Author Correction: Early endosome autoantigen 1 regulates IL-1β release upon caspase-1 activation independently of gasdermin D membrane permeabilization

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

. Caspase-1 induces the release of EEA1. (A) Canonical NLRP3 activation by hypotonic solution (90 mOsm, 1 h) in LPS-primed THP-1. Cell lysate and concentrated cell-free supernatant were analyzed by immunoblot for EEA1, Rab5, LAMP1 and IL-1β. Arrows denotes a smaller product of EEA1 upon hypotonic stimulation. IL-1β released to the extracellular medium was also detected by ELISA (bottom). (B) Immunoblot analysis of EEA1 and caspase-1 on cell lysate and cell-free supernatant from LPS-primed THP-1 (wild-type, CASP1 −/− or CASP4 −/− ) after hypotonic stimulation. Where indicated, Ac-YVAD was added to inhibit caspase-1. (C) In vitro processing assay for recombinant human EEA1 expressed in HEK293 cells and incubated with either recombinant caspase-1, or with concentrated supernatant or cell lysates from wild type (WT) or Casp1/11 −/− LPS-primed BMDM activated with nigericin. Where indicated, Ac-YVAD was added to inhibit caspase-1. Dashed line separates two different blots. (D) In vitro processing assay for recombinant human EEA1 WT, D132A, D127A, or D127A/D132A mutants expressed in HEK293 cells and incubated with recombinant caspase-1. (E) Representative high resolution deconvolved images of LPS-primed BMDMs after incubation with hypotonicity and stained with fluorescent probe for active caspase-1 (FLICA, green) and the endosome marker EEA1 (red) and nuclei (DAPI, blue); bar, 5 µm. (F) Co-immunoprecipitation of EEA1 and the p10 subunit of active caspase-1 in cell supernatants of LPS-primed THP-1 macrophages after NLRP3 stimulation www.nature.com/scientificreports www.nature.com/scientificreports/ upon activation, caspase-1 cleaves EEA1 at Asp 127/132 and induces the release EEA1 protein from the cell. The release of EEA1 was independent of gasdermin D and also occurs on the alternative NLRP3 activation pathway that resulted in a pyroptosis-independent release of IL-1β. Deficiency of EEA1 in macrophages resulted in a decrease of IL-1β release upon inflammasome activation, without affecting pyroptosis. EEA1 release into a pro-inflammatory environment with IL-1β could promote the initiation of autoimmune diseases.

Results
EEA1 is processed and released by caspase-1. Macrophage swelling activates the NLR containing a Pyrin domain 3 (NLRP3) inflammasome and leads to the release of IL-1β upon caspase-1 activation 8,24,25 . Examination of THP-1 supernatants after NLRP3 inflammasome activation by hypotonic solution revealed the release of the endosome docking factor EEA1, but not Rab5 or the late-endosomal/lysosomal marker LAMP1 (Fig. 1A). Interestingly, hypotonic activation of the NLRP3 inflammasome resulted in the appearance of an additional lower molecular weight form of EEA1 in THP-1 cell lysates and supernatants, representing the 31.07 ± 15.24% and 7.97 ± 0.52% of the total, respectively (Fig. 1A). The processing and release of EEA1 was completely inhibited when the specific caspase-1 inhibitor Ac-YVAD was used or when THP-1 cells deficient on caspase-1, but not caspase-4, were used (Fig. 1B). Next, direct processing of EEA1 by caspase-1 was investigated by using an in vitro cleavage assay. We found that recombinant human caspase-1 was able to process recombinant EEA1 (Fig. 1C) and similarly, incubation of recombinant EEA1 with the supernatant from mouse bone marrow derived macrophages (BMDM) activated with nigericin, where native active caspase-1 is released 8 , also resulted in the cleavage of EEA1 (Fig. 1C). EEA1 processing was not observed using both supernatant or cell lysate from caspase-1/11-deficient mouse BMDMs (Fig. 1C). All these data show that EEA1 is a direct substrate of caspase-1. Bioinformatics analysis of human EEA1 sequence with ExPASy Peptide Cutter software 26 and GraBCas software found three Asp at positions 132, 738 and 762 as the best candidates for caspase-1 cleavage, however ExPASy Peptide Cutter software and GraBCas software did not find any Asp in EEA1 as candidate for caspase-11 cleavage. Cleavage at Asp 132 would result in a predicted EEA1 C-terminal fragment of 147 kDa ( Supplementary Fig. S1), similar to the size found for EEA1 processing in cellular experiments and in vitro cleavage assays. Asp 132 was confirmed as the cleavage site of caspase-1 after Edman N-terminus sequencing of EEA1 incubated with recombinant caspase-1. However, mutation of D132A result in a similar caspase-1 processing than wild-type EEA1 (Fig. 1D). EEA1 processing during apoptosis in different cell types has been reported to occur at Asp 127 and Asp 132 27 . Double EEA1 mutant D127A/D132A, but not single mutants, was not processed by caspase-1 (Fig. 1D).
