Cord blood was collected after healthy full/term deliveries, and peripheral blood mononuclear cells (PBMCs) were isolated by performing density gradient centrifugation with Ficoll^® Paque Premium media. Afterwards, CD34⁺ stem cells were isolated by applying magnetic sorting with a Miltenyi Biotech CD34⁺ micro bead kit according to the manufacturer’s instructions. The purity of each batch of enriched cells was confirmed with flow cytometry. Only isolates with a purity of > 95% were used for experiments. CD34⁺ stem cells were stored in liquid nitrogen until use. Informed consent forms were obtained from all patients as required by the Declaration of Helsinki. Ethics approval was obtained from the ethics committee of the University of Bonn and the Medical University of Graz Institutional Review Board (26-520).

Thawed CD34⁺ hematopoietic stem cells were differentiated into LCs over 8 days in cell culture. We used RMPI 1640 with GlutaMAXTM, 50 µM 2-mercaptoethanol, 10% FBS and 1% penicillin-streptomycin as a differentiation medium with GM-CSF (300 IU/ml), mIgE (1 µg/ml), FLT3L (10 ng/ml), SCF (10 ng/ml), TGF-β (0.5 ng/ml) and TNF-α (20 U/ml). After 8 days of differentiation, LCs were collected and enriched with a Miltenyi Biotech CD207⁺ micro bead kit according to the manufacturer’s instructions. The purity of the generated CD207⁺ cells was confirmed with flow cytometry. In further experiments, only isolates with a purity of > 90% were used.

Antibodies used for flow cytometry are listed in Table 1. Intracellular staining was performed with a BD Cytofix/Cytoperm Kit (Beckton Dickinson, Germany) according to the manufacturer’s protocol. Experiments performed in Bonn used 7-AAD (Biolegend, USA) as viability stain and were performed on a FACSCantoTM flow cytometer. Experiments performed in Graz used Viability™ 405/520 Fixable Dye (Miltenyi Biotec, Germany) and were performed on a CytoFlex LX (Beckman Coulter, Germany). Data was analyzed with FACSDiva™ (Beckton Dickinson, Germany) or FlowJo v10 (FlowJo, USA).

RNA was isolated from harvested cells by using the TRIzol^® reagent and following the manufacturer’s protocol. The RNA concentration and purity were measured with a NanoDrop™ spectrophotometer (Thermo Fisher Scientific, MA, USA). The RNA was then transcribed to cDNA by using the SuperScript™ reverse transcriptase.

qRT-PCR was performed with the SYBR^® Green Supermix with ROXTM according to the manufacturer’s instructions. The primers used are listed in Table 2.

LCs were stained with 5 μM CellTracker™ Blue CMHC Dye in clear, serum-free media. Live SA from 1 ml stock (OD₆₀₀ = 0.5) was stained with 5 μg/ml SA polyclonal antibody and 5 μg/ml goat anti-rabbit IgG secondary antibody Alexa Fluor 546. Live SE from 1 ml stock (OD₆₀₀ = 0.5) was stained with 5 μg/ml SE monoclonal antibody and 5 μg/ml goat anti-mouse IgG secondary antibody Alexa Fluor 488. Then, 100 μl of stained live bacteria suspension were added to 5 ml of LCs. Cells were imaged with a Nikon Eclipse Ti2-E inverted microscope, taking one image every 10 min. NIS-Elements C software was used for the image analysis. The antibodies used for live imaging are listed in Table 3.

SA (DSM 20372) and SE (DSM 1798) cultures were purchased from the Leibniz Institute DSMZ (German Collection of Microorganisms and Cell Cultures GmbH) in Braunschweig Science Campus, Braunschweig-Süd, Germany. bacteria were grown on tryptic soy agar, and colonies were then inoculated into tryptic soy broth (TSB). bacteria were grown to an OD₆₀₀ of 0.6, then harvested. For experiments with inactivated bacteria, these were fixed in formalin (1.5%, 1 h, RT) and then heat-inactivated (80 °C, 10 min). For experiments with live bacteria, the overnight culture was used to inoculate fresh RT TSB at a dilution of 1:200. bacteria used for infections were harvested at an OD₆₀₀ of 0.5. Serial dilutions of the inoculum were plated onto tryptic soy agar to quantify the titre (viable bacteria).

