Endothelial cell PECAM-1 confers protection against endotoxic shock

Matthias Maas, Michelle Stapleton, Carmen Bergom, David L. Mattson, Debra K. Newman, Peter J. Newman


Platelet endothelial cell adhesion molecule-1 (PECAM-1; CD31) is a 130-kDa member of the Ig superfamily that is expressed on platelets and leukocytes and is highly enriched at endothelial cell-cell junctions. Previous studies showed that this vascular cell adhesion and signaling receptor functions to regulate platelet activation and thrombosis, to suppress apoptotic cell death, to mediate transendothelial migration of leukocytes, and to maintain the integrity of the vasculature. Because systemic exposure to the bacterial endotoxin LPS triggers an acute inflammatory response that involves many of these same processes, we compared the pathophysiological responses of wild-type versus PECAM-1-deficient mice to LPS challenge. We found that PECAM-1-deficient mice were significantly more sensitive to systemic LPS administration than their wild-type counterparts and that the lack of PECAM-1 expression at endothelial cell-cell junctions could account for the majority of the increased LPS-induced mortality observed. The diverse functional roles played by PECAM-1 in thrombosis, inflammation, apoptosis, and the immune response may make this molecule an attractive target for the development of novel therapeutics to manage and treat endotoxic shock.

  • platelet endothelial cell adhesion molecule-1
  • sepsis
  • endotoxin

platelet endothelial cell adhesion molecule-1 (PECAM-1; CD31) is a 130-kDa member of the Ig superfamily that is expressed on platelets and leukocytes and is highly enriched at endothelial cell-cell junctions (16, 17). PECAM-1 possesses homophilic adhesive properties within its extracellular domain and signals via recruitment of cytosolic proteins such as the protein-tyrosine phosphatase SHP-2 to its cytoplasmic tail (reviewed in Ref. 19). Studies in mice deficient in PECAM-1 (6) have revealed a number of interesting physiological roles for this molecule. Platelets from PECAM-1−/− mice are hyperaggregable in response to collagen and von Willebrand factor (12, 22, 24), PECAM-1-deficient leukocytes become trapped within the perivascular basement membrane during transendothelial migration (6, 27), endothelial cells lacking PECAM-1 show increased rates of apoptosis after irradiation (10), maturation of PECAM-1-deficient B cells is compromised (28), and PECAM-1−/− mice are prone to develop autoimmune disease (11, 28). Finally, there is growing evidence that PECAM-1 functions to maintain the integrity of the vasculature in vivo, because deficiency in murine endothelial PECAM-1 prolongs tail vein bleeding time (14) and delays reestablishment of the vascular permeability barrier after histamine challenge (11).

Sepsis, a systemic inflammatory response to infection, is the leading cause of death in noncardiac intensive care units. The mortality rate associated with the most severe form of sepsis, septic shock, is >40% because of accompanying organ dysfunction and refractory hypotension (13, 15). LPS (endotoxin) is a structural component of the outer membrane of gram-negative bacteria and has been used extensively to induce endotoxic shock in a variety of animal models (5). LPS administration triggers a systemic inflammatory response that includes blood vessel dysfunction, endothelial cell apoptosis, and coagulation system activation with fibrin deposition.

Because LPS administration initiates a broad range of physiological changes, some of which may be regulated by PECAM-1-mediated adhesion and/or signaling, we hypothesized that PECAM-1-deficient mice might be more susceptible to endotoxic shock. The purpose of the present investigation, therefore, was to compare the pathophysiological responses of wild-type versus PECAM-1-deficient mice to LPS challenge. We found that PECAM-1-deficient mice were significantly more sensitive to systemic LPS administration than their wild-type counterparts and that the lack of PECAM-1 expression in the vasculature accounted for the majority of the increased LPS-induced mortality observed. Given the diverse functional roles played by PECAM-1 in thrombosis, inflammation, apoptosis, and the immune response, these findings suggest that agents that target PECAM-1 may be potentially attractive and novel therapeutics for the management and treatment of endotoxic shock.



