The Neurobeachin‐like 2 protein (NBEAL2) controls the homeostatic level of the ribosomal protein RPS6 in mast cells

The Beige and Chediak‐Higashi (BEACH) domain‐containing, Neurobeachin‐like 2 (NBEAL2) protein is a molecule with a molecular weight of 300 kDa. Inactivation of NBEAL2 by loss‐of‐function mutations in humans as well as deletion of the Nbeal2 gene in mice results in functional defects in cells of the innate immune system such as neutrophils, NK‐cells, megakaryocytes, platelets and of mast cells (MCs). To investigate the detailed function of NBEAL2 in murine MCs we generated MCs from wild type (wt) and Nbeal2−/− mice, and deleted Nbeal2 by CRISPR/Cas9 technology in the murine mast cell line MC/9. We also predicted the structure of NBEAL2 to infer its function and to examine potential mechanisms for its association with interaction partners by using the deep learning‐based method RoseTTAFold and the Pymol© software. The function of NBEAL2 was analysed by molecular and immunological techniques such as co‐immunoprecipitation (co‐IP) experiments, western blotting, enzyme‐linked immunosorbent assay and flow cytometry. We identified RPS6 as an interaction partner of NBEAL2. Thereby, the NBEAL2/RPS6 complex formation is probably required to control the protein homeostasis of RPS6 in MCs. Consequently, inactivation of NBEAL2 leads to accumulation of strongly p90RSK‐phosphorylated RPS6 molecules which results in the development of an abnormal MC phenotype characterised by prolonged growth factor‐independent survival and in a pro‐inflammatory MC‐phenotype.

Pleckstrin homology (PH), a Beige and Chediak-Higashi (BEACH) module and a seven-bladed WD40 repeat domain (WDR) [2].The structural diversity, the expression of several NBEAL2 variants as well as the localisation in different subcellular compartments [3] indicate that NBEAL2 is involved in a multiplicity of cellular processes such as protein sorting, signal transduction, protein biosynthesis and/or protein degradation.This is reflected by the fact that NBEAL2 inactivation in mice and humans results in diverse functional defects in different cell types of the haematopoietic system such as the instability of α-granules in megakaryocytes and platelets [4,5], abnormalities in the phenotype and function of neutrophils and NK cells, as well as to a decreased expression of CTLA-4 on T-cells [6,7].These functional defects characterise the Gray platelet syndrome (GPS) and lead to splenomegaly, macrothrombocytopenia, increased susceptibility to infections or to the development of autoimmunity [3,[5][6][7][8].We recently found that NBEAL2 deficiency in murine mast cells (MCs) results in a paucity of granules and the stabilisation of transcription factors which are essential for MC differentiation [9].MCs are located in peripheral tissues and are one part of the first line of defence in the immune system.Under physiological conditions the activation of MCs by the IgE/FcεRI axis or by alarmins such as IL-33 and/or ATP [10] in absence or presence of growth factors [11][12][13][14][15][16][17][18] results in degranulation and thus in the release of histimines and proteases as well as to the de novo synthesis of cytokines and chemokines which mediate neutrophile attraction of to the side of infection [13,19].However, under pathophysiological conditions, the expression of constitutively active c-Kit mutants and/or of defective regulators of the transcription machinery results in the development of mastocytosis or MCs tumours [20][21][22].Therefore, the controlled differentiation and activation of MCs is essential to prevent the formation of abnormal and hyperinflammatory MC phenotypes [22].In this work, we describe a NBEAL2-mediated mechanism which prevents the development of abnormal MC phenotypes.Thereby, NBEAL2 inactivation in MCs results in a prolonged growth factor-independent survival and a potentiated production of pro-inflammatory cytokines.These characteristics are caused by a defective protein degradation machinery which results in accumulations of proteins which are involved in the protein biosynthesis such as p90RSK and RPS6.In summary, our data show that functional intact NBEAL2 is essential to prevent the development of abnormal MC phenotypes by balancing the protein level of the p90RSK-RPS6 module.

Bone marrow-derived mast cell generation
Bone marrow (from the femurs and tibias) was obtained from age (8-13 weeks) and sex matched wt, Nbeal2 +/À and Nbeal2 À/À mice.Cells were cultured in IMDM (Gibco) (with 10% FCS, 100 U/mL penicillin, 100 mg/mL streptomycin, 50 mM 2-mercaptoethanol) (complete IMDM) supplemented with 20 ng/mL IL-3 (conditioned media from X63Ag-653 BPV-rmIL-3).For removal of adherent cells from the cultures, suspension cells were collected by centrifugation and resuspended in fresh media every second day within the first 4 weeks.After these 4 weeks, cultures were tested for the percentage of FcεRI + /c-Kit + cells.When cultures contained 95% FcεRI + /c-Kit + bone marrow-derived mast cells (BMMCs), cultures were used for experiments during the subsequent 4 weeks.In this time, medium was refreshed twice a week.

Expansion of peritoneal cavity-derived MCs
Peritoneal cavities of wt and Nbeal2 À/À mice were flushed with ice cold PBS and cells were collected by centrifugation.For expanding peritoneal cavity-derived mast cells (PCMCs), isolated peritoneal cavity cells were cultured in complete IMDM (Gibco) supplemented with murine SCF and IL-3 (both 10 ng/mL) (Peprotech) for 21 days.Purity of PCMCs was determined by staining for c-Kit + /FcεRI + cells.PCMC cultures with a purity of 90%-95% c-Kit + /FcεRI + cells were used for experiments.Pooled peritoneal cells from five mice represented one biological replicate.Therefore, three biological replicates consist of 15 wt or 15 Nbeal2 À/À mice.

