PSTPIP1-LYP phosphatase interaction: structural basis and implications for autoinflammatory disorders

Vol.:(0123456789) 1 3 Cellular and Molecular Life Sciences (2022) 79:131 https://doi.org/10.1007/s00018-022-04173-w ORIGINAL ARTICLE PSTPIP1‑LYP phosphatase interaction: structural basis andimplications forautoinflammatory disorders JoséA.Manso2 · TamaraMarcos1· VirginiaRuiz‑Martín1· JavierCasas1· PabloAlcón2 · MarianoSánchezCrespo1· YolandaBayón1 · JoséM.dePereda2 · AndrésAlonso1 Received: 14 September 2021 / Revised: 14 January 2022 / Accepted: 27 January 2022 / Published online: 12 February 2022 © The Author(s) 2022 Abstract Mutations in the adaptor protein PSTPIP1 cause a spectrum of autoinflammatory diseases, including PAPA and PAMI; however, the mechanism underlying these diseases remains unknown. Most of these mutations lie in PSTPIP1 F-BAR domain, which binds to LYP, a protein tyrosine phosphatase associated with arthritis and lupus. To shed light on the mechanism by which these mutations generate autoinflammatory disorders, we solved the structure of the F-BAR domain of PSTPIP1 alone and bound to the C-terminal homology segment of LYP, revealing a novel mechanism of recognition of Pro-rich motifs by proteins in which a single LYP molecule binds to the PSTPIP1 F-BAR dimer. The residues R228, D246, E250, and E257 of PSTPIP1 that are mutated in immunological diseases directly interact with LYP. These findings link the disruption of the PSTPIP1/LYP interaction to these diseases, and support a critical role for LYP phosphatase in their pathogenesis. Keywords Immunology· Auto-inflammation· LYP· PSTPIP1 Introduction Pyogenic arthritis, pyoderma gangrenosum and acne (PAPA) is a rare autoinflammatory disease caused by mutations in the gene that codes for the proline–serine–threonine phosphatase interacting protein 1 (PSTPIP1) [1]. PAPA is characterized by the presence of sterile arthritis with a rich neutrophilic infiltrate in the joints, ulcerative lesions in the skin (pyoderma gangrenosum), and acne. PSTPIP1 mutations generate other autoinflammatory diseases, like PSTPIP1-associated myeloid-related proteinemia inflammatory (PAMI) syndrome, previously known as hyperzincemia and hypercalprotectinemia [2, 3]. The autoinflammatory diseases caused by PSTPIP1 mutations are now collectively termed as PAID (PSTPIP1-associated inflammatory diseases). Moreover, the collection of diseases associated with PSTPIP1 mutations has expanded to include common variable immunodeficiency (CVID) [4]. PSTPIP1 is an adaptor protein expressed in most immune cell lineages. It presents two structural domains in its sequence, an N-terminal Fer/CIP4 homology-Bin/Amphiphysin/Rvs (F-BAR) domain and a C-terminal SH3 domain, connected by a linker region (Fig.1A). The function of F-BAR domains seems to be membrane shape coordination with the cytoskeleton in various cellular processes such as endocytosis and cell division [5–7]. F-BAR domains form homo-dimers with an elongated and slightly curved shape. Positively charged residues on the concave surface of F-BAR domains interact with the polar heads of the phospholipids in the cellular membrane [8, 9]. The F-BAR domain of PSTPIP1 binds to the PEST subfamily of protein tyrosine phosphatases (PTPs): PTPN12, also known as PTP-PEST, PTPN18, and the lymphoid phosphatase (LYP), encoded by the gene PTPN22 [10]. These phosphatases bind to PSTPIP1 through a conserved motif in their C-terminus, thereby called C-terminal homology domain (CTH). LYP phosphatase is involved in autoimmune diseases with a significant inflammatory component, like arthritis or systemic lupus erythematosus [11]. The F-BAR domain of PSTPIP1 also binds pyrin, a sensor that activates the pyrin inflammasome in response to RhoA inactivation by bacterial toxins. Cellular andMolecular Life Sciences * Andrés Alonso [email protected]a.es 1 Unidad de Excelencia Instituto de Biología y Genética Molecular (IBGM), CSIC-Universidad de Valladolid, c/ Sanz y Forés 3, 47003Valladolid, Spain 2 Instituto de Biología Molecular y Celular del Cáncer (IBMCC), CSIC-Universidad de Salamanca, Campus Unamuno, 37007Salamanca, Spain J.A.Manso et al. 1 3 131 Page 2 of 17 A B C Anisotropy log [F-BAR (M)] Wild type E250Q W232A +++ E257G E277D + E257K + R405C + V344I + D246N V5 V5 myc myc TL 3xmyc-LYP -+ ++ PSTPIP1 A230T E250Q + W232A + D266N PSTPIP1 + G258A + E250K + R52Q IP anti-myc ++++++ ++ + T68M V122I Q219H R228C T274M D289H Y345C D384G G403E G403R -10-9-8 -7 -6 -5 0.04 0.08 0.12 A230T F-BAR SH3 1 287359 416 W232A D266N E250Q/E250K A230T E257G/E257K G258A T274M D289H R405C V344I G403E/G403R D384G E277D V122I D246N R228C R52Q T68M Y345C Q219H PSTPIP1 D log(kd (M)) -5 -8 -7 -6 A230T W232A E250Q Wild type ns **** **** E PSTPIP1 R228C W232A D246N E250K E250Q E257G E257K 0.0 0.2 0.4 0.6 0.8 1.0 Mander´s Colocalization Index (M2) **** **** **** **** **** ** **** PSTPIP1‑LYP phosphatase interaction: structural basis andimplications forautoinflammatory… 1 3 Page 3 of 17 131 Mutations in pyrin cause the autoinflammatory disease Familial Mediterranean Fever [12]. Additionally, the SH3 domain of PSTPIP1 interacts with ABL kinase [13], WASP [14], CD2 [15] and FasL [16], and may participate in the interaction with pyrin [17]. PSTPIP2, a paralog of PSTPIP1, presents mutations in mice that end up causing an autoinflammatory bone disease similar to chronic osteomyelitis [18, 19]. PSTPIP2 also has an F-BAR domain that shares 49% sequence identity with the F-BAR domain of PSTPIP1, followed by a short tail containing several tyrosines known to be phosphorylated, and not conserved