Megalencephalic leukoencephalopathy with subcortical cysts protein-1 modulates endosomal pH and protein trafficking in astrocytes: relevance to MLC disease pathogenesis.

Megalencephalic leukoencephalopathy with subcortical cysts (MLC) is a rare leukodystrophy caused by mutations in the gene encoding MLC1, a membrane protein mainly expressed in astrocytes in the central nervous system. Although MLC1 function is unknown, evidence is emerging that it may regulate ion fluxes. Using biochemical and proteomic approaches to identify MLC1 interactors and elucidate MLC1 function we found that MLC1 interacts with the vacuolar ATPase (V-ATPase), the proton pump that regulates endosomal acidity. Because we previously showed that in intracellular organelles MLC1 directly binds Na, K-ATPase, which controls endosomal pH, we studied MLC1 endosomal localization and trafficking and MLC1 effects on endosomal acidity and function using human astrocytoma cells overexpressing wild-type (WT) MLC1 or MLC1 carrying pathological mutations. We found that WT MLC1 is abundantly expressed in early (EEA1(+), Rab5(+)) and recycling (Rab11(+)) endosomes and uses the latter compartment to traffic to the plasma membrane during hyposmotic stress. We also showed that WT MLC1 limits early endosomal acidification and influences protein trafficking in astrocytoma cells by stimulating protein recycling, as revealed by FITC-dextran measurement of endosomal pH and transferrin protein recycling assay, respectively. WT MLC1 also favors recycling to the plasma-membrane of the TRPV4 cation channel which cooperates with MLC1 to activate calcium influx in astrocytes during hyposmotic stress. Although MLC disease-causing mutations differentially affect MLC1 localization and trafficking, all the mutated proteins fail to influence endosomal pH and protein recycling. This study demonstrates that MLC1 modulates endosomal pH and protein trafficking suggesting that alteration of these processes contributes to MLC pathogenesis.

Introduction

Megalencephalic leukoencephalopathy with subcortical cysts (MLC, OMIM 604004) is a rare congenital and incurable leukodystrophy characterised by early-onset macrocephaly, ataxia, seizures, degeneration of motor functions and mild cognitive decline. Magnetic resonance imaging (MRI) indicates diffuse signal abnormality, swollen appearance of the white matter and the presence of subcortical cysts (Singhal et al., 2003; Topku et al., 1998; Van der Knaap et al., 1995a,b). MLC patients have a slowly progressive clinical course that can be worsened by minor head trauma and fever (Ben-Zeev et al., 2001; Bugiani et al., 2003; Riel-Romero et al., 2005). Histopathological analysis of MLC patient brains revealed intramyelinic vacuole formation, alterations of the blood–brain barrier structure and astroglial activation (Van der Knaap et al., 1995a,b, 1996). Enlarged intracellular vacuoles localized in the astrocyte end-feet contacting blood vessels have also been described (Duarri et al., 2011).

Almost 75% of MLC patients carry different types of mutations in the MLC1 gene suggesting that functional alterations of the MLC1 gene product are the leading cause of this disease. However, to date no correlation between genotype and phenotype has been found (Leegwater et al., 2001, 2002; Patrono et al., 2003). Recently, mutations in the HEPACAM/GLIALCAM gene encoding an adhesion-like molecule of unknown function have been found in a substantial fraction of MLC affected patients without MLC1 mutations, unveiling genetic heterogeneity of MLC disease (Boor et al., 2006; Jeworutzki et al., 2012; López-Hernández et al., 2011).

The MLC1 gene encodes an oligomeric and highly hydrophobic protein which shows low homology with some ion channels and transporters (Boor et al., 2005; Leegwater et al., 2001; Teijido et al., 2004). In the central nervous system (CNS), MLC1 is mainly expressed in perivascular and subpial astrocytes, particularly in astrocytic end-feet contacting blood vessels and meninges (glia limitans) and in astrocytic intracellular organelles (Ambrosini et al., 2008; Boor et al., 2007; Duarri et al., 2011; Teijido et al., 2004). Bergmann glia and ependymal cells lining the ventricles also express MLC1 (Ambrosini et al., 2008; Boor et al., 2007; Duarri et al., 2008; Teijido et al., 2004). Outside the CNS, MLC1 has been detected in monocytes and lymphocytes (Boor et al., 2005; Duarri et al., 2008). Although myelin vacuolation is a typical feature of MLC disease, the myelin forming cells, oligodendrocytes, do not express MLC1 (Boor et al., 2005; Schmitt et al., 2003), suggesting that myelin degeneration may be secondary to astrocyte dysfunction. Indeed, the tissue distribution and structural features of MLC1 protein and MLC-associated brain damage suggest a possible role for MLC1 in the regulation of fluid and/or ion homeostasis, a function that in the CNS is mainly carried out by astrocytes (Parpura and Verkhratsky, 2012). Consistent with this hypothesis, we have shown recently that MLC1 is part of a macromolecular complex associated to the sodium, potassium-ATPase pump (Na, K-ATPase) which includes the inward rectifier potassium channel 4.1 (Kir4.1), the water channel aquaporin-4 (AQP4), the transient receptor potential cation channel subfamily V, member 4 (TRPV4), the cytoskeletal anchoring protein syntrophin and the membrane raft-associated protein caveolin-1 (Brignone et al., 2011; Lanciotti et al., 2012). We have also provided evidence that MLC1 is involved in the astrocytic response to changes in the extracellular ion composition and cooperates with TRPV4 to activate intracellular calcium influx during hyposmotic stress (Lanciotti et al., 2012). Most importantly, we have found that these interactions and pathways are affected by MLC1 pathological mutations (Lanciotti et al., 2012). The TRPV4-mediated calcium influx is the first and essential step required for the activation of astrocyte regulatory volume decrease (RVD) which is needed to rescue the rapid and temporary cell swelling induced by hyposmosis (Benfenati et al., 2007, 2011). Interestingly, defects in a RVD-induced chloride current have been noted in rat astrocytes following siRNA-mediated MLC1 downregulation and in MLC patient-derived lymphoblasts (Ridder et al., 2011). Altogether, these results support the hypothesis that MLC1 is an ion channel involved in the astrocyte response to osmotic imbalance and regulation of cell volume.

In initial experiments aimed at identifying MLC1 protein interactors and MLC1-associated functional pathways, we found that MLC1 interacts with the vacuolar ATPase (V-ATPase), the proton pump responsible for endosomal acidification (Forgac, 2007). This finding and the observation that in rat primary astrocytes and human astrocytoma cells MLC1 binds the Na, K-ATPase which, among other functions, controls early endosomal pH (Brignone et al., 2011; Lanciotti et al., 2012), prompted us to investigate the possible role of MLC1 in organelle acidification and protein trafficking and the effects of MLC disease-associated pathological mutations on these pathways.

Materials and methods

Cell cultures and treatments

Astrocyte-enriched cultures (about 95% purity) were generated from 1- or 2-day-old newborn rats and maintained in culture as previously described (Agresti et al., 1991). By using a retroviral bicistronic vector (pQCXIN, Takara Bio Europe Clontech, France) and the packaging cell line (GP2, Hek293) retroviral particles carrying recombinant WT or mutated MLC1 (S246R, S280L, C125R) were generated as previously described (Lanciotti et al., 2012). By infection with recombinant retroviral particles, astrocytoma cell lines overexpressing WT and mutated MLC1 and a control cell line infected with the empty virus were generated, as previously described (Lanciotti et al., 2012). Cells were exposed to hyposmotic solution, as described previously (Brignone et al., 2011). Cells stimulated with 100 nM bafilomycin A1 (Sigma-Aldrich, St. Louis, MO) in serum-free (SF) medium for 3, 6 and 48 hours (h) were used for immunofluorescence stainings and western blot analysis as described below. The procedures for human monocyte isolation and culture are described in the Supplementary material.