Deletion of NBEAL2 in HEK293 and MC/9-cells by CRISPR/Cas9
For deletion of NBEAL2 in cell lines we used the lenti-CRISPR v2 plasmid (Addgene #52961) containing target-specific gRNAs.Forward and reverse oligos encoding the respective gRNA were annealed and ligated into the BsmBI-restricted lentiCRISPR v2 vector.For the knockout (KO) of NBEAL2 in HEK293 cells, parental cells were seeded (500 000 cells/6-well) in DMEM (Sigma) without antibiotics (supplemented with 10% FCS) and were transiently transfected with a mixture of four different gRNA constructs targeting human (h) NBEAL2 by using PEI reagent (Sigma-Aldrich, 408 727, diluted to 10 μg/mL, pH 7.2, adjusted with HCl).On Day 4 after transfection, cells were selected by using puromycin (1 μg/mL) (Sigma-Aldrich #P8833) in DMEM (Sigma) (supplemented with 10% FCS, 100 U/mL penicillin, 100 μg/mL streptomycin, 50 mM β-mercaptoethanol) (complete DMEM medium) for 1 week.After selection, single cell clones were grown and NBEAL2 deletion was evaluated by western blotting.We randomly chose one NBEAL2 knockout (KO) HEK293 clone for transfection experiments.As control cells, we used a HEK293 cell pool transfected with lentiCRISPR v2 without the insertion of a gRNA and selected with puromycin in complete DMEM (Sigma).For deletion of murine Nbeal2 in the murine MC/9 mast cell line, we generated viral particles in HEK293T cells.The viral particle-containing supernatants were used to infect parental wt MC/9 cells [24].After three rounds of infections cells were selected by using puromycin (1 μg/mL) in RPMI (Sigma) (with 10% FCS, 100 U/mL penicillin, 100 mg/mL streptomycin, 50 mM β-mercaptoethanol) (complete RPMI medium) supplemented with 20 ng/mL IL-3 (conditioned medium from X63Ag-653 BPV-rmIL-3 cells) for 1 week.Single cell clones were generated (cultured in complete RPMI medium, supplemented with 20 ng/mL IL-3) and were analysed for NBEAL2 expression by western blotting.Randomly selected Nbeal2-KO MC/9 were used for further experiments.As control cells, a puromycin selected cell pool of MC/9 cells transduced with virus prepared from lentiCRISPR v2 without the insertion of a gRNA was used.

HEK293 cell transfection
NBEAL2-KO HEK293 cells were seeded (1.5 Â 10 6 cells in six-well plates) in DMEM (Sigma) without antibiotics (supplemented with 10% FCS and 50 mM β-mercaptoethanol).The next day, cells were transfected with different NBEAL2 constructs such as PBW-NBEAL2 (a kind gift from Dr. J. A. Gurrero; Department of Haematology, University of Cambridge, Cambridge, United Kingdom), PB-NBEAL2, BW1-NBEAL2 or BW2-NBEAL2 by using the Lipofectamin 2000 (Thermo Scientific) standard transfection protocol for adherent cells with OptiMem (Thermo Scientific).Twenty-four hours later, cells were lysed and the protein concentration was determined by using the BCA-kit (Pierce).

Immunoprecipitation
Lysates from transfected NBEAL2-KO HEK293, wt or Nbeal2 À/À BMMCs were treated overnight with an anti-RPS6 (sc-74 459) (5 μg/500 μg total protein content in the samples) (Santa Cruz Biotechnology) or the mouse isotype control (#31903) (ThermoFisher).The next day washed protein-G Sepharose 4B beads (#101243) (ThermoFisher) were added and incubated for 4 h at 4 C on a turning wheel.Afterwards beads were washed with lysis butter and subsequently with PBS and were then boiled in 6Â Laemmli sample buffer.

Survival experiments
By seeding in IL-3-free complete IMDM (Gibco) or RPMI (Sigma) medium, BMMCs, PCMCs or MC/9 cells (1 Â 10 6 cells/ml) were IL-3 starved.Cells were then left unstimulated or were stimulated with murine IL-3 (as indicated in the figures).Afterwards, cells were treated with vehicle (DMSO), inhibitors (as indicated in the figures) (all Selleckchem) or were left untreated.After the indicated time points (shown in the figures) cells were harvested, treated with trypan blue (#15250061) (ThermoFisher) for exclusion of dead cells and counted using a Neubauer chamber.Furthermore, harvested cells were treated with propidium iodide (PI) (#421301) (BioLegend) to evaluate the percentage of dead cells.Samples were recorded with the LSR II flow cytometer (BD) and analysed by FlowJo 10 (BD).

Flow cytometry
BMMCs and PCMCs were treated with rat-IgG (Jackson) to block unspecific antibody binding.Afterwards, cells were stained with PE-conjugated anti-c-Kit (#105808) and FITC-conjugated anti-FcεRI (#134306) (all BioLegend).To exclude dead cells, PI was added to the stained samples 5 min prior to analysis.Samples were recorded with the LSR II flow cytometer (BD) and analysed by FlowJo 10 (BD).

