Trehalose upregulates progranulin expression in human and mouse models of GRN haploinsufficiency: a novel therapeutic lead to treat frontotemporal dementia

Bafilomycin A1 (BafA1), PP242, and Torin1 were obtained from Tocris (R&D Systems, Minneapolis, MN). Chloroquine diphosphate and Actinomycin D (ActD) were obtained from Sigma (St. Louis, MO). Suberanilohydroxamic acid (SAHA or vorinostat) and rapamycin were obtained from LC Laboratories (Woburn, MA). Trehalose (dihydrate) was obtained from Sigma or Brooklyn Premium (Brooklyn, NY). BafA1, PP242, Torin1, rapamycin, and SAHA were dissolved in DMSO and stocks were frozen at −20 °C. Chloroquine and trehalose were dissolved in ultrapure Milli-Q water (EMD Millipore) and frozen at −20 °C or filtered (0.22-μm) and stored at 4 °C, respectively. Trehalose stocks were made up fresh as needed.

Human embryonic kidney cells (HEK293T; American Type Culture Collection), human HeLa cells (American Type Culture Collection), human neuroglioma cells (H4; American Type Culture Collection), and mouse neuroblastoma cells (N2a; American Type Culture Collection) were cultured in high glucose Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10 % FBS, 1 % penicillin/streptomycin, and 1 % Gluta-max. Human neuroblastoma cells (SH-SY5Y; American Type Culture Collection) were cultured in DMEM/Ham’s F12 1:1 medium supplemented with 10 % FBS and 1 % penicillin/streptomycin. The TFEB-GFP stable HeLa cell line [12] was a kind gift provided by Dr. Shawn Ferguson (Yale University) and was cultured in high glucose Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10 % FBS, 1 % penicillin/streptomycin, and 1 % Gluta-max. The HAP1 human haploid cell lines were purchased from Horizon Discovery Group. The TFEB knock-out cell line was produced by CRISPR/Cas9 gene editing which introduced a frameshift mutation into the coding sequence (7 bp deletion in exon 2) of TFEB. HAP1 cells were cultured in Iscove’s Modified Dulbecco’s Medium (IMDM) supplemented with 10 % FBS and 1 % penicillin/streptomycin. Primary mouse cortical neurons (embryonic day 18) were a generous gift from Dr. Randy Hall (Emory University). All cell lines were maintained at 37 °C with 5 % CO₂.

Human dermal fibroblasts were collected under protocol 00064365 as approved by the Emory University Institutional Review Board. Informed written consent was obtained for all research subjects. A 2-mm dermal punch biopsy was taken from each subject under sterile conditions. The biopsy sample was placed in culture medium (DMEM with 4.5 mg/mL glucose, L-glutamine, and sodium pyruvate) with or without 0.6 % v/v fungizone (Gibco). Culture media was supplemented with 10 % FBS (Atlanta Biologicals) and 1 % penicillin/streptomycin (Gibco) and filtered through a 0.2-μm syringe filter prior to use. In a culture hood, skin biopsies were rinsed 3× with sterile PBS and the skin biopsy was then transferred to a single well of a 6-well culture dish in ~50-100 μL PBS or culture medium. The tissue was cut into several smaller pieces using a sterile rounded-tip scalpel blade. A sterile glass coverslip (25 mm) was placed on top of the skin pieces and pressed down onto the dish to secure the sample. Carefully, 3 mL of culture medium was added to the well and the plate was placed in an incubator at 37 °C with 5 % CO2. The cultures were allowed to sit undisturbed for 6–7 days and then half of the media was removed and replaced with fresh media. Fibroblasts typically emerged by day 10 and were ready to passage by day 12–16. Fibroblast cultures were passaged as needed using 0.25 % trypsin with EDTA (Invitrogen) and gradually scaled up to larger culture dishes in culture medium without fungizone. Individual cell plugs from each line were then cryopreserved at low passage numbers, typically at P < 5. Fibroblast lines were generated from 3 patients with the following GRN mutations: c.1477C > T (R493X) (designated GRN #1, GRN #2) and c.592_593delAG (R198GfsX19) (designated GRN #3).

