Distinct Modes of Balancing Glomerular Cell Proteostasis in Mucolipidosis Type II and III Prevent Proteinuria

Abstract

Visual Abstract

Background 

The mechanisms balancing proteostasis in glomerular cells are unknown. Mucolipidosis (ML) II and III are rare lysosomal storage disorders associated with mutations of the Golgi-resident GlcNAc-1-phosphotransferase, which generates mannose 6-phosphate residues on lysosomal enzymes. Without this modification, lysosomal enzymes are missorted to the extracellular space, which results in lysosomal dysfunction of many cell types. Patients with MLII present with severe skeletal abnormalities, multisystemic symptoms, and early death; the clinical course in MLIII is less progressive. Despite dysfunction of a major degradative pathway, renal and glomerular involvement is rarely reported, suggesting organ-specific compensatory mechanisms.

Methods 

MLII mice were generated and compared with an established MLIII model to investigate the balance of protein synthesis and degradation, which reflects glomerular integrity. Proteinuria was assessed in patients. High-resolution confocal microscopy and functional assays identified proteins to deduce compensatory modes of balancing proteostasis.

Results 

Patients with MLII but not MLIII exhibited microalbuminuria. MLII mice showed lysosomal enzyme missorting and several skeletal alterations, indicating that they are a useful model. In glomeruli, both MLII and MLIII mice exhibited reduced levels of lysosomal enzymes and enlarged lysosomes with abnormal storage material. Nevertheless, neither model had detectable morphologic or functional glomerular alterations. The models rebalance proteostasis in two ways: MLII mice downregulate protein translation and increase the integrated stress response, whereas MLIII mice upregulate the proteasome system in their glomeruli. Both MLII and MLIII downregulate the protein complex mTORC1 (mammalian target of rapamycin complex 1) signaling, which decreases protein synthesis.

Conclusions 

Severe lysosomal dysfunction leads to microalbuminuria in some patients with mucolipidosis. Mouse models indicate distinct compensatory pathways that balance proteostasis in MLII and MLIII.

A proteome adapted to cellular needs depends on the balanced interplay between protein synthesis, protein folding, and protein degradation by the autophagosomal, lysosomal, and the ubiquitin proteasome pathways. The renal glomerulus comprises mesangial cells, glomerular endothelial cells, and podocytes that together form a functional syncytium to build and regulate the glomerular filtration barrier. How glomerular cells balance their proteome in the setting of injury is unknown. Further, it is not known how unbalanced proteostasis affects glomerular cells. The extent of glomerular cell involvement varies strongly in lysosomal storage disorders, a heterogeneous group of diseases caused by unbalanced proteostasis secondary to lysosomal dysfunction. Although glomerular involvement is common and constitutes a key contributor to the morbidity and mortality in patients with Fabry disease, a monogenic lysosomal storage disorder caused by mutations in the GLA gene¹ encoding α-galactosidase A, glomerular/renal impairment is usually rare in other lysosomal storage diseases. Nonetheless, renal dysfunction has been reported in certain patients with nephrosialidosis,² infantile sialic acid storage disease,³ cystinosis,⁴ mucopolysaccharidosis I,⁵ as well as with mucolipidosis (ML),^6–8 leading to the question of whether glomerular cell involvement was the result of lysosomal impairment or a coincidence in these cases.

For lysosome biogenesis, newly synthesized lysosomal enzymes are equipped with mannose 6-phosphate (M6P)–targeting signals. Subsequently, M6P-containing lysosomal enzymes bind to M6P receptors, mediating their vesicular transport from the trans-Golgi network via the endosomal compartment to lysosomes.⁹ The key enzyme in the formation of M6P residues is the Golgi-resident hexameric (α₂β₂γ₂) GlcNAc-1-phosphotransferase. The GNPTAB gene encodes the common membrane-bound α/β-subunit precursor that is cleaved by site-1 protease into the individual, catalytically active α- and β-subunits; whereas the soluble, regulatory γ-subunit is encoded by GNPTG.^10–12

ML type II (MLII) and III (MLIII) are rare lysosomal storage disorders whose classification is based on age of onset, clinical symptoms, and severity. More specifically, mutations in GNPTAB resulting in inactive GlcNAc-1-phosphotransferase cause the severe MLII disease, whereas a few mutations in GNPTAB and mutations in GNPTG, which are accompanied by a residual GlcNAc-1-phosphotransferase activity, cause the less-progressive diseases MLIII alpha/beta or MLIII gamma, respectively.¹³ Consequently, in cells from patients with MLII and MLIII, the generation of M6P-targeting signals on lysosomal enzymes is impaired and leads to their missorting and hypersecretion into the extracellular compartment.¹⁴ The subsequent intracellular deficiencies of multiple lysosomal enzymes result in the accumulation of various nondegradable macromolecules in dysfunctional lysosomes, which impair cellular homeostasis. Clinical manifestations of patients with MLII include progressive neurodegeneration, severe skeletal abnormalities, organomegaly, and death in the first decade of life.¹³ MLIII alpha/beta and MLIII gamma are less severe and progressive forms compared with MLII, and mainly affect skeletal and joint tissues.¹³ Although the clinical onset and manifestations are highly variable in affected individuals, MLIII gamma seems to be less severe than MLIII alpha/beta.

In the case of patients with MLII and MLIII, only sporadic cases of renal involvement (mostly glomerular involvement) have been reported.^6–8 Thereby, two autopsy cases of patients with MLIII alpha/beta described a foamy transformation of podocytes in a 45-year-old woman⁸ and in a man with normal renal function up to his death.⁷ In a large kindred family with MLIII alpha/beta, two of the five affected children exhibited nephrotic-range proteinuria with FSGS and podocytes that had a foamy appearance, presumably due to lysosomal storage accumulation.¹⁵ Besides glomerular involvement, tubular involvement has been suggested to occur in a patient with MLII, who presented with a slightly increased aminoaciduria and low mol wt proteinuria, indicative of representing a defective proximal tubular function.¹⁶ However, kidney and glomerular impairment has not been observed in patients with MLIII gamma thus far.¹⁵

To evaluate whether the sporadic reports of a glomerular phenotype in patients with ML were due to dysfunction of GlcNAc-1-phosphotransferase, we analyzed the urine of patients with ML for protein loss. We generated a new mouse model for MLII; used an established model of MLIII; and evaluated whether MLII and MLIII mutants developed a renal phenotype, focusing on the glomerular phenotype. We hypothesized that absence of a glomerular phenotype in the setting of dysfunction of such a major degradative pathway as the lysosomes necessitated the activation of unknown organ-/cell-specific compensatory mechanisms to balance proteostasis and thereby preclude/attenuate glomerular cell involvement.

Methods

Antibodies

Primary antibodies used for the study were rat anti–UCH-L1 (immunofluorescence [IF] microscopy 1:50; Western blot [WB] 1:250, self-made clone U104¹⁷), guinea pig anti-nephrin (IF 1:100; Acris), mouse anti-ubiquitin (WB 1:250; Millipore), rabbit anti-ubiquitin (IF 1:300; Novus), rabbit anti–K48-polyubiquitin (WB 1:1000; Abcam), rabbit anti-β5 (WB 1:1000; X. Wang, University of South Dakota), rabbit anti-LMP7 (WB 1:5000; E.K.), rabbit anti-Limp2 (WB 1:1000, IF 1:1000; P. Saftig, University of Kiel, Kiel, Germany), rabbit anti-Lamp2 (WB 1:1000, IF 1:400; Sigma-Aldrich), rabbit anti-LC3B (WB 1:1000, IF 1:50; Sigma-Aldrich), mouse anti–β-actin (WB 1:10,000; Sigma-Aldrich), goat anti–kidney injury molecule 1 (IF 1:1000; R&D Systems), rabbit anti–smooth muscle actin (IF 1:400; Abcam), rabbit anti–cleaved caspase 3 (IF 1:100; Cell Signaling), rhodamine wheat germ agglutinin (IF 1:400; Vector), rabbit anti-pS6 (IF 1:100, WB 1:1000; Cell Signaling), rabbit anti-S6 (WB 1:1000), rabbit anti–p-PERK (IF 1:100; Invitrogen), rabbit anti-ATF4 (WB 1:1000, IF 1:100; Bioss), rabbit anti-NRF1 (WB 1:1000, IF 1:100; Cell Signaling), rabbit anti–p(S40)-NRF2 (WB 1:1000, IF 1:50; Bioss), rabbit anti–eukaryotic translation initiation factor 2 α (anti-eIF2α; WB 1:1000), rabbit anti–p-eIF2α (WB 1:1000, IF 1:50; Cell Signaling), and guinea pig anti-p62 (IF 1:100; Progen). All secondary antibodies used were either biotinylated or horseradish peroxidase- or fluorescent dye–conjugated, affinity-purified donkey antibodies (Jackson ImmunoResearch).