RNA sequencing and analysis
Total RNA from wt or Nbeal2 À/À BMMCs was isolated by using Qiazol reagent (Qiagen).Sequencing of RNA samples was performed using Illumina's next-generation sequencing methodology [25].In detail, total RNA was quantified and quality checked using Bioanalyzer 2100 instrument in combination with RNA 6000 nano kit (both Agilent Technologies).Libraries were prepared from 500 ng of input material (total RNA) using NEB-Next Ultra II Directional RNA Library Preparation Kit in combination with NEBNext Poly(A) mRNA Magnetic Isolation Module and NEBNext Multiplex Oligos for Illumina (Unique Dual Index Primer Pairs) following the manufacturer's instructions (New England Biolabs).Quantification and quality check of libraries was done using a Tapestation 4200 instrument and a DNA 1000 assay (Agilent Technologies).Libraries were pooled and sequenced using a NovaSeq 6000 SP 100 cycle runs.System runs in 101 cycle/single-end/standard loading workflow mode.Sequence information was converted to FASTQ format using bcl2fastq v2.20.0.422.Sequencing resulted in around 62 million reads per sample.Mapping of reads was done to GRCm38.100(PMID: 29155950) using Top Hat (v2.1, parameters: --no-convert-bam-nocoverage-search -x 1 -g 1) [26].The reads per gene were counted using feature Counts (v1.6.5, parameters: -s 2) [27].The counts were imported into the statistical computing environment R (v.3.6.3).R package DESeq2 v1.26 [28] was applied in order to find differentially expressed genes (DEG).For each gene, a p-value was calculated using Wald significance test including correction for multiple testing following Benjamini and Hochberg.DEGs were defined by corrected p < 0.05.

Western blot quantifications
All western blots were performed at least three times.Thereby, three western blot experiments were performed with samples from three biological replicates (three different BMMC or PCMC cultures derived from wt or Nbeal2 À/À mice).For western blot quantifications, we used ImageJ (National Institute of Health, USA).For protein expressions or stability experiments the intensities of protein bands were determined and normalised to the respective tubulin band.To determine the expression levels of proteins between wt and Nbeal2 À/À lysates the wt sample was set as 1.To compare protein stabilities in wt and Nbeal2 À/À MCs after inhibitor treatments, the vehicle-treated wt and Nbeal2 À/À controls were set as 1.In all cases, the fold increase or decrease of protein levels were calculated.

Statistical analysis
All experiments were performed at least three times (if not otherwise stated in the legends).For western blot or flow cytometry experiments, one representative experiment is shown.For quantifications, the summary of at least n = 3 biological replicates (if not otherwise stated in the legends) of wt and Nbeal2 À/À BMMCs is shown.For ELISA and survival experiments, we show the summary of the results obtained from at least n = 3 (if not otherwise stated in the legends) biological replicates.In addition, for ELISA experiments, every biological replicate was split into at least 6 technical replicates per condition of wt and Nbeal2 À/À MCs.One biological BMMCs replicate consists of pooled bone marrow cells from n = 2 mice.One biological replicate from wt and Nbeal2 À/-PCMCs consists of pooled cells obtained from the peritoneal cavity flushing of n = 5 mice.Shown is the mean ± SEM.For statistical analysis we used SigmaPlot 13.0 (Systat Software, Inc.).We used unpaired students t-test and statistical significance was accepted for p ≤ 0.05 (*p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ns: not significant).

Nbeal2 deficiency in MCs resulted in prolonged IL-3-independent survival and accumulated p90RSKs and RPS proteins
In this work, we aimed to determine the role of NBEAL2 in MCs.First, we characterised the used anti-NBEAL2 antibody.In lysates of wt BMMCs, we detected NBEAL2 in bands with different molecular weights among them the 300 kDa full length NBEAL2.However, all bands disappeared in Nbeal2 À/À BMMCs confirming the specificity of the used anti-NBEAL2 antibody (Figure S1A).The identity of these NBEAL2 bands is unknown.However, we speculate the expression of different NBEAL2 isoforms and/or the appearance of NBEAL2 degradation products in BMMCs.Given its easier detection we show the band between 180 kDa and 300 kDa to prove NBEAL2 deficiency in all following western blot experiments (Figure S1A; red frame).
Next, we investigated the influence of NBEAL2 deficiency on MC survival, which depends on IL-3.As shown in Figure 1a, NBEAL2 deficiency improved the survival in absence of IL-3 in BMMCs.Notably, these effects were dependent on the bi-allelic loss of Nbeal2 since Nbeal2 +/À BMMCs showed the same phenotype as wt BMMCs and died quickly in absence of IL-3 (Figure 1a).Confirming these data, we detected increased percentages of dead BMMCs in wt and Nbeal2 +/À compared to Nbeal2 À/À cultures (Figure 1b).Given the similar survival phenotypes of wt and Nbeal2 +/À BMMCs, we focused on the comparisons of wt and Nbeal2 À/À MCs in this work.Thereby, we found that in absence of IL-3, the c-Kit + /FcεRI + MCs disappeared in wt but not in Nbeal2 À/À cultures (Figure S1B) which was in accordance to the survival experiments.
BMMCs represent MCs with an immature phenotype.Therefore, we tested peritoneal cavity-derived MCs (PCMC).PCMCs are fully differentiated connective tissue MCs (CTMC), which were isolated from the peritoneal cavity of wt and Nbeal2 À/À mice.As already shown for wt and Nbeal2 À/À BMMCs, wt PCMCs died quickly in absence of IL-3 resulting in decreased percentages of dead cells if compared to Nbeal2 À/À PCMC cultures (Figure S1C,D).
Next, we aimed to determine the mechanism for this abnormal survival phenotype found in MCs with deleted Nbeal2.MC survival depends on RSKs, STATs, IKKs, PI3Ks, PKB/Akt and MAP-kinases such as ERK1/2 and p38 [17,[29][30][31].Notably, compared to wt BMMCs and wt PCMCs, we detected accumulated p90RSKs, p70S6K and of ribosomal proteins in Nbeal2 À/À BMMCs (Figure 1c,d) and PCMCs (Figure S1E).Due to the increased amount of these proteins, we also detected their increased phosphorylation in Nbeal2 À/À BMMCs (Figure 1c) and Nbeal2 À/À PCMCs (Figure S1E).In contrast to this, the protein levels of STAT5, IKK2/IκB, PKB/Akt, MAPkinases were comparable between wt and Nbeal2 À/À BMMCs (Figure 1e).Together, these data show that deletion of Nbeal2 resulted in dominant, abnormal MCs characteristics detectable in immature (BMMCs) and fully differentiated MCs (PCMCs).These phenotypes comprise a prolonged IL-3-independent survival and the accumulation of RSKs and of ribosomal proteins.However, these characteristics could be caused by the permanent absence of NBEAL2 in MCs, differentiated from the bone marrow or isolated and expanded from the peritoneal cavity from Nbeal2 À/À mice, rather than being a direct result from NBEAL2 deficiency.To clarify the direct role of NBEAL2 we used the IL-3-dependent MC-line MC/9, which is homologous to in vitro generated wt BMMCs [32,33].We deleted Nbeal2 in MC/9 cells by using the CRISPR/Cas9 technique and randomly selected two single cell Nbeal2-KO clones (clone 1 and 2 Nbeal2-KO MC/9).Compared to in vitro generated wt BMMCs, the pattern of western blot NBEAL2 bands was different in wt MC/9 cells with the prominent full length 300 kDa NBEAL2 band which disappeared in the Nbeal2-KO MC/9 cell clones (compare Figure S1A to Figure S1F).The presence of different NBEAL2 bands in BMMCs and MC/9 cells is unknown but could be explained by different sources of BMMCs (immature MCs generated from bone marrow) and MC/9 cells (mature MCs derived from the liver) [32,33].Nevertheless, the two selected Nbeal2-KO clones were further characterised.In presence of IL-3, the proliferation of control cells and Nbeal2-KO MC/9 clones was similar (Figure S1G; +IL-3).In contrast to this, the absence of IL-3 significantly decreased the proliferation of the wt controls, whereas the proliferation of the Nbeal2-KO MC/9 cell clones was unaffected (Figure S1G; ÀIL-3).Thereby, NBEAL2 deficiency was accompanied by a slightly increased level of RPS6 and the correlating increased phosphorylations (Figure S1H).Together, these data demonstrated that NBEAL2 deficiency abolishes the strict IL-3 dependency of MCs and results in accumulation of proteins of the protein biosynthesis machinery in MCs.