Immunofluorescence and confocal microscopy analysis

Astrocytoma cells grown on polylysine-coated coverslips were incubated in isosmotic, hyposmotic or bafilomycin-containing solution for different time lengths, fixed for 10 minutes (min) with 4% paraformaldehyde and washed with PBS. After 1 h of incubation with blocking solution (5% BSA in PBS), cells were incubated for 1 h at room temperature (RT) with the following primary antibodies (Abs) diluted in PBS containing 0.025% Triton X100: affinity purified anti-MLC1 polyclonal Ab (pAb) (1:50, Atlas AB, AlbaNova University Center, Stockholm, Sweden), anti-V-ATPase B2 monoclonal Ab (mAb) (D11; 1:50, Santa Cruz Biotechnology, Inc., Santa Cruz, CA), anti-EEA1 mAb (1:50, BD Transduction Laboratories, Lexington, KY), anti-Rab5 mAb (D11; 1:25, Santa Cruz Biotechnology), anti-Rab11 mAb (clone47; 1:25, Millipore, Temecula, CA), anti-Lamp-2 mAb (1:100, Abcam, Cambridge, UK), anti-transferrin receptor (TfR) mAb (1:80, Abcam) and anti-TRPV4 (H-79) pAb (1:10, Santa Cruz Biotechnology). As secondary Abs, a biotinylated goat anti-rabbit IgG H + L Ab (4.3 μg/ml; Jackson Immunoresearch Laboratories, West Grove, PA) followed by streptavidin-TRITC (2 μg/ml; Jackson), and Alexa Fluor 488 goat anti-mouse IgG Ab (1:300, Invitrogen, Milan, Italy) were used. Coverslips were washed, sealed in Vectashield medium (Vector Lab, Burlingame, CA) and analyzed with a laser scanning confocal microscope (LSM 5 PASCAL, Carl Zeiss, Jena, Germany). In the colocalization experiments fluorescence intensity profiles or fluorochrome colocalization analysis based on Manders' overlap coefficient (MOC, Manders et al., 1993) was evaluated with the profile analysis tool of the LSM 5 PASCAL or NIH ImageJ software. MOC value range is 0–1 (0: no colocalization, 1: all pixels colocalize).

Immunostaining of human brain tissue

Post-mortem brain tissue from a person without neurological disease was obtained from the UK MS Tissue Bank at Imperial College London, and stained as previously described (Brignone et al., 2011). Briefly, sections were incubated overnight (ON) at 4 °C with a mixture of rabbit anti-MLC1 pAb (1:250, ATLAS) and anti-V-ATPase B2 mAb (D11) (1:100, Santa Cruz Biotechnology) or anti-GFAP mAb (1:20, Pharmingen BD Biosciences, Milan, Italy). After extensive washings, sections were incubated for 1 h at RT with a mixture of fluorescein-conjugated goat anti-mouse and rhodamine-conjugated goat anti-rabbit Abs and images were analyzed with laser scanning confocal microscope (LSM 5 PASCAL).

Protein extract preparation and western blotting

Cytosolic and membrane fractions from cultured astrocytes and total protein extracts from astrocytoma cell lines were obtained as previously described (Lanciotti et al., 2010, 2012). Equal amounts of proteins (30 μg) were resolved on SDS-PAGE using gradient (4–12%) precasted gels (Invitrogen), and transferred onto a nitrocellulose membrane. Nitrocellulose membranes were blotted ON at 4 °C using anti-MLC1 pAb (1:1500, in-house generated; Ambrosini et al., 2008), anti-V-ATPase A1 mAb (H-140;1:1000, Santa Cruz Biotechnology), or anti-TRPV4 pAb (1:200, Alomone Labs, Israel) in PBS/3% BSA followed by extensive washings and then incubated for 1 h with horseradish peroxidase-conjugated anti-mouse or anti-rabbit Abs (1:10000; Thermo Scientific, MO), for 1 h at RT. Immunoreactive bands were visualized using an enhanced chemiluminescence reagent (Thermo Scientific), according to the manufacturer's instructions, and exposed on X-ray films. Densitometric analyses of WB bands were performed using ImageJ software.

Pull-down and LC–MS/MS analysis for the identification of MLC1 protein interactors

Cytosolic and membrane protein extracts obtained from cultured rat astrocytes were subjected to pull-down assay using recombinant His-tagged MLC1 bound to Ni-NTA resin and control empty resin, as previously described (Ambrosini et al., 2008; Brignone et al., 2011; Lanciotti et al., 2010, 2012). Proteins pulled-down with recombinant MLC1 and control unspecific proteins eluted from control resin were separated by SDS-PAGE, stained with Coomassie Colloidal Blue (Invitrogen) for proteomic analysis and analyzed by RP-LC–MS/MS. Protein bands present in eluates from MLC1 resin but not in control empty resin were excised and destained with 50 mM ammonium bicarbonate/acetonitrile (1:1 v/v) solution (Merck Darmstadt, Germany) and then dried by acetonitrile treatment. Protein cysteine residues were reduced by 10 mM dithiothreitol (Sigma-Aldrich Canada Markham, ON, Canada) in 50 mM ammonium bicarbonate (ICN Biomedicals Aurora, OH) for 45 min at 56 °C, then alkylated by 55 mM iodoacetamide (Sigma-Aldrich) for 30 min in the dark at RT in 50 mM ammonium bicarbonate. Finally, samples were dried by centrifugation under vacuum and in-gel digestion was performed by incubating gel particles with 12.5 ng/ml trypsin (sequencing grade modified porcine trypsin, Promega, Madison, WI) in 25 mM ammonium bicarbonate at 37 °C ON under stirring. Peptide mixtures were analyzed by nanoflow reversed-phase liquid chromatography tandem mass spectrometry (RP-LC–MS/MS) using an HPLC Ultimate 3000 (DIONEX, Sunnyvale, CA) connected on line with a linear Ion Trap (LTQ, Thermo Electron, San Jose, CA). Peptides have been desalted in a trap column (Acclaim PepMap 100 C18, LC Packings, DIONEX) and then separated in a reverse phase column, a 10-cm long fused silica capillary (Silica Tips FS 360-75-8, New Objective, Woburn, MA), slurry-packed in-house with 5 μm, 200 Å pore size C18 resin (Michrom BioResources, CA). Peptides were eluted using a linear gradient from 96% A (H2O with 5% acetonitrile and 0.1% formic acid) to 60% B (ACN with 5% H2O and 0.1% formic acid) for 40 min, at 300 nl/min flow rate. Analyses were performed in positive ion mode and the HV Potential was set up around 1.7–1.8 kV. Full MS spectra ranging from m/z 400 to 2000 Da were acquired in the LTQ mass spectrometer operating in a data-dependent mode in which each full MS scan was followed by five MS/MS scans where the five most abundant molecular ions were dynamically selected and fragmented by collision-induced dissociation (CID) using a normalized collision energy of 35%. Target ions already fragmented were dynamically excluded for 30 s. Tandem mass spectra were matched against a database containing rat proteins downloaded from Swiss Prot (http://www.uniprot.org/downloads) and through SEQUEST algorithm incorporated in Bioworks software (version 3.3, Thermo Electron). The parameters for the database search were: fully tryptic cleavage constraints, with the possibility to have one miss cleavage permitted, static carbamidomethylation on cysteine residues and methionine oxidation as variable modification. Data were searched with 1.5 Da and 1 Da tolerance respectively for precursor and fragment ions. A peptide has been considered legitimately identified when it achieved cross correlation scores of 1.8 for [M + H]1 +, 2.5 for [M + 2H]2 +, 3 for [M + 3H]3 +, and a peptide probability cut-off for randomized identification of P < 0.001.

Co-purification of histidine-tagged MLC1 interactors

Lysates obtained from three 175 cm2 flasks of confluent astrocytoma cell lines stably overexpressing histidine-tagged WT or mutated MLC1 (U251-HisMLC1) and control cells infected with empty viral vectors were incubated ON at 4 °C with 100 μl (50% v/v suspension) of Ni-NTA Agarose (Qiagen, Hilden, Germany). After extensive washings (10 bed volumes of 10 mM–25 mM imidazole, 0.2% Triton X-100, 0.15 M NaCl, 20 mM Tris–HCl pH 7.4), protein elution was carried out using imidazole at a concentration of 50 and 200 mM (Lanciotti et al., 2010). The eluted proteins were analyzed by SDS-PAGE and WB.

