluminescent Substrate (Pierce) and Amersham Biosciences Hyperfilm (GE Healthcare).
Immunofluorescence of Skeletal Muscle Tissue and Large Vessels-Gastrocnemius muscles from WT, APN-KO, Tcad-KO, AdipoR1-KO, and AdipoR2-KO mice were embedded in optimum cutting temperature compound. Cryostat sections of 5 m were cut. Specimens were fixed by 4% paraformaldehyde for 15 min. Nonspecific binding sites were blocked with 3% horse serum and 3% bovine serum albumin in PBS for 1 h at room temperature. Then samples were incubated with both goat polyclonal anti-T-cadherin antibody (R&D Systems) and rabbit polyclonal anti-adiponectin antibody (ThermoScientific, Billerica, MA) overnight at 4°C followed by incubation with anti-goat IgG conjugated with Alexa Fluor 488 (Invitrogen) for detection of T-cadherin and anti-rabbit IgG conjugated with Alexa Fluor 594 (Invitrogen) for detection of adiponectin. DAPI was used as a nuclear stain. Images were recorded using a Zeiss LSM 710-LIVE DuoScan (Carl Zeiss Microscopy, LLC, Thornwood, NY) confocal microscope.
Immunofluorescence of Muscle Capillaries-Rectus femoris muscle from WT mice was frozen in optimum cutting temperature compound, and 24-m cryostat sections were cut, collected on slides, and fixed with 30% acetic acid, 70% ethanol for 5 min at room temperature. Samples were incubated simultaneously with rabbit polyclonal anti-adiponectin antibody (Affinity BioReagents, Golden, CO), goat polyclonal anti-Tcadherin antibody (R&D Systems), and rat monoclonal anti-CD31 antibody (BD Biosciences) overnight at 4°C followed by incubation with the following secondary antibodies: donkey IgG (Cy5-conjugated; Jackson ImmunoResearch Laboratories, West Grove, PA) for detection of adiponectin, rat IgG (Alexa Fluor against goat IgG, Cy3-conjugated; Jackson ImmunoResearch Laboratories) for detection of T-cadherin, and rabbit IgG (Alexa Fluor 488-conjugated; Invitrogen) for detection of CD31. All antibodies were diluted in Dako Antibody Diluent (Dako, Carpinteria, CA). Hoechst (Sigma-Aldrich) was used as a nuclear stain. Images were recorded using a Zeiss LSM 710 NLO (Carl Zeiss Microscopy, LLC) confocal microscope. Z-stack image processing used Volocity 3D Image Analysis software (PerkinElmer Life Sciences).
Cell Culture-Human umbilical vein endothelial cells (HUVECs; American Type Culture Collection, Manassas, VA) were cultured in EGM-2 medium up to passage 6. Cells were transfected with 40 nM siRNA (ThermoScientific, siGENOME SMART Pool) and Lipofectamine RNAiMAX (Invitrogen) for 72 h. For the migration assay, cells were serum-depleted overnight in EBM-2 ϩ 0.5% FBS. In a 6-well plate, confluent HUVECs were scratched with a 200-l pipette tip and cultured for 16 h in DMEM ϩ 1% FBS Ϯ 10 g/ml recombinant fulllength adiponectin protein (BioVendor, Inc., Asheville, NC). Cells were photographed at hours 0 and 16 to determine the extent of migration, which was expressed as percent gap closure. To measure HUVEC proliferation, transfected cells were cultured in a 96-well plate. Growth medium consisting of DMEM ϩ 5% FBS Ϯ 10 g/ml recombinant adiponectin protein was added 24 h prior to the start of the BrdU incorporation assay. This BrdU ELISA-based assay was performed according to the manufacturer's instructions (Roche Diagnostics).
Real Time Quantitative PCR-RNA was isolated from HUVECs using RNeasy Micro (Qiagen, Germantown, MD) according to the kit instructions. RNA was isolated from epididymal adipose tissue and gastrocnemius muscle using an RNeasy Plus Universal kit (Qiagen) and RNeasy Fibrous Tissue Mini kit (Qiagen), respectively. cDNA was synthesized from 500 ng of RNA with the Superscript III First-Strand Synthesis System (Invitrogen). For HUVEC samples, real time PCR was performed using SYBR Green PCR Mastermix (Invitrogen). For tissue samples, real time PCR was performed using TaqMan PCR Mastermix. Expression of mRNA was determined for T-cadherin, AdipoR1, AdipoR2, and adiponectin. Expression was normalized relative to GAPDH or ␤-actin.
