RESULTS
Parallel collection of IS, BAL and blood, and chest CT imaging in 2-year-olds with CF. The ability to collect IS from young children with CF was shown in prior studies but focused on a limited set of cytokines [15] and on microbiology [17]. To assess the potential of IS to yield data on an extended set of inflammatory biomarkers, we collected it in parallel to BAL from the RML and LIN, blood and chest CT, all within a single study visit. We attempted and achieved collection of all samples and data on 11 subjects (demographics inTable 1 ), except for two subjects from whom we were unable to obtain IS, thus lowering our analyzable set of IS samples to 9 (Table S1 ). Chest CT scoring confirmed that this cohort was at an early stage of airway disease, with the total disease score (%Dis) ranging from 0 to 5.9% (Table S2 ).
Similarities and differences in soluble immune mediators in IS and BAL. Prior studies of BAL have shown that onset of neutrophilic inflammation is a key determinant of structural lung damage in young children with CF [18]. Thus, we selected 11 mediators impacting neutrophil recruitment and activation (Figure 1 ). Five of those mediators showed similar levels in IS and BAL, namely CXCL1, CXCL8 (IL-8), IL-1α, IL-1RA, and IL-6. Six of those mediators showed differential levels between IS and BAL: CXCL5, G-CSF, IL-1β, and TNF-α were higher in IS than in one or both of the BAL fractions while IL-18 was lower in IS than in LIN BAL. Finally, IL-10 was below the limit of detection in most BAL samples but was measurable in most IS samples. We also assessed 9 mediators impacting recruitment and activation of monocytes/macrophages and T cells (Figure 2 ). Four of those mediators showed similar levels in IS and BAL, namely CCL2, CXCL10, M-CSF, and VEGF-A. Two of those mediators, CXCL11 and TNFSF10, showed higher levels in IS than in BAL.Finally, CCL4 and IFN-γ were below detection levels in most RML BAL samples while GM-CSF was below detection limit in most IS samples.
Correlations in soluble immune mediator levels among collected fractions. To explore potential relationships between the three airway fractions collected, we assessed correlations between IS and RML BAL, IS and LIN BAL, and RML and LIN BAL in the 20 immune mediators measured (Figure S2 ). Six neutrophil-associated mediators had significant positive correlations between IS and one or both of the BAL fractions, namely, CXCL8, G-CSF , IL1α, IL-1RA, IL-6 and TNF-α. Two monocyte/macrophage-associated mediators had significant positive correlations between IS and one or both of the BAL fractions, namely, M-CSF and VEGF-A. Between RML and LIN BAL fractions, mediators with significant positive correlations included four neutrophil-associated mediators (CXCL5, CXCL8, IL-1β, and IL-6) and three monocyte/macrophage -associated mediators (CXCL11, GM-CSF and VEGF-A).
Next, we assessed cross-correlations of the three airway fluid fractions with plasma (Figure S3 ). In all three airway samples, we observed positive correlations between CXCL8 and IL-1β and CXCL11 with TNFSF10. In IS and RML BAL, TNF-α and TNFSF10 correlated positively and G-CSF and IL-1RA correlated negatively. In IS and LIN BAL, CCL2 and IL-6 correlated positively. In RML and LIN BAL, CXCL11 correlated positively with CCL2, CXCL10, IL-1β and TNF-α and IL-1RA correlated positively with TNF-α. Plasma cross-correlations differed from those in airway samples except for positive correlations between IL-10 and IL-18 observed in both plasma and IS; IL-1RA and TNF-α observed in plasma and both BAL fractions; and CXCL8 and IL-1α as well as IL-1RA with IL-18 observed in both plasma and RML BAL.
Cellular analysis of IS yields monocyte/macrophage, T cell, and neutrophil subsets, with phenotypes similar to BAL. By flow cytometry, we gated live monocytes/macrophages, T cells and neutrophils in BAL as we showed before [11], and successfully applied the same gating strategy to IS (Figure S4 ). Relative frequencies of neutrophils and monocytes/macrophages were similar in IS and BAL fractions, with a predominance of monocytes/macrophages in both. In contrast, the frequency of T cells was significantly lower in IS compared to BAL (Figure 3A ). Neutrophils were activated in IS compared to blood based on increased surface expression of CD66b (secondary granule exocytosis marker). However, surface CD63 (primary granule exocytosis marker) was not significantly increased in IS compared to blood, likely due to the very early stage of airway disease in this 2-year-old cohort. CD16 expression was reduced on BAL, but not IS, neutrophils compared to blood (Figure 3B ). Monocytes/macrophages in IS had significantly higher surface levels of CD115 (M-CSF receptor), while surface expression of CD163 (scavenger receptor) trended lower, as compared to BAL. Monocytes/macrophages in IS also showed decreased surface expression of CD16, which is sensitive to NE-mediated cleavage (Figure 3C ).
We further analyzed neutrophil populations by determining the proportion of neutrophils demonstrating the GRIM phenotype (CD63high CD16low, as shown inFigure S4 ). The prevalence of these highly exocytic neutrophils was not significantly different between IS and BAL. Although there was substantial dynamic range among airway samples, all blood samples had <1% GRIM neutrophils. Despite the high prevalence of GRIM neutrophils in some IS samples, soluble NE (following release from the primary granules of neutrophils) was lower in IS than in BAL (Figure 4A ). Extracellular NE can be scavenged by neutrophils and monocytes/macrophages, which we assessed by surface staining with flow cytometry. In contrast to the difference with soluble NE measurement, surface NE on neutrophils and monocytes/macrophages did not differ between IS and BAL, while blood cells had lower surface NE, as expected (Figure 4B ).