Myoferlin-Mediated Lysosomal Exocytosis Regulates Cytotoxicity by Phagocytes

Phagocytes (such as macrophages, neutrophils, and dendritic cells) internalize a variety of foreign particles, bacteria, and dead cells into a subcellular compartment called a phagosome, where they are digested to yield a series of peptides following the fusion of the phagosome with a lysosome (1). Phagocytosis and the subsequent digestion of enclosed particles are essential processes for eukaryotes to protect themselves from bacterial and viral infections and for the maintenance of homeostasis (2). Lysosomes are filled with more than 50 different hydrolytic enzymes, which are capable of digesting most cellular components, including proteins, nucleic acids, carbohydrates, and lipids (3). It has been reported that phagocytes activated by bacillus Calmette–Guérin infection secrete lysosomal enzymes with cytotoxic activity against tumor cells, which is called heterolysis (4). During contact with the activated phagocytes, the target tumor cells undergo degenerative changes, including the clumping of nuclear chromatin, vacuolation, and retraction of cytoplasm, which is inhibited by preventing lysosomal exocytosis. However, the molecular mechanisms that regulate lysosomal exocytosis by phagocytes are largely unknown.

Several hematopoietic cells have cell-specific secretory lysosomes, which are secreted in response to external stimuli (5), including neutrophil azurophil granules, platelet dense granules, eosinophil granules, basophil and mast cell histamine granules, and CTL lytic granules. Among these cells, CTLs have been most extensively studied in terms of the exocytic machinery of the secretory lysosomes, which cause the release of membrane-bound Fas ligand, the pore forming protein perforin, and granzyme serine protease to induce the killing of target cells (6). In CTLs, lysosomal exocytosis is mediated by a complex formed of small GTPase Rab27A and its effector Munc 13-4 that tethers the lytic granules to the plasma membrane to allow fusion (7). Mutations in the Munc 13-4 gene cause familial hemophagocytic lymphohistiocytosis type 3, which causes immunodeficiency due to defects in the lysosomal exocytosis of cytolytic enzymes by CTLs (8). In several cell types, lysosomal exocytosis is also regulated by the Ca²⁺ sensor synaptotagmin-VII (Syt VII) (9). Syt VII is present on the lysosomal membrane and regulates Ca²⁺-triggered exocytosis by binding to the plasma membrane, a process that is required for repairing damaged plasma membranes. Mice deficient in Syt VII exhibit defective membrane repair, which results in myopathy (10). Both Munc 13-4 and Syt VII possess C2 domains that mediate Ca²⁺-dependent binding to phosphatidylserine (PS) in the inner leaflet of the plasma membrane, which is a critical process for membrane fusion (11). Myoferlin is a type II transmembrane protein with seven C2 domains in the cytoplasmic region (12, 13). The first C2 domain (C2A domain) binds PS in a calcium-dependent manner (14). Myoferlin was identified as a protein expressed in the plasma membrane of myoblasts undergoing fusion and has been implicated in the repair of injured plasma membranes (15, 16). Myoferlin is also known to regulate various aspects of tumor progression and cancer cell motility (17, 18). It is believed that injury-induced Ca²⁺ influx through membrane lesions triggers endocytosis and the generation of endocytosed vesicles expressing myoferlin, which fuse with the injured membrane to yield a membrane patch (16). This characteristic of myoferlin led us to hypothesize that it may be a regulator of calcium-dependent lysosomal exocytosis by phagocytes.

