Published online before print February 3, 2005, doi: 10.1161/01.RES.0000158285.57191.60
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2005 American Heart Association, Inc.
UltraRapid Communication |
Angiopoietin-1 Promotes Cardiac and Skeletal Myocyte Survival Through Integrins
From the Division of Vascular Biology (S.M.D., N.I., J.R.O., B.E.H., M.A.R.), Children’s Hospital, Boston; Harvard-MIT Division of Health Sciences and Technology (B.E.H.), Cambridge; the Division of Cardiovascular Medicine (M.A.R.), Brigham and Women’s Hospital, Boston; and the Department of Chemical Engineering (M.A.R.), Massachusetts Institute of Technology, Cambridge, Mass.
Correspondence to Maria Rupnick, MD, PhD, Vascular Biology Division, Children’s Hospital, Research Building, Rm RB11-211, 1 Blackfan Circle, Boston, MA 02115. E-mail maria.rupnick{at}childrens.harvard.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Cardiac myocyte loss, regardless of insult, can trigger compensatory myocardial remodeling leading to heart failure. Identifying mediators of cardiac myocyte survival may advance clinical efforts toward myocardial preservation. Angiopoietin-1 limits ischemia-induced cardiac injury. This benefit is ascribed to angiogenesis because the receptor, tie2, is largely endothelial-specific. We propose that direct, non-tie2 interactions of angiopoietin-1 on cardiac myocytes contribute to this cardioprotection. We found that mouse C2C12 skeletal myocytes lack tie2, yet dose-dependently adhered to angiopoietin-1 and angiopoietin-2 similarly to laminin, fibronectin, vitronectin, and more than to collagen-I, -III, and -IV. Adhesion was divalent cation-mediated (Mn2+, Ca2+, not Mg2+), blocked with EDTA/EGTA, RGD-based peptides, and select integrin subunit antibodies. Similar findings were obtained with human skeletal myocytes (HSMs) and freshly isolated rat neonatal cardiac myocytes (NCMs). Furthermore, angiopoietin-1 conferred significant survival advantage exceeding that of most cell matrices, which was not fully explained by differences in cell adhesion. Angiopoietin-1 promoted survival of serum-starved C2C12, HSM, and NCM (MTT, trypan blue) and prevented taxol-induced apoptosis (caspase-3). Immobilized and soluble angiopoietin-1 phosphorylated AktS473 and MAPKp42/44, (not FAKY397) in C2C12 more than in endothelial cells and more than did angiopoietin-2 or cell matrices. EDTA, RGD-based peptides, and some integrin antibodies blocked these responses. Angiopoietin-1 activated HSM and NCM AktS473 and MAPKp42/44 survival pathways. We propose that this novel function contributes to developmental and cardioprotective actions of angiopoietin-1 presently attributed to vascular effects alone. Angiopoietin-1 may prove therapeutically valuable in cardiac remodeling by supporting myocyte viability and preserving pump function. The full text of this article is available online at http://circres.ahajournals.org.
Key Words: angiopoietin-1 • angiopoietin-2 • cardiac myocytes • adhesion molecules • myocyte apoptosis • skeletal myocytes
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
There is growing consensus that cardiomyocyte (CM) apoptosis contributes to many cardiac diseases (eg, ischemia,1 infarction,2 hypertension,3 myocarditis,4 transplant rejection,5 and heart failure6,7). Research efforts are directed at defining the incidence of CM death, the contributions to cardiac dysfunction, and the consequences of inhibiting apoptosis. Incentive is based on the rationale that CM loss reduces contractile mass of the heart and may be a preventable catalyst of heart failure. In support of this concept, low levels of CM apoptosis (23 CM/105 nuclei) cause lethal cardiomyopathy in mice.8 CM apoptosis rates are higher in cardiomyopathy patients (80 to 250 CM/105 nuclei) compared with healthy hearts (1 to 10 CM/105 nuclei).9,10 Further, ischemic preconditioning upregulates bcl-2, a cytoprotective protein, and is linked to reduced apoptosis.11 Despite amassing experimental and clinical evidence, mechanisms and signaling pathways in CM apoptosis are largely unexplored. Identifying regulatory mediators may lead to novel therapies to preserve myocardial function after injury. Toward this goal, we introduce a novel function for angiopoietin-1 as a direct cardiac myocyte survival factor.
Angiopoietin-1 (Ang1) and angiopoietin-2 (Ang2) are secreted proteins that bind the tie2 receptor, which is largely limited to endothelium.12,13 Angiopoietins are most noted as regulators of vascular maturation. Ang1 associates with matrix and acts locally, whereas Ang2 freely diffuses.14 Ang1 promotes endothelial cell survival15,16 and increases mural cell17 and matrix18 contacts with vessels to establish quiescence and promote maturation. Ang2 often favors vessel destabilization allowing angiogenesis or regression,19 at high concentrations acts as a weak agonist,20 and is an agonist on lymphatic endothelium.21
A role for angiopoietins in cardiovascular health and disease is emerging. Ang1 and tie2 knockouts result in embryonic lethal cardiac defects, including impairments in vascular maturation and in endocardial and trabeculae formations.22 A similar phenotype is produced by transgenic overexpression of Ang2, a competitive Ang1 antagonist.19 Postnatally, as the heart transitions from extensive neonatal remodeling to limited adult remodeling, Ang1 levels increase.23 The endothelial-specific tie2 receptor12,13 is constitutively activated in adult hearts,24 suggesting Ang1 may serve a maintenance function for the vasculature.
