All animal experiments were in compliance with the German animal protection law in a protocol approved by the Landesverwaltungsamt Sachsen-Anhalt (file number: 203.h-42502-2-901, University of Magdeburg). The ethics committee of Hannover Medical School (MHH) approved the parts of the study involving patients (No. 10499-BO-K-2022).

Age- and sex-matched animals were used for the experiments. C57BL/6 wild-type (WT) mice were obtained from Janvier (Le Genest Saint Isle, France), and C57BL/6 Cyld ^−/− mice were kindly provided by Dr. Ramin Massoumi (Department of Laboratory Medicine, Malmö, Sweden) (44). All animals were kept under specific pathogen-free (SPF) conditions in an isolation facility at the Otto-von-Guericke University Magdeburg (Magdeburg, Germany).

Patients with AD were recruited at the Department of Dermatology and Allergy of Hannover Medical School (MHH). The work described was carried out in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki), and patients gave their written informed consent prior to the study.

Wild-type (strain SH1000) and methicillin-resistant (strain MW2) S. aureus were grown in Luria broth (LB, Oxoid, Germany), and aliquots of log-phase cultures were stored at −80°C. For infection of cells, fresh log-phase cultures were prepared from frozen stock.

The THP-1 cells (clone TIB-202) were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). The cells were cultured in a Roswell Park Memorial Institute (RPMI) cell culture medium supplemented with 10% fetal calf serum (FCS) and 1% penicillin/streptomycin.

CYLD was knocked down by the CRISPR/Cas9 method using SG cell line 4D-nucleofector X kit (#V4XP-4024, Lonza, Basel, Switzerland). For 1 × 10⁶ cells, 210 pmol of duplexed gRNA was mixed with 70 pmol of Cas9 Nuclease V3 (#1081059, IDT, Coralville, IA, USA) for RNP complex synthesis, and the complex was mixed with 70 pmol of EE buffer (#1075916, IDT) to form electroporation mix. Electroporation was performed using the preset program DZ100 in the Lonza nucleofection system.

THP-1 cells were differentiated into macrophages through Phorbol 12-myristate 13-acetate (PMA) (50 ng/mL) treatment for 48 h. The macrophages were either maintained in an unpolarized state (M0) or polarized into the M1 phenotype by stimulation with IFN-γ (20 ng/mL) and lipopolysaccharide (LPS) (10 pg/mL).

Peripheral blood mononuclear cells (PBMCs) were isolated using Ficoll density gradient centrifugation followed by magnetic cell separation of CD14⁺ monocytes (#480094, MojoSort, BioLegend, San Diego, CA, USA). CD14⁺ monocytes were then cultured for 6 days in Dulbecco’s modified Eagle medium (DMEM) medium supplemented with 50 ng macrophage colony-stimulating factor (M-CSF). Cells were harvested and stimulated as per experimental requirements. MDMs at day 7 polarized into M1 phenotype by stimulation with IFN-γ (20 ng/mL) and LPS (10 pg/mL).

In vitro infection of THP-1 macrophages and MDMs was performed as previously described (38). THP-1 macrophages and MDMs were infected with S. aureus at a multiplicity of infection (MOI) of 1:1. The dose of infection was verified by plating on Luria Broth (LB) agar. One hour post-infection, the extracellular bacteria were killed by incubating the cells with 30 µg/mL gentamicin (Sigma-Aldrich, St. Louis, MO, USA) for 30 min. The cells were thereafter washed with phosphate-buffered saline (PBS) to remove the extracellular bacteria and further cultivated in a medium containing gentamicin for the indicated time points. NF-κB was inhibited in THP-1 macrophages by treating the cells with IKK inhibitor VII (1 µM; Calbiochem, Darmstadt, Germany) for 24 h before the infection.

The bacterial load in infected THP-1 macrophages and MDMs was enumerated as previously described (38). In brief, 24 h p.i., S. aureus-infected cells were washed twice with PBS to remove the antibiotics, the cells were then lysed with 0.1% Triton X-100, and serial dilution was made and plated on LB agar. Bacterial colonies were enumerated microscopically after incubation at 37°C for 24 h and 48 h.

