All cells were cultured at 37°C and 5% CO₂. HC15 cells (ATCC-CRL-1964) were purchased from LGC Standards GmbH (Teddington, U.K.). Cells were cultured at 0.25 to 1.5 x 10⁶ cells/mL in RPMI 1640 (72400054, Thermo Fischer Scientific Inc., Waltham, MA, USA) (10% FBS (F7524, Merck KGaA, Darmstadt, Germany), 1% P/S (P4333, Merck), pH 7.6-7.8) with half medium changes done every day to ensure slightly alkaline conditions. For differentiation, a full medium change was performed. Cells were centrifuged for 5 min at 300 g, washed once with PBS and resuspended at 5 x 10⁵ cells/mL in RPMI 1640 (10% FBS, 1% P/S, pH 7.6-7.8) supplemented either with 0.5 mM SB (sc-202341, Santa Cruz Biotechnology, Inc., Dallas, TX, USA) (DHC15) or with 0.5 mM SB and 10 ng/mL IL-5 (NBP2-34897, Biotechne GmbH, Wiesbaden Nordenstadt, Germany) (IHC15) for 5 days, in line with what has been described before (32, 35, 40). For proteomics measurements, 50 ng/mL IL-5 were used to differentiate IHC15 cells. During differentiation, no medium change was performed. If not indicated differently, cells were incubated at 1 x 10⁶ cells/mL for all experiments. Activation was performed in fresh medium with 10 µM phorbol-12-myristate-13-acetate (PMA, P1585, Merck) for 90 min.

Venous blood from self-proclaimed healthy volunteers was collected into tubes containing 3.2% sodium citrate (SAR-011606001, Sarstedt, AG & Co. KG).

Eos were isolated from 50 mL whole blood from healthy volunteers using a customized protocol. In brief, erythrocytes were sedimented in 10 mL batches using the Sedimentation Kit II (130-126-357, Miltenyi Biotech B.V. & Co. KG, Bergisch Gladbach, Germany). Each resulting pellet was resuspended in 200 µL MACS^® Separation Buffer (130-091-221, Miltenyi Biotech), with 80 µL Erythrocyte Depletion Microbeads from the human StraightFrom^® Whole Blood peripheral blood mononuclear cell (PBMC) Isolation Kit (130-126-359, Miltenyi Biotech) and 40 µL human CD61 Microbeads (130-051-101, Miltenyi Biotech) and incubated for 10 min at 4°C for magnetic labelling. Magnetic separation was performed as described in the human StraightFrom^® Whole Blood PBMC Isolation Kit. Eosinophils were isolated from the pooled flow throughs using the human Eosinophil Isolation Kit (130-092-010, Miltenyi Biotech). Quality control was done using flow cytometry by staining with CD45-APC/Cyanine7 (2D1, mouse IgG1, 368516, Biolegend Inc., San Diego, CA, USA), Siglec-8-PE (mouse 7C9, IgG1, 347104, Biolegend) and CD16-BV 510™ (B73.1, mouse IgG1, 360730, Biolegend) (for gating see Supplementary Figure 1A ). This protocol resulted in a recovery of up to 2.14 x 10⁵ cells/mL of whole blood (median 5.46 × 10⁴ cells/mL, 95% confidence limits: [5.03 x 10⁴ cells/mL – 7.87 x 10⁴ cells/mL], data not shown) and a median eosinophil (= CD45⁺Siglec-8⁺) purity of 97.3% [94.8% - 97.2%] ( Supplementary Figure 1B ). For proteomics, Eos were isolated using the MACSxpress Eosinophil isolation kit (130-104-446, Miltenyi Biotech) followed by a magnetic erythrocyte depletion kit (130-098-196, Miltenyi Biotech), which resulted in a recovery of up to 2.07 x 10⁴ cells/mL whole blood (median 9.40 x 10³ cells/mL [8.93 x 10² cells/mL – 2.22 x 10⁴ cells/mL]) and a median purity of 99.2% [98.5% - 99.9%] (data not shown). If not indicated differently, cells were incubated at 1 x 10⁶ cells/mL for all experiments. Activation was performed in fresh medium with 10 µM PMA for 90 min.