The Ang/tie2 system is also implicated in cardiac remodeling under pathological conditions. Expressions of Ang2 increase after myocardial infarction25,26 and hypoxia/reoxygenation injury,27 and plasma Ang2 levels are higher in patients with heart failure.28 In contrast, Ang1 expressions are reduced in hypoxic myocardium.26 Ang1 overexpression in mice, rats, and rabbits with acute myocardial infarcts reduced infarct sizes and preserved ejection fractions.29–31 These benefits were ascribed solely to the role of Ang1 in angiogenesis because the tie2 receptor is endothelial-specific.12,13 However, fibroblasts were recently shown to adhere to Ang1 and Ang2 via integrins (
5ß1, vß5),32 raising the possibility that other nonendothelial cell types, such as CM may directly interact with angiopoietins.33
Integrins are cell surface adhesion receptors composed of and ß subunits, which combine to form at least 24 heterodimers with different, although often overlapping, ligands and signaling properties.34 Integrins are also regulated by dynamic spatial and temporal expression patterns35 and subunit isoforms. For example, integrin adhesion, expression, and activation shift during cardiac development,36 hypertrophy,35 infarct,37 and failure.38 Information traffic via integrins is bidirectional, enabling cells to interact with the extracellular matrix (ECM)/environment. Integrins thereby mediate numerous vital CM activities such as cell shape, adhesion, apoptosis, anoikis, hypertrophy, survival, differentiation, contraction, and conduction.39
In this study, we present the first evidence that cardiac and skeletal myocytes adhere to Ang1 and Ang2 via integrins. We show that Ang1 markedly promotes CM survival under stress, and Ang1 protects CM from apoptosis. This raises the possibility that the cardioprotective benefits of Ang1 overexpression in ischemic hearts are due, in part, to direct interactions between Ang1 and CM, mediated via integrins. If so, angiopoietin regulation may serve as a novel target for preserving myocyte viability after cardiac insults, impeding heart failure development.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Details of the following methods can be found in the Expanded Materials and Methods section of the online data supplement available http://circres.ahajournals.org.
Cell Culture
We chose to use the C2C12 cell line (American Tissue Culture Collection) because these transformed mouse skeletal myoblasts, once differentiated, have features of CM such as expression of cardiac isoforms of contractile proteins and well-organized myofibrils.40 Further, using a cell line eliminates the possibility of contaminating cells found in primary cultures. C2C12 cells were grown in high-glucose (4.5 g/L) DMEM (HG-DMEM) (Gibco) supplemented with 10% heat-inactivated fetal calf serum (FCS) (Hyclone)/0.01 mol/L HEPES (Gibco)/L-glutamine-penicillin G-streptomycin sulfate (GPS) (Gibco) (37°C, 5% CO2). For adhesion, survival, and caspase-3 assays, C2C12 myoblasts were differentiated to myocytes by confluence.41 Cells that were 4 to 7 days postconfluence were used. Microvascular endothelial Ms1 cells were cultured in low glucose (1.0 g/L) DMEM (LG-DMEM) supplemented with 10% FCS/GPS (37°C, 10% CO2). Human skeletal muscle primary myoblasts (Cambrex) were grown in Clonetics SkGM BulletKit Medium plus growth factors (37°C, 5% CO2) and differentiated to myocytes by confluence and growth factor withdrawal as per manufacturers instructions.
Cardiac Myocyte Isolation
Cardiac myocytes were isolated from the ventricles of Sprague-Dawley P2 rat pups (Charles River Laboratories, Wilmington, Mass) following published procedures.42 Briefly, ventricles were minced, digested in 0.6 mg/mL trypsin/EDTA at 4°C overnight, then 1 mg/mL collagenase for 30 minutes 37°C, and preplated. Myocytes were collected, rinsed, and grown in LG-DMEM/2% FCS.
RT-PCR
Cells were immersed in RNALater (Ambion). Total RNA was purified using RNeasy mini kits (Qiagen). cDNA was made using Advantage RT-for-PCR kits (Clontech) with RNA (1 µg). PCR with tie1, tie2, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Invitrogen) primers were performed as we have described.43
Cell Adhesion Assay
Assays were conducted as we44 and others32 have described with some modifications. Human recombinant Ang1 and Ang2 (R&D Systems) were dissolved in 0.1% bovine serum albumin (BSA), fraction V/phosphate-buffered saline (PBS) (Fisher), and then further diluted in PBS at 0 to 400 nmol/L. BSA controls contained comparable amounts of 0.1% BSA/PBS diluted in PBS. Ang1, Ang2, and BSA control solutions were used to coat wells in 96-well flat-bottom-tissue culture plates (room temperature, 1 hour). Wells were also coated with human fibronectin, vitronectin, collagen I, collagen III, and mouse laminin and collagen IV as per the manufacturers’ instructions. Wells were blocked for at least 30 minutes with 0.5% heat-inactivated BSA (10 minutes, 80°C)/PBS, rinsed three times with PBS, and prepared cells were added.
