Impaired β-glucocerebrosidase activity and processing in frontotemporal dementia due to progranulin mutations

Abstract

Loss-of-function mutations in progranulin (GRN) are a major autosomal dominant cause of frontotemporal dementia. Most pathogenic GRN mutations result in progranulin haploinsufficiency, which is thought to cause frontotemporal dementia in GRN mutation carriers. Progranulin haploinsufficiency may drive frontotemporal dementia pathogenesis by disrupting lysosomal function, as patients with GRN mutations on both alleles develop the lysosomal storage disorder neuronal ceroid lipofuscinosis, and frontotemporal dementia patients with GRN mutations (FTD-GRN) also accumulate lipofuscin. The specific lysosomal deficits caused by progranulin insufficiency remain unclear, but emerging data indicate that progranulin insufficiency may impair lysosomal sphingolipid-metabolizing enzymes. We investigated the effects of progranulin insufficiency on sphingolipid-metabolizing enzymes in the inferior frontal gyrus of FTD-GRN patients using fluorogenic activity assays, biochemical profiling of enzyme levels and posttranslational modifications, and quantitative neuropathology. Of the enzymes studied, only β-glucocerebrosidase exhibited impairment in FTD-GRN patients. Brains from FTD-GRN patients had lower activity than controls, which was associated with lower levels of mature β-glucocerebrosidase protein and accumulation of insoluble, incompletely glycosylated β-glucocerebrosidase. Immunostaining revealed loss of neuronal β-glucocerebrosidase in FTD-GRN patients. To investigate the effects of progranulin insufficiency on β-glucocerebrosidase outside of the context of neurodegeneration, we investigated β-glucocerebrosidase activity in progranulin-insufficient mice. Brains from Grn^-/- mice had lower β-glucocerebrosidase activity than wild-type littermates, which was corrected by AAV-progranulin gene therapy. These data show that progranulin insufficiency impairs β-glucocerebrosidase activity in the brain. This effect is strongest in neurons and may be caused by impaired β-glucocerebrosidase processing.

Keywords: Frontotemporal dementia; Glycosphingolipid; Lysosome; Neuronal Ceroid Lipofuscinosis; Progranulin; β-Glucocerebrosidase.

Fig. 1

Deficits in GCase Activity in Inferior Frontal Gyrus from Patients with FTD-GRN. Lysates from inferior frontal gyrus of controls and FTD-GRN patients were analyzed for activity of sphingolipid-metabolizing enzymes. a, Simplified diagram of metabolism of gangliosides, with a summary of observed phenotypes in FTD-GRN cases. Lipids are shown in black, with the enzymes that metabolize each lipid shown in blue. GM1, 2, and 3 = GM1, 2, and 3 ganglioside, β-Gal = β-galactosidase, β-Hex = β-Hexosaminidase, HexA = β-Hexosaminidase A, GLA = α-galactosidase A, GCase = β-glucocerebrosidase, Neu = neuraminidase (activity not measured in this study). b–f, Enzymatic activity in FTD-GRN patient brains as measured by fluorogenic assays for b, β-Gal, c, β-Hex, d, HexA, which was increased (* = t test, p = 0.0474), e, GLA, and f, GCase, which was decreased (** = t test, p = 0.0015). n = 5 controls, 7 FTD-GRN