Animals

Gnptg^ko (hereafter called MLIII) mice have been described previously.¹⁸ A Gnptab-targeting vector¹⁹ used for generation of MLII mice contained a neomycin resistance cassette between exons 16 and 17, allowing selection for recombination in targeted embryonic stem cells, and two loxP sites located in introns 13 and 16 (Supplemental Figure 1A). Targeted embryonic stem cells were injected into C57BL6/J blastocysts and subsequently implanted into the uterine horns of C57BL6/JxCBA foster female mice according to standard protocols. Offspring chimeric males from three clones were further crossed with Cre-expressing mice, resulting in mice (hereafter called MLII) that carry the GnptabΔ^ex14–16 allele lacking the floxed region (exons 14, 15, and 16, and the neomycin cassette). Heterozygous MLII mice were then inbred to generate homozygous MLII animals. Experiments were performed with mice of a mixed C57BL6/J-129/SvJ genetic background, with littermates used as controls. Mice were housed in a pathogen-free animal facility at the University Medical Center Hamburg-Eppendorf, and experimental procedures were performed according to the institutional guidelines. For genotyping of mice, genomic DNA from tail biopsies was extracted using the KAPA Mouse Genotyping Hot Start Kit (VWR) and analyzed by multiplex PCR using primers F1 (5′-CAT CCC ACC GAC TCA GGA AG-3′), F2 (5′-GAA ATG TTG CAC CAA ACT GG-3′), and R (5′-GCA GCA GTG CCC ATC TGA TA-3′) (Supplemental Figure 1B). To confirm deletion of exons 14–16 at the transcriptional level, total RNA was isolated from liver wild-type and MLII mice, converted to cDNA, and sequenced (Seqlab) using primers F3 (5′-CGC GGG CTC CTC CTG GCG TGG-3′) and R2 (5′-GAC TCC GCT GTG GTG GTG TCG AGG AG-3′) (Supplemental Figure 1C). Furthermore, quantitative PCR (qPCR) using exon-specific TaqMan primers revealed normal Gnptab transcript levels for exons 1 and 2, whereas no transcripts were detectable using primers for the exons 14 and 15 (Gnptab, exons 1–2: Mm01773334_m1; Gnptab, exons 14–15: Mm02599733_g1; Supplemental Figure 1D). For renal analyses, MLII and MLIII mice were older than 15 weeks and predominantly analyzed at 30–40 and 60–90 weeks of age, respectively. For bone analyses, 20-week-old wild-type and MLII mice were used.

Generation of the Human α/β-del Construct

Generation of the expression constructs for the C-terminally myc-tagged, full-length α/β-subunit precursor (α/β-myc) of GlcNAc-1-phosphotransferase using the expression vector pcDNA3.1D-TOPO was previously described.²⁰ Deletion of the amino acids 906–1083 from the α/β-myc construct was performed by site-directed mutagenesis using mutagenic primers F-del (5′-GTA TTT CCA AGA TCT TCT CGA CCC ACC GGT CAC TAA AAG TCT AG-3′) and R-del (5′-CTA GAC TTT TAG TGA CCG GTG GGT CGA GAA GAT CTT GGA AAT AC-3′) and Phusion Polymerase (Thermo Fisher Scientific). The mutated plasmid DNA was commercially sequenced to confirm proper deletion (Seqlab).

Immunoblot Analysis

Preparation of cell extracts, measurements of GlcNAc-1-phosphotransferase activity, and SDS-PAGE followed by immunoblotting for the human analysis were performed as described.^18,21 All other immunoblots were performed with isolated glomeruli or total kidney from 30- to 90-week-old MLII or MLIII mice and their respective littermates as follows. Samples were lysed in T-Per (Thermo Fisher Scientific) containing 1 mM sodium fluoride, 1 mM sodium vanadate, 1 mM calyculin A, and cOmplete (Roche), and denatured in SDS solubilization buffer. Samples were separated on a 4%–12% Mini Protean TGX gel (Bio-Rad) in a Tris-glycine migration buffer (0.25 M Tris, 1.92 M glycine, 1% SDS, pH 8.3). Protein transfer was performed in transfer buffer (0.192 M glycine, 25 mM Tris base, 20% ethanol in water [H₂O]) in a TransBlot Turbo System (Bio-Rad). After the transfer, all proteins were visualized by Ponceau staining. Polyvinylidene difluoride membranes (Millipore) were blocked (5% nonfat milk) before incubation with primary antibodies diluted in SuperBlock blocking reagent (Thermo Fisher Scientific) or nonfat milk. Binding was detected by incubation with horseradish peroxidase–coupled secondary antibodies (1:10,000, 5% nonfat milk). Protein expression was visualized with ECL SuperSignal (Thermo Fisher Scientific) according to manufacturer’s instructions on an Amersham Imager 600 (GE Healthcare). Immunoblots were analyzed using software from ImageJ.²² Ponceau and β-actin stainings of the same membrane are shown and were used as loading control and for densitometric normalization. Bands of the same membrane are shown; fine dashed white lines indicate where bands were not adjacent to one another on the membrane.

Skeletal Analysis

Dissected skeletons were fixed in 3.7% PBS-buffered formaldehyde for 18 hours at 4°C and stored in 80% ethanol. All skeletons were first analyzed by contact radiography (Faxitron x-ray) to measure the length of the lumbar spine, femora, and tibia. For nondecalcified bone histology, the lumbar vertebral bodies L1 to L4 of each mouse were dehydrated in ascending alcohol concentrations and then embedded in methyl methacrylate as described previously.²³ Sections of 4 μm thickness were cut sagittally on a Microtec rotation microtome (Techno-Med GmbH) and stained by von Kossa/van Gieson and by toluidine blue procedures. Histomorphometry was performed according to the American Society for Bone and Mineral Research guidelines²⁴ using the OsteoMeasure system (Osteometrics).

Lysosomal and Proteasomal Activity Assays

Photometric activity measurements of the lysosomal enzymes β-hexosaminidase, β-galactosidase, and arylsulfatase B in the serum of mice were described elsewhere.^18,19 For measurement of β-hexosaminidase activity in glomeruli, samples were lysed in T-Per (Thermo Fisher Scientific) supplemented with Protease cOmplete Inhibitor Cocktail without EDTA (Roche). Protein concentration was determined by spectrophotometry (DeNovix DS-11), and 6 µl of samples were incubated with 24 µl H₂O and 30 µl substrate buffer (10 mM 4-nitrophenyl N-acetyl-β-D-glucosaminide, 0.1 M sodium citrate, 0.2% BSA, pH 4.6) per well for 1 hour at 37°C in the dark. The reaction was stopped with 240 µl stop solution (0.4 M glycine, pH 10.4) per well. Activity was measured in duplicates using a microplate spectrophotometer (EL 808; BioTek) at 405 nm.

For measurement of the main proteolytic activity of the proteasome, isolated glomeruli were lysed in T-Per (Thermo Fisher Scientific) supplemented with Protease cOmplete Inhibitor Cocktail without EDTA (Roche). Protein concentration was determined by spectrophotometry (DeNovix DS-11). For the measurement of proteasomal (chymotrypsin-like) activity, 10 mg total protein was diluted in incubation buffer (20 mM HEPES, 0.5 mM EDTA, 5 mM dithiothreitol, 0.1 mg/ml ovalbumin in H₂O, pH 7.8) to a final volume of 50 µl. Samples were preincubated in incubation buffer for 2 hours at 4°C. Following preincubation, the substrate Suc-LLVY-AMC (Calbiochem) was added to the samples at a final concentration of 60 µM and to an end volume of 100 µl. Proteasomal activity was measured in triplicate at 355 and 460 nm in a Mithras LB 940 fluorescent spectrophotometer after incubation at 37°C for 2 hours in the dark.