PECAM-1-deficient mice (6) that have been backcrossed for >12 generations onto a C57/BL6 background were maintained in a pathogen-free facility under the supervision of the Animal Resource Center of the Medical College of Wisconsin. All experiments were approved by the local institutional animal care and use committee at the Medical College of Wisconsin and were performed with age- and sex-matched wild-type controls as previously described (22). For survival studies, PECAM-1-deficient and wild-type mice were injected intraperitoneally with 30 mg/kg LPS O55:B5 (Sigma, St. Louis, MO). This dose had previously been determined for this lot of LPS to be lethal for ∼50% of wild-type mice. Animals were allowed free access to food and water, and survival was recorded every 12 h for 14 days. If mice showed signs of irreversible organ dysfunction, prolonged absence of voluntary responses to external stimuli, or persistent convulsions, they were killed and counted as dead.

Creation of radiation chimeras.

Chimeric mice expressing PECAM-1 on either their endothelial or their hematopoietic cell compartment were created by collecting bone marrow cells from the femur and tibia of wild-type or PECAM-1−/− mice into X-Vivo 15 medium (Atlanta Biologicals, Norcross, GA). Small mononuclear cells were enriched by gradient centrifugation through Fico/Lite-LM (density 1.086 g/l; Atlanta Biologicals), resuspended in X-Vivo 15, and injected (2 × 106 total cells) retroorbitally into isoflurane-anesthesized, lethally irradiated (11 Gy, Shepherd Mark I cesium irradiator; J. L. Shepherd, San Fernando, CA) recipient mice 24 h after irradiation. Mice were rephenotyped by flow cytometry 12 wk after bone marrow transplantation to ensure that full engraftment had taken place, with both CD3-positive leukocytes and platelets examined. Irradiated wild-type mouse recipients reconstituted with PECAM-1−/− bone marrow expressed PECAM-1 solely on their endothelial cells but not on blood cells, whereas PECAM-1−/− mouse recipients reconstituted with wild-type marrow expressed PECAM-1 on their blood cells but not on their endothelium. Lethally irradiated wild-type control mice in which wild-type marrow was transplanted were also produced to control for posttransplant sensitivity to endotoxin.

Hematologic data.

Blood samples were obtained from the inferior vena cava of anesthetized mice (120 mg/kg ketamine, 15 mg/kg xylazine ip). Blood samples were anticoagulated with 106 mM sodium citrate (final dilution 1:7) or EDTA. Hematologic parameters (hemoglobin, hematocrit, red blood cell count, white blood cell count, and platelet count) were measured on an automated Heska animal blood counter (Heska, Fort Collins, CO) that has a specific profile for mouse blood. Plasma was obtained by centrifugation at 2,200 g for 15 min. Samples were shock frozen in liquid nitrogen and stored at −80°C until further use. Plasma concentrations of IL-6, IL-1β, TNF-α, and soluble E-selectin were determined with ELISA kits from R&D Systems (Minneapolis, MN). Body temperature was measured rectally with a traceable thermometer (Fisher Scientific).

Plasma volume measurement.

Plasma volume was determined by injecting anesthetized (120 mg/kg ketamine, 15 mg/kg xylazine ip) mice retroorbitally with 600 μg of Evans blue dye (Sigma) in 200 μl of normal saline solution. Evans blue dye binds tightly to albumin. Two minutes after the end of injection (2.5 min after start of injection) blood was drawn from the inferior vena cava over 30 s and anticoagulated with EDTA. Absorbance of plasma diluted 1:20 with normal saline was spectrophotometrically determined at 620 nm, and the concentration of Evans blue dye was calculated with a standard curve. Plasma volume was calculated according to the formula plasma volume = injected Evans blue (g)/plasma Evans blue (g/l) − 200 μl (injection volume).

Determination of vascular permeability.

FITC-labeled high-molecular weight (MW) dextran (MW 200 × 104, 100 mg/kg; Sigma) and rhodamine labeled low-MW dextran (MW 4 × 104, 100 mg/kg; Sigma) were injected retroorbitally under isoflurane anesthesia in mice 8 h after treatment with LPS or vehicle (0.9% NaCl). Fifteen minutes after dextran injection, the animals were killed and heart and lungs were removed en bloc and immediately embedded in tissue freezing medium (Fisher Scientific), frozen in liquid nitrogen, and stored at −80°C until further use. Tissue sections (10 μm) were cut and analyzed on a Nikon microscope equipped with a digital camera (Spot), an ultraviolet light source, and filters to visualize FITC and rhodamine fluorescence. All images were taken with the same camera settings and exposure times.

Myeloperoxidase activity.