C2C12 myocytes, primary rat neonatal cardiac myocytes (NCMs), and primary human skeletal myocytes (HSMs) were detached with trypsin/EDTA, rinsed in serum-free medium, plated onto immobilized proteins, and incubated for 40 minutes. Wells were rinsed, attached cells fixed, toluidine blue-stained, solubilized, and absorbances measured (650 nm). Values were corrected for background myocyte adhesion to BSA wells. For some assays, adhesion was challenged with EDTA, EGTA, GRGESP (RGE), or GRGDSP (RGD) peptides, or adhesion-blocking integrin subunit antibodies. Assays were also done in which divalent cations were depleted and replaced44,45 (CaCl2, MgCl2, MnCl2, or no divalent cations).
Cell Survival Assays
A trypan blue assay and an MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (Sigma) assay with improved sensitivity were used to measure myocyte viability46 on immobilized angiopoietins compared with the various extracellular matrices. C2C12, HSM, and NCM were prepared and 96-well plates were coated as described (adhesion assay) using 200 nmol/L solutions to coat the plate surfaces.
Caspase-3 Assay
Apoptosis was measured in C2C12 myocytes, NCMs, and HSMs using a Caspase Colorimetric Kit (Promega) as per manufacturer’s instructions. Myocytes were prepared and 96-well plates were coated as described (adhesion assay). Cells were incubated in serum-free medium on untreated wells (plate) or wells coated with 200 nmol/L Ang1, Ang2, matrix component, or BSA for 1 day. Apoptosis was then induced by taxol47 in an overnight incubation, and caspase-3 activity was measured.
Cell Signaling
Phosphorylation of Akt (protein kinase B) serine 473 (pAktS473), mitogen-activated protein kinase (MAPK) p42/44 (ERK 1/ERK 2) threonine 202/tyrosine 204 (pMAPKp42(T202)/44(Y204)), and focal adhesion kinase (FAK) tyrosine 397 (pFAKY397) were measured in myocytes (C2C12, HSM, NCM) incubated either on immobilized or in soluble molecules (200 nmol/L Ang1, Ang2, laminin, fibronectin, vitronectin, collagen I, -III, and -IV, BSA) or on the plate alone (unmodified tissue culture wells), and examined by Western blotting. We conducted two types of soluble studies in which activation of Akt and MAPKp42/44 were measured. In one, we added 200 nmol/L soluble molecules to cells (C2C12, Ms1) in suspension. In the other, we plated cells (C2C12, HSM, Ms1) onto tissue culture plates, serum-starved cells, added wortmannin to some wells, then added (3.6 or 200 nmol/L) soluble molecules, and made protein lysates.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Skeletal and Cardiac Myocytes Adhere to Ang1 and Ang2
C2C12 mouse skeletal myocytes (Figure 1A), human skeletal myocytes (HSMs, Figure 1B), and rat neonatal cardiac myocytes (NCMs, Figure 1C) all adhered to immobilized Ang1 and Ang2 in a concentration-dependent manner. In contrast, undifferentiated C2C12 myoblasts did not attach to either angiopoietin (data not shown). The number of myocytes adhering to Ang1 versus Ang2 was similar among the cell types.
|
We compared myocyte adhesion to angiopoietins to six ECM/basement membrane (BM) components prominent in the heart (Figure 1). C2C12 myocyte adhesion curves for Ang1, Ang2, laminin, fibronectin, and vitronectin were similar, whereas little adhesion occurred on collagen-I, -III, or -IV (Figure 1A). In contrast, HSMs adhered to all the matrix components and at lower surface coating concentrations than to Ang1 or Ang2 (Figure 1B). However, when the plates were coated with 200 nmol/L of each molecule, the number of adherent HSMs plateaued and was similar on all the surfaces. At that concentration, NCM also adhered to Ang1 and Ang2 similarly to most other matrix molecules, with the exception of superior adhesion to collagen–IV (Figure 1C). The concentration for maximum adhesion of myocytes to Ang1 and Ang2 is equal to or less than that for fibroblasts to Ang1/Ang232 and for endothelial cell (EC) to angiopoietin-like protein-3 (Angptl3),48 which does not bind tie2.
Based on these findings, 200 nmol/L coating concentrations were used for the matrices in subsequent assays. This enabled the functional effects of each matrix to be assessed at comparable cell densities. The exceptions were poor adhesion of C2C12 to the collagens and greater adhesions of NCM to collagen-I and -IV.
Our data shows that mouse, rat, and human myocytes (primary cells, cell lines) interact with human Ang1 and Ang2. This likely reflects the highly conserved amino acid sequence. Ang1 homology for human versus mouse is 97% and for human versus rat is 96%. Human Ang2 is 85% identical to mouse and 86% identical to rat sequence.
Myocyte Adhesion to Ang1/Ang2 Does Not Involve the Tie2 Receptor and Is Inhibited by RGD and EDTA
We conducted RT-PCR of tie2, tie1 (related receptor that does not bind angiopoietins), and GAPDH (housekeeping gene) using Ms1 (mouse microvascular endothelial cells), HMVECs (human microvascular endothelial cells), and rat left ventricle (LV) tissue as positive controls (tie2, tie1). C2C12 myocytes, HSM, and NCM did not express mRNA for tie2 or tie1 (Figure 2A). To determine whether tie2 expression was conditional, we conducted RT-PCR and real-time PCR for tie2 and GAPDH with C2C12 myoblasts and myocytes that were cultured in full and serum-free medium. Tie2 mRNA was not detectable under any condition used in our assays, whereas GAPDH mRNAs were abundant (data not shown). Thus, angiopoietin/myocyte interactions do not involve tie2.
|
We measured Ang1 and Ang2 mRNA expressions in C2C12 myocytes, NCM, and HSM using mouse, rat, and human heart as positive controls. Skeletal and cardiac myocytes express Ang1 mRNAs (Figure 2A). Only C2C12 myocytes express Ang2 transcripts.