in PSTPIP1 [20]. PSTPIP2 also binds to PEST phosphatases via its F-BAR domain [21, 22]. More than 20 missense mutations have been identified in PSTPIP1 in patients with autoinflammatory diseases [23], the majority of which are located in the F-BAR domain (Fig.1A) [23–25]. Most of these mutations are currently uncharacterized, and thereby, it remains unknown how such a diverse array of changes in PSTPIP1 gene, mostly inherited in a dominant fashion, alter PSTPIP1 function to cause the above referred immune diseases. Wise and co-workers identified PSTPIP1 as the gene that originates PAPA syndrome and described the first mutations associated to this disease in PSTPIP1, A230T and E250Q [1], suggesting that the interaction of PSTPIP1 with PEST phosphatases could be critical to develop PAPA. Thus, we studied the PSTPIP1/LYP interaction to gain a better understanding of the mechanism underlying PSTPIP1- related autoinflammatory diseases. We determined the previously unknown structure of the isolated F-BAR domain of PSTPIP1. Furthermore, we solved the structure of the F-BAR domain bound to the LYP CTH motif, which is the first highresolution structure of an F-BAR domain bound to its ligand. This structure shows a new way of interacting Pro-rich peptides with proteins, in which the F-BAR dimer of PSTPIP1 binds to one molecule of the CTH peptide of LYP. Analysis of several mutations identified in patients show that only mutations R228C, E250Q, E250K, E257K, and D246N block the binding to LYP, and do not affect PSTPIP1 oligomerization. As a result, we conclude that disrupting the interaction between LYP and PSTPIP1 could play a key role in the mechanism underlying the autoinflammatory diseases caused by PSTPIP1 mutation. Materials andmethods Antibodies andcell lines The hemagglutinin (HA) monoclonal antibody (mAb) was from Covance (Berkely, CA, USA). The anti-V5 mAb was from Invitrogen (Carlsbad, CA, USA). The myc Ab (9E10) was from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA). PSTPIP1 Ab was generated against the whole protein produced in bacteria [26]. The anti-LYP goat polyclonal Ab was from R&D Systems, Inc. (Minneapolis, MN, USA). The anti-GFP Ab was from eBioscience (San Diego, CA, USA). HEK293 were maintained at 37°C in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, 2mM l-glutamine, 100 U/mL penicillin G, and 100µg/ mL streptomycin. Transient transfection of HEK293 cells was carried out using the calcium phosphate precipitation method [27]. Plasmids andmutagenesis PSTPIP1 (Uniprot O43586) wild type sequence in pcDNA3.1/V5-His and the mutants A230T and E250Q were a kind gift from Dr. Carol Wise. pEF3xmyc-LYP has been described previously [26]. Standard molecular biology techniques were used to generate the different constructs used in this study. Point mutations were created by PCR using the QuikChange method. All constructs and mutations were verified by nucleotide sequencing. For recombinant expression of PSTPIP1 in bacteria the cDNA coding for the region 1–289 of human PSTPIP1 was cloned in a modified version of the pET15b vector that codes for a fusion protein with an N-terminal poly-His tag and a site recognized by the tobacco etch virus (TEV) protease [28]. Immunoprecipitation, SDS PAGE andimmunoblotting These procedures were done as reported before [29]. Briefly, cells were lysed in TNE lysis buffer: 20mM Tris/HCl (pH Fig. 1 Interaction of LYP with of PSTPIP1 mutants associated to immune diseases. A Domain structure of PSTPIP1 with the mutations studied on this work. Mutations described in patients suffering from autoinflammatory diseases and CVID are indicated on top, while W232A mutation that is not related to disease is on the lower part. B HEK293 cells were transfected with PSTPIP1-V5 wild type and several mutants associated with autoinflammatory diseases, as indicated in the top of the panels, and with 3xmyc-LYP. Lysates were immunoprecipitated with myc antibody (Ab) and PSTPIP1 bound to LYP was detected by IB with Ab for the V5 epitope. Expression of the proteins was verified by IB in total lysates (TL) with the same Abs. C Dissociation constants (kd) of the interaction of the F-BAR domain of PSTPIP1, wild type and indicated mutants, with fluorescein-LYP-CTH. Data are means ± standard deviation (SD), n = 3 independent experiments, one shown in D. Statistical comparison to the kd of the wild type F-BAR was analyzed using ANOVA followed by Dunnett’s test. ****p < 0.0001. D Representative binding isotherms of the F-BAR domain of PSTPIP1, wild type and mutants, to fluorescein-labeled LYP-CTH (2nM), measured by fluorescence anisotropy. Data points are means ± SD of measurement replicates; lines represent the fitted binding curves. E HEK293 cells co-transfected with LYP-mCherry (red) and wild type or indicated mutants of PSTPIP1-eGFP (green). Manders’ colocalization indexes (M2) were calculated. Error bars represent SEM (n > 10). **p < 0.01; ****p < 0.0001, by Student’s t test ◂ J.A.Manso et al. 1 3 131 Page 4 of 17 7.4), 150mM NaCl, 5mM EDTA containing 1% NP-40, 1mM Na3VO4, 10µg/mL aprotinin and leupeptin, and 1mM PMSF, and clarified by centrifugation at 15,000rpm for 10min. The clarified lysates were preadsorbed on protein G-Sepharose and then incubated with Ab and protein G-Sepharose beads for 1h. Immune complexes were washed three times in TNE buffer and suspended in SDS sample buffer. Proteins resolved by SDS-PAGE were transferred electrophoretically to nitrocellulose membranes, which were immunoblotted with optimal dilutions of specific Abs, followed by the appropriate