One control and one GRN (GRN #3) fibroblast line were reprogrammed into iPSCs using the Cytotune 2.0 Kit (Life Technologies) per the manufacturer’s protocol. In brief, early passage fibroblast (P < 10) were grown to approximately 50–80 % confluency in fibroblast medium consisting of 10 % ES-qualified FBS (Life Technologies, 0.1 mM NEAA, 55 μM β-mercaptoethanol, high glucose DMEM (Life Technologies). On Day 0, fibroblasts were transduced with three Sendai viruses encoding Klf4–Oct3/4–Sox2 (KOS), hc-Myc, and hKlf4, each at an MOI of 5). Cells were fed with fibroblast medium every other day for 7 days. On day 7, cells were passaged onto vitronectin (Life Technologies) coated dishes at a density of 2.5 × 10⁵ - 5.0 × 10⁵ cells/well. Beginning on day 8, cells were fed every day in Essential 8 medium (Life Technologies). Resulting iPS colonies were manually picked and transferred to a dish coated with either vitronectin or Matrigel (BD). iPSCs were maintained on Matrigel coated dishes and mTesR1 medium (Stem Cell Technologies) and passaged every 5–7 days.

Prior to differentiation, iPSC colonies were treated with 10 μM ROCK inhibitor, Y-27632 (Stem Cell Technologies), for ~1 h. iPSCs were then treated with Accutase (Stem Cell Technologies) for ~8 min to obtain a single cell suspension. Cells were spun out of Accutase and resuspended in N2B27 differentiation medium (1:1 Advanced DMEM-F12/Neurobasal, 1× N2, 1× B27, 0.2 % Penstrep, 1× Glutamax, 110 μM β-mercaptoethanol) and seeded in 10 cm Ultra-Low Attachment dishes (Corning) in order to form embryoid bodies. Cells were maintained as embryoid bodies throughout the differentiation and were fed every 2 days. For the first 2 days, the differentiation medium contained a final concentration of 3 μM CHIR99021 (Stem Cell Technologies), 10 μM SB431542 (Stem Cell Technologies), 10 μM LDN193189 (Stemgent), 0.4 μg/mL Ascorbic Acid (Sigma), 10 μM Y-27632. On Day 2, 1 μM Retinoic Acid (Sigma) and 500 nM Smoothened Agonist (Millipore) were added to the medium. On Day 4, CHIR99021 was removed from the medium. On Day 8, SB and LDN were removed from the medium and the following supplements were added: 10 ng/mL BDNF (Peprotech), 10 ng/mL GDNF (Peprotech), and 10 μM DAPT (Tocris). On day 18, embryoid bodies were disassociated to single cells using papain/DNase (Worhtington Bio) and plated on polyornithine/laminin coated glass coverslips or cell culture plates. Neurons were typically used for experiments around 1-week post-plating.

HEK293T cells were plated in a 96-well plate at approximately 1.5 × 10⁴ cells per well in complete medium. The following day, cells were transfected with 100 ng of the GLuc-ON human PGRN promoter (Gene Accession: NM_001012479; promoter length: 1253 bp; sequence verified) dual reporter (GLuc/SEAP) plasmid (Genecopoeia) using Mirrus LT1 transfection reagent. 24 h after transfection, a commercially available library of ~100 reported chemical enhancers or inhibitors of the autophagy pathway (Enzo Screen-Well library; BML-2837) were applied to the cells overnight for 16 h. Final concentration for all drugs were 1 μM except for bafilomycin A1 (50 nM), chloroquine (50 μM), and trehalose (100 mM) based on previously published reports for increasing progranulin expression or activating autophagy (see main text). After treatment, the media was collected and transferred to a new 96-well plate, spun briefly to remove cell debris, and 10 μL of each sample was transferred to a new 96-well black plate. Secreted GLuc activity normalized to SEAP was measured for each compound using the SecretePair Dual Luminescence Assay Kit (GeneCopoeia) on a BioTek Synergy plate reader and compared to mock-treated (DMSO) cells. The screen was performed in two independent experiments and the average fold-increase compared to mock treatment is reported. Staurosporine (1 μM) was used as a control for overt cell toxicity and compounds whose individual GLuc or seAP values were below those of staurosporine were omitted.