Cell surface protein biotinylation and isolation

Four 75 cm2 flasks of 90–95% confluent U251 astrocytoma cell lines overexpressing WT or S280L MLC1, untreated or treated for 30 min with hyposmotic solution as previously described (Brignone et al., 2011), were incubated with Sulf-NHS-SS-Biotin (0.25 mg/ml, Thermo Scientific, Rockford, IL) for 30 min at 4 °C. The biotinylation reaction was terminated by the addition of 500 μl of Quenching solution (Thermo Scientific) per flask. Following biotinylation, the cells were harvested and solubilized in Lysis Buffer (Thermo Scientific). The cells were further disrupted by brief sonication and centrifuged for 2 min at 10,000 ×g. Solubilized biotinylated membrane proteins were incubated for 60 min at room temperature with NeutrAvidin® Agarose (Thermo Scientific) in column, as indicated by the manufacturer. The NeutrAvidin® Agarose bound biotinylated proteins were eluted from the column with 450 μl SDS-PAGE Sample Buffer (62.5 mM Tris–HCl, pH6.8, 1% SDS, 10% glycerol, 50 mM DTT). Fractions containing eluted protein were analyzed by Western Blot, as described above.

Endosomal pH measurement by FITC-dextran

Endosomal pH was measured by means of the video-imaging technique with the pH sensitive probe FITC-dextran. FITC-dextran (Sigma-Aldrich) loading was achieved by exposing astrocytoma cells grown on glass coverslips for 30 min to a loading solution with the following composition (mM): 122 NaCl, 3.3 KCl, 1.2 KH2PO4, 1.3 CaCl2, 1 MgCl2, 10 d-glucose, 25 NaHCO3, 0.07 FITC-dextran (70 KD) (37 °C, CO2 5%, pH 7.4). The short incubation period in a bicarbonate buffered solution allowed the endocytosis of FITC-dextran and the formation of vesicles in which the control of physiological luminal pH was made possible. After a strong wash out to remove unspecifically bound probe, the glass coverslip was placed in a stainless steel chamber (Attofluor cell chamber, Molecular Probes, Invitrogen) positioned on the stage of an inverted microscope (Axiovert 35, Zeiss, Germany), and bathed in an extracellular solution similar to the loading solution with HEPES replacing NaHCO3. During recording, FITC-dextran loaded cells were exposed every 2 s to the excitation wavelengths of 480 and 440 nm by means of a monochromator (Polychrome II, Photonics, Germany), and the emission light at 510 nm was collected by a 40 × (1.35 NA) UApo/340 oil immersion objective (Olympus, Japan) and recorded by a CCD camera (PCO, Sensicam, Germany). This allowed the recording of 1280 × 1024 pixels/frame with spatial resolution of 170 nm/pixel. A local perfusion system (Rapid Solution Changer 100; Biologic) allowing the rapid switch between different solutions was used. After a 60 s period of recording in the extracellular solution, the cells were bathed with a KCl-based calibration solution containing the ionophore nigericine (10 μM) at pH 7.5 and 6.5, each for 60 s. The calibration solution had the following composition (mM): 120 KCl, 10 NaCl, 1 CaCl2, 1 MgCl2, 10 glucose and 10 HEPES buffer. The ratios of the fluorescence intensity values at 480 nm and 440 nm were converted to pH values by linear regression fitting to the line obtained with the calibration pH values 7.5 and 6.5. The pH value of the single FITC-dextran-containing endosomal vesicles was calculated from the average pH recorded in the extracellular solution. Recording and analysis of the data were made possible by the use of the Imaging Workbench software package (INDEC Systems, CA). For further data processing and presentation the Origin 7.5 software package (Microcal software, USA) was utilized. Data are shown as mean ± SEM values of pH values, and statistical significance was calculated using the Student's t-test.

Transferrin (Tf) internalization and recycling assay

Transferrin internalization and recycling assay was performed as previously described (Bacac et al., 2011), with minor modifications. Briefly, subconfluent U251 astrocytoma cells grown on polylysine-coated glass coverslips were washed twice with PBS and then starved for 2 h in internalization medium (IM) (DMEM, 0.01% BSA) to deplete endogenous transferrin. Where indicated, hyposmotic cell treatment was performed as previously described (Brignone et al., 2011). Then, cells were cooled on ice for 5 min to block endocytosis, rinsed using ice-cold PBS, and incubated with ice-cold IM containing human transferrin conjugated to Alexa Fluor 488 (50 μg/ml, Molecular Probes, Invitrogen) for 1 h at 4 °C to allow cell-surface binding of transferrin. Cells were then washed on ice using cold PBS–BSA 0.5% to remove unbound transferrin, and the probe was chased at 37 °C for indicated time-periods in the presence of 100 × excess unlabeled iron saturated human holo-transferrin (Sigma-Aldrich) to prevent fluorescent transferrin re-internalization. At the end of each time-point, cells grown on coverslips were fixed on ice using 4% PFA and analyzed using a laser scanning confocal microscope (LSM 5 PASCAL).

Results

MLC1 interacts with proteins involved in endosomal pH regulation

A proteomic MS/MS analysis of MLC1 molecular interactors was performed on cytosolic and membrane protein fractions from rat primary astrocytes that were pulled down with recombinant His-tagged-MLC1 protein bound to NiNTA resin selectively linking histidine residues (see Materials and methods). Among the potential MLC1 interactors identified (manuscript in preparation), we found that 2 subunits (a1 and B2) of vacuolar ATPase (V-ATPase), the proton pump responsible for endosomal acidification (Forgac, 2007), were pulled-down by His-MLC1 in 2 out of 3 experiments (Table 1). In the same experiments we also detected the alpha and beta subunits of the Na, K-ATPase pump (Table 1), thus confirming the results obtained in rat astrocytes and human astrocytoma cells using the yeast two-hybrid system and biochemical assays (Brignone et al., 2011; Lanciotti et al., 2012). Because Na, K-ATPase, besides controlling plasma-membrane potential and cell volume, is also involved in the regulation of endosomal pH (Fuchs et al., 1989), these findings suggested that MLC1 might play a role in this process.

Table 1

MLC1 protein interactors identified by mass spectrometry in the membrane protein extracts derived from rat astrocytes.

Protein name	Short name	Probability	Seq Cov %	MW (kDa)	Accession n	Peptides n
V-type proton ATPase 116 kDa subunit a isoform 1	VPP1	1.64E − 06	4	96	P25286	2
V-type proton ATPase subunit B, brain isoform	VATB2	1.21E − 06	11	57	P62815	3
Sodium/potassium-transporting ATPase subunit alpha	Atp1a1	2.10E − 11	19	113	P06685	6
Sodium/potassium-transporting ATPase subunit beta	AT1B3	2.48E − 09	17	32	Q63377	3

Probability corresponds to the Sequest probability score defined as the probability that the peptide assignment is a random match to the spectral data: the value is reported as − 10 ∗ log (probability). % Seq Cov indicates the percentage of the protein sequence covered by matching peptides based on the number of matching amino acids over the total number of amino acids in the protein. MW is for Molecular Weight and is expressed in kDa. In the last column the number of peptides mapped for each identified protein is reported.

Using a mAb recognizing the a1 subunit of V-ATPase, the interaction between MLC1 and V-ATPase was confirmed by WB analysis of proteins pulled-down from rat primary astrocytes, particularly in the membrane protein fraction (Fig. 1a). V-ATPase also co-purified with His-tagged WT MLC1 over-expressed in a human astrocytoma cell line (Fig. 1b). In astrocytoma cell lines overexpressing 3 MLC1 missense mutations (S246R, S280L and C125R) (Lanciotti et al., 2012), the amount of V-ATPase associated with MLC1 was markedly lower than in WT MLC1 astrocytoma cells (Figs. 1b, c), indicating that each of these pathological mutations affects MLC1 molecular interactions. Double immunostaining for MLC1 and V-ATPase revealed that in WT MLC1 astrocytoma cells MLC1 and V-ATPase colocalized in perinuclear vesicles (Fig. 1d, arrows; colocalization analysis, as described in the Materials and methods section, is shown in Fig. S1). In post-mortem human brain tissue V-ATPase immunoreactivity was detected in cytoplasmic vesicles of GFAP+ astrocytes (Fig. 1e, arrows in magnification) and in astrocytic end-feet surrounding blood vessels where it colocalized with MLC1 (Fig. 1f, arrows in magnification). Altogether, these findings suggest that in astrocytes MLC1 might interact with V-ATPase in intracellular organelles.

