Thirty-six patients with PCD (range 2–26 years, average age 13 years; for details, see Table 1) were enrolled between 2012 and 2016 at the university hospitals of KU Leuven. Patients were only included when they were clinically stable (no change in cough or sputum, no fever, no change in therapy for a period of at least 2 weeks, change in FEV₁ < 10% since the last measurement). The study protocol (S57236[ML11095]) was approved by the ethical committee of the university. All patients underwent a combination of diagnostic testing: all had abnormal ciliary activity after cell culture of nasal biopsies. Electron microscopy of the cilia showed outer dynein arm deficiency (Dynein) in 16, microtubular disarrangement with inner dynein deficiency and central pair abnormalities (CP) in 1, absence of the central pair in 2, ciliary aplasia in 1, and normal ultrastructure (NU) in 16 patients. Genetic analysis confirmed disease causing mutations in 24 patients, did not provide evidence for mutations in seven patients and was not performed in five patients.

Because neutrophil migration needs to be performed with freshly isolated cells and because maximally two patients could be evaluated on the same day, we included in each experiment PMN from a healthy adult control. This study, thus, included two control groups: a pediatric control group (n = 21; range 4–18 years, average age 10 years) that corresponded best to the enrolled patients and an adult control group (n = 19; range 23–41 years; average age 29) because it was impossible to organize blood donations of healthy children and children with PCD on the same day. Healthy pediatric controls were included as there are no available data on pediatric values of neutrophil function and expression of receptors in function of age. The healthy controls (or their parents) all signed informed consent and their blood samples were processed identically to those of the patients (same transport time, conditions, and handling time).

Blood samples were collected in EDTA⁺ tubes. Peripheral blood granulocytes and peripheral blood mononuclear cells (PBMCs) were separated and isolated from whole blood by density gradient centrifugation as described (11). First, the blood sample was diluted with D-PBS (Lonza, Belgium) and gently layered on Ficoll-sodium diatrizoate (Lymfoprep, Axis-Shield PoC AS, Oslo, Norway). After centrifugation (400 × g, 30 min, 20°C, without break), PBMCs and plasma were separated (top layers) from the granulocytes and red blood cells (bottom layer). After collection, PBMCs were washed twice, counted in a hemocytometer and were then ready for use. The red pellet of neutrophils and erythrocytes was gently mixed with a starch solution (6% Dextran; Sigma) and the mixture was incubated for 30 min at 37°C to cause aggregation and sedimentation of the red blood cells. After a wash step with D-PBS, the residual red blood cells in the granulocyte preparation were removed by performing a hypotonic shock. After two additional wash steps, the neutrophilic granulocytes (>95% of the total granulocytic fraction) were counted in a hemocytometer and were then ready for use.

Neutrophil migration toward CXCL5, CXCL8, LTB4, and C5a was measured in a 48-well micro chemotaxis chamber (Neuro Probe, Gaithersburg, MD, USA) as described (12). LTB4 and CXCL8 (72 AA) were bought from Peprotech (Rocky Hill, NY, USA), complement component C5a from Sigma (St. Louis, MO, USA) and CXCL5 from R&D Systems (Abingdon, UK). In every assay, a reference adult control (healthy member of the laboratory staff) was included in order to allow normalization. Migration of the PCD PMN was expressed relative to migration of the reference adult control. The same approach (i.e., standardization to the healthy adult control) was used to study the migration of PMN from healthy children. Before we tested the migration of neutrophils from patients with PCD, we first tested dose–response curves of healthy neutrophils to determine the optimal concentration of chemoattractant to be tested with patient cells.

Flow cytometry was used to compare the expression of chemoattractant receptors on PMNs of patients with PCD and healthy controls. PMNs {3 × 10⁵ cells, diluted in FACS buffer [D-PBS + 2% Fetal Calf Serum (FCS)]} were stained with the following monoclonal antibodies: unlabeled anti-CXCR1, anti-CXCR2, and anti-C5aR; PE-labeled anti-CD16 and anti-BLT1. The cells were incubated with the antibodies for 30 min and afterward washed three times with FACS buffer. Cells incubated with unlabeled antibodies were subsequently stained with goat anti-mouse PE-labeled antibody and again incubated for 30 min on ice. After three additional washing steps, the cells were fixed with FACS buffer containing 0.4% paraformaldehyde. Cell suspensions were applied to a FACSCalibur flow cytometer (BD Biosciences) and the results were analyzed by CellQuest software (BD Biosciences). The FSC/SSC gate settings for PMN were verified with anti-CD16 antibodies.