Sample Collection, Serum, and Urine Analyses

Kidney removal and blood collection was performed in 15- and 30- to 40-week-old MLII mice and 60- to 90-week-old MLIII mice. Glomeruli were isolated using Dynabead perfusion as described.²⁵ Samples were stored at −80°C before further analysis. Urine samples from patients with MLII, MLIII alpha/beta, and MLIII gamma as well as from healthy individuals were collected at the University Medical Center Hamburg-Eppendorf (Hamburg, Germany) and the Hospital de Clínicas de Porto Alegre, Porto Alegre (Brazil).

Mouse urine was collected in metabolic cages. Patient urine and mouse serum and urine were analyzed by standard methods using an autoanalyzer (Hitachi 717; Roche) in the Department of Clinical Chemistry at the University Medical Center Hamburg-Eppendorf. Proteinuria in MLII and MLIII mice and patients was assessed by Coomassie blue staining of creatinine-adapted urine as described before.²⁶ Urine albumin content was quantified using a commercially available ELISA system (Bethyl for murine urine and Dunn Labortechnik GmbH for human urine) according to the manufacturer’s instructions, using an ELISA plate reader (BioTek), as described.²⁶ Urinary albumin values were standardized against urine creatinine values of the same individuals determined according to Jaffe and plotted.

Morphologic Analysis

Kidney cortex was embedded in paraffin for light or high-resolution confocal microscopy. Sections of 1 µm were cut on a rotation microtome and stained with Periodic acid–Schiff (PAS) reagent (Sigma-Aldrich) for light microscopic evaluation according to the manufacturer’s instructions. For IF stainings, 2-µm paraffin sections were deparaffinized and antigen retrieval was performed by microwave boiling (10 mM citrate buffer, pH 6.1) or by protease XXIV (5 µg/ml; Sigma-Aldrich) digestion. Unspecific binding was blocked in 5% horse serum for 30 minutes. Primary antibody incubations (5% horse serum, overnight, 4°C) were followed by incubation with biotinylated or AF488- or Cy3-coupled secondary antibodies (1:400, 30 minutes). Stainings were evaluated with an LSM510 META microscope for conventional microscopy or with an LSM800 with Airyscan for high-resolution confocal microscopy using the LSM or ZENblue software (all Zeiss, Oberkochen, Germany). Nuclear levels of TFEB, ATF4, NRF1, and p-NRF2 and cytoplasmic levels of p-PERK within glomerular cells were quantified using ImageJ of five individual glomeruli from n≥3 mice per group. Mean fluorescent intensities (MFIs) were normalized to the respective nuclear or cytoplasmic area and are indicated as mean±SEM.

For electron-microscope analyses, small cortical samples were perfusion fixed in 4% buffered paraformaldehyde with 1% glutaraldehyde. Tissue was postfixed with 1% osmium in 0.1 M phosphate buffer (1 hour at room temperature [RT]), stained with 1% uranyl acetate (1 hour at RT in 70% ethanol), dehydrated, and embedded in epoxy resin (Durcupan; Sigma-Aldrich). Ultrathin sections were cut (Ultramicrotome UC6; Leica) and contrasted with lead citrate. Micrographs were generated with a transmission-electron microscope (TEM 910; Zeiss).

Primary Culture Podocytes

Glomeruli from MLII, MLIII, and littermate control mice were isolated by magnetic bead isolation. A total of 800 glomeruli were plated on collagen type 1–coated, 35-mm dishes (Becton Dickinson) in culture medium (37°C, 5% carbon dioxide, RPMI 1640 supplemented with 10% FCS, 15 mmol/L HEPES buffer solution, 1 mmol/L sodium pyruvate; Gibco, Grand Island, NY) and podocytes were allowed to grow out. After 7 days, podocytes were fixed with 4% paraformaldehyde for 8 minutes at RT, washed with PBS, and processed for high-resolution confocal microscopy. Unspecific binding was blocked with 5% horse serum in 0.05% Triton X-100 for 30 min. Rabbit-LC3B (1:50) and guinea pig p62 (1:100) were incubated overnight in blocking buffer. Following washes in PBS, Cy2-anti-rabbit, Cy5-anti-guinea pig (both 1:200; Jackson ImmunoResearch Laboratories), AF568-phalloidin (1:200; Molecular probes), and Hoechst (1:1000; Molecular Probes) were incubated for 30 minutes in blocking buffer at RT. Following washes, stainings were covered with fluoromount and analyzed using a LSM800 with Airyscan using ZENblue software (all Zeiss).

qPCR Analysis

Total mRNA was extracted from whole kidney and glomeruli using phenol/chloroform. The tissue was lysed with Tungsten Carbide Beads (Qiagen) and TRIzol (Ambion) in a TissueLyser II (Qiagen) at 30 Hz for 1 minute. The RNA was separated with 1/6 vol chloroform, precipitated with isopropanol at 4°C for 30 minutes, and the RNA pellet was washed three times with 80% ethanol. The RNA was then dissolved in purified H₂O. Extracted RNA (200 ng) was then reverse transcribed using Random Hexamer Primer (Invitrogen) and MMLV Reverse Transcriptase (NEB). mRNA expression was quantified with a QuantStudio 3 (Applied Biosystems) AbiPrism NN8860 using SYBR green as recently described.²⁷ The exon-spanning primer pairs for murine cDNA used are listed in Supplemental Table 2. Glyceraldehyde-3-phosphate dehydrogenase or 18S was used as an internal control to correct for small variations in RNA quality and cDNA synthesis as described by AbiPrism. Amplicons of random samples for each primer pair were determined by automatic PCR sequencing to demonstrate the specificity of the PCR reaction (data not shown). Relative gene expression was calculated using the ΔΔCT method.

Statistical Analysis

Bone results were expressed as mean±SD, and the mean values were compared using the t test. Differences with P<0.05 were considered statistically significant. Renal group measures were given as mean±SEM, and significance was set at P<0.05. The means were compared using the two-tailed nonparametric Mann–Whitney U test to enable robust conclusions on the significance of effects in case of departures from normality associated with small sample sizes. The replicates used were biologic replicates, which were measured using different samples derived from distinct mice. All animals were littermates and were blindly assigned to the experimental groups.

Results

Growth Retardation and Low Bone Mass in MLII Mice

In this study, we used a novel mouse model for the MLII disease, which was generated by deletion of exons 14–16 in the mouse Gnptab gene (Supplemental Figure 1). The resulting in-frame deletion of a 207 amino acid fragment in the α/β precursor of GlcNAc-1-phosphotransferase (p.E855_L1062del) harbors the site-1 protease cleavage site at the position K907/D908, which is mandatory for proteolytic activation of GlcNAc-1-phosphotransferase.¹⁰ Accordingly, the deletion of a corresponding fragment from a human GNPTAB expression construct prevented the cleavage of the mutant p.E906_L1083del, which led to a strong reduction of GlcNAc-1-phosphotransferase activity to 4% compared with the wild-type protein (Supplemental Figure 2). In cells from patients with MLII and a severe clinical presentation, the GlcNAc-1-phosphotransferase activity was found to be <10%, which resulted in missorting and hypersecretion of lysosomal enzymes into the extracellular space.¹³ Accordingly, in serum of MLII mice, activities of the lysosomal enzymes β-hexosaminidase, β-galactosidase and arylsulfatase B were increased 12- to 36-fold compared with wild-type littermates (Figure 1A). Notably, the lack of M6P formation on lysosomal enzymes in MLII cells prevents endocytosis of hypersecreted lysosomal enzymes via M6P receptors located at the plasma membrane and results in the absence of these enzymes in lysosomes.