Fig. 1

Interaction between WT MLC1 and V-ATPase.

a. Pull-down assay shows the interaction between MLC1 and V-ATPase (a1 subunit) in rat primary astrocytes. Cytosolic and membrane fractions from primary cultures of rat astrocytes (input) were pulled-down by His-MLC1 and His-empty vector (used as control) bound to NiNTA-agarose and eluted with 0.1 M glycine, pH 3. Western blot (WB) analysis shows that V-ATPase interacts with His-MLC1 in both the cytoplasmic (cyt) and membrane (mem) fractions. A very low level of unspecific binding is observed only in the membrane fraction of the NiNTA-agarose bound with His empty vector. Molecular weight markers are indicated on the left (kDa). b. His co-purification assay was performed to verify the association between MLC1 and V-ATPase in human astrocytoma cells (U251) overexpressing His-tagged WT and mutated MLC1. V-ATPase is co-eluted and enriched with 200 mM imidazole from NiNTA-agarose bound with proteins of WT MLC1 astrocytoma (WT). A decrease in the binding is observed when MLC1 S246R, C125R and S280L NiNTA-agarose samples have been used, while no interaction is detected using samples derived from astrocytoma cells infected with empty vector (used as control). One representative experiment out of 3 performed is shown. Molecular weight markers are indicated on the left (kDa). c. Densitometric analysis of V-ATPase protein bands revealed by WB in (b). d. Double immunofluorescence staining of astrocytoma cells overexpressing WT MLC1 with anti-MLC1 pAb (red) and anti-V-ATPase mAb (green) reveals colocalization of MLC1 and V-ATPase in intracellular vesicles, mainly in perinuclear areas (arrows). e. Immunofluorescence staining of normal human brain tissue with GFAP mAb (green) and anti-V-ATPase pAb (red) shows V-ATPase expression in intracytoplasmic vesicles of GFAP+ astrocytes (arrows in the microscopic field shown at high power magnification on the right). f. Immunofluorescence staining of normal human brain tissue with anti-MLC1 pAb (red) and anti-V-ATPase mAb (green) reveals that MLC1 and V-ATPase colocalize in perivascular astrocyte end-feet (arrows in the microscopic field shown at high power magnification on the right). Scale bars: 10 μm.

MLC1 localization in specific endosomal compartments and effects of pathological mutations

We (Lanciotti et al., 2012) and others (Duarri et al., 2008) have previously shown that the 3 MLC1 missense mutations analyzed here differently affect MLC1 intracellular localization. Similarly to WT MLC1, the S246R mutated protein reaches the plasma membrane, while the S280L and C125R mutations cause intracellular retention of MLC1 protein (Lanciotti et al., 2012) (see also Figs. 1a–d). To understand whether pathological mutations could affect MLC1 localization and molecular interactions in the cytoplasmic vesicular compartment, astrocytoma cell lines overexpressing WT or mutated MLC1 were immunostained with anti-MLC1 Ab in combination with Abs specific for endosomal markers. Double immunostainings for MLC1 and EEA1 or Rab5, two proteins expressed in organelles of the early endosomal compartment (Jovic et al., 2010), revealed that WT MLC1 colocalized with both markers (Figs. 2a and e, respectively). When using Abs to proteins expressed in the recycling (Rab11) and late endosomal-lysosomal (Lamp-2) compartments, we observed that WT MLC1 immunoreactivity largely colocalized with Rab11+ recycling endosomes (Fig. 2i) and to a much lesser extent with Lamp-2+ organelles (Fig. 2m). While the distribution of S246R mutant protein in intracellular organelles was similar to that of WT MLC1 (Figs. 2b, f, j, n), lower amounts of S280L and C125R mutants were found in EEA1, Rab5+ and Rab11+ organelles when compared to WT MLC1. Conversely, the same mutant proteins were more abundantly localized in Lamp-2+ organelles relative to the WT protein (Manders' colocalization coefficient and statistical analysis is shown in Fig. S2).

Fig. 2

Intracellular localization of WT and mutated MLC1 in human astrocytoma cell lines.

a–d. Double immunofluorescence staining for MLC1 (red) and early endosome antigen (EEA1) (green) shows colocalization of MLC1 and EEA1 in the perinuclear area of MLC1-WT (a) and S246R mutant (b) astrocytoma cells but not of S280L (c) and C125R (d) mutant astrocytoma cells. e–h. Double immunofluorescence for MLC1 (red) and Rab5 (green) reveals protein colocalization in WT (e) and S246R (f) MLC1 astrocytoma cells, but only occasionally in S280L (g) and C125R (h) mutants. i–l. Double immunofluorescence for MLC1 (red) and Rab11 (green) shows that WT (i) and S246R (j) MLC1 proteins are abundantly present in Rab11+ recycling endosomes while colocalization is markedly reduced in S280L (k) and C125R (l) mutants. m–p. Double immunofluorescence for MLC1 (red) and the lysosomal marker Lamp-2 (green) reveals a partial colocalization in WT (m) and mutated MLC1 expressing astrocytoma cells (n–p). Note that Rab5 and Rab11 immunoreactivities have a more diffuse intracellular distribution in WT and S246R MLC1 astrocytoma cells compared to S280L and C125R mutants. Scale bars: 10 μm.

To summarize, this set of experiments allowed us to conclude that intracellular MLC1 is expressed mainly in the early endosomal and Rab11+ recycling compartments and that pathological mutations can affect both MLC1 membrane localization and MLC1 distribution in specific endosomal compartments.

To exclude that MLC1 distribution in early and recycling organelles and the effects of MLC1 mutations observed in astrocytoma cells were due to tumor cell-specific trafficking pathways and/or overexpression of recombinant proteins we analyzed the distribution of endogenous MLC1 in blood monocyte-derived macrophages (Boor et al., 2005; Duarri et al., 2008; Petrini et al., 2013) that were obtained from healthy donors and MLC patients. In macrophages from healthy donors MLC1 immunoreactivity was present in the cell membrane, the endoplasmic reticulum cisternae and intracellular vesicles where it colocalized with Rab11 and EEA1 (Figs. S3a,d). In macrophages from two MLC patients, each carrying a different MLC1 mutation (see Materials and methods), MLC1 immunoreactivity was almost undetectable in the plasma membrane, as previously reported (Petrini et al., 2013), and showed reduced colocalization with cytoplasmic organelle markers, particularly with Rab11 (Figs. S3b,c,e,f).

MLC1 traffics through the Rab11+ perinuclear recycling compartment

The data presented above indicate that MLC1 is abundantly distributed in perinuclear vesicles, mainly in Rab11+ organelles. Since Rab11+ perinuclear organelles are part of the perinuclear (or pericentrosomal) recycling compartment (PNRC) which is important for slow recycling of proteins and storage of membrane proteins, including ion and water channels (Mukherjee et al., 1997; Sheff et al., 1999; Ullrich et al., 1996), we investigated trafficking of MLC1 through this compartment. In WT MLC1 astrocytoma cells MLC1 localization in PNRC was confirmed by the finding that MLC1 immunoreactivity overlapped with that of the transferrin receptor (TfR), which is specifically expressed in Rab11+ organelle (Presley et al., 1997; Ullrich et al., 1996) (Fig. 3, colocalization analysis is shown in Fig. S4).

Fig. 3

WT MLC1 is localized in the perinuclear recycling compartment (PNRC).

Double immunofluorescence staining of WT MLC1 astrocytoma cell lines with anti-MLC1 pAb (red) and anti-transferrin receptor mAb (TfR; green), known to localize in PCRN, shows overlap of MLC1 and TfR immunoreactivities in the perinuclear area (arrows). Scale bar: 10 μm.