Cell culture supernatants were collected from the experiments and sterile filtered. Cytokines were measured with LEGENDplex™ cytokine bead assays according to the manufacturer’s instructions. The following kits were used: LEGENDplex™ Human Proinflammatory Chemokine Panel 1 (13-plex), LEGENDplex™ Human Inflammation Panel 1 (13-plex) and LEGENDplex™ Human Th Cytokine Panel (12-plex).

Peripheral blood was collected from healthy volunteers. PBMCs were isolated with density gradient centrifugation as for cord blood. CD4⁺ T cells were isolated by performing magnetic sorting with a Miltenyi Biotech CD4⁺ micro bead kit (130-096-533) according to the manufacturer’s instructions. The purity of the enriched cells was assessed with flow cytometry. In further experiments, only isolations with a cell purity of ≥ 95% were used. Isolated CD4⁺ T cells were stored in liquid nitrogen until use.

Cord blood-derived LCs were primed with live SA or live SE cultures for 1 hour, then harvested, thoroughly washed and counted. 1×10⁵ viable LCs were co-cultured with 5×10⁵ T cells (LC:TC = 1:5) in 96-well round bottom tissue culture plates in RMPI 1640 10% FBS, 5% penicillin-streptomycin. Cells were harvested after five days and analyzed with flow cytometry. The cell culture medium was analyzed by performing cytokine ELISA. All experiments were performed in duplicate.

Gene expression (qRT-PCR) and cytokine measurements (ELISA) were visualized and analyzed with R [27] (v4.4.0, packages stringr, colorspace, RColorBrewer, dplyr, tidyr, readxl, pheatmap, ggplot2, ggpubs, ggbeeswarm, dendsort, nlme, emmeans, rstatix) and Tibco Spotfire (v14.4.0). Normality and scedasticity was tested with a Kolmogorov-Smirnov test with stats::ks.test() and a Bartlett test with stats::bartlett.test(), additionally applying the Benjamini-Hochberg (BH) multiple test adjustment method with stats::p.adjust(). The data distribution improved sufficiently after performing a log₁₀ transformation, and only transformed data were used for the subsequent statistical analysis. A hierarchical clustering analysis was performed centred and scaled to unit variance base::scale() over both the parameters and the samples. Dendrograms were calculated by performing standard clustering stats::hclust(dist()) (Lance-Williams dissimilarity update with complete linkage) and sorted dendsort::dendsort() according to the average distance between the subtrees at every merging point. Heatmaps were plotted using pheatmap::pheatmap(). For the univariate analysis, linear models were fitted by applying the generalized least squares method [28] with nlme::gls() on log₁₀-transformed data with BH multiple test adjustment.

All other statistical analyses were performed using GraphPad Prism V8 software (GraphPad Software). Differences between the two groups were calculated with a two-tailed Student’s t-test. Multiple groups were compared by applying a one-way analysis of variance (ANOVA), corrected with Tukey’s multiple comparison test.

In mice and humans, different types of cells have been called “LCs” or “LC-like cells”. However, there are significant differences between real LCs (embryonically derived and self-renewing), mo-LCs (monocyte-derived) and mucosal LCs (bone marrow-derived). In our study, we generated LCs from human cord-blood-derived CD34⁺ embryonic stem cells. We used a protocol previously developed by our group to generate LCs matching the phenotype found in the skin of AD patients [29]. Our LCs are TLR2⁺, FCεRI⁺ and CD83-, i.e. immature cells. The LC quality and phenotypes were controlled for each batch with flow cytometry (Suppl. material 1: fig. S1).

First, we exposed our LCs to heat-killed SA and SE overnight and analyzed LC activation with qPCR and flow cytometry (Fig. 1A). As expected, exposure to SA led to LC maturation, as indicated by the upregulation of CD83 and CCR7 and downregulation of CCR6, TLR2 and FCεRIα (Fig. 1B, C). Exposure to SE led to the same LC maturation; while slight quantitative differences were observed (e.g. SA exposure leads to the higher expression of CCR6 and lower expression of CD83), the overall trend is very much the same.

Next, we analyzed the expression of key transcription factors, receptors and cytokines with qPCR (Fig. 1D, violin plots in Suppl. material 1: fig. S2). Again, exposure to both SA and SE led to the initiation of the relevant maturation-associated genetic programs: TRLs were downregulated as well as the transcription factors PU1 and YY1, while JAK1 and JAK3 were upregulated. For all observed genes, the only significant difference was a slightly higher expression of IL-4Rα in SA-treated as compared to SE-treated LCs. All other analyzed genes showed no difference between these two groups. Thus, both heat-killed SA and SE activate LCs in the same way.