Figure 1.: Growth retardation and osteoporosis in MLII mice. (A) Relative enzyme activities of the lysosomal enzymes β-hexosaminidase (β-Hex), β-galactosidase (β-Gal), and arylsulfatase B (Arsb) were measured in serum from wild-type (WT) and MLII mice (WT=1, mean±SD, *P≤0.05, n≥10). (B) MLII mice show flat facial profile and skeletal abnormalities. (C) Body weight of WT and MLII mice (mean±SD, *P≤0.05, n≥12). (D) Contact radiograph quantification of femur length (FeL) and tibia length (TiL) of WT and MLII mice (mean±SD, *P≤0.05, n=4). (E) Representative toluidine blue staining of undecalcified tibia sections from WT and MLII mice. The growth plate (GP) areas are indicated. Scale bars, 20 µm. (F) Quantification of the GP width of the same mice (n=4, mean±SD). (G) Representative Kossa/Gieson staining of undecalcified vertebra sections from WT and MLII mice. Scale bars, 1 mm. (H) Quantification of the vertebral trabecular bone volume per tissue volume (BV/TV), trabecular thickness (Tb.Th), and trabecular number per mm (Tb.N) of the same mice (mean±SD, *P≤0.05, n=4). (I) C-terminal telopeptides of type I collagen (CTX-I) in the serum of WT and MLII mice (mean±SD, *P≤0.05, n≥4).

Skeletal alterations and growth retardation are clinically relevant complications in patients with MLII²⁸ and were also present in two other Gnptab-targeted mouse models.^29,30 Likewise, our MLII mice showed back deformities, prominent skeletal abnormalities, a flat facial profile, and a reduced body weight compared with wild-type littermates (Figure 1, B and C). For in-depth skeletal examination, we first analyzed wild-type and MLII littermates by contact x-rays. By quantifying the lengths of femurs and tibias, we observed that the size of these skeletal elements was reduced in MLII mice (Figure 1D). Because skeletal growth primarily depends on the coordinated differentiation of chondrocytes within the growth plates, we next applied histology with subsequent toluidine blue staining of calcified tibia sections (Figure 1E). Here it was evident that the growth plate chondrocytes were strikingly enlarged in MLII mice. Moreover, there was a significant increase of the growth plate width in tibia sections of MLII mice compared with wild-type littermates (Figure 1F).

Because patients with MLII are characterized by progressive osteoporosis,¹³ we next analyzed calcified spine sections from wild-type and MLII littermates (Figure 1G). Histomorphometric quantification of trabecular bone parameters demonstrated that the bone volume to tissue volume ratio was significantly reduced in MLII mice compared with wild-type littermates (Figure 1H). This osteopenic phenotype was primarily reflected by a twofold reduction of trabecular number, in addition to trabecular thickness that was also significantly reduced compared with wild-type littermates (Figure 1H). Furthermore, there was a nearly twofold increase of bone resorption compared with wild-type littermates, as assessed by measuring the serum concentrations of the bone resorption biomarker CTX-1 (Figure 1I). Thus, the generated MLII mouse model shows the typical severe biochemical and skeletal features of the human disease and is suitable for further investigation of whether the defects in the GlcNAc-1-phosphotransferase are causative of the glomerular phenotype described in a subset of patients.

Lysosomal Dysfunction in Glomerular Cells of MLII and MLIII Mice

Next, we evaluated whether glomerular cells of MLII and MLIII mice exhibit lysosomal dysfunction. We first measured the activity of the lysosomal enzyme β-hexosaminidase in glomeruli and found a significant reduction of 25% in MLII glomeruli and of 45% in MLIII glomeruli (Figure 2A). In lysosomal storage disorders, the lysosomal accumulation of partially undegraded catabolic products in lysosomes leads to an increase in size and number of lysosomes. Accordingly, immunoblot analyses demonstrated an increase of lysosomal and autophagosomal markers in glomeruli of both MLII (Figure 2B) and MLIII mice (Figure 2C). The most prominent increase was noted for the lysosomal integral membrane protein 2 (Limp2) in MLII and MLIII glomeruli. Levels of the lysosomal-associated membrane protein 2 (Lamp2) and of the marker of autophagosomes, the microtubule-associated proteins 1A/1B light chain 3B (LC3), were less strongly elevated. The total amount of LC3 was increased in MLII and MLIII glomeruli and the ratio of the lipidated autophagosomal membrane-bound LC3-II form to the cytoplasmic LC3-I form was also elevated, indicating autophagosome formation. High-resolution confocal evaluations revealed an enlargement of Limp2-positive vesicles in glomerular cells, which was most prominent in endothelial cells but less apparent in podocytes and mesangial cells (Figure 2D). Within the renal cortex, tubulointerstitial cells also exhibited a remarkable accumulation of enlarged Limp2-positive lysosomes (Supplemental Figure 3). Surprisingly, no drastic alterations were noted in the IF expression of the Lamp2-positive lysosomes, although an increased Lamp2 expression in glomeruli of MLII and MLIII mice was shown by immunoblotting. Lamp2-positive lysosomes were most prominent in proximal tubular cells, but not obviously different in MLII and MLIII kidneys compared with wild-type littermates. In glomeruli, an enhanced occurrence of Lamp2-positive lysosomes was seen in podocytes and endothelial and mesangial cells of MLII and MLIII mice (Supplemental Figure 4), although no drastic increase in lysosome size could be detected, as was the case for Limp2-positive lysosomes. We next performed ultrastructural analyses of glomeruli from MLII (Figure 3A) and MLIII (Figure 3B) mice in comparison with wild-type littermates (Supplemental Figure 5) to assess lysosomal storage accumulation in renal/glomerular cells of the mutant mice. Indeed, in both ML models, electron-lucent storage material was found in numerous lysosomes of podocytes and endothelial cells, whereas mesangial cells contained electron-dense undegraded material in lysosomes. Of note, endothelial cells contained multivesicular bodies filled with lysosomes, which might relate to the large Limp2-positive vesicular structures detected by high-resolution confocal microscopy. Podocyte foot processes and endothelial fenestrations were mostly intact in MLII and MLIII mice, despite the occurrence of focal irregularities and splitting of the glomerular basement membrane and the accumulation of electron-dense and vesicular material within the glomerular basement membrane.

Figure 2.: Lysosomal upregulation and enlargement in glomeruli of MLII and MLIII mice. (A) Lysosomal β-hexosaminidase (β-Hex) activities in isolated glomeruli (nmol/mg per hour) corresponding to 100 µg protein extracts were measured for 60 minutes (mean±SEM, *P≤0.05, Mann–Whitney U test, n≥9). (B and C) Immunoblot analysis for lysosomal (Limp2, Lamp2) and the autophagosome marker LC3 in isolated glomeruli from (B) MLII and (C) MLIII mice in comparison to wild-type (WT) littermates. Graphs exhibit densitometric quantification with pooled values of four independent experiments (n≥10), the densitometric quantification is given for the ratio LC3-II/LC3-I as a measurement of autophagosome formation; values are expressed as mean±SEM (**P≤0.01, Mann–Whitney U test). (D) High-resolution confocal micrographs exhibiting Limp2-positive lysosomes (green) in MLII and MLIII mice in comparison to a wild-type littermate. Glomerular filtration barrier was visualized by staining for the slit membrane protein nephrin (red). White arrows point toward large Limp2-positive lysosomes. ec, glomerular endothelial cell nucleus; mc, mesangial cell nucleus; p, podocyte nucleus; Rel., relative.

Figure 3.: Glomerular cells and the basement membrane of MLII and MLIII mice exhibit lysosomal storage accumulations. Electron-microscopy evaluation in (A) MLII and (B) MLIII mice reveals lysosomal enlargement with storage material (red arrows) of varying electron density depending on the glomerular cell type. Podocytes (PC) and parietal epithelial cells (PEC) contain (A, a and b; B, a) electron-lucent storage vesicles; (A, c; B, c and d) endothelial cells display electron-lucent storage material in lysosomes and prominent multivesicular bodies (MVB) full of lysosomes with electron-dense storage material; (A, e; B, e) mesangial cells exhibit lysosomes filled with electron-dense storage material (white asterisks) in (A, e) MLII and with less dense storage material in (B, e) MLIII. White arrows point toward irregularities, splitting, and electron-dense and vacuolar accumulations within (B, a, b) the glomerular basement membrane (gbm) and (A, b) the Bowman’s capsule. Endothelial fenestrations and podocyte foot processes are mostly intact. BC, Bowman's capsule; bm, basement membrane; c, capillary space; EC, endothelial cell; ERY, erythrocyte; FIB, tubulointerstitial fibroblast with prominent lysosomal accumulations (A,d, white arrows); MC, mesangial cell; PTC, proximal tubular cell; u, urinary space.