anti-IgG-peroxidase-conjugate. Blots were developed by the enhanced chemiluminescence technique with Pierce ECL Western Blotting substrate (Thermo Scientific, Rockford IL) according to the manufacturer’s instructions. In vitro pull‑down assays Recombinant His-PSTPIP1 full-length produced in bacteria (2μg) was combined with 2μg of recombinant GST-CTH wild type or mutated in R799 or W805 to A, and 10 μL of glutathione sepharose beads (Cytiva, Marlborough, MA) in binding buffer: 5mM Tris/HCl (pH 7.4), 50mM NaCl, 0.1% NP-40. After incubation for 1h at 4°C, beads were washed three times in the same buffer and suspended in SDS sample buffer. The complexes were then processed for immunoblotting as explained above for the immunoprecipitation. Colocalization analysis HEK 293 cells expressing LYP-mCherry wild type and PSTPIP1-eGFP wild type or indicated mutants were nuclei counterstained with the DNA binding dye DAPI. All images were captured with a Leica confocal system TCS SP5X inverted microscope with a HCS Plan Apo CS 63X/1.4 NA oil immersion lens. Leica Application Suite Advanced Fluorescence software was used for the capture, and ImageJ for image presentation. Colocalization was analyzed with JACoP plugin in ImageJ. After background subtraction; negative pixel values were clipped to zero. Positive values were selected by Costes automatic thresholding, removing the bias of visual interpretation [30]. Colocalization index Manders’ M1 and M2, were calculated. Manders’ split coefficients are based on the Pearson’s correlation coefficient but avoid issues relating to absolute intensities of the signal, since they are normalized to total pixel intensity [31]. These coefficients vary from 0 (nonoverlapping) to 1 (100% colocalization). The index M2 is defined as the percentage of above-background pixels from the second, red channel (LYP) that overlap abovebackground pixels from the first, green channel (PSTPIP1). This index is sensitive to changes in the background but not to differences in the intensity of overlapped pixels and is suitable to apply in images with a high and very clear signal to background ratio. Protein purification Proteins were produced in the Escherichia coli strain BL21 (DE3). They were purified by nickel-chelating affinity chromatography, the His-tag was cleaved by digestion with TEV protease and removed as described [32]. Peptides of human LYP (Uniprot Q9Y2R2-1) corresponding to the region 787–807 (GFANRFSKPKGPRNPPPTWNI) were custom synthesized either labeled with fluorescein at the N-terminus (Thermo Scientific, Germany) or unlabeled (Genosphere Biotechnologies, France). Fluorescence‑based binding assay Binding of the F-BAR of PSTPIP1 to a peptide of the CTH of LYP labeled with fluorescein was analyzed by fluorescence anisotropy. Fluorescein-LYP-CTH in 20mM Tris/ HCl (pH 7.5), 150mM NaCl, 0.1mg/mL bovine serum albumin (BSA) were titrated with the F-BAR of PSTPIP1, wild type and point mutants. The fluorescence anisotropy was measured at 25ºC in a Fluoromax-3 spectrofluorometer (HORIBA-Jobin–Yvon) with Glan–Thompson polarizers using a 3 × 3mm quartz cuvette. The samples were excited at 490nm and the emission was collected at 520nm. The apparent dissociation constant (kd) was estimated by fitting a one-to-one binding model as described [33] using the SigmaPlot program. Alternatively, to determine the stoichiometry of the interaction, titrations were done under saturation conditions. Fluorescein-LYP-CTH at 1μM, which was ~ 50 times above the kd value, was titrated with PSTPIP1. Under these conditions, PSTPIP1 binds to the CTH in a linear fashion until all the CTH sites are occupied. The binding stoichiometry was estimated from the intersection of asymptotes of the ascending and plateau regions. Isothermal titration calorimetry (ITC) ITC experiments were carried out at 25°C using a VP-ITC system (MicroCal, Northampton, MA, USA). Solutions of the F-BAR of PSTPIP1 were loaded on the sample cell at a known concentration between 26.2 and 30.6μM in 20mM sodium phosphate (pH 7.5), 150mM NaCl, 5% (v/v) dimethyl sulfoxide (DMSO). Solutions of the unlabeled peptide of the CTH of LYP, in the same buffer as the PSTPIP1 samples to avoid buffer mismatch and at known concentrations between 146.8 and 155.2μM, were loaded in the injection syringe. Sample solutions were degassed and thermostated at 25 °C using a MicroCal ThermoVac sample station prior to use. Titrations were done by one initial injection of 3 μL PSTPIP1‑LYP phosphatase interaction: structural basis andimplications forautoinflammatory… 1 3 Page 5 of 17 131 followed by 27 sequential injections of 10 μL. A reference power of 15 μcal s−1, stirring speed of 307rpm and 240s spacing were selected. Heat exchange from the first injection was not used in the analysis. Data were analyzed using the Origin 7 ITC software package (MicroCal Software, Northampton, MA, USA), corrected by the heat of injection calculated from the basal heat remaining after saturation and confirmed by titration into buffer only as control. The single set of identical sites model was applied to obtain the binding enthalpy (∆H), entropy (∆S), stoichiometry (N), and association constant (ka = 1/kd) using a nonlinear squares algorithm. Crystallization andstructure determination oftheG258A mutant oftheF‑BAR domain ofPSTPIP1 Crystals of the F-BAR domain of PSTPIP1, residues 1–289, carrying the G258A mutation were obtained by hanging drop vapor diffusion at room temperature. A protein solution at 1.2mg/mL in 10mM Tris/HCl (pH 7.5), 150mM NaCl, 0.5mM dithiothreitol (DTT) was mixed with an equal volume of the crystallization