Immunoprecipitation-Confluent HUVECs were cultured with serum-free EBM-2 medium (Lonza, Hopkinton, MA) for 16 h and then treated with 25 g/ml high molecular weight recombinant adiponectin (rAPN) (BioVendor) for 60 min. Cells were washed with PBS and cross-linked with 2.5 mmol/liter dimethyl 3,3Ј-dithiopropionimidate dihydrochloride (Sigma-Aldrich) for 60 min at room temperature. Cells were washed with PBS and treated with lysis buffer containing protease and phosphatase inhibitor mixtures (Thermo Scientific). Cell lysates were incubated with normal rabbit IgG (Cell Signaling Technology, Inc.) or anti-adiponectin (Abcam, Cambridge, MA) antibodies at 4°C overnight, and then magnetic protein G beads (Dynabeads Co-Immunoprecipitation kit, Invitrogen) were added. After a 1-h incubation (room temperature), the beads were washed, and target antigen was eluted with elution buffer and SDS sample buffer (Bio-Rad). Protein was detected by Western blotting using the following primary antibodies: T-cadherin (R&D Systems), adiponectin (Abcam), and ␤-actin (Cell Signaling Technology, Inc.).
Statistical Analysis-Data are presented as mean Ϯ S.E. Analyses were performed using GraphPad Prism software. For LDPI analysis, a two-way repeated measured analysis of variance was performed with Bonferroni post hoc tests. When directly comparing three or more groups, a one-or two-way analysis of variance was performed with post hoc Student's t tests. A p value of less than 0.05 was considered statistically significant.

T-cadherin and Adiponectin Are Co-localized in Skeletal
Muscle Tissue-It has been reported previously that T-cadherin is important for binding and localizing adiponectin to tumor vasculature (31) and cardiac tissue (47). To examine this activity in skeletal muscle, immunofluorescence analysis of murine gastrocnemius muscle was performed. Localization of adiponectin and T-cadherin at the membranes of myocytes (Fig. 1A) as well as capillaries (Fig. 1B) was readily apparent in wild-type mice. Adiponectin and T-cadherin also co-localized on the luminal surface of the larger, muscular blood vessels (Fig.  1C). Strikingly, adiponectin was not detected in the muscle of either APN-KO mice or Tcad-KO mice, suggesting that T-cadherin is required for its localization to these structures. Also of note, T-cadherin expression in muscle was markedly reduced in samples isolated from APN-KO mice. In contrast to these observations with the Tcad-KO mice, deficiency of AdipoR1 or AdipoR2 had little or no effect on the localization of adiponectin to muscle tissue. Likewise, AdipoR1 or AdipoR2 deficiency had no apparent effect on T-cadherin expression.
T-cadherin Localizes Adiponectin to Skeletal Muscle Tissue-Western blot analysis of gastrocnemius muscle was performed to quantify levels of adiponectin and T-cadherin in the different experimental strains of mice (Fig. 2, A-C). As expected, adiponectin was not detected in the muscle lysates of APN-KO mice. Similar to the immunohistochemical analysis, adiponectin protein was also not detected in muscle lysates from Tcad-KO mice (Fig. 2, A and B). Similarly, tissue levels of T-cadherin were reduced in APN-KO mice (Fig. 2, A and C), suggestive of a feedback regulatory loop between adiponectin and T-cadherin. Muscle T-cadherin is detected as a double band representing the mature and propeptide forms of the protein (22). Adiponectin deficiency led to reduction in the expression of both forms with nearly complete disappearance of the higher molecular weight propeptide-containing isoform ( Fig.  2A). Similar levels of T-cadherin mRNA expression in gastrocnemius muscle isolated from wild-type and APN-KO mice were observed (Fig. 2D), indicating that the regulation of T-cadherin is post-transcriptional.
Consistent with a decrease in tissue-localized adiponectin, mice deficient in T-cadherin have elevated levels of circulating, unbound serum adiponectin (Fig. 2E). By Western blot analysis, it appears that the high molecular weight isoform of adiponectin is selectively elevated in the serum of Tcad-KO mice (Fig.  2F). This observation supports the notion that T-cadherin preferentially binds the high molecular weight isoform (21). As expected, serum adiponectin protein was not detected by ELISA or Western blot in adiponectin-deficient mice. Despite markedly elevated adiponectin levels in the serum of the Tcad-KO mice, adipose tissue levels of adiponectin transcript and protein were unaffected by T-cadherin deficiency (Fig. 2,  G-I). Furthermore, adiponectin mRNA expression in muscle was negligible compared with that of adipose tissue and did not differ between groups. Also of note, the distribution of adiponectin isoforms in adipose tissue was not different between wild-type and Tcad-KO mice (Fig. 2J). Taken together, these data suggest that T-cadherin is critical for tissue localization of adiponectin and that the elevated circulating adiponectin found in the Tcad-KO mouse results from its liberation from tissue rather than an increase in its synthesis.