Neutros were isolated from 8 mL whole blood from healthy volunteers using the MACSxpress^® whole blood neutrophil isolation kit (130-104-434, Miltenyi Biotech) followed by magnetic erythrocyte depletion. The manufacturers’ instructions were followed and quality control was done using flow cytometry by staining with CD45-APC/Cyanine7, Siglec-8-PE and CD16-BV 510™. Median neutrophil (= CD45⁺CD16⁺) purity was 99.6% [99.4% – 99.9%] over all experiments (data not shown).

PBMCs were isolated from 50 mL whole blood from healthy volunteers. In brief, blood was layered over Ficoll-Paque™ Plus (17144003, Cytiva, Marlborough, MA, USA) in a 2:1 ratio and centrifuged at 800 g for 30 min without brake. Subsequently, the enriched PBMC layer was collected. Isolated PBMCs were washed with PBS prior to further use.

RNA was isolated using the mirVana miRNA Isolation Kit (AM1561, Thermo Fischer Scientific) following the manufacturers’ protocol. For lysis of 1 x 10⁶ cells, 600 µL Lysis solution and 60 µL mRNA homogenate additive were used. RNA extraction was done using 600 µL Acid-Phenol: Chloroform, pH 4.5 (with IAA, 125:24:1, AM9722, Thermo Fischer Scientific). RNA was eluted with 100 µL hot nuclease free water (95°C) and the final RNA concentration was measured on a Nanodrop 2000 (Thermo Fischer Scientific). Reverse transcription into cDNA was performed using the LunaScript^® RT SuperMix (E3010L, New England Biolabs GmbH, Ipswich, MA, USA) on a DNA-engine Thermal cycler (Bio-Rad Laboratories GmbH, Hercules, CA, USA). In a 20 µL reaction, 250 ng RNA were transcribed resulting in a final concentration of 12.5 ng/µL cDNA. For running the qPCR, the Luna^® Universal qPCR Master Mix (M3003L, New England Biolabs) was used. In each reaction, 12.5 ng cDNA template (1 µL) were employed. The primers used for target gene recognition are depicted in Table 1 . The RT-qPCR was run on a Lightcycler^® 96 (F. Hoffmann-La Roche Ltd, Basel, Switzerland) and the RNA expression level for each target gene was calculated as 2^-ΔΔCT.

For leukocyte staining, the cells obtained from the human StraightFrom^® Whole Blood PBMC Isolation Kit before isolating eosinophil with the human Eosinophil Isolation Kit were used. After washing twice with PBS, 1 x 10⁶ cells were cytospun on a glass slide for 5 min at 200 g, fixed with M-Fix^® fixation spray (1039810102, Merck) and airdried at 37°C for 5 min. Cells were stained using the Hemacolor^® staining kit (111661, Merck) according to the manufacturers’ instructions and mounted using Eukitt mounting medium (03989, Merck) Microscopy was performed on an Axioscope connected to an AxioCam ERc5s (Carl Zeiss Meditec AG, Oberkochen, Germany) with a magnification of 40 x. On each slide, 5 pictures of randomly chosen areas were taken.

Prior to antibody staining, cells were fixed with 0.5% paraformaldehyde (PFA, P6148, Merck) in PBS for 15 min at room temperature. Subsequently, cells were washed with PBS by centrifugation for 5 min at 300 g and resuspended at 1 x 10⁶ cells/mL in FACS buffer (1% FBS, 2 mM EDTA (V4231, Promega GmbH, Walldorf, Germany) in PBS). Unspecific binding was blocked by incubation with TruStain FcX™ (422301, Biolegend) for 10 min at room temperature. Cells were washed and incubated with monoclonal antibodies α-CD193 (CCR3)-PE (5E8, mouse IgG2b, 310705, Biolegend), α-CD11b- BV 510™ (ICRF44, mouse IgG1, 301333, Biolegend), α-CD63-PB™ (H5C6, mouse IgG1, 353012, Biolegend), α-CD41-PE (HIP8, mouse IgG1, 303705, Biolegend) and α-CD62P-BV 510 (AK4, mouse IgG1, 304936, Biolegend) or isotype controls PE mouse IgG2b (MPC-11, 400314, Biolegend), BV 510™ mouse IgG1 (MOPC-21, 400172, Biolegend), PB™ mouse IgG1 (MOPC-21, 400151, Biolegend) for 15 min at room temperature in the dark. After washing, cells were measured using a CytoFLEX flow cytometer (Beckman Coulter GmbH, Krefeld, Germany). Data analysis was performed using Flow Jo v10.10.0.