NBEAL2 regulates the homeostasis of p90RSKs and RPS proteins
Accumulated proteins in NBEAL2 deficient MCs could be explained by accumulated mRNAs.Therefore, we performed RNA sequencing (RNA-Seq) to determine the genome-wide transcription levels in wt and Nbeal2 À/À BMMCs.In sum, we found 5687 differentially expressed genes (DEGs).Among them all RPS genes, whose expression was slightly upregulated in Nbeal2 À/À BMMCs (Figure 1f).The transcript levels of the RPS-upstream kinases, p90RSK and p70S6K, were either unchanged or even slightly downregulated in Nbeal2 À/À BMMCs compared to wt MCs (Figure 1f).Furthermore, the transcript levels of STAT5, PKB/Akt, ERK1/2, p38 or IKK2/IκB were either not among the DEGs or were also slightly downregulated in Nbeal2 À/À BMMCs (Figure 1f).Therefore, the RPS transcript levels correlated with their accumulated protein levels in Nbeal2 À/À MCs.These data showed that NBEAL2 regulates the levels of RPS proteins either by controlling transcript stabilities, and/or by regulating translational-or post-translational processes.To test whether NBEAL2 regulates protein levels, we treated BMMCs with the protein-biosynthesis inhibitor cycloheximide (CHX) and followed the protein stabilities.Since 10 μM CHX minimally increased the amounts of dead cells after 8 h (Figure S2A) we treated BMMCs with 10 μM CHX for maximal 4 h.Thereby, CHX treatment resulted in degradation of p90RSK, p70S6K, and RPS proteins in wt BMMCs but only slightly in Nbeal2 À/À BMMCs (Figure 2a,b).In contrast to that, CHX did not affect the stability of STAT5, IKK2/IκB, PKB, p38 and ERK1/2 in wt and Nbeal2 À/À BMMCs (Figure 2c).This indicated that NBEAL2 deficiency did not result in a general defect of the protein degradation machinery.
This suggestion was supported by the fact that CHX-and the proteasome inhibitor MG132 induced similar kinetics of overall ubiquitinations in wt and Nbeal2 À/À BMMCs (Figure S2B).In summary, these data showed that NBEAL2 deficiency results in a defective degradation machinery for selected proteins among them p90RSKs, p70S6K and RPS6.
Accumulated p90RSKs mediate the prolonged survival of Nbeal2 À/À BMMCs NBEAL2 controls the stability of p90RSKs, p70S6K and RPS6, which are essential for ribosome functionality.
To investigate their role in the abnormal behaviour of Nbeal2 À/À BMMCs, we used specific inhibitors for p90RSKs and p70S6K the upstream kinases of RPS6.In contrast to the p70S6K inhibitor LY2584702 (Figure S2C,D) the p90RSK inhibitor LJH685 abrogated the prolonged IL-3-independent survival of Nbeal2 À/À BMMCs (Figure 2d) without inducing cell death (Figure 2e) demonstrating that NBEAL2 limits the survival of MCs by controlling the stability of the p90RSK-RPS6 module.