T-cadherin Is Required for Limb Revascularization-Because of the dependence of adiponectin tissue localization on T-cadherin expression and the dependence of T-cadherin expression on adiponectin levels, further studies were performed to determine the functional significance of the protein/protein interaction in a model of peripheral artery disease (45). Wild-type, APN-KO, and Tcad-KO mice underwent analysis in a model of unilateral hind limb ischemia where a small segment of the femoral artery, nerve, and vein was ligated and excised. Limb perfusion was then monitored for 28 days using LDPI. At the time of surgery (9 -11 weeks old), male APN-KO and Tcad-KO mice were of similar body weight to WT mice and did not exhibit metabolic abnormalities under the conditions of our assays (supplemental Fig. 1, A-C). Impaired recovery of the laser Doppler signal was observed in both adiponectin-deficient and Tcad-KO mice relative to isogenic wild-type mice (Fig. 3, A  and B). Statistically significant reductions in LDPI were observed at 7-, 14-, 21-, and 28-day time points.
A common clinical assessment of limb function in patients with peripheral artery disease is walking distance on a treadmill (48). Thus, to assess functional recovery from hind limb ischemia, mice were subjected to treadmill running until exhaustion at 10 days postsurgery. Whereas no difference in running ability between strains was detected at base line (Fig. 3C), after chronic ischemia, APN-KO and Tcad-KO mice displayed reduced maximum treadmill running distance compared with wild-type mice (Fig. 3D). Wild-type mice equally utilized both limbs for running, whereas APN-KO and Tcad-KO mice favored their non-ischemic limb (shown as a video in supplemental Fig. 2). Collectively, these data show that the decreases in reperfusion as assessed by LDPI in APN-KO and Tcad-KO mice correspond to functional limitations in limb performance. These data also show that APN-KO mice and Tcad-KO mice display similar phenotypes with regard to their revascularization response to chronic ischemia.
T-cadherin Is Essential for Adiponectin-mediated Revascularization-We have reported previously that elevating levels of serum adiponectin by adenoviral overexpression rescues the impaired blood flow recovery of adiponectin-deficient mice FIGURE 1. T-cadherin and adiponectin are co-localized in skeletal muscle tissue. A, representative confocal images of adiponectin and T-cadherin immunofluorescence in gastrocnemius muscle isolated from wild-type, APN-KO, Tcad-KO, AdipoR1-KO, and AdipoR2-KO mice. DAPI was used as a nuclear stain. B, skeletal muscle isolated from wild-type mice was stained with antibodies targeting adiponectin, T-cadherin, and an endothelial cell marker (CD31). Hoechst was used as a nuclear stain, and white arrows indicate co-localization in confocal images. C, major vessels in skeletal muscle were observed for both adiponectin and T-cadherin expression. Co-localization was observed on the luminal surface of vessels.
in the hind limb ischemia model (8). Thus, to determine causality for a functional interaction between T-cadherin and adiponectin, an in vivo rescue experiment was performed via the expression of exogenous adiponectin. As detailed in Fig. 2E, Tcad-KO mice had higher levels of serum adiponectin than wild-type mice at base line. As high base-line levels of adiponectin would confound an adiponectin rescue experiment in the Tcad-KO mice, a strain that was deficient in both T-cadherin and adiponectin (TA-dKO) was prepared for this purpose. Thus, the rescue of adiponectin-null mice can be directly compared in APN-KO and TA-dKO mice. As described previously (47), TA-dKO mice are viable and breed with normal Mendelian frequency.
