A hampered chemoattractant-induced cytoskeletal rearrangement in granulocytes of patients with unexplained severe chronic and relapsing infections of the upper and lower airways. In vitro restoration by G-CSF exposure

Abstract

Granulocytes play a major role in host defense against bacterial infections. Severe inborn defects in granulocyte function are associated with fulminant bacterial infections in early childhood. Subtle disturbances in granulocyte function might contribute to an enhanced susceptibility to bacterial infections in adulthood. We investigated chemoattractant (N-formyl-methionyl-leucyl-phenylalanine, fMLP and casein) induced cytoskeletal rearrangements (polarization) of blood granulocytes in 77 adults with chronic and recurrent therapy-resistant infections of the upper and lower airways. These infections could not be explained by B- and/or T-cell defects or local anatomic abnormalities. Besides polarization, chemotaxis of blood granulocytes was measured in 33 patients, as well as granulocyte superoxide production in eight patients.

The chemoattractant-induced cytoskeletal rearrangement in patient blood granulocytes was significantly lower as compared to healthy control values with both fMLP and casein as stimuli. About two-thirds of the patients showed a defective polarization response to fMLP. Granulocyte colony-stimulating factor (G-CSF) when added in vitro corrected the defective polarization responses; responses in the normal range were not enhanced. The chemotactic motility of patient blood granulocytes was also slightly, but significantly lowered. However, it did not correlate to the lowered polarization. Granulocyte superoxide production was comparable in patients and in healthy controls.

Our data thus show that subtle abnormalities in chemoattractant-induced cytoskeletal and motile function of blood granulocytes are frequent in patients with severe therapy-refractory bacterial infections of the upper and lower airways.

INTRODUCTION

Although local mucosal and anatomical abnormalities are a major reason for chronic relapsing infections of the upper and/or lower airways, the cause of these infections remains unexplained in many patients. Severe impairments in host defence, such as X-linked agammaglobulinaemia and common variable immunodeficiency are rare causes of chronic recurrent airway infections. More subtle disturbances in humoral immunity have also been suggested as a cause of at least a part of the remaining unexplained cases of chronic recurrent airway infections, e.g. deficiencies in immunoglobulin isotype, IgG subclasses or specific antibodies have been observed in patients (in most studies children) with chronic recurrent (airway) infections [1–6]. However, the role of these subtle defects of humoral immunity in the pathogenesis of chronic recurrent infections of the airways has not been completely elucidated [7]. Moreover they have only been found in a proportion (12–58%) of the patients.

Next to humoral immunity, granulocyte function is important in host defence against bacterial infections. One of the properties of granulocytes is their capability to polarize, i.e. to change shape from a round to elongated or triangular form with sometimes large cytoplasmic protrusions. Important tasks of human granulocytes are dependent on this ability to undergo impressive morphological changes: passage through small capillary vessels requires the rapid acquisition of an elongated shape [8]; emigration from blood to inflamed tissues necessitates drastic squeezing between endothelial cells [8], ingestion of foreign particles usually involves close adaption of the phagocyte to its prey [8], and displacement along a chemotactic gradient is associated with characteristic morphological patterns [9–11]. The process of cell polarization is based on a rearrangement of the interior cytoskeleton of the cell [12]. Chemokines and chemoattractants such as N-formyl-methionyl-leucyl-phenylalanine (fMLP) induce such rearrangements by binding to serpentine G-protein coupled receptors in the cytomembrane of granulocytes. This stimulates a complex cascade of 2nd messenger systems including the phosphatidyl inositol (PI) cycle and the Rho-family of GTP-ases, which results – amongst other things – in a coupling of f-actin and other cytoskeletal building blocks to the membrane of the cell [13]. In this way a submembraneous myoactin cell cortex is formed, enabling the cell to change shape [14].