We propose that integrins mediate myocyte interactions with Ang1 and Ang2. Integrin adhesions require divalent cations with 3 to 5 divalent cation binding sites per integrin heterodimer.49 Integrins also require an exposed aspartic (D) or glutamic acid (E) in the ligand.50 There are different D-based (LDV, RTD, REDV, KRLDGS) and E-based (LRE)49 motifs, but RGD-based motifs are most common. Requiring divalent cations and antagonism by RGD peptides would support a role for integrins in mediating myocyte-angiopoietin interactions.
C2C12 myocyte adhesion to Ang1 and Ang2 were significantly blocked by RGD-based peptides (74.1±7.5%, 89.2±10.1%), EDTA (96.0±1.5%, 95.8±3.7%), and EGTA (95.5±5.8%, 91.8±3.0%) (Figure 2B). EDTA nonspecifically chelates divalent cations, whereas EGTA has a higher affinity for Ca2+ than Mg2+. To define the divalent cations mediating C2C12 myocyte adhesion to Ang1 and Ang2, we examined the effects of Ca2+, Mg2+, Mn2+, and no divalent cations.44,45 Mn2+ and Ca2+ supported adhesion to Ang1 and Ang2, but Mg2+ did not (Figure 2C). Ca2+ occupies divalent cation sites in many integrins.51 Mn2+ activates integrins, increasing the kinetic "on" rate.45,52 A determinant of integrin high-affinity binding (firm adhesion) to a ligand is activation from a low to high affinity state, involving a conformational change.50 Overall, these findings suggest that myocyte adhesion to Ang1 and Ang2 is integrin-mediated.
Integrins Mediate Skeletal and Cardiac Myocyte Adhesion to Ang1 and Ang2
To assess whether integrins mediate the attachment of C2C12 myocytes, NCM, and HSM to Ang1 and Ang2, we conducted adhesion assays using adhesion-blocking integrin antibodies. C2C12 myocyte adhesion to Ang1 was inhibited by 6 (93.5%±5.6%), ß3 (53.9%±13.6%), ß1 (44.1%±11.5%), 1 (39.9%±4.1%), and v (35.1%±2.7%), but not 4 or 5 antibodies (Figure 3A). Adhesion to Ang2 was blocked by ß3 (71.1%±2.5%), ß1 (66.7%±1.8%), 6 (61.3%±1.7%), 1 (39.3%±8.8%), and 5 (36.9%±5.9%), but not 4 or v antibodies (Figure 3A). Integrin subunits 6 and ß3 appear to be key mediators of C2C12 myocyte adhesion. Subunit 6 can pair with ß1 or ß4. Currently, no anti-mouse adhesion-blocking ß4 integrin subunit antibodies are commercially available, so we were unable to further analyze 6 involvement using this method. Integrin subunit ß3 only complexes with v, except for platelets where ß3 complexes with IIb. In line with our results that ß3 mediates adhesion, vß3 mediates HMVEC adhesion to Angptl3.48 NCM adhesion to Ang1 was inhibited by ß1 (81.3%±3.8%) and ß3 (67.1%±14.4%), but not 1 antibody (Figure 3B). NCM adhesion to Ang2 was also blocked by ß1 (67.5%±7.7%) and ß3 (91.9%±6.9%), but not 1 antibody (Figure 3B). Again, further analysis was limited by a lack of commercially available anti-rat integrin antibodies.
|
To better define the integrins involved, we examined HSMs, which enabled use of human antibodies (1 to 6, v, ß3 to ß4, vß3, vß5, vß6). HSM adhesion to both Ang1 and Ang2 was markedly blocked by antibodies (2, 5, v) followed by (ß3, ß4) (Figure 3C). Anti-6 significantly inhibited adhesion to Ang1, but not Ang2. EDTA considerably blunted HSM adhesion to Ang1 and Ang2, showing the need for divalent cations (Figure 3C).
In mouse (C2C12), rat (NCM), and human (HSM) myocyte studies, ß3 antibodies, consistently blocked adhesion, whereas antibody LM609 (vß3) did not. This suggests that although vß3 may be a receptor for Ang1/Ang2 on myocytes, the interaction may be somewhat different than that seen in other systems. For example, LM609 inhibited the interaction between Angptl3 and HMVECs.48
HSM studies also implicated 2ß1, 5ß1, 6(ß4 and/or ß1), and perhaps, vß1, as integrins that mediate myocyte interactions with Ang1 and Ang2. Fibroblasts adhere to Ang1 and Ang2 mainly via 5ß1 and vß5 integrins.32 The only prevalent CM integrin not tested was 7 because of a lack of a commercially available antibody.