For immunofluorescence staining, 12-well chamber slides (81201, Ibidi GmbH, Gräfelfing, Germany) were precoated with Cell-Tak (354240, Corning Inc., Corning, NY, USA) as indicated in the manufacturers’ protocol. Subsequently, 2 x 10⁵ cells in 150 µL RPMI 1640 were added to each chamber and incubated at 37°C for 30 min. Between each of the following steps, cells were washed 3 x 3 min in PBS. The attached cells were fixed using 4% PFA in PBS, permeabilized with 0.2% Tween^® 20 (#9127.2, Carl Roth) in PBS and blocked using 2% bovine serum albumin (BSA) (8076.4, Carl Roth), 2% goat serum (ab-7481, Abcam plc., Cambridge, Great Britain), 0.01% Tween^® 20 in PBS. All antibodies and dyes were diluted in antibody diluent (0.1% BSA, 0.01% Tween^® 20 in PBS). Incubation with the primary antibodies α-EMBP (1:100, rabbit, PA5-102628, Thermo Fischer Scientific) and α-eosinophil peroxidase (EPX) (1:200, AHE-1, mouse IgG1, MAB1087, Merck) was performed overnight at 4°C, followed by incubation with the corresponding secondary antibodies (α-mouse AF488 (1:2 000, goat, A-11001, Thermo Fischer Scientific) and α-rabbit APC (1:250, goat, A10931, Thermo Fischer Scientific)) for 1 h at room temperature in the dark. Next, cells were stained with 2.5 µg/mL wheat germ agglutinin CF^® 568 (29077-1, Biotium, San Francisco, CA, USA) and 125 ng/mL DAPI (6335.1, Carl Roth). The slides were airdried, the silicone chambers were removed and the slides were mounted using Prolong™ Gold Antifade Mountant (P10144, Thermo Fischer Scientific). Stained slides were stored at 4°C until analysis on a Thunder Imager (Leica Microsystems, Wetzlar, Germany) with a magnification of 40 x. Of each chamber, 5 pictures of randomly chosen areas were taken.

Cell migration capacity was assessed using a 24-well transwell system (6.5 mm Transwell^® with 3.0 μm PC membrane insert, CLS-48EA, Corning). To the upper chamber, 3 x 10⁵ cells in 100 µL RPMI 1640 (10% FBS, 1% P/S, pH 7.6-7.8) were added. If indicated, adherence receptors on eosinophilic cells were blocked in the upper chamber as follows: CD63 was blocked using 5 µg/mL α-CD63-PE (H5C6, mouse IgG1, 353003, Biolegend). CD11a was blocked using 5 µg/mL α-CD11a (HI111, mouse IgG1, 301202, Biolegend). CD11b was blocked using 5 µg/mL α-CD11b (ICRF44, mouse IgG1, 301302, Biolegend). PSGL-1 was blocked using 50 µg/mL α-CD162 (KPL1, mouse IgG1, 328802, Biolegend). SLC44A2 was blocked using 20 µg/mL α-SLC44A2 (rabbit, LS−C750149, Vector Laboratories, Inc., Newark, CA, USA). TREM-1 was blocked using 10 µM LP17 (MedChemExpress, Monmouth Junction, NJ, USA). LP17 was incubated on cells for 2 h at 37°C and in the last 10 min of incubation, all blocking antibodies were added to their respective samples. The indicated antibody concentrations were determined by titration and the lowest blocking concentration was used. Subsequently, 100 nM N-formyl-Met-Leu-Phe (fMLP, F3506, Merck) was supplied to the lower chamber of the transwells. The cells were incubated with fMLP for 1 h at 37°C. In order to detach the migrated cells from the lower face of the membrane, 3 µM EDTA was added to the lower chamber and the cells were incubated for an additional 5 min at 37°C, before the transwells were removed. The remaining medium in the lower chambers was mixed thoroughly and the migrated cells were counted in a Neubauer improved counting chamber in triplicate for each sample or by counting events/µL in 200 µL medium using a flow cytometer. The relative number of migrated cells was determined as a ratio of migrated/total cells.