NIH3T3 cells were obtained from RIKEN BioResource Research Center, (Ibaraki, Japan), and cultured in DMEM (Nacalai Tesque, Kyoto, Japan) supplemented with 10% FBS (Biowest). Bone marrow–derived macrophages (BMDMs) were prepared as previously described (19). pCDNA3.1-Myoferlin HA was a gift from W. Sessa (20) (plasmid no. 22443; Addgene). GFP was fused to the N terminus of myoferlin by amplifying the GFP locus on pCAG-GFP using the following primers: forward (Fw) (5′-GCCACCATGGTGAGCAAGGG-3′) and reverse (Rv) (5′-GATATCTTGTACAGCTCGTCCA-3′). The product was cloned into pGEM-T Easy Vector (Promega), digested by NotI and EcoRV sites, and subsequently inserted into the pCDNA3.1-Myoferlin HA. The Δ C2A mutant of myoferlin was constructed by depleting the amino acid region from 2 to 115 loci. Each plasmid DNA was introduced into NIH3T3 cells by lipofection using FuGENE6 (Promega). Retroviral 29-mer short hairpin RNA (shRNA) plasmids that targeted the mouse Myoferlin coding sequence (5′-GTTCCATTCAGCCACCATGTTGCAAGATG-3′ for shRNA 1, or 5′-GACAATGACAGTGATGACGTGGAGAGCAA-3′ for shRNA 2) and a scrambled control shRNA plasmid were established using the HuSH shRNA cloning plasmid pRS (OriGene). These retroviral plasmids were introduced by lipofection using FuGENE6 into Platinum-E packaging cells (21). Forty-eight hours after transfection, the culture supernatant was used to infect NIH3T3 cells in the presence of 10 μg/ml polybrene, thereby establishing stable transformants.

Gene-trap 129 embryonic stem cells that targeted the Myoferlin chromosomal gene (clone D135C09) were obtained from Helmholtz Zentrum Munchen, (Munich, Germany) and introduced into C57BL/6 embryos to produce chimeric mice. Chimeric mice with a high ES contribution were crossed with C57BL/6 mice to produce Myoferlin^+/− mice. Myoferlin^−/− mice were generated by crossing Myoferlin^+/− parents, and the phenotypes of the Myoferlin^+/+ and Myoferlin^−/− littermates were analyzed. The genotype of the Myoferlin gene was determined by PCR with the following primers: wild-type (WT)–Fw (5′-AGGCAACACTGCACACTCAA-3′) and WT-Rv (5′-GCCAGACGCTGTGACAGTTA-3′) for the WT allele, and knockout (KO)-Fw (5′-GAGAAGAAGAGGAGGAACCC-3′) and KO-Rv (5′-GTGATTGACTACCCGTCAGC-3′) for the mutant allele. All mice were housed in a specific pathogen-free facility, and all animal experiments were performed according to a protocol approved by Kanazawa University, Kanazawa, Japan.

Total RNA was isolated from cells using a GenElute Mammalian Total RNA Miniprep Kit (Sigma-Aldrich), and cDNA was synthesized with ReverTra Ace (TOYOBO, Osaka, Japan), according to the manufacturer’s instructions. The cDNA products were amplified with a LightCycler 96 (Roche) using SYBR Select Master Mix (Thermo Fisher Scientific). The primers used for real-time PCR were as follows: Myoferlin-Fw (5′-TGCTCATCCTGTTGCTGTTC-3′), Myoferlin-Rv (5′-GTTCTTCATTGCTTGCGTGA-3′); β-actin–Fw (5′-TGTGATGGTGGGAATGGGTCAG-3′), β-actin–Rv (5′-TTTGATGTCACGCACGATTTCC-3′). The data were analyzed using the ΔCt (threshold cycle) method and normalized against levels of β-actin RNA expression in each sample.

The lysosomal enzymes secreted by Ca²⁺ stimulation were quantified as described previously (22). In brief, the NIH3T3 transformants or BMDMs (1 × 10⁵ cells in 24-well plates) were cultured in PBS containing NAADP (100 μM; R&D Systems), ML-SA1 (1 μM; Wako Pure Chemical Industries, Osaka, Japan), or ionomycin (100 nM; Wako Pure Chemical Industries) for the indicated periods. The cultured PBS was incubated for 15 min at 37°C with 4 mM 4-methyl-umbellyferyl-N-acetyl-β-D-glucosaminide (Sigma-Aldrich) in 20 mM sodium citrate phosphate buffer (pH 4.5). A 1:1 mixture of 2 M Na₂CO₃ and 1 M glycine was added to the reaction solution to terminate the reaction. The fluorescence produced by 4-methyl-umbellyferyl-N-acetyl-β-d-glucosaminide when reacted with the β-hexosaminidase (also known as N-acetyl-β-d-glucosaminidase) secreted in the cultured PBS was measured using a Synergy HT multidetection microplate reader (BioTek Instruments) at an excitation wavelength of 360 nm and an emission wavelength of 460 nm. To quantify the lysosomal enzymes secreted in vivo, the peritoneal fluid was collected with 10 ml PBS at 2 d after i.p. injection of 70 μg Escherichia coli. After centrifugation at 1500 rpm for 5 min, the activity of β-hexosaminidase was measured in the supernatants as described above.

