Heart and Circulatory Physiology

Monocyte chemotactic protein-1 directly induces human vascular smooth muscle proliferation

Craig H. Selzman, Stephanie A. Miller, Michael A. Zimmerman, Fabia Gamboni-Robertson, Alden H. Harken, Anirban Banerjee


Although monocyte chemotactic protein-1 (MCP-1) is best known for its ability to recruit mononuclear cells, few studies have examined the effects of this chemokine on other events in the vascular response to injury. The purpose of the present study was to determine the influence of MCP-1 on human vascular smooth muscle (VSMC) proliferation. MCP-1 induced concentration-dependent VSMC proliferation as measured by bromodeoxyuridine (BrdU) uptake. Direct cell counting demonstrated a twofold increase in VSMC after stimulation with MCP-1. This mitogenic effect was similar to that observed with the prototypical atherogenic cytokine platelet-derived growth factor. Immunohistochemistry and Western blot analysis revealed that MCP-1 increased both proliferating nuclear cell antigen and cyclin A expression. Whereas MCP-1 did not promote nuclear factor-κB activation, MCP-1-induced VSMC proliferation appeared to be dependent on phosphotidylinositol 3-kinase activation. In conclusion, MCP-1 directly induces VSMC growth, which is associated with activation of cell cycle proteins and intracellular proliferative signals. Within the inflammatory paradigm of vascular remodeling, these data suggest that MCP-1 is more than simply a chemokine but also a potent mitogen for VSMC proliferation.

  • chemokine
  • platelet-derived growth factor
  • proliferating cellular nuclear antigen
  • nuclear factor-κB
  • phosphotidylinositol 3-kinase

the inflammatory response to vascular injury has evolved into a unifying theme in the pathogenesis of intimal hyperplasia (19, 22). Both direct (angioplasty) and indirect (nicotine) insults promote endothelial cell damage and dysfunction. The wounded lining of the vessel thus serves as a nidus for adherence of circulating platelets, monocytes, and lymphocytes with subsequent release of multiple cytokines, chemokines, and growth factors. Vascular smooth muscle cells (VSMC) respond to these mitogens by proliferating, migrating into the intima, and secreting matrix products (12).

Monocyte chemotactic protein-1 (MCP-1) is one member of a large family of endogenous chemokines that recruits circulating monocytes to areas of vessel injury (17, 18). On adherence, monocytes transmigrate into the vessel wall where they phenotypically transform into macrophages. In addition to acting as reservoirs for other cytokines and growth factors, macrophages ingest cholesterol and oxidize lipids to assist in the development and instability of atheromatous plaques. In several vascular experimental models, antagonism of MCP-1 or its receptor, CCR2, appears to inhibit lesion development (2, 5, 7, 8). Cumulatively, these investigators target the influence of MCP-1 on mononuclear cell recruitment as the effector of vascular protection.

Whereas its chemotactic properties are well recognized, we sought to investigate the influence of MCP-1 on another aspect of vascular injury, specifically VSMC proliferation. Indeed, VSMC not only secrete MCP-1 (29) but also express the CCR2 receptor (9). To our knowledge, three studies examined the effect of MCP-1 on VSMC proliferation in vitro. These studies utilized models of rat VSMC and offer conflicting conclusions: MCP-1 either is a mitogen (14), has no effect (30), or inhibits VSMC proliferation (10). The purpose of the present study was to determine the direct influence of MCP-1 on human VSMC proliferation.


Cells isolation and culture.

Human VSMC were isolated from the thoracic aorta and iliac vessels of transplant donors as previously described (21). Procurement of human VSMC was approved by the Colorado Multiple Institutional Review Board (no. 96-161). Purity of isolation was demonstrated by the typical “hill and valley” morphology revealed by phase-contrast microscopy with uniform phallodin staining for F-actin and α-smooth muscle actin (Sigma Immunochemicals; St. Louis, MO). Routine staining for von Willebrand factor and CD14 (Sigma) consistently demonstrated the lack of endothelial cell or macrophage contamination, respectively. VSMC were grown to confluence in a tissue culture flask in a 37°C, 5% CO2 incubator with “complete media.” Complete media consisted of 5% fetal bovine serum (Summit Biotechnology; Ft. Collins, CO), 5% human cord serum (graciously provided by Dr. Lawrence D. Horwitz, University of Colorado, Denver, CO), Dulbecco's modified Eagle's medium (Sigma), and an antibiotic-antimycotic solution containing 10,000 U/ml penicillin G, 10,000 mg/ml streptomycin sulfate, and 25 mg/ml amphotericin (GIBCO-BRL; Grand Island, NY). “Control media” contained all of the listed ingredients except for a reduction in the serum component to only 5% fetal bovine serum. “Serum-free media” actually contained 0.5% fetal bovine serum. Recombinant human MCP-1 and recombinant human platelet-derived growth factor (PDGF) (R&D Systems; Minneapolis, MN) were reconstituted in control media. All studies were conducted by using cells from three separate donors from passages 2–4.