Cells were rinsed 2× in PBS and lysed in ice-cold RIPA buffer (50 mM Tris–HCl, pH 8.0, 150 mM NaCl, 1 % Triton-×100, 0.1 % SDS, 0.5 % sodium deoxycholate) in the presence of protease and phosphatase inhibitor cocktail (PPIC, Pierce). RIPA lysates were sonicated at 20 % amplitude for 5 cycles of 2 s on/2 s off on ice using a sonic dismembrator (QSonica, LLC; Newton, Ct). In some experiments cells were first lysed in CYTO buffer (50 mM Tris–HCl, pH 8.0, 150 mM NaCl, 0.5 % Triton X-100 + PPIC) on ice for 10 min followed by centrifugation at max speed for 10 min at 4 °C. The resultant supernatant contained cytoplasmic and membrane proteins. The pellet was rinsed with CYTO buffer, re-centrifuged for 2 min and the supernatant was discarded. The pellet was then resuspended in NUC buffer (RIPA; 50 mM Tris–HCl, pH 8.0, 150 mM NaCl, 1 % Triton-×100, 0.1 % SDS, 0.5 % sodium deoxycholate + PPIC) and sonicated as above to obtain a nuclear extract. GAPDH and Histone H3 were used as protein markers for cytoplasmic/membrane and nuclear extraction efficiencies, respectively. Total protein was measured by BCA assay (Pierce) and Western blot samples were prepared with 4× loading buffer (125 mM Tris, pH 6.8, 8 % LDS, 40 % glycerol, Orange G) and heat-denatured at 70 °C for 15 min. Samples of equal protein were run on a range of Bio-Rad TGX mini-gels and transferred to PVDF or Nitrocellulose membranes. Membranes were blocked with LiCor Odyssey blocking buffer (1:1 TBS/Blocking Buffer) for 1 h at room temperature followed by incubation with primary antibody (diluted in 1:1 TBST/Blocking Buffer) over night at 4 °C with gentle rocking. HRP-conjugated secondaries (Jackson Labs or Cell Signaling technologies) diluted in 5 % milk/TBST or LiCor fluorescent secondaries diluted in 1:1 TBST/Blocking Buffer were used. West Dura (Pierce) substrate was used for chemiluminescent detection. Blots were imaged using an Odyssey Fc (LiCor) and analyzed using Image Studio software (Ver 3.1) for densitometry analysis. The following primary antibodies were used for Western blot: LC3A/B (1:1000; CST), total 4EBP1 and P-4EBP1 (1:1000; CST), human PGRN (1:1000; R&D), PGRN/PCDGF (1:700; Invitrogen), mouse PGRN (1 μg/mL; R&D), total S6 and P-S6 (1:1000; CST), p62 (1:1000; BD), TFEB (1:1000; CST), TUJ1 (1:2,000; Covance). Tubulin (1:20,000; Epitomics), Actin (1:10,000; Epitomics), GAPDH (1:5,000; Sigma), and Histone H3 (1:5,000; Millipore) were used as loading controls.

Cells were plated in 6-cm dishes in complete medium. Two days after plating, cells were washed 2× with serum-free media and treated over night with vehicle or drug at the indicated concentrations in serum-free media. Cell supernatants were collected the following day at the indicated times and immediately spun at 6000 rpm for 5 min at 4 °C to clear debris. The supernatant (500 μL) was transferred to a 0.5 mL Amicon Ultra concentrator (50 kDa cutoff) and spun for 5–10 min at 14,000 × g. The spin filter was placed upside down in a fresh tube and spun at 1000 × g for 2 min to collect the concentrated sample. The concentrated samples were normalized to total protein in the cell lysates to account for differences in cell numbers, mixed with 4× loading buffer to a final concentration of 1×, and heated at 70 °C for 15 min to denature.

