Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Granulocyte colony-stimulating factor (G-CSF) is a glycoprotein hormone that induces proliferation and differentiation of neutrophil progenitor cells in humans and other primates. It has been purified (1, 2), cloned (2) and produced on a large scale using recombinant techniques (5). Currently, recombinant human granulocyte colony-stimulating factor (rhG-CSF) is widely used clinically to reduce the incidence and duration of neutropenia in patients treated with oncologic chemotherapy. On the other hand, studies investigating the effect of rhG-CSF on mature neutrophils in vitro have shown that rhG-CSF has effects on mature neutrophils, such as prolongation of life span, acceleration of chemotaxis, superoxide anion production, and augmentation of adherence (6). Transient decreases in circulating neutrophil numbers after G-CSF injection have been reported (10); however, the precise mechanism remains unclear. Recent reports have suggested that treatment with rhG-CSF in patients receiving cytotoxins can be associated with pulmonary toxicity (13). Because neutrophil sequestration in the lung induced by intravascular inflammatory mediators can result in injury to the alveolocapillary membrane (17, 18), and since development of adult respiratory distress syndrome (ARDS) in patients treated with granulocyte-macrophage colony-stimulating factor has been reported (19), it is possible that G-CSF initiates acute lung injury.

The purpose of this study is to determine whether intravenous injection of rhG-CSF causes neutrophil sequestration in the microvasculature of the lung, increases pulmonary vascular permeability, and induces acute lung injury in rabbits. To investigate the possible mechanisms underlying G-CSF-induced neutrophil sequestration, the effect of pretreatment of rhG-CSF on neutrophil deformability and expression of adhesion molecules (CD11/CD18 complex) were studied. Neutrophil deformability was determined by measuring the pressure required to pass neutrophils through a polycarbonate microfilter (20). The surface expression of adhesion molecules was examined by flow cytometry analysis using monoclonal antibodies to CD11a, CD11b, CD11c, and CD18 (25). These studies show that intravenous injection of rhG-CSF caused a sudden neutropenia and neutrophil sequestration in the lung, possibly by inducing a decrease in neutrophil deformability due to actin polymerization and upregulation of CD11/CD18 expression. However, a single injection of rhG-CSF neither increased pulmonary vascular permeability nor induced massive lung injury in rabbits.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

In Vivo Studies Evaluating Neutrophil Sequestration and Vascular Permeability

Effect of intravenous injection of rhG-CSF on circulating neutrophil counts. New Zealand white rabbits weighing 3.2 ± 0.1 (SD) kg were anesthetized using ketamine hydrochloride (80-100 mg/kg i.m.) and xylazine, and catheters were placed in the left carotid artery and the right jugular vein. The animals received a tracheostomy and breathed room air. Heparin (200 U/kg) was injected into the left carotid artery. After baseline (time 0) samples of blood were drawn, an intravenous injection of rhG-CSF (filgrastim; Kirin Brewery Co., Tokyo, Japan), 25 µg/4 ml saline (n = 7) or 4 ml saline (n = 7) was given into the right jugular vein. Arterial and venous blood (1 ml each) was sampled for cell counts at 2, 4, 7, 11, 15, 30, 60, and 90 min after rhG-CSF or saline injection. Blood volume was replaced by saline at each time point. Circulating blood cell counts were corrected for hematocrit and expressed as the percent change in corrected values from the baseline (time 0) values.

Effect of rhG-CSF on extravascular protein and lung water. The albumin concentration in the bronchoalveolar lavage fluid (BALF) and lung water were evaluated at 90 min in both G-CSF-treated and saline-treated groups. Immediately after blood samples at 90 min were collected, animals' hearts were stopped using an arterial injection of saturated potassium chloride (KCl) and the animals were exsanguinated prior to the surgical removal of the thoracic organs. Bronchoalveolar lavage (BAL) was performed three times using 15 ml saline delivered to the right lower lobe. Total cell counts, differential leukocyte counts of the BALF, and the albumin concentration in the epithelial lining fluid (ELF) corrected by diffusion of urea N (26) were compared between the two groups. Lung water was evaluated by using the wet/dry lung weight ratios in the right upper lobe. Prior to the BAL procedure, the right upper lobe was resected, free blood was gently removed by paper towels, and wet lung weight was measured immediately. The tissues were then dried in a vacuum drying oven at 80°C for 72 h. The weight of the dried lung tissue was determined and the wet/dry lung weight ratio was calculated.