To study MLC1 recycling through the PNRC WT MLC1 astrocytoma cells were treated with bafilomycin, a specific inhibitor of V-ATPase which blocks endosomal acidification and leads to inhibition of TfR recycling to the cell membrane through PNRC without affecting TfR endocytosis and transport to the perinuclear compartments (Baravalle et al., 2005; Presley et al., 1993, 1997). Incubation of astrocytoma cells with 100 nM bafilomycin for 3, 6 and 48 h induced a progressive disappearance of WT MLC1 from the plasma membrane and its accumulation in Rab11+ vacuolar structures, as revealed by double immunostaining with anti-MLC1 and anti-Rab11 Abs (Fig. 4a). After 48-h stimulation all the membrane-associated MLC1 were recruited, along with TfR, to Rab11+ cytoplasmic organelles (Fig. 4b). These structures appeared morphologically swelled due to defective acidification, as already described in other bafilomycin-treated cells (Bacac et al., 2011; Baravalle et al., 2005).

Fig. 4

Effect of bafilomycin on WT MLC1 protein localization and expression.

a. Double immunofluorescence stainings for MLC1 (red) and Rab11 (green) were performed in WT MLC1 astrocytoma cells that were grown in basal culture conditions (CTR) or treated with 100 nM bafilomycin (BAF) for 3, 6 and 48 h. Note the progressive disappearance of MLC1 from the plasma membrane and concomitant accumulation of the protein in Rab11+ intracellular vacuolar structures. b. Double immunostaining for MLC1 (red) and transferrin receptor (TfR, green) shows that after 48-h treatment with 100 nM bafilomycin, all MLC1 proteins colocalize with TfR in intracellular vacuolar structures. Scale bars: 10 μm.

When MLC1 mutant expressing astrocytoma cells were analyzed after bafilomycin treatment we found that, similarly to WT MLC1, the plasma membrane-associated S246R protein accumulated in Rab11+ intracellular organelles (data not shown). Conversely, no accumulation of the mutant S280L and C125R proteins, which fail to reach the plasma membrane, was observed in TfR+ or Rab11+ organelles (Figs. S5a,b). Using WB to monitor MLC1 protein levels we found that 48-h treatment with bafilomycin caused accumulation of the membrane-associated dimeric form of WT MLC1 and S246R proteins and a decrease of S280L and C125R mutants (Fig. S6a). Since bafilomycin did not increase MLC1 mRNA in WT MLC1 astrocytoma cells (Fig. S6b), accumulation of WT MLC1 and S246R mutant is likely caused by blockade of protein degradation consequent to bafilomycin-induced arrest of lysosomal function, as previously reported (Baravalle et al., 2005). In contrast, since S280L and C125R mutants are mainly retained in the endoplasmic reticulum (ER), their decrease may be caused by ER associated protein degradation (ERAD) (Duarri et al., 2008), which is unaffected by bafilomycin treatment (Ishida et al., 2009). To summarize, these data indicate that, similarly to some water and ion channels (McEwen et al., 2007; Takata, 2006; Takata et al., 2008; Van de Graaf et al., 2006, 2008), in astrocytoma cells MLC1 traffics between the plasma membrane and Rab11+ organelles of the PNRC.

Hyposmosis stimulates MLC1 recycling to the plasma membrane

MLC1 recycling and the effect of pathological MLC1 mutations on this process were analyzed in hyposmotic conditions that in primary rat astrocytes favor MLC1 transport to the plasma membrane (Brignone et al., 2011; Lanciotti et al., 2012). To this end, MLC1 distribution in the recycling endosomal compartment was analyzed by immunostaining of astrocytoma cells expressing WT or mutated MLC1 with anti-MLC1 and anti-Rab11 Abs in basal conditions and after incubation in hyposmotic solution. A progressive increase in membrane expression of WT MLC1 was observed between 5 and 30 min after exposure to hyposmotic solution, leading to strong labeling of membranes in astrocytic end-feet and thin fillopodia and at astrocyte–astrocyte contacts in the majority of MLC1+ cells (Figs. 5a,b). Hyposmotic treatment also induced intracellular redistribution of Rab11+ vesicles from their typical perinuclear localization throughout the cell body and cytoplasmic extensions where they colocalized with MLC1 (Fig. 5b). This result was confirmed by analysis of the florescence intensity profile of double stained MLC1+/Rab11+ cells (Figs. 5c,d) and by WB quantitation of hyposmosis-induced translocation of WT MLC1 to the plasma membrane using enriched plasma membrane protein fractions obtained after selective biotinylation of cell surface proteins (Fig. 5e). Interestingly, in the same experiments we found that only the dimeric form of MLC1 (60 kDa) is expressed at the plasma membrane after hyposmotic stress while the monomeric MLC1 component (36 kDa) is not detectable anymore (Fig. 5e), confirming that MLC1 dimerization is important for MLC1 functionality (Brignone et al., 2011).

Fig. 5

Effects of hyposmotic stress on WT MLC1 intracellular traffic.

a,b. Astrocytoma cell lines overexpressing WT MLC1 were incubated in control (CTR) or hyposmotic (HYPO) solution for 30 min and then stained with Abs to MLC1 (red) and Rab11 (green). In control cells (a) MLC1 immunoreactivity is found in the cell membrane and in the perinuclear cytoplasm, where it colocalizes with Rab11+ vesicles (merge). Hyposmotic stress (b) induces a marked increase in MLC1 immunoreactivity throughout the cell body and processes, in the plasma cell membrane and at astrocyte-astrocyte contacts (asterisks), and the redistribution of Rab11+ vesicles along the astrocyte cell body and cytoplasmic extensions (arrowheads). Scale bars: 10 μm. c,d. Immunofluorescence pixel intensity along the white dotted arrows drawn in representative cells in a and b was obtained using the profile analysis tool of the LSM 5 PASCAL. After hyposmotic stress, the fluorescence intensity peaks of MLC1 and Rab11 re-distribute from the central perinuclear area, typically observed in control conditions (arrow in c), to a more peripheral cytoplasmic localization and toward the plasma membrane (arrows in d). A strong increase in MLC1 fluorescence intensity is observed after hyposmotic stress (compare red lines in c and d). e. WB of total cell proteins (Input) and of enriched surface proteins after biotinylation experiments (Eluate) from WT MLC1 astrocytoma cells maintained in control (CTR) and (HYPO) hyposmotic conditions. Although hyposmotic stress does not affect the total amount of WT MLC1 protein (CTR versus HYPO in Input lanes), it strongly increases the amount of the dimeric (60 kDa) membrane-associated MLC1 component (CTR versus HYPO in Eluate lanes). Note that in the surface protein fraction the monomeric (36 kDa) MLC1 component is not detectable anymore. One representative experiment out of three performed is shown. Molecular weight markers are indicated on the left (kDa).

In contrast, hyposmotic treatment did not induce plasma membrane translocation of the S280L MLC1 mutant protein, representative of MLC1 mutants not expressed in the plasma membrane, as revealed by both immunofluorescence staining and WB analysis of biotinylated proteins (Figs. S7a,b). Immunofluorescence staining of S246R MLC1 astrocytoma cells showed that, similar to WT MLC1, this mutant protein is expressed at the plasma membrane already in control conditions but that its membrane localization does not increase after 30-min incubation in hyposmotic solution (Fig. S8). Since hyposmosis did not alter MLC1 mRNA levels (Fig. S9) these results demonstrate that the increase in plasma membrane expression of MLC1 WT induced by hyposmotic stress is mainly due to mobilization from intracellular organelles, primarily belonging to the Rab11+ PRNC. The hyposmosis-induced recycling process is hampered by pathological MLC1 mutations, independently of whether the mutated MLC1 protein localizes in the plasma membrane in isosmotic conditions.

MLC1 is involved in the regulation of early endosome pH

MLC1 interaction with proteins involved in the regulation of organelle acidification, like V-ATPase and Na, K-ATPase, and the high levels of expression of MLC1 in early and recycling endosomal compartments support the idea of a possible involvement of MLC1 in endosomal function and modulation of organelle acidity. To assess organelle pH in astrocytoma cell lines, we took advantage of the pH-sensitivity of FITC fluorescence emission in dynamic imaging experiments. After 30-min incubation of astrocytoma cells in the presence of FITC-conjugated dextran (see Materials and methods), endocytic vesicles containing the dye could be observed as fluorescence emitting dots. These were mainly localized around the cell nucleus, which was visualized with the chromatin selective dye Hoechst 33258 (Figs. 6a,b), and often clustered on one side of the nucleus.

