CDK7, CDK9, or CDK11 inhibition reduces neutrophil-driven inflammation and tissue damage in experimental autoimmune models

Abstract

Background

Cyclin-dependent kinases (CDKs) are involved in basic cellular processes like regulation of cell-cycle progression and transcription. However, recent data also indicate a specific role in terminally differentiated neutrophils by promoting reactive oxygen species (ROS) release, degranulation, neutrophil extracellular trap formation, or apoptosis. Since these mechanisms are implicated in multiple autoimmune diseases, we aimed to delineate the role of CDKs in immune complex (IC)-mediated autoimmune diseases both in vitro and in vivo.

Methods

We analyzed CDK activity and gene expression in unstimulated and IC-stimulated neutrophils. Subsequently, we investigated the effect of pharmacological CDK inhibition on IC-activated neutrophil functions. To analyze the inhibitors in a more translational approach, we proceeded with the systemic and topical application of the effective inhibitors in a murine antibody transfer-induced local epidermolysis bullosa acquisita (EBA) model. The most effective inhibitor, MC180295, was validated in two other IC-mediated models of autoimmune disease: KBxN serum-transfer arthritis (KBxN STA), which also, but not exclusively, depends on neutrophils and immune thrombocytopenia (ITP), which is considered less neutrophil-dependent.

Results

We found 14 CDKs expressed in unstimulated cells, while the IC-stimulation showed an upregulation of CDK2 and CDK4 expression. Inhibitors selectively targeting CDK1, CDK2, CDK4/6, CDK7, CDK9, CDK11, and CDK12 showed effects on different neutrophil functions (surface activation marker expression, ROS release, adhesion, apoptosis) in vitro. In the predominantly neutrophil-driven EBA model, we observed a reduction of disease severity upon treatment with CDK7, CDK9, or CDK11 inhibitors. Inhibiting these CDKs with topical THZ2, MC180295, or OTS964, respectively, also improved the clinical phenotype. In line with our hypothesis, MC180295 impaired the development of KBxN STA, but not ITP.

Conclusions

CDK7, CDK11, and especially CDK9 inhibition show therapeutic potential in IC-driven neutrophil-mediated diseases such as inflammatory arthritis and EBA.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12964-026-03219-x.

Background

Cyclin-dependent kinases (CDKs) are a group of serine-threonine kinases that were discovered for their involvement in the regulation of the eukaryotic cell cycle [1–3]. CDKs are crucially contributing to the correct replication and subsequent separation of the chromosomal DNA during mitosis [4] and are highly conserved across eukaryotes, from yeast to mammals [5]. They were termed for their critical interaction with specific regulatory protein binding partners, the cyclins [6]. The latter are expressed in a cyclic fashion during the cell cycle, while the CDKs themselves are stably expressed throughout the cell cycle [3]. There are 20 CDKs and 29 cyclins described in mammals [7], as well as other regulating proteins such as intrinsic CDK inhibitors (CKI) [8]. Mechanistically, activation of CDKs not only requires the interaction with a cyclin, also phosphorylation at a specific tyrosine residue in the activation loop is necessary [9, 10]. This phosphorylation of at least CDK1/2/4/6/9/12/13 is mediated by the CDK-activating kinase (CAK) complex [11], consisting of CDK7, cyclin H and MAT1 [12–14]. Importantly, CDK inactivation is also regulated by phosphorylation [15] or binding of CKIs [16, 17].

Beyond the well-described role of CDKs in direct cell-cycle regulation, there is rising evidence that CDKs are also involved in regulating the RNA polymerase II transcription cycle [11, 18]. It was thus proposed that there are two groups of CDKs: the cell-cycle CDKs (subfamilies CDK1, 2, 3, 4, and 6) and the transcriptional CDKs (subfamilies CDK7, 8, 9, 10, 11, 12, and 13) [7]. The remainder is referred to as atypical CDKs, although these kinases are insufficiently characterized [19]. Though this may be a helpful concept, this dichotomy is probably oversimplified as more and more studies demonstrate roles for traditional cell-cycle CDKs in transcriptional regulation and vice versa [18, 20–22].

Considering that CDKs are not only expressed in proliferating cells, but also in terminally differentiated cells like neutrophils [23], it seems a reasonable assumption that their role extends beyond mere cell cycle regulation. Neutrophils are known as the first line of defense against pathogens [24]. Once recruited to the tissue of interest, they are activated by the binding of immune complexes (IC) to their fragment crystallizable gamma receptors (FcγR). Downstream kinase signaling cascades are activated, triggering effector functions such as reactive oxygen species (ROS), protease and cytokine release to kill the invading pathogens in a controlled manner. Subsequently, the neutrophils are cleared mostly by macrophages under inflammatory conditions [25, 26]. Notably, a role for several CDKs in regulating neutrophil functions has been described: CDK2 is involved in migration [27], CDK4/6 in formation of neutrophil extracellular traps (NETs) [28], CDK5 in degranulation [29], and CDK7/9 in apoptosis [22, 30]. Still, little is known about the distinct functions of different CDKs in direct comparison, and most studies rely on broad-range small-molecule CDK inhibitors (CDKi). Moreover, the potential involvement of CDKs in FcγR-mediated signaling remains largely unexplored, representing a significant gap in our understanding of CDK biology.

An increasing number of studies point towards the relevance of CDKs in various human disorders, underlining the need for a better comprehension of these kinases. Deregulated CDK activity has been shown to be associated not only with e.g. neurodegenerative diseases [31] and cancer [19], but also with autoimmune diseases like rheumatoid arthritis (RA) [32] or systemic lupus erythematosus [33]. In contrast to infectious diseases, however, there is no pathogen in autoimmunity, but autoantibodies develop against self-antigens due to a break of tolerance [34]. To unravel the function of CDKs in the context of neutrophil-mediated FcγR-dependent signaling, we investigated the kinase signaling pathways upon IC stimulation in human neutrophils in vitro. Furthermore, we chose three well-established murine in vivo models of antibody-mediated autoimmune diseases: Epidermolysis bullosa acquisita (EBA), KBxN serum-transfer arthritis (KBxN STA) as a model for RA, and immune thrombocytopenia (ITP). EBA belongs to the group of pemphigoid diseases (PD), which is a group of rare autoimmune bullous skin diseases characterized by autoantibodies targeting structural proteins of the dermal-epidermal junction. More specifically, the autoantigen in EBA is collagen VII (COL7) [35]. Neutrophils are the central effector cells in PD [36–39], causing the uncontrolled release of inflammatory mediators. Thereby, tissue damage occurs, causing the degradation of the dermal-epidermal adhesion complex and subepidermal split formation [38, 40]. Clinically, patients present with pruritus, erosions, and tense, painful blisters on different sites of the body, e.g. trauma-prone areas or the extensor skin surface [41]. Currently, the patients are most commonly treated with systemic corticosteroids, often in combination with additional immunosuppressants, bearing a high risk for adverse events due to general immunosuppression [42, 43].

Another more prevalent neutrophil-dependent autoimmune disease with the need for novel therapeutic options is RA, modelled by KBxN STA [44]. Similarly to EBA, neutrophils do play a critical role in pathogenesis [45, 46], but other cell types like macrophages [47] or mast cells [48, 49] are also crucially involved in the effector phase. Joint inflammation, bone and cartilage destruction are major hallmarks of RA [50].