Serum level of G-CSF. Serum level of G-CSF was measured at 0, 0.5, 1, 2, 15, 30, 60, 90, and 180 min by radioimmunoassay (n = 3). Two milliliters of arterial blood were taken and circulating neutrophil counts corrected for hematocrit were examined simultaneously. Circulating neutrophils were expressed as the percent change in corrected values from the baseline (time 0) values.

Histology of the Lung; Arterial-Venous (A-V) Differences in Circulating Neutrophils within 2 min

Experiments examining the immediate A-V differences in neutrophil numbers immediately after injection of rhG-CSF and the histology of the lung at 30 min were performed in the same manner described earlier. Arterial and venous blood samples (1 ml each) for cell counts were simultaneously obtained at baseline (time 0) and 0.5, 1, and 2 min after rhG-CSF (n = 6) and saline (n = 6) injection. The histology of the lung was evaluated at 30 min. KCl was injected arterially to stop the animal's heart. Immediately after the chest was opened, the base of the heart was tied off and the left hilum was clamped to prevent blood loss from the lungs. After the thoracic organs were removed, 2.5% glutaraldehyde in phosphate buffer was instilled from the trachea at 30 cm H₂O pressure. Then the left lung was cut at the midsagittal line, sectioned at 2 µm, and stained with hematoxylin and eosin. The morphology of the lung was examined and the concentration of neutrophils relative to red blood cells (RBC) in the capillaries (< 20 µm in diameter) and the small vessels (20-50 µm in diameter) was quantitated in 10 randomly selected fields as previously described (27).

In Vitro Studies of Neutrophil Deformability and CD11/CD18 Expression

Neutrophil isolation. Rabbit neutrophils were isolated from peripheral blood using previously described methods (28). In brief, dextran (1-2 × 10⁵ mol wt, 1.7% final concentration) was combined with blood anticoagulated with acid-citrate-dextrose to sediment the RBC. After hypotonic lysis in water for 20 s, the neutrophils were separated from the mononuclear cells by centrifugation through Histopaque (Sigma 1007; Sigma, St. Louis, MO). The neutrophil purity was > 95%.

Filtration assay. The effect of G-CSF on neutrophil deformability was examined using the microfilter technique (24), which measured the pressure needed to pass a bolus of neutrophils through a polycarbonate filter with a uniform pore diameter of 5 µm (Nucleopore, Pleasanton, Canada). The roles of both the actin and the microtubular components of the cytoskeleton were examined. Microtubular assembly was inhibited by colchicine, and F-actin formation was inhibited by cytochalasin B (24). Isolated neutrophils from each rabbit were divided into 0.2-ml aliquots of 4 × 10⁶/ml, and deformability was examined in the following five groups (n = 5 in each group): group 1, neutrophils + saline (control); group 2, neutrophils + G-CSF (250 ng/ml); group 3, neutrophils + 50 µM colchicine + G-CSF (250 ng/ml); group 4, neutrophils + 0.2% dimethyl sulfoxide (DMSO) + G-CSF (250 ng/ml); group 5, neutrophils + 10 µM cytochalasin B + G-CSF (250 ng/ml). Isolated neutrophils were incubated with colchicine (Sigma; final concentration, 50 µM in saline) for 30 min at room temperature before treatment with G-CSF. Cytochalasin B (Sigma) was dissolved in DMSO and mixed with the isolated neutrophils for 5 min at room temperature (final concentration, 10 µM cytochalasin B and 0.2% DMSO). Neutrophil deformability was measured after 2 min of treatment with saline or G-CSF (final concentration, 250 ng/ ml) at 37°C. The rabbit neutrophils (0.2 ml of 4 × 10⁶/ml buffer) were injected as a bolus from the port located above the polycarbonate filter, immediately followed by continuous infusion of phosphate buffer at a constant flow rate of 2 ml/min.