Fig. 6

WT and mutated MLC1 differently regulate endosomal pH.

a,b. Fluorescence images of astrocytoma cells pre-loaded with FITC-dextran (green) and the chromatin selective dye Hoechst 33258 (blue), to depict endosomes and nuclei, respectively. b. Higher magnification of the area selected in (a), depicting the polarized localization of FITC-dextran loaded endosomes. Scale bar: 23 μm. c. The bar graph shows the mean ± SEM pH values in control astrocytoma cells infected with empty vector (CTR) and astrocytoma cells overexpressing WT and mutated MLC1; the number of recorded cells for each cell line ranged between 43 and 83. Significant differences between CTR and MLC1 overexpressing cells were calculated using Student's t test; *P < 0.05. d. The graph shows the time-course of pH changes in labeled endosomes of WT and S280L MLC1 astrocytoma cells recorded in a representative experiment. Note that the pH recorded in the time lag before the application of calibration solutions is more basic in WT compared to S280L MLC1 astrocytoma cells.

As expected, membrane permeabilization induced by the ionophore nigericine revealed the pH sensitivity of most FITC-dextran loaded vesicles. When the pH of FITC-dextran loaded vesicles was tested in control astrocytoma cells infected with an empty vector and WT MLC1 overexpressing astrocytoma cells, large variations in pH values were calculated within a single cell (data not shown), suggesting heterogeneity of endocytic vesicles. Interestingly, endocytic vesicles of WT MLC1 astrocytoma cells had a more basic pH (6.76 ± 0.05) while those of cells expressing the S280L and C125R mutants had a more acidic pH (6.37 ± 0.04 and 6.36 ± 0.05, respectively) compared to mock infected control cells (6.49 ± 0.03). The pH value of endocytic vesicles of S246R mutant astrocytoma cells (6.50 ± 0.02) was undistinguishable from that of control cells (Figs. 6c,d). These findings indicate that WT MLC1 overexpression increases endosomal pH and that this effect is not reproduced by overexpression of mutated MLC1.

To understand whether a specific endosomal compartment is affected by MLC1-induced changes in vesicular pH, WT MLC1 astrocytoma cells pre-loaded for 30 min with FITC-dextran (as described above) were stained with anti-MLC1 Ab and Abs to specific organelle markers (Rab5, EEA1, Rab11, Lamp-2). We observed that FITC-dextran mainly localized in MLC1+ vesicles, in EEA1+ early endosomes (Figs. 7a,b, arrows) and, to a lesser extent, in Rab5+ organelles (Fig. 7c, arrows). In the same cells no colocalization was found between FITC-dextran and Rab11 or Lamp-2 immunoreactivities (Figs. 7d and e, respectively), indicating that the changes in organelle pH measured by FITC-dextran affect predominantly the EEA1+/Rab5+ early endosomal compartment.

Fig. 7

Characterization of FITC-dextran positive vesicles in WT MLC1 astrocytoma cells.

WT MLC1 astrocytoma cells were pre-loaded with FITC-dextran for 30 min to identify endosomes in which pH changes have been recorded (see Fig. 6) and then labeled with anti-MLC1 Ab or Abs specific for endosomal organelles. a,b. MLC1 (red, a) and EEA1 (red, b) immunoreactivities are found in FITC-dextran positive vesicles (green) (arrows). c. Immunostaining for Rab5 (red) reveals a slightly lower degree of localization in the FITC-dextran positive vesicles (green) (arrows). d, e. No overlap is found between FITC-dextran (green) and Rab11 (red, d) or Lamp-2 (red, e) immunoreactivities. Scale bars: 10 μm.

In support of these findings, electron microscopy analysis of cytosolic extracts derived from rat primary astrocytes (see Materials and methods) confirmed that MLC1 is abundantly expressed in the membrane of EEA1+ organelles, in EEA1+ intraorganelle compartments and in the membranes of Rab5+ vesicles (Fig. S10), a localization compatible with a functional role of endogenous MLC1 in the early endosomal compartment.

MLC1 overexpression increases transferrin and TRPV4 channel recycling

Because control of the luminal ionic composition of endolysosomal vesicles is an essential regulatory step for protein sorting to the degradative or recycling compartments (Grant and Donaldson, 2009; Luzio et al., 2007) we next investigated whether MLC1 over-expression, by altering endosomal pH, could influence protein trafficking. Since WT MLC1 overexpression induced a minor acidification of early endosomes (see above), a condition favoring protein sorting toward the recycling pathway (Grant and Donaldson, 2009; Jovic et al., 2010; Yamashiro and Maxfield, 1987 and reference therein) we monitored protein recycling in MLC1 overexpressing astrocytoma cells using a classical assay based on the recycling properties of transferrin (Tf). This experiment was performed in basal conditions and after hyposmotic stress which favors MLC1 recycling to the plasma membrane, as shown above. Astrocytoma cell lines expressing WT MLC1, S246R or S280L mutated proteins were incubated with Alexa Fluor 488-conjugated Tf at 4 °C to allow binding to surface TfR. After washing out the excess of fluorochrome-labeled Tf (T0), we further incubated the cells at 37 °C for 20 min and overnight in the presence of excess unlabeled Tf to monitor Tf trafficking. By comparing Tf-associated fluorescence in WT and mutant MLC1 astrocytoma cells at these different time points we did not find any appreciable difference in Tf trafficking among cells expressing WT or mutated MLC1 in control conditions (data not shown). However, when MLC1 expressing astrocytoma cells were pre-incubated in hyposmotic medium for 30 min to activate MLC1 recycling before incubation with Alexa Fluor 488-conjugated Tf, we observed that compared to T0, Tf-associated fluorescence in WT MLC1 positive cells started to decrease after 20 min (not shown) and disappeared almost completely after overnight incubation, indicating extracellular release of Tf molecules, as previously reported (Bacac et al., 2011; Takahashi et al., 2012) (Figs. 8a,d). In contrast, a considerable amount of Tf-associated fluorescence was detectable in discrete intracellular aggregates in S246R and S280L MLC1 astrocytoma cells after overnight incubation, suggesting that WT but not mutant MLC1 accelerated Tf recycling (Figs. 8b,c,e,f). These results were confirmed by fluorescence intensity analysis with confocal microscope (Fig. 8g).

Fig. 8

Transferrin recycling assay in WT and mutant MLC1 astrocytoma cell lines after hyposmotic stress.

Astrocytoma cell lines expressing WT MLC1, S246R or S280L mutated protein were pre-incubated with Alexa Fluor 488-conjugated transferrin (Tf) at 4 °C to allow binding to surface TfR. After washing out labeled Tf (T0), cells were incubated at 37 °C overnight (ON) in the presence of excess unlabeled Tf and then stained with anti-MLC1 Ab (red). a,b,c. Alexa Fluor 488-conjugated Tf (green) shows comparable binding to the surface of WT and mutated MLC1 astrocytoma cells (T0). d,e,f. After overnight (ON) incubation with excess unlabeled Tf, Tf disappears almost completely from the surface of WT MLC1 astrocytoma cells (d), whereas it is still present in S246R and S280L mutant astrocytoma cells (e,f). Scale bars: 10 μm. g. The bar graph shows the mean ± SEM values of the Alexa Fluor 488-conjugated Tf fluorescence intensity in the different astrocytoma cell lines after ON incubation; 10 to 15 random fields (field area = 230 μm2) were analyzed. Significant differences between WT MCL1 and mutated (S246R, S280L) astrocytoma cells were calculated using Student's t test; *P < 0.05. **P < 0.005, ***P < 0.0005.