Gasdermin D is dispensable for EEA1 release. Activation of caspase-1 through the canonical NLRP3 inflammasome activation by ATP, nigericin, and uric acid crystals resulted in the processing and release of EEA1 ( Fig. 2A), in parallel to caspase-1-dependent pyroptosis induction (Fig. 2B). Likewise, the non-canonical NLRP3 activation by live E. coli induced the release of EEA1 ( Fig. 2A). In this pathway, intracellular LPS binds directly to caspase-11, which processes gasdermin D and evoke pyroptosis and subsequent activation of casapse-1 by NLRP3 inflammasome activation 29 . Moreover, activation of caspase-1 by other inflammasomes (NLRC4 or NLRP1) resulted in the processing and release of EEA1 ( Supplementary Fig. 3A,B). Following ATP stimulation, the processing and release of EEA1 were completely inhibited in macrophages deficient for either P2X7 receptor, NLRP3 or ASC ( Supplementary Fig. S3C). However, the release of EEA1 was occurring in gasdermin D-deficient macrophages (Fig. 2C). In the absence of gasdermin D we found a small release of IL-1β that was blocked after caspase-1 inhibition (Fig. 2D,E). However, deficiency of gasdermin D blocked pyroptosis, measured as the release of LDH or inflammasome proteins (NLRP3 and ASC) (Fig. 2F,G), and plasma membrane pore formation, measured by Yo-Pro uptake (Fig. 2H). This suggest that gasdermin D induced plasma membrane permeabilization and pyroptosis execution is not responsible for EEA1 release upon caspase-1 activation, but were responsible for the majority of IL-1β released from macrophages. EEA1 is released during the alternative NLRP3 activation pathway. Human monocytes release IL-1β in the absence of pyroptotic cell death via the alternative activation pathway of NLRP3 13 . We found that upon alternative activation of NLRP3 in human monocytes there was processing and release of EEA1, and this process was blocked after caspase-1 inhibition (Fig. 3A). In parallel, mature IL-1β and caspase-1 were also released from monocytes (Fig. 3A,B). The release of cytokines and EEA1 from human monocytes was independent of pyroptosis, as the amount of extracellular LDH was not increased in supernatants (Fig. 3C). These data support the conclusion that EEA1 is released as consequence of caspase-1 activation in the absence of pyroptosis. with hypotonicity. Nitrocellulose membranes were cropped after protein blotting and then inmunodetection was carried out separately for each antibody. Complete membrane fragments are showed in all cases. Blots and pictures are representative of three to four independent experiments. ELISA data are presented as average ± sem from n = 3 independent experiments. www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ EEA1 knock-down impaired IL-1β release but not pyroptotic cell death. We next examined the effect of knocking-down EEA1 expression on IL-1β release. In cells transduced with lentivirus bearing a specific short hairpin RNA against Eea1, EEA1 expression was suppressed by 78.79 ± 2.11% compared to cells transduced with a lentivirus containing a non-target encoding sequence (Fig. 4A,B). Following canonical stimulation of NLRP3 inflammasome by ATP, EEA1 knock-down resulted in a decrease, but not completely blocking, of the release of IL-1β (45.46 ± 6.65% of release respect to non-target lentivirus) (Fig. 4A,C). However, the pyroptotic release of the inflammasome components NLRP3 and ASC (Fig. 4A) and LDH (Fig. 4D) was not affected by EEA1 silencing. Altogether the data show that caspase-1 could control the release of IL-1β by EEA1, independently and in parallel to gasdermin D induced pyroptotic cell death.