Bromodeoxyuridine proliferation assay.

VSMC were trypsinized (0.05% trypsin-EDTA, GIBCO-BRL) and plated at a density of 3,000 cells/well on 96-well gelatin-coated plates with complete media. After 24 h, the media was removed and replaced with serum-free media for 48 h to achieve synchronous growth arrest. Forty-eight hours after the experimental stimulation, VSMC bromodeoxyuridine (BrdU) uptake was determined by using the Cell Proliferation ELISA, BrdU assay (Boehinger Mannheim; Mannheim, Germany). This assay is a nonradioactive alternative to tritiated thymidine, which has been previously demonstrated to be a marker for VSMC proliferation (11). BrdU incorporation into DNA is measured by photometric analysis using a microplate reader (Bio-Rad; Hercules, CA) and is quantified by optical density (λ). Results represent experiments done in three separate donors performed in triplicate.

Cell counting.

VSMC were plated at a density of 20,000 cells/well on 24-well plates with complete media for 24 h. Media were substituted with serum-free media for 48 h to achieve synchronous growth arrest. Forty-eight hours after stimulation, VSMC were washed twice with PBS (Sigma) and incubated with 200 μl of 0.05% trypsin for 5 min at 37°C. After the trypsin was deactivated with 50 μl of fetal bovine serum, cells were aspirated into tubes and centrifuged at 1,100 rpm for 5 min. The supernatant was decanted, and cells were resuspended in 1 ml of PBS. VSMC were then counted by using a Coulter model ZM analyzer (Coulter; Hialeah, FL). Each experiment was done in duplicate on three separate occasions.


VSMC were plated with complete media on Lab-Tek Chamber Slides (Nalge Nunc International; Naperville, IL). Twenty four hours later, serum-free medium was substituted to allow growth arrest. VSMC were incubated with experimental agents for 24 h. Slides were fixed with a 70% methanol-30% acetone solution (Fisher Scientific; Pittsburgh, PA) for 10 min, air dried, and blocked with 10% goat serum for 60 min at room temperature. Subsequently, cells were incubated at 4°C overnight with mouse monoclonal anti-proliferating cell nuclear antigen (PCNA, Oncogene; Cambridge MA), 1:40 dilution with PBS-1% BSA. After three washes with PBS, cells were incubated in Texas-Red WGA (1:250 dilution) and Alexa-488 goat anti-mouse IgG (1:250 dilution, Molecular Probes; Eugene, OR) for 45 min in the dark at room temperature. After washing was completed, nuclei were stained with bis-benzimide (2.5 μg/ml). Fluorescent images were observed and photographed with a Leica DMRXA confocal microscope. Images were analyzed with the use of Slidebook software.

Western immunoblots.

Human VSMC were plated at a density of 20,000 cells/well on a six-well plate in complete media for 24 h. After growth arrest, VSMC were stimulated with experimental agents for 24 h. Whole cell lysates were obtained in equal volumes of sample buffer (100 nM Tris · HCl, 2% SDS, 0.02% bromophenol blue, and 10% glycerol). After boiling, each sample was electrophoresed on 4–20% Tris · HCl Ready Gel (Bio-Rad). Proteins were then eletrophoretically transferred onto a nitrocellulose membrane and cross linked with UV Stratagene 1800 (Stratagene; La Jolla, CA). Protein loading was determined with the use of ponceau staining and computer-assisted analysis of band density (National Institutes of Health Image 1.59b4). Membranes were blocked for 1 h at room temperature in 5% milk-PBS solution and incubated overnight at 4°C with rabbit polyclonal anti-cyclin A (Santa Cruz Biotechnology; Santa Cruz, CA), 1:200 dilution in 5% milk-PBS and 0.1% Tween (PBST) (Sigma Chemical). Membranes were washed in PBST and incubated in peroxidase-labeled secondary antibody (goat anti-rabbit 1:10,000 dilution in 5% milk-PBST solution) for 1 h at room temperature. Membranes were washed in PBST and then PBS. Antigen-antibody complexes were revealed by enhance chemiluminescence system. Quantification of the immunoblot was performed by band densitometric analysis (NIH Image 1.59b4). Band density values are expressed as absolute values normalized for protein loading using ponceau density.