Conversely, ITP is also IC-driven, but depends less on neutrophils. In ITP, autoantibodies target platelets leading to their phagocytosis mainly by macrophages, thus causing platelet depletion [51–53].

Since kinases are generally well-druggable targets [54, 55] and considering that there are several CDKi already in clinical use, e.g. for cancer treatment [56], a potential repurposing of the inhibitors could accelerate the process of clinical approval. Thus, the characterization of CDK signaling cascades in neutrophils can help to identify potential therapeutic targets for PD, but also potentially for other neutrophil-driven autoimmune diseases like RA. This could provide patients with targeted treatment options and therefore drastically ameliorate the safety profile of the therapy by allowing to avoid broader immunosuppression.

Methods

Sampling of human biomaterials

Whole blood was obtained from healthy volunteers after written informed consent. All experiments using human samples were approved by the local ethics committee (AZ #20–463, University of Lübeck, Lübeck, Germany) and were performed in accordance with the Declaration of Helsinki.

Chemicals

All standard chemicals were supplied by Carl Roth (Karlsruhe, Germany) or Sigma-Aldrich (Taufkirchen, Germany). CDKi were supplied by Selleck Chemicals (Houston, Texas, USA) or MedChemExpress (Monmouth Junction, New Jersey, USA). For in vitro assays, 10 mM CDKi was dissolved in dimethyl sulfoxide (DMSO) and further diluted to final concentrations of 10 µM, 1 µM, 0.1 µM, and 0.01 µM. DMSO at a final concentration of 0.1% was included in the controls. For in vivo experiments, CDKi were prepared as stated in Table 1. The respective vehicle only served as the respective control. Additional targets, references for the in vivo dosage and stage of drug development are given in Supplemental Table 1.

Table 1.

CDKi used in this study. The vehicle, dosage, route and frequency for systemic in vivo treatment are stated, where applicable

substance	main target	vehicle	dosage [mg/kg]	route	frequency
Ro 3306	CDK1	10% DMSO/ 30% PEG-300/ 10% Tween-80	4	p.o.	daily
OTS964	CDK11B	10% DMSO in corn oil	40	p.o.	daily
Palbociclib HCl	CDK4/6	sodium lactate buffer (50 mM, pH 4.0)	100	p.o.	daily
THZ1 2HCl	CDK7	10% DMSO/ 90% 5% dextrose in water	10	i.p.	twice daily
THZ2	CDK7	10% DMSO in corn oil	10	i.p.	daily
MC180295	CDK9	5% DMSO/ 40% PEG-300/ 5% Tween-80/ 50% water	20	i.p.	daily
LY3143921 hydrate	CDC7	-	-	-	-
BMS-265,246	CDK1/2	-	-	-	-
Purvalanol B	CDK1/2/5	-	-	-	-
BSJ-4-116	CDK12	-	-	-	-
SR-4835	CDK12/13	-	-	-	-
K03861	CDK2	-	-	-	-
AS2863619	CDK8/19	-	-	-	-
BAY1251152	CDK9	-	-	-	-

Multiplex kinase activity profiling

To assess CDK activity, a published dataset of a multiplex kinase activity profiling using the PamGene™ technology (BJ‘s-Hertogenbosch, The Netherlands) was reanalyzed. Briefly, lysates of human PMNs stimulated with immobilized COL7-anti-COL7-IgG1 IC for 1 min, 5 min, or 15 min were compared to lysates of unstimulated PMNs [57]. The analysis was performed using the BioNavigator software (version 0.17.52) as previously described [57].

Isolation of human immune cells by magnetic activated cell sorting (MACS)

For the isolation of human neutrophils by MACS, blood from healthy young (< 30 y) donors was drawn into citrate blood collection tubes. After 1:2 dilution with 1% (w/v) polyvinyl alcohol and sedimentation for 25 min, the supernatant (up to 40 mL) was transferred on top of a layer of 8 mL of Ficoll Paque PLUS (Cytiva, Marlborough, Massachusetts, USA). Density gradient centrifugation [58] was performed for 24 min at 850 g and room temperature (RT) without brake. The lymphocyte layer was processed for B and CD4⁺ T cell isolation, whereas the granulocyte pellet was used for neutrophil purification. The lymphocyte layer was resuspended in phosphate-buffered saline (PBS) and centrifuged at 300 g for 10 min at 4 °C for platelet removal. Another washing step with PBS was performed at 400 g for 10 min at 4 °C. Next, the purified lymphocytes were subjected to MACS according to the manufacturer’s instructions using the CD4⁺ T cell isolation kit (130-096-533, Miltenyi Biotec, Bergisch Gladbach, Germany) and the B cell isolation kit II (130-091-151, Miltenyi Biotec). B and CD4⁺ T cell purities were assessed by flow cytometry and were > 80%.

For erythrocyte lysis, the granulocyte pellet was resuspended in distilled water, mixed for 45 s, and then neutralized with 2x PBS. Two washing steps with PBS were performed for 10 min at 300 g and RT. Subsequently, the granulocytes were subjected to eosinophil removal by MACS according to the manufacturer’s instructions (130-092-010, Miltenyi Biotec). Neutrophil purity was assessed by flow cytometry and was > 98%.

Bulk RNA sequencing

For bulk RNA sequencing (RNAseq), 100,000 MACS-purified neutrophils from healthy volunteers were sampled and frozen at -80 °C in 1 mL Trizol (Invitrogen, Waltham, Massachusetts, USA). An equal number of cells from each donor was stimulated for 6 h at 37 °C and 5% CO₂ with immobilized IC consisting of human COL7E-F and anti-human COL7 IgG1 as previously described [59]. All samples were sent to Single Cell Discoveries (Utrecht, The Netherlands) for RNA extraction and sequencing. Data analysis was performed in RStudio (Posit Software, Boston, Massachusetts, USA; version 2024.12.1) as previously described [60]. Differentially expressed genes were identified using DESeq2 [61] on transcripts where the sum of the counts over all samples was > 10, and p_adj < 0.1 was considered significant. Normalized counts were extracted after DESeq2 analysis. Gene set enrichment analysis was conducted using ClusterProfiler referring to the Gene Ontology (GO) database [62, 63]. The data for this study have been deposited in the European Nucleotide Archive (ENA) at EMBL-EBI under accession number PRJEB103987 (https://www.ebi.ac.uk/ena/browser/view/PRJEB103987).