Ang1 Promotes Skeletal and Cardiac Myocyte Survival
To assess the function of integrin-mediated myocyte adhesion to Ang1 and Ang2, we conducted trypan blue and MTT-based viability/survival assays. Myocytes were incubated in serum-free medium on the various matrices (200 nmol/L) for 1 day. Overall, Ang1 promoted C2C12 myocyte, HSM, and NCM viability (trypan blue) better than or similar to the other matrices tested (Table).
|
In MTT survival assays, C2C12 myocytes were incubated in serum-free medium on wells coated with 200 nmol/L Ang1, Ang2, BSA, fibronectin, or laminin, or plate (unmodified tissue culture well). Photomicrographs (day 1 and 4 shown) (Figure 4A) show that myocytes on BSA did not spread, but rather remained rounded, aggregated, and died. In contrast, myocytes on Ang1, Ang2, and plate spread, maintaining a normal morphology the first 2 days. Thereafter, myocytes on Ang1 approached confluence, whereas those on Ang2 and plate began elongating, forming aggregates, and dying. Ang1 promoted C2C12 myocyte survival better than all other conditions (Figure 4B and 4C).
|
To assess if Ang1 enables cardiac myocyte survival, we conducted MTT-based viability assays using freshly isolated rat NCM. Similar to C2C12 myocytes, photomicrographs (day 4 shown) show that NCM on BSA were largely rounded, yet those on Ang1, Ang2, and plate spread and maintained a normal morphology (Figure 4D). However, only NCM on Ang1 had increased survival compared with the other conditions (Figure 4E).
Adhesion to Ang1 Prevents Skeletal and Cardiac Myocyte Apoptosis
We determined the effects of Ang1, Ang2, and some matrix components on myocyte apoptosis by measuring caspase-3 activity after taxol treatment in serum-free conditions on the various matrices. Adhesion to Ang1 markedly prevented C2C12 myocyte (Figure 5A), HSM (Figure 5B), and NCM (Figure 5C) apoptosis. For C2C12 myocytes, Ang1 prevented caspase-3 activation better than Ang2, laminin, collagen-I and -III, BSA, and plate (*P
0.01), similar to collagen IV and vitronectin, and slightly less than fibronectin (P=0.0006). In HSMs, Ang1 prevented caspase-3 activation considerably better than Ang2, collagen IV, fibronectin, laminin, vitronectin, BSA, and plate (*P0.001), but was not as protective as collagens-I or -III (P0.01). Ang1 prevented NCM caspase-3 activation more than Ang2, fibronectin, laminin, vitronectin, BSA, and plate (P0.02), similar to collagen IV, and less than collagens-I and -III (P0.01). We evaluated the percentage of caspase-3 activation suppressed by Ang1 compared with plate (C2C12, 50.5%; HSM, 8.0%; NCM, 41.5%).
|
This data demonstrates that Ang1 prevented myocyte (mouse, rat, human) apoptosis under adverse conditions as well as or better than most other matrices tested. This cytoprotection is not explained by differences in cell adhesion, because at 200 nmol/L, there are similar numbers of adherent C2C12 (Figure 1A), HSMs (Figure 1B), and NCMs (Figure 1C) to Ang1, Ang2, and the other matrices, except for less C2C12 myocytes on the collagens and more NCMs on collagens-I and -IV. These data suggest a novel, direct role for Ang1 in protecting myocytes from injury.
Ang1 Activates Akt and MAPKp42/44 Signals in Skeletal and Cardiac Myocytes
We used Western blotting to determine whether myocyte interactions with Ang1 and Ang2 activate Akt, MAPKp42/44, and FAK signals, because these pathways increase survival and prevent apoptosis of CM. Comparisons were made among myocytes plated on the various immobilized matrices or suspended (Sus) in serum free media. Ang1 significantly induced C2C12 myocyte phosphorylation of Aktser473 (Figure 6A) and MAPKp42/44 (Figure 6B) more than any other matrix (*P0.0001 Ang1 versus other matrix), whereas FAK activation was not affected (Figure 6C). Ang1 also activated Aktser473 (Figure 7A) and MAPKp42/44 (Figure 7B) in HSMs more than all other conditions (*P0.001). In NCMs as well, Ang1 promoted phosphorylation of Aktser473 (Figure 8A) and MAPKp42/44 (Figure 8B) more than all other ECM (*P0.05), but did not alter FAK activation (Figure 8C). The ability of Ang1 to increase Akt and MAPKp42/44 phosphorylation in myocytes suggests a novel mechanism that may have therapeutic value in preserving CM viability and function after injury.