The above findings prompted us to investigate whether MLC1 could also influence the recycling of known MLC1-interacting proteins. We focussed our attention on TRPV4 because, among the MLC1 interactors identified so far, it is the only one for which a functional interaction with MLC1 has been demonstrated (Lanciotti et al., 2012). Preliminary experiments indicated that in WT MLC1 astrocytoma cells TRPV4 is localized in Rab11+ perinuclear vesicles (Fig. S11). Using the Tf recycling assay to analyze TRPV4 distribution in astrocytoma cells during recycling we found that at T0 in WT MLC1 cells and in cells carrying S280 MLC1 mutation TRPV4 had a similar distribution, mainly in the cytoplasmic compartments where it partially colocalized with Alexa Fluor 488-conjugated Tf (Figs. 9a,c). After 20 min of Tf internalization (data not shown) and particularly after overnight incubation (Figs. 9b,d), TRPV4-Tf colocalization almost disappeared in WT-MLC1 astrocytoma cells but not in cells expressing the S280L mutant, where TRPV4 still colocalized with Tf in discrete vesicles in the cell body and cytoplasmic processes (Fig. 9d, arrows). These experiments suggest that WT and mutated MLC1 can differently influence TRPV4 intracellular localization and recycling. In order to verify whether TRPV4 recycling to the plasma membrane was modulated by MLC1 expression we performed biotinylation assays and WB analysis of cell surface proteins derived from WT MLC1 and S280L overexpressing astrocytoma cells, in control conditions and after hyposmotic stress. These experiments showed that in WT MLC1+ cells low levels of TRPV4 were expressed in the plasma membrane in control conditions and that hyposmosis consistently increased TRPV4 translocation to the plasma membrane. In contrast, in mutant expressing astrocytoma cells TRPV4 was never found among biotinylated surface proteins, neither in control nor after hyposmotic stimulation (Fig. 9e). Overall, these results concur to demonstrate that in astrocytoma cells WT, but not mutant MLC1, influences TRPV4 trafficking and plasma membrane expression.

Fig. 9

TRPV4 recycling in WT and S280L MLC1 astrocytoma cell lines after hyposmotic stress.

Astrocytoma cell lines expressing WT or mutated S280L MLC1 proteins were incubated with Alexa Fluor 488-conjugated Tf at 4 °C to allow binding to surface TfR. After washing out labeled Tf (T0), cells were incubated at 37 °C for 20 min and overnight (ON) in the presence of excess unlabeled Tf and then stained with anti-TRPV4 pAb. a,c. TRPV4 (red) and Alexa Fluor 488-conjugated Tf (green) partially colocalize in intracellular vesicles in WT MLC1 and S280L mutant astrocytoma cell plasma membranes at T0. b,d. After ON incubation with unlabeled Tf, TRPV4-Tf colocalization disappears in WT MLC1 astrocytoma cells (b) but not in S280L astrocytoma cells where TRPV4 still colocalizes with Tf in clustered intracytoplasmic vesicles (d, arrows). Scale Bars:10 μm. e. WB of total cell proteins (input lanes) and of enriched surface proteins after biotinylation (eluates) reveals that 30-min incubation in hyposmotic solution induces an increase in surface expression of TRPV4 in WT MLC1 astrocytoma cells but not in cells expressing S280L mutation. Molecular weight markers are indicated on the left (kDa).

Discussion

Understanding the functional role of MLC1 protein in astrocytes and the effects of MLC1 mutations leading to MLC disease is an essential step toward the identification of disease mechanisms and the development of effective therapies for patients affected by this rare childhood-onset leukodystrophy. Using human astrocytoma cell lines stably overexpressing WT or mutated MLC1 proteins to study the localization and function of MLC1 in endosomal organelles, we have uncovered a role for MLC1 in pH regulation and protein trafficking in the endocytic compartment and described the impact of pathological MLC1 mutations on this pathway. Although of tumoral origin, this is a useful and reliable experimental model to study the pathophysiological role of MLC1 as the results obtained so far in human astrocytoma cells have been reproduced in rat primary astrocytes and MLC patient-derived blood cells (Lanciotti et al., 2010; and this study).

Intracellular MLC1 localizes in early and recycling endosomes and traffics along the Rab11+ perinuclear recycling compartment

We have shown that WT MLC1, but not mutated MLC1 proteins (C125R, S280L), which show defective plasma membrane localization, is abundantly expressed in early endosomal organelles identified by EEA1 and Rab5 markers (reviewed by Jovic et al., 2010; Maxfield and McGraw, 2004). Of interest, Rab5 also labels primary early endosomal vesicles, both clathrin-coated vesicles (Zerial and McBride, 2001) and caveolin-positive vesicles. The latter are responsible for caveolar-dependent endocytosis (Aoki et al., 2007; Hagiwara et al., 2009) which represents the main endocytosis route for MLC1 and its associated proteins in astrocytes (Lanciotti et al., 2010) before their sorting to EEA1+ endosomes (Fig. 10). In line with the present findings, colocalization of MLC1 and EEA1 was detected in human brain tissue (Ambrosini et al., 2008). The present results indicate that MLC1 also accumulates in perinuclear Rab11+ endosomal vesicles. The latter define a specific intracellular compartment known as perinuclear (or pericentrosomal) recycling compartment (PNRC) where many proteins, including ion and water channels and receptors, are stored and from where these molecules can be mobilized and recruited to the plasma membrane in physiological conditions (Innamorati et al., 2001; Mukherjee et al., 1997; Parent et al., 2009; Sheff et al., 1999; Takata et al., 2008; Ullrich et al., 1996). The finding that hyposmosis increases the mobilization of Rab11+/MLC1+ vesicles from perinuclear areas toward astrocytic plasma membrane and end-feet suggests the presence of an intracellular pool of MLC1 that is capable of recycling back to the cell surface in stress conditions. We also observed accumulation of MLC1 in Rab11+ and TfR+ organelles after astrocytoma cell treatment with the V-ATPase inhibitor bafilomycin that selectively inhibits protein trafficking along the degradative-lysosomal pathway but not the transport of TfR to PNRC (Baravalle et al., 2005). This finding confirms Rab11+ vesicle-mediated bidirectional trafficking of MLC1 from the plasma membrane to the PNRC and vice versa. Interestingly, in a previous study we observed that also in rat primary astrocytes endogenous MLC1 was localized in perinuclear vesicles and that treatment with nocodazole, a drug which disrupts microtubule organization, abolished MLC1 perinuclear accumulation (Lanciotti et al., 2010). Because microtubule disruption affects the perinuclear localization of Rab11+ vesicles and of membrane proteins that undergo PNRC-mediated storage and recycling (Baravalle et al., 2005; Vossenkämper et al., 2007), these findings suggest that also in primary astrocytes endogenous MLC1 accumulates in this compartment. In different cell types the presence of perinuclear pools of membrane proteins has been reported to exert an important role in the replenishment of the constitutively internalized proteins and for the maintenance of steady-state surface levels of receptor and transporter proteins like transferrin (Ullrich et al., 1996), thromboxane and dopamine receptors (Li et al., 2012; Thériault et al., 2004), glucose transporters (Ishiki and Klip, 2005; Widmer et al., 2005) ion and water channels like Kv1.5 (McEwen et al., 2007), AQP2 (Nedvetsky et al., 2007; Takata, 2006), TRPV5-6 (Van de Graaf et al., 2006, 2008) and junctional proteins, like E-cadherin (Balzac et al., 2005) and claudin-1 (Dukes et al., 2011). By showing that MLC1 is recycled through this pathway and that its recycling toward the plasma membrane is stimulated by hyposmosis, this study unveils the importance of endosomal recycling in the regulation of MLC1 function and impairment of this process by pathological MLC1 mutations.

Fig. 10

Schematic representation of MLC1 intracellular trafficking.

A model of MLC1 intracellular trafficking is proposed on the basis of our previous results (Lanciotti et al., 2010) and the data presented in this paper. MLC1 is internalized via caveolae-mediated endocytosis and traffics through Rab5+ and EEA1+ early endosomes where it is sorted to the recycling or degradative pathway. Most of the intracellular MLC1 protein is stored in the perinuclear Rab11+ recycling vesicles from which it is recycled to plasma membrane in stress condition (Hyposmosis).

Localization of endogenous MLC1 in EEA1+ and Rab11+ organelles and the effects of pathological mutations on MLC1 intracellular distribution were also observed in monocyte-derived macrophages from healthy donors and MLC1 mutated patients. These findings allow us to exclude that the intracellular distribution and trafficking properties of WT MLC1 as well as the defects induced by MLC1 mutations in astrocytoma cells might be due to MLC1 overexpression or to the tumoral nature of the cells.