In this study we investigated blood granulocyte polarization in response to two chemotactic stimuli in a total of 77 patients with unexplained chronic and relapsing infections of the upper (n = 59) and lower airways (n = 18) and found it lowered as compared to the polarization of blood granulocytes of healthy controls (n = 72). In addition to granulocyte polarization we studied the capability of patient blood granulocytes to move along a chemotactic gradient using a commercially available microchemotaxis assay with fMLP as chemoattractant. The latter assay was used in 33 randomly chosen patients from the total group of 77 patients (26 chronic purulent rhinosinusitis patients, seven chronic purulent bronchitis patients). This function was also lowered in the patients. Finally superoxide production in response to fMLP was tested, but in only a limited number of patients (n = 8), since the assay did not show any abnormalities.

We have previously observed, in preliminary experiments, that the decreased blood granulocyte polarization response found in some patients was correctable by granulocyte colony-stimulating factor (G-CSF) when added to the test in vitro. We therefore also tested the effects of in vitro G-CSF addition on granulocyte polarization in all 77 patients.

MATERIALS AND METHODS

Patient selection

Patients with chronic relapsing bacterial infections of the upper and/or lower airways to be tested in the polarization assay were selected according to the following criteria:

(a) Patients with chronic purulent rhinosinusitis

• Present symptoms of chronic sinusitis such as facial pain, nasal congestion and abnormal nasal discharge, present persistently or recurrently for at least 6 months [15].

• Absence of response to functional endoscopic sinus surgery (FESS).

• Absence of response to conservative treatments (nasal congestants, > two courses of appropriate antibiotics of at least 10 days each).

• Abnormalities at sinus CT scan in at least ethmoid and maxillary sinus [16].

• No preexisting anatomic abnormalities.

• Sinuscopy showing local or generalized mucosal swelling.

• Confirmed bacterial infection of paranasal sinuses: positive culture of discharge and/or biopsy taken during sinoscopy.

• No usage of corticosteroids, antibiotics, immunosuppressors or immunomodulators at the time of testing.

• No allergy as major cause for rhinosinusitis.

(b) Patients with chronic purulent bronchitis

• Chronic production of purulent sputum (minimally 3 months a year).

• No response to antibiotic treatment or frequent recurrence despite at least two courses of targetted antibiotic therapy.

• Minimal duration of the disease 2 years.

• Sputum culture positive.

• No usage of antibiotics at the time of testing.

• No preexisting anatomic abnormalities (e.g. lung sequester).

Excluded in both groups were also patients with metabolic defects (e.g. diabetes, cystic fibrosis) and patients with well- characterized immunodeficiency syndromes, such as a HIV infection, complement deficiency, patients with a previous diagnosis of disorders of phagocyte function (chronic granulomatous disease, Chediak–Higashi syndrome, myeloperoxidase deficiency) or severe hypo- or agammaglobulinaemia (IgG < 4 g/l). The absolute neutrophil count had to be >1·0 × 10⁹/l.

Serum levels of IgG, IgA and IgM were routinely determined by nephelometry, and IgG-subclass levels were determined with a radial immunodiffusion assay (Department of Clinical Chemistry, University Hospital Dijkzigt, Rotterdam, The Netherlands).

Patients with chronic purulent bronchitis were recruited from the Department of Pulmonology and the Department of Immunology of the University Hospital Dijkzigt, and from pulmonary departments of the St. Franciscus Ziekenhuis and the Havenziekenhuis in Rotterdam (in the period 1994–95). Patients with chronic purulent rhinosinusitis were recruited from the University Hospital Dijkzigt Rotterdam, the University Medical Centre Utrecht and the University Hospital Nijmegen (in the period 1997–2000).

On the basis of the inclusion and exclusion criteria we selected 59 patients with chronic and relapsing infections of the upper airways, 34 women and 25 men, aged 43·8 ± 10·0 years (mean ± standard deviation, s.d., range 24–69 years). The majority of patients had a long history of sinonasal complaints (mean duration 11·5 years, s.d. 10·2, range 1–42 years) and had been operated 4·5 (mean, s.d. 2·7) times. The included patients had all been given conventional treatments, such as antibiotics, nasal decongestants, FESS, frontal sinus surgery and occasionally Caldwell-Luc procedures, yet they still suffered from their disease as the treatments were unsuccessful.