Proteins were isolated from S. aureus-infected THP-1 macrophages as previously described (38). In brief, cells were lysed using the lysis buffer (38). The lysates were centrifuged to remove the cell debris and the protein for Western blotting. For the Western blotting, equal amounts of proteins were separated on sodium dodecyl sulfate (SDS)–polyacrylamide gels as described previously (38), and the proteins were then transferred onto polyvinylidene fluoride membranes. To block the non-specific binding of antibodies, the membrane was incubated either with Blotto B [1% nonfat dry milk + 1% bovine serum albumin (BSA)], 5% nonfat dry milk, or 5% BSA for 1 h. The proteins were stained for GAPDH, phospho-STAT1 Y701, phospho-STAT1 S727, STAT-1, MyD88, IRAK-4, phospho-IRAK-4, TRAF6, NOD2, phospho-RIPK2, RIPK2, p65phospho-p65, p65, phospho-p38, p38, phospho-ERK1/2, ERK1/2, phospho-JNK, JNK, IKKγ/NEMO, CYLD, and K63-linkage-specific polyubiquitin overnight (all antibodies were obtained from Cell Signaling Technology, Frankfurt, Germany). The following day, membranes were washed using Tris-buffered saline with 0.1% Tween 20 (TBST) incubated with anti-mouse or anti-rabbit secondary antibodies (Dako, Hamburg, Germany) for 1 h. The blots were washed in TBST and developed using an ECL Plus kit (GE Healthcare, Freiburg, Germany). Western blotting (WB) images were captured using the Intas Chemo Cam Luminescent Image Analysis system^® (INTAS Science Imaging Instruments, Göttingen, Germany) and analyzed using the LabImage 1D software^® (Kapelan Bio-Imaging Solutions, Leipzig, Germany) (38).

Immunoprecipitation of CYLD and STAT1 was performed as described previously (38). In brief, proteins from uninfected and S. aureus-infected THP-1 macrophages were lysed using the lysis buffer as described. The protein samples were precleared using Gamma Bind™ G Sepharose™ beads (GE Healthcare Europe GmbH, Freiburg, Germany) to remove the proteins that non-specifically bind to the beads. The proteins were thereafter incubated with anti-STAT1 (1:100) or anti-CYLD (1:100) antibodies at 4°C overnight. IgG antibody was used as a negative control. The protein antibody complex was precipitated using fresh Gamma Bind™ G Sepharose™ beads at 4°C overnight. The protein antibody complex was then washed and incubated with 1× lane marker reducing sample buffer and heated at 99°C for 5 min. Thereafter, samples were centrifuged, and the supernatant was used to detect STAT1, CYLD, K63, TRAF6, RIPK2, IKK/NEMO, and K63-linked ubiquitin by WB. GAPDH was used as the input control (38).

The production of NO₂ by S. aureus-infected THP-1 macrophages was determined using the Griess Assay Kit (Promega, Mannheim, Germany) as previously described (38). In brief, the supernatant from infected and non-infected cells was incubated first with sulfanilamide solution and thereafter with N-(1-naphthyl)ethylenediamine solution in the dark for 10 min. The concentration of NO₂ was determined using a Synergy^® microplate reader (Biotek, Berlin, Germany).

ROS in S. aureus-infected THP-1 macrophages were determined using a ROS detection kit (Enzo Life Sciences, Lörrach, Germany) as described previously (38). In brief, S. aureus-infected THP-1 macrophages were washed twice with the washing buffer. Thereafter, the cell pellet was resuspended in the detection reagent and incubated in the dark at 37°C for 30 min. The samples were analyzed by Cytek Aurora flow cytometry (Fremont, CA, USA).

Lactate dehydrogenase (LDH) enzyme activity was determined in S. aureus-infected THP-1 macrophages using CyQUANT LDH Cytotoxicity Assay (#C20300, Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s protocol.

Stable knockout of CYLD was generated by the CRISPR/Cas9 system using the gRNA Hs.Cas9.CYLD.1.AA:TCACTGACGGGGTGTACCAA and SG cell line 4D-nucleofector X kit (#V4XP-3024, Lonza) according to the manufacturer’s protocol.