Human polymorphonuclear granulocyte (PMN) purification for real-time quantitative reverse transcription polymerase chain reaction (qRT-PCR) and functional assays

For ROS release assay, flow cytometry, adhesion assay, and qRT-PCR, PMNs were isolated from EDTA-treated human whole blood performing a PolymorphPrep™ (Serumwerk Bernburg AG, Bernburg, Germany) density gradient centrifugation as previously described [59, 60, 64]. The cells were diluted to 4 × 10⁵ cells/mL in 1% (w/v) bovine serum albumin (BSA) in PBS. Neutrophil purity was assessed by flow cytometry for all experiments and was higher than 90% for all samples.

qRT-PCR

For gene expression analysis by qRT-PCR, human PMNs were stimulated with human COL7E-F anti-human COL7 IgG1 IC for 6 h at 37 °C and 5% CO₂ as stated above. Alongside unstimulated PMNs, B cells and CD4⁺ T cells isolated by MACS as described above, the cells were collected with a cell scraper in PBS and centrifuged for 7 min at 700 g and RT. The lysates were resuspended in 700 µL RLT buffer (Qiagen, Hilden, Germany) and frozen at -80 °C until RNA extraction. RNA extraction was performed using the innuPREP RNA Mini Kit 2.0 (Analytik Jena, Jena, Germany) according to the manufacturer’s instructions. Any remaining DNA was digested with DNase I (Sigma-Aldrich). cDNA was synthesized using the RevertAid Reverse Transcriptase (Thermo Fisher Scientific, Waltham, Massachusetts, USA) according to the manufacturer’s instructions. TaqMan™ gene expression assays (see Table 2) were purchased from Thermo Fisher Scientific and used according to the manufacturer’s instructions. qRT-PCR was performed on an Applied Biosystems 7900HT Fast Real-Time PCR System (Thermo Fisher Scientific). The results were analyzed using the 2^−ΔΔCt method [65] and normalized to the mean values of the housekeeping genes [66].

Table 2.

TaqMan™ gene expression assays used for qRT-PCR

	Catalogue no.	Assay ID	Gene symbol
CDKs	4453320	Hs00938777_m1	CDK1
	4453320	Hs01548894_m1	CDK2
	4448892	Hs01077186_g1	CDK3, TEN1-CDK3
	4453320	Hs00364847_m1	CDK4
	4453320	Hs00358991_g1	CDK5
	4453320	Hs01026371_m1	CDK6
	4448892	Hs00361486_m1	CDK7
	4448892	Hs00992501_g1	CDK8
	4448892	Hs00977896_g1	CDK9
	4448892	Hs02341397_m1	CDK11A
	4448892	Hs07291776_gH	CDK11B
	4448892	Hs00901732_m1	CDK12
	4448892	Hs00953418_m1	CDK14
	4448892	Hs01039930_g1	CDK19
	4453320	Hs00177487_m1	CDC7
housekeeping genes	4331182	Hs01060665_g1	ACTB
	4331182	Hs02786624_g1	GAPDH
	4331182	Hs02800695_m1	HPRT
	4331182	Hs00272002_m1	GNB2L1

ROS release assay

Plates were coated with IC consisting of human COL7 E-F at 2.5 µg/mL and anti-human COL7 IgG1 at 1.8 µg/mL as previously described [59, 67]. PMNs were added at 2 × 10⁵ cells per well in the presence or absence of CDKi. Only antibody (with cells), only antigen (with cells), or only cells served as negative controls, while IC (with cells, but without CDKi) served as positive control. DMSO was adjusted to a final concentration of 0.01%. All samples were measured in duplicates. Luminol was automatically injected into each well to reach a final concentration of 90 µg/mL, and chemiluminescence was measured for 2 h every 2 min at 37 °C in a GloMax^® Discover microplate reader (Promega, Walldorf, Germany). The area under the curve (AUC) was calculated, and the results were normalized to IC-stimulated cells without CDKi.

Adhesion assay

PMN adhesion was investigated using impedance-based measurements as described before [68, 69]. A 96-well E-plate (Agilent, Santa Clara, California, USA) was coated with IC as described for the ROS release assay. Inhibitor solutions were prepared and added to the plate before 2 × 10⁵ cells were transferred to each well. Only antibody (with cells), only antigen (with cells), or only cells served as negative controls, while IC (with cells, but without CDKi) served as positive control. DMSO was adjusted to a final concentration of 0.01%. All samples were measured in duplicates. Impedance was measured in an xCELLigence real-time cell analysis device (Agilent) every 2 min for 120 min expressed in arbitrary units as cell index. Data analysis was performed as for the ROS release assay.

Flow cytometry

Human PMNs were isolated as previously described and stimulated with IC, prepared in the same manner as for the ROS release assay, for 2 h at 37 °C and 5% CO₂. Subsequently, they were stained for the flow cytometric analysis of the following surface markers using standard flow cytometry procedures: CD14 (clone HCD14), CD16 (clone 3G8), CD18 (clone 1B4/CD18), CD45 (clone HI30), CD62L (clone DREG-56), and CD193 (clone 5E8). Viability was assessed by Annexin V (B266195) and Zombie NIR staining (B308933) [60]. Only antibody (with cells), only antigen (with cells), or only cells served as negative controls, while IC (with cells, but without CDKi) served as positive control. DMSO was adjusted to a final concentration of 0.01%. All samples were measured in duplicates. Analyses for activation markers were performed on neutrophils identified as FSC-SSC gated CD45⁺CD16⁺CD14⁻CD193⁻ singlets. Annexin V⁻ Zombie NIR⁻ cells were considered viable, Annexin V⁺ Zombie NIR⁻ cells were considered apoptotic. The positive control corresponds to values of up to 3%. All dyes were purchased from BioLegend (San Diego, California, USA). Measurements were performed on a MACSQuant^® Analyzer 10 (Miltenyi Biotec) and analysis was conducted using MACS Quantify software (version 2.13.3, Miltenyi Biotec).

Study approval

Animal experiments were approved by local authorities of the Animal Care and Use Committee (government of Schleswig-Holstein and government of Lower Franconia) and the Institutional Animal Care and Use Committee at Rush University. All experiments were performed by certified personnel (AZ 108/08–15, AZ 97 − 11/20 and 55.2.2-2532-2-2085) following the ARRIVE guidelines.

Animal experimentation

C57BL/6J mice (Charles River, Wilmington, Massachusetts, USA) were bred in a specific pathogen–free environment and provided standard mouse chow and acidified drinking water ad libitum. Mice of both sexes, aged 8 to 14 weeks, were used for experimental EBA models with clinical examinations performed under anesthesia, using intraperitoneal (i.p.) administration of a ketamine (75 mg/kg, Sigma-Aldrich) and medetomidine (1 mg/kg, Vetoquinol, Ismaning, Germany) mixture, antagonized after ca. 30 min by atipamezole (5 mg/kg, Vetoquinol). On the final day, the mice were anesthetized using 15 mg/kg xylazine (Sigma-Aldrich) and 100 mg/kg ketamine. KBxN STA and ITP models were performed with female C57BL/6J aged 12 weeks. For anesthesia, isoflurane was used and mice were killed by CO₂.

Local antibody-transfer induced experimental EBA

Specific anti–mouse COL7^C IgG from rabbit serum was isolated as previously described [40, 70–72]. Experimental EBA was induced by ear-base injection with 100 µg rabbit anti–mouse COL7^C IgG. For systemic treatment, CDKi were administered as indicated in Table 1 at 100 µL/20 g mouse. For topical treatment in independent experiments, THZ2, MC180295, or OTS964 were applied at 65 µg/100 µL in acetone/DMSO (20:1) per ear, beginning one day before the anti–mouse COL7^C IgG injection and treatment was performed daily for four days. Vehicle-treated mice served as controls in all cases. The mice were clinically evaluated every day until day 3, and the percentage of ear thickness was assessed using a Mitutoyo 7301 dial thickness gauge (Neuss, Germany) by a blinded person, as well as the affected ear surface area (AESA). Ears were collected on the final day of the experiment, and half of each ear was embedded in 4% PBS-buffered paraffin or frozen in liquid nitrogen and stored at -80 °C. Histological analysis of hematoxylin and eosin (H&E) staining and direct immunofluorescence staining for murine IgG and C3 were performed as described elsewhere [72]. Scoring of H&E-stained sections was performed to determine (i) the magnitude of epidermal thickness with values from 0 (10–20 μm), 1 (20–40 μm), 2 (40–100 μm) to 3 (> 100 μm), (ii) the percentage of dermal-epidermal split formation (length of split formation/length of dermal-epidermal junction), with values from 0 (no split), 1 (< 2%), 2 (20 − 5%) to 3 (> 5%) and (iii) the dermal infiltration with values from 0 to 3 corresponding to no, mild, moderate or severe degree, respectively. The H&E score was calculated as the mean of these three parameters.