H4 or primary fibroblast cells were fixed with 4 % paraformaldehyde for 15 min. Cells were then permeabilized with ice-cold methanol for 10 min. After blocking with 0.1 % BSA in PBS, cells were incubated with goat anti-PGRN (1:300; R&D) in blocking buffer overnight at 4 °C. After washing with PBS, cells were incubated in secondary antibodies conjugated to Cy3 (1:300; Jackson Immunoresearch) and DAPI (1:1000; Life Technologies) for 1 h at room temperature. Slides were mounted using Vectashield Hard Set (Vector Laboratories, Inc.; Burlingame, CA). iPSC-neurons were fixed with 4 % paraformaldehyde for 15 min. Cells were then permeabilized with ice-cold methanol or 0.1 % triton X-100 in PBS for 10 min. After blocking with 0.1 % BSA or 5 % Normal Donkey Serum (NDS), cells were incubated with the following primary antibodies in blocking buffer overnight at 4 °C: PGRN (Goat polyclonal, 1:300; R&D), LC3A/B (1:100; CST), Tuj1 (1:2000; Covance). After washing with PBS, cells were incubated in secondary antibodies conjugated to Cy5 (1:300; Jackson Immunoresearch) or Alexa fluor 488 (1:300; Life Technologies) for 1 h at room temperature. Slides were mounted using ProLong Gold Antifade Reagent with DAPI (Life Technologies). Images were collected with a Zeiss LSM 510 NLO META system (Emory University Integrated Cellular imaging Microscopy Core) or EVOS FL Cell Imaging System (Life Technologies).

RNA was extracted using QiaShredder lysis buffer and purified using RNeasy spin columns from Qiagen. Purified RNA (900 ng) was used as a template to produce cDNA using AB high Capacity RNA-to-cDNA kit (Life Technologies) according to the manufacturer’s protocol on a Bio-Rad C1000 Thermo Cycler. Approximately 20 ng of cDNA was used for quantitative real-time PCR (qPCR) experiments using Eppendorf plates and AB Power-SYBR mix (20 μL final reaction volume) on an Eppendorf Mastercycler ep realplex S. Primers were obtained from IDT and used at a final concentration of 200 nM. The following cycle conditions were used for all genes: 95 °C for 10 min followed by 40 cycles of 95 °C for 15 s and 54 °C for 60 s with a final extension step of 95 °C for 15 s and 54 °C for 60 s. The ΔΔCt method was used to calculate fold changes in RNA levels compared to vehicle treated cells after normalization to a reference gene, U36-B. The primer sets used were reported previously [13] as follows: human U36B-F, 5′-CGAGGGCACCTGGAAAAC-3′; human U36B-R, 5′-CACATTCCCCCGGATATGA-3′; human GRN-F, 5′-CAGGGACTTCCAGTTGCTGC-3′; human GRN-R, 5′-GCAGCAGTGATGGCCATCC-3′.

Mouse progranulin ELISAs were purchased from Adipogen (San Diego, CA) and carried out according to the manufacturer’s protocol. Plasma samples were diluted 1:500 in 1× Diluent buffer. Brain lysates were diluted 1:50 in 1× Diluent buffer. All ELISA measurements were done in duplicate and fell within the standard curve generated by the provided recombinant progranulin standard. Mouse plasma and brain lysate samples were randomized and loaded blind to the researcher. For brain and plasma ELISAs, a positive (Grn+/+) and negative (Grn−/−) control sample were also run to verify specificity of the ELISA.