Surface Expression of Adhesion Molecules, CD11/18 Complex on Neutrophils

Human whole blood samples from healthy volunteers were used to investigate the effect of rhG-CSF on the expression of CD11a, CD11b, CD11c, and CD18 on neutrophils using flow cytometry analysis. Three groups were studied (n = 4 in each group): group 1, human whole blood + G-CSF (250 ng/ml); group 2, human whole blood + G-CSF (25 ng/ ml); group 3, human whole blood + saline. Heparinized human whole blood was incubated with the high (250 ng/ ml) or low (25 ng/ml) concentration of rhG-CSF or with saline at 37°C. The expression of adhesion molecules on neutrophils was measured at baseline (time 0) and 5, 30, and 60 min. The monoclonal antibodies used in this study were anti-CD11a, CD11b, CD11c, CD18, and an irrelevant antibody, each of which was fluorescein isothiocyanate-conjugated (Dako Ltd., High Wycombe, UK). CD11/CD18 expressions were measured as previously described (25). A monoclonal antibody (20 µl) was added to 100 µl of whole blood and incubated for 30 min at 4°C. Samples were lysed and fixed with fluorescent-activated cell sorter lysing solution and washed twice with phosphate-buffered saline. Cells were then analyzed by flow cytometry in which polymorphonuclear leukocytes were selectively gated. The specific fluorescence was calculated by subtracting the mean fluorescent intensity of the control antibody from that of the anti-CD11/CD18 antibody. Levels of labeled CD11a, CD11b, CD11c, and CD18 were expressed as the percent increase from the baseline values.

Statistics

The results are expressed as means ± SEM. Two groups of animals, G-CSF-treated and saline-treated, were compared using the Student's standard two-tailed t test. A repeated-measures analysis of variance was used to compare A-V differences in circulating neutrophils. Probability values of less than 0.05 were considered significant.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Effect of Intravenous Injection of G-CSF on Circulating Blood Cell Counts

Intravenous injection of rhG-CSF (25 µg/4 ml saline) caused the circulating white blood cell (WBC) counts to decrease to 60% from the baseline values at 2 min in both the arterial and venous blood. This decrease continued to 90 min after G-CSF injection (Figure 1). Intravenous injection of rhG-CSF caused circulating neutrophils to fall to 10.3 ± 3.6%, 3.4 ± 1.6%, and 3.9 ± 1.2% of the baseline value after 2, 4, and 7 min in venous blood, respectively (Figure 2). This neutropenia continued for 15 min and then gradually increased to 30% of baseline at 90 min. Immediately after G-CSF injection, neutrophil counts in venous blood were higher than those in arterial blood at 0.5 and 1 min, which was a significant difference at 0.5 min (Figure 3, P < 0.05). At later time points, no significant differences between arterial and venous samples were observed. In contrast, circulating lymphocyte counts were unchanged after intravenous injection of G-CSF (Figure 4).

View larger version (18K): [in this window] [in a new window]   Figure 1.   Changes in the circulating WBC counts in arterial and venous blood after intravenous injection of rhG-CSF (25 µg/ 4 ml saline). Intravenous injection of rhG-CSF (25 µg/ 4 ml saline) caused the circulating WBC counts to decrease at 2 min in both the arterial and venous blood. This decrease continued to 90 min. The data are expressed as percentages of baseline value.

View larger version (15K): [in this window] [in a new window]   Figure 2.   Changes in the circulating neutrophil counts in arterial and venous blood after intravenous injection of rhG-CSF (25 µg/ 4 ml saline). Injection of G-CSF caused a sudden neutropenia, with a gradual increase to 30% of baseline at 90 min. There was no significant difference in the circulating neutrophils between arterial and venous samples at any time point between 2 and 90 min. The data are expressed as percentages of baseline value.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Changes in the circulating neutrophil counts in arterial and venous blood immediately after intravenous injection of rhG-CSF (25 µg/ 4 ml saline). Neutrophil counts in venous blood were higher than those in arterial blood at 0.5, 1, and 2 min, and this was significant at 0.5 min (P < 0.05). The data are expressed as percentages of baseline value.

View larger version (17K): [in this window] [in a new window]   Figure 4.   Changes in the circulating lymphocyte counts in arterial and venous blood after intravenous injection of rhG-CSF (25 µg/ 4 ml saline). Circulating lymphocyte counts were unchanged in both the arterial and venous blood after G-CSF injection. The data are expressed as percentages of baseline value.