|
|
|
Ang1-Mediated Myocyte Activation of Akt and MAPK Occurs via Integrins
To assess whether Ang1 activation of myocyte Akt and MAPKp42/44 occurs via integrins, we preincubated C2C12 myocytes with EDTA, GRGDSP or GRGESP peptides, integrin antibodies (4, 6, ß3), or PBS, and then plated them on 200 nmol/L immobilized Ang1 or vitronectin. We chose vitronectin for comparison because it induced the second highest Akt phosphorylation after Ang1 (Figure 6A). Furthermore, cell interactions with vitronectin via integrins can be disrupted with RGD-based peptides and EDTA.32 Again, suspended myocytes were used as a negative control. EDTA significantly reduced myocyte Akt (Figure 9A and 9B) and MAPKp42/44 (Figure 9C and 9D) activation when incubated on Ang1 (Figure 9A, Akt; Figure 9C, MAPK) or vitronectin (Figure 9B, Akt; Figure 9D, MAPK) (*P0.002). GRGDSP peptide was consistently more effective at blocking Ang1- (Figure 9A, Akt; Figure 9C, MAPK) and vitronectin- (Figure 9B, Akt; Figure 9D, MAPK) induced cell signaling, than GRGESP peptide (#P0.05). GRGESP peptide reduced myocyte Akt and MAPKp42/44 activation, which may reflect nonspecific interference with integrin adhesion.53
|
Ang1-induced Akt phosphorylation was significantly blocked by anti-ß3 (*P=0.001), with a trend for anti-6 (P=0.06), and was not effected by anti-4 (Figure 9E). Ang1 activation of MAPKp42/44 was blocked by anti-ß3 more than anti-6 (*P0.001), but not anti-4 (Figure 9F). Thus, antibodies to anti-4 did not affect myocyte adhesion (Figure 3A) or signaling, but antibodies to ß3 and 6 reduced both myocyte adhesion (Figure 3A) and signaling. This data demonstrates that Ang1-induced activation of myocyte Akt and MAPKp42/44 is integrin-mediated.
Soluble Ang1 Activates Myocyte Akt and MAPKp42/44 Signaling
To determine whether Ang1-induced phosphorylation of Akt and MAPKp42/44 required cell adhesion, we placed C2C12 myocytes and Ms1 microvascular endothelial cells in suspension in serum-free medium and added 200 nmol/L Ang1, Ang2, fibronectin, laminin, vitronectin, or PBS. Western blotting showed that Ang1 markedly phosphorylated Akt (Figure 10A) and MAPKp42/44 (Figure 10B) in both cell types considerably more than did the other soluble conditions (*P0.001 Ang1 versus other conditions). Activation was greater in myocytes (65.7-fold Akt; 3.5 fold MAPK) than endothelial cells (3.4-fold Akt; 2.1-fold MAPK). Thus, Ang1 potently activated cytoprotective signaling pathways in myocytes in a non-tie2, cell adhesion–independent manner.
|
Ang1 has been shown in several reports to activate endothelial cell Akt15,54,55 and MAPKp42/44.56,57 In all of these endothelial cell studies, similar assays were conducted. Endothelial cells were grown in tissue culture dishes, serum-starved, then soluble Ang1 was added and Akt and/or MAPKp42/44 phosphorylations were assessed. To further compare the effects of soluble Ang1 on myocytes versus endothelial cells, we also conducted this type of assay. We found that Ang1 phosphorylated Akt (Figure 11) and MAPKp42/44 (Figure 12) on C2C12 myocytes, HSMs, and Ms1 endothelial cells. Wortmannin, a phosphatidylinositol 3'-kinase (PI3-K) inhibitor, effectively blocked Akt (Figure 11) and MAPKp42/44 (Figure 12) activation in both myocytes and endothelial cells. Studies have also shown that wortmannin blocks MAPKp42/44 phosphorylation in myoblasts.58 Thus, Ang1-induced myocyte Akt and MAPKp42/44 activation appear to be PI3-K–mediated. We found that Ms1 endothelial cell, but not myocyte, Akt (Figure 11) and MAPKp42/44 (Figure 12) were phosphorylated by 3.6 nmol/L Ang1, a commonly used dose in endothelial cell studies.15,54–57 Ang1 (200 nmol/L) increased phosphorylation of Akt HSM (4.2-fold), C2C12 myocytes (2.4-fold), and Ms1 endothelial cells (2.1-fold). For MAPKp42/44, Ang1 (200 nmol/L) increased HSMs (3.5-fold), C2C12 myocytes (2.3-fold), and Ms1 (3.8-fold).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In this study, we show that skeletal and cardiac myocytes adhere to Ang1 and Ang2 via integrins and that Ang1 promotes survival under adverse conditions. This finding revises the current view that angiopoietins act essentially on the vasculature via the endothelial cell (EC)–specific receptor, tie2. With the recognition that angiopoietins act on CMs, ECs, and fibroblasts,32 an expanded role for these molecules in cardiac health and disease may emerge. Angiopoietins appear well-positioned to mediate myocardial remodeling, which requires coordination among these various cell types for proper cardiac function. Interest in mechanisms that regulate balanced remodeling is increasing with the appreciation that disproportionate changes in these cellular components occur in pathological cardiac remodeling and may ultimately contribute to heart failure. Furthermore, our findings indicate that this novel angiopoietin role extends beyond the heart. Cytoprotective effects on skeletal myocytes (HSM, C2C12 myocytes) suggest that Ang1 may also directly mediate skeletal muscle survival59,60 during insults such as limb ischemia from peripheral vascular disease. We propose that this new angiopoietin function acts in concert with effects on the vasculature to protect the heart and peripheral tissues under adverse conditions.