In the study group of severe chronic and relapsing infections of the lower airways, 18 patients, 12 women and six men, aged 50·7 ± 17·8 years (mean ± s.d., range 23–80 years) were included. Duration of disease was 9·4 years, s.d. 6·9 years, range 1–23 years.

Sputum and nasal discharge cultures contained a plethora of microbes, mainly Staphylococcus species, predominantly Staph. aureus, H. influenzae, Strep. pneumoniae and Pseudomonas species, predominantly Ps. aeruginosa.

Total serum immunoglobulin and subclass levels were within the normal range in virtually all patients, except for slightly lowered levels of IgG and IgA in two patients and one patient with chronic purulent bronchitis, respectively. We are of the opinion that these slightly lowered levels did not explain the susceptibility of these patients for severe airway infection (in support of this notion treatment with intravenous immunoglobulin started in one had no effect).

Healthy controls consisted of laboratory staff (n = 145 blood samples, n = 72 controls, ages of controls 20–50 years, note 17 controls were used multiple times).

Isolation of granulocytes

Heparinized blood was obtained by venepuncture, immediately brought to our laboratory (< 2 h) and mixed 1 : 1 with phosphate-buffered saline (PBS) containing 0·1% bovine serum albumin (BSA, Organon Technika, Boxtel, The Netherlands). Granulocytes and erythrocytes were separated from mononuclear cells with Ficoll-Paque (1·079 g/ml; Pharmacia, Uppsala, Sweden) density gradient centrifugation. Granulocytes were obtained by lysing the erythrocytes in ice-cold NH₄Cl-buffer for 20 min. Subsequently, the cells were washed twice and resuspended in RPMI 1640 (Gibco, Breda, The Netherlands) supplemented with penicillin and streptomycin.

Polarization assay

The granulocyte polarization assay was developed analogous to the polarization assay for monocytes described by Cianciolo [17] and by this laboratory [18,19]. Granulocytes (0·2 × 10⁶) were incubated in a total volume of 0·25 ml in polypropylene tubes (Falcon, Oxford, CA) with or without a chemotactic stimulus at 37°C in a waterbath. After 15 min, cells were fixed by adding 0·25 ml ice-cold phosphate-buffered 10% (w/v) formaldehyde solution. Stimuli used were: N-formyl-methionyl-leucyl-phenylalanine (fMLP) in concentrations of 0·1, 1, 10, 100 and 1000 nm, and casein (concentrations: 0·004, 0·008, 0·016, 0·032 and 0·064 w/v). To investigate the possible effect of G-CSF on granulocyte polarization, rhuG-CSF (Roche, Basel, Switzerland) was added in a concentration of 0·2 μg/ml in samples from the bronchitis patients and from patients with rhinosinusitis in concentrations of 0·008, 0·04, 0·2, 1·0 μg/ml together with 10 nm FMLP in the polarization assay. The cells were kept at 4°C until counting in a haemocytometer. Of each tube 200 cells were counted separately by two experienced investigators using an ordinary light microscope. A cell was considered polarized if it showed multiple broadened lamellipodia and/or a triangular or elongated shape. The percentage of polarized cells was recorded. The mean value of the two values recorded blindly by the two experienced observers was taken as the final value. Although this method seems to be a subjective microscopical quantification, it has been proven to be the most sensitive and reliable polarization assay as compared to morphometric and flow cytometric methods [20]. This has also been our experience, with intra- and interassay variations hardly ever exceeding 5% polarization (n > 100). If the data were discrepant in this way (less than 2% of assays), outcomes were not used and the assay was carried out again.