Effect of rhG-CSF Injection on Pulmonary Vascular Permeability

The cell and protein concentration in the BALF and the wet/dry lung weight ratio at 90 min are shown in Table 1 and Figure 5. The total cell counts in the saline-treated control group were 2.1 ± 0.9 × 10⁵/ml and those in the G-CSF-treated group were 1.6 ± 0.4 × 10⁵/ml. The fraction of each leukocyte in the BALF was unchanged after G-CSF administration. In particular, no increase in the number of neutrophils was observed after G-CSF injection (Figure 5). The albumin concentrations in the ELF (corrected by urea N) in saline-treated animals and G-CSF-treated animals were 32.7 ± 11.4 mg/ml and 21.2 ± 5.0 mg/ ml, respectively. The wet/dry lung weight ratios in the control and G-CSF groups were 5.09 ± 0.19 and 5.07 ± 0.37, respectively. There was no significant difference in either the albumin concentration or the wet/dry ratio between the groups.

View larger version (13K): [in this window] [in a new window]   Figure 5.   Leukocyte subtypes in BALF in the saline control group and the G-CSF group at 90 min after intravenous injection of rhG-CSF (25 µg/ 4 ml saline). There was no increase in the number of neutrophils after the injection of G-CSF.

Serum Level of G-CSF

The serum level of G-CSF and neutrophil counts in arterial blood are compared in Figure 6. The concentration of G-CSF reached over 300 ng/ml serum within 1 min, then gradually decreased to 100 ng/ml at 180 min. Circulating neutrophil counts in arterial blood mirrored the serum level of G-CSF, increasing as the G-CSF levels declined, and they returned to the baseline value at 180 min.

View larger version (18K): [in this window] [in a new window]   Figure 6.   Serum level of G-CSF and circulating neutrophil counts in arterial blood after intravenous injection of G-CSF. Circulating neutrophil counts in arterial blood mirrored the serum level of G-CSF, decreasing immediately after the injection of G-CSF, then gradually increased as the G-CSF levels decreased.

Morphology of the Lung

The histology of the lung was evaluated at 30 min after G-CSF or saline injection. Light microscopic observation revealed no obvious morphologic changes after G-CSF treatment. High-power light microscopic observations revealed that neutrophils were sequestered within the microvasculature in the lung following injection of G-CSF. We quantitated the number of neutrophils in pulmonary capillaries and small vessels by calculating the concentration of neutrophils per 1,000 RBC (Table 2). G-CSF induced a significant increase in neutrophil numbers, which measured 2.5-fold in capillaries and 3.9-fold in small vessels.

Effect of G-CSF on Neutrophil Deformability and the Role of the Cytoskeleton

Neutrophil deformability was examined to determine the effect of G-CSF (250 ng/ml) and the role of the cytoskeleton (Figure 7). The peak pressure in the G-CSF-treated group was significantly higher than that in the control group (27.5 ± 2.1 cm H₂O versus 21.5 ± 1.9 cm H₂O, P < 0.05). The peak pressure after pretreatment with colchicine (50 µM) was 28.1 ± 1.7 cm H₂O, not significantly different from G-CSF-treated neutrophils with no exposure to colchicine. DMSO, the solvent for cytochalasin B, also did not alter G-CSF-induced decreases in deformability (27.0 ± 1.3 cm H₂O). On the other hand, pretreatment of cytochalasin B (10 µM) completely prevented the increase in peak pressure induced by G-CSF (P < 0.001).

View larger version (17K): [in this window] [in a new window]   Figure 7.   Neutrophil deformability as defined by peak pressure across the filter. The peak pressure was significantly increased after treatment with G-CSF (P < 0.05). Pretreatment with colchicine and DMSO had no measurable effect on G-CSF-induced increases in peak pressure. Pretreatment with cytochalasin B completely prevented the increase in peak pressure induced by G-CSF. P < 0.05 between * and **;  P < 0.01 between ** and ***.

