Two groups of subjects were studied: 1) a ‘pwCF group’ (CF/median age = 15, n=22) with mild disease (median % FEV1 predicted = 87, where only one patient was under modulator (Orkambi) at the time of sampling); 2) a control group consisting of young healthy subjects (median age = 29, n= 9). Patients and healthy controls were matched for sex (50% and 55% male, respectively). The genetics, infectious and ventilatory status of all pwCF are listed in Supplementary Table S1 . The Institutional review board for human studies of the APHP (Hôpital Necker, Assistance Publique des Hopitaux de Paris) approved the protocols and written consent was obtained from the subjects or their surrogates.

An average volume of 10 ml of blood samples from healthy subjects and pwCF was diluted ½ with PBS and deposited on 20 ml of Ficoll (Lymphocyte Seperation Medium, Sigma Aldrich). After 30 minutes of centrifugation at 800g without brake and at room temperature, the ring of PBMCs was recovered at the interface between the Ficoll and the serum. The cells recovered were washed in PBS, centrifuged for 10 minutes at 1500 g, then resuspended in 1 ml of PBS and counted on a Kova cell after a 1/10 dilution with trypan blue. Finally, 10 ^ 6 cells per 1 ml of 10% SVF-DMSO were placed in cryopreservation ampoules and stored at -80°C until use.

Two days before infection, bacteria (PAO1 strain) stored at -80° C were seeded on LB agar (lysogeny broth) and incubated overnight at 37° C. Bacteria were then precultured in 10 ml of liquid LB medium at 37° C, with overnight stirring. The pre-culture was then diluted (1/10^th-1/5^th) at 37° C with stirring for 1h30-2h in order to achieve exponential growth. PAO1 concentration was then determined with a spectrophotometer (densitometry at 600nm) and used to infect PBMCs (see below).

PBMCs from control patients and pwCF stored at -80° C were thawed, washed in Dulbecco′s Modified Eagle′s Medium (DMEM) and centrifuged for 10 minutes at 1500 rpm. After counting with trypan blue on Kova cells, 2.10^6 cells were cultured (comprising approximately 1.2. 10^6 lymphocytes (T cells, B cells and NK cells) and 0.8.10^6 of myeloid CD11b+ cells) in DMEM free medium, in a 24-well cell culture plate (Corning ™ 3524) and stimulated with PMA/iono or infected with PAO1 at a multiplicity of infection of 1 (MOI). After 4 hours of stimulation/infection, PBMCs were recovered and centrifuged for 10 minutes at 1500g, cells were treated with RNA lysis buffer, or used for FACS analysis (see below). Simultaneously, supernatants (diluted ½ or ¼, in duplicates) were either analysed separately by ELISA (IL-1b, TNF-a, IL-8, IFN-g, IL-13, IL-17 DuoSet R&D Diagnostics) or pooled (within each HC or CF arm of the experiment) and analysed (40 mediators) using a customised Thermofisher Luminex Human Procartaplex Mix&Match (catalog numbers: PPX-35-MXXGTEK and PPX-14-MXNKTXF), according to the manufacturer’s instructions.

PBMC cells were lysed with lysis buffer containing 1% β-Mercaptoéthanol. RNA isolation was then performed with the PureLink RNA Mini Kit (12183018A, Ambion, Life Technologies, Asnières sur Seine, France), following the manufacturer’s instructions and as described previously (26).

qPCR primers were: HPRT= house-keeping gene:R: 5’ATCCAACACTTCGTGGGGTC3’; F: 5’ TTGCTTTCCTTGGTCAGGCA3’; TNF: R:5’ TGAGGTACAGGCCCTCTGAT 3’;F:5’CCCGAGTGACAAGCCTGTAG3’;IFNg:R:5’CTGGGATGCTCTTCGACCTC3’; F:5’AGTGATGGCTGAACTGTCGC3’; IL-13: R: 5’ GTCTCTGAACCCTTGGCTCC 3’; F:5’GTCTCTGAACCCTTGGCTCC3’; IL-22:R:5’CCAAGAGGGCCAAGAGAAGG3’;F: 5’ CAGTCACCAGTTGCTCGAGT 3’;IL-17:R5’GGATCTCTTGCTGGATGGGG3’; F: 5’ATCTCCACCGCAATGAGGAC3’;IL1b:R5’TCAACACGCAGGACAGGTAC3’;F:5’ GCTTGGTGATGTCTGGTCCA3’; IL-6: R:5’GGTCAGGGGTGGTTATTGCA