For siRNA-mediated knockdown of STAT1, WT and CYLD ^−/− THP-1 macrophages were transfected with 20 nm of STAT1-specific siRNA (Dharmacon, Lafayette, CO, USA) according to the manufacturer’s instruction. Scrambled siRNA at 20 µM was used as a control. A total of 50 µL of the siRNA mixture was added to each well in a 24-well plate containing 4 × 10⁵ macrophages and incubated at 37°C for 48 h. The efficiency of siRNA-mediated STAT1 silencing was controlled by WB.

OCT-embedded tissues were cut into 6-mm cryosections, air-dried, and fixed at 4°C in acetone for 10 min. After washing, horseradish peroxidase (HRP) and Alkaline Phosphatase (AP) blocking solution was applied for 20 min (Invitrogen, Waltham, MA, USA), followed by Superblock (Thermo Scientific, Waltham, MA, USA). Rabbit polyclonal anti-CYLD (ab33929, Abcam, Cambridge, UK) or rabbit Ig (Dako X 0936, Agilent, Santa Clara, CA, USA) were applied in equal concentrations. Mouse monoclonal anti-CD3 (Agilent) was used to stain the cellular skin infiltrate. After applying the Envision+ HRP conjugate (Agilent) with anti-rabbit K4009 and anti-mouse K005, the AEC substrate kit was applied (AEC, 3-amino-9-ethylcarbazole; Zytomed Systems, Bargteheide, Germany). Images were taken using a Pannoramic MIDI II Sola (Sysmex, Norderstedt, Germany).

Data from a previous study that included single-cell RNA sequencing were used for the investigation of CYLD expression on a single-cell level (Zhang et al., 2023). For a comprehensive explanation, refer to the corresponding methods section of the study. Briefly, skin punch biopsies were processed using the Miltenyi Biotech skin dissociation kit for subsequent cell isolation and CD2 enrichment, followed by pooling and loading onto the chromium chip (10x Genomics, Pleasanton, CA, USA). The CellRanger pipeline version 3.1.0 was employed to align reads to the human reference genome GRCH38. The expression matrix generated was then analyzed using the Seurat package version 4.2.3. UMAP and feature plots were generated using their respective functions.

The statistical significance was determined using the software Prism 9 with respective tests as mentioned in the figure legends, and p-values of ≤0.05 were considered significant. All experiments were performed at least twice.

The expression of CYLD in different organs and cells under homeostatic conditions has been summarized in “The Human Protein Atlas” (https://www.proteinatlas.org/ENSG00000083799-CYLD). These data show that CYLD is expressed in all organs including the skin, with strong protein and mRNA expression in keratinocytes, Langerhans cells, and immune cells including T cells, macrophages, and B cells but not in melanocytes (43). CYLD is expressed in all leukocyte populations including granulocytes and monocytes (https://www.proteinatlas.org/ENSG00000083799-CYLD/immune+cell). Transcriptome analysis of skin biopsies from 10 AD patients identified CYLD mRNA in T cells, natural killer cells ( Figure 1A ), and different macrophage subtypes including M2 and dendritic cells (DCs) ( Figure 1B ). mRNA expression in keratinocytes, fibroblasts, vascular endothelial cells, and pericytes appeared weaker but was present as well, while CYLD expression was also detectable in PBMC samples of the same donors including different myeloid cells ( Figures 1C, D ). Our histological analysis of CYLD expression in AD patients revealed the expression of CYLD in CD3⁺ T cells as well as non-T cells ( Figure 1E ). CYLD expression appeared particularly strong in the lesional skin of a patient with clinical signs of superinfected eczema ( Figure 1E , patient AD3). Our further histological analysis confirmed the widespread CYLD expression in the cellular infiltrate and keratinocytes of inflamed AD skin lesions as compared to healthy skin ( Supplementary Figure S1 ). Thus, CYLD is expressed in the skin of both healthy persons and AD patients with expression in identical cell types including skin macrophages and monocytes in the blood.