PLTs were isolated from 50 mL whole blood from healthy volunteers as described before (45). In brief, blood was centrifuged to collect platelet-rich plasma. After addition of prostaglandin I₂ and dilution with Tyrode buffer, washed platelets were collected in a second centrifugation step. The platelet pellet was resuspended in Tyrode buffer and quality control and platelet counting were performed using flow cytometry by staining with α-CD45-APC (HI30, mouse IgG1, 304012, Biolegend), α-CD41-AF488 (HIP8, mouse IgG1, 303724, Biolegend) and α-CD62P-BV 510. The median PLT (= CD41⁺) purity was 99.5% [99.2% - 99.7%] with a mean baseline activation (= CD41⁺CD62P⁺) of 0.570% [0.410% - 0.740%] (Data not shown).

PLTs were activated by incubation with 25 ng/mL thrombin receptor activator peptide 6 (TRAP, HY-P0078, MedChemExpress) at 1 x 109 cells/mL for 5 min. Cells were resuspended to 1 x 10⁶ cells/mL in 100 µL fresh medium. If indicated, adherence receptors on eosinophilic cells were blocked prior to the addition of platelets as described the transwell migration assay section. Subsequently, 2 x 10⁷ resting or activated PLTs were added without washing for a cell to PLT ratio of 1:200. After 15 min of incubation cells were stained as described in the flow cytometry section. PLT-eosinophilic cell-aggregates were identified as CD41/CD45-double positive events.

24-well plates were coated with 0.2% porcine gelatin (G6144, Merck) in Hank’s balanced salt solution (14175095, HBSS; Thermo Fischer Scientific). The gelatin was aspirated and the plates were airdried prior to addition of cells. HUVECs were cultured in Endothelial Cell Growth Medium 2 (C-22011, PromoCell GmbH, Heidelberg, Germany). For experiments, HUVECs were seeded at 1 x 10⁵ cells/well in EM2 into coated 24-well plates. In case of stimulation, 100 pM IL-4 (200-04, Thermo Fischer Scientific) and 100 pM TNFα (11343015, ImmunoTools GmbH, Friesoythe, Germany) were added during seeding. The cells were grown to confluency for 24 h at 37°C, 5% CO₂. Eos and cell line cells were prestained with 2.5 µM carboxyfluoresceinsuccinimidylester (CFSE, C34570, Thermo Fischer Scientific) in HBSS for 5 min at room temperature in the dark. After staining, cells were washed twice with HBSS, resuspended at 1 x 10⁶ cells/mL in HBSS and added to washed HUVECs at 500 µL per well. If indicated, 5 x 10⁷ PLTs were added to respective wells for a cell to PLT ratio of 1:100 and cells were incubated for 30 min at 37°C. After adherence, the supernatant was carefully removed and wells were washed twice with HBSS. Cells were fixed, permeabilized and stained with DAPI as described in the microscopy section. For microscopy, wells were filled with 500 µL PBS. Stained plates were stored at 4°C. Microscopy was performed on a Thunder Imager (Leica Microsystems, Wetzlar, Germany) with a magnification of 40 x. CFSE⁺ Cells were counted in 10 randomly chosen sections of each well. In some experiments, adherence receptors on eosinophilic cells were blocked prior to their addition to HUVEC cells as described in the transwell migration section. In these cases, the experiment was performed in 96-well plates with 2 x 10⁴ HUVECs/well and 10⁵ eosinophilic cells and the addition of 10⁷ PTLs. For microscopy, wells were filled with 100 µL PBS and cells were counted in 5 randomly chosen sections of each well.

Cells were centrifuged 300 g for 5 min at room temperature and supernatants from 1 x 10⁶ cells were collected. The following kits were used to determine the secretion levels of a total of 11 biomarkers: Custom Legendplex™ (IL-5 (740043), IL-4 (740540), IL-13 (740047), IL-2 (740934)) (1:1, Biolegend), Human IL-33 DuoSet ELISA (DY3625B, 1:1, Bio-techne), Human CCL11 (C-C chemokine 11/Eotaxin) DuoSet ELISA (DY320, 1: 1, Bio-techne), Human IL-6 DuoSet ELISA (DY206, 1: 1, Bio-techne), Human IL-8 ELISA Max™ Deluxe Set (431504, 1: 1, Biolegend), Human CCL5 (C-C chemokine 5/regulated on activation, normal T-cell expressed and secreted (RANTES)) ELISA Max™ Deluxe Set (440804, 1: 3, Biolegend), Human IL-12 (p70) ELISA Max™ Deluxe Set (431704, 1: 1, Biolegend), Human EPX ELISA Kit (NB-E11396A, 1:50 (cell line), 1:1 (Eos), Novatein Biosciences Inc., Woburn, MA, USA). All experiments were run as described in the manufacturers’ protocols with diluting the samples as indicated.