NF-κB immunoassay.

An ELISA was employed to investigate nuclear factor-κB (NF-κB) activity in VSMC (15). This assay is based on the specific binding of the active form of NF-κB from cellular extracts to a NF-κB consensus site oligonucleotide attached to an ELISA plate. The primary antibody recognizes an epitope of the p65 subunit, accessible only after degradation of inhibitory-κB with subsequent release of p65 from its stable cytoplasmic heterodimer. A secondary horseradish peroxidase-conjugated antibody provides colorometric readout quantified by spectrophotometry. Positive controls for NF-κB p65 subunit were provided from cellular extracts previously evaluated by both ELISA and electromobility gel shift assay (15). To enhance the sensitivity of the assay, both wild-type and mutated consensus oligonucleotides were employed in each reaction.

Human VSMC were plated at 5 × 106 cells/well and treated as previously described. Nuclear extracts of VSMC were prepared 1 h following stimulation. An aliquot of each sample was used for cell counting, and samples were centrifuged at 1,000 rpm for 10 min at 4°C. All samples were then incubated on ice for 15 min in abuffer A (10 nM HEPES, 1.5 mM MgCl2, and 10 mM KCl, pH 7.9). After cytoplasmic fractions were removed by 15 passages through a 25-gauge needle, nuclei were centrifuged at 4°C for 6 min at 600 g. The pellet was incubated on ice for 15 min inbuffer B (20 mM HEPES, 0.42 NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, and 25% glycerol, pH 7.9) and subsequently centrifuged at 4°C for 10 min at 12,000 g. Supernatants were collected, and protein content was measured via the Commassie protein assay (Pierce; Rockford, IL). Twenty-five micrograms of total protein (per 5 × 106 VSMC) were loaded into each well and assayed according to the manufacturer's directions (Active Motif; Carlsbad, CA). Quantification of the NF-κB p65 subunit was expressed as mean absorbence (λ) per sample.

Statistical analysis.

Data are presented as means values ± SE. ANOVA with Bonferroni-Dunn post hoc analysis was used to analyze differences between experimental groups. Statistical significance was accepted within 95% confidence limits.


MCP-1 and VSMC proliferation.

We utilized several techniques to investigate the effect of MCP-1 on human VSMC proliferation in vitro. With each experiment, we simultaneously tested a well-known VSMC mitogen PDGF (10 ng/ml) (25) to demonstrate the relative potency of MCP-1 as a direct proliferative agent. To find the optimal dose of MCP-1, we performed dose-response experiments by using BrdU uptake as a marker of DNA synthesis. As shown in Fig. 1, MCP-1 induced concentration-dependent BrdU uptake. As expected, PDGF stimulated VSMC proliferation compared with control (1.9 ± 0.21 vs. 1.00 ± 0.01, P < 0.05). Compared with control VSMC, MCP-1 induced VSMC proliferation in concentrations as low as 0.1 ng/ml (1.40 ± 0.14, P < 0.05 vs. control). Maximal MCP-1 stimulation was observed at 1 ng/ml (1.65 ± 0.16, P < 0.05 vs. control). Higher doses of MCP-1 appeared to have no influence on VSMC proliferation. Quantitatively, PDGF did promote more vigorous BrdU uptake than VSMC stimulated with MCP-1; however, this difference was not significant. From these results, a concentration of 1 ng/ml of MCP-1 was used in subsequent experiments.

Fig. 1.