Microchemotaxis assay

Granulocyte chemotaxis was tested in a commercially available microchemotaxis assay; we used the 48-well microchemotaxis chamber (Neuro Probe Inc., Cabin John, MD) as described by Falk et al. [21]. The chemotactic solution, i.e. 25 μl of 10^–7m fMLP in RPMI-0·5% BSA (sinusitis series) or in RPMI-0·5% FCS (bronchitis series) or as control RPMI-BSA or RPMI-FCS alone, was placed in the bottom wells. (FCS was used in our laboratory in the microchemotaxis assay in 1997 when we started the collection of the bronchitis series. We tested all bronchitis patients and their controls with the same badge of FCS. We later switched to BSA – a better defined supplement – in our chemotaxis assay, and all our rhinitis patients and their controls were tested with the same batch of BSA.) The upper chambers were filled with 50 μl cell suspensions containing 20 000 granulocytes. The test was always performed in quadruplicate. The wells were separated by a polycarbonate filter with 3 μm pores (Nuclepore, Pleasanton, CA). After 30 min incubation at 37°C and 5% CO₂, the chemotaxis filters were removed and nonmigrated cells were swept off using a rubber policeman. Subsequently the filters were air dried and stained with Coomassie blue. Migrated cells were counted automatically using a VIDAS-RT image-analysing computer (Kontron Elektronik GmbH, Neufarn, Germany). Granulocyte chemotaxis was calculated from the mean number of migrated cells in four wells per 1 mm² in the presence of fMLP minus the mean number of migrated cells without fMLP.

Cytochrome c reduction assay

Superoxide production was measured as the superoxide dismutase (SOD) inhibitable reduction of Cytochrome c as described by Pick and Mizel [22]. The assay was performed in 96-well tissue culture plates in an ELISA-reader (Molecular Devices, Menlo Park, CA) at 37°C. In each well 100 000 granulocytes with a stimulus and with or without SOD (Sigma) in a total volume of 200 μl were incubated for 15 min. As a stimulus phorbol myristate acetate (PMA, Sigma, 100 ng/ml), serum treated zymosan (STZ, 1·25 mg/ml) [23] and fMLP (1 μm) were used. Optical densities (OD) were measured at intervals of 18 s. at 550 nm and at 540 nm (reference wave length). From the optical densities measured, the maximum rate of superoxide production was calculated in mOD/min using Softmax (Molecular Devices) software. This value was transformed into nmol/min using the molar extinction coefficient of cytochrome c (21 × 10³m⁻¹cm⁻¹).

Statistics

Results are shown as mean ± standard deviation (s.d.) or standard error of the mean (s.e.m.). Since patient data were assumed not to show normal distributions, the Wilcoxon test was used for statistical analysis between patients and controls (Instatt computer programme). P-values < 0·05 were considered significant. Dose– and time–response curves were first checked by anova (Instatt computer programme).

RESULTS

Polarization of blood granulocytes

The percentage blood granulocytes polarizing upon stimulation with a chemoattractant for the 59 patients with severe purulent rhinosinusitis, the 18 patients with severe purulent bronchitis and the 145 healthy control samples is depicted in Fig. 1. With fMLP and casein as stimuli clear dose–response kinetics were obtained. FMLP-induced granulocyte polarization was significantly lower in both the patients with severe purulent rhinosinusitis as well as the patients with chronic infections of the lower airways as compared to the healthy control values, e.g. 39 (66%) out of the 59 rhinosinusitis patients and eight (44%) out of the 18 bronchitis patients showed maximal polarization responses below 30% of control granulocyte polarization (i.e. the average of the healthy controls minus 1 × s.d.) after stimulation with the optimal concentration of 10 nm fMLP. With 3·2% casein used as stimulus (optimal dosage) the granulocytes of the patients also showed significantly lower responses when compared to the healthy controls, but the absolute differences between the patients groups and the healthy controls were smaller than with fMLP as a stimulus (Fig. 1). It must be noted that there was no difference between healthy controls and patients in spontaneous polarization (without a stimulus) (6·5 ± 4·1%versus 5·4 ± 3·4%, n = 59, P > 0·05 in severe purulent rhinosinusitis and 9·8 ± 4·4%versus 10·0 ± 5·3%, n = 18, P > 0·5 in severe purulent bronchitis).

Fig. 1.

Chemoattractant-induced granulocyte polarization (corrected for spontaneous polarization) of patients with chronic recurrent infections of the upper (♦) and lower airways (▪) and healthy controls (•). Concentration–response curves of the responses to fMLP (horizontal axis, in nm) and dose–response curves to casein (horizontal axis, in w/v concentration) are given. anova showed statistical significant dose–response kinetics (P < 0·001 in all instances). Data points represent mean ± standard error of the mean (s.e.m.). For fMLP, statistical significant differences (Wilcoxon test, P < 0·01) between patient groups and healthy controls existed for 10 and 100 nm, for casein such statistical significant difference (P < 0·05) was found at 3·2%.