Prior to statistical analysis, outliers were removed in Graphpad prism version 10.2.3 using the ROUT method with the Q value set to 1%.

Subsequently, statistical analyses were computed in R version 4.3.3 using the package “rstatix”. Due to the small sample sizes all data was assumed to be non-normally distributed. All tests were computed in an unpaired manner. Two groups with one variable were compared using a Mann-Whitney U test. For more than two groups with one variable, Kruskal-Wallis testing followed by Dunn’s multiple comparisons with Bonferroni correction was performed. Due to the lack of a fitting non-parametric alternative testing strategy, statistical significance between more than two groups with more than one variable was determined by two-way ANOVA followed by Tukey’s multiple comparisons with Bonferroni correction. In the case of two groups with more than one variable, while all possible multiple comparisons were computed, only comparisons untreated vs. treated within one differentiation protocol (between treatments) or between differentiation protocols of one treatment (between differentiations) were depicted in the graphs. Statistical significance was assumed with a p-value ≤0.05.

For all data, effect sizes with 95% confidence intervals were computed in R using the packages “effsize” and “apaTables”. For Mann-Whitney U tests, Cliff’s delta was used, for Kruskal-Wallis testing, Eta² was calculated and for two-way ANOVA, partial Eta² was determined. Details on all significances, multiple comparisons and effect sizes with confidence intervals are depicted in Supplementary Tables 4 - 11 .

Graphs were plotted using Graphpad prism version 10.2.3. The data is depicted as boxplots with a horizontal line at the median, the 25th to 75th percentiles as hinges and whiskers extending to the lowest and highest datapoints. N depicts the number of independent experiments in each experiment, representing cells of one passage or one healthy donor, respectively.

To determine the value of HC15 cells as a model for eosinophilic research, in this study we evaluated the two most commonly used differentiation methods, namely SB = DHC15 and SB+IL-5 = IHC15 and compared the cell line to Eos, Neutros and PBMCs in a set of descriptive and functional assays (see Figure 1 for the workflow of differentiation, as well as for an overview over all conducted tests and analyses). To assess the biological differences of the differentiated cell line and Eos, we performed proteomics, followed by TOST test and found the proteomes of the IHC15 cells and Eos to be significantly similar ( Table 2 ).

We clustered the proteomes using a PCA, to analyze the data in more detail ( Figure 2A ). Generally, all cell line samples clustered together independent of their differentiation status, closer to Eos than to Neutros in the first component. A list of the 15 highest contributing proteins in component 1 is given in Supplementary Table 12 . Out of the 11 highest contributing proteins were 5 eosinophilic or neutrophilic granule proteins, namely lactotransferrin, eosinophil cationic protein, neutrophil defensin 3, azurocidin and neutrophil elastase. In addition, several adhesion proteins highly contributed to the similarities of the cell line cells to Eos. In contrast, in component 2 the cell line cells were more similar to Neutros than to Eos. Nonetheless, differentiation of the HC15 cell line induced a slight shift towards Eos in this component. Out of the top 15 contributing proteins in component 2, many were intracellular signaling molecules contributing to different generic cellular pathways ( Supplementary Table 13 ).

The observations made in the PCA analysis were also reflected in a differential protein abundance depicted in volcano blots comparing the proteomes of IHC15 cells and Eos or Neutros, respectively. These showed a higher number of DAPs between the IHC15 and Neutros proteomes than between the IHC15 and Eos proteomes ( Figures 2B, C ).

Subsequently, a GSVA was performed ranking the differential abundance and respective p-value of proteins from specific gene sets in IHC15 cells to Eos, to evaluate enrichment. The gene sets analyzed were chosen to cover a broad range of general eosinophilic functions. While the IHC15 cells did not show significant differences to Eos in gene sets representing eosinophils pathway, differentiation, chemotaxis or migration, they were significantly different to Eos in gene sets representing eosinophil-mediated immunity and activation ( Figure 2D ).