Human vascular smooth muscle cells (VSMC) proliferation via bromodeoxyuridine (BrdU). VSMC were stimulated with various doses of monocyte chemotactic protein-1 (MCP-1) for 48 h, and BrdU incorporation was analyzed by photometric immunoassay. A maximal response was demonstrated at 1 ng/ml. This increase in BrdU uptake was comparable to that observed after platelet-derived growth factor (PDGF, 10 ng/ml) stimulation [vs. control (Cont), *P < 0.05].

Whereas MCP-1 induced BrdU uptake, we sought to validate this assay of VSMC proliferation by also directly counting cell numbers following stimulation (Fig. 2). Compared with control, MCP-1 induced a twofold increase in VSMC numbers (56,303 ± 3,935 vs. 27,299 ± 1,505 cells/ml, P < 0.05). The relative proliferative influence of MCP-1 was similar to that induced by PDGF (54,975 ± 5,161 cells/ml).

Fig. 2.

Human VSMC proliferation via direct cell counting. VSMC were stimulated for 48 h with MCP-1 (1 ng/ml) or PDGF (10 ng/ml). MCP-1 induced a twofold increase in cell numbers compared with control. Proliferative effect of MCP-1 was similar to that observed with PDGF (vs. control, *P < 0.05).

MCP-1 and the cell cycle.

Having demonstrated that MCP-1 increases VSMC numbers, we next wanted to examine the possible mechanisms promoting MCP-1-induced VSMC proliferation. We first interrogated the influence of MCP-1 on the cell cycle machinery. PCNA expression was examined with two different techniques (Fig. 3). Immunohistochemical staining with a monoclonal antibody against PCNA qualitatively revealed an increase in PCNA expression in VSMC stimulated with MCP-1 or PDGF. With the use of confocal microscopy, PCNA expression was quantified by calculating the percentage of PCNA-positive VSMC over 10 randomly selected fields. Compared with control, MCP-1 increased VSMC expression of PCNA threefold (13.1 ± 4.7 vs. 46.6 ± 7.4% VSMC with PCNA, P < 0.05). Interestingly, stimulation with PDGF resulted in nearly 80% of VSMC expressing PCNA. These immunohistochemical observations were corroborated by Western blot immunoanalysis. Compared with control, VSMC stimulated with MCP-1 or PDGF increased PCNA protein expression. As previously demonstrated, PDGF appeared to have a more vigorous effect on PCNA expression than MCP-1.

Fig. 3.

MCP-1 and proliferating cell nuclear antigen (PCNA) expression. A: immunohistochemical detection of PCNA following stimulation with MCP-1 (1 ng/ml) or PDGF (10 ng/ml) for 24 h. Labeled PCNA is stained green; VSMC membranes are stained red. Compared with control VSMC (a), MCP-1 (b) and PDGF (c) stimulation qualitatively increased the number of PCNA-stained VSMC. B: PCNA expression was quantified immunohistochemically using confocal microscopy. Compared with control, MCP-1 PCNA expression was increased 3.5-fold following MCP-1 stimulation (*P < 0.05). PDGF appeared to stimulate PCNA expression more than both control and MCP-1 treated cells. C: these observations were corroborated by Western blot immunoblots, which demonstrated increased PCNA expression in both MCP-1 and PDGF-treated VSMC.

We further examined the influence of MCP-1 on cyclin A expression. Twenty-four hours after stimulation, cyclin A protein expression was assessed by Western immunoanalysis (Fig.4). Qualitatively, both MCP-1 and PDGF increase cyclin A expression. Compared with control, densitometric quantification of immunoblots reveal that MCP-1 increases cyclin A expression nearly twofold (0.99 ± 0.3 vs. 1.91 ± 0.1,P < 0.05). As seen with PCNA expression, PDGF had an even more robust influence on cyclin A expression than MCP-1.

Fig. 4.

MCP-1 and cyclin A expression. A: 24 h after stimulation, cyclin A protein expression was assessed by Western blot. B: densitometric quantification of immunoblots demonstrated that MCP-1, while not to the same level induced by PDGF, increased cyclin A expression 1.8-fold over control (*P< 0.05).

MCP-1 and intracellular signaling.