Renal and Glomerular Function Is Intact in MLII and MLIII Mice

Glomerular cells of MLII and MLIII mice demonstrate a comparable extent of lysosomal dysfunction. We therefore evaluated whether we could determine alterations of renal/glomerular function in MLIII (Figure 4A) and MLII (Figure 4B) mice with an advanced general phenotype in comparison to wild-type control littermates. Measurement of the serum parameters (Supplemental Figure 6) of renal retention, such as creatinine and BUN, demonstrated an elevation of BUN in the MLII mice, whereas MLIII mice showed no difference to wild-type littermates. Serum electrolytes (sodium, potassium, chloride, and calcium) were normal in MLIII mice, whereas potassium levels were elevated in the sera of MLII mice. Serum markers of nephrotic syndrome (cholesterol, triglycerides, and albumin) were normal in MLIII mice. MLII mice, however, exhibited normal serum lipids and decreased serum albumin. Corroborating the mostly unaltered serum parameters, glomerular, tubular, and tubulointerstitial morphology appeared normal in MLII and MLIII mice, as evaluated by light microscopy of PAS staining (Supplemental Figure 7). Furthermore, staining for specific markers of renal injury such as kidney injury molecule 1 (a highly sensitive marker for tubular injury), smooth muscle actin (a sensitive marker for renal fibrosis), and cleaved caspase 3 (a specific marker for apoptotic cells) did not highlight discrete renal injury (Supplemental Figure 7), which might have been missed in PAS staining. To assess glomerular filtration barrier permeability to protein, the urine albumin-creatinine ratio was determined (Figure 4, A and D), and mouse urine was analyzed by creatinine-adapted Coomassie staining (Figure 4, B and E). However, these parameters remained normal in both mutant mice, indicating no glomerular and/or proximal tubular alterations in MLII and MLIII mice. Accordingly, analysis of nephrin localization in MLII (Figure 4C) and MLIII (Figure 4F) mice by high-resolution microscopy did not show foot process effacement, corroborating the findings of the electron-microscope evaluations. Taken together, these data suggest that glomerular function of both MLII and MLIII mice is not impaired and that the elevated BUN and the decreased serum albumin levels observed in MLII mice were most likely not of renal but of hepatic origin.

Figure 4.: Glomerular filtration barrier of MLII and MLIII mice is normal. Quantification of the urinary albumin-creatinine ratio by ELISA in (A) 60–90 week old MLIII, (D) 30–40 week old MLII, and respective wild-type (WT) littermates (mean±SEM, Mann–Whitney U test, n≥7, three pooled independent experiments). Representative Coomassie blue staining of creatinine-adapted urine from (B) MLIII, (E) MLII, and respective wild-type littermates. High-resolution confocal microscopy of the slit membrane protein nephrin localization in (C) MLIII, (F) MLII, and wild-type littermates. Note the even meandering nephrin pattern demonstrating mostly normal podocyte foot processes.

Patients with MLII but Not MLIII Exhibit Microalbuminuria

Having observed normal glomerular function in MLII and MLIII mice, we next assessed proteinuria as an indicator of glomerular filtration barrier alterations in urinary samples from patients with MLII, MLIII alpha/beta and MLIII gamma in comparison to a healthy control cohort (Figure 5, Supplemental Table 1 indicates individual patient characteristics and urine values). Of note, patients with MLIII alpha/beta harbor mutations in the GNPTAB gene, allowing residual GlcNAc-1-phosphotransferase activity and therefore less severe lysosomal dysfunction, which is comparable to patients with MLIII gamma.⁵ To this end, Coomassie blue staining of creatinine-adapted urine did not show significant loss of high or low mol wt proteins to the urine in patients with MLIII gamma or MLIII gamma. Patients with MLII, however, showed very low levels of albumin loss (Figure 5A). Quantification of the absolute urinary albumin content (Figure 5B) did not reveal significant albuminuria in patients with MLII and MLIII in comparison to healthy controls. However, calculating the albumin-creatinine ratio (Figure 5C) suggested the occurrence of microalbuminuria in patients with MLII. Together, these findings suggest that glomerular cells were capable of compensating for severe lysosomal dysfunction both in mice and in humans, thereby maintaining glomerular function.

Figure 5.: Patients with MLII but not MLIII exhibit microalbuminuria. Urine of a healthy control (Ctrl) cohort (n=9) and patients with MLIII gamma (n=3), MLIII alpha/beta (n=5), and MLII (n=3) was analyzed for (A) low and high mol wt proteinuria by Coomassie blue staining of creatinine-adapted urine and (B) urinary albumin content (mg/L) by ELISA to human albumin. (C) The albumin-creatinine ratio (ACR, in mg/g) was determined after normalization to the respective urinary creatinine content. White arrow indicates albumin band.

Lysosomal Dysfunction Is Differentially Compensated for by the Ubiquitin Proteasome System in MLII and MLIII Mice

The fact that pronounced lysosomal dysfunction of glomerular cells was not associated with any major functional disruption of the glomerular filtration barrier prompted us to search for an explanation as to how glomerular cells of MLII and MLIII mice overcome the proteotoxic stress triggered by nondegraded material. A strong crosstalk between the autophagosomal/lysosomal and ubiquitin proteasomal system (UPS) is evident in situations of impaired podocyte autophagy, where the UPS steps in to compensate.³¹ In this way, abnormal protein accumulations can be prevented, which would otherwise exert proteotoxic stress. Aiming to assess the extent of such compensation by the UPS in MLII and MLIII glomeruli, we performed immunoblot analysis and demonstrated a pronounced upregulation of the proteasomal system in glomeruli of MLIII mice (Figure 6A), but surprisingly not in glomeruli of MLII mice (Figure 6B), despite the more severe general phenotype of MLII mice. In particular, MLIII glomeruli showed elevated levels of both the standard proteasome (β5 subunit) and the stress-induced immunoproteasome (LMP7 subunit). High-resolution confocal microscopy corroborated the immunoblot findings, demonstrating pronounced β5 expression in podocytes of MLIII but not MLII mice (Figure 6C). The chymotrypsin-like activity, which encompasses proteolytic activity mediated by the β5 and the LMP7 subunit of the proteasome, was only slightly reduced in MLII glomeruli (Figure 6D) and maintained at wild-type levels in MLIII glomeruli (Figure 6E), suggesting the elevated proteasome expression was sufficient to compensate for proteasomal impairment in MLIII glomeruli. We also evaluated the expression levels of the deubiquitinating enzyme UCH-L1, which stabilizes cellular monoubiquitin levels,³² and whose de novo expression in podocytes is associated with podocyte injury.³³ As expected, UCH-L1 levels were significantly increased in glomeruli of MLIII mice (Figure 6A). Strikingly, UCH-L1 was strongly downregulated in MLII glomeruli in comparison to wild-type littermates (Figure 6B). Because MLII glomeruli did not show a compensatory proteasomal and UCH-L1 upregulation and displayed a slightly reduced overall proteasomal activity, we expected accumulation of polyubiquitinated proteins in MLII mice, resulting from noncompensated lysosomal dysfunction. Therefore, we performed immunoblot analysis of MLII and MLIII glomeruli to assess the extent of accumulation of polyubiquitinated and K48-linked polyubiquitinated (linkage targets to proteasomal degradation) proteins. However, neither MLII (Figure 6, F and G) nor MLIII (Figure 6, H and I) mice demonstrated a significant accumulation of polyubiquitinated or K48-polyubiquitinated proteins in glomeruli. Rather, polyubiquitin levels were decreased in MLII mice and comparable to wild-type littermates in MLIII mice, whereas K48 polyubiquitin levels remained unaffected in both mutants. Confocal analyses revealed high ubiquitin levels in podocytes in comparison to mesangial and endothelial cells (Supplemental Figure 8), with the rare occurrence of ubiquitin aggregating preferentially in podocytes of MLII mice. Taken together, these data suggest that both MLII and MLIII mice do not accumulate protein up to proteotoxic levels in glomerular cells. Whereas MLIII mice upregulate the UPS system to compensate for lysosomal dysfunction, MLII mice seem to overcome proteostasis unbalance by other compensatory mechanisms.