Adiponectin was administered to APN-KO and TA-dKO mice using hydrodynamic delivery of a plasmid vector that expresses murine adiponectin (pLEV113-mADIPO-hFc). In this procedure, the plasmid is rapidly delivered in a 0.9% saline solution via tail vein injection. This results in the efficient transduction of hepatocytes that express the protein of interest (46). One week after hydrodynamic injection, mice were subjected to  hind limb ischemia surgery. At the time of surgery, blood was collected by tail vein to determine the concentration of serum adiponectin. Adiponectin levels were undetectable in control plasmid-injected APN-KO and TA-dKO mice ( Table 1). The impaired revascularization response observed in TA-dKO mice injected with control plasmid was not different from that of APN-KO or Tcad-KO mice injected with control plasmid. Deficiency of both adiponectin and T-cadherin did not exacerbate the impairment in revascularization beyond that of the single KO mice. APN-KO mice injected with adiponectin plasmid reached physiological levels of adiponectin 1 week after treatment ( Table  1). As anticipated, replenishment of adiponectin improved blood flow recovery in adiponectin-deficient mice (Fig. 4, A and C). However, adiponectin delivery did not improve blood flow recovery in TA-dKO mice (Fig. 4, B and C). These data suggest that expression of T-cadherin is required for the revascularization actions of adiponectin in the hind limb ischemia model.

T-cadherin Is Required for Adiponectin-induced Migration and Proliferation of Endothelial
Cells-Revascularization as a result of chronic ischemia is largely due to an adaptive angiogenesis and/or arteriogenesis response involving endothelial cell migration and proliferation (49). Thus, cell culture experiments were initiated to further delineate the functional relationship between T-cadherin and adiponectin. It has been reported previously that adiponectin promotes endothelial cell migration and proliferation in vitro (9,50,51). The dependence of in vitro angiogenic activities of adiponectin on T-cadherin expression in endothelial cells was evaluated. Using low passage HUVECs, T-cadherin mRNA and protein expression was reduced by siRNA directed against this transcript but not by control siRNA (Fig. 5A). The expression of other reported adiponectin-binding proteins, AdipoR1 and AdipoR2, was not altered by knockdown of T-cadherin. In a scratch-induced migration assay, knockdown of T-cadherin had little or no effect on HUVEC migratory activity. The addition of recombinant adiponectin promoted migration in cells transfected with control siRNA but not in those that received T-cadherin-targeting siRNA (Fig. 5B). In an assay of HUVEC proliferation, knockdown of T-cadherin had a small effect on BrdU incorporation into DNA. Notably, adiponectin-induced proliferation was not observed in HUVECs deficient in T-cadherin (Fig. 5C). These data show that expression of T-cadherin is required for the ability of adiponectin to promote endothelial cell migration and proliferation in vitro.
One mechanism for the revascularization actions of T-cadherin presented here may be its ability to bind and localize adiponectin to the vascular endothelium. To determine whether a direct interaction occurs between T-cadherin and adiponectin, a co-immunoprecipitation experiment was performed. After overnight serum starvation, HUVECs were cultured in the presence or absence of high molecular weight rAPN. After a brief cross-linking, cell lysates were incubated with an antibody targeting adiponectin. Protein complexes were isolated and visualized by Western blotting (Fig. 5D). An adiponectin⅐T-cadherin co-immunoprecipitation complex was observed in HUVECs treated with high molecular weight adiponectin, suggesting a direct protein interaction. Thus, adiponectin binds T-cadherin on endothelial cells, corroborating previous reports that T-cadherin directly binds high molecular weight adiponectin in co-transfected human embryonic kidney 293 cells (21) and C2C12 murine myotubes (47).
Although AdipoR1 and AdipoR2 are not essential for localization of adiponectin to skeletal muscle (Figs. 1A and 2A), we assessed any functional importance for these receptors in the proangiogenic action of adiponectin in vitro. Similar to previous experiments with T-cadherin siRNA knockdown, HUVECs were treated with siRNA targeting AdipoR1 or AdipoR2. Expression was reduced by Ͼ88% compared with control siRNA-treated cells. Deficiency of either AdipoR1 or AdipoR2 prevented the migratory (Fig. 5E) and proliferative actions of adiponectin (Fig. 5F), indicating that expression of AdipoR1 or AdipoR2 is functionally significant for the revascularization actions of adiponectin.

DISCUSSION
Adiponectin is almost exclusively produced by adipose tissue, and it acts on cardiovascular tissues in a paracrine manner. Although it is widely recognized that adiponectin has vascular protective and proangiogenic actions (9 -12), the membrane proteins that mediate these actions are unknown. Here we assessed the functional interaction between adiponectin and T-cadherin, a glycosylphosphatidylinositol-anchored membrane protein, using both in vitro and in vivo assays. Our findings show that T-cadherin is required for localization of adiponectin to skeletal muscle and that Tcad-KO mice are phenotypically similar to APN-KO mice in that both strains display impaired blood flow recovery compared with wildtype mice in a model of hind limb revascularization. The impaired revascularization phenotype could be rescued by the administration of adiponectin in APN-KO mice but not in mice that were lacking T-cadherin. These studies confirm the critical role of T-cadherin in ischemia-induced revascularization and are the first to demonstrate that T-cadherin functions are essential in mediating the proangiogenic activity of adiponectin.