We next sought to examine the influence of MCP-1 on two intracellular signaling pathways that are known to be important to VSMC proliferation. To assess the influence of MCP-1 on NF-κB, we stimulated VSMC and measured NF-κB activity by using an enzyme-linked immunoassay (Fig.5). We utilized a newer immunoassay that is 10-fold more sensitive than the traditional electromobility shift assay to quantitate activation of NF-κB (15). VSMC stimulated with PDGF had a vigorous NF-kB response compared with control cells (0.41 ± 0.32 vs. 0.09 ± 0.01 mean λ/sample, P < 0.05). This effect was similar to that seen with a classic NF-kB stimulus, LPS (100 ng/ml). Interestingly, MCP-1 had no effect on nuclear translocation of the NF-kB p65 subunit (0.12 ± 0.02). These results represent experiments performed on two separate VSMC donors, in duplicate through two separate passages.

Fig. 5.

MCP-1 and nuclear factor-κB (NF-κB) activation. One hour after stimulation, VSMC nuclear extracts were prepared and assayed for the NF-κB p65 subunit. Results are presented as mean absorbence/per 5 × 106 VSMC. Both PDGF and the well-known NF-κB stimulant LPS markedly increase identified p65 (*P < 0.05). Compared with control, MCP-1 appears to have no effect on nuclear translocation of the NF-κB p65 subunit.

To interrogate the importance of phosphotidylinositol 3-kinase (PI3 kinase) signaling to MCP-1-induced VSMC proliferation, we treated VSMC with a specific PI3 kinase antagonist, LY-294002 (Santa Cruz Biotechnology) and measured BrdU uptake (Fig.6). As previously demonstrated, MCP-1 increased BrdU uptake compared with control (0.25 ± 0.03 vs. 0.44 ± 0.06 λ, P < 0.05). When treated with LY-294002, MCP-1 invoked BrdU uptake only to the level of control (0.21 ± 0.02). When given to control cells, LY-294002 did not appear to affect BrdU uptake (0.24 ± 0.03 λ).

Fig. 6.

MCP-1 and phosphotidylinositol 3 (PI3) kinase signaling. Twenty-four hours after stimulation, MCP-1 increased BrdU uptake nearly twofold (*P < 0.05). With concomitant administration of the PI3 kinase inhibitor LY-294002, the proliferative effect of MCP-1 was reduced to the level of control. When given alone, LY-294002 had no effect on BrdU uptake.


Three studies have previously investigated the influence of MCP-1 on VSMC proliferation in vitro, and they offer disparate conclusions. Ikeda and colleagues (10) demonstrated dose-dependent inhibition of [3H]thymidine uptake in rat aortic VSMC after exposure to MCP-1. They reported maximal inhibition at a dose of 100 ng/ml; no difference was identified at our ideal dose of 1 ng/ml. Direct cell counting was performed, but an inhibitory effect was not detected until 4 days of culture. Additional experiments suggested that their observations were not related to VSMC release of prostaglandins or nitric oxide. Porreca and colleagues (14) similarly evaluated the influence of MCP-1 on rat aortic VSMC proliferation. Compared with Ikeda's group, these investigators altered culture conditions: longer preincubation, longer phase of cell quiescence, and lower cell density. They demonstrated a dose-dependent increase in VSMC proliferation as assessed by both [3H]thymidine uptake and cell numbers. A maximal effect was observed at 100–200 ng/ml. Most recently, Watanabe and colleagues (30) investigated the interaction between MCP-1 and serotonin on rabbit aortic VSMC proliferation. These investigators demonstrated no effect of MCP-1 (dose range 25–200 ng/ml) on VSMC proliferation as assessed by [3H]thymidine uptake. MCP-1 did, however, augment the mitogenic influence of serotonin.

In the present study, we demonstrate that MCP-1 induces human VSMC proliferation as assessed by BrdU uptake and cell counting. From previously conflicting reports, our results must be interpreted with several caveats. In their discussion, Porreca and colleagues (14) mention the importance of specific culture criteria in distinguishing their findings from Ikeda's study. Indeed, we performed our experiments after a 48-h period of growth arrest. Care was also taken to seed VSMC to avoid confluency before stimulation. Certainly, variations in culture technique might affect our observations. Importantly, our model examined human VSMC. All of the above-mentioned studies utilized human MCP-1 to stimulate rat or rabbit aortic VSMC. Quite possibly, differing results may reflect unique interactions between human protein and other species' chemokine receptors. Finally, Porreca and colleagues (14) identified their optimal dose of MCP-1 at either 100 or 200 ng/ml. We observed maximal VSMC stimulation at 1 ng/ml while still noting a difference in growth at a dose as low as 100 pg/ml. We can only speculate the relevance of these doses. In native, noninjured vessels, MCP-1 expression is very low (5, 13). Two hours after balloon injury in rat carotid arteries, MCP-1 concentrations increase to nearly 130 pg/mg protein (5). Extrapolating from these observations, the doses identified in our experiments appear to be relevant.