Figure 6.: Glomerular cells in MLIII but not in MLII mice induce the proteasome system to alleviate proteostatic stress. Isolated glomeruli of (A, H, and I) MLIII, (B, F and G) MLII, and wild-type (WT) littermates were analyzed by immunoblot for the expression levels of (A and B) the main proteolytic β5 subunit of the standard 26S proteasome, the main proteolytic subunit LMP7 of the i26S immunoproteasome, and the deubiquitinating enzyme UCH-L1, for (F and H) polyubiquitin (pUB) levels, and for (G and I) levels of K48-polyubiquitinated proteins (K48 pUB). Graphs exhibit densitometric quantifications of the mean±SEM, Mann–Whitney U test, n≥9, three pooled independent experiments, *P≤0.05. (C) Confocal micrographs showing the β5 proteasomal subunit of the standard proteasome (green) expression in glomerular cells, podocyte slit membrane stained for nephrin (red) and DNA (blue). Note the accentuated expression of β5 in podocytes (p) of MLIII mice. (D and E) Chymotrypsin-like activity (CTL) of the proteasome (assesses the cleavage activity of β5 and LMP7 proteasomal subunits) measured in isolated glomeruli of (D) MLII, (E) MLIII, and wild-type littermates (mean±SEM, Mann–Whitney U test, n≥6, two pooled independent experiments, *P≤0.05). ec, endothelial cells; mc, mesangial cell; Rel., relative; epoxo, epoxomicin.

Pathways Regulating Protein Translation Are Differentially Regulated in MLII and MLIII Mice

The findings that (1) lysosomal deficiency resulted in an upregulation of the UPS in MLIII but not in MLII glomeruli, and (2) MLII glomeruli and kidneys (Supplemental Figure 9, A and B) exhibited decreased ubiquitin levels prompted us to investigate whether proteostasis (i.e., the balance of protein synthesis and protein degradation) was maintained through other unknown mechanisms in MLII mice. We hypothesized that, in MLII kidneys, proteostasis was maintained through a decrease of protein synthesis rather than by an enhancement of proteasomal degradation. The extent of protein translation is adapted to the cellular needs by two major pathways, the mammalian target of rapamycin complex 1 (mTORC1) pathway and the so-called integrated stress response (ISR). Therefore, mTORC1 activity enhances protein translation through the S6Kinase-S6-4E-BP1 pathway, whereas the opposing activation of the ISR results in translational repression.³⁴ Central to this mechanism is the phosphorylation of eIF2α by the endoplasmic reticulum (ER) stress sensor PERK and/or by cytoplasmic stress sensors (such as HRI, PKR, and GCN2) resulting in repression of general protein translation. Analysis of total kidney (Supplemental Figure 9C) and isolated glomeruli suggested that protein translation was dampened in MLII but not MLIII mice (Figure 7). More specifically, MLII mice exhibited a mild repression of the mTORC1-associated transcripts Fasn, Vegfa, and Hif1a in the kidney (Supplemental Figure 9C) and decreased total protein levels of the mTORC1 target S6 in glomeruli (Figure 7A). The latter corroborates the decreased pS6 levels especially in podocytes, as revealed by high-resolution confocal microscopy (Supplemental Figure 10). MLIII mice only exhibited convincing mTORC1 pathway repression at the IF level with pS6 signal being reduced in podocytes (Supplemental Figure 10). The prominent nuclear translocation of the transcription factor TFEB (a strong inductor of lysosomal biogenesis in response to decreased mTORC1 activity) in MLII and MLIII glomerular cells was in conjunction with the decreased p-S6 expression in both mutants (Figure 7B). Quantification of the nuclear mean fluorescent intensity (MFI) in MLIII glomerular cells exhibited an MFI of 220±16 for nuclear TFEB expression in comparison to a nuclear MFI of 105±12 in wild-type littermates (n≥3 mice per group, P<0.001, two-tailed t test). MLII glomerular cells exhibited a nuclear MFI for TFEB of 263±38 in comparison to a nuclear MFI of 71±11 in wild-type littermates (n≥3 mice per group, P≤0.01, two-tailed t test). This suggests that in both MLII and MLIII glomerular cells, repression of mTORC1 results in TFEB translocation to the nucleus, resulting in transcription of autophagosomal and lysosomal genes, which corroborates the observed enhanced protein levels for lysosomal and autophagosomal proteins for both mutants in Figure 2. Predictably, TFEB translocation would result in lysosomal exocytosis, lysosomal biogenesis, and increased autophagic flux, which most likely represent part of a compensatory response to lysosomal insufficiency.

Figure 7.: Glomerular cells of MLII but not of MLIII mice activate the integrated stress response (ISR) to alleviate proteostatic stress. (A) Immunoblot for the mTORC1 target S6 and its phosphorylated form p-S6 in MLII and MLIII mice demonstrates downregulation of total S6 in MLII and MLIII glomeruli, whereas the ratio of p-S6/S6 is not altered. Graphs show densitometric quantification normalized to β-actin of the same membrane, the relative levels to respective wild-type littermate controls are shown (mean±SEM, Mann–Whitney U test, n≥7, *P≤0.05). (B) High-resolution confocal microscopy for the transcription factor (TF) TFEB, which translocates to the nucleus to initiate the transcription of lysosomal genes in the setting of decreased mTORC1 activity. White arrows point toward podocyte (p) and endothelial cells (ec) with nuclear TFEB signal. (C) Activation of the ISR as a pathway to downregulate protein translation was evaluated by qPCR in MLIII and MLII glomeruli. Graph shows the relative gene of interest (GOI) expression to respective wild-type (WT) littermate controls (dashed line); internal controls used were 18S and glyceraldehyde-3-phosphate dehydrogenase (mean±SEM, Mann–Whitney U test, n≥5, *P≤0.05). (D) Immunoblot for the expression and activity (phosphorylation) levels of eIF2α as the core event of the ISR leading to a decrease in global protein synthesis and to the induction of selected genes such as the transcription factor ATF4 to promote cellular recovery in isolated glomeruli of MLIII and MLII mice. Graphs represent densitometric quantification normalized to β-actin of the same membrane, relative levels to respective wild-type littermate controls are shown (mean±SEM, Mann–Whitney U test, n≥7, *P≤0.05). High-resolution confocal microscopy exhibits expression of (E) activated p-eIF2α and its downstream target the transcription factor ATF4, (F) the activated ER stress sensor p-PERK, and (G) the transcription factors NRF1 and activated p-NRF2, which induce the transcription of proteasome genes. Podocyte slit membrane was visualized by staining for nephrin (red), DNA was counterstained with Hoechst (blue). Note the enhanced signal for p-PERK in MLII endothelial cells, for NRF1 and to a lesser extent for p-NRF2 in MLIII podocytes, of p-eIF2α in MLII podocytes and endothelial cells, and the marked ATF4 nuclear signal in MLII glomerular cells (white arrows). Rel., relative.

Besides the repression of the mTORC1 pathway, MLII mice exhibited a strong induction of the ISR in total kidneys (Supplemental Figure 9C) and glomeruli. Transcript levels and phosphorylation of eIF2α was enhanced in MLII but not in MLIII glomeruli by qPCR (Figure 7C), immunoblot (Figure 7D), and IF (Figure 7E). Quantification of the cellular MFI for p-eIF2α in MLIII glomerular cells exhibited am MFI of 88±5 in comparison to an MFI of 104±10 in control littermates (n≥3 mice per group, P=0.13, two-tailed t test). MLII glomerular cells exhibited an MFI for p-eIF2α of 164±18 in comparison to an MFI of 49±3 in control littermates (n≥3 mice per group, P<0.001, two-tailed t test). Phosphorylated eIF2α inhibits conventional protein biosynthesis and promotes the cap-independent translation of various stress proteins including the transcription factor ATF4. In turn, ATF4 decreases mTORC1 activity by inducing the transcription of 4E-BP1³⁵ and induces apoptosis in the nonhomeostatic unfolded protein response (UPR) by enhancing CHOP levels. In conjunction, MLII glomerular cells and kidneys exhibited elevated ATF4 transcript and a marked nuclear translocation of this transcription factor in MLII mice (Figure 7E). Quantification of the MFI in MLIII glomerular cells exhibited an MFI of 476±40 for nuclear ATF4 expression in comparison to an MFI of 398±27 in wild-type littermates (n≥3 mice per group, P=0.09, two-tailed t test). MLII glomerular cells exhibited an MFI of 794±38 for nuclear ATF4 expression in comparison to 246±23 in wild-type littermates (n≥3 mice per group, P<0.001, two-tailed t test). Strikingly, CHOP was transcriptionally elevated only in glomeruli of MLII mice and repressed in MLII kidneys, suggesting lysosomal dysfunction in tubulointerstitial cells of MLII kidneys does not result in pathologic UPR with cell death, corroborating the absence of renal injury/apoptosis marker expression in MLII kidneys (Supplemental Figure 7).