NIH3T3 cells (1 × 10⁵ cells) that expressed GFP-myoferlin were cultured in 3.5-cm glass-bottom dishes, incubated with LysoTracker Red DND-99 (50 nM; Thermo Fisher Scientific), or coexpressed either with RFP-LAMP1 or RFP-CD63 fusion protein. For intracellular staining, NIH3T3 cells (2 × 10⁴ cells) cultured in a Millicell EZ SLIDE eight-well glass slide (Merck Millipore), were fixed in 4% paraformaldehyde solution, and permeabilized with ice-cold acetone. After blocking with PBS containing 10% goat serum (Sigma-Aldrich) and 1% BSA, the cells were stained in PBS containing 1% BSA with FITC anti-LAMP1 Ab (1D4B; BioLegend). After staining, the cells were mounted with coverslips using VECTASHIELD mounting medium (Vector Laboratories) and were observed by a confocal microscopy (FV10i; Olympus).

For intracellular LAMP1 staining, BMDMs were incubated with an unconjugated LAMP1 Ab to inhibit the staining of extracellular LAMP1, then permeabilized with FOXP3 Fix/Perm Buffer Set (BioLegend) and stained with FITC anti-LAMP1 Ab. For the staining of mouse peritoneal exudate cells, the cells were collected 2 d after i.p. injection with 1 mg pHrodo Red E. coli BioParticles (Thermo Fisher Scientific), and then stained with FITC anti-CD11b (M1/70; BioLegend) and Alexa Fluor 647 anti–Ly-6G (1A8; BioLegend) and analyzed using a BD FACSVerse system.

Lysosomes were isolated from NIH3T3 cells by using a Lysosome Enrichment Kit for Tissues and Cultured Cells (Thermo Fisher Scientific), according to the manufacturer’s protocol. Isolated lysosomes or whole cells were lysed in RIPA buffer (0.1% SDS, 0.5% DOC, 150 mM NaCl, 50 mM Tris (pH 7.4), 2 mM EDTA, 1% Triton X-100), which was supplemented with a protease inhibitor mixture (Roche) and/or a phosphatase inhibitor mixture (Sigma-Aldrich). Protein concentrations were determined with a Pierce 660nm Protein Assay Reagent (Thermo Fisher Scientific). Western blot analysis was performed by using Abs against myoferlin (159; Sigma-Aldrich), LAMP1 (1D4B; BioLegend), syntaxin6 (C34B2; Cell Signaling Technology), or β-actin (AC74; Sigma-Aldrich). The Western blot signals were detected with ImageQuant LAS 4000 mini Biomolecular Imager (GE Healthcare), and the intensity was quantified by ImageQuant TL Analysis Toolbox software.

E. coli DH5α were grown at 37°C in lysogeny broth until the midexponential phase (OD.0.8–1.0). The E. coli were collected by centrifugation at 4000 × g for 10 min and inactivated with 50 ml of 70% (v/v) ethanol for 30 min. The E. coli were centrifuged, washed and suspended in PBS, then followed by UV irradiation treatment for 30 min. For the cytotoxicity assay, the peritoneal fluid was collected with 10 ml DMEM containing 1% FBS at 2 d after i.p. injection of the killed E. coli. After centrifugation at 1500 rpm for 5 min, the supernatant was condensed 10 times using an Amicon Pro Purification System and was used as a culture medium for NIH3T3 cells (2 × 10⁴ cells in 24-well plates). After 2 d, the survival of attached NIH3T3 cells was examined using WST-1 Cell Proliferation Reagent (Roche), according to the manufacturer’s protocol, with an iMark Microplate Absorbance Reader (Bio-Rad Laboratories). For CFU fibroblast assay, NIH3T3 cells (1× 10³ cells) were cultured with the mouse peritoneal fluids for 2 d, and the colonies were counted.

