Serum samples from CF patients (n = 76) were obtained at the Institute of Medical Microbiology, University Hospital Münster, Germany. They comprise two cohorts, a multicenter study (n = 62) (5, 15) as well as a two center study that was conducted in Münster (n = 14). Samples from healthy individuals were obtained from in-house volunteers (n = 46). The median age of CF patients was 14.9, 52 patients (68.5%) were male, 24 (31.5%) were female. In 44 subjects from the multicenter study the S. aureus nasal colonization status was known; 28 were S. aureus nasal carriers and 16 were non-carriers. The median age of the healthy subjects was 23, 13 (28.2%) were male and 33 (71.8%) female; 16 (34.8%) were persistent S. aureus carriers. Five CF patients (from the two center study) and five healthy volunteers additionally donated peripheral EDTA blood samples. All blood donors gave informed consent (Approvals of the responsible Ethics Committees; Greifswald: IIIUV 23/06a, BB007/17; Münster: 2007-496-f-S, 2014-054-f-S).

Recombinant Spls were generated as described (14). Lyophilized staphopain A and B were purchased from Sigma-Aldrich and reconstituted in PBS. When used in cell culture assays, the proteins were denatured at 95°C for 30 min.

Enzyme-linked immunosorbent assays (ELISAs) were performed as previously described (14). Briefly, wells of 96-well microtiter plates (MaxiSorp, Nunc) were coated with 5 μg/mL recombinant Spls (50 μL/well). Serum samples were diluted 1:5 and added in duplicate wells. The bound IgE-antibodies were detected with biotinylated rabbit anti-human IgE antibody (10 µg/mL; antibodies online) followed by Streptavidin-HRPO (1:333; Dianova). TMB substrate reagent was added for 10 min, and the reaction was stopped with 20 µl 2N sulfuric acid. The optical densities (OD) were measured at 450 nm in Infinite M200 Pro (Tecan Austria GmbH). Negative controls were processed without the addition of serum. Each assay was repeated on two separate days.

PBMCs were isolated from 30 mL of whole blood using standard gradient methods and cryopreserved until analysis. After thawing, CD14+ monocytes were isolated from PBMCs by positive selection using CD14 MicroBeads (Miltenyi Biotec 30-050-201). Untouched T cells were isolated from the CD14-negative fraction using PAN T cell isolation kits (Miltenyi Biotec 130-096-535). The purity of the isolated T cells was assessed by flow cytometry and was > 95%.

The purified T cells were co-incubated with irradiated CD14+ feeder cells at a ratio of 10:1 in RPMI medium (PAN Biotech, P04-17500) supplemented with 5% human serum (PAN Biotech, P30-2401), 100 IU/mL penicillin, 200 µg/ml streptomycin, 4 mM glutamine, 50 µM β-mercaptoethanol, 1.0 mM sodium pyruvate, 0.1 mM non-essential amino acids (Sigma, M7145-100M). Cells were seeded in 24-well flat bottom plates and stimulated with a cocktail of recombinant SplA, SplB, SplD, SplE, and SplF (5 mg/mL each). On day 5, 750 µL of the medium was replaced by fresh medium supplemented with 20 IU/mL human recombinant IL-2 (Miltenyi Biotec).

On day 9, the supernatant was taken and stored at -80°C until analysis. The cytokine concentrations in the supernatant were measured using a 13-plex cytometric bead array (LEGENDplex Human Inflammation Panel, BioLegend 740721), and cytokine concentrations were determined with the corresponding LEGENDplex software.

The T cells were harvested in PBS and stained using fluorochrome-conjugated antibodies (Supplementary Table 1A). NIR (Biolegend, 423106) was used to exclude dead cells. Data were acquired on an LSR II (BD Bioscience, San Jose, CA, USA) and FlowJo (Treestar, Ashland, OR, USA) software was used for analysis. FSC-A vs. FSC-H blots identified singlets. After gating on live T cells (NIR-CD3+), CD4+ Th cell subsets were identified by their chemokine receptor expression patterns as shown in (Supplementary Table 1B).