NBEAL2 limits the production of cytokines
Next, we investigated the consequences of NBEAL2 deficiency and the resulting p90RSKs and RPS6 accumulation on induced cytokine responses in MCs.First, we tested IL-3 and SCF, two MC-growth and survival factors, which typically induce a slight STAT-and MAPkinase-dependent cytokine production in MCs [16].Interestingly, in Nbeal2 À/À MCs, IL-3 and SCF induced a strong production of IL-6 and IL-13 whereas these growth factors only induced weak cytokine responses in wt MCs (Figure 3a,b).Furthermore, and in correlation with their accumulated protein levels, the activation of p90RSKs, p70S6K and the phosphorylation of RPS6 were increased in SCF-(Figure 3c) and IL-3-(Figure 3d) stimulated Nbeal2 À/À compared to wt BMMCs.Thereby, only the inhibition of p90RSKs (by LJH685) (Figure 3e) but not the inhibition of p70S6K (by LY2584702) (Figure 3f) reduced the strong IL-3-and SCF-induced cytokine production in Nbeal2 À/À BMMCs.
In contrast to IL-3 and SCF, the alarmin IL-33 induces IKK2/IκB-p65/RelA-dominated signalling pathways resulting in an effective and strong cytokine production in MCs [10,34,35].Interestingly, the IL-33-induced production of IL-6 and IL-13 which was found in wt MCs was amplified in MCs with deleted Nbeal2 (Figure 4a-c).Thereby, the IL-33-induced activation of the IKK/IκB-p65/RelA signalling was not affected by Nbeal2 deletion (Figure 4d).However, correlating to their accumulations, we detected a slightly increased activation level of p90RSKs, p70S6K and of RPS6 (Figure 4e) which resulted in an p90RSKs-but not of p70S6K-mediated amplified cytokine production in Nbeal2 À/À BMMCs (Figure 4f,g).In summary, our data show that NBEAL2 limits the SCF-, IL-3-and IL-33-induced cytokine production in MCs by controlling homeostatic levels and thus the activity of p90RSKs and RPS6.

p90RSKs stabilise RPS6 by regulating Nbeal2
Due to these findings, we focused our investigations on p90RSKs.Unexpectedly, the p90RSK inhibitor LJH685 strongly destabilised p90RSKs in wt BMMCs and to a lesser extent in Nbeal2 À/À MCs (Figure 5a,b).Thereby, LJH685 permanently blocked the phosphorylation of RPS6 in wt and Nbeal2 À/À MCs (Figure 5a,b).Beside the blocked RPS6 phosphorylations, LJH685 induced a transient loss of the RPS6 stability in wt BMMCs but not in Nbeal2 À/À BMMCs (Figure 5a,b).These experiments revealed a link between p90RSKs, NBEAL2 and RPS6.Thereby, inactivation of RSKs in wt BMMCs resulted in a transient NBEAL2-dependent RPS6 degradation which was nearly absent in Nbeal2 À/À BMMCs.Consequently, we concluded that in wt MCs, p90RSKs balances the NBEAL2dependent RPS6 degradation to ensure adequate RPS6 levels (Figure 5c), whereas in MCs with deleted Nbeal2, the levels of RPS6 are not regulated and are thus accumulated (Figure 5d).

NBEAL2 structurally resembles to the F-box protein β-TrCP
Our data demonstrated that NBEAL2 controls RPS6 degradation.However, NBEAL2 does not contain ubiquitinconjugating or -ligase activities.Therefore, we hypothesised that NBEAL2 rather is a RPS6 binding protein translocating RPS6 to an ubiquitination ligase complex.The PH-BEACH-WDR (PBW) module in full length NBEAL2 (FL-NBEAL2) (Figure S3A) is known to mediate protein/protein interactions [4].Hence, we assumed that the PH-BEACH-WDR (PBW-NBEAL2) part of NBEAL2 (Figure S3B) is most likely the part that associates with RPS6.Given these facts we focused our investigations on the PBW-module (PBW-NBEAL2).First, we wanted to get structural insights into the molecular organisation of PBW-NBEAL2, which would allow to predict a model for RPS6 binding.To do so, we used the deep learning-based method, RoseTTAFold [36] on the Rosetta structure prediction service (https://robetta.bakerlab.org/)which calculated five potential PBW-NBEAL2 models.The alignment of the five obtained PBW-NBEAL2 models with the Pymol© software revealed closely overlapping structures (Figure S3C).Therefore, we randomly selected one of these models.The overall structure of the PBW-NBEAL2 has a clamp-like shape (Figure 6a,b).In this model, the PH domain is in close proximity to the WDR domain (Figure 6a-c) which contains seven propeller-like arranged WD40 blades (Figure 6d).WDR domaincontaining proteins play an important role in protein degradation processes [37].Among them, the F-box protein β-TrCP which interacts with Skp1 and Cullin, and thus forms the F-box-containing ubiquitin ligase complex (SCF β-TrCP ).In this complex, β-TrCP binds phosphorylated substrate proteins via its WDR domain, which is required to transfer ubiquitin molecules from the ubiquitin conjugating E2 enzyme to the phosphorylated and β-TrCP-bound substrates [38].Interestingly, the predicted overall structural organisation of PBW-NBEAL2 resembles to β-TrCP associated with Skp1 (Figure S3D,E).These overall structural similarities supported the idea  F I G U R E 4 NBEAL2 limits the IL-33-induced cytokine production.(a) Wt and Nbeal2 À/À BMMCs were left unstimulated (C as control) or were stimulated with different amounts of IL-33 (as indicated).Supernatants were analysed by ELISA (The SEMs of n = 3 biological replicates of wt and Nbeal2 À/À BMMCs are shown.We performed unpaired students t-tests between wt and Nbeal2 À/À BMMCs stimulated with different amounts of IL-33).(b) Wt and Nbeal2 À/À BMMCs were left unstimulated (C as control) or were stimulated with IL-33 for different time points (The SEMs of n = 3 biological replicates of wt and Nbeal2 À/À BMMCs are shown.We performed unpaired students ttests between wt and Nbeal2 À/À BMMCs on the displayed time points).(c) Wt and Nbeal2 À/À PCMCs were left unstimulated (C as control) or were stimulated with IL-33 for 24 h.(The SEMs of n = 3 biological replicates of wt and Nbeal2 À/À BMMCs are shown.We performed unpaired students t-tests between wt and Nbeal2 À/À BMMCs).(d, e) Wt and Nbeal2 À/À BMMCs were left unstimulated (C as control) or were stimulated with IL-33.Lysates were analysed by western blotting.(f, g) Wt and Nbeal2 À/À BMMCs were treated with vehicle (DMSO) or with LJH685 (10 μM) (f) or LY2584702 (50 μM) (g) and stimulated with IL-33 (f, g) for 24 h.Supernatants were analysed by ELISA [The SEMs of n = 4 (f; IL-6), n = 7 (g; IL-6) and n = 3 (g; IL-13) biological replicates of wt and Nbeal2 À/À BMMCs are shown.We performed unpaired students t-tests between DMSO-and inhibitor-treated and IL-33-stimulated wt and Nbeal2 À/À BMMCs].that NBEAL2 also acts as such a substrate recognition molecule.Thereby, the closed proximity between the PH and the WDR domain indicated that both domains might stabilise the potential interaction between PBW-NBEAL2 and RPS6 (Figure 6e).