Discussion
Gasdermin D pore formation on the plasma membrane and subsequent pyroptotic cell death upon caspase-1 activation has been described as a pathway for unconventional IL-1β release 11,12,30,31 . Oligomeric NLRP3 inflammasome particles containing ASC aggregates are also released from macrophages via pyroptotic cell death 8,25,32 . However, in some circumstances, gasdermin D pore formation has been found as responsible for IL-1β release in the absence of pyroptotic cell death 19 , explaining the release of IL-1β in the absence of cell death. Here we found that upon inflammasome activation, caspase-1 cleaves the endosomal docking-protein EEA1 at Asp 127/132 and also induce its release. We cannot completely rule out the participation of caspase-11 in EEA1 processing after non-canonical inflammasome activation, however, human EEA1 sequence did not contain any Asp as candidate for caspase-11 cleavage, compared with the three Asp found as the best candidates for caspase-1 cleavage. The release of EEA1 was independent on gasdermin D and pyroptosis execution and account for the release of a pool of IL-1β. EEA1 release also occurs during the release of IL-1β in human monocytes in the absence of cell death upon the activation of the NLRP3 inflammasome alternative pathway.
Caspase-1 controls the unconventional release of the pro-inflammatory cytokine IL-1β as an important effector molecule of the inflammasome. Different unconventional secretion pathways have been described for IL-1β release, these involve intracellular vesicular trafficking (such as exocytosis of lysosomes or autophagosomes), the release of microvesicles and exosomes, or ultimately pyroptotic cell death induced by gasdermin D pore formation on the plasma membrane 9,10,12,31,33-36 . EEA1 represents an additional pathway to release IL-1β that is important for a pool of non-pyroptotic IL-1β release, as the release of IL-1β described in monocytes or neutrophils 13,37 . In fact, we found that the activation of the alternative pathway of NLRP3 in human monocytes also induces the www.nature.com/scientificreports www.nature.com/scientificreports/ release of EEA1. However, whether gasdermin D induced IL-1β release in the absence of pyroptosis 19 applies to the alternative NLRP3 activation pathway found in human monocytes is not known. It seems unlikely that EEA1 could control the unconventional release mechanisms already described for IL-1β 33,34 . Multivesicular bodies are the origin of exosomes after fusing with plasma membrane 38,39 and maturation of endosomes to multivesicular bodies, lysosomes or autophagolysosomes lack EEA1 protein association 40 . However, EEA1 could be controlling an endosomal-related release of IL-1β, since it is a key controller protein of endosomal docking, clustering and fusion via its association to small GTP-binding proteins of the Rab family [41][42][43] . Although we found EEA1 cleavage and release controlled by caspase-1, we cannot formally conclude that EEA1 cleavage by caspase-1 is dispensable for its release. The generation of macrophages carrying double EEA1 mutant D127A/D132A will be key to determine if caspase-1 processing of EEA1 is a key step for its release. However, due to the differential release of EEA1 www.nature.com/scientificreports www.nature.com/scientificreports/ forms (unprocessed vs. processed) from the different experiments presented in our study, we cannot conclude that EEA1 processing is a determinant step for its release.