Expression of CD11/CD18 Complex on Neutrophils

The surface expression of CD11/CD18 heterodimer was measured using flow cytometry to determine the effect of high (250 ng/ml) and low (25 ng/ml) concentrations of G-CSF. The expression of CD11a was unchanged after G-CSF treatment (Figure 8A). The expression of CD11b on neutrophils was upregulated by both high and low concentrations of G-CSF. There was a 30% increase at 5 min and a 40% increase at 30 and 60 min, and no difference was found between high and low concentrations (Figure 8B). Similar increases were seen in the expression of CD11c and CD18 (Figures 8C and 8D).

View larger version (10K): [in this window] [in a new window]   View larger version (11K): [in this window] [in a new window]   View larger version (10K): [in this window] [in a new window]   View larger version (10K): [in this window] [in a new window]   Figure 8.   The surface expression of CD11a, CD11b, CD11c, and CD18 after treatment with high (H; 250 ng/ml) and low (L; 25 ng/ml) concentrations of G-CSF and saline (cont). The data are expressed as the percent increase from the baseline value. The expression of CD11a was not increased after G-CSF treatment (A). The expression of CD11b, CD11c, and CD18 increased after G-CSF treatment between 5 and 60 min without differences between high and low concentrations of G-CSF (B, C, D).

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

This study clearly demonstrates that intravenous injection of rhG-CSF caused a rapid neutropenia and neutrophil sequestration within the microvasculature of the lungs in rabbits. This phenomenon is asssociated with a decrease in the deformability of neutrophils due to changes in the actin filaments, as well as the upregulation of CD11/CD18 expression. However, injection of G-CSF did not induce neutrophil emigration or albumin leakage into alveolar space. Wet/dry lung weight ratios and light microscopic observation showed that G-CSF-induced neutrophil sequestration did not induce a massive lung injury.

A single injection of rhG-CSF immediately caused a decrease in circulating neutrophils in rabbits. This early effect of G-CSF injection has been reported (10), although the details in the time course immediately after G-CSF injection have not been examined. Our data showed that intravenous injection of G-CSF caused a sudden decrease in circulating neutrophils by 2 min, in both arterial and venous blood samples. At 0.5 min, the earliest time point examined, there was a significant difference between arterial and venous samples. The number of neutrophils was higher in the venous than in the arterial blood, indicating that the neutrophils rapidly sequestered in the microvasculature of the lung after G-CSF administration. This neutropenia continued for 30 min. Then, circulating neutrophils gradually increased to 30% of the baseline value in venous blood at 90 min and to baseline value after 180 min. The time course of this recovery corresponded to the time course of the decrease of G-CSF from the blood.

G-CSF induces activation of neutrophils by binding to specific receptors on their surface (29). Evidence of this activation has been identified by Katoh and coworkers (11), who found that the rapid fall in circulating neutrophils following rhG-CSF administration was accompanied by increased expression of neutrophil C3bi receptors, another function of the CD11b/CD18 heterodimer. Circulating lymphocyte counts were not influenced by G-CSF injection. In contrast, infusion of zymosan-activated plasma, a source of C5a, caused a less rapid and less profound fall in circulating lymphocyte counts compared with neutrophil counts (24). These findings correspond with the expression of G-CSF receptors and action of G-CSF, which is limited to neutrophils and neutrophil progenitor cells (30).

In order to begin to examine the mechanisms important in G-CSF-induced sequestration, our studies examined changes in neutrophil deformability, as well as in the expression of neutrophil adhesion molecules. Neutrophil deformability was examined by measuring the pressure generated by the passage of neutrophils through a polycarbonate filter with a uniform pore diameter of 5 µm 2 min after treatment with G-CSF or saline. The peak pressure in G-CSF-treated neutrophils was significantly higher than that in saline-treated neutrophils, suggesting that G-CSF treatment caused a rapid stiffening of neutrophils and that more pressure was needed to pass neutrophils through the microfilter. These stiffened neutrophils are likely to have greater difficulties in traveling through the pulmonary capillaries, many of which are narrower than the spherical diameter of neutrophils (31), resulting in sudden sequestration of neutrophils and neutropenia. The role of the cytoskeleton in this change was examined. Inhibition of microtubule assembly by colchicine did not alter the G-CSF-induced decrease in neutrophil deformability. In contrast, inhibition of F-actin formation by cytochalasin B markedly reduced this stiffening. These data indicate that rhG-CSF induces a rapid decrease in neutrophil deformability, which is mediated through the polymerization of F-actin but not through changes in microtubule assembly. The peak pressure after the treatment with cytochalasin B was smaller than that in the control group, confirming previous studies suggesting that F-actin formation is essential both in normal neutrophil deformability and stimulus-induced decreases in neutrophil deformability (24). Similar effects on neutrophil deformability have been observed when C5a, formyl-methionyl-leucyl-phenylalanine, and interleukin-8 are the stimuli, suggesting that this mechanism of sequestering neutrophils is common to many stimuli that act through different G-protein-linked receptors (20, 27, 32).