Left lungs were surgically removed, frozen in liquid nitrogen, and stored at −80°C until further processing. Lung samples were homogenized and sonicated in 50 mM potassium phosphate buffer, pH 6.0 containing 0.5% hexadecyltrimethylammonium bromide (Sigma). After three freeze-thaw cycles, samples were sonicated a second time, and the supernatant was collected after centrifugation at 20,000 g for 15 min. Protein (30 μg, 10 μl) was added to 190 μl of potassium phosphate buffer containing 0.166 mg/ml o-dianisidine dihydrochloride (Sigma) and 0.0025% hydrogen peroxide (Sigma). Change in absorbance at 450 nm over 5 min was measured, and the final concentration of myeloperoxidase was determined by comparison with the absorbance obtained with myeloperoxidase standards (Calbiochem, San Diego, CA).

Measurement of blood pressure and heart rate.

Small telemetry units with catheters (Data Sciences International, Arden Hills, MN) were implanted via the left carotid artery into the thoracic aorta. Mice were anesthetized with pentobarbital sodium (50 mg/kg ip). By aseptic techniques, a ventral midline incision was made in the skin over the throat and the left carotid artery was isolated. The tip of the catheter was inserted into the carotid until its tip projected into the thoracic aorta, and the catheter was secured in place with 6-0 surgical silk. The transducer-transmitter-battery was tunneled subcutaneously and placed along the right flank. Incisions were closed with sterile 4-0 monofilament suture. Mice were allowed to recover from the procedure for 24 h and then challenged with LPS as described above. Blood pressure and heart rate was measured every 5 min. Baseline blood pressure was determined from the mean of 15 independent measurements taken immediately before LPS injection. Change in blood pressure after LPS administration was determined from the mean of five measurements taken every 3 h.

Statistical analysis.

Survival data (Figs. 1 and 5) were compared with the Mantel-Haenszel log rank test and Graphpad Prism 4.0 Software. Data in Figs. 24 are presented as means ± SE. If data within groups fulfilled criteria for Gaussian distribution (Kolmogorov-Smirnov), an n-way ANOVA was performed with subsequent Tukey post hoc test with Graphpad Instat 3.0 and Prism 4.0 software. The acceptable level of significance was P < 0.05.

Fig. 1.

Platelet endothelial cell adhesion molecule-1 (PECAM-1)-deficient mice are highly sensitive to endotoxin. PECAM-1−/− and wild-type mice were injected intraperitoneally with 15 μg/g (left) or 30 μg/g (right) LPS, and survival was noted every 6 h. The data shown were accumulated from 2 (15 μg/g) and 4 (30 μg/g) independent experiments, each giving similar results. A log rank test between groups showed statistical significance of P < 0.02 (15 μg/g) and P < 0.0001 (30 μg/g).

Fig. 2.

PECAM-1-deficient mice have an exaggerated inflammatory response to LPS. Hematologic, plasma cytokine level, and lung myeloperoxidase (MPO) activity data from PECAM-1−/− and PECAM-1+/+ mice at various time points after LPS challenge are shown. Platelet (Plt) count (n = 14–25/data point; A) and white blood cell (WBC) count (n = 14–25/data point; B) were obtained from EDTA-anticoagulated blood samples. C: soluble (s)E-selectin in plasma samples was determined by ELISA (n = 5 for time 0 and n = 10–12 for 2, 8, and 20 h after LPS). D: body temperature (n = 6–16/data point) was measured rectally. E: lung MPO activity (n = 7/data point). Plasma cytokine levels of IL-6 (F), IL-1β (G), and TNF-α (H) (n = 5 for time 0 and n = 10–12 for 2, 8, and 20 h after LPS) were determined by ELISA. Statistically significant differences between PECAM-1−/− and PECAM-1+/+ mice at each time point: *P < 0.05, **P < 0.01, ***P < 0.001.


PECAM-1-deficient mice are highly sensitive, and show an exaggerated inflammatory response, to endotoxin.