The transcript levels of the eIF2α kinases and cytosolic stress receptors PKR and GCN2 were enhanced in MLII kidneys and glomeruli (Figure 7C, Supplemental Figure 9C). Further, eIF2α phosphorylation is the consequence of the unfolded protein response (UPR) because, upon endoplasmic reticulum (ER) stress, the ER stress receptor PERK phosphorylates eIF2α. MLII glomeruli exhibited an activation of the UPR at the transcriptional and protein level. The ER stress sensors IRE1α and PERK transcripts Ire1a and Perk were upregulated (Figure 7C), and high-resolution confocal microscopy demonstrated an enhanced p-PERK expression in glomerular cells, especially in endothelial cells of MLII mice (Figure 7F). MLIII glomerular cells exhibited comparable MFI for p-PERK in comparison to control littermates (MLIII, 65±5; wild type, 66±5; n≥3 mice per group, P=0.9, two-tailed t test). MLII glomerular cells exhibited enhanced cellular MFI for p-PERK in comparison to control littermates (MLII, 304±23; wild type, 41±3; n≥3 mice per group, P≤0.01, two-tailed t test). Taken together, our data indicate that a combination of the cytosolic stress– and the UPR-dependent pathway is responsible for eIF2α phosphorylation and for ATF4 induction in MLII kidneys. Given that MLII glomeruli balance their proteostasis by repression of protein translation through decreasing the activity of the mTORC1 pathway and by induction of the ISR rather than the UPS, we validated this finding by assessing transcript levels and the nuclear translocation of the transcription factors Nrf1 and Nrf2 in glomerular cells. Both Nrf1 and Nrf2 induce the transcription of UPS genes and were most prominently located in the nucleus of glomerular cells in MLIII mice and less in MLII mice (Figure 7G). Quantification of nuclear MFI demonstrated enhanced NRF1 intensities in MLIII nuclei (MLIII MFI, 334±32; wild type, 223±20; n≥3 mice per group, P≤0.01, two-tailed t test) and to a similar extent in MLII nuclei (MLII MFI, 375±25; wild type, 175±13; n≥3 mice per group, P≤0.01, two-tailed t test). Comparable enhanced expression levels were observed for p-NRF2 in MLIII nuclei (MLIII MFI of p-NRF2, 211±19; wild type, 127±10; n≥3 mice per group, P≤0.002, two-tailed t test), which were not apparent in MLII nuclei (MLII MFI, 230±19; wild type, 212±15; n≥3 mice per group, P=0.47, two-tailed t test). Together, these nuclear levels of both NRF1 and NRF2 corroborate the strong upregulation of the proteasome system in MLIII but not MLII kidneys as observed at the protein level (Figure 6).

Discussion

In this study, we demonstrate that mice with lysosomal dysfunction in glomerular cells fail to develop a glomerular cell injury. We suggest (Figure 8) that in moderate lysosomal dysfunction the glomerular cell proteostasis is mainly maintained by (1) upregulation of the UPS, and (2) by suppression of mTORC1 signaling; whereas in severe lysosomal dysfunction the glomerular proteostasis is mainly balanced by general suppression of protein translation through (1) suppression of mTORC1 signaling, and (2) through activation of the ISR. Interestingly the extent of pathway activation varied among the glomerular cell types, pointing toward a cell-specific adaptation to lysosomal deficiency in glomeruli.

Figure 8.: Proposed compensatory mechanisms activated in MLII and MLIII glomerular cells to maintain proteostasis. (A) The Golgi-resident GlcNAc-1-phosphotransferase conjugates M6P residues to lysosomal enzymes, which then results in proper sorting of lysosomal enzymes to lysosomes after binding to M6P receptors. (B) In MLIII γ, the γ subunit of GlcNAc-1-phosphotransferase is mutated, resulting in a partial missorting of lysosomal enzymes outside the cell and to mild lysosomal dysfunction. In MLIII γ glomerular cell proteostasis (the balance between protein synthesis and degradation) is maintained (1) by upregulation of the ubiquitin proteasome system (UPS) by Nrf1/Nrf2 activation, and (2) by suppression of mTORC1 signaling, resulting in decreased protein translation through the S6Kinase-S6 pathway. (C) In MLII α/β, the α and β subunits of the GlcNAc-1-phosphotransferase are mutated, resulting in a severe missorting of lysosomal enzymes outside the cell and to severe lysosomal dysfunction. In MLII α/β glomerular cells, proteostasis is mainly maintained by suppression of protein translation by (1) suppression of mTORC1 signaling, and (2) activation of the integrated stress response (ISR) by the endoplasmic reticulum (ER) stress sensors and by cytosolic stress receptors. UPS transcripts and proteins are downregulated as a result of the “generally” suppressed protein translation. Ub, ubiquitin.

Renal involvement is rare in lysosomal storage diseases, and only sporadic cases have been reported for patients with MLII or MLIII, giving rise to the question of whether the predominant glomerular involvement observed in a few patients was a pure coincidence or a direct consequence of the perturbed glomerular cell protein homeostasis due to the GNPTAB or GNPTG mutations. This open question is attributable to the fact that, although rapid progress has been made in understanding the genetic basis of lysosomal storage disorders, the factors underlying the disruption of cell metabolic and signaling pathways associated with these conditions are less well understood. It is commonly accepted that, despite the broad variability of lysosomal storage diseases, the storage of similar types of macromolecules (sphingolipids, mucopolysaccharides, or glycoproteins) often results in similar biochemical, pathologic, and clinical phenotypes, suggesting the nature and distribution of the stored material is the major lesion-defining factor.³⁶ In line with this, the electron density of the accumulated storage material differed between podocytes/endothelial cells and mesangial cells in MLII and MLIII mice in our study, suggesting lysosomal dysfunction resulted in the accumulation of distinct types of macromolecules in a glomerular cell-specific manner. Until now, the biochemical nature of the accumulated lysosomal storage material within glomerular cells of MLII and MLIII mice or patients has not been identified and was not the scope of this study. In skeletal cells from MLII mice, mainly electron-lucent material resulting from, e.g., defective glycosaminoglycan degradation was detectable, whereas electron-dense lipid material was found in MLII brains^19,29 and in MLII and MLIII glomerular mesangial cells (this study). However, given the plethora of missorted lysosomal enzymes, we suspect the presence of a heterogenous mixture of abnormal storage material within the different glomerular cells.

Gnptg and Gnptab genes are ubiquitously expressed in mice. However, in the kidney, relative transcript levels for Gnptg are higher than for Gnptab.¹⁸ Whether this has a functional consequence in the kidney is unknown. In our investigations, the extent and site of renal cell lysosomal alteration appeared comparable between mice with Gnptg (MLIII) and Gnptab (MLII) mutations. In both models, Limp2-positive lysosomes were found to be more abundant and enlarged, predominantly in renal tubulointerstitial cells, podocytes, endothelial cells, mesangial cells, and juxtaglomerular cells. However, Lamp2-positive lysosomes, which are by number most prominent in proximal tubular cells in the healthy mouse, did not appear enlarged.