solution 100mM Bis–Trispropane (pH 6.5), 15% (w/v) PEG 3350, 250mM sodium citrate. Prior to data collection crystals were transferred to 100mM Bis–Tris-propane (pH 6.5), 15% (w/v) PEG 3350, 250mM sodium citrate, 18% glycerol and were cooled by immersion in liquid nitrogen. These and all other diffraction data were collected at 105K on the I03 beamline of Diamond Light Source synchrotron (Didcot, UK). Diffraction intensities of this and other crystals (see below) were integrated, reduced, and converted into structure factor amplitudes with the programs XDS [34], Pointless [35], Aimless [36] and Staraniso [37] as implemented in the autoPROC pipeline [38]. Crystals belong to the space group P212121 (TableS1) and contain two PSTPIP1 molecules in the asymmetric unit (AU), which correspond to ~ 53% solvent content. Diffraction data was notably anisotropic. The lowest resolution limit was ~ 2.9Å in the direction b* and the highest limits were 1.97Å and 2.09 in the directions a* and c*, respectively. Staraniso was used to apply an anisotropic cut-off and correction of the intensity data, and to calculate the structure amplitudes using Bayesian estimation. The structure was phased by molecular replacement (MR) using Phaser [39]. Initially, the MR was solved using a mixed-atom homology model built with SCWRL [40] using as template the crystal structure of Hof1p F-BAR domain (PDB ID 4WPE), which shares 14% sequence identity with PSTPIP1. Starting from this MR solution the structure was refined with Phenix Refine [41] against the anisotropically corrected data, alternated with manual model building in Coot [42]. Yet, it was not possible to improve the refinement of this model; the free R factor was ~ 47%. At this point, we perform normal mode analysis (NMA) on the partially refined structure using the elNemo sever, a web interface to do elastic network model-based NMA [43]. Eleven conformations that represent global motions corresponding to the lowest-frequency mode (an amplitude perturbation in the direction of a single normal mode of 500 was applied by using a 50 step size) were generated. Each of these conformations was used as search model to phase again the structure by MR. The model with best MR solution scores was automatically rebuilt using Phenix Autobuild [44]. The resulting model was refined as described above. Torsion angle non-crystallographic symmetry (NCS) restraints were used. Two TLS (Translation-Libration-Screw-rotation) groups, one per molecule, were refined. The refined model includes residues 5–289 of each of the two PSTPIP1 molecules in the AU, and 310 water molecules. The structure has 99.3% residues in most favored regions of the Ramachandran plot and the remaining in additionally allowed regions (TableS1). Crystallization andstructure determination ofthewild type F‑BAR domain ofPSTPIP1 Crystals of the wild type F-BAR domain of PSTPIP1 were obtained similarly as for the G258A mutant. Briefly, a protein solution at 0.5mg/mL in 10mM Tris/HCl (pH 7.5), 100mM NaCl, 0.5mM DTT was mixed with an equal volume of 100mM Bis–Tris-propane (pH 6.0), 17% (w/v) PEG 3350, 150mM sodium citrate and was equilibrated against the latter. Crystals were flash cooled in liquid nitrogen using 19% glycerol as cryoprotectant. Data were collected and processed as described above. A high multiplicity dataset was obtained by combining two sets of 2400 and 1200 images, 0.15° oscillation per image, collected on two separate regions of the same crystal. Crystals were isomorphic to those of the G258A mutant. The diffraction was strongly anisotropic with approximate resolution limits of 2.12, 4.32, and 2.17Å in the directions a*, b* and c*, respectively. The structure was refined against the anisotropically corrected data as for the G258A mutant, using the structure of the mutant as the starting model. The refined structure had a free R factor of 24.1% (TableS1). Crystallization andstructure determination oftheF‑BAR domain ofPSTPIP1 incomplex withLYP To obtain crystals of the PSTPIP1/LYP complex, the peptide of the CTH of LYP (residues 787–807) was soaked into preformed crystals of the wild type F-BAR of PSTPIP1. A 4.2mM solution of the peptide in 100mM Bis–Tris-propane (pH 6.0), 20% PEG 3350, 200mM sodium citrate was added to a crystallization drop with the PSTPIP1 crystals. After J.A.Manso et al. 1 3 131 Page 6 of 17 1h incubation at room temperature the crystals were briefly transferred into a cryoprotectant solution containing 20% glycerol and were flashed cooled. Diffraction data were collected and processed as for the previous crystals. A dataset was obtained by combining three sets of 1800 images, 0.1° oscillation per image, measured in three regions of a crystal. Crystals were isomorphic to those of the wild type and G258A mutant of PSTPIP1. The diffraction was also highly anisotropic; the approximate resolution limits were 4.05Å in the direction b* and ~ 2.1Å in directions a* and c*. The structure of the complex was refined against the anisotropically corrected data as for the isolated PSTPIP1 structures. The structure of the wild type F-BAR was used as starting model. The LYP segment was built using Coot and 2mFobs-DFcalc maps. Refinement converged to a free R factor of 24.1%. The refined model (TableS1) includes residues 5–289 and 3–289 of the two PSTPIP1 molecules in the AU, respectively, residues 793–806 of LYP, 162 waters, and two molecules of glycerol. Sequence analysis Sequences similar to the PEST phosphatase-binding site of PSTPIP1/2 were identified using HMMER (v3.3) [45]. A hidden Markov model profile build