KBxN STA

KBxN STA was induced by injection of arthritogenic KBxN serum in female C57BL/6J mice at 10 µL/g body weight as described before [73]. 1–2 h before and daily after arthritis induction, mice were treated i.p. with 20 mg/kg body weight MC180295 or vehicle control. The severity of joint swelling was monitored to evaluate clinical disease over the course of seven days. Peripheral blood was collected retro-orbitally on day 0 and day 7, and inflammatory cytokines were quantified in the serum by LegendPlex multiplex assay (Mouse Inflammation 13-plex, BioLegend) as a parameter of systemic inflammation. On day 7, mice were sacrificed, and forelegs were prepared for histological analysis. H&E staining was performed to evaluate inflammatory infiltrates and Safranin-O staining to assess bone and cartilage damage, as previously described [73, 74]. Stained sections were evaluated by light microscopy using a Zeiss Axiovert 200 microscope and the AxioVision LE4.6 software (Olympus Life Science, Waltham, Massachusetts, USA).

Immune thrombocytopenia (ITP)

Platelet depletion in the presence of kinase inhibition was assessed in female C57BL/6J mice. Platelet depletion was induced by injection of 0.35 µg/g anti-platelet IgG2c (clone 6A6) as described [53]. Treatment with vehicle control or MC180295 (20 mg/kg body weight i.p.) was performed 24 h and 4 h before the induction of platelet depletion. Platelet counts in peripheral blood were assessed immediately before 6A6-IgG2c injection, and at 4 h and 24 h post-injection in peripheral blood collected retro-orbitally. Platelet counts were determined using an ADVIA^® 2120i hematocytometer (Siemens Healthineers, Erlangen, Germany).

Immunofluorescence stainings

Paraffin-embedded ear sections of MC180295-treated and control antibody transfer-induced EBA mice or paw sections of KBxN STA mice were subjected to immunofluorescence staining for Ly6G or F4/80 as described previously [75] and mounted with DAPI Fluoromount-G™ (Southern Biotech, Birmingham, Alabama, USA). Images were taken at a Keyence BZ-X810 Fluorescence microscope (Osaka, Japan) and quantified using ImageJ software version 1.53p (National Institutes of Health, Bethesda, Maryland, USA).

Statistical analysis

Data were analyzed using Prism 10.4.1 (GraphPad Software, San Diego, USA). For all in vitro assays, Kruskal-Wallis test with Dunn’s multiple comparisons test was performed if not indicated differently. Dose response curves and IC₅₀ values were calculated using the non-linear fit inhibitor vs. response (three parameters) function or for EC₅₀ values using the agonist vs. response (three parameters) function where an activating response was observed. For all in vivo experiments, two-way ANOVA or mixed-effects analysis (in the case of missing values) with Šidák’s multiple comparisons test were performed unless stated differently. A value of p ≤ 0.05 was considered significant. The data are displayed as Tukey box-and-whisker plots or else as mean ± standard deviation unless described differently.

Results

CDKs are active in human neutrophils and expression of CDK2 and CDK4 changes upon IC stimulation

CDKs are essential to drive cell cycle progression in proliferating cells [19]. However, at least certain CDKs seem to be functionally relevant also in terminally differentiated cells like neutrophils [22, 27–29]. Since there are conflicting data published on this topic [22, 28], we first set out to clarify which CDKs and regulating proteins are expressed at mRNA level in human neutrophils and which CDKs are modulated in their activity in IC-stimulated human neutrophils. Concerning kinase activity, a published dataset was reanalyzed unveiling changes in the activity of several CDKs. Mostly, upregulation in activity was observed after 5 and 15 min of stimulation, especially for CDKL5, CDK11A, CDK17, CDK7, CDKL1, and CDK13 (Fig. 1A). In total, 17 CDKs were detected by this method. To gain further insight into the gene expression levels of CDKs, an RNAseq experiment on neutrophils from healthy volunteers was performed. In addition, neutrophils were stimulated with immobilized IC for 6 h and compared to unstimulated neutrophils to reveal any transcriptional changes. Figure 1B shows only significantly altered CDKs, cyclins, and selected regulators after IC stimulation, whereas Fig. 1C shows the best-described CDK-cyclin complexes involved in the cell cycle and the transcriptional cycle irrespective of whether their expression changes significantly upon IC stimulation or not. From the 20 CDKs that are overall expressed in humans, we found 16 CDK transcripts (considering both paralogues of CDK11) expressed in unstimulated neutrophils and six significantly changed by IC stimulation. While CDK13 (p_adj = 0.0051) and CDK19 (p_adj = 1.92E-05) expression was downregulated after 6 h, CDK2 (p_adj = 6.79E-07), CDK4 (p_adj = 1.80E-08), CDK7 (p_adj = 0.0096), and CDK12 (p_adj = 0.0209) expression was upregulated. Regarding cyclins, 16 out of 29 genes were expressed. On the one hand, the CDKs with best established roles in direct regulation of the cell cycle, namely CDK1, CDK2, CDK4, and CDK6, were all detected by RNAseq except CDK1. The gene expression of CDK2 and CDK4 was induced upon IC stimulation. In addition, their partner cyclins were also detected except cyclin E (CCNE), which binds to CDK2 during late G1 phase to promote G1-S transition [76, 77]. Cyclin A (CCNA) was unchanged after IC stimulation, whereas cyclin B (CCNB, p_adj = 0.0279) and D (CCND, p_adj = 0.0011) showed an increased expression. On the other hand, expression of the transcriptional kinases CDK7, CDK9, CDK12, CDK13 and CDK19 was observed, but not of CDK18. Their cyclin binding-partners were all detected with cyclin H (CCNH, p_adj = 0.0012) and T (CCNT, p_adj = 0.0003) elevated upon IC treatment. Alongside CDKs, the expression of other genes such as CCL3/4 (p_adj = 0.0209), FRTA1/2 (p_adj = 0.0209) and many others was significantly modulated (Fig. S1A). In addition, the gene expression of other CDK-regulating proteins like RB1 (p_adj = 4.89E-05), EP300 (p_adj = 0.0112) or MED12/13 (p_adj = 0.0011/ p_adj = 0.00296) was significantly elevated after IC stimulation (Fig. S1B). Furthermore, GO pathway analysis of all genes significantly changed by IC stimulation revealed an activation of granulocyte-specific pathways and a suppression of metabolic pathways (Fig. 1D). To exclude that these effects are confounded by dead or dying cells, an Annexin V/Zombie NIR staining was performed, showing that the viability of the cells is not significantly influenced by the stimulation (Fig. S1C).