Fig. 2

Lower Levels of Mature GCase Protein and Accumulation of Incompletely Glycosylated GCase in Brains from FTD-GRN Patients. a–c, Lysates of inferior frontal gyrus from FTD-GRN cases had lower levels of normal molecular-weight GCase protein than controls as detected by a rabbit polyclonal anti-GCase antibody (b, t test, p = 0.0059), and a mouse monoclonal anti-GCase antibody (c, t test, p = 0.0017). The polyclonal antibody also detected a lower molecular weight band that was present in all but one FTD-GRN case, but rarely in controls. d, To determine whether this lower molecular weight band could represent incompletely glycosylated GCase, lysates were treated with Endo H or PNGase F. The lower-molecular weight GCase band (red arrowhead) was approximately the same molecular weight as Endo H-treated GCase. Complete removal of glycosylation with PNGase F collapsed the multiple GCase bands into one band just over 50 kDa, indicating that the low-molecular weight form of GCase detected by the polyclonal antibody may consist of incompletely glycosylated GCase. The higher molecular weight band detected by the mouse monoclonal antibody was determined to be non-specific because it did not shift with PNGase F treatment (data not shown). e, Unlike normal molecular weight GCase, there were no group differences in deglycosylated (PNGase F-treated) GCase (t test, p = 0.9594), showing that while FTD-GRN brains contain less normal molecular weight, active GCase than controls, they have similar total levels of GCase protein. f, The low-molecular weight form of GCase accumulated in the sarkosyl-insoluble fraction of FTD-GRN brains, while normal molecular weight GCase was only detectable in the soluble fractions. Levels of sarkosyl-insoluble GCase did not statistically differ between controls and FTD-GRN patients (g, Mann-Whitney test, p = 0.1490), but the insoluble band was detectable in 6/7 FTD-GRN patients, but only reliably detectable in 1/5 controls. This stands in contrast to α-tubulin (f, h), which did not differ between groups. Levels of insoluble GCase were very high in some FTD-GRN patients, so data in g are shown in log scale. n = 5 controls, and 7 FTD-GRN. * = p < 0.05 and ** = p < 0.1 by Tukey’s post-hoc test. GCase = β-glucocerebrosidase, Endo = Endoglycosidase H, PNG = PNGase F

Fig. 3

Lower GCase Immunolabeling in FTD-GRN. We performed GCase immunostaining with the rabbit polyclonal antibody used in the preceding biochemical experiments and a second mouse monoclonal antibody. a, using homozygous GBA L444P fibroblasts, we confirmed that the rabbit polyclonal antibody detected an immature form of GCase (also see Additional file 2: Figure S2), but the mouse monoclonal antibody was selective for mature, fully glycosylated GCase. We analyzed GCase immunolabeling in each group in layer III, which contained many strongly-labeled pyramidal neurons in controls. b, c, FTD-GRN patients did not exhibit significantly less GCase immunolabeling with the rabbit polyclonal antibody (t test, p = 0.2042). d, e, In contrast, FTD-GRN patients had significantly less GCase immunolabeling with the mouse monoclonal antibody selective for mature GCase (t test, p = 0.0285). Low power images of GCase immunoreactivity throughout the cortex are shown in b and d, with 100 μm scale bars, with representative 20x images of GCase immunostaining below with a 20 μm scale bars. The numbers for each image reference the patients described in Table 1

Fig. 4

Loss of Neuronal GCase in FTD-GRN. a, GCase immunostained sections were counterstained with hematoxylin to identify neurons, which revealed both fewer neurons (t test, p = 0.0437) and less neuronal area (b, t test, p = 0.0458) in layer III of FTD-GRN patients than controls, consistent with neurodegeneration. c, Within these hematoxylin-labeled neurons, FTD-GRN patients had less GCase immunolabeling than controls (t test, p = 0.031). This observation survived correction for the lower total neuronal area (d, t test, p = 0.048), indicating that the reduced GCase in FTD-GRN brains is unlikely to be explained solely by neuronal loss. n = 5 controls and 7 FTD-GRN patients. e, The intensity of fluorescent GCase immunolabeling in cortical neurons was also reduced in FTD-GRN patients, with quantification of neuronal GCase intensity shown in f, cumulative probability plots and g, density plots. These data were analyzed first by Kolmogorov-Smirnov test, showing significantly lower neuronal GCase in FTD-GRN (p < 0.0001). A more conservative mixed-model regression of the fluorescent intensity distribution data confirmed that the distribution of FTD-GRN neurons was skewed toward less intense labeling than controls, with significant group differences found in the 60th (p = 0.007), 70th (p = 0.001), 80th (p = 0.005) and 90th (p = 0.022) deciles. n = 519 neurons from 5 controls and 262 neurons from 6 FTD-GRN patients. All images are labeled with the corresponding case number from Table 1. Representative 40X images of GCase and NeuN immunostaining are shown in e with a 20 μm scale bar