In careful retrospective analysis one of our patients with purulent bronchitis met the criteria for the hyper IgE syndrome [24]. This patient had a defective granulocyte polarization (18% polarized granulocytes at the optimal fMLP concentration) and showed a strongly elevated IgE level (3419 U/ml). Until 4 years ago, this patient had suffered from eczema. Excluding this patient from analysis had no noteworthy effects on the here described outcomes of the study. All other patients had neither severe nor minor known immunodeficiencies.

In vitro addition of G-CSF improved the defective fMLP-induced granulocyte polarization in the patient groups. The following values were found at 0·2 μg/ml G-CSF added: for purulent bronchitis 32·9 ± 13·3% polarized cells without versus 42·3 ± 9·5% with G-CSF (n = 7, P = 0·02), for purulent rhinosinusitis 26·1 ± 10·8% without versus 36·1 ± 9·9% with G-CSF (n = 59, P = 8 × 10^–7). This was mainly due to a restoring effect on the polarization of patient granulocytes if it was defective, i.e. if it was below 30% (i.e. the average of healthy controls minus 1 × s.d.). Figure 2 shows that both in rhinosinusitis and bronchitis patients with such lowered granulocyte polarization, in vitro addition of G-CSF almost completely restored polarization. In patients showing percentages of polarized granulocytes in the normal range, G-CSF had no enhancing effect on polarization (Fig. 2). G-CSF also had no effect on the percentages of polarized granulocytes in healthy controls (Fig. 2).

Fig. 2.

Granulocyte polarization in response to 10 nm fMLP in the absence (□) or presence of G-CSF in two concentrations () 0·2μg/ml; ▪ 1·0μg/ml). G-CSF effects on granulocyte polarization responses of healthy controls are compared to those of patients with a granulocyte polarization above (normal) and below (lowered) 30% granulocyte polarization. Bars represent the mean ± standard deviation (s.d.). *P < 0·05; **P < 0·005 (Wilcoxon test).

We additionally investigated whether the combined presence of rhinosinusitis and bronchitis had any effect on the found defects in fMLP-induced granulocyte polarization. All together there were 29 patients having both rhinosinusitis and bronchitis. There were no statistical significant differences between polarization values of these patients and those with rhinosinusitis or bronchitis only. There was also no relationship between other clinical parameters (gender, age, duration of disease, specific microbes, or used antibiotics) and defective polarization values.

In patients with chronic purulent bronchitis we were able to test whether steroid treatment had any effect. Granulocyte polarization was comparable in patients on prednisone treatment (n = 5) and those without (n = 13) prednisone (39·6 versus 37·1%, P > 0·5). fMLP induced optimal granulocyte polarization was also comparable in patients with normal lung function (n = 8) and patients with an obstructive lung function (n = 6, FEV1 and FEV1/FVC < 80% predicted): 37·6%versus 36·8% (P > 0·5). Patients with (n = 9) and without bronchiectasis (n = 7) also had comparable granulocyte polarization: 34·3%versus 43·0% (not significant).

Chemotaxis of blood granulocytes

Because we found the interassay variation in the granulocyte microchemotaxis assay to be considerable, the chemotactic migratory capability of the blood granulocytes of a given patient was always determined together with the granulocyte chemotaxis of a paired healthy control performed in the same assay. In total 33 of our patients (random selection) were tested in the microchemotaxis assay. Figure 3 shows the results. Also in this assay chemoattractant-induced motility disturbances were detected: we found a significantly lowered granulocyte chemotaxis towards the chemoattractant fMLP in the 26 patients with chronic purulent rhinosinusitis as compared to healthy controls. For the seven patients with chronic purulent bronchitis tested in the microchemotaxis assay, results were also significantly lower as compared to controls, i.e. a value of 202 (s.d. 97) cells/mm² membrane versus 404 (s.d. 205) cells/mm² membrane for healthy controls (n = 7) was found (P = 0·05, paired analysis) (higher values were found in this set of experiments because FCS instead of BSA was used in the assay, see Materials and Methods).