All animal work was conducted with prior Institutional Animal Care and Use Committee (IACUC) approval, and was performed in accordance with PHS guidelines. All procedures were performed under conditions designed to minimize pain and distress. Emory University is an Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) approved institution, and follows the current version of the Guide for the Care and Use of Laboratory Animals (8th Edition), as adopted by the Office of Laboratory Animal Welfare (OLAW). Twenty Grn heterozygous mice (Grn+/−) obtained from our colony by crossing Grn knockout mice (Grn−/−) [14] with wild-type C57bl6J (originally acquired from The Jackson Laboratory, Bar Harbour, Maine) were used in the study. We elected to use both male and female mice in approximately equal numbers based on the number of animals available in the colony and to comply with recent NIH guidance on maintaining gender balance in biomedical animal studies. For treatments, solutions of 2 % sucrose or 2 % trehalose were made up in water (same regularly provided to animals in the vivarium) and filtered through a 0.2-μm stericup filter. Fresh water was changed out once per week. Weights of the water bottles were measured before and after each exchange to estimate amount of water consumed per group. Mice were weighed periodically to monitor changes in weight gain per group. After 65 days of treatment, mice were euthanized by decapitation after anesthetization with isoflurane. Blood was collected at time of death in a reservoir containing 100 μL of 0.5 M EDTA. Plasma was separated by centrifugation at 3000 rpm for 15 min at 4 °C and stored at −80 °C. Whole brain was harvested and separated into hemispheres after removal of the cerebellum. One half of the brain was drop fixed in 4 % paraformaldehyde (PFA) and the other half was placed in a tube, snap frozen in liquid nitrogen, and stored at −80 °C. Frozen brain tissue was ground under liquid nitrogen with a mortar and pestle to create a uniform powder. For protein extraction, approximately 50 mg of tissue powder was homogenized in lysis buffer containing 50 mM Tris–HCl (pH 7.4), 150 mM NaCl, 1 % Triton X-100 and complete protease and phosphatase inhibitor cocktail (Pierce) and sonicated briefly. Protein concentrations were determined by BCA assay (Pierce) and Western blot samples were made up as described above. Approximately 25 μg of total protein per sample was loaded onto a 12 % TGX mini gel (Bio-Rad) for Western blotting as described above.

All values are expressed as the mean ± SEM. For experiments where two groups were compared, a standard two-tailed Student’s t-test was used to measure significance. For comparisons of more than two groups, one-way analysis of variance (ANOVA) was used followed by Dunnet’s or Tukey’s comparison post-hoc test. For correlation analysis, Pearson’s r was used. All statistical analyses were performed in GraphPad Prism 6.02 (GraphPad Software, La Jolla, California, USA). A P-value <0.05 was considered significant.

PGRN partially localizes to late endosomal and lysosomal compartments within neurons [15, 16] and loss of PGRN causes lysosome dysfunction; therefore, we hypothesized that modulators of this pathway may also regulate GRN expression [11, 17, 18]. To test this hypothesis, we expressed a dual reporter construct (Fig. 1a) in HEK293T cells to monitor activity of the human GRN promoter and screened a custom chemical library of autophagy-lysosome modulators (see Methods section). We identified several compounds that increased reporter activity ~2-fold (Fig. 1b), comparable to suberoylanilide hydroxamic acid (SAHA), a histone deacetylase inhibitor (HDAC) which was previously identified as a robust regulator of GRN transcription in a high-throughput screen (HTS) of the Prestwick chemical library [13]. The top hits from our screen were Torin1, a mechanistic target of rapamycin (mTOR) inhibitor [19], and trehalose, an mTOR-independent activator of autophagy [20]. The mTOR kinase is a principle negative regulator of autophagy and numerous autophagy activators have been discovered that work via inhibiting mTOR and downstream signaling pathways (reviewed in [21]). We found that several other compounds that reportedly inhibit mTOR also increased GRN promoter activity (Additional file 1: Table S1), suggesting this signaling pathway may be involved in regulating GRN expression.

Next, we used a secondary cellular screen to determine whether identified compounds increased endogenous levels of PGRN. We treated human H4 neuroglioma cells with SAHA, trehalose, or the mTOR inhibitors rapamycin, PP242, or Torin1 for 24 h and monitored autophagy activation and PGRN expression by immunoblot (Fig. 1c). The conversion of the microtubule-associated protein light chain 3 from a non-lipidated form (LC3-I) to a lipidated membrane-bound form (LC3-II) correlates with the formation of autophagosomes and is used as a marker for autophagy [22]. All compounds resulted in increased levels of LC3-II and decreased levels of p62, an autophagic substrate, indicating autophagy activation. Trehalose had no effect on phosphorylation of S6 ribosomal protein (P-S6), a downstream target of mTOR, confirming that activation of autophagy was independent of mTOR inhibition. Further, trehalose and mTOR inhibitors increased both intracellular and secreted PGRN (Fig. 1c and Additional file 1: Figure S1a). Importantly, trehalose also increased PGRN in human (Fig. 1d) and mouse (Fig. 1e) neuroblastoma cell lines, and in mouse primary cortical neurons (Fig. 1f). In contrast, mTOR inhibitors failed to increase PGRN in these cell lines even though increased LC3-II was observed (Fig. 1d–f). These data indicate a mechanistic difference between trehalose and mTOR inhibitors in cells of the neuronal lineage and that mTOR-inhibition and PGRN upregulation are independent pathways. Thus, we focused on trehalose as our lead molecule because it had consistent and robust effects on PGRN levels across multiple disease-relevant cell types.