To increase the depth of our analysis further and find more specific similarities between the cell line and Eos, we assessed exclusively expressed proteins among the samples by plotting their gene IDs in Venn diagrams without considering abundance. We found that the differentiated cells generally shared more proteins with Eos than the undifferentiated cells, and IL-5 slightly increased the number of exclusively shared proteins with Eos ( Figures 2E, F , Supplementary Figure 2A ). A pathway enrichment on the exclusively shared proteins extracted from the Venn diagrams was conducted using Metascape. The goal was to identify shared pathways between differentiated cells (± IL-5) and Eos to validate and expand the results previously observed in the GSVA analysis. The common pathways shared in all cells were hemostasis, bacterial defense, immune response and programmed cell death. Myeloid cell differentiation was only found to be shared between DHC15 and Eos, but not IHC15. In contrast, pathways associated to regulation of cell morphogenesis, adhesion and migration were only found in IHC15 cells and Eos, but not in DHC15 ( Figures 2G, H ).

Lastly, to identify differences between Eos and the cell line cells, we plotted a Venn Diagram simultaneously comparing gene ID lists from HC15, DHC15, IHC15 and Eos ( Supplementary Figure 2A ). In this analysis, 131 proteins could be identified that were expressed exclusively in Eos but not in any of the cell line cells. The gene IDs from these proteins were used in pathway enrichment analysis ( Supplementary Figure 2B ). Of the pathways found as enriched from the proteins exclusively abundant in Eos, several were related to immune response and activation, further confirming the results of the GSVA indicating a difference between the cell line cells and Eos in these gene sets.

In summary, our data indicate that the cell line cells are more similar to Eos than to Neutros and the differentiated cell line cells are more similar to Eos than the undifferentiated cell line cells. Most differences of the cell line cells to Eos were found in immunity and activation pathways.

The proteomics data revealed a high similarity between IL-5-differentiated cells and Eos in overall eosinophilic and differentiation associated proteins. We aimed to evaluate this observation by investigating an array of eosinophilic features, such as morphology and the profiles of surface markers, TFs and granule proteins in detail.

According to our proteomics data, IL-5-differentiated cells showed a high similarity to Eos in regards to chemotaxis and migration. By studying aspects of cell migration individually, we aimed to get a clearer picture of how the cell line cells migrate compared to Eos.

First, we determined the mRNA expression levels of an array of important adhesion markers, such as SEPLG (encoding PSGL-1), SLC44A2, CD63, ITGAM (encoding CD11b), CD40, ITGAL (encoding CD11a) and TREM1 (46). Most adhesion markers (SELPLG ( Figure 6A ), SLC44A2 ( Figure 6B ), CD63 ( Figure 6C ) and ITGAM ( Figure 6D ) were expressed significantly higher in Eos than in the cell line cells. CD40 ( Figure 6E ) and ITGAL ( Figure 6F ) showed a trend towards higher expression in Eos, whereas TREM1 ( Figure 6G ) showed lower expression in Eos compared to the cell line cells. Moreover, differentiation induced a significant increase in the expression of SLC44A2 in the cell line cells. A non-significant trend towards increased expression in DHC15 and IHC15 cells, respectively compared to HC15 cells was also found in CD40 and SLC44A2. In PBMCs, high levels of CD40 compared to all other cell types were found, while the expression of ITGAL in PBMCs was comparable to Eos. All other markers were expressed in PBMCs at a similar level as in the cell line cells.

Next, we evaluated the cell lines’ ability to migrate towards the chemotactic stimulus fMLP using a transwell assay. Our results show that differentiation increased unspecific cell migration which was generally higher in the cell line cells than in Eos. Furthermore, in Eos, a significantly increased migration could be observed after the addition of a chemotactic stimulus. In IHC15 cells, a similar, but non-significant trend could be observed. ( Figures 7A, B ).

To study the influence PLTs might have on migration of the cell line cells, we analyzed the interaction between the cells and PLTs with and without PLT stimulation (see Supplementary Figure 4A for gating strategy of platelet stimulation). All cell line cells showed a similar baseline PLT binding capacity. Without stimulation, about 40% of the cell line cells were bound to PLTs after 15 min of incubation, whereas 20% of Eos bound PLTs. Activation of PLTs significantly increased the capacity of binding to the differentiated cell line cells and Eos to a similar extend. A similar trend was visible in the HC15 cells, which was however not significant ( Figures 7C-E , for gating strategy see Supplementary Figure 4B ).