Since AD is characterized by a type 2 inflammation including M2 macrophage polarization (19) and activation of the IFN-γ/STAT1 and TLR2/4 pathways synergize in the control of S. aureus, we first determined whether M1 polarization of macrophages improves the control of S. aureus-infected macrophages of AD patients to the same extent as in healthy controls. In these experiments, we isolated CD14⁺ monocytes from the blood of AD patients and healthy controls, and we differentiated these cells into macrophages by M-CSF treatment followed by IFN-γ/LPS treatment. Control of S. aureus was significantly impaired in MDMs from all investigated AD patients ( Figure 2A ). Since CYLD can impair the cell-intrinsic control of intracellular bacteria (37, 38), we determined the CYLD expression in the infected MDMs. WB analysis showed that the impaired capacity of M1-polarized MDMs of AD correlated with an increased CYLD protein expression in these infected cells ( Figure 2B ).

Thus, AD patients are characterized by an overexpression of CYLD in macrophages and impaired control of S. aureus. To corroborate the inhibitory function of CYLD on the control of S. aureus in M1-polarized macrophages of AD patients, we deleted CYLD by siRNA in IFN-γ/LPS-primed (M1) macrophages of healthy blood donors and AD patients ( Figure 2C ). Upon infection, the control of S. aureus was significantly improved in CYLD-deleted macrophages of AD patients as compared to control siRNA-treated macrophages ( Figure 2D ). Thus, CYLD is upregulated in MDMs of AD patients and impairs the control of S. aureus in M1-polarized MDMs, demonstrating that CYLD is a macrophage intrinsic inhibitor of the control of S. aureus.

To determine the mechanisms of how CYLD impairs intrinsic immunity to S. aureus, we established CYLD-deficient THP-1 monocytes by CRISPR/Cas9 ( Figure 3A ). Upon PMA-mediated differentiation into macrophages and subsequent IFN-γ/LPS stimulation, CYLD-deficient THP-1 cells expressed higher levels of CD80, a marker for M1 macrophages, and TNF and IL-6 mRNA, two prototypic M1 macrophage cytokines important for the control of S. aureus ( Figure 3B ). In IFN-γ/LPS-stimulated THP-1 macrophages, CYLD deletion significantly improved control of methicillin-sensitive and methicillin-resistant S. aureus, whereas no differences between the two genotypes were observed for unstimulated macrophages ( Figures 3C, D ). Analysis of anti-bacterial nitric oxide and reactive oxygen species revealed that ROS were significantly increased in infected CYLD-deficient THP-1 24 h p.i., whereas NO levels did not differ ( Figures 3E–G ). A kinetic analysis of bacterial loads revealed that CYLD does not influence the entry of S. aureus into the macrophages ( Figure 3H ) but impairs intracellular bacterial control, which leads to the death of macrophages ( Figure 3I ). These data illustrate that CYLD impaired M1 macrophage differentiation, control of S. aureus, and ROS production.

Since M1-polarized macrophages showed improved control of S. aureus and anti-bacterial activity in CYLD-deficient macrophages, we determined the impact of CYLD on i) IFN-γ-induced STAT1 activation, ii) the effect of LPS (TLR4) and S. aureus (TLR2)-induced MyD88 signaling, and iii) S. aureus-activated NOD2 pathways.

WB analyses showed that CYLD impaired STAT1 phosphorylation at tyrosine 701 and serine 727, which cooperatively led to the induction of IFN-γ-induced genes (Sadzak et al., 2008; Varinou et al., 2003) ( Figure 4A ). Co-immunoprecipitation experiments newly identified that CYLD directly interacted with STAT1 ( Figures 4B, C ). STAT1 immunoprecipitation and subsequent WB analysis of K63 ubiquitin revealed that S. aureus infection induced increased K63 ubiquitination of STAT1 in CYLD-deficient macrophages as compared to WT macrophages at 2 h p.i. ( Figure 4D ). In agreement with published data in other models (29, 30, 32, 38), CYLD also interacted with TRAF6, IKKγ/NEMO, and RIPK2 in M1-polarized S. aureus-infected THP-1 macrophages ( Figure 4E ). This was associated with impaired MyD88 and NOD2/RIPK2 signaling, leading to reduced downstream NF-κB activation shown by impaired phosphorylation of p65 in S. aureus-infected CYLD-competent macrophages ( Figure 4F ). Thus, CYLD inhibits simultaneously key pathways leading to reduced activation of the transcription factors NF-κB and STAT1.