We found that RGD-based peptides and EDTA blocked myocyte adhesion to Ang1 and Ang2 (Figures 2 and 3
). Fibroblast32 adhesion to Ang1 and Ang2, and EC48 adhesion to Angptl3 are also inhibited by RGD-based peptides and EDTA. These features support integrin involvement. However, Ang1, Ang2, and Angptl3 lack an RGD site. We examined mouse, rat, and human Ang1 and Ang2, and found no exact matches to 22 well-known integrin-binding motifs.53 However, there were areas resembling integrin motifs in the fibrinogen-like (fbg-lk) domain of Ang1 and Ang2 that were conserved among mouse, rat, human, and other species (eg, QHREDGS). This region resembles KRLDGS (fibrinogen) or REDV (fibronectin) integrin motifs. A QHREDGS or similar amino acid (AA) sequence exists in human Ang1 and 2, mouse Ang1 to 3 and 5, pig Ang1 and 2, chicken Ang2 isoforms A and B, zebrafish Ang1 to 3, Caenorhabditis elegans angiopoietin (Ang)-related protein, human Angptl1 to 6, mouse Angptl3, and yellow fever mosquito Ang-related protein. Mouse Angptl4 and African clawed frog Ang-related protein have a nearby RGD site. An exception is human Ang4, which lacks the D, and has a diverging function from its counterpart.61 We propose that select myocyte integrins adhere to Ang1 and Ang2 via the fbg-lk domain, possibly at or including the AA’s QHREDGS. In support of this concept, ECs adhere to via vß3 via the Angptl3 fbg-lk domain,48 and fbg-lk domains of Angptl3, Ang1, and Ang2 are very homologous.19,48,62
Our studies implicate integrins (eg, 6ß4, 6ß1, vß3, vß1, 5ß1, 2ß1) for mediating myocyte adhesion to Ang1 and/or Ang2. Our findings together with those of Carlson et al (fibroblasts)32 suggest that each cell type has a characteristic set of integrins that engages the angiopoietins, which may partially overlap among cell types. Our most complete analysis was conducted on skeletal myocytes (HSM) owing to the greater availability of anti-human integrin antibodies. There are fewer available anti-rat antibodies, so additional integrins may yet be identified that mediate cardiac myocyte adhesion to Ang1 and Ang2. Furthermore, the expression of many integrin subunits shift during myocyte differentiation.63 This may account for our finding that C2C12 differentiation into myocytes is vital for attachment to Ang1 and Ang2.
Promising integrin subunit candidates include ß3 and 6. ß3 antibodies blocked mouse, rat, and human myocyte adhesion to Ang1 and Ang2 (Figure 3A through 3C) and prevented Ang1 induction of Akt and MAPKp42/44 phosphorylations (Figure 9E and 9F). ß3 is important in CM focal adhesion complexes and is active in cardiac hypertrophy.64 Also, 6 inhibited myocyte attachment to Ang1 and Ang2 (Figure 3A and 3C) and abrogated MAPKp42/44 activation (Figure 9F). 6 is vital to CM function, prevalent on CM, and is not expressed by cardiac fibroblasts.65 Isoforms (6A and 6B) are expressed in fetal and adult hearts,36 and levels shift during cardiac development.66,67 The blocking effects of ß3 and 6 antibodies support a link between myocyte-Ang1 interactions and activation of cell signaling (Figures 6 through 12 ).
We examined the impact of Ang1 on signaling pathways critical to CM survival and protection from apoptosis. Akt phosphorylation is necessary and sufficient for CM cytoprotection. Constitutively active Akt reduced infarct size (64%), CM apoptosis (84%),68 and heart failure69 in vivo. In vitro Akt phosphorylation prevented DNA fragmentation,70 caspase activation, cytochrome c release,71 and CM apoptosis.70 Phosphorylated MAPKp42/44 activated CM prosurvival signals72 and reduced reperfusion injury,73 ß-adrenergic stimulation38-induced CM apoptosis, and heart failure.74 Thus, Ang1 activation of human, mouse, and rat myocyte Akt and MAPKp42/44 signals may increase myocyte survival and prevent apoptosis.
A direct role for Ang1 in preventing caspase-3 activation in CM may ultimately advance efforts toward preserving cardiac function. In a mouse model of ischemia/reperfusion injury, cardiospecific caspase-3 overexpression increased CM apoptosis, poly-ADP ribose polymerase (DNA repair enzyme) degradation, DNA fragmentation, and infarct size.75 Alternatively, inhibiting caspase-3 prevented CM apoptosis76 and reduced ischemia injury,77,78 infarct size,79 postinfarct remodeling,80 and heart failure.81 Activated caspase-3 can degrade CM contractile proteins (myosin light chain-1,82 sarcomeres,83 and myofibrils75) without inciting cell death. This contributes to LV remodeling, dilation,84 and reduced LV function in ischemia,84 reperfusion injury,75 infarct, and heart failure.82
Our data show that Ang1 and Ang2 are not EC-specific and can use non-tie2 receptors. Ang1 and Ang2 can act directly on cardiac and skeletal myocytes through integrins and promote myocyte survival (Table, Figure 4) through specific pathways such as caspase-3 inhibition (Figure 5) and Akt and MAPKp42/44 phosphorylation (Figures 6 through 12). These actions are similar to those reported for Ang1 as an EC survival factor,15,16,54,56,85 namely inhibition of caspase,54,55 and activation of Akt,15,54,56 MAPKp42/44,55 PI-3K, survivin, and eNOS.86,87 Whether there is an integrin-mediated component to these EC effects in addition to the tie2 receptor interactions has not been explored. Precedents suggest that this may be possible. Hepatic fibrinogen/Ang-related protein, which does not bind tie2, prevents HUVEC apoptosis in serum-free media.88 Angptl3, which also does not bind tie2, adheres to HMVECs via integrin vß3, activating Akt, MAPKp42/44, and some FAK.48 HUVEC adhesion to Ang1 and Ang2 activates MAPKp42/44 and FAK,32 and EDTA- and RGD-based peptides nullify these signals, implicating integrins as mediators of angiopoietin signals on ECs. Thus, in the heart, Ang1 may act on CMs and cardiac ECs to promote survival and maintain cardiovascular health.