Fig. 1.

Cell-cycle and transcriptional CDKs are expressed in human neutrophils and IC-stimulation partially alters their expression. A Human PMNs were stimulated with COL7 anti-COL7 IgG1-IC for 1 min, 5 min, or 15 min and their kinase activity was compared to unstimulated PMNs by PamGene multiplex kinase activity profiling. * mean specificity score ≥ 1, ** mean specificity score ≥ 1 and mean significance score ≥ 0.5. B Human neutrophils were purified and unstimulated cells (unstim) plus cells stimulated for 6 h with COL7 anti-COL7 IgG1-IC (6 h IC) were analyzed by bulk RNAseq. Significantly differentially expressed genes among CDKs, cyclins, and selected regulators were determined and log2 fold changes (log2 FC) are displayed for the individual samples. C Expression in unstimulated and stimulated neutrophils was compared for the individual CDKs on the basis of DESeq2-extracted normalized counts. D Top 20 GO activated and suppressed pathways after gene set enrichment analysis of genes significantly differentially expressed upon IC stimulation. Adjusted p value (p. adjust) < 0.1. Gene ratio equals the ratio of differentially expressed genes in the given GO term, count refers to the absolute number. n = 3

qRT-PCR confirms the expression of 14 CDKs from RNAseq in human PMNs

To validate the results from RNAseq, an independent qRT-PCR experiment was performed on unstimulated and IC-stimulated human PMNs. Data are provided in Additional file 1. All tested CDKs were detected, although CDK1, CDK3, CDK5, CDK6, CDK8, CDK11A, and CDC7 were expressed at a relatively low level (2^−ΔΔCt < 0.1) under both conditions. The expression of CDK2 and CDK4 was increased more than threefold (p = 0.0159) or sixfold (p = 0.0079), respectively, after IC stimulation, while CDK9 expression decreased by half (p = 0.0317) and CDK19 expression by approximately 70% (p = 0.0159). CDK3 and CDK11B showed a trend to be reduced (both p = 0.0556), but all other CDKs remained unaffected by IC stimulation (Fig. 2A). The results support the RNAseq data which also indicate an IC-mediated induction of CDK2 and CDK4 transcription, but a reduction of CDK9 and CDK19 transcription (see section above). For comparison with cells which still have the potential to enter the cell cycle and proliferate, resting B and CD4⁺ T cells were also analyzed in qRT-PCR (Fig. 2B). Strikingly, CDK1, CDK2, CDK4, CDK5, CDK6, CDK8, and CDC7 were expressed at higher levels in B and/or T cells, whereas CDK4, CDK9, CDK11B, CDK14, and CDK19 were increased in PMNs.

Fig. 2.

CDK2 and CDK4 expression are induced upon IC stimulation in human PMNs. A Human PMNs were isolated, stimulated for 6 h with COL7 anti-COL7 IgG1-IC and subjected to qRT-PCR detecting 14 distinct CDKs and CDC7 using the the 2^−ΔΔCt method. B In parallel, B cells and CD4⁺ T cells were purified and analyzed in the same way for comparison with unstimulated PMNs. n = 5. Mann-Whitney test (A) or Kruskal-Wallis test with Dunn’s multiple comparisons test (B). * p ≤ 0.05. ** p ≤ 0.01. *** p ≤ 0.001

Pharmacological inhibition of multiple specific CDKs partially protects in vitro from neutrophil effector functions in a dose-dependent manner

To functionally evaluate the role of CDKs in neutrophil kinase signaling, we selected CDKi based on an extensive literature review and tested them in several functional neutrophil assays, covering multiple steps from neutrophil activation to cell death. Therefore, COL7-anti-COL7-IgG1 IC were immobilized on plates, incubated with PMNs from healthy donors and the respective experiment was performed. The surface expression of CD62L, an L-selectin which is shed upon neutrophil activation [78, 79], was analyzed first. Data are provided in Additional file 2. The CDK9i MC180925 (Fig. 3, p = 0.0014), CDK7i THZ1 2HCl (Fig. S2, p = 0.0057), and CDK1i Ro 3306 (Fig. S2, p = 0.0014) significantly reduced CD62L shedding, while the other inhibitors tested did not (Fig. 3, S1). CDK7 inhibition with THZ2 instead significantly reduced CD18 expression (Fig. 3, p = 0.0022), which is a β2-integrin responsible for neutrophil adhesion [80]. Similar effects were observed for Ro 3306 (p = 0.0076) and THZ1 2HCl (Fig. S3, p = 0.0066). Subsequently, adhesion measurements demonstrate a significant reduction only for THZ2 (Fig. 3, S4, p = 0.0022). Furthermore, THZ2 (p = 0.0004), MC180295 (p = 0.001), OTS964 (p = 0.0117), Ro 3306 (p = 0.0042), BMS-265,246 (p = 0.0494), K03861, palbociclib HCl (p = 0.0026), THZ1 2HCl (p = 0.0007), and BAY1251152 (p = 0.0101) inhibited ROS release (Fig. 3, S5). Regarding apoptosis, a significant increase was measured for THZ2 (p = 0.0009), OTS964 (p = 0.0301), and BSJ-4-116 (p = 0.0177, Fig. 3, S6). For BMS-265,246 (p = 0.0164), statistical significance was reached, although the values did not exceed those of the negative controls (Fig. S6). However, when looking at the absolute numbers, there are no statistically significant differences. Therefore, the small average percentage of apoptotic cells, which is only up to 4%, is unlikely to be biologically relevant (Fig. S7). Although viability was significantly lower, especially at 10 µM (THZ2 p = 0.001, OTS964 p = 0.0135, MC180295 p = 0.0386, BMS-265246 p = 0.0707, THZ1 2HCl p = 0.0176, BSJ-4-116 p = 0.0231, Ro 3306 p = 0.0494), more than 99% of the cells remained viable in all cases (Fig. 3, S8). Taken together, CDK inhibition impairs various aspects of IgG-induced neutrophil activation, with ROS release being significantly reduced across multiple inhibitors at doses not substantially affecting viability. Thus, the inhibitory potential of THZ2, MC180295, and OTS964, in particular, is highlighted.

Fig. 3.