Previous studies showed that absence of PECAM-1 results in platelet and leukocyte hyperresponsiveness, exaggerated anaphylactic responses, increased susceptibility to endothelial cell apoptosis, and vascular dysfunction. Because sepsis and endotoxic shock initiate physiological responses that involve many of these same parameters, we sought to determine whether PECAM-1-deficient mice might show increased sensitivity after exposure to endotoxin. As shown in Fig. 1, intraperitoneal administration of 15 μg/g body wt LPS (the LD50 for C57BL/6 PECAM-1-deficient mice) had virtually no effect on the survival of wild-type littermate control mice (Fig. 1, left). Likewise, administration of 30 μg/g LPS (the LD50 for C57BL/6 wild-type mice) resulted in the death of nearly all PECAM-1−/− mice within 48 h—a value that was highly statistically significant (P < 0.0001). No biologically significant differences were found in platelet count (Fig. 2A), in the plasma level of soluble E-selectin (Fig. 2C), or in body temperature (Fig. 2D). Moreover, LPS had no measurable effect on the bronchoalveolar lavage (BAL) fluid in either normal or knockout mice. BAL protein was not elevated, no increase in BAL cell number was seen, and there were no neutrophils in the BAL fluid (not shown). However, there was a small, but statistically significant, difference in total leukocyte count 20 h after LPS injection (Fig. 2B), and we observed consistently higher levels of neutrophil-derived myeloperoxidase in isolated whole lung lysates of PECAM-1-deficient mice 8 h after LPS administration (Fig. 2E). Given the lack of neutrophils in BAL fluid, these data suggest that the increased myeloperoxidase activity was due to neutrophils that had marginated into the vascular pool or into the interstitium, similar to previous findings (6, 27) of delayed neutrophil transmigration in PECAM-1-deficient animals. Although increased lung neutrophils are unlikely to account for the differential survival of PECAM-1-deficient versus wild-type mice shown in Fig. 1, they likely reflect the exaggerated acute inflammatory response of PECAM-1-deficient mice to LPS, as further evidenced by elevated levels of the proinflammatory cytokines IL-6, IL-1β, and TNF-α found in the plasma of PECAM-1−/− mice 8 and 20 h after LPS administration (Fig. 2, F–H). Together, these results indicate that PECAM-1-deficient mice exhibit an exaggerated systemic inflammatory response in response to LPS challenge.

Failure of PECAM-1-deficient mice to recover from LPS-induced septic shock.

Although not statistically significant at all time points, PECAM-1-deficient mice in general had a lower body temperature than wild-type mice after exposure to LPS (Fig. 2D), consistent with the observation that septic shock was developing earlier and more severely in these mice (Fig. 1). Additional evidence that PECAM-1−/− mice are hypersensitive to endotoxin was obtained by monitoring blood pressure and heart rate in the thoracic aorta with a small catheter inserted through the left carotid artery and connected to a transponder that had been implanted within the right flank of the mouse (3). As shown in Fig. 3A, the baseline mean blood pressure in wild-type and PECAM-1−/− mice was virtually identical [111 ± 5 (n = 9) and 111 ± 2 (n = 11) mmHg, respectively] and the mice exhibited a similar decrease in blood pressure during the first 24 h after LPS administration (after 12 h: 81 ± 5 and 70 ± 4 mmHg, respectively). After that, however, the mean blood pressure curves spread significantly, with wild-type mice recovering substantially, whereas the blood pressure of PECAM-1-deficient mice remained low or continued to decline. In keeping with the observed hypotension, the heart rate of PECAM-1−/− mice also became significantly slower over the same period of time, falling to a mean of 313 ± 13 beats/min within the first 12 h compared with 456 ± 25 beats/min in PECAM-1+/+ mice (Fig. 3B). Thus PECAM-1-deficient mice fail to recover from LPS challenge and progress more readily to endotoxic shock than their wild-type counterparts.

Fig. 3.

Failure of PECAM-1-deficient mice to recover from endotoxin-induced hypotension. Mean blood pressure (BP; A) and heart rate (B) were measured in PECAM-1+/+ (wild type, WT) and PECAM-1−/− mice for 48 h after LPS challenge. Note that BP and heart rate of both phenotypes start out the same but deteriorate and fail to recover in PECAM-1-deficient mice. Two-way ANOVA testing yielded significant differences between WT and PECAM-1−/− mice for BP (P < 0.0001) and for heart rate (P < 0.0001).

Increased vascular permeability in PECAM-1-deficient mice after LPS challenge.