The WDR domain of NBEAL2 mediates the interaction with RPS6
First, we tried to isolate the RPS6/NBEAL2 heterodimer from BMMCs lysates.However, the low precipitation rate of RPS6 from wt BMMCs (Figure S4A, see red frame) did not allow an efficient isolation and detection of the RPS6/NBEAL2 complex.To detect and to characterise the hypothetical RPS6/NBEAL2 complex, we used CRISPR/Cas9-generated NBEAL2-KO HEK293 cells (Figure S4B) which were transfected with different NBEAL2 constructs.For the generation of these NBEAL2 constructs, we used the C-terminally FLAG-tagged PBW-NBEAL2 construct [4] (Figure 6f).Our predicted PBW-NBEAL2 structure indicated that both, the PH and the WDR domain might be important for RPS6 complexation.Hence, from the FLAG-tagged PBW-NBEAL2 construct (Figure 6f), we either deleted the WDR domain to obtain the PH-BEACH (PB) construct, or we deleted the PH domain and the N-terminal part of the BEACH domain (which was suggested as a regulatory motif) to obtain the BEACH-WDR 1 and 2 (BW1 and BW2) constructs (Figure 6g).These constructs were transfected into NBEAL2-KO HEK293 cells (Figure S4C,D) and endogenously expressed RPS6 was immunoprecipitated (IP).In RPS6 precipitates, we only hardly detected the PB-NBEAL2 (Figure 6h), but we still detected the PBW-, BW1-and the BW2-NBEAL2 constructs that still contain the WDR domain (Figure 6i).These data demonstrated that the WDR but not the PH domain is essential for the interaction of NBEAL2 with RPS6-which was in contrast to our assumed model (see Figure 6e).This showed that comparably to β-TrCP, the WDR domain of NBEAL2 is essential to interact with its substrate proteins (Figure 6j,k).PBW-NBEAL2 binds RPS6 independently from its phosphorylation The WDR domain of β-TrCP binds its client proteins in dependency of their phosphorylations [38].To test whether the phosphorylation of RPS6 is required for binding to PBW-NBEAL2, we transfected NBEAL2-KO HEK293 cells with PBW-NBEAL2 and used the RSK inhibitor LJH685 to block RSK-mediated RPS6 phosphorylations.As shown in Figure 6l, LJH685 did not reduce the association of RPS6 with PBW-NBEAL2, while it efficiently inhibited the phosphorylation of RPS6.This demonstrated that in contrast to β-TrCP, binding of client proteins to PBW-NBEAL2 occurs independently from their phosphorylation status.Given that electrostatic potentials (EPs) determine the interaction between proteins, we expected that the comparison of the EPs of the WDR domains of PBW-NBEAL2 and β-TrCP might explain the different client protein binding modalities of these proteins.The EPs of the WDR front side of β-TrCP is predominantly negatively charged with a positively charged central pore [38].However, the backside consists of large central positively charged cleft (Figure 6m).This cleft is able to bind phosphorylated and therefore negatively charged client proteins such as phosphorylated IκBα [38].In contrast to this, the WDR of PBW-NBEAL2 consists of a negatively charged front side with a positively charged core, and a diffusively charged back side (Figure 6n).The predominantly negatively charged WDR domain of PBW-NBEAL2 demonstrated that binding of proteins is mediated independently from their phosphorylations.Therefore, we concluded that in contrast to β-TrCP, NBEAL2 does not discriminate between substrates due to their phosphorylation status (Figure 6o,p).