Figure 1. 

Exposure to heat-killed SA and SE causes LCs to activate and mature: A) CD34⁺ stem cell-derived LCs were exposed for 24 h to heat-killed SA or SE; LCs were harvested and analyzed with qPCR; B) violin plots depict the distribution of cytokine cDNA levels. Each dot represents an individual biological replicate (n = 6). Significant differences between groups were determined by linear mixed-effects models with generalized least squares, followed by the application of a Benjamini-Hochberg correction for multiple comparisons. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001; C) flow cytometry; D) expression levels were normalized to actin. One-way ANOVA, dots represent biological replicates, n = 7–9.

Next, we collected the supernatant of our cell cultures and analyzed the secreted cytokines by performing multiplex ELISA (Fig. 2A, B). While the gene expression and expression of surface markers hardly differed between the SA and SE groups, significant differences were observed in the secreted cytokines. Exposure to both SA and SE promoted the secretion of pro-inflammatory cytokines in LCs while this secretion was not observed in the untreated control. However, regarding almost all analyzed cytokines, SA-treated LCs (SA-LCs) secreted significantly higher amounts than SE-treated LCs (SE-LCs). Specifically, SA-treated LCs secreted more CCL11, CCL5, CXCL1, IP-10, TNFα, IL-1β, IL-12p70, IFN-α2, CXCL11, CCL2, CCL4, IL-10, CXCL9, CCL20, CXCL5 and IL-18. SE-LCs also secreted all observed cytokines, but most secreted less of these than SA-LCs. Thus, while SA induces the stronger secretion of cytokines, in principle, exposure of LCs to both SA and SE triggers the same type of reaction. The differences seen are only quantitative.

Figure 2. 

Exposure to heat-killed SA and SE causes LCs to secrete pro-inflammatory cytokines. LCs were stimulated with heat-killed SA or SE for 24 h; then cell culture supernatants were collected and secreted cytokines were analyzed via multiplex ELISA: A) heat-map and B) violin plots depict the distribution of secreted cytokine levels. Each dot represents an individual biological replicate (n = 6). Significant differences between groups were determined by linear mixed-effects models with generalized least squares, followed by the application of a Benjamini-Hochberg correction for multiple comparisons. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001, n = 6.

In vivo observations correlate high levels of SA with increased inflammation and high levels of SE with reduced inflammation. Contrary to this observation, our first set of experiments showed that LCs react the same way to SA and SE. However, these experiments were performed with heat-killed cells.

In order to better reflect the in vivo situation, we next co-cultured LCs with living SA or SE. We observed the LCs by performing live cell imaging for 4 h. Additionally, LCs were collected after 1 h for flow cytometry analysis.

LCs co-cultured with SA displayed fast, mesenchymal migration. They were also observed to actively engulf SA. LCs co-cultured with SE displayed slow, amoeboid movement and did not engulf bacteria (Fig. 3A, B). Co-culturing LCs with SA also results in rapid cell death, while exposure of LCs to SE only slightly decreases the LC viability at later timepoints (Fig. 3C). This is probably due to SA’s ability to survive phagocytosis and to kill phagocytes from within [30].

SA-LCs were strongly activated, as shown by the upregulation of HLA-DR, CD80, CD86 and CCR7. Interestingly, SE-LCs did not display an upregulation of any of these markers. LCs definitely reacted to exposure to SE, as evidenced by the upregulation of CD1b and CCR5 (also upregulated in SA to comparable levels). However, these cells did not mature (Fig. 3D and Suppl. material 1: fig. S3). Thus, live bacteria affect LCs very differently than heat-killed bacteria.

Figure 3. 

Exposure to live SA activates LCs, while exposure to live SE does not. LCs were co-cultured with live SA or SE: A) live microscopy of LCs co-cultured with SA (red); B) live microscopy of LCs co-cultured with live SE (green); C) viability of LCs co-cultured with live bacteria; D) flow cytometry analysis of LC activation after 1 h. One-way ANOVA, dots represent biological replicates, n = 4.

We next collected the supernatants from the co-cultures of LCs and living bacteria and analyzed the cytokines secreted into this supernatant after 1 h, 2 h and 3 h of exposure, respectively, with multiplex ELISA. Already after 1 h of co-culture, SA-LCs and SE-LCs show highly distinct cytokine profiles. These differences are only stronger at later timepoints (Fig. 4A, B).