Fig. 3.

The chemotactic response to fMLP expressed as number of migrated cells/mm² membrane (mean) in the microchemotaxis assay at various timepoints (horizontal axis) of granulocytes of 26 patients with chronic recurrent infections of the upper airways (•) and healthy controls (○). anova showed statistical significant concentration response kinetics (P < 0·001) *P = 0·034 (Wilcoxon test).

It is worthy to note that we could not find a significant correlation between the impaired polarization and the reduced chemotaxis in individual patients (r = – 0·30, n = 26, P > 0·10).

Superoxide production of blood granulocytes

In earlier experiments we had established that 1 μm fMLP, 100 mg/ml PMA, and 1·25 mg/ml STZ were optimal stimuli in the superoxide production assay. No differences in optimal superoxide production were observed between healthy controls and patients with either fMLP, PMA or STZ as a stimulus (fMLP: 248 ± 150 versus 266 ± 98 pmol/min; PMA: 518 ± 266 versus 524 ± 256 pmol/min; STZ: 168 ± 28 versus 176 ± 36 pmol/min, n = 8, P > 0·5). When patients with a lowered granulocyte polarization (< 40%, five of eight patients tested) were compared to those with a normal polarization we also did not observe a difference in superoxide production.

DISCUSSION

We investigated the cytoskeletal and motile function of granulocytes in patients with unexplained chronic recurrent infections of the upper and/or lower airways. Our patient selection was based on the presence of chronic/recurrent severe refractory rhinosinusitis or bronchitis, not explained by well-defined (partial) immunodeficiencies or present anatomic or metabolic abnormalities. The bacterial infection was proven by positive bacterial culture. These patients constitute a significant clinical problem, since despite intensive and frequent therapy the infections do not resolve and many of the patients are disabled and unable to work [25].

We here demonstrate that patients having the above-described characteristics show a diminished blood granulocyte polarization towards two tested chemotactic stimuli (fMLP and casein), and that G-CSF when added in vitro restored this granulocytic defect. Patients’ blood granulocytes also had a somewhat (yet statistically significant) lower chemotactic migration potential. In a limited series of experiments patients’ granulocytes showed a normal superoxide production also when stimulated with fMLP. One of the limitations of our study is that granulocytes of nasal lavage fluids have not been studied (simultaneously), hence defective functions of the peripheral cells may not be taken as proof of a defect at the local level.

To our knowledge a defective granulocyte polarization in patients with severe infections of the upper and lower airways has not been reported in the literature before.

What is known about the connection between granulocyte polarization and granulocyte chemotaxis? Displacements along a chemotactic gradient are associated with characteristic morphological patterns of polarization [9–11], and it has been suggested that cell polarization reflects an early event in the process of chemotaxis [11,17]. However, we did not find a positive correlation between the chemotaxis disturbances and polarization defects of the granulocytes of our patients. We therefore view polarization and chemotaxis as related but not identical processes. Apparently other mechanisms – apart from cell polarization – are involved in the complex migration of cells in a chemotactic gradient. It is also clear that the polarization assay can not replace the chemotaxis assay and vice versa.

With regard to chemotactic motility disturbances in patients with severe bacterial airway infections, a reduced neutrophil chemotaxis has been reported in children with recurrent pulmonary infections, recurrent otitis media, recurrent furunculosis or recurrent episodes of fever, be it in a clear minority of patients [26,27]. In a study on 150 patients with chronically infected bronchiectases, neutrophil chemotaxis, measured in a Boyden chamber, was defective in only three of the patients [28]. In a similar study [29] including only 10 patients neutrophil chemotaxis was normal, again indicating the low prevalence of such chemotactic disturbances in patients with severe bacterial infections of the airways. Our data showing a marginally lowered granulocyte chemotaxis in patients with severe airway infections are in accordance with such view that defects might be present, but are not prominent. A reduced granulocyte chemotaxis is however, a well-recognized phenomenon in patients with severe bacterial infections of the airways when occurring in combination with burn injuries, viral infections and rare diseases such as the Wiskott–Aldrich syndrome, Leukocyte Adhesion Deficiency, the Kartagener syndrome, the Chediak Higashi syndrome and the hyperimmunoglobulin E syndrome [30–32] (see one of our patients, Results section).