DISCUSSION
We show that inactivation of NBEAL2 resulted in an abnormal MC phenotype, which was characterised by the stabilisation of p90RSKs and RPS6.This indicated that in MCs, NBEAL2 attains constant and adequate RPS6 levels to avoid an overload of MCs with RPS6 molecules.To achieve that we postulate that by binding RPS6, NBEAL2 facilitates the supply of RPS6 to a yet unidentified ubiquitination system resulting in RPS6 degradation.Since inactivation of p90RSKs mediates RPS6 degradation in wt but not in Nbeal2 À/À BMMCs, this shows that p90RSKs, is important for this mechanism.Thereby, we postulate that the p90RSK activity regulates the function of NBEAL2 and thus the level of RPS6 molecules.Under homeostatic conditions, the basally active p90RSK balances and controls the function of NBEAL2.This results in homeostatic degradation of RPS6 to control the level of RPS6 molecules.Thus, NBEAL2 prevents an overload of the system with RSK-phosphorylated, and consequently highly active, RPS6 molecules (Figure S5A).In contrast to that, inactivation of p90RSK leads to a NBEAL2-dependent RPS6 degradation (Figure S5B).Under these conditions, the p90RSK-mediated suppression of NBEAL2 is abrogated which unleashes the function of NBEAL2 and results in an excessive RPS6 degradation (Figure S5B).However, when the p90RSK activity is increased by external stimuli, the function of NBEAL2 is blocked by highly active p90RSK.Consequently, the level of p90RSK-phosphorylated RPS6 molecules increases due to the decreased RPS6 degradation (Figure S5C).This afterwards enhances the protein biosynthesis rate of the ribosomes.Therefore, p90RSK not only mediates RPS6 phosphorylations but also regulates the level of phosphorylable RPS6 molecules via NBEAL2.By this mechanism, MCs can adapt the level of RPS6 molecules and thus the functionality of ribosomes to extracellular conditions.How p90RSK controls the function of NBEAL2 is unknown.We speculate that p90RSK phosphorylates NBEAL2 and thus controls its function.In this hypothetical model, the p90RSK-mediated NBEAL2 phosphorylation decreases the interaction between NBEAL2 and RPS6 and thus stabilises RPS6.In contrast to that, nonphosphorylated NBEAL2 is more active and interacts with RPS6 and thus mediates its degradation.However, the involved ubiquitin-ligase which mediates RPS6 ubiquitination and degradation is unknown.Interestingly, NBEAL2 interacts with the ubiquitin ligases HUWE1 and UBR5 [4] which both belong to the HECT-ubiquitin ligase family [39].Given that UBR5 regulates the stability of ribosomal proteins [40] we speculate that UBR5 ubiquitinates RPS6 associated with NBEAL2 and thus mediates RPS6 degradation (Figure S6A,B).Therefore, NBEAL2 deletion or inactivating NBEAL2 mutations which prevent RPS6 binding or the potential interaction to UBR5, block RPS6 degradation and thus leads to RPS6 accumulations.Consequently, the increased protein levels of RSKs and RPS6 mediated by NBEAL2 inactivation resulted in an amplified cytokine production (Figure S6C,D) and a prolonged growth-factor independent survival.These data suggest that by restricting protein stabilities, NBEAL2 regulates the MC survival and prevents the development of a proinflammatory, MC phenotypes.This fits to the findings that the loss of the NBEAL2 gene by deletion of the 3p21.3chromosome region in humans is associated with the development of mastocytosis [41] which is mostly accompanied by increased activities of p90RSK and RPS6.However, Nbeal2 inactivation alone is not sufficient to dysregulate the distribution of MCs or to induce higher susceptibility to allergic reaction, and did also not result in abnormalities in red/white blood cell count in mice [9,42].This indicates that NBEAL2 inactivation alone does not result in fully degenerated MC phenotypes.Therefore, we speculate that NBEAL2 inactivation is only one step on the route to the development of mastocytosis or MC tumours.Together, our findings with the fact that NBEAL2 inactivation results in defects in the granule formation in megakaryocytes, platelets [3] and neutrophils [6] and is associated with increased susceptibilities to bacterial and viral infections [6,43] indicate that intact and functional NBEAL2 is pivotal to suppresses the development of cancer diseases and immune deficiencies [44].

CONCLUSION
We found that NBEAL2 maintains the adequate homeostatic level of the p90RSK-RPS6 module, whose expression level and activity regulates the protein biosynthesis rate of ribosomes.Mechanistically, NBEAL2 interacts with RPS6 via its WDR domain.We speculate that this association is prerequisite to mediate an ubiquitin-dependent degradation of RPS6 which limits or adapts the protein biosynthesis to extra cellular conditions.Consequently, by maintaining adequate levels of ribosomal proteins, NBEAL2 protects MCs from aberrant development.
U R E 2 NBEAL2 limits the stability of RSKs and RPS proteins.(a-c) Wt and Nbeal2 À/À BMMCs were treated with vehicle (C = DMSO) or CHX (10 μM).Cells were lysed after the indicated time points and were analysed by western blotting.(b) The statistical analysis of the western blots in (a) is shown [The SEMs of n = 6 (for p90RSK, p70S6K and RPS6), n = 4 (for RPS28), n = 5 (for RPS15 and RPS8) biological replicates of wt and Nbeal2 À/À BMMCs are shown.We performed unpaired students t-tests between wt and Nbeal2 À/À BMMCs on the displayed time points].(d, e) Wt and Nbeal2 À/À BMMCs were treated with vehicle (DMSO) or LJH685 (10 μM) without IL-3 (ÀIL-3).Cells were counted every consecutive day (d) or were treated with PI to measure the percentage of dead cells (PI + ) by flow cytometry (every second day) (e) (The SEMs of n = 10 biological replicates of wt and Nbeal2 À/À BMMCs are shown.We performed unpaired students t-tests between wt and Nbeal2 À/À BMMCs on the displayed time points).
U R E 3 NBEAL2 limits the IL-3-and SCF-induced cytokine production.(a, b) Wt and Nbeal2 À/À BMMCs (a) or PCMCs (b) were left unstimulated (C as control) or were stimulated with IL-3 or SCF.After different time points (a) or after 24 h (b) supernatants were collected and analysed by ELISA.(The SEMs of n = 3 biological replicates of wt and Nbeal2 À/À BMMCs (a) or PCMCs (b) are shown.We performed unpaired students t-tests between wt and Nbeal2 À/À BMMCs on the corresponding time points).(c, d) Wt and Nbeal2 À/À BMMCs were left unstimulated (C as control) or were stimulated with SCF (c) or IL-3 (d).Lysates were analysed by western blotting.(e, f) Wt and Nbeal2 À/À BMMCs were treated with vehicle (DMSO) or with LJH685 (10 μM) (e) or with LY2584702 (50 μM) (f) and were then stimulated with IL-3 or SCF.Supernatants were analysed by ELISA [The SEMs of n = 6 (e) and n = 3 (f) biological replicates of wt and Nbeal2 À/À BMMCs are shown.We performed unpaired students t-tests between DMSO-and inhibitor-treated and SCF or IL-3 stimulated Nbeal2 À/À BMMCs].