To examine whether various phagocytes express myoferlin, we examined the expression profiles of myoferlin by quantitative PCR. Myoferlin was highly expressed in macrophages, such as thioglycollate-elicited peritoneal macrophages, BMDMs and bone marrow-derived dendritic cells (BMDCs), macrophage cell lines (J774.1, RAW 264, and BAM3), neutrophils, and NIH3T3 cells but not in T or B cell lines (WR19L or BaF3 cells) (Fig. 1A). As NIH3T3 cells are the most susceptible to transfection among these cells, we first examined the subcellular localization of myoferlin by expressing GFP-myoferlin fusion protein in NIH3T3 cells and found that myoferlin was colocalized with various lysosomal markers, including LysoTracker, LAMP1, or CD63 (Fig. 1B, Supplemental Fig. 1A). Western blot analysis confirmed that both myoferlin and LAMP1 were enriched in isolated lysosomal fractions, whereas neither syntaxin6 (a Golgi marker) nor β-actin (a cytosol marker) was enriched (Fig. 1C).

To clarify the functions of myoferlin in lysosomes, we established NIH3T3 transformants expressing shRNA against myoferlin (Supplemental Fig. 1B). LAMP1 staining revealed that the myoferlin knockdown cells contained significantly more lysosomes compared with the control cells (Fig. 2A). The increase of lysosomes was confirmed by FACS staining with LysoTracker, which showed a 2.5-fold increase in the myoferlin knockdown cells (Fig. 2B). Electron microscopic analysis of the NIH3T3 transformants showed that the myoferlin knockdown cells carried many more cytoplasmic vesicles filled with debris and additional membranous materials compared with the control cells (control 5.9 ± 1.8 per cell versus myoferlin-knockdown 28.6 ± 5.5 per cell; n = 20; p < 0.02, Student t test) (Fig. 2C), which is consistent with a previous report showing that autolysosomes accumulate in myoferlin-deficient cells (23). However, induction of autophagy was not altered by myoferlin deficiency (Supplemental Fig. 1C).

Based on these results, we considered the possibility that the accumulation of lysosomes might be due to impaired lysosomal exocytosis by myoferlin knockdown. To examine whether myoferlin promotes lysosomal exocytosis, we quantified the amount of β-hexosaminidase, which is one of the most common lysosomal enzymes, secreted by calcium stimulation. When NIH3T3 transformants were treated with NAADP, which induces calcium release from endolysosomes through two-pore channels (24), β-hexosaminidase was rapidly secreted, but the secretion was significantly impaired by myoferlin knockdown (Fig. 3A). The secretion of β-hexosaminidase from myoferlin knockdown cells was restored by expressing human myoferlin protein but not by its deletion mutant that lacks the C2A domain, suggesting that the binding of the C2A domain to PS is essential for the myoferlin activities (Fig. 3B). We then generated Myoferlin^−/− mice to examine the lysosomal exocytosis by phagocytes. These mice were viable and exhibited normal development. BMDMs were prepared from Myoferlin^+/+ and Myoferlin^−/− mice, and we confirmed the accumulation of lysosomes in Myoferlin^−/− BMDMs by FACS staining of intracellular LAMP1 (Fig. 3C). Next, we compared the secretion of β-hexosaminidase from BMDMs upon various calcium stimuli. As shown in Fig. 3D, Myoferlin^−/− BMDMs secreted a significantly lower amount of β-hexosaminidase than Myoferlin^+/+ BMDMs in response to various calcium stimuli such as NAADP, ionomycin (a calcium ionophore), and ML-SA1 (a mucolipin1 channel agonist), although the increase of cytoplasmic Ca²⁺ levels were comparable (Fig. 3E). The occurrence of lysosomal exocytosis can also be quantified by expression levels of cell surface LAMP1. After ionomycin treatment, the intensity of cell surface LAMP1 levels was increased on Myoferlin^+/+ BMDMs, but it was much lower on Myoferlin^−/− BMDMs (Fig. 3F), which was consistent with the secretion of β-hexosaminidase. These results indicate that myoferlin plays a crucial role in lysosomal exocytosis by phagocytes upon calcium stimulation. To investigate whether myoferlin also plays a role in phagocytosis, we cocultured Myoferlin^+/+ or Myoferlin^−/− BMDMs with FITC-labeled small or large beads, but we did not find any effect of myoferlin deficiency on the uptake of these beads (Supplemental Fig. 2A). During phagocytosis, myoferlin was gradually recruited to phagosomes (Supplemental Fig. 2B), but the phagosome–lysosome (P-L) fusion was not impaired in Myoferlin^−/− BMDMs when acidification of phagosomes was examined with engulfed pHrodo Red E. coli BioParticles (Supplemental Fig. 2C). This is probably because only a small increase of cytoplasmic Ca²⁺ levels occurred during phagocytosis compared with the treatment with calcium mobilizers (data not shown), which might be insufficient to activate the function of myoferlin. Accordingly, the ability of killing- engulfed bacteria was not impaired in Myoferlin^−/− BMDMs (Supplemental Fig. 2D). These data indicate that myoferlin has a minimal effect on P-L fusion, suggesting that lysosomal exocytosis is a primary role for myoferlin.