Absolute and relative hypovolemia are important contributors to the hypotension that accompanies endotoxic shock (21). Previous studies have shown that absence of PECAM-1 from endothelial cell-cell junctions can compromise the integrity of the vasculature, either in response to physical injury (14) or in response to the inflammatory mediator histamine (11). Therefore, we hypothesized that fluid leakage from the vasculature into the surrounding tissues might be contributing to the increased hypotensive state observed in PECAM-1−/− versus PECAM-1+/+ mice after LPS challenge. This was tested in a number of ways. First, we compared absolute plasma volume and hematocrit in PECAM-1+/+ versus PECAM-1−/− mice at 0, 2, and 8 h after LPS administration. As shown in Fig. 4a, the mean total plasma volume in PECAM-1−/− mice was significantly lower than in wild-type mice 8 h after LPS administration, and the hematocrit was correspondingly higher (Fig. 4b). Evidence that the decreased plasma volume in PECAM-1−/− mice was caused by increased vascular permeability was also obtained by injecting mice systemically with a solution containing FITC-labeled high-MW (200 × 104) dextran and rhodamine-labeled low-MW (4 × 104) dextran and examining lung tissue for evidence of vascular leakage 8 h after LPS challenge. As shown in Fig. 4, c and d, increased penetration of rhodamine-conjugated low-MW dextran from the vessels into their immediate surroundings could be readily observed in PECAM-1-deficient mice. Together, these data suggest that loss of PECAM-1 from the endothelium compromises the permeability barrier, leading to hypovolemia, hypotension, and endotoxic shock after challenge with LPS.

Fig. 4.

Increased vascular permeability in PECAM-1-deficient mice after LPS challenge. a: Plasma volume was determined from spectrophotometric readings of plasma samples taken at 0, 2, and 8 h after injection of 600 μg of Evans blue dye and expressed as % body weight (BW). b: Hematocrit obtained from EDTA-anticoagulated blood at the indicated time points after LPS challenge (n = 14–25/data point). PECAM-1+/+ (c) and PECAM-1−/− (d) mice were injected with rhodamine-labeled low-molecular weight dextran (red) and FITC-labeled high-molecular weight dextran (green) 0 and 8 h after LPS challenge. After 15 min of circulation, lungs and hearts were harvested en bloc and immediately frozen in tissue freezing medium with liquid nitrogen; 10-μm sections were cut, and lung tissue was analyzed by fluorescent microscopy. Note that vessels derived from PECAM-1−/− mice (d) were substantially more permeable to low-molecular weight dextran (red arrows) 8 h after exposure to endotoxin than their wild-type counterparts (c). Statistically significant differences between PECAM-1−/− and PECAM-1+/+ mice at each time point: **P < 0.01.

Endothelial cell PECAM-1 rescues PECAM-1-deficient mice from endotoxic shock.

The above observations predict that specific restoration of PECAM-1 to the endothelial compartment might be able to reverse the sensitivity of PECAM-1−/− mice to endotoxic shock. We addressed this question by creating radiation chimeras that selectively express PECAM-1 either on their endothelial cells or on their bone marrow-derived blood cells. PECAM-1-deficient mice that had been transplanted with PECAM-1+/+ marrow produced >95% PECAM-1-positive, CD3-positive lymphocytes (Fig. 5B), whereas PECAM-1+/+ mice transplanted with PECAM-1−/− marrow produced <5% PECAM-1+/+ lymphocytes (Fig. 5C). Other hematopoietically derived blood cells had similar degrees of thorough reconstitution (not shown). When challenged with 30 mg/kg LPS, lethally irradiated PECAM-1+/+ mice that had been reconstituted with PECAM-1+/+ marrow showed somewhat greater mortality than did nonirradiated, nontransplanted wild-type control mice (compare Figs. 1 and 5D)—a finding not uncommon in mice having undergone total bone marrow ablation-reconstitution only 3 mo beforehand (23). Of more importance, however, was the finding that chimeric mice with PECAM-1+/+ endothelium and PECAM-1−/− blood cells were nearly as resistant to endotoxic shock as were the wild-type → wild-type reconstituted controls (Fig. 5D), indicating that expression of PECAM-1 on endothelial cells alone is sufficient to rescue PECAM-1-deficient mice from their sensitivity to LPS-induced endotoxic shock. In contrast, the degree of resistance to LPS in mice expressing normal levels of PECAM-1 on their blood cells but none on their endothelial cells (Fig. 5B) was statistically insignificant compared with that observed in totally PECAM-1-deficient animals (compare Figs. 1 and 5D). These data demonstrate that endothelial PECAM-1 is sufficient to confer protection against endotoxic shock.