Ang1 and Ang2 promoted adhesion of all the myocytes studied (Figure 1). However, only Ang1 increased myocyte survival (Figure 4, Table) and regulated cytoprotective signaling (Figures 5 through 12
Soluble Ang1 added to myocytes and endothelial cells grown in tissue culture plates activated Akt (Figure 11) and MAPKp42/44 (Figure 12). Our study is the first to use human recombinant Ang1 to measure myocyte and endothelial cell signaling. Previous studies15,54–57,89 used a variant form of Ang1, termed Ang1*.90 Ang1* contains the first 73 amino acids of Ang2 fused to the 77th amino acid of Ang1 with a cysteine to serine mutation at amino acid 265.91 Ang1* was engineered because it is easier to produce and purify than Ang1. However, sequences within the first 76 amino acids of Ang1 are critical for multimerization and tie2 activation,92 and thus may alter the activities of Ang1.
Our studies show similar effects of human Ang1 on human, mouse, and rat myocytes. Several in vivo studies demonstrate that human Ang1 is effective in other animals. In myocardial infarction models, transfection of human Ang1 plasmids30 or adenoviral vectors29 reduced infarct zones in mice and rats, respectively. Plasmid-driven overexpression of human Ang1 reduced hindlimb ischemia in rabbits93,94 and gastric ulcers in rats.95
Mechanisms-of-action and functional significance of angiopoietin–myocyte interactions in vivo remain to be determined. Various ischemia models have shown a protective benefit from overexpressing Ang1. Ang1 overexpression in the heart resulted in smaller infarct zones and preserved cardiac function following myocardial infarctions.29–31 Improved myoblast survival and reduced tissue necrosis with hindlimb ischemia96 and enhanced survival of muscle59,60 and skin97 flaps have also been reported. These studies showed increases in perfusion and/or capillary density, which were thought to explain the protective effects of Ang1 on the muscle and other tissues. However, our findings suggest that nonvascular mechanisms may also contribute to the superior outcomes. We show that Ang1 directly promotes survival of cardiac and skeletal myocytes (Figure 4, Table), prevents caspase-3 activation (Figure 5), and activates cell survival pathways (Figures 6 through 12
In summary, we demonstrated that angiopoietins directly interact with myocytes via integrins to mediate cell adhesion and survival. It is worth noting that Ang1 activated cytoprotective signaling cascades far more potently than any of the other integrin ligands tested, whether in an immobilized (Figures 6 through 9
These findings support the concept that Ang1-myocyte interactions are biologically relevant. We propose that this new angiopoietin function acts in concert with effects on the vasculature to protect the heart. Furthermore, understanding the role of Ang1 in CM survival may lead to novel therapies to stabilize and maintain CM number and function after cardiac insults, thereby impeding heart failure.
). The observations that both immobilized (Figures 6 through 9) and soluble (Figures 10 through 12) Ang1, but not Ang2, activated myocyte Akt and MAPKp42/44 signaling within 30 minutes, suggests that this distinction is not due to differences in coating affinities or protein degradation. Furthermore, Ang1-induced signal activation remained intact in suspended myocytes (Figure 10), eliminating differences in cell adhesion as an explanation. Cell adhesion was not required for this response. The different survival effects of the angiopoietins may relate to the region in the fbg-lk domain of each molecule that we propose mediates adhesion. These regions share 77.8% homology, which is greater overlap than that found in the full-length proteins 62% homology. We are exploring the possibility that this region of Ang2 may resemble Ang1 enough to have a similar binding site conformation, but vary in key residues needed to induce signaling. On endothelium, the mechanism-of-action of Ang2 is complex. Ang2 is a tie2 antagonist on vascular endothelium,19 and at high concentrations is a weak agonist.20 On lymphatic endothelium, Ang2 acts as a tie2 agonist.21 Similar complexities likely characterize the actions of Ang2 on myocytes via integrins. ). This suggests that Ang1 directly supports myocyte survival in vivo under adverse conditions (eg, ischemia). ) or soluble (Figures 10 through 12) form. In fact, when cells were held in suspension and incubated with soluble Ang1, Ang1 was more efficacious at inducing anti-apoptotic pathways in myocytes than in endothelial cells; for which it is a well-recognized survival factor.
Acknowledgments
This work was supported in part by NIH grants K02-HL071840-01 and R21-CA107976-01 (to M.A.R.), and NIH grant K01-DK063970-01A2 and Harvard Medical School Scholars in Medicine Award (to S.M.D.). We thank Thomas Michel MD, PhD, Cardiovascular Division, Brigham and Women’s Hospital, and Douglas B. Sawyer, MD, PhD and David Pimintel, MD, Whitaker Cardiovascular Institute, Boston University School of Medicine for providing rat neonatal cardiac myocytes used in preliminary studies for this work.
Footnotes
Original received May 18, 2004; resubmission received December 17, 2004; revised resubmission received January 24, 2005; accepted January 25, 2005.
References
Top
Abstract
Introduction
Materials and Methods
Results
Discussion
References