F
I G U R E 5 p90RSK regulates RPS6 stability.(a) Wt and Nbeal2 À/À BMMCs were treated with vehicle (C as control; DMSO) or with LJH685 (10 μM).Cells were lysed after the indicated time points and were analysed by western blotting.(b) The statistical analysis of the western blots in (a) are shown [The SEMs of n = 6 (for RSKs and RPS6) and n = 3 (for pRPS6) biological replicates of wt and Nbeal2 À/À BMMCs are shown.We performed unpaired students t-tests between wt and Nbeal2 À/À BMMCs on the displayed time points].(c) The proposed model of RPS6 degradation in wt MCs is shown.(1) The protein biosynthesis results in the production RPS6 molecules.(2) p90RSKs suppresses and thus balances the function of NBEAL2.Therefore, (3) NBEAL2 limitedly binds RPS6 and (4) mediates its homeostatic degradation.This results in (5) NBEAL2-controlled homeostatic RPS6 levels which can be phosphorylated an activated by p90RSKs.Consequently, NBEAL2 prevents the occurrence of accumulated and active RPS6 molecules.(d) The model of RPS6 stability in Nbeal2 À/À MCs is shown.(1) The protein biosynthesis results in the RPS6 production.(2) Due to the loss of functional NBEAL2 the homeostatic RPS6 degradation is blocked and thus results in RPS6 accumulation.(3) p90RSKs phosphorylate the accumulated RPS6 molecules.This results in an increased amount of phosphorylated RPS6 molecules which enhance the protein biosynthesis.
Structural organisation of NBEAL2.(a) (Upper panel) The cartoon of the linear structural organisation of PBW-NBEAL2 is shown.(Lower panel) The predicted structural organisation of PBW-NBEAL2 is shown.This model was calculated with the Rosetta structure prediction service (https://robetta.bakerlab.org/)and modelled with the Pymol software ® [36].(b) Cartoon of PBW-NBEAL2, derived from the calculated PBW-NBEAL2 model (shown in (a)) generated with the Pymol software ® .(c) Rotated PBW-NBEAL2 model.(d) The isolated view (in brown) and the seven WD40 blades (in colours) of the WDR domain are shown.(e) Cartoon of the assumed RPS6 binding by PBW-NBEAL2 is shown (left panel: side view; right panel: front view).(f) Cartoon of the linear domain structure of FLAG-tagged PBW-NBEAL2 (PBW) is shown.(g) Cartoons of the linear domain structures of the generated mutants are shown: PB-NBEAL2 (PB), BW1-NBEAL2 (BW1) and BW2-NBEAL2 (BW2) with their FLAG-tags are shown.(h) NBEAL2-KO HEK293 cells were transfected with: PBor PBW-NBEAL2 constructs.(i) NBEAL2-KO HEK293 cells were transfected with: BW1-, BW2-or the PBW-NBEAL2 constructs.In all cases (h, i) RPS6 was immunoprecipitated (RPS6-IP) with an anti-RPS6 antibody (RPS6-AK) and the precipitates and input lysates were analysed by western blotting.(j, k) Cartoons of substrate (Sub.)binding mechanism of β-TrCP (j) and the predicted RPS6 binding to PBW-NBEAL2 (k) are shown.(l) NBEAL2-KO HEK293 cells were transfected with PBW-NBEAL2.Cells left untreated were treated with vehicle (DMSO) or with LJH685 (10 μM).RPS6 was immunoprecipitated (RPS6-IP) with an anti-RPS6 antibody (RPS6-AK) and the precipitates and input lysates were analysed by western blotting.All transfection experiments were performed in n = 3 independent experiments.Shown is one representative experiment.(m, n) The isolated front-side-and back-side-views of the WDR domains of β-TrCP (m) and PBW-NBEAL2 (n) with their electrostatic potentials (EPs) are shown (generated with the PlugIn APBS Electrostatics in the Pymol software ® ).(o, p) Cartoons of phospho-substrate (P-Sub.)binding to β-TrCP (o) and the predicted binding of phospho-RPS6 (P-RPS) to PBW-NBEAL2 (p) are shown.
P.W. designed research, performed experiments, analysed data, and made the figures.L.Z., B.S., and R.M. designed research, performed experiments, and analysed data.C.K. and M.G. performed experiments.B.N. contributed essential material.N.A. performed experiments, analysed data, and edited the manuscript.J.D. designed research, performed experiments, analysed data, wrote and edited the manuscript.S.D. developed the concept, designed the research, performed experiments, analysed data, made the figures, drafted, wrote and edited the manuscript.