We implicate several different mechanisms for MCP-1-induced VSMC proliferation. Whereas we did not perform flow cytometric studies, Porreca and colleagues (14) previously reported that MCP-1 promoted a twofold increase in the percentage of VSMC in S-phase, with a corresponding decrease of cells in the G0/G1phase. Our study extends these observations by examining specific mitotic marker proteins. Immunohistochemical and Western blot immunoassays suggest that MCP-1 directly influences the cell cycle. PCNA is a 36-kDa polypeptide, which is only expressed in the nuclei of cells that are actively proliferating. Cyclin A is only one of a myriad of molecules that assist in regulating the cell cycle (27). It is especially important in governing the G1/S transition and the S phase of mitosis (6). Whereas we demonstrate that MCP-1 induces cyclin A protein expression, we acknowledge that there are likely several other cell cycle proteins that may be involved.

Previous studies have suggested that MCP-1 activates VSMC calcium-dependent protein kinase C and mitogen-activated protein kinase pathways (14, 30). We looked at two additional intracellular signaling intermediates as candidate targets for MCP-1. Classically, NF-κB exist in the cytoplasm as a heterodimer of its two subunits, p50 and p65, and its inhibitory protein, IκB (26). On activation, IκB is degraded, allowing the p65 subunit to migrate to the nucleus and bind its target DNA. We have previously demonstrated a causal relationship between cytokine and growth factor-induced VSMC proliferation and activation of NF-κB (23, 24). As such, we were intrigued to observe that MCP-1 did not increase p65 translocation into stimulated VSMC nuclei. To our knowledge, there are no reports that implicate NF-κB as a transduction intermediate in MCP-1 and its receptor signaling.

When activated, the PI3 kinase pathway is important in mitogen-induced cellular proliferation, quite possibly by promoting entry into the S phase of the cell cycle (4, 16). Whereas MCP-1 gene expression is mediated by PI3 kinase (1), MCP-1 activation of PI3 kinase has been limited to monocytic cell lines (28). We used a specific, reversible inhibitor, LY-294002, to demonstrate that MCP-1-induced VSMC proliferation is, in part, PI3 kinase dependent.

We performed most experiments using PDGF, a well-known VSMC mitogen, as an external control. Compared with MCP-1-stimulated cells, PDGF seemed to promote more vigorous intracellular responses for PCNA and cyclin A. Surprisingly, the proliferative responses appeared to be similar. Whereas Porecca and colleagues (14) demonstrated MCP-1-induced VSMC proliferation, the relative increase was markedly lower than that afforded by VSMC cultured in 10% fetal calf serum. Our results suggest that MCP-1 rivals PDGF as a proliferative agent. We can only speculate why the PDGF-induced increase in cell cycle proteins did not translate into more VSMC growth than MCP-1. Intuitively, whereas sharing many transduction intermediates, PDGF receptors initiate different signals than those activated by the CCR2 receptor.

MCP-1 is clearly important in the early stages of vessel repair by recruiting mononuclear cells and assisting their transmigration. However, the role of MCP-1 in atherogenesis is likely multifunctional. MCP-1 has prothrombotic properties by inducing VSMC tissue factor production (20) and appears to promote phenotypic differentiation of VSMC (3). Interleukin-8 is another well-known chemokine that has previously been demonstrated to have mitogenic effects on VSMC (31). Similarly, MCP-1 is not only an important chemokine, but a potent growth factor for human VSMC.


  • Address for reprint requests and other correspondence: C. H. Selzman, Division of Cardiothoracic Surgery, Box C-310, Univ. of Colorado Health Sciences Center, 4200 East Ninth Ave., Denver, CO 80262 (E-mail:craig.selzman{at}UCHSC.edu).

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • June 20, 2002;10.1152/ajpheart.00188.2002


View Abstract