We report that adiponectin is present on the cell surfaces of the vascular endothelium and myocytes in gastrocnemius muscle tissue of wild-type mice. Adiponectin localization was not detected in the muscle of Tcad-KO mice. Based upon these data, it is reasonable to hypothesize that T-cadherin facilitates the ability of adiponectin to promote vascular function through its localization of this adipokine to target tissues. Consistent with these observations, it has also been reported that T-cadherin is important for the localization of adiponectin to tumor  vasculature (31) and the heart (47). Adiponectin is abundantly present in the serum of wild-type mice, but its levels in serum are elevated ϳ4-fold in mice that lack T-cadherin. Because the expression of adiponectin protein and transcript by adipose tissue is not influenced by a deficiency in T-cadherin, it would appear that ϳ75% of the organism's total adiponectin is bound to tissue via a T-cadherin-dependent mechanism. In contrast to T-cadherin, the localization of adiponectin to skeletal muscle tissue was not dependent on expression of either AdipoR1 or AdipoR2. Thus, although AdipoR1 may have important metabolic effects in skeletal muscle (16,19,20), it appears that T-cadherin is primarily responsible for binding and localizing adiponectin to that tissue. Furthermore, based upon these data, the presence of AdipoR1 and AdipoR2 in the TA-dKO is not sufficient to enable a revascularization response to the administration of adiponectin. However, because only single knockout mouse models were examined in this study, we cannot exclude an effect of simultaneous AdipoR1 and AdipoR2 deficiency on adiponectin localization. Further evidence of a functional T-cadherin/adiponectin interaction comes from the observation that T-cadherin protein expression in skeletal muscle was markedly repressed in APN-KO mice. In contrast, T-cadherin mRNA expression was not altered in APN-KO mice compared with wild type, indicating that this regulation is post-transcriptional. Denzel et al. (47) also observed the coordinate regulation of adiponectin and T-cadherin expression in heart. Similarly, a correlation between adiponectin and T-cadherin expression was described in a model of liver fibrosis where elevated serum adiponectin was associated with elevated hepatic expression of T-cadherin (52). Thus, it is tempting to speculate that adiponectin signals through T-cadherin to support the expression of T-cadherin via a positive feedback loop mechanism. Accordingly, altered expression of T-cadherin under physiological or pathological conditions could then lead to changes in the amounts of circulating and tissue-localized adiponectin.
Adiponectin is an unusual receptor ligand because it circulates at very high levels and has a complex structure. In humans, adiponectin levels in serum range from 3 to 30 g/ml, and it represents ϳ0.01% of the total serum protein (53). In contrast, growth factors and cytokines that interact with conventional cell surface receptors are present in the circulation at levels that are 3 orders of magnitude lower. Adiponectin comprises a globular head and a collagenous tail that allow the protein to form stable trimers, hexamers, and higher order oligomers (360 -540 kDa). These properties suggest that adiponectin is structurally and functionally similar to collectin proteins that interact with a variety of macromolecules at relatively low affinities. Collectin proteins, such as C1q and lung surfactant proteins, either circulate at high levels or are localized to mucosal surfaces in the lung and gastrointestinal tract where they function to protect tissues from environmental stress (54). Accordingly, collectin-like properties including an ability to opsonize apoptotic cells and bind bacterial lipopolysaccharides have been attributed to adiponectin (55,56). On the other hand, it is also clear that adiponectin can activate intracellular signaling cascades that control cellular phenotype (e.g. Refs. 9 and 12), suggesting that these effects are mediated by specific receptors on the cell surface. However, given its structure and abundance, it is likely that the interactions of adiponectin receptor will be atypical. In particular, it is difficult to imagine a classical, high affinity ligand/receptor interaction for adiponectin because the receptor would be in a constant state of saturation at physiological levels of this adipokine. In view of these considerations, T-cadherin may serve as a membraneassociated signal transducer for adiponectin. Like adiponectin, T-cadherin is abundantly present on the surface of some cell types, and it may serve as a low affinity receptor for adiponectin. Alternatively, T-cadherin may have a co-receptor function and be required for the proper presentation of adiponectin to a classical high affinity receptor, such as Adi-poR1 or AdipoR2. Recently, Denzel et al. (47) identified T-cadherin as critical for the cardiac protective effects of adiponectin in models of cardiac hypertrophy and ischemiareperfusion. Taken together with the current study, this work highlights the critical role for T-cadherin in mediating the protective effects of adiponectin in the cardiovascular system.