3’;F:5’CCAGAGCTGTGCAGATGAGT3’;IL8:R5’TCAGCCCTCTTCAAAAACTTCTC3’;F:5’CATACTCCAAACTTTCCACCCC3’;IL-10:R:5’ GCCACCCTGATGTCTCAGTTT; F:5’ACGGCGCTGTCATCGATTT;P.a opRL:R:5’ACCGGACGCTCTTTACCATA3’;F: CGAGTACAACATGGCTCTGG

Gene expression was expressed as: dCT = CT gene of interest – CT HPRT. Because dCT values are inversely correlated with gene expression levels, the Y axis of the PRISM panels are reversed, for a more intuitive visualisation of RNA expression (i.e. higher expression: ‘up’, lower expression: ‘down’). When appropriate, RNA expression from stimulated/infected- (with either PMA/iono or PAO1) HC or CF samples was expressed compared to their unstimulated control, using the formula: fold increase =2^-(ddCT).

FACS analysis of PBMCs (isolated post Ficoll gradient, see above) was performed for the quantification of the different cell types, using either membrane (CD45, CD3, CD4, CD8, CD11, CD14, CD15, CD16, CD66, CD19, NKp44, CD7, CD127, CD56, TCRgd, CD33), or intra-cellular markers (EOMES, IFNg, IL-13, IL-17, IL-22).

Briefly, 5.10⁶ to 10.10⁶ PBMCs were used for each condition, resuspended in 1 ml PBS, and the relevant mix of antibodies added (see Supplementary Table S2 ).

For membrane receptors analysis, 4 antibody mixes were used: the 1^st ( Figure 1 ) was used to assess the PBMCs content of monocytes/CD14+, T (CD3), B lymphocytes (CD19) (all -FITC-labelled, named ‘Lin+’, panel A), as well as γδ T cells (panel B), NKT cells (panels C-E), NK CD56 Bright/dim cells (panel F) and innate lymphocytes (ILCs, panels F-K). The 2^nd was used to analyse separately T CD4+, T CD8+ and B cells ( Figure 2 ). The 3^rd and 4^th mixes independently analysed monocytes (CD11b+CD14+) and neutrophils (CD15+CD66+) ( Supplementary Figure S1 ).

For intra-cellular cytokines analysis, Golgi Plug was added prior to stimulation/infection of cells (BD Biosciences, 1/1000 dilution). Then, after 4hrs, one mix allowed for the assessment of IFN-γ, IL-13, IL-17 and IL-22 in Lin + cells ( Supplementary Figure S2A ), and in NK and ILC cells (panels B).

After incubation and/or permeabilization and washing, cells were fixed (Fix/Perm/eBioscience FoxP3 kit), centrifuged, washed (Perm Wash), and resuspended in PBS before analysis. For compensation purposes, mono-labelled tubes were used, using Ultra Comp eBeads (Invitrogen). Acquisition of events (around 20,000) were performed on an LSR-Fortessa and analysed with BD FACSDIVA™ and FlowJo (Treestar, OR) softwares.

PBMC supernatants enzymatic activities were analysed in 384 black wells plates in a final volume of 30 µL, using the procedure described (26). Briefly, neutrophil elastase (NE) activity was measured using the fluorogenic substrate Methoxysuccinyl-Ala-Ala-Pro-Val-7-amido-4-methylcoumarin, (Sigma), excitation and emission wavelength being 460 and 370 nm, respectively, while pan-metalloprotease activity was assessed with the substrate 5-FAM-Ser-Pro-Leu-Ala-Gln-Ala-ValArg-Ser-Ser-Ser-Arg-Lys(5-TAMRA)-NH2 (Enzo Life Sciences), excitation and emission wavelength being 485 and 535 nm, respectively. Fluorescence was read over a 3 h period, with a Varioskan™ Flash Multimode Reader (Thermo Scientific).