We next examined whether myoferlin promotes secretion of lysosomal enzymes in vivo. The peritoneal injection of pHrodo Red E. coli BioParticles induced the recruitment of CD11b⁺/Ly-6G⁺ polymorphonuclear cells [i.e., neutrophils to elicit phagocytosis of the BioParticles (Fig. 4A, 4B)]. There were no differences in the number of recruited cells in Myoferlin^+/+ and Myoferlin^−/− mice. We then compared the activity of lysosomal enzymes in the peritoneal fluid that was secreted by the neutrophils after the bacterial infection. Two days after peritoneal injection of the E. coli BioParticles, the increase of the lysosomal enzyme activity was much greater in the peritoneal fluid of Myoferlin^+/+ mice than Myoferlin^−/− mice (Fig. 4C). We then investigated whether the activity of lysosomal enzymes correlated with the cytotoxicity of the peritoneal fluid. Again, we injected large numbers of killed E. coli into the Myoferlin^+/+ mouse peritoneal cavity to promote the recruitment of neutrophils to elicit phagocytosis. Two days later, the peritoneal fluid was collected by injecting DMEM containing 1% FBS and was concentrated 10 times by diafiltration. When NIH3T3 cells were cultured with the medium for 3 d, 80% of the cells were killed by the medium (Fig. 4D). This cytotoxic effect was also observed using the medium containing peritoneal fluid obtained without E. coli injection, but when the media were serially diluted, the peritoneal fluid obtained after E. coli injection had greater cytotoxicity than that without E. coli injection. This result is consistent with the activity of lysosomal enzymes with or without E. coli injection (Fig. 4C). The cytotoxic effect was significantly impaired, and viable cells formed many more colonies when cultured with the peritoneal fluid obtained after E. coli injection from Myoferlin^−/− mice compared with that from Myoferlin^+/+ mice (Fig. 4E, 4F). Overall, these results demonstrate that myoferlin plays a crucial role in the release of cytotoxic lysosomal contents from phagocytes, which may increase the likelihood of cellular injuries in bacterial infections.