Specifically, SA-LCs express more IFN-α2, IL-18, IL-1β, IP-10 and (at early timepoints only) TNFα and CCL5. In contrast, SE-LCs express (at later timepoints) more IL-10, IL-6, IL-8, CCL2 and CCL4. Cytokines that were identified but where amounts did not significantly change are shown in Suppl. material 1: fig. S4A.

Thus, LCs secrete cytokines upon exposure to both SA and SE, but the secretory profiles of these LCs are very different. Of note, LC death in the SA culture impacts the secretory profile. The observations of early (1 h) increases in cytokine levels in these cultures may be partly due to their release upon LC death. However, in AD patients where SA infiltrates the epidermis and releases exotoxins there, LC death may also be expected.

Figure 4. 

LCs co-cultured with live SA or SE show different cytokine profiles. LCs were co-cultured with living SA or SE for 1, 2 or 3 h, then supernatant was collected and analyzed with multiplex ELISA. A) heatmap and B) violin plots of secreted cytokines. One-way ANOVA, dots represent biological replicates, n = 4.

Next, we investigated the downstream effect of SA-/SE-LC co-culturing. We exposed LCs to live SA or SE for 1 h, then isolated and washed the LCs. We then co-cultured them together with T cells for five days. The flow cytometry analyses show that exposure to SA-primed LCs induced the activation and proliferation of T cells. T cells co-cultured with SE-primed LCs did not proliferate. They did however still upregulate CD25, indicating that some form of activation occurred (Fig. 5A, B).

The multiplex ELISA of the cell culture supernatant shows that SA-primed cultures contained secretions of high levels of the interleukins IL-5, IL-6, IL-9 and IL-22. SE-primed cultures show no significant upregulation of these cytokines compared to un-primed controls. SE-priming of LCs, however, induced very high secretions of IL-10, which was not found in SA-primed cultures (Fig. 5C). Cytokines that were identified but where amounts did not significantly change are shown in Suppl. material 1: fig. S4B.

Of note, patients suffering from AD display increased serum levels of IL-5 and IL-6 [31]. T cells producing IL-22 are key players in the pathophysiology of AD, directly causing epidermal hyperproliferation [32]. Thus, LCs primed with SA display a cytokine profile similar to that found in LCs of AD-patients, while LCs primed with SE do not.

Figure 5. 

SA-primed LCs cause T-cell proliferation and secretion of cytokines. LCs were exposed to SA or SE for 1 h, then washed and co-cultured with T cells for 5 d. A, B) flow cytometry measurements of activation and proliferation of T cells; C) multiplex ELISA measurements of secreted cytokines. One-way ANOVA, dots represent biological replicates, n = 4.

The authors have declared that no competing interests exist.

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of the University Bonn (Nr. 204/09) and the Medical University of Graz Institutional Review Board (26-520).

The authors declared that no clinical trials were used in the present study.

The authors declared that no experiments on humans or human tissues were performed for the present study.

The authors declared that no informed consent was obtained from the humans, donors or donors’ representatives participating in the study.

The authors declared that no experiments on animals were performed for the present study.

The authors declared that no commercially available immortalised human and animal cell lines were used in the present study.

PW was supported by the Austrian Science Fund (FWF - W1241). YP was supported by the China Scholarship Council and received research support from the company Shanghai Biocelline Enterprise Co. Ltd, China. The project was supported by the Christine Kühne Center for Allergy Research and Education, Davos, Switzerland.

YP: Conceptualization; Funding acquisition; Material preparation; Planning and performing the experiments; Data collection and analysis; Writing original draft. MH: Planning and performing the experiments; Data analysis; Writing and editing. NB: Data analysis; Software; Visualization. MA: Live imaging. HS: Collect and prepare human stem cells. TB: Conceptualization; Funding acquisition; Supervision; Review and editing. PW: Conceptualization; Funding acquisition; Supervision; Review and editing.

Yi Pan https://orcid.org/0000-0001-6723-1370

Mathias Hochgerner https://orcid.org/0000-0003-4987-4393

Natalie Bordag https://orcid.org/0000-0002-9505-6271

Herbert Strobl https://orcid.org/0000-0002-8070-4977

Thomas Bieber https://orcid.org/0000-0002-8800-3817

Peter Wolf https://orcid.org/0000-0001-7777-9444

All of the data that support the findings of this study are available in the main text or Supplementary Information.