using the regions 225–262 of PSTPIP1 and PSTPIP2 was used to search the UniProtKB/ Swiss-Prot reference proteomes (Release 2020/01) [46]. This identified similarity with regions of Schizosaccharomyces pombe Cdc15 (Uniprot entry Q09822, residues 234–271) and Imp2 (Uniprot Q10199, residues 221–258). Structure analysis Protein–protein contacts were analyzed with the PISA server [47]. Calculation of electrostatic potential was performed with the program APBS [48]. Molecular figures were prepared with PyMol [49]. The Cdc15/Cdc12 complex was modeled using the structure of LYP bound to PSTPIP1 as template. The PSTPIP1/LYP complex was superimposed onto the structure of Cdc15 by fitting the α4 helices of the F-BAR dimers. Next, the residues in LYP were mutated to the equivalent residues in Cdc12; the side chains that were changed were modeled as energetically favorable conformations. Results Effect ofPSTPIP1 mutations associated toPAPA ontheinteraction withLYP andother PEST phosphatases PSTPIP1 mutations A230T and E250Q, which were initially identified as the cause of PAPA autoinflammatory disease, restricted PSTPIP1 binding to the tyrosine phosphatase PTP-PEST [1], implying that the interaction between PSTPIP1 and PEST phosphatases is important in the etiology of PAPA autoinflammatory syndrome. Furthermore, since the discovery of A230T and E250Q mutations in PSTPIP1 in 2002, over 20 missense mutations have been linked to autoinflammatory diseases (Fig.1A) [23]. Given that LYP is involved in inflammatory autoimmune diseases such as arthritis and lupus, we wanted to see how PSTPIP1 mutations affected its interaction with LYP. Toward this end, transiently transfected HEK293 cells with full-length PSTPIP1 and LYP were used to assess, by immunoprecipitation, the effect on PSTPIP1/ LYP interaction of an ample set of mutations identified in PSTPIP1 in patients with autoinflammatory diseases. The mutation W232A, which prevents PSTPIP1 from interacting with PTP-PEST [50] and LYP [26], but is unrelated to illness, was also included. Our results showed that mutations R228C, D246N, E250Q, E250K and E257K, in addition to W232A, reduced the interaction with LYP (Fig.1B). To exclude any possible indirect effect of the inhibitory mutations on the interaction between PSTPIP1 and LYP in HEK293 cells, we also analyzed their direct binding invitro. Using the purified F-BAR domain of PSTPIP1 (residues 1–289) and a synthetic peptide of the CTH of LYP (residues 787–807) labeled with fluorescein, we measured the affinity of the interaction (Fig.1C, D). The wild type F-BAR and the A230T mutant bound to LYP with similar affinity; their apparent dissociation constants (kd) were ~ 20nM. On the other hand, the binding affinities of the mutants E250Q and W232A were drastically lower (~ 60-fold), in agreement with the loss of interaction observed in the immunoprecipitation experiments. As our results partially differed from those reported previously for PTP-PEST [1], we tested the interaction of this phosphatase and PTPN18 with PSTPIP1 mutants to see whether this discrepancy was merely due to the phosphatase used in the assay (Supplemental Fig. S1). In these experiments, the effect of the assayed point mutations in PSTPIP1 on the interaction with PTP-PEST and PTPN18 was identical to that observed for the interaction with LYP. Collectively, our data suggest that the PEST phosphatases LYP, PTP-PEST and PTPN18 interact with PSTPIP1 in a similar way, and that the amino acids R228, W232, D246, E250 and E257 of PSTPIP1 are important for the association with the PEST phosphatases, indicating that a cluster of residues in PSTPIP1 F-BAR domain is critical for the interaction with these phosphatases. Given that PSTPIP1 is located in the membrane through the F-BAR domain, we tested whether the mutations that alter the interaction with LYP, affected its recruitment to the membrane. We transfected HEK293 cells with full-length LYP-mCherry and PSTPIP1-eGFP wild type or mutated PSTPIP1‑LYP phosphatase interaction: structural basis andimplications forautoinflammatory… 1 3 Page 7 of 17 131 in amino acids that make contact with LYP. Colocalization between LYP and PSTP1 constructs was studied by confocal microscopy using the Manders’ colocalization index M2 [31] (Fig.1E and Supplemental Fig. S2). These data indicate that PSTPIP1 mutations that reduce the interaction with LYP, also diminished its recruitment to the membrane, what would affect accessibility of LYP to its substrates and subsequently its dephosphorylation by LYP. Structure oftheF‑BAR domain ofPSTPIP1 To better understand the molecular basis of the physiological function of PSTPIP1 in the immune system, and its role in autoinflammatory diseases, we elucidated the 3D structure of the F-BAR domain (residues 1–289) using X-ray crystallography. Initially, we solved the structure of the G258A mutant, and subsequently that of the wild type protein. The two structures were almost identical. After superimposition, the root mean square deviation (rmsd) between all the equivalent Cα atoms was 0.275 Ǻ, which was similar to the estimated errors of the coordinates of the structures (0.23–0.24 Ǻ). Since the G258A structure was refined to a higher resolution, hereafter, we refer to it. The asymmetric unit of the crystal contains two molecules of PSTPIP1 that form the characteristic elongated crescent-shape of F-BAR dimers (Fig.2A) [50]. Each protomer consists of five α-helices, with the longer helices, α2, α3, and α4, forming a helical bundle. The dimerization interface is mainly formed by α2, α4, and the shorter helix, α5. Part