CDK7, CDK9, and CDK11 inhibition attenuate neutrophil effector functions and trigger apoptosis. Different neutrophil functions and characteristics were analyzed after COL7 anti-COL7 IgG1-IC stimulation of human PMNs for 2 h and inhibition with three CDKi: (A) CD62L shedding, (B) CD18 expression, (C) adhesion, (D) ROS release, (E) apoptosis, and (F) viability. The results were normalized to the positive control stimulated with IgG1-IC but without inhibitor indicated by the red line. The dose-response curves were calculated and the IC₅₀/EC₅₀ values are indicated if in the tested concentration range. Cells only, antigen only (Ag), and antibody only (Ab) served as negative controls. Kruskal-Wallis test with Dunn’s multiple comparisons test was performed. * p ≤ 0.05. ** p ≤ 0.01. *** p ≤ 0.001. For apoptosis experiment using THZ2 n = 6, rest n = 5. AUC, area under the curve

Both systemic and topical treatment with CDK7, CDK9 or CDK11 inhibitors reduces disease severity in the local antibody-transfer induced EBA model

Following the successful in vitro validation of the selected CDKi, a local antibody-transfer induced EBA model was performed to test these compounds as systemic treatment (Fig. 4A). Data are provided in Additional file 3. CDK9 inhibition by MC180295 completely prevented disease induction (p = 0.0004 on day 3), which is exemplified by representative images showing the ameliorated clinical presentation of the mice and the reduced inflammation in the H&E staining despite IgG and C3 deposition along the dermal-epidermal junction (Fig. 4B). The AESA was significantly reduced starting at day 1 and increasing further until day 3, which is also reflected by the H&E score (Fig. 4C, p < 0,0001). CDK7 inhibition by both THZ2 (Fig. 4D) and THZ1 2HCl (Fig. 4E) reduced the AESA (p = 0.0488 and p = 0.0008 on day 3, respectively), but only THZ2 treatment significantly decreased the H&E score (p = 0.0031). Systemic CDK11 inhibition by OTS964 treatment also ameliorated the disease severity (p = 0.0328), but the H&E score did not change significantly between the treated and the control mice (Fig. 4F, p = 0.3732). However, treatment with Ro 3306 and palbociclib HCl had no significant effect (p = 0.2422, Fig. S9). To test which immune cells are infiltrating the skin, Ly6G and F4/80 stainings were performed on ear sections from MC180295-treated and control mice. The stainings showed a deposition of Ly6G⁺ neutrophils along the dermal-epidermal junction in control mice, which was absent in treated mice (p < 0.0001). F4/80⁺ macrophages were observed in the dermis of control mice, and to a lesser extent, in treated mice (Fig. 4G, p = 0.0001).

Fig. 4.

Systemic and topical CDK7i, CDK9i, and CDK11i treatment reduces disease severity in local EBA. A EBA was induced by injection of anti-mCOL7^c IgG and the mice were treated with CDKi either systemically or topically. B Exemplary clinical and histological images are shown with the scale bars corresponding to 100 µM. Asterisks indicate cell infiltrates and arrows indicate binding of C3/IgG to the dermal-epidermal junction. C The affected ear surface area (AESA) and the H&E score were evaluated in the systemic treatment group for MC180295, (D) THZ2, (E) THZ1 2HCl, and (F) OTS964. G Stainings for Ly6G and F4/80 were performed on ear sections from mice systemically treated with MC180295 and control mice. The scale bar refers to 100 μm. Ly6G or F4/80 are depicted in red, DAPI in blue. H For the topical model, AESA was analyzed as well. Two-way ANOVA with Šidák’s multiple comparisons test was performed for AESA evaluation, except for the systemic treatment with THZ2 and MC180295 a mixed-effects analysis with Šidák’s multiple comparisons test was conducted. For H&E scores and stainings, Mann-Whitney test was carried out. * p ≤ 0.05. ** p ≤ 0.01. *** ≤ p 0.001. **** p ≤ 0.0001. n = 8

The most promising inhibitors from the systemic treatment in local antibody-transfer induced EBA were selected for topical treatment in the same model. THZ2 was chosen over THZ1 HCl since both inhibit CDK7, but THZ2 showed better results in vitro and almost completely protected from disease development in vivo. All three inhibitors tested prevented disease induction when applied topically (Fig. 4H).

CDK9 inhibition with MC180295 is beneficial in neutrophil-dependent KBxN STA, but not in neutrophil-independent ITP

The CDKi that showed the most prominent effect in the local antibody-transfer induced EBA model, MC180295, was validated in two other IC-driven models that differ regarding the role of neutrophils in their respective pathogenesis, namely KBxN STA and ITP. Data are provided in Additional file 4. In KBxN STA (Fig. 5A), systemic treatment with MC180295 significantly reduced the disease score of paw swelling (p < 0.001 on day 7, Fig. 5B). This was reflected in histological analysis showing reduced proteoglycan degradation and thus cartilage damage as compared to control mice upon staining with Safranin-O. H&E staining revealed pronounced immune cell infiltration in the controls, but not or to a lesser extent in the treated mice as observed in representative images (Fig. 5C). Out of 13 cytokines tested in the serum, two were significantly changed on the final day between the two groups. Both IL-6 (p = 0.0159) and IL-10 (p = 0.0317) were reduced on day 7 in the MC180295-treated group compared to the control group (Fig. 5D). All other measured cytokine levels were not significantly altered over the course of the observation period of 7 days (Supplemental Table 2).

Fig. 5.

CDK9 inhibition with MC180295 improves disease severity in KBxN STA, but not in ITP. A To induce KBxN STA, mice were injected with arthritic KBxN serum and treated with MC180295 for seven days. B The course of the disease score is depicted and (C) reflected by the clinical presentation of the paws and the histological stainings. The scale bars represent 100 μm. Asterisks indicate proteoglycan staining (Safranin-O) or cell infiltration (H&E). D Cytokine levels were measured in serum and significantly changed cytokines are shown. E Stainings for Ly6G and F4/80 were performed on paw sections from mice systemically treated with MC180295 and control mice. The scale bar refers to 100 μm. Ly6G or F4/80 are depicted in red, DAPI in blue. F ITP was induced by anti-platelet IgG2c-induced platelet depletion and MC180295 was applied at 24 h and 4 h before depletion. G Absolute and (H) relative platelet counts were monitored for up to 24 h and values normalized to 100% at 0 h are indicated. B, G, H For arthritis score and platelet counts, two-way ANOVA with Šidák’s multiple comparisons test was performed. (D, E) For cytokine levels and stainings, Mann-Whitey test was conducted. * p ≤ 0.05. ** p ≤ 0.01. **** p ≤ 0.0001. (A-E) n = 5 (treatment) and n = 4 (control). (F-H) n = 4

To investigate which immune cells infiltrate the tissue, Ly6G and F4/80 stainings were performed, revealing the reduced infiltration of neutrophils in treated mice compared to control mice (p = 0.0303). However, macrophages were not significantly affected by the treatment with MC180295 (Fig. 5E).

In contrast to KBxN STA, in ITP (Fig. 5F), where neutrophils are not central to disease pathogenesis, prophylactic treatment with MC180295 did not show any effects on platelet count 24 h after platelet depletion (Fig. 5G-H), demonstrating the variable effect of CDKi in the different disease models.

Discussion

IC-induced FcγR-mediated kinase signaling plays a major role in intracellular signal transduction by neutrophils under pathological conditions, e.g., in autoimmune diseases such as EBA or RA. However, the role of CDKs in neutrophils is largely unknown. A small number of studies have reported the expression of CDKs on protein level and suggested functional relevance in these cells [22, 28, 81]. To the best of our knowledge, we show in the present study for the first time a comprehensive, functional analysis of IC-triggered FcγR-dependent kinase signaling using a variety of selective CDKi. Figure 6 summarizes our main findings.

Fig. 6.