In this study, we found that myoferlin is a regulator of lysosomal exocytosis by phagocytes. Myoferlin belongs to the ferlin family, the members of which are defined by their sequence similarity to Caenorhabditis elegans fer-1 (13). Fer-1 is localized to the membrane of specialized vesicles called membranous organelles (MOs) in spermatids, where it regulates the Ca²⁺-dependent fusion of MOs with the plasma membrane during spermiogenesis (25). In fer-1 mutants, MOs do not fuse with the plasma membrane, thereby causing the abnormal accumulation of MOs and infertility. It is known that mammals possess at least six members of the ferlin family of proteins, including dysferlin, otoferlin, and myoferlin, which share similar domain structures, with a single transmembrane domain in the C terminus and multiple C2 domains in the N terminus (13). Dysferlin is the most thoroughly studied ferlin, and its mutations cause limb-girdle muscular dystrophy 2B and Miyoshi myopathy (26). Dysferlin localizes to the cytoplasmic vesicles and plays a critical role in muscle membrane repair by mediating the fusion of vesicles with the plasma membrane, thereby serving as a membrane patch in the disrupted regions (26). Otoferlin is highly expressed in the cochlea of the ear and has been implicated in synaptic vesicle exocytosis, in which it acts as the major Ca²⁺ sensor to trigger the fusion of the synaptic vesicles and plasma membrane (27). Thus, the common function of ferlin family proteins is to mediate the fusion of cytoplasmic vesicles with the plasma membrane in exocytosis. Myoferlin was identified as the protein that shares the highest homology with dysferlin (12). Despite their similarity and their common expression in skeletal muscle, myoferlin and dysferlin are thought to share few overlapping functions. Previously, dysferlin was shown to regulate monocyte adhesion and phagocytosis (28, 29). However, we could detect dysferlin expression only in NIH3T3 cells but not in any macrophages or neutrophils, and lysosomal exocytosis was intact in dysferlin-deficient BMDMs (data not shown). Given the low expression levels of other ferlin family proteins in phagocytes, myoferlin is likely the only ferlin protein that regulates lysosomal exocytosis by phagocytes.

Myoferlin regulates lysosomal exocytosis in a Ca²⁺-dependent manner. In this study, we used a calcium mobilizer to increase cytoplasmic Ca²⁺ levels and thus to activate the function of myoferlin. In physiological settings, various immune or inflammatory responses elevate Ca²⁺ levels in phagocytes (30). For example, it is well known that signals from cell surface receptors such as Fc receptors activate the phospholipase C γ and inositol 1,4,5-triphosphate signaling pathways, thereby causing the release of Ca²⁺ from the endoplasmic reticulum (31). Such Ca²⁺ elevation may trigger myoferlin-mediated lysosomal exocytosis in response to bacterial infection. In contrast, myoferlin did not appear to control phagocytosis or phagosome maturation although it was gradually recruited to phagosomes. Because we detected only a small increase of cytoplasmic Ca²⁺ levels during phagocytosis and P-L fusion occurs calcium independently (32), myoferlin, a Ca²⁺-dependent protein, plays only a minimal role in these processes. However, as we observed the accumulation of autolysosomes filled with debris in the myoferlin knockdown cells, it is speculated that myoferlin on phagolysosomes might promote expelling of indigestible debris generated inside phagolysosomes by exocytosis (33, 34). It has also been shown that phagocytes have a potent cytotoxic capacity that is induced by mAbs against target cells (e.g., tumor cells and pathogens). This is known as Ab-dependent cell-mediated cytotoxicity (35). The mAbs activate Fc receptors to trigger the signaling pathway that induces Ca²⁺-dependent lysosomal exocytosis. Neutrophils have a particularly strong Ab-dependent cell-mediated cytotoxic activity, during which they secrete large amounts of cytotoxic molecules, including hydrolytic enzymes, oxidative metabolites, and host defense peptides such as defensins (35). It would be useful to study whether myoferlin regulates the release of these molecules from neutrophils. In any case, our demonstration that myoferlin deficiency decreased the cytotoxicity by phagocytes is clinically important and may facilitate the development of novel therapeutic approaches based on myoferlin-mediated lysosomal exocytosis.

We thank H. Ohgi and J. Song for technical assistance in the initial stage of this project and Y. Okayasu for secretarial assistance. We also thank H. Omori for electron microscopy analysis and Y. Esaki and S. Nishioka for generating the myoferlin-deficient mice.

The authors have no financial conflicts of interest.