of helices α3 and α4 protrude from the central dimerization core region forming the so-called wings, which are bent ~ 30° with respect to the longitudinal axis of the central region when viewed from the concave or convex sides, resulting in a tilted shape. The F-BAR dimer of PSTPIP1 bears structural resemblance to the F-BAR of Imp2 of Schizosaccharomyces pombe (PDB ID 5C1F, 27% sequence identity) [51]; after superimposition, the rmsd for 556 Cα atoms was 2.4 Ǻ. PSTPIP1 is also very similar to the F-BAR domains of Hof1p of Saccharomyces cerevisiae (PDB ID 4WPE, rmsd 2.7 Ǻ for 550 Cα atoms, 15% sequence identity) [52], Cdc15 of S. pombe (PDB ID 6XJ1, rmsd 2.9 Ǻ for 532 Cα atoms, 20% sequence identity) [53], and the human Cdc42-inter- actin protein 4 (CIP4) (PDB ID 2EFK, rmsd 3.1 Ǻ for 534 Cα atoms, 20% sequence identity) [8]. The minor differences between PSTPIP1 and these F-BAR structures are most noticeable in the wings (Supplemental Fig. S3). The 90º α5 α4 α3 α3 α4 α5 α1 C C ~215 Å α1 AB 180º Convex side E214 D222 Concave side K55 R109 K113 R116 R37 K34 R33 K117 K128 K136 K135 K142 K147 K168 K172 K176 R62 K63 R52 R160 R149 +7 kT/e 0 kT/e -7 kT/e Convex side E247 E251 E257 D262 E250 D266 D246 Wing Side view Central core α3 C α4 α5 N C N α1 α3 α4 α1 α5 α2 α2 Concave side Convex side Dimer 2-fold axis Fig. 2 Structure of the F-BAR domain of PSTPIP1. (A) Two orthogonal views of a ribbon representation of the structure of PSTPIP1. Protomers of the dimer are colored in orange and blue, respectively. (B) Surface representation of the F-BAR dimer colored by the electrostatic potential. Clusters of basic residues in the concave surface are encircled with dashed lines. Acidic and basic residues on the surface are labeled in one of the protomers J.A.Manso et al. 1 3 131 Page 8 of 17 curvature of the F-BAR dimer of PSTPIP1 is most similar to that of Cdc15 and CIP4, and it is only slightly straighter than the F-BAR domains of Imp2 and Hof1p. This is supported by the fact that the liposome tubules induced by PSTPIP1 (diameter 60–80nm) [54] and the tubules formed by Imp2 [51] and CIP4 [55] present a comparable diameter. The concave side of the F-BAR dimer is overall electropositive and contains two clusters of basic residues in each protomer, one in the core region, and the other near the tip of the wing (Fig.2B). Similar electro-positive patches are observed in the F-BAR domains of Imp2 [51], FCHo2 (PDB ID 2V0O) [56], the Gem-interacting protein (PDB ID 3QWE), and the RhoGAP Rgd1p (PDB ID 4WPC) [52] (Supplemental Fig. S4); which are important for binding to membranes with anionic groups. On the other hand, the central area of the convex surface has strong electronegative character due to the presence of multiple acidic residues. Large acidic surfaces are observed in the F-BAR domains of other proteins, such as Cdc15, growth arrest-specific 7 (GAS7), and formin-binding protein 17 (FBP17), but are absent in several F-BAR domains (Supplemental Fig. S4). The residues of PSTPIP1 critical for binding to PEST phosphatases (see above) are located in this electronegative area, supporting the notion that the acidic convex area is important for PSTPIP1 function. Structure oftheF‑BAR domain ofPSTPIP1 incomplex withLYP To visualize directly the PSTPIP1/LYP interaction we produced crystals of the F-BAR domain of PSTPIP1 in complex with a peptide of the CTH of LYP, residues 787–807. An additional continuous electron density, adjacent to PSTPIP1, was assigned to residues 793–806 of LYP (Supplemental Fig. S5). No electron density was observed for the first six residues of the CTH, suggesting that they do not engage in specific contacts with PSTPIP1. A single copy of LYP binds to the convex surface of the PSTPIP1 dimer at the rim of the dimerization interface formed by helices α4 of the two protomers (Fig.3A). LYP buries ~ 625 Ǻ2 (2.1%) of the surface area of the PSTPIP1 dimer and establishes distinct contacts with each of the PSTPIP1 protomers; hereafter we refer to them as A and B. The LYP-binding site is characterized by three hydrophobic pockets aligned on the surface of PSTPIP1 (Fig.3B, C). The first segment of LYP, residues 793–800, binds in an extended conformation. P795 docks in a first pocket that is delimited by V243 and D246 of each protomer. D246 of protomer B also forms hydrogen bonds (H-bond) with the backbone amide groups of K796 and G797 of LYP. P798 is cradled in a shallow second pocket. The side chain of the adjacent R799 makes ionic contacts with E250 and E257 of the protomer B; in addition, the main-chain amide and carbonyl of R799 make H-bonds with the side chain of N236 of protomer A. The C-terminal part of LYP (801–805) interacts with a deeper pocket formed between R228 and W232 of protomer A. P801 and P802 adopt a poly-proline II helix conformation; the ring of P801 contacts W232 in parallel. P802-W805 form a type I β-turn stabilized by a H-bond between the carbonyl of P802 and the amide group of W805. This creates a stacking of the rings of P802 and W805. In turn, W805 makes a stacking cation-π interaction with R228 of PSTPIP1. P803 and T804 of LYP are exposed to the solvent and do not contact PSTPIP1. Nonetheless, they might contribute to the stabilization of the LYP backbone conformation, because Pro and Thr frequently appear in these positions of type I β-turns [57]. The side chains of K794 and K796 of LYP do not engage in specific contacts with PSTPIP1. Nevertheless, K794, K796, and R791 (the latter is disordered in the structure) might favor the association of LYP with PSTPIP1 via electrostatic complementarity with