Successful inhibition of neutrophil functions and experimental autoimmune models using CDKi. Out of 20 CDKs that are described in humans, 14 were detected in our gene expression analysis. After choosing selective CDKi, different in vitro neutrophil functional assays were performed. The number of effective inhibitors is indicated. Subsequently, all CDKi that inhibited ROS release were tested as systemic treatment in the local antibody transfer-induced EBA model, except for BAY1251152 to avoid redundancy. Three CDKi reduced the disease severity and were thus applied topically in the same model. Additionally, the most promising CDKi was validated in KBxN STA and ITP. Created in https://BioRender.com

Since phosphorylation is the most important regulatory mechanism for kinase activity, we performed an activity-based kinome profiling for CDKs and identified almost all CDKs to be active in human IC-stimulated neutrophils. Interestingly, we observed multiple CDKs as expressed at RNA level, with a major overlap to the kinase activity profiling. Several transcriptional CDKs including CDK7 were substantially upregulated, supporting the relevance of these CDKs in neutrophils. Those kinases that were not detected in the activity profiling were also not expressed based on the RNAseq data, except CDK15. This discrepancy might be explained by the high structural and functional similarity of the CDKs. Furthermore, we found that CDK transcription may change in IC-stimulated neutrophils. The expression of the two traditional cell-cycle CDKs CDK2 and CDK4 increased, whereas the expression of the two transcriptional CDKs CDK9 and CDK19 decreased. Concerning CDK9, it has been shown that inhibition leads to the reactivation of epigenetically silenced genes, e.g. SYNE1, GLDC, MUC20, ANPEP, SUSD4, and COLEC11 [82]. All of these genes have been linked to the immune system [83], e.g. NETosis [84–86], apoptosis [87], complement inhibition [88], or neutrophil degranulation [89]. Therefore, CDK9 inhibition might further influence these processes.

The trends found in RNAseq are clearly supported by our qRT-PCR analysis in an independent dataset. Furthermore, the qRT-PCR data highlight the elevated expression of CDK9, 11B and 19 in neutrophils compared to B and CD4⁺ T cells, making them especially promising targets in neutrophil-dependent diseases. The tendencies concerning gene expression levels in unstimulated cells found in our RNAseq and qRT-PCR analyses are in line with publicly available RNAseq data of unstimulated neutrophils from six healthy donors sorted by flow cytometry ([72], Human Protein Atlas proteinatlas.org]). Several CDKs like CDK2, CDK4, and CDK11 were also detected in a kinase activity measurement of IC-stimulated PMNs [59].

Since CDKs only become fully active after binding to their cyclin binding partners, we also examined cyclins in our RNAseq analysis. Most prominently, we found both CDK7 and its binding partner cyclin H significantly increased upon IC activation. Strikingly, MNAT1 (MAT1 RNA) was not detected, suggesting the requirement of a phosphorylation-dependent activation of CDK7 [90, 91].

Furthermore, several regulators of CDKs were differentially expressed upon IC stimulation in human neutrophils, hence adding another layer of complexity to the regulation of CDKs in autoimmune disease. Both MED12 and MED13 were transcribed more, possibly promoting the formation of a transcriptional pre-initiator complex after association with CDK8/19 and cyclin C [92]. Moreover, regulators of chromatin remodeling required for transcriptional activation like RB1 [93] and EP300 [94, 95] were detected at higher RNA levels in stimulated cells.

In addition to the analysis focusing on CDKs and their regulators, we performed a GO pathway analysis for the enriched gene sets from unstimulated versus IC-stimulated neutrophils. Interestingly, our pathway analysis revealed that many metabolic pathways including ATP synthesis or catabolic processes are suppressed, while immune pathways such as granulocyte migration, chemokine and cytokine pathways were activated.

Thus, we hypothesized that targeting CDKs by selective small-molecule inhibitors might reduce neutrophil effector functions in IC-stimulated neutrophils in vitro and consequently disease severity in IC-driven neutrophil-dependent models of autoimmune disease in vivo. Although there are multiple studies published on CDK inhibition in different contexts, they rarely focus on neutrophils and utilize relatively unselective CDKi. However, pan-CDK inhibition is often associated with severe side effects [96]. Here, we therefore chose more selective CDK inhibitors to investigate their role on neutrophil functions upon IC activation.

Hence, we analyzed several functions that are critical for neutrophil recruitment as well as apoptosis. As multiple studies suggest that ROS release might be a predictor of in vivo efficacy in experimental EBA [59, 60, 64, 97], all CDKi that significantly reduced ROS release were tested as systemic treatment in the local antibody transfer-induced EBA model (Ro 3306, palbociclib HCl, THZ1 2HCl, THZ2, OTS964). Moreover, inhibition of CDK1, CDK2, CDK4/6, CDK7, CDK9, or CDK12 influenced other neutrophil functions such as expression of the surface activation markers CD18 and CD62L, and adhesion, highlighting a role for CDKs in neutrophil recruitment and activation. Viability was only slightly influenced, with values still in the range of the negative controls. However, some inhibitors (THZ2, OTS964, BSJ-4-116) increased apoptosis, which is a highly controlled form of cell death leaving the cell membrane intact [25]. In line with our data, a role in regulating apoptosis has already been suggested for CDK7 in neutrophils [22], and for CDK11 in Chinese Hamster Ovary cells and T-cell lines [98]. Concerning neutrophils, this may protect the tissue from toxic substances like ROS which are released during alternative forms of cell death and has been described as a potential mechanism of action of CDKi [81, 99]. Indeed, THZ2 and OTS964 increased apoptosis while reducing ROS. However, other inhibitors reduced ROS release without affecting apoptosis (Ro 3306, K03861, palbociclib HCl, THZ1 2HCl, BAY1251152), implicating the involvement of different mechanisms or signaling pathways. Interestingly, CDK9 inhibition by MC180295 or BAY1251152 did not significantly influence apoptosis, whereas CDK9 inhibition by promiscuous inhibitors such as flavopiridol or R-roscovitine resulted in neutrophil apoptosis induction [22, 30]. Thus, this indicated an inhibitor-specific, probably synergistic effect caused by targeting several CDKs at once.

Subsequently, the most promising CDKi were applied systemically in the local antibody-transfer induced EBA model. Here, inhibition of CDK7 by THZ1 2HCl or THZ2, CDK9 by MC180295, or CDK11 by OTS964 were found to significantly reduce the disease severity. Differences in potency between THZ1 2HCl and THZ2 are likely due to pharmacokinetic differences [100]. Remarkably, CDK7, CDK9, and CDK11 are all described as transcriptional kinases [3]. This is in line with our expectation considering that neutrophils are non-proliferating cells. Inhibition of CDK1 with Ro 3306 or CDK4/6 with palbociclib HCl, in contrast, did not reduce the AESA. Although palbociclib HCl showed promising effects in vitro and is licensed as treatment for breast cancer, where it improves the neutrophil-to-lymphocyte ratio in combination with the aromatase inhibitor letrozole [101], other data indicate an opposite role for CDK6 in regulating NET formation as CDK6^−/− mice failed to produce NETs upon PMA stimulation [28]. Additionally, CDK6^−/− mice exhibited a higher amount of primary granule proteins, suggesting CDK6 as a regulator of neutrophil granularity [102]. These data might explain the outcome seen in EBA.

The most effective CDKi, MC180295, was next administered in two other models to verify the neutrophil-dependent effects: KBxN STA, which is neutrophil- and macrophage-dependent, and ITP, which is mostly neutrophil-independent. In KBxN STA, MC180295 significantly reduced disease severity and levels of several inflammatory cytokines. These findings complement a study describing protective effects on human chondrocytes and cartilage explants by CDK9 inhibition with flavopiridol or small interfering RNA. Less pro-inflammatory cytokines are released, preventing cartilage degradation [32]. Of note, recent data show the induction of neutrophil reverse migration from the inflammation site back to the circulation upon CDK9 inhibition in zebrafish larvae, hence contributing to the resolution of inflammation [103].