Moreover, the adherence of the cells to stimulated HUVECs was tested. Since it is known that PLTs assist this process in a physiological setting, this experiment was performed with and without the addition of PLTs. Overall, more eosinophilic cells adhered in a PLT-rich setting than without PLTs (data not shown). In a PLT-rich environment, stimulation of HUVECs tended to increase the binding capacity of all cells. However, this was only significant in the case of Eos. Hence, our data indicate that the cell line cells overall, but cells differentiated without IL-5, specifically have a similar ability to adhere to HUVECs as Eos ( Figures 7F-H ).

Lastly, to investigate the influence of different adhesion receptors on the migratory functions of the cell line cells, we repeated the all chemotaxis and migration assays while blocking the adhesion markers studied before (see Figure 6 ). Since we did not find CD40 to be present on the surface of Eos (data not shown), we did not include its blocking into our experiments. However not significant, the blocking of SLC44A2, CD11b and TREM-1 lead to a slight reduction of Eos migration. This trend could also be seen in the undifferentiated cell line cells ( Supplementary Figures 5A, B ). Furthermore, PLT-aggregate formation was found to be strongly dependent on PSGL-1, as the blocking of this receptor led to a strong reduction of PLT binding in all cell types ( Supplementary Figures 5C-E ). In the case of cell adherence to stimulated HUVECs in a PLT-rich environment, blocking of the adherence receptors seemed to have little to no effect ( Supplementary Figures 5F, G ).

In summary, the cell line cells presented with a different surface marker expression than Eos. Nonetheless, the differentiated cell line cells showed a similar profile of chemotaxis, PSGL-1-dependent PLT-aggregate formation and PLT-dependent adherence as Eos.

After analyzing gene sets that were similar between IL-5-differentiated cells and Eos in our proteomics data, we set to investigate the differing gene sets, in particular activation and immunity. Eosinophil activation involves the increased exposure of adhesion markers, as well as the release of inflammatory mediators and granule proteins by degranulation (19). We activated the cells with the protein kinase C activating compound PMA to observe the exposure of the adhesion receptors CD11b and CD63 and the secretion of CCL5 and EPX (47).

In the cell line cells, PMA activation did not induce significant changes in CD63 exposure. However, in DHC15 cells, PMA induced an exposure of CD11b which could also be seen as a trend in IHC15 cells. In contrast, in Eos, PMA activation induced a significant increase of CD63 but not of CD11b ( Figures 8A-D , for gating strategy see Supplementary Figures 6A-C ). IHC15 cells showed a significantly enhanced CCL-5 secretion in response to PMA, a trend which could also be seen in Eos. Nonetheless, the total amount of CCL5 secreted from the IHC15 cells was about ten times higher than from that Eos. Furthermore, while the cell line cells showed an about ten times higher baseline secretion of EPX than Eos ( Figure 8E ), the EPX secretion was not affected by PMA in any of the cells ( Figure 8H ).

In summary, the differentiated cell line cells showed a different response to PMA-induced activation than Eos. While the surface marker exposure on the cell line cells in response to PMA was less pronounced and to some extend opposite to that on Eos, the baseline granule protein secretion from all cell line cells and the PMA-induced secretion of CCL5 from differentiated cells was increased compared to Eos.

Our proteomics data suggest that IL-5-differentiated cells might differ from Eos in proteins associated with eosinophil-mediated immunity. We compared the CCL-5 and EPX release between the cell line cells and Eos (data from experiment depicted in Figures 8E-H ). Indeed, in line with the proteomics data the CCL5 and EPX release was significantly increased in the cell line cells compared to Eos ( Figures 9A, B ).

We therefore investigated the secretion profiles of other cytokines of different functionality. We chose an array of cytokines involved in Th1, Th2, proinflammatory and anti-inflammatory signaling or granulocyte/eosinophil related pathways. Of the tested cytokines, only IL-8 could be detected in the supernatants, which showed a similar secretion profile as CCL5 ( Figures 9A-C , Supplementary Figure 7 ).

In summary, our data indicate that the cytokine profiles of cell line cells significantly differ from those of Eos, with cell line cells having a significantly higher EPX release and the differentiated cell line cells having a significantly higher IL-8 and CCL-5 release.