Data were analysed with GraphPad PRISM software 9.3.1. Statistical analysis was performed with either t-tests, followed by non-parametric Mann-Whitney tests when two groups were compared, or by Kruskal Wallis tests, followed by post-hoc Dunn’s multicomparison tests when several groups were studied. A dendrogam/heatmaps analysis was also performed, encompassing all 40 Luminex analytes, using the ClustVis 2.0 program (https://biit.cs.ut.ee/clustvis/?s=mnEhQJZfoKkEurv). A heatmap was then obtained, using the same program.

The cellular content of Ficoll-derived PBMCs from a population of young pwCF (see Supplementary Table S1 ) and of control young healthy patients (HC) was analysed by FACS. For both ethical and practical reasons, it was not possible to exactly age-match the two populations, with a median age of 15 (25% percentile =7.5) and 29 (25% percentile =24.0), for CF and HC subjects, respectively. To assess any potential age-related confounding factor, we correlated all measures of cell concentration (Lin+, TCRgd, iNKT, Lin-, CD56 Bright, CD56 Dim, total ILCs, ILC1, ILC2, ILCP) with the age of the HC subjects ( Supplementary Figure S3 ) and that of pwCF ( Supplementary Figure S4 ), and found no correlation, except for two variables (described in the paragraph below).

After appropriate gating of live CD45+ cells (see in Figure 1 the FACS gating strategy), we analysed CD3+ CD4+ CD19 + CD14+ markers (all FITC-labelled, namely Lin+ cells) and showed, expectedly, that these cells (mostly TCR/BCR-bearing lymphocytes and monocytes) constituted the bulk of live CD45+ cells. Interestingly, there was a trend (p=0.07) for a reduction in Lin + cells and their % related to CD45+ cells in the CF group ( Figure 3A , Supplementary Figure S5A ). This down-regulation was not due to a reduction in either γδ T cells (3B, S5B) nor NKT cells (3C, S5C), as assessed in the same mix, but could be ascribed, as analysed with an independent mix (FACS strategy described in Figure 2 ) to ‘classical’ CD3+ T cells ( Figure 3L , Supplementary Figure S5L ) and, to a lesser extent to CD14+CD66- (most likely representing monocytes, Figure 3M , Supplementary Figure S5M ). Interestingly, in CF (but not HC) PBMCs, there was a negative correlation between the % of Lin + and the age of the patients ( Supplementary Figure S4A ).

When total Lin- cells (i.e not TCR-bearing lymphocytes, nor monocytes) were considered ( Figure 3D , Supplementary Figure S5D ), CD56 dim ‘cytotoxic’ NK cells ( Figure 3E ) were, as expected, the remaining major cell population, independently of the phenotype, averaging between 10 and 15% of CD45+ cells ( Supplementary Figure S5E ), and there was no difference in NK concentration between HC and CF individuals. Interestingly, ‘inflammatory’ CD56 high NK cells ( Figure 3F , Supplementary Figure S5F ), which represent a small portion of total NKs, were increased in pwCF.

When total ILCs and their precursor ILCPs were analysed ( Figure 3G–J , Supplementary Figure S5G–K ), their proportion was relatively low (albeit higher than iNKT and Tgd cells, Figures 1B, C , Supplementary Figure S5B, C ), and again pwCF had a higher proportion of these cells, compared to the Control group ( Figure 3G , Supplementary Figure S5G ). When ILCs were subdivided into ‘ILC1’, ‘ILC2’, ‘ILC3’, ‘ILCPs’, ILC3s were completely absent in both groups (not shown). When averaging ILCs, overall the % of ILC2s ( Supplementary Figure S5I ) surpassed that of ILC1s (roughly 4 fold, 4H), and there were obvious differences between the subject groups. Indeed, the % of HC ILC1s and ILC2s represented roughly 4% and 55% of all ILCs, respectively, whereas these proportions were 12.5% and 38% in pwCF, demonstrating a shift towards more ‘inflammatory’ ILCs in CF. Despite this overall trend, as for Lin+ cells (see above), there was a negative correlation between the % of ILC1s and the age of pwCF ( Supplementary Figure S4H ), and conversely, a positive correlation for HC ILC1s with age ( Supplementary Figure S3H ).