During pyroptotic cell death, the release of IL-1β via gasdermin D pore formation and subsequent osmotic cell lysis 44 overlaps to the release of IL-1β induced by EEA1. Silencing EEA1 expression decreases the release of mature IL-1β from macrophages, without affecting other markers of pyroptosis. Furthermore, deficiency of gasdermin D was not affecting EEA1 release upon caspase-1 activation. A recent gasdermin D-independent form of lytic cell death has been described upon NLRP3 inflammasome activation 45 , but EEA1 controlled IL-1β release will not explain this pathway as EEA1 did not affect cell lysis. Furthermore, Monteleone et al. 46 have recently described that IL-1β maturation is sufficient for slow, gasdermin D-independent secretion of ectopic IL-1β from non-pyroptotic macrophages, but the speed of IL-1β release is boosted by inflammasome activation, via caspase-1 and gasdermin D. Mature IL-1β was found concentrated in plasma membrane ruffles enriched with phosphatidylinositol-4,5-biphosphate (PIP2), and this was a prerequisite for the slow and non-pyroptotic IL-1β secretion 46 . PIP2 is important for endocytic turnover trafficking and early endosomal homeostasis 47 , suggesting that it could be also important for the EEA1 controlled IL-1β release.
Autoantibodies to EEA1 have been described in patients with subacute cutaneous systemic lupus erythematosus, polyarthritis, rheumatoid arthritis, and neurological diseases 48,49 . It is poorly known how EEA1 could be exposed to the extracellular environment to induce autoantibody production. Although it is commonly assumed that these proteins could be present on apoptotic debris 50 , apoptosis and removal of apoptotic debris are immunologically silent process. Here we imply caspase-1 activation as controlling the release of EEA1 into a pro-inflammatory milieu that could favor autoantibodies generation against EEA1.
Taking together, our results point that caspase-1 induces the release of EEA1 independently and parallel to pyroptotic cell death, accounting for a pool of IL-1β release in the absence of gasdermin D plasma membrane permeabilization. During this process, EEA1 release could be important for the development of autoimmune diseases.
Reagents. E. coli LPS O55:B5, DAPI, ATP, triton X-100, nigericin, anti-V5 antibody, PMA and puromycin resistant MISSION ® shRNA Lentiviral Transduction Particles were from Sigma-Aldrich; Recombinant human caspase-1 and caspase-1 inhibitor Ac-YVAD-AOM were from Merk-Millipore; MSU crystals were from Enzo Life Sciences; flagellin from S. typhimurium from Invivogen; the receptor-binding protein protective antigen and metalloprotease lethal factor were from List Biological; the protein-delivery reagent PULSin was from PolyPlus Transfection;Yo-Pro-1 from Life Technologies. Antibodies for caspase-1 p10, IL-1β, ASC and LAMP1 were from Santa Cruz Biotechnology. Mouse monoclonal anti-NLRP3 was from AdipoGen. Mouse monoclonal anti-Rab5 and anti-EEA1 were from BD Biosciences. All HRP-conjugated secondary antibodies were from GE Healthcare and fluorescent-conjugated secondary antibodies were from Life Technologies.
Human EEA1 construct 56 was sub-cloned into pcDNA3.1/V5-His TOPO vector (Life Technologies) with a Nt c-Myc and Ct V5 epitope tags. EEA1 D127A and D132A mutations were incorporated by overlapping PCR. The full coding region of the constructs used in this study were verified by DNA sequencing. EEA1 expression vector was transfected into HEK293 cells with Lipofectamine 2000 according to the manufacturer's instructions (Life Technologies).
Amino terminal sequencing of cleaved EEA1. Over expressed recombinant human EEA1 was extracted from transfected HEK293 cells by incubation in cold lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 1% NP-40; pH 7.5). After centrifugation, the supernatant was incubated with anti-V5-tag mAb-Magnetic Beads (MBL International Corporation) for 3 h at 4 °C. Magnetic beads were isolated and washed on a magnetic rack, and then, www.nature.com/scientificreports www.nature.com/scientificreports/ suspended in ICE buffer (0.1 M HEPES, 0.1% CHAPS, 10% Sucrose, 2.9 mM DTT; pH 7.5) with 1 U recombinant human caspase-1 for 2 h at 37 °C. Magnetic beads were then isolated, washed and suspended in 40 µl of reducing loading sample buffer (Invitrogen), boiled and subjected to SDS-PAGE. Gels were blotted to a PVDF membrane and proteins were revealed by Coomassie brilliant blue protein staining (BioRad). Cleaved EEA1 was cut and sent to Protein Chemistry Laboratory of the CIB-CSIC (Madrid, Spain), where it was processed and subjected to N-terminal sequencing by Edman's sequential degradation in a Procise 494 protein sequencer (Perkin Elmer). Eea1 gene silencing. For Eea1 gene silencing, bone marrow cells were infected with short hairpin RNA (shRNA) lentiviral transduction particles, and experiments were performed after 7 days of macrophage differentiation. In brief, 10 6 of bone marrow cells were plated in 6 well plates (day 0). At day 3, medium was removed and new medium containing 8 µg/ml hexadimethrine bromide (Sigma-Aldrich) and 10 MOI of lentiviral particles (EEA1 specific or non-target control) were added. On next day, medium with 3 µg/ml of puromycin was added, and cells were used at day 7.