the acidic surface around the binding site. Finally, the conformation of the PSTPIP1 residues involved in LYP binding is very similar in the free and bound structures (Supplemental Fig. S6). The pre-organized conformation of the LYP-binding site might favor the interaction. The residues of PSTPIP1 that participate in the LYP- binding area are conserved in PSTPIP2 (Fig.3D), with the exception of V233, M240, and L254, which are in the periphery of the interface and do not make specific contacts with LYP. The LYP-binding interface is more conserved (78% identity between PSTPIP1 and PSTPIP2), than the rest of their F-BAR domains (49% identity), suggesting functional conservation. On the other hand, the residues of LYP that make the key interactions with PSTPIP1 are conserved in PTP-PEST and PTPN18 (Fig.3E). Collectively, the conservation of the interaction interfaces supports the notion that the structure of the PSTPIP1/LYP complex represents a general model for the binding of the PEST phosphatases LYP, PTP-PEST, and PTPN18, to PSTPIP1 and PSTPIP2. To verify experimentally the LYP residues critical for the interaction we mutated in the CTH sequence amino acids R799 and W805 to Ala. Invitro binding assays were conducted with His-PSTPIP1 recombinant full-length protein obtained from bacteria and GST-CTH proteins. Mutations R799A and W805A reduced the interaction with PSTPIP1, in agreement with structural data (Fig.3F). We also analyzed whether the characteristics of the LYP-binding site of PSTPIP1/2 are present in other F-BAR domains. Using a profile hidden Markov model method, sequence similarity was detected between the α4 segments of PSTPIP1/2 and the S. pombe proteins Cdc15 and Imp2. The sequence identity in the α4 between PSTPIP1 and Cdc15 and Imp2 is 59 and 42%, respectively; while the overall identity is lower (27 and 32%). The key residues of the LYP binding PSTPIP1‑LYP phosphatase interaction: structural basis andimplications forautoinflammatory… 1 3 Page 9 of 17 131 Fig. 3 Structural determinants of the PSTPIP1/LYP interaction. A Orthogonal views of the overall structure of the complex. The two protomers of the PSTPIP1 dimer, A (orange) and B (blue), and LYP (magenta) are colored similarly throughout the figure. B Interaction of LYP, shown as sticks, with PSTPIP1, shown as surface. Residues of PSTPIP1 that participate in the interface are labeled. The three hydrophobic pockets of the binding site are highlighted by dashed-line circles. C Two views showing contacts between LYP and PSTPIP1. H-bonds and salt bridges are shown as dashed lines. D Sequence alignment of the regions of human PSTPIP1 and PSTPIP2 (Uniprot Q9H939) that contain the LYP-binding site. Residues from protomers A and B, which form the interface, are indicated by inverted orange triangles and blue circles, respectively. Residues whose solvent exposed surface was buried ≤ 20% by LYP are indicated by open symbols. Conserved residues are shown in green boxes. E Multiple sequence alignment of the CTH regions of phosphatases LYP, PTP-PEST (Uniprot Q05209) and PTPN18 (Uniprot Q99952). Residues of LYP that make specific contacts with PSTPIP1 are indicated by inverted triangles. Conserved residues are shown in green boxes. (F) His-PSTPIP1 fulllength recombinant protein was subjected to pulldown assays with the indicated CTH peptide fused to GST. The presence of PSTPIP1 in the precipitates was visualized by IB with antibody against PSTPIP1, and GST– peptides were detected with an antibody against GST A C B E D F J.A.Manso et al. 1 3 131 Page 16 of 17 34. Kabsch W (2010) Xds. Acta Crystallogr D Biol Crystallogr 66:125–132 35. Evans P (2006) Scaling and assessment of data quality. Acta Crystallogr D Biol Crystallogr 62:72–82 36. Evans PR, Murshudov GN (2013) How good are my data and what is the resolution? Acta Crystallogr D Biol Crystallogr 69:1204–1214 37 Tickle IJ, Flensburg C, Keller P, Paciorek W, Sharff A, Vonrhein C, Bricogne G (2017) STARANISO in. Global Phasing Ltd, Cambridge 38. Vonrhein C, Flensburg C, Keller P, Sharff A, Smart O, Paciorek W, Womack T, Bricogne G (2011) Data processing and analysis with the autoPROC toolbox. Acta Crystallogr D Biol Crystallogr 67:293–302 39. McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn MD, Storoni LC, Read RJ (2007) Phaser crystallographic software. J Appl Cryst 40:658–674 40. Krivov GG, Shapovalov MV, Dunbrack RL Jr (2009) Improved prediction of protein side-chain conformations with SCWRL4. Proteins 77:778–795 41. Adams PD, Afonine PV, Bunkoczi G, Chen VB, Davis IW, Echols N, Headd JJ, Hung LW, Kapral GJ, Grosse-Kunstleve RW, McCoy AJ, Moriarty NW, Oeffner R, Read RJ, Richardson DC, Richardson JS, Terwilliger TC, Zwart PH (2010) PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr 66:213–221 42. Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot. Acta Crystallogr D Biol Crystallogr 66:486–501 43. Suhre K, Sanejouand YH (2004) On the potential of normal-mode analysis for solving difficult molecular-replacement problems. Acta Crystallogr D Biol Crystallogr 60:796–799 44. Terwilliger TC, Grosse-Kunstleve RW, Afonine PV, Moriarty NW, Zwart PH, Hung LW, Read RJ, Adams PD (2008) Iterative model building, structure refinement and density modification with the PHENIX AutoBuild wizard. Acta Crystallogr A 64:61–69 45. Potter SC, Luciani A, Eddy SR, Park Y, Lopez R, Finn RD (2018) HMMER web server: 2018 update. Nucleic Acids Res 46:W200–W204 46. (2019) UniProt: a worldwide hub of protein knowledge. Nucleic Acids Res 47:D506–D515 47. Krissinel E, Henrick K (2007) Inference of macromolecular assemblies from crystalline state. J Mol Biol 372:774–797 48. Jurrus E, Engel D, Star K, Monson K, Brandi J, Felberg LE, Brookes DH, Wilson L, Chen J, Liles K, Chun M, Li P, Gohara DW, Dolinsky T, Konecny R, Koes DR, Nielsen JE, Head-Gordon T, Geng W, Krasny R, Wei GW, Holst MJ, McCammon JA, Baker NA (2018) Improvements to the APBS biomolecular solvation software suite. Protein Sci 27:112–128 49. The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC in 50. Daumke O, Roux A, Haucke V (2014) BAR domain scaffolds in dynamin-mediated membrane fission. Cell 156:882–892 51. McDonald NA, Takizawa Y, Feoktistova A, Xu P, Ohi MD, Vander Kooi CW, Gould KL (2016) The tubulation activity of a fission yeast F-BAR protein is dispensable for its function in cytokinesis. Cell Rep 14:534–546 52. Moravcevic K, Alvarado D, Schmitz KR, Kenniston JA, Mendrola JM, Ferguson KM, Lemmon MA (2015) Comparison of Saccharomyces cerevisiae F-BAR domain structures reveals a conserved inositol phosphate binding site. Structure 23:352–363 53. Snider CE, Chandra M, McDonald NA, Willet AH, Collier SE, Ohi MD, Jackson LP, Gould KL (2020) Opposite surfaces of the Cdc15 F-BAR domain create a membrane platform that coordinates cytoskeletal and signaling components for cytokinesis. Cell Rep 33:108526 54. Tanaka-Takiguchi Y, Itoh T, Tsujita K, Yamada S, Yanagisawa M, Fujiwara K, Yamamoto A, Ichikawa M, Takiguchi K (2013) Physicochemical analysis from real-time imaging of liposome tubulation reveals the characteristics of individual F-BAR domain proteins. Langmuir 29:328–336 55. Frost A, Perera R, Roux A, Spasov K, Destaing O, Egelman EH, De Camilli P, Unger VM (2008) Structural basis of membrane invagination by F-BAR domains. Cell 132:807–817 56. Henne WM, Kent HM, Ford MG, Hegde BG, Daumke O, Butler PJ, Mittal R, Langen R, Evans PR, McMahon HT (2007) Structure and analysis of FCHo2 F-BAR domain: a dimerizing and membrane recruitment module that effects membrane curvature. Structure 15:839–852 57. Hutchinson EG, Thornton JM (1994) A revised set of potentials for beta-turn formation in proteins. Protein Sci 3:2207–2216 58. Willet AH, McDonald NA, Bohnert KA, Baird MA, Allen JR, Davidson MW, Gould KL (2015) The F-BAR Cdc15 promotes contractile ring formation through the direct recruitment of the formin Cdc12. J Cell Biol 208:391–399 59. Li SS (2005) Specificity and versatility of SH3 and other prolinerecognition domains: structural basis and implications for cellular signal transduction. Biochem J 390:641–653 60. Senju Y, Itoh Y, Takano K, Hamada S, Suetsugu S (2011) Essential role of PACSIN2/syndapin-II in caveolae membrane sculpting. J Cell Sci 124:2032–2040 61. Hansen CG, Howard G, Nichols BJ (2011) Pacsin 2 is recruited to caveolae and functions in caveolar biogenesis. J Cell Sci 124:2777–2785 62. Yao G, Su X, Nguyen V, Roberts K, Li X, Takakura A, Plomann M, Zhou J (2014) Polycystin-1 regulates actin cytoskeleton organization and directional cell migration through a novel PC1-Pacsin 2-N-Wasp complex. Hum Mol Genet 23:2769–2779 63. Nikki M, Merilainen J, Lehto VP (2002) FAP52 regulates actin organization via binding to filamin. J Biol Chem 277:11432–11440 64. Begonja AJ, Pluthero FG, Suphamungmee W, Giannini S, Christensen H, Leung R, Lo RW, Nakamura F, Lehman W, Plomann M, Hoffmeister KM, Kahr WH, Hartwig JH, Falet H (2015) FlnA binding to PACSIN2 F-BAR domain regulates membrane tubulation in megakaryocytes and platelets. Blood 126:80–88 65. Salzer U, Kostan J, Djinovic-Carugo K (2017) Deciphering the BAR code of membrane modulators. Cell Mol Life Sci 74:2413–2438 66. Nguyen LP, Tran SC, Suetsugu S, Lim YS, Hwang SB (2020) PACSIN2 interacts with nonstructural protein 5A and regulates hepatitis C virus assembly. J Virol 94:e01531–e01619 67. Garabedian MV, Stanishneva-Konovalova T, Lou C, Rands TJ, Pollard LW, Sokolova OS, Goode BL (2018) Integrated control of formin-mediated actin assembly by a stationary inhibitor and a mobile activator. J Cell Biol 217:3512–3530 68. Liu Y, McDonald NA, Naegele SM, Gould KL, Wu JQ (2019) The F-BAR domain of Rga7 relies on a cooperative mechanism of membrane binding with a partner protein during fission yeast cytokinesis. Cell Rep 26:2540-2548 e4 69. Tarricone C, Xiao B, Justin N, Walker PA, Rittinger K, Gamblin SJ, Smerdon SJ (2001) The structural basis of Arfaptin-mediated crosstalk between Rac and Arf signalling pathways. Nature 411:215–219 70. Nakamura K, Man Z, Xie Y, Hanai A, Makyio H, Kawasaki M, Kato R, Shin HW, Nakayama K, Wakatsuki S (2012) Structural basis for membrane binding specificity of the Bin/Amphiphysin/Rvs (BAR) domain of Arfaptin-2 determined by Arl1 GTPase. J Biol Chem 287:25478–25489 71 Holzinger D, Roth J (2019) PAPA syndrome and the spectrum of PSTPIP1-associated inflammatory diseases. In: Efthimiou P (ed) Auto-inflammatory syndromes: pathophysiology, diagnosis, and management. Springer International Publishing, Cham, pp 39–59 72. Wang D, Hoing S, Patterson HC, Ahmad UM, Rathinam VA, Rajewsky K, Fitzgerald KA, Golenbock DT (2013) Inflammation in mice PSTPIP1‑LYP phosphatase interaction: structural basis andimplications forautoinflammatory… 1 3 Page 17 of 17 131 ectopically expressing human pyogenic arthritis, pyoderma gangrenosum, and acne (PAPA) syndrome-associated PSTPIP1 A230T mutant proteins. J Biol Chem 288:4594–4601 73. Cote JF, Chung PL, Theberge JF, Halle M, Spencer S, Lasky LA, Tremblay ML (2002) PSTPIP is a substrate of PTP-PEST and serves as a scaffold guiding PTP-PEST toward a specific dephosphorylation of WASP. J Biol Chem 277:2973–2986 Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.