The morphological shape changes that PMNs rapidly undergo when stimulated with chemotactic stimuli were examined as described (12). 50 µl of shape change buffer (HBSS without calcium and magnesium; supplemented with 10 mM HEPES) or 50 µl in shape change buffer diluted chemokine was added in duplicate to a 96-well plate. Shape change buffer was used as negative control. After adding the PMNs to the plate (3 × 10⁴ cells/50 μl), the cells were fixed with 100 µl/well 4% paraformaldehyde in shape change buffer at time points 0 and 1 min. For each condition, at least 200 cells/stimulus were microscopically (200× magnification) evaluated by an additional independent researcher who was blinded for the experimental conditions. We calculated the percentage of non-activated (perfectly round) cells and activated cells (showing cellular extensions and an irregular cell shape). Before we tested the activation of neutrophils from patients with PCD, we first tested dose–response curves of healthy neutrophils to determine the optimal concentration of CXCL5 and CXCL8 to be tested with patient cells.

Via time-lapse microscopy, the μ-slide chemotaxis assay allows to study directionality, velocity, and total distance covered during the migration of PMNs toward a concentration gradient of chemokines (13). A microfluidic chamber (μ-slide VI, IBIDI, München, Germany) was used to create a stable concentration gradient of CXCL8. PMNs of patients with PCD and adult controls (3 × 10⁶ cells/ml) were suspended in RPMI 1640 + 2 mM HEPES + 0.5% HSA (μ-slide chemotaxis buffer) and after injection of the cells in the channel, the microfluidic chamber was incubated at 37°C for 30 min to allow the PMNs to settle down. Perpendicular on the channel with the cells, a concentration gradient of CXCL8 (200 ng/ml in μ-slide chemotaxis buffer) was created. Every 90 s, a snapshot of the cells was made with an inverted microscope (10× phase-contrast objective; Zeiss Axiovert 200 M) for 2 h. Constant temperature (37°C) and CO₂ concentration (5%) were maintained throughout the recording. Migration of 20 randomly picked PMNs of each donor was tracked with the ImageJ manual tracking plug-in and data were analyzed with the IBIDI chemotaxis and migration tool. The optimal concentration of CXCL8 was determined in pilot experiments with healthy neutrophils.

Freshly isolated PBMCs (containing both lymphocytes and monocytes) were diluted in induction medium (2 × 10⁶ c/ml; RPMI 1640 + 2% FBS + 0.01% gentamycin) and seeded in 48-well plates. Cells were stimulated with 500 ng/ml lipopolysaccharide (LPS), 10 µg/ml peptidoglycan (PGN) or 100 ng/ml recombinant human IL-1β at 37°C and 5% CO₂. After 24 h, the cell supernatants were collected and stored at −20°C. CXCL8 (14) and IL-1β (R&D Systems) concentrations in the cell supernatants, and known to be mainly produced by the monocytes, were determined by ELISA (detection limit 10 pg/ml CXCL8 and 5 pg/ml IL-1β). The IL-1β Duoset ELISA principally measures the active cytokine and is only marginally cross-reactive with pro-IL-1β, according to the manufacturer.

Normal distribution of the data was verified by the D’Agostino & Pearson normality test. Since the results were not normally distributed, non-parametric statistical tests were performed. First, non-parametric one-way ANOVA (Kruskal–Wallis test) was performed and afterward pairwise comparisons (Mann–Whitney U test) were performed to detect statistical differences between two groups using GraphPad software (GraphPad Software Inc., La Jolla, CA, USA). Significant differences detected by the Mann–Whitney U test are indicated on the figures and in the text. The chi-square test was applied to test whether receptor expression levels were more often reduced in patients compared to controls. Finally, Pearson correlation analysis was executed to assess a possible correlation between CXCR2 expression levels or CXCL8 response and lung function. A p-value <0.05 was regarded statistically significant.

Increased CXCL8 levels were previously detected in PCD sputum (3, 15), but migration of PCD PMN toward CXC chemokines has not yet been investigated. Although PMN from some patients with PCD displayed increased migratory capacity in the Neuro Probe microchamber chemotaxis assay, overall PCD PMN showed reduced migration toward CXCL5 and CXCL8 (Figures 1A,B, respectively) compared to healthy PMN (p = 0.0011 and p = 0.0003, respectively), whereas migration toward LTB4 and C5a was not altered (Figures 1C,D, respectively). Non-normalized chemotaxis results for CXCL5 and C5a are shown in Figure S1 in Supplementary Material. In order to detect whether reduced chemokine-induced migration only occurred in a subset of patients, chemotactic responses to CXCL8 and CXCR2 expression levels were displayed according to the structural cilium characteristics (NU, versus Dynein, versus CP) (Figure 2). Figure 2A shows that decreased migration to CXCL8 is observed in all subgroups and that the patients with an abnormal central pair show apparently less variability. Our study group included three such patients and PMN from one patient were analyzed on three different dates in the Neuro Probe chemotaxis assay and by flow cytometry. The results for this patient were congruent for the three testing dates.