Next, we characterized the activity of trehalose in cultured cells in more detail. Using bafilomycin A1 (BafA1) to block autophagosome fusion with lysosomes, we found that trehalose increased autophagic flux in H4 cells (Additional file 1: Figure S2a-b), consistent with previous reports that trehalose likely acts at the step of autophagy initiation [23, 24]. In addition, treatment of H4 cells with trehalose had no overt effect on lysosomal acidification based on LysoTracker Red staining, unlike the vacuolar-ATPase inhibitor BafA1 (Additional file 1: Figure S2c). Therefore, it is likely that trehalose and the previously reported BafA1 have different mechanisms of action for increasing PGRN [25]. Finally, dose- or time-dependent treatment of H4 cells with trehalose did not affect cell viability (Additional file 1: Figure S2d). Overall, these results confirm that trehalose is well tolerated by mammalian cells.

Next, we found that trehalose significantly increased PGRN in a dose- (Fig. 2a–b) and time- (Fig. 2e–f) dependent manner that coincided with increased LC3-II (Fig. 2a, c and Fig. 2e, g). Once again, this effect was independent of mTOR inhibition as measured by phosphorylated 4EBP-1, a direct target of mTOR (Fig. 2a, d and Fig. 2e, h). A significant increase in LC3-II was observed prior to an increase in PGRN, indicating that activation of autophagy may precede PGRN upregulation. The dose- and time-dependent effects of trehalose on PGRN were confirmed in both human and mouse neuroblastoma cell lines as well (Additional file 1: Figure S3a-c). H4 cells treated with trehalose showed a robust increase in intracellular PGRN by immunocytochemistry, confirming the immunoblot data (Fig. 2i). Importantly, we observed a dose-dependent increase of secreted PGRN in the media of H4 cells treated with trehalose (Fig. 2j–k). Finally, we found that trehalose increased GRN mRNA in a dose-dependent manner in H4 cells (Fig. 3a) and human neuroblastoma cells (Additional file 1: Figure S3d) using quantitative real-time PCR (qPCR). Treatment of H4 cells with actinomycin D, an inhibitor of mRNA transcription, blocked the trehalose-mediated increase of PGRN (Fig. 3b). These data provide further evidence that trehalose primarily acts by increasing GRN gene transcription and supports the validity of our initial GRN promoter reporter assay.

The transcription factor EB (TFEB) has been described as a master regulator of autophagy-lysosomal gene expression [26, 27] and implicated in regulation of GRN transcription [26, 28]. In addition, loss of PGRN has been linked to TFEB activation [29] and trehalose has been reported to activate TFEB in cultured cells [30, 31]. Therefore, we sought to explore whether TFEB plays a role in mediating the trehalose-induced upregulation of PGRN. First, we measured the levels of GRN mRNA and PGRN protein in cultured HeLa cells stably overexpressing TFEB-GFP [12] compared to wild-type (WT) HeLa cells. We found that the TFEB-GFP HeLa cells had a small, but significant increase in both GRN mRNA and PGRN protein compared to wild-type HeLa cells (Fig. 4a–c), indicating that TFEB overexpression modulates PGRN expression via direct or indirect mechanisms.