We analyzed 76 CF patients from two cohorts, a multi-center study (n = 62) (5, 15) and a two-center study (n = 14), as well as 46 healthy adults. All patients were persistently colonized and recurrently infected with S. aureus in their airways. We quantified Spl-specific IgE in the sera by ELISA. The antigens SplA, SplB, SplD, SplE, and SplF were tested, and we found IgE binding to all of them to be strongly elevated in CF patients compared with controls (Figure 1A). Only a minority of the healthy adults had measurable concentrations of Spl-specific serum IgE. The difference was robust and remained highly significant when the two patient cohorts were tested separately (see Supplementary Figure 1). There was pronounced variability in the patterns of each patient’s IgE binding to the five Spls. This likely reflects the patients’ history of exposure to these enzymes: Spl-specific IgE tended to be higher in CF patients that were persistently colonized with S. aureus not only in the lung but also in the nose; in the case of SplA this difference reached significance (Table 1). While all Spls are encoded in one operon that is present in around 80% of clinical S. aureus isolates, the composition of this operon is variable, indicating that in S. aureus-infected CF patients the immune system is confronted with different subsets of the Spl proteins (16).

Since protease activity is common in allergens, we next analyzed the IgE response against two other cysteine proteases of S. aureus, staphopain A (ScpA) and staphopain B (SspB). In contrast to the remarkably increased IgE response to the Spls in CF, the staphopain-specific IgE serum levels did not differ between CF patients (n = 14; subjects of the two-center study) and healthy controls (n = 46) (Figure 1B). This highlights that the ability to induce a strong IgE response is a specific property of the Spls rather than a general feature of S. aureus proteases or, in fact, staphylococcal antigens in general and corroborates earlier findings that S. aureus antigens can elicit immune responses of different quality in the same individual (14, 17). It is plausible to assume that the Spls’ proteolytic activity has a role in causing the type 2 bias of the specific antibody response. The Spls of S. aureus are known to have distinctive and very selective preferences for cleavage motifs, indicating a narrow substrate range (18, 19). However, the knowledge about the Spls’ pathophysiological substrates is very limited (20, 21). The extensive and long-term exposure of the CF patients’ airways to S. aureus drives a strong antibody response to many S. aureus antigens, documented by high specific IgG titers (5, 15). However, this pronounced humoral immune reaction to the bacteria cannot be the only reason for the sensitization to the Spls in CF, which is very selective.

Immunoglobulin class switch to IgE requires the help of antigen-specific Th2 cells. Therefore, we studied the Spl-specific T cell memory response in CF patients (n = 5) and healthy controls (n = 5) and compared cytokine production and phenotype of the Spl-stimulated T cells. We isolated and co-cultured T cells and CD14⁺ antigen-presenting cells from peripheral blood, stimulated them with a mixture of recombinant SplA, SplB, SplD, SplE, and SplF (each at 5 µg/mL) for nine days and then measured cytokines in the cell culture supernatants.

In healthy controls, type 2 cytokines (IL-4, IL-5, IL-13) were of low concentration or below the threshold of detection. In comparison, release of these cytokines was significantly increased in all cultures from CF patients. Similarly, IL-6 production was significantly higher in T cells isolated from CF patients than in those from controls, whereas IL-10 release tended to be lower in CF (Figure 2). We did not observe significant differences for IFNγ, IL-17A, IL-17F (Figure 2) nor for TNF, IL-9, IL-21 or IL-22 (not shown).

At the same time point, nine days after Spl stimulation, we assessed the phenotype of the T cells by flow cytometry and determined the proportions of CD4⁺ T cell subtypes according to their chemokine receptor expression. Focusing on changes in the T cell subtype composition due to Spl exposure, we found a stronger Th2 and Th17 cell response in CF patients, whereas Th1 cells dominated the reaction in the control individuals (Figure 3). The slightly increased percentage of Th17 cells in CF patients was not reflected in the release of IL-17 in cell culture. This is not easily explained. It may reflect the known plasticity of Th17 cells, which developed differently in CF patients than in controls (22, 23).