Immunocytochemistry and microscopy. Macrophages stimulated on cover slips were washed twice with PBS, fixed with 4% formaldehyde/4% sucrose in PBS for 5 min at room temperature, and then washed three times with PBS. Cells were blocked with 1% bovine serum albumin (Sigma) and permeabilized with 0.1% triton X-100 in PBS for 30 min at room temperature before incubating with primary antibody. Primary antibodies used were anti-EEA1(1:1000, BD Biosciences), anti-LAMP1 (1:1000, Santa Cruz Biotechnology), and anti-Rab5 (1:500, Cell Signalling). Primary antibodies were incubated for 16 h at 4 °C. Cells were washed and incubated with appropriate fluorescence-conjugated secondary antibody (1:200) for 2 h at room temperature, then rinsed in PBS. To stain for active caspase-1, macrophages were incubated with the fluorophore inhibitor of caspase-1, green fluorescent peptide 5-carboxyfluorescein-Tyr-Val-Ala-Asp-fluoromethyl ketone (FAM-YVAD-fmk, FLICA), according to the manufacturer's recommendations (Immunochemistry Technologies). Labeling lysosomes with LysoTraker Red DND-99 (Life Technologies) was performed following manufacturer's recommendations. In both protocols, cells were fixed with 4% formaldehyde, blocked, permeabilized and co-stained for EEA1 or LAMP1 as described above. All coverslips were mounted on slides with Fluoroshield with DAPI (Sigma-Aldrich). Images were acquired either with a Delta Vision RT (Applied Precision) restoration microscope using a Coolsnap HQ (Photometrics) camera, with a 60×/1.42 Plan Apo or 100×/1.40 Uplan Apo objectives, a Z optical spacing of 0.2 μm and the 360 nm/475 nm, 490 nm/528 nm and555nm/617 nm filter sets (Chroma 86000v2). Raw images were deconvolved using Softworxor ImageJ (NHI) software and maximum intensity projections of these deconvolved images are shown in the results. Co-localization analysis was performed using the Coloc 2 plugin for ImageJ (NHI).

Co-immunoprecipitation and Western blotting.
Detailed methods used for Western blot analysis have been described previously 24 . For co-immunoprecipitation, equal amounts of cell supernatant were incubated with 1.5 µg of anti-EEA1 or normal mouse IgG 1 (1 h, 4 °C) and then with protein-G-agarose beads (Merk-Millipore, 1 h, 4 °C). Beads were washed and heated (5 min, 80 °C) with reducing loading sample buffer (Invitrogen).
IL-1β ELISA. IL-1β release was measured by ELISA for mouseIL-1β (R&D Systems) or human IL-1β (eBiosciences) following the manufacturer's instructions and read in a Synergy Mx plate reader (BioTek). The concentration of IL-1β was estimated using a standard with known concentrations of recombinant IL-1β.
Lactate dehydrogenase (LDH) release and Yo-Pro-1 uptake. LDH release was measured using the Cytotoxicity Detection kit (Roche) following the manufacturer's instructions, and expressed as percentage of total cell LDH content. Yo-Pro-1 uptake was performed as already described 9 .
Statistical analysis. Statistics were calculated using Prism software (GraphPad). Comparisons of two groups were analyzed using two-tailed t-test and comparisons of multiple groups were analyzed by mean of analysis of variance (ANOVA) with Bonferroni's post-test.
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.