To determine the functional importance of CYLD-regulated NF-κB and STAT1 activation for the control of S. aureus, we stimulated THP-1 macrophages with IFN-γ/LPS for M1 polarization and treated the cells prior to infection with the IKK VII inhibitor to inhibit activation of NF-κB and with STAT1 siRNA to inhibit STAT1. Inhibition of NF-κB activation and STAT1 resulted in a strong increase of colony-forming units (CFUs) of S. aureus in both WT and CYLD-deficient THP-1 macrophages, and both treatments abolished the differences between the two genotypes ( Figures 5A, B ). Thus, the inhibition of both STAT1 and NF-κB by CYLD is critical for the control of S. aureus in macrophages.

The data presented identify that CYLD is increasingly expressed in macrophages of AD patients and inhibits the control of S. aureus in human macrophages. To further validate an inhibitory function of CYLD in S. aureus infection and to evaluate whether systemic CYLD inhibition may be a therapeutic option to ameliorate S. aureus infection, we infected Cyld-deficient and WT mice with S. aureus. Cyld-deficient mice had significantly reduced weight loss ( Figure 6A ) and improved pathogen control in the liver and kidney at the acute and chronic stages of infection ( Figures 6B, C ). In good agreement with the human macrophages, CYLD-deficient murine bone marrow-derived macrophages showed better clearance of S. aureus compared to WT macrophages ( Figure 6D ). This qualifies CYLD as a relevant therapeutic target to ameliorate S. aureus infections.

CYLD is known to negatively regulate various signaling pathways, including NF-κB (61, 62), MAPK (38), and JAK/STAT (37, 38, 63) pathways, which are crucial for the inflammatory response. Our previous studies have identified CYLD as an inhibitor of immune responses during bacterial (37, 38) and parasitic infections (64), and inhibition of CYLD resulted in a positive disease outcome. Inhibition of CYLD in hepatocytes promoted IL-6/STAT3-mediated fibrin production and prevented the spread of L. monocytogenes (37). Ablation of CYLD in macrophages enhanced NF-κB-mediated production of reactive oxygen species, resulting in improved clearance of L. monocytogenes (38). Depletion of CYLD in T cells attenuates T-cell responses, prevents blood–brain barrier disruption, and protects mice from experimental cerebral malaria (64). Furthermore, Lim et al. (2007) showed that CYLD deficiency protected mice from S. pneumoniae pneumolysin (PLY)-induced Acute lung injury (ALI) and lethality (34). In the current study, we identified CYLD as an inhibitor of the anti-S. aureus immune response in human and murine macrophages. These studies highlight the potential of CYLD as a therapeutic target in infectious diseases. However, it should be noted that CYLD negatively regulates NF-κB-dependent inflammation during non-typeable H. influenzae infection and protects mice from deleterious inflammatory responses (65). Similarly, CYLD impairs inflammation and protects mice from E. coli-induced pneumonia (35), suggesting that the protective or detrimental function of CYLD in infectious diseases depends on the underlying infection.

M1 macrophages are crucial in defending against pathogens through phagocytosis and the secretion of pro-inflammatory cytokines. Several bacteria, including S. aureus (66), Mycobacterium (67), Salmonella (68), and Coxiella (69), promote M2 polarization to evade the pro-inflammatory response and enhance their survival within macrophages. Ablation of CYLD would foster M1 polarization, enhancing their antimicrobial activities and leading to improved outcomes in these bacterial infections.

Commercially available small molecule inhibitors that target the USP family of deubiquitinases, such as NSC-687852, NSC112200, and WP1130, could be used to inhibit CYLD. More specifically, subquinocin, a small molecule inhibitor primarily targeting CYLD, could serve as a potential therapeutic strategy against infectious diseases (70). Additionally, proteolysis-targeting chimeras (PROTACs), which utilize the ubiquitin–proteasome system to target specific proteins for degradation (71), could be used to specifically target CYLD for proteasomal degradation, making it a promising therapeutic option.