Nevertheless, compared to the almost complete protection from disease development observed in the local antibody transfer-induced EBA model, the MC180295-treated KBxN STA mice showed a residual mean arthritis score of 2 on the final day, representing still 22% of that observed in the control mice, in contrast to only 5% in the EBA model. This finding might reflect the different functional relevance of neutrophils in the two distinct models. While EBA is described to be primarily neutrophil-driven [38], neutrophils also play a critical [45], yet not exclusive, role in KBxN STA [47]. In the latter model, the function of macrophages, NK cells, mast cells, and synovial fibroblasts during the effector phase of the disease is also well established [44] and might explain the differences in efficacy seen on both models. A study conducted with LPS-stimulated human monocyte-derived macrophages demonstrated a reduction of IL-6 and TNF after treatment with the more promiscuous CDK9i AT7519 or roscovitine, supporting a role for macrophages in the resolution of inflammation [104]. In addition, administration of the CDK9i PC585 and PC579 in murine arthritis models led to the downregulation of the anti-apoptotic protein MCl-1, thus favoring apoptosis [105]. As delayed apoptosis is a feature of RA [46], CDK inhibition might be a reasonable approach in this context.

By staining for neutrophils and macrophages, we were able to show that neutrophil infiltration is reduced in MC180295-treated mice compared to control mice in EBA and STA, while macrophage infiltration is only diminished in EBA. This finding underlines the importance of neutrophils in these models, but highlights also the involvement of macrophages.

Conversely, no effects on platelet depletion or recovery were observed in ITP. Contrasting with EBA and KBxN STA, neutrophils are considered dispensable for the effector phase of ITP. Rather other immune cells like tissue-resident macrophages are involved [51–53]. This highlights the role for CDK9 in IC-induced neutrophil-dependent diseases like EBA and KBxN STA.

Regarding potential limitations of this study, only one concentration of each inhibitor was used in vivo. Thus, any effects which might be seen at different concentrations remain unclear. However, these concentrations had successfully been used in mouse models before and were hence deemed safe to apply. For example, MC180295 demonstrated an anti-tumoral effect in NOD.Cg-Prkdc^scidIl2rg^tm1Wjl/SzJ mice lacking mature B and T cells [82] and palbociclib is already in clinical use for the treatment of metastatic breast cancer [106]. Furthermore, our in vitro data indicate a mechanistic effect of CDKi rather than a mere cytotoxic effect.

Moreover, the effect of the CDKi on murine CDKs compared to human CDKs is not always known. This is exemplified by the CDK1i Ro 3306, which is effective in reducing ROS release from human PMNs but does not affect disease severity in the murine EBA model. Notably, CDK1 was not detected at the level of gene expression or activity, suggesting that the observed effects in vitro are also due to selectivity for other targets.

Additionally, only one inhibitor per target (except for CDK7 since both inhibitors were equally promising) was tested. We are aware that other inhibitors with different selectivity or bioavailability might be more effective, nevertheless the use of further inhibitors would have exceeded the scope of this explorative study. The way is now paved for further studies grounding on this work.

Instead of pharmacological inhibition, genetic knockouts could be a more precise alternative to address our research questions. Unfortunately, genetic deletions of many CDKs are embryonically lethal, with CDK1 being indispensable for early embryogenesis [107, 108]. CDK4- and CDK6-deficient mice display problems in embryonic development [109, 110] similar to CDK8^−/− [111], CDK11^−/− [112] or CDK12^−/− mice [113]. Therefore, we consider pharmacological inhibition to be a practicable and clinically relevant alternative, though future studies with neutrophil-specific inducible depletion may be informative.

Conclusions

Taken together, we have demonstrated the functional relevance of CDKs including CDK7, CDK9, and CDK11 in neutrophil IC-induced kinase signaling both in vitro and in vivo. Our findings contribute to the exploration of potential novel treatment options for autoimmune diseases such as PD or RA.

Abbreviations

AESA

Affected ear surface area

AUC

Area under the curve

BSA

Bovine serum albumin

CAK

CDK–activating kinase

CDK

Cyclin–dependent kinase

CDKi

Small–molecule CDK inhibitor

CKI

Intrinsic CDK inhibitor

COL7

Collagen VII

DMSO

Dimethyl sulfoxide

EBA

Epidermolysis bullosa acquisita

ENA

European Nucleotide Archive

FcγR

Fragment crystallisable gamma receptor

GO

Gene Ontology

H&E

Hematoxylin and eosin

i.p.

Intraperitoneal

IC

Immune complex

ITP

Immune thrombocytopenia

KBxN STA

KBxN serum–transfer arthritis

MACS

Magnetic activated cell sorting

NET

Neutrophil extracellular trap

PBS

Phosphate–buffered saline

PD

Pemphigoid disease

PMN

Polymorphonuclear granulocyte

p.o.

Per os

qRT-PCR

Real–time quantitative reverse transcription polymerase chain reaction

RA

Rheumatoid arthritis

RNAseq

Bulk RNA sequencing

ROS

Reactive oxygen species

RT

Room temperature

TBS

Tris–buffered saline

Authors’ contributions

Investigation: MS, LV, APM, DM, SJ, MR, AKS, SMV, YW, HG, JL, writing (original draft): MS, visualization: MS, LV, AKS, formal analysis: CO, NE, AV, MS, AKS, LV, KA, resources: RV, GV, FP, XY, RJL, KTA, conceptualization: RJL, AL, KB, supervision: AL, KB, KTA, software: AV, project administration: AL, KB, writing (review & editing): all.

Funding

Open Access funding enabled and organized by Projekt DEAL. This research was funded by the Cluster of Excellence “Precision Medicine in Chronic Inflammation” (EXC 2167), the Research Training Group “Defining and Targeting Autoimmune Pre-Disease” (RTG 2633), and the Collaborative Research Center “Pathomechanisms of Antibody-mediated Autoimmunity” (CRC 1526), all from the Deutsche Forschungsgemeinschaft and the Schleswig-Holstein Excellence-Chair Program from the State of Schleswig Holstein.

Data availability

The RNAseq dataset generated and analyzed during the current study is available in the European Nucleotide Archive (ENA) at EMBL-EBI under accession number PRJEB103987 ( [https://www.ebi.ac.uk/ena/browser/view/PRJEB103987](https:/www.ebi.ac.uk/ena/browser/view/PRJEB103987)), all other data generated or analyzed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

Whole blood was obtained from healthy volunteers after written informed consent. All experiments using human samples were approved by the local ethics committee (AZ #20–463, University of Lübeck, Lübeck, Germany) and were performed in accordance with the Declaration of Helsinki. Animal experiments were approved by local authorities of the Animal Care and Use Committee (government of Schleswig-Holstein and government of Lower Franconia) and the Institutional Animal Care and Use Committee at Rush University. All experiments were performed by certified personnel (AZ 108/08-15, AZ 97-11/20 and 55.2.2-2532-2-2085) following the ARRIVE guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.