Expression levels of CXCR1 (CXCL8 receptor) and CXCR2 (CXCL5 and CXCL8 receptor) on PCD PMN and Ad CO, measured by flow cytometry, did not significantly differ (Figures 3A,B); CXCR2 levels on PCD PMN only tended to be lower (median relative expression level 72%; p = 0.3383 PCD versus Ad CO). CXCR2 was downregulated in about 65% of the PCD patients. We performed a chi-square test to evaluate whether the proportion of individuals with a reduced CXCR2 expression differed between the PCD patients, on the one hand, and the healthy Ad CO or Ped CO, on the other hand. We defined three categories: <90, 90–110, and >110% CXCR2 expression level (relative to the reference Ad CO) and tested whether PCD patients and Ad COs were differently distributed (p = 0.11, chi-square test). Non-normalized CXCR2 flow cytometry profiles of neutrophils are shown in Figure S2 in Supplementary Material. Expression levels of the C5a and LTB4 receptors, BLT1 and C5aR, were not statistically different (Figures 3C,D). We explored a possible correlation between CXCR2 expression and the clinical parameter forced expiratory volume in 1 s (FEV₁) and CXCR2 expression and the clinical parameter forced vital capacity (FVC) of the patients. However, we found no correlation between these lung function parameters and CXCR2 expression, but FEV₁ correlated with the chemotactic response to CXCL8 (Pearson correlation coefficient r = 0.43, p = 0.02) (Figure S3 in Supplementary Material). Finally, we did not detect a clear effect of chronic infection or drug treatment on CXCR2 expression levels or CXCL8 response (data not shown). However, there was a big variability in disease severity, type of chronic infection, and treatment between the included patients.

Subsequently, we microscopically analyzed changes in cell shape 1 min after stimulation of PCD PMN with CXCL5 or CXCL8. Figures 4A,B show the response to CXCL5 (n = 5) and CXCL8 (n = 7), respectively. Three out of five patients displayed a reduced response to CXCL5, whereas the PMN from all patients tested were less responsive to CXCL8 (p = 0.0478).

In a second round of experiments using time-lapse microscopy, we monitored during 120 min migration of PMN from patients that showed a reduced chemotactic response to CXCL8 in the Neuro Probe chemotaxis chamber. The Neuro Probe chemotaxis assay and flow cytometry (CXCR1/2 expression) were repeated in parallel with the μ-slide chemotaxis assay (Figure 5A). We observed that PCD PMN migrated more slowly and, therefore, covered shorter distances toward CXCL8 than Ad CO PMN (Figures 5B–D). The directionality of migration was comparable for control and PCD PMN (data not shown). The decreased chemotactic response was, however, not associated with defective spontaneous migration, because the mean accumulated distance in response to buffer stimulation was similar for control and PCD PMN, confirming published findings (7).

Finally, it is known that CXCL8 levels in PCD sputum are increased (3, 15). To test the hypothesis that this is due to overproduction of CXCL8 by PCD monocytes, we stimulated PCD PBMCs with several inflammatory mediators. The relative distribution of monocytes and lymphocytes in the PBMC fraction was similar in patients compared to healthy controls. On average, 9.7 ± 0.8% (mean ± SEM) CD14+ cells were present in PCD patients compared to 9.1 ± 0.6% (mean ± SEM) CD14+ cells in the Ad CO (p = 0.6933). All inducers tested stimulated production of CXCL8 by monocytes from healthy controls and patients with PCD (Figures 6A,B). Exact, non-normalized cytokine levels are shown in Table 2. The highest production of CXCL8 (402 ng/ml) was induced by LPS in PCD PBMCs. CXCL8 production after LPS treatment was significantly (p < 0.05) enhanced in PCD compared to Ped CO (Figure 6A). Also PGN provoked higher CXCL8 production by PCD monocytes compared to monocytes of Ad CO (323 and 260 ng/ml, respectively, p < 0.05), however, due to the number of individuals tested, this difference was not statistically significantly for PCD versus Ped CO (p = 0.0823) (Figure 6B). IL-1β stimulation also led to higher CXCL8 production by PCD monocytes compared to both healthy control groups (75, 25, and 15 ng/ml, respectively; not shown). Since IL-1β is an important CXCL8 inducer (16), we also measured IL-1β in the PBMC supernatants (Figures 6C,D). IL-1β production was higher in PCD compared to Ad CO upon LPS stimulation (16 and 7 pg/ml for PCD and Ad Co, respectively; p < 0.01) (Figure 6C), but not upon PGN stimulation (Figure 6D). Thus, PCD monocytes produce more CXCL8 and IL-1β in inflammatory conditions and IL-1β provoked higher CXCL8 production by PCD monocytes in turn. The increased CXCL8 production can downregulate CXCR2 expression and cause CXCR2 desensitization on PMN, providing an explanation for the reduced migration toward CXCL8. We must admit that not all patients showed reduced CXCR2 levels. However, it has been shown that in some circumstances chemokine receptors, although being upregulated, transmit migratory responses less efficiently (17–19).