The effects of low and high concentrations of rhG-CSF on the expression of CD11/CD18, the leukocyte adhesion complex, were studied using human blood because of a lack of antibodies of rabbits. The expression of labelled CD11b, CD11c, and CD18 increased after activation by G-CSF, but CD11a did not. Others have reported that CD11a is not a useful marker of neutrophil activation, as there is no granular pool of this molecule (25). The level of expression of CD11b (Mac-1) increased 30% by 5 min and 40% by 30 min, and lasted until at least 60 min after either dose of G-CSF. CD11c and CD18 expression showed similar changes after G-CSF treatment. This prolonged increase in expression of neutrophil integrins might be important in CD18-dependent adhesion that is needed to maintain the accumulation of neutrophils within the pulmonary microvasculature (33). Studies in which rabbits were pretreated with the blocking anti-CD18 antibody 60.3 showed that CD11/CD18 did not mediate the initial neutropenia induced by intravascular infusion of complement fragments, but was required for neutrophils to remain sequestered throughout the duration of the infusion (33). The upregulation of CD11b/CD18 and CD11c/CD18 observed in our study may be simply a marker of neutrophil activation, and does not necessarily imply that functional increases in neutrophil-endothelial cell adhesion occurred. Considering both, however, our working hypothesis is that G-CSF-induced decreases in neutrophil deformability mediate the initial sequestration of neutrophils within the pulmonary microvasculature at 0.5 min, whereas adhesive interactions between neutrophils and endothelial cells mediate prolonged sequestration.

Pulmonary vascular permeability and histopathologic changes in the lung after rhG-CSF administration were investigated. Because G-CSF influences various biologic activities of neutrophils, and activated neutrophils are thought to play a central role in the pathogenesis of microvascular injury in the lung, several studies have focused on understanding whether rhG-CSF potentiates acute lung injury, such as the capillary leak syndrome. It has been reported that G-CSF does not enhance endotoxin-induced lung injury in guinea pigs (34) and sheep (35); however, G-CSF potentiates alpha-naphthylthiourea- and HCl-induced lung injury in rats (36). Recently, treatment with rhG-CSF has been associated with pulmonary toxicity in a variety of clinical settings (13, 36). White and Cebon reported that transient hypoxemia occurred during neutrophil recovery in febrile patients receiving oncologic chemotherapy (13). Schilero and colleagues described arterial desaturation after successive injections of G-CSF in a patient with ARDS and antibiotic-induced neutropenia (37). Our study showed that intravenous injection of rhG-CSF did not induce either neutrophil emigration or albumin leakage into alveolar spaces. These findings suggest that neutrophil sequestration within pulmonary capillaries and small vessels induced by the dosage of G-CSF used in this study does not produce a massive lung injury in rabbits, although this dose of G-CSF is greater than that given to humans when calculated per kilogram. Although a single injection of rhG-CSF alone did not increase pulmonary vascular permeability, G-CSF is reported to have priming effects on the production of superoxide anion by neutrophils and other proinflammatory effects. These observations suggest that G-CSF may, in certain disease states where numerous inflammatory mediators are present, enhance the inflammatory cascade through effects on neutrophil function.

Because this study focused on the acute effect of G-CSF administration, cell proliferation in the bone marrow or neutrophil release from the bone marrow are thought to be little involved (38). In clinical settings, repeated administration of rhG-CSF may provide a greater supply of neutrophils from the bone marrow by shortening the period of cell maturation and margination there. Activated neutrophils, including immature forms, might be sequestered into the lung. The ability of repeated administration of G-CSF to exacerbate acute lung injury has not yet been fully evaluated.