Fig. 5.

Endothelial cell PECAM-1 rescues PECAM-1-deficient mice from endotoxic shock. Radiation chimeras that exclusively express PECAM-1 either on their endothelium or on their circulating blood cells were created as described in materials and methods. A: flow cytometric analysis of lymphocytes from PECAM-1−/− (knockout, KO) and PECAM-1+/+ (WT) mice. Twelve weeks after bone marrow transplant, >95% of the lymphocytes from PECAM-1−/− mouse that had been engrafted with PECAM-1+/+ bone marrow cells had become positive for PECAM-1 expression (B); the converse was found for PECAM-1+/+ mice that had been engrafted with PECAM-1−/− bone marrow cells (C). D: survival of endothelial (EC) PECAM-1−/−, blood cell PECAM-1+/+ mice (n = 16), blood cell PECAM-1−/−, EC PECAM-1+/+ mice (n = 15), and EC PECAM-1+/+, blood cell PECAM-1+/+ mice (n = 10) mice after LPS challenge. Mice were injected intraperitoneally with 30 mg/kg BW LPS, and survival was noted every 6 h. Data shown were combined from 4 independently performed experiments, each yielding similar results. ***Statistical significance by log rank test (P < 0.001).

It is noteworthy that no structural or functional abnormalities have been reported in the developing or adult vasculature of PECAM-1-deficient mice (6) and that PECAM-1−/− mice have normal blood pressure and heart rate under basal conditions (Fig. 3). Similar to many other in vivo models involving genetically modified mice, overt phenotypic consequences of PECAM-1 deficiency do not seem to be manifested unless the animal is subjected to physical or pathological challenge. Thus, despite the fact that PECAM-1−/− platelets are hyperresponsive to a variety of stimuli (4, 12, 22), and that PECAM-1-deficient animals have a tendency to produce larger thrombi more rapidly at sites of laser-induced vascular injury (8), unchallenged PECAM-1-deficient mice exhibit no obvious signs of thrombotic disease. Similarly, PECAM-1 deficiency has not been associated to date with a tumorigenic phenotype, despite the fact that this receptor has been shown in a number of cellular systems to afford significant protection against apoptotic cell death (2, 7, 9, 10, 20). Given the reported role of PECAM-1 in cell survival, careful studies were also done to look for increasing endothelial and leukocyte apoptosis in PECAM-1-deficient mice after LPS challenge, but none was found (data not shown). We are now pursuing a role for PECAM-1 in regulating cellular responses to cytokine signaling.

The major finding of the present work—that blood pressure, plasma volume, and vascular integrity all become compromised in PECAM-1-deficient mice after LPS challenge—provides further evidence that this vascular cell adhesion and signaling molecule contributes importantly to homeostasis, most notably under conditions of physiological stress. Whether the ability of PECAM-1 to suppress inflammation and maintain an intact vascular permeability barrier after systemic exposure to endotoxin is due to its homophilic adhesive properties (1, 26), its concentration at endothelial cell junctions (18, 25), or its ability to recruit and activate cytosolic signaling molecules to its cytoplasmic domain (19) are interesting and important questions that remain to be addressed. In any event, the unique combination of functional roles for PECAM-1 in thrombosis, inflammation, apoptosis, and the maintenance of vascular integrity makes it a novel and potentially attractive therapeutic target for the treatment of endotoxic shock.


This work was supported by Grant MA 6 1 03 02 from “Innovative Medizinische Forschung,” Muenster, Germany, Grant MA 2604/1–1 from the Deutsche Forschungsgemeinschaft, Germany (to M. Maas), and National Heart, Lung, and Blood Institute Grant HL-40926 (to P. J. Newman and D. K. Newman).


The authors are grateful to Drs. Hartmut Weiler, Berend Iserman, Drazen Zagorac, and Demin Wang for helpful discussions during the course of this investigation. We also thank Drs. Steven Albelda and Melpo Christofidou-Solomidou (University of Pennsylvania, Philadelphia, PA) for murine BAL analysis.

Portions of this work were presented in abstract form at the 45th annual meeting of the American Society of Hematology, December 6–10, 2003.


  • 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.


View Abstract