Although it is clear that T-cadherin localizes adiponectin to various tissues, it is not well understood how T-cadherin initiates intracellular signaling because it lacks a transmembrane domain and intracellular domain. However, it has been reported that T-cadherin can function independently of adiponectin by promoting LDL-induced intracellular calcium release (57), facilitating release of insulin from pancreatic ␤-cells (58) and inhibiting insulin signaling in endothelial cells via the PI3K/Akt/mammalian target of rapamycin axis (33).
Because T-cadherin is found to co-localize with caveolin, it is possible that interactions between T-cadherin and other proteins in caveolae or lipid rafts may result in intracellular signaling (59). This concept of receptor internalization is consistent with a report of T-cadherin nuclear localization (60). T-cadherin has also been shown to promote endothelial survival via binding interactions with integrin-linked kinase (61) and Grp78 (62). However, these studies did not evaluate T-cadherin signaling downstream of adiponectin or a functional relationship between T-cadherin and AdipoR1 and/or AdipoR2.
Peripheral artery disease is a serious condition of reduced lower limb perfusion that is estimated to affect 15-20% of the elderly population (63). The development of peripheral artery disease is associated with smoking, advanced age, and metabolic syndrome, and it represents a major unmet clinical need (64). Our results suggest that adiponectin has therapeutic benefit for ischemic limb disease and that this effect occurs via an interaction with T-cadherin. Consistent with this hypothesis, it has been reported that circulating levels of high molecular weight adiponectin are inversely associated with development of peripheral artery disease (65). Moreover, genome wide association studies have linked T-cadherin with peripheral artery disease risk factors and co-morbidities (37,44,66). Interestingly, polymorphisms in the T-cadherin gene are also associated with circulating levels of adiponectin in human subjects (34 -39), and these reports are reminiscent of observations made here of elevated serum adiponectin levels in Tcad-KO mice. Collectively, these findings highlight the importance of the T-cadherin/adiponectin interaction in vascular homeostasis and suggest that further studies on this system could lead to a better understanding of ischemic diseases.

FIGURE 5. siRNA knockdown of T-cadherin blocks adiponectin-induced migration and proliferation of endothelial cells in vitro.
A, HUVECs (p3-p6) were transfected with siRNA targeting T-cadherin or a control unrelated sequence for 72 h. Cells were lysed, and mRNA expression of T-cadherin, AdipoR1, and AdipoR2 was assessed by real time PCR. GAPDH mRNA expression was used as a housekeeping gene. T-cadherin protein expression was assessed by Western blot. Tubulin was used to control for equal protein loading. B, following overnight serum starvation in EBM-2 ϩ 0.5% FBS, confluent siRNA-transfected cells were scratched using a 200-l pipette tip in a 6-well plate. Cells were cultured for 16 h in DMEM ϩ 1% FBS Ϯ 10 g/ml full-length rAPN. Images were taken at t ϭ 0 h and t ϭ 16 h to determine the extent of cellular migration into the scratched area. Migration assay results are reported as percent gap closure. Representative images are shown. C, siRNA-transfected HUVECs were transferred onto a 96-well plate in the presence of DMEM ϩ 5% FBS Ϯ 10 g/ml rAPN. Cell proliferation was assessed by a BrdU incorporation ELISA-based assay and reported as -fold change relative to control siRNA-transfected cells cultured in the absence of rAPN. D, for the co-immunoprecipitation study, HUVECs were cultured in serum-free EBM-2 medium for 16 h and then treated with 25 g/ml high molecular weight rAPN for 60 min. Cells were washed with PBS and cross-linked with 2.5mmol/liter dimethyl 3,3Ј-dithiopropionimidate dihydrochloride for 60 min at room temperature. Cell lysates were used for immunoprecipitation with control IgG or APN antibody, and protein complexes were analyzed by Western blotting. ␤-Actin was used to control for equal protein loading. Ig, immunoglobulin; WB, Western blot; IP, immunoprecipitation. HUVECs were transfected with siRNA targeting AdipoR1, AdipoR2, or a control unrelated sequence for 72 h. Cellular migration (E) and BrdU incorporation (F) were assessed as above in the presence or absence of 10 g/ml rAPN. Error bars represent S.E. AU, arbitrary units.