With respect to the cellular mechanisms responsible for the impaired granulocyte polarization observed in our patients, it seems unlikely that these are due to a decreased number or avidity of fMLP receptors, since superoxide production in response to fMLP was normal and since polarization defects were also visible with casein as chemoattractant. Polarization of granulocytes can also be induced by compounds such as IL-8 or C5a [33] and unfortunately we have not used these compounds in this series of experiments. Chemoattractant signalling activates a plethora of molecules involved in intracellular signal transmission and cytoskeletal rearrangements (see introduction). It should be noted, however, that fMLP induces different signal transduction pathways for chemotaxis and for superoxide production in human neutrophils [33,34]. Hence, our data might be explained by a hampered signal transduction for processes directly involved in cell motility in our patients’ neutrophils. Changes in cell shape are considered to be driven by cytoplasmic cortico-cytoskeletal building blocks and motor proteins, whereas the plasma membrane behaves as a passive structure [35]. Actin filaments play a dominant role in active cell shape changes, and exposing cells to microfilament inhibitors results in major alterations of shape control [36–38]. Indeed, cytochalasin B exposure abolished the polarization response of granulocytes in our studies (data not shown). Our data thus urge for further studies on defects in intracellular signalling cascades downstream from membrane receptors.

With regard to the aetiology of the cytoskeletal and motile defects of the patient granulocytes, defects may be inborn (intrinsic) or induced by factors influencing the cell’s motility [31]. An inborn defect is less likely, since the susceptability to infection did not run in the family and since in most patients the onset of the airway infections was in adulthood (data not shown). For inborn defects, we would have expected a positive family history and a start of the infections in childhood. With regard to environmental influences, factors influencing phagocyte function can be produced by bacteria colonizing the airways, by airway epithelial cells or by immune cells themselves. Several bacteria have been shown to produce factors that affect leucocyte function, e.g. pyocyanine of Ps. aeruginosa inhibits neutrophil superoxide production [39]. The production of cytokines by airway epithelial cells or various types of immune cells may also down-regulate neutrophil function [40]. By the production of such factors, bacterial infection might cause an alteration of host defenses in the respiratory tract, predisposing to further infection, and the establishment of a vicious circle, as has been hypothesized before [41,42].

G-CSF clearly restored the polarization of granulocytes to normal when the defective granulocytes of patients were exposed to the cytokine in vitro. G-CSF has been reported to have a number of effects on neutrophils. It enhances haematopoietic neutrophil progenitor growth, neutrophil release from the bone marrow, and primes several mature neutrophil effector functions [43]. G-CSF has also been reported to co-operate in the induction of fMLP receptors on the myeloid HL60 cell line [44], to prime fMLP-induced membrane depolarization in neutrophils [45], to stimulate granulocyte migration towards fMLP [46] and to enhance fMLP receptor-mediated, but not protein kinase C- mediated, neutrophil superoxide generation [47].

Considering the importance of the restorative effect of G-CSF on the defective granulocyte polarization for host defence in vivo, studies in animal models have provided interesting data. In mice, burn injury as well as superimposed Pseudomonas sepsis caused a depressed chemotaxis. Interestingly, in vivo administration of rhG-CSF restored chemotaxis and resulted in an improved survival of the mice [48]. In addition, G-CSF and GM-CSF have been shown to reverse neutrophil dysfunction in influenza A virus infection in chinchillas [49]. In experimental C. albicans infection in the rat, G-CSF also enhanced pulmonary antifungal activity [50].

The here reported restoration of defective granulocyte polarization by G-CSF when added in vitro has urged us to treat the here reported group of patients with purulent rhinosinusitis with rhG-CSF or placebo in a double-blind fashion. Our next report will address the outcome of this treatment.