Besides the biochemical nature of the abnormally stored material in lysosomes, the compensatory potential of the affected cell types to deal with this storage stress could define/shape the characteristics of the clinical presentation. Of all glomerular cells, podocytes have most frequently been observed to exhibit an altered morphology in lysosomal storage disorders, usually described as having a foamy appearance,^7,8,15 most likely due to the accumulation of storage material. However, this foamy podocyte appearance does not necessarily associate with a decreased renal/glomerular function in patients with ML.⁸ This differs from patients with Fabry disease, in which renal involvement is frequent and observed in several renal cell types, including podocytes, mesangial and interstitial cells, vascular endothelial and smooth muscle cells, as well as the tubular cells of the proximal and distal tubules and loop of Henle.¹ Despite our thorough evaluation of the glomerular phenotype in MLII and MLIII mice, no significant signs of altered glomerular function were discernable, although glomerular cells exhibited enlarged lysosomes with storage material. Besides the focal glomerular basement alterations, podocyte foot processes and endothelial cell fenestrations were inconspicuous, relating to the absence of proteinuria. Similarly, patients with MLII, MLIII alpha/beta, and MLIII gamma did not exhibit gross protein loss to their urine, only microalbuminuria was observed in patients with MLII. Therefore, we suggest the glomerular cell protein homeostasis was most likely not sufficiently perturbed by the lysosomal storage disorder to result in a disease promoting loss of glomerular filtration barrier integrity. Although the direct comparison of the murine MLII/MLIII models with the situation in patients with MLII/MLIII is limited due to possible species-based differences in glomerular cell metabolism,^37,38 we are confident enough to conclude that the absence of a significant glomerular function impairment in MLII and MLIII mice relates to the human disease. This conclusion is based on the fact that several lysosomal enzymes are missorted to the extracellular space in MLII and MLIII mice as well as in patients with MLII and MLIII. However, it remains unresolved as to why in Fabry disease (which is the result of low or deficient enzyme levels of one single lysosomal enzyme, i.e., α-galactosidase A) and not in patients with MLII and MLIII, glomerular cell involvement is prominent and leads to Fabry nephropathy with podocyturia,³⁹ proteinuria, and decreased GFR.

The autophagosomal/lysosomal system and the UPS strongly crosstalk, thus mutually compensate for the impairment of one another.⁴⁰ In many cases, lysosomal storage disorders are accompanied by the appearance of ubiquitinated protein aggregates, a finding frequently made in neurons.⁴¹ Further, podocytes deficient for ATG5, which is essential for autophagosome formation, compensate for autophagosome deficiency by upregulation of the UPS.³¹ Only upon impairment of the UPS at 28–52 weeks of age does ubiquitin accumulate and does mild proteinuria occur in ATG5-deficient podocytes.³¹ We therefore carefully assessed the status of the UPS in glomeruli of MLII and MLIII mice. In MLII mice, the latest age assessable was 40 weeks due to their severe general ML phenotype. MLIII mice, which presented a milder general phenotype, were evaluated for the development of a renal phenotype until 90 weeks of age. To our surprise, only MLIII mice exhibited a compensatory upregulation of the UPS. In MLIII glomerular cells, the extent of UPS upregulation was most likely sufficient to preserve protein homeostasis in the setting of lysosomal dysfunction, because glomerular ubiquitin levels and proteasomal activity were comparable to control littermates. In contrast, MLII mice failed to upregulate the UPS at the protein level in glomeruli, an astonishing difference given the fact that the same cellular degradative mechanism was affected in MLII and MLIII mice. Rather, ubiquitin levels were decreased on the transcriptional and more clearly on the protein level in MLII glomeruli and kidneys. Whereas abnormal accumulation of ubiquitin was prevented in MLII mice, the nuclear translocation of the UPS gene activating transcription factors Nrf1 and Nrf2 was less apparent in MLII compared with MLIII mice but still above wild-type littermate levels, suggesting a global downstream compensatory mechanism (distinct from the UPS pathway) was activated to balance protein homeostasis in the setting of severe lysosomal dysfunction.

We also evaluated the expression levels of the deubiquitinating enzyme UCH-L1, which stabilizes cellular monoubiquitin levels³² and whose de novo expression in podocytes is associated with podocyte injury.³³ Only MLIII mice exhibited enhanced levels of UCH-L1, mirroring the induction of proteasome subunits to maintain proteostasis. In MLII glomeruli, however, UCH-L1 levels were strongly repressed at the protein level, whereas proteasome subunits were at levels comparable to control littermates, suggesting a distinct involvement of UCH-L1 in these lysosomal storage disorders. In line with this, the UCH-L1 repression observed in MLII glomeruli has already been reported in fibroblasts of patients with lysosomal storage disorders, namely sialidosis, sialic acid storage disease, galactosialidosis, GM1 gangliosidosis, Morquio disease type A and B, and Gaucher disease.³⁶ There, it has been shown that reestablishment of UCH-L1 expression protected patient fibroblasts from apoptosis,³⁶ however, no apoptosis per se could be detected in either MLII or MLIII glomeruli. On the other hand, further evidence links UCH-L1 to autophagosomal/lysosomal function, because a mutated form of the enzyme deficient for hydrolase activity was shown to inhibit macroautophagy by associating with Lamp2.⁴² Furthermore, UCH-L1 has also been suggested to destabilize mTORC1,⁴³ which is a sensor for dysfunctional lysosomes.⁴⁴ mTORC1 forms a lysosome-nucleus signaling axis involving the transcription factor TFEB. In the absence of lysosomal stress, mTORC1 is recruited to the lysosomal surface, where it becomes activated and phosphorylates TFEB to prevent its translocation to the nucleus. In situations of lysosomal stress, mTORC1 detaches from the lysosome and remains in the cytosol in an inactive state, whereas TFEB is no longer phosphorylated. This results in TFEB translocation to the nucleus, where it activates gene expression programs that boost lysosomal function and autophagy. MLII and, to a lesser extent, MLIII kidneys exhibited signs of mTORC1 pathway repression in the form of (1) decreased levels of total and phosphorylated glomerular S6; (2) decreased renal levels of the mTORC1-related transcripts Fasn, Vegfa, and Hif1a; and (3) enhanced nuclear translocation of TFEB resulting in elevated levels of lysosomal and autophagosomal proteins such as Lamp2, Limp2, and LC3. A protective effect of mTORC1 downregulation in MLII and MLIII glomerular cells could arise from enhanced autophagy, because LC3 levels were increased in both mutants. However, autophagy is also dysfunctional in situations of lysosomal impairment because degradation of autophagosomal content requires the transfer of lysosomal enzymes to autophagosomes from intact lysosomes. This process is dysfunctional in MLII and MLIII mice due to the missorting of lysosomal enzymes such that the degradation of autophagosomal content is impaired, resulting in accumulation of p62 (Supplemental Figure 11) as a sign of impaired autophagosomal flux, suggesting that enhanced autophagosomal flux cannot compensate for lysosomal deficiency in MLIII and MLII mice. Therefore, we hypothesize that the reduced activation of the mTORC1-S6Kinase-S6-4E-BP1 pathway, which ultimately represses protein synthesis, is the most likely contributor to balancing proteostasis in situations of lysosomal dysfunction. Accordingly, downregulation of mTORC1 was more evident in MLII (decreased levels of UPS-related transcripts and of ubiquitinated proteins) than in MLIII kidneys. Additionally, our investigations demonstrated a significant repression of protein synthesis in MLII kidneys through activation of the ISR, which was not activated in MLIII kidneys. Finally, UCH-L1 has recently been shown to bypass mTORC1 to directly promote translation initiation by an unknown mechanism.⁴⁵ Taken together with this finding, the strongly repressed UCH-L1 could represent a yet undefined pathway that lessens proteotoxic stress in MLII kidneys through decreasing protein synthesis independently of mTORC1 signaling.

In summary (Figure 8), our study suggests that, despite a comparable level of lysosomal dysfunction between MLII and MLIII kidneys, no impairment of glomerular function ensues due to compensatory mechanisms, which balance proteostasis. Importantly, the compensatory mechanisms activated within glomerular cells differ significantly between the two mouse models. MLII glomerular cells rebalance proteostasis by decreasing protein synthesis through upregulation of the ISR, whereas MLIII glomerular cells balance proteostasis by enhancing the degradative capacity of the proteasome. Further studies are necessary to understand the equilibrium homeostasis and crosstalk of the intricate mechanisms ensuring proteostasis in renal cells.

Disclosures

All authors have nothing to disclose.

Funding

This work was funded by Deutsche Forschungsgemeinschaft (DFG; German Research Foundation) grants CRC 877 project B3 (to T. Danyukova and S. Pohl), PO 1539/1-1 (to S. Pohl), and CRC 1192 project B3 and ME 2108/10-1 (to C. Meyer-Schwesinger); and by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (Brazilian National Council for Scientific and Technological Development; to N. Ludwig). T.B. Huber was supported by DFG grants CRC 1192 and HU 1016/8-2, Bundesministerium für Bildung und Forschung (Federal Ministry of Education and Research) grant STOP-FSGS 01GM1901C, and European Research Council grant 61689, DNCure.