Fig. 5

Brains from FTD-GRN Patients Do Not Accumulate Glucosylceramide or Glucosylsphingosine. Lipids were extracted from lysates of inferior frontal gyrus of control and FTD-GRN patients and analyzed by high performance liquid chromatography/mass spectrometry. No group differences were detected in glucosylceramide isoforms (a, MANOVA effect of group, p = 0.954), total levels of glucosylceramide (b, t test, p = 0.8470), or glucosylsphingosine (c, t test, p = 0.4199). n = 5 controls and 5 FTD-GRN. GlcCer = glucosylceramide, GlcSph = glucosylsphingosine

Fig. 6

Progranulin Interacts with GCase. HEK-293 cells were co-transfected with constructs expressing HA-tagged human progranulin and/or myc-flag-tagged human GCase. HA-tagged progranulin was then immunoprecipitated from cell lysates with an anti-HA antibody. a, Flag-tagged GCase co-immunoprecipitated with progranulin, indicating interaction of the two proteins. b, Consistent with the co-immunoprecipitation of GCase with progranulin, we detected strong proximity ligation (PLA) signal in HEK-293 cells co-transfected with human progranulin and human GCase constructs. c, The specificity of the Flag-HA PLA signal was confirmed by the presence of significantly more PLA puncta from cells co-transfected with the progranulin-HA and GCase-Flag constructs than in cells transfected with only one of the constructs, or from cells that underwent PLA in the absence of the HA and Flag antibodies (ANOVA effect of experimental condition, p < 0.0001, *** = p < 0.001 and **** = p < 0.0001 by Dunnett’s post-hoc test). The scale bars represents 5 μm. GCase = β-glucocerebrosidase, GRN = progranulin, PLA = proximity ligation assay

Fig. 7

Grn^−/− Mice Have GCase Activity Deficits That Are Corrected by Restoration of Progranulin with an AAV Vector. Frontal cortical lysates from 7 to 10 month-old wild-type, Grn^+/−_, and Grn^−/− littermates were analyzed for activity of GSL-metabolizing enzymes. A simplified diagram of ganglioside metabolism, with a summary of observed phenotypes in Grn^−/− mice is shown in a. Lipids are shown in black, with the enzymes that metabolize each lipid shown in blue. GM1, 2, and 3 = GM1, 2, and 3 ganglioside, β-Gal = β-galactosidase, β-Hex = β-Hexosaminidase, HexA = β-Hexosaminidase A, GLA = α-galactosidase A, GCase = β-glucocerebrosidase, Neu = neuraminidase (activity not measured in this study). b–f, At this age, Grn^−/− mice exhibited significant increases in activity of β-Gal (b, ANOVA, p = 0.0024), β-Hex (c, ANOVA, p < 0.0001), HexA (d, ANOVA, p < 0.0001), and GLA (e, ANOVA, p < 0.0001). However, Grn^−/− mice exhibited a decrease in GCase activity (f, ANOVA, p = 0.0099). n = 6–24 per genotype. * = p < 0.05 ** = p < 0.1, **** = p < 0.0001 by Dunnett’s post-hoc test. g, Analysis of ventral striatum samples from a cohort of Grn^−/− mice treated with AAV-GFP or AAV-progranulin in the medial prefrontal cortex revealed that restoring progranulin to Grn^−/− mice normalized GCase deficits (ANOVA genotype x treatment interaction, p = 0.0122), but that AAV-Grn reduced GCase activity in wild-type mice. h, Analysis of multiple brain regions from a second group of AAV-treated mice produced similar results, (RM ANOVA, genotype x virus x brain region, p = 0.042). Subsequent tests of the effects of virus within genotype confirmed the opposite effects of AAV-Grn in each genotype. In wild-type mice, AAV-Grn reduced GCase activity (RM ANOVA effect of virus, p = 0.0088), but in Grn^−/− mice, AAV-Grn increased GCase activity to near wild-type levels (RM ANOVA effect of virus, p = 0.0072). * = p < 0.05 and **** = p < 0.0001 by Sidak’s post-hoc test. n = 8–11 mice per group in g, with 4 uninjected Grn^−/− mice included as a reference, and 4–6 mice per group in h. β-Gal = β-galactosidase, β-Hex = β-Hexosaminidase, HexA = β-Hexosaminidase A, GLA = α-galactosidase A, GCase = β-glucocerebrosidase, Neu = neuraminidase (activity not measured in this study)