Remarkably, ILCPs, which are ‘poised’ precursor cells able to differentiate into ILCs at local mucosal sites, albeit at low concentration in peripheral blood ( Figure 3J ), represented 25% and 38% of HC and CF ILCs, respectively, suggesting an overall increased ILC- poiesis in CF blood ( Supplementary Figure S5K ).

Finally, we assessed the concentration of neutrophils, hereby named ‘low-density neutrophils’ (LDNs), given their presence in the upper band of Ficoll gradient (where PBMCs concentrate), and showed that their concentration was clearly increased in CF PBMCs, as assessed with the 2 neutrophil markers CD15 and CD66, Figure 3L ).

Overall, this first part of our study demonstrated a clear phenotypic bias in CF PBMCs, with, quantitatively, a trend towards a reduction in the number and proportion of adaptive T cells and an increase in ‘pro-inflammatory cells’ (inflammatory CD56 Bright NK cells, ILCs, and LDNs).

After assessing with appropriate markers the numbers and % of Lin+ and Lin- populations in PBMCs of HC and CF patients (see above), we set up to stimulate these PBMCs with a non specific stimulus (PMA/ionomycin), in a 4hrs protocol which did not induce cytotoxicity (not shown). An example of the FACS strategy (for Lin+, NK, Tgd and ILC cells) is depicted in Supplementary Figure S2 . The intra-cellular levels of the cytokines IFN-g, IL-13, IL-17 and IL-22, signatures of ‘type1’, ‘type2’ and ‘type3’ responses, respectively, were then measured and quantitative results are shown in Figure 4 .

Following PMA/ionomycin stimulation, irrespective of the ‘Healthy’ or ‘CF’ phenotype, there was in Lin+ cells ( Figures 4A–E ) a gradient of cytokine expression, with the type 1 cytokine IFN-g being the most prevalent cytokine, followed by IL-13, and IL-17. By contrast, IL-22 expression was more heterogeneous and was not PMA/ionomycin-responsive (Figure 4D).

In NK CD56 dim and CD56 Bright cells, the % of IFN-g+ cells was again prevalent (Figure 4F), post stimulation, compared to other cytokines (Figures 4G, H, K, L), with IL-22 expression being intermediate (Figures 4I, M).

Interestingly, ILC cells followed a different pattern, by producing mostly IL-13 in response to PMA/iono (about 20% of ILCs, Figure 4O ), in keeping with the phenotypic characterization showing that ILC2s were prevalent among the ILCs (see Figure 3I ). By comparison, the induction of IFN-g and IL-17 was less important (about 2 and 0.5% of ILCs, respectively, see Figures 4N, P), while there was no IL-22 induction in ILCs following PMA/iono stimulation (Figure 4Q).

When HC and CF PBMCs were compared side by side, there was, at homeostasis, no obvious difference between CF and HC PBMCs. However, following PMA/ionomycin stimulation, there was a reduction in the % of CF Lin+ PBMCs expressing IFN-g and IL-13 ( Figures 4A, B ), and conversely, an increase in CF in the % of Lin+ IL-17+ (Figure 4C) and Lin+ IL-17+IL-22+ cells (Figure 4E). For Lin- cells (CD56 and ILCs), no major differences were noted between Healthy and CF PBMCs upon PMA/iono stimulation.

We then set up to assess, in an independent set of experiments, whether the behaviour of HC and CF PBMCs was similar when stimulated with a more ‘specific’ stimulus, i.e P.a live bacteria (PAO1 strain).