MLC1 is involved in the regulation of early endosome acidity

The identification of V-ATPase and Na, K-ATPase, also known to regulate early endosome pH, as proteins interacting with MLC1 (this study and Brignone et al., 2011, respectively) led us to hypothesize a role for MLC1 in the control of organelle pH. FITC-dextran measurement coupled with vesicle immunostainings revealed a decrease in the acidification of early endosomes (EEA1+ and Rab5+) in WT MLC1 astrocytoma cells compared to mock infected control cells. Importantly, overexpression of all 3 pathological MLC1 mutants (S246R, S280L, C125R) did not increase endosomal pH, indicating altered regulation of this pathway in MLC disease. Because the S246R mutation does not affect membrane expression of MLC1, these data suggest that reduced MLC1 localization in the plasma membrane is not the sole indicator of a pathological phenotype. Altogether, these data confirm the hypothesis that MLC1 is involved in the regulation of organelle acidity, possibly by limiting vesicle acidification. A similar behavior has been described for the Na, K-ATPase which exerts an essential role in the maintenance of the slightly acidic pH typical of early endosomes in which it is localized (Cain and Murphy, 1988; Cain et al., 1989; Feldmann et al., 2007; Grabe and Oster, 2001) and whose direct interaction with MLC1 has been demonstrated by our group (Brignone et al., 2011). Future experiments will aim to clarify the exact mechanisms through which MLC1 can influence endosomal pH and the molecular and functional relationships between MLC1 and V-ATPase. Because it is known that abnormal acidification can lead to endosome enlargement (Bacac et al., 2011; Forgac, 2007; Martina et al., 2009), these data lead to the hypothesis that the formation of intracellular vacuoles observed in MLC1 silenced rat astrocytes and in astrocytes in the brain of MLC patients (Duarri et al., 2011) might be due to endosome swelling caused by dysregulation of organelle pH.

The mildly acidic pH of the endocytic pathway is strictly regulated by a variety of ion channels, transporters and exchangers present in the endosomal membranes (Scott and Gruenberg, 2011) where also endogenous MLC1 is localized (Fig. S10). Along with V-ATPase other proteins can influence vesicle acidification, including different components of the transient receptor potential (TRP) type, of the two pore (TPC) type of calcium channels (Abe and Puertollano, 2011; Martina et al., 2009; Morgan et al., 2011) and of the chloride channels, that mainly control late/lysosomal compartment acidity (Edwards and Kahl, 2010; Faundez and Hartzell, 2004). We cannot exclude that, in addition to an effect on the activity of its molecular interactors (V-ATPase, Na, K-ATPase), MLC1 may function itself as an ion channel and directly influence organelle proton influx.

MLC1 influences protein recycling

Early endosome organelles constitute the sorting station where the fate of internalized proteins and lipids is decided by a complex interplay of molecular and structural determinants mainly leading to changes in intra-organelle pH (Jovic et al., 2010). There is broad evidence that the acidification of endocytotic vesicles is essential for the regulation of uncoupling of receptor–ligand complexes, intracellular membrane flow and protein sorting toward the recycling or degradative pathway (Jovic et al., 2010; Luzio et al., 2007) and that inhibition or overexpression of proteins involved in the regulation of endosomal pH affects protein trafficking (Bacac et al., 2011; Hara-Chikuma et al., 2005; Lelouvier and Puertollano, 2011; Martina et al., 2009; Smith and Lippiat, 2010). Our experiments indicate that WT but not mutated MLC1 favors transferrin and TRPV4 channel recycling. The present results are in agreement with a functional role of MLC1 in endosomal pH regulation, since recycling vesicles are characterised by a slightly less acidic pH compared to vesicles sorted toward the lysosomal degradative pathway (Grant and Donaldson, 2009; Scott and Gruenberg, 2011). The demonstration that S280L and S246R MLC1 mutations do not promote transferrin and TRPV4 channel recycling supports the hypothesis that pathogenic MLC1 mutations influence endosomal maturation and protein sorting decision.

We also show for the first time that TRPV4 is expressed in Rab11+ perinuclear vesicles and is subjected to recycling pathway, as previously reported for TRPV5 and TRPV6, two transient receptor cation channels belonging to the same family of vanilloid-type of transient receptors (Van de Graaf et al., 2006). The influence of MLC1 on the recycling rate of its molecular partner TRPV4 could explain the molecular mechanism underlying the functional effect of MLC1 on the TRPV4-mediated calcium influx that we have recently described (Lanciotti et al., 2012). Defective TRPV4 recycling in hyposmotic conditions in MLC1 mutant expressing astrocytoma cells could be responsible for a dysfunctional astrocyte response to osmotic stress. The analysis of proteins exposed on the cell surface in basal and hyposmotic conditions revealed a marked reduction in plasma membrane localization of TRPV4 in mutant MLC1 expressing astrocytoma cells. Interestingly, defects in TRPV4 function have been recently reported to be responsible for cyst formation in the autosomal recessive polycystic kidney disease (Zaika et al., 2013), suggesting that also in MLC disease cyst formation could be caused by TRPV4 functional alterations. To date, TRPV4 is the only identified MLC1 interactor whose activity is modulated by MLC1 in rat astrocytes and human astrocytoma cells (Lanciotti et al., 2012). Recently, mutations in the gene encoding Hepacam/Glialcam, an adhesion-like molecule of unknown function, have been found in a considerable percentage (> 50%) of MLC patients without mutations in MLC1 and Hepacam/Glialcam protein has been reported to regulate specifically MLC1 expression at astrocyte–astrocyte junctions (Duarri et al., 2011). WB analysis and biochemical assays indicated that Hepacam/Glialcam is expressed in astrocytoma cell lines and binds MLC1, and that MLC1 mutations differently affect this interaction (Lanciotti et al., 2012). Further studies are needed to clarify MLC1-Hepacam/Glialcam interaction in the astrocytoma cell model.

Conclusion

This study shows that in astrocytes MLC1 is expressed in early endosomes and recycled through the Rab11+ perinuclear compartment (Fig. 10). MLC1 localization in the early endosomal compartment is functionally relevant to modulate organelle acidity and protein recycling, including recycling of the MLC1 interactor TVPR4, in stress conditions like hyposmosis. It is possible that trafficking of other MLC1 interacting proteins, like Na, K-ATPase complex proteins and junction proteins [i.e. ZO-1 (Duarri et al., 2011)], is influenced by MLC1 and that dysregulation of protein trafficking following changes in brain homeostasis induces or amplifies brain damage in MLC patients carrying mutated MLC1 genes. It is known that some junction proteins are continuously being endocytosed and recycled back to the plasma membrane (Chalmers and Whitley, 2012; Dukes et al., 2012) via caveolar-dependent internalization and Rab11 recycling pathway (Desclozeaux et al., 2008; Dukes et al., 2011; Nighot and Blikslager, 2012) and that their traffic can be modulated by the activity of ion channels responsible for endosomal acidification (Nighot and Blikslager, 2012). Abnormal recycling of junction proteins can lead to the formation of intracellular vacuoles (Dukes et al., 2012) similar to those observed in the brain of MLC patients (Duarri et al., 2011). We hypothesize that MLC1-mediated recycling becomes relevant when MLC1 is recruited to the plasma membrane due to an increased functional demand of the astrocyte. In particular, regulation of MLC1 in the astrocyte membrane could be critical for the astrocyte response to changes in extracellular osmolarity.

Endocytosis and recycling are essential processes in the regulation of the expression of cell surface molecules that mediate glial cell differentiation and neuronal–glial interactions during brain development (Chen et al., 2011; Shilo and Schejter, 2011; Yap and Winckler, 2012). Modification of these processes may result in brain oedema and disturbance of myelin formation, which are observed also in other leukodystrophies associated with specific defects in astrocyte maturation and function (Bugiani et al., 2011; Messing et al., 2012). By increasing our understanding of the molecular mechanisms that lead to brain damage in MLC patients and could be shared with other leukodystrophies, this study opens new perspectives for the identification of therapeutic targets.

Funding

The financial support of ELA Foundation (grants ELA 2006-001/4 and ELA 2009-002C5A to EA) and Telethon Italy (grant n. GGP1118) is gratefully acknowledged. AL is the recipient of an ELA foundation fellowship (grant n. 2012-021F2).

Conflict of interest statements

The authors have no conflicting financial interests.