Under basal conditions, TFEB is normally phosphorylated by several kinases, including mTOR, which retains TFEB in the cytoplasm. Upon nutrient starvation or pharmacological inhibition of mTOR, TFEB is dephosphorylated and translocates to the nucleus where it activates transcription of its gene targets [12, 32, 33]. To test whether trehalose activates TFEB, we treated TFEB-GFP HeLa cells with Torin1 or trehalose and monitored TFEB-GFP localization using fluorescence microscopy. While Torin1 robustly increased TFEB nuclear localization by 2 h, trehalose treatment only resulted in a diffuse nuclear TFEB-GFP signal by 24 h of treatment (Fig. 4d). Fractionation of treated TFEB-GFP HeLa cells into cytoplasmic/membrane and nuclear fractions revealed that trehalose treatment increased nuclear TFEB levels slightly whereas Torin1 significantly increased nuclear TFEB levels (Fig. 4e–f). In contrast, trehalose treatment robustly increased PGRN levels, significantly more than Torin1 treatment (Fig. 4e–f), indicating that the amount of TFEB in the nucleus does not correlate with PGRN expression.

Finally, to directly address the role of endogenous TFEB on trehalose-mediated PGRN upregulation we utilized a human haploid cell line (HAP1) deficient in TFEB (TFEB KO) that was generated using CRISPR/Cas9. First, we confirmed by immunoblot that TFEB was not expressed in HAP1 TFEB KO cells (Fig. 4g). We then treated HAP1 wild-type (WT) or TFEB KO cells with vehicle or trehalose and measured GRN mRNA and PGRN protein levels. TFEB KO cells showed no significant decrease in GRN mRNA or PGRN protein levels compared to WT cells (Fig. 4h, j). However, trehalose treatment significantly increased GRN mRNA levels in both cell lines to similar levels (Fig. 4h). Accordingly, we found that trehalose was able to robustly increase PGRN protein expression in TFEB KO cells to the same level as in WT cells (Fig. 4i–j). Taken together, these results demonstrate that trehalose is a relatively weak activator of TFEB and that activation of TFEB is not responsible for the trehalose-induced upregulation of GRN/PGRN expression in cultured cells.

We then asked if trehalose could rescue PGRN deficiency in an in vitro GRN haploinsufficient model. Human primary fibroblast cell lines were generated from 3 control (CTL) or 3 progranulin mutation (GRN) carriers with the following GRN mutations: c.1477C > T (R493X) (designated GRN #1, GRN #2) and c.592_593delAG (R198GfsX19) (designated GRN #3). Intracellular PGRN protein was reduced ~50 % in the GRN lines compared to CTL lines (Fig. 5a–b). However, secreted PGRN levels were reduced by nearly 80 % in GRN lines compared to CTL lines (Fig. 5a, c), which is comparable to circulating PGRN levels in human patients with GRN mutations [34‐36] and may reflect a defect in exosome secretion [37]. Treatment of GRN fibroblasts with trehalose significantly increased intracellular and secreted PGRN compared to vehicle treated GRN fibroblasts (Fig. 5a–c). Additionally, GRN mRNA from GRN fibroblasts, which was reduced by ~55 % compared to CTL fibroblasts, was significantly increased to near CTL levels after trehalose treatment (Fig. 5d). Immunocytochemistry of primary human fibroblasts revealed a robust increase in PGRN staining in GRN cells treated with trehalose, similar to vehicle treated CTL cells (Fig. 4e).

Next, we reprogrammed one CTL and one GRN (GRN #3) fibroblast line to generate induced pluripotent stem cells (iPSC) and derived neurons from each line using an embryoid body-based differentiation protocol (Fig. 6a–b). Treatment of the GRN embryoid bodies (EBs) with trehalose restored PGRN to CTL levels, whereas PP242 failed to increase PGRN (Fig. 6c), similar to what we observed in neuronal-like cells. In GRN iPSC-neurons, PGRN expression was detected in the cell body and processes (Additional file 1: Figure S4a) and expression was reduced compared to the CTL line (Fig. 6d). Treatment of GRN iPSC-neurons with trehalose for 24 h lead to a significant increase in PGRN (Fig. 6d–e) and LC3-II (Fig. 6d and Additional file 1: Figure S4b), indicating autophagosome induction. Overall, these data show that trehalose can rescue PGRN deficiency in cell-based models of GRN haploinsufficiency, including iPSCs and iPSC-derived neurons.

To extend our observations to an in vivo model, we treated three-month old Grn+/− mice with 2 % trehalose water (Trehalose, n = 7; 3 M/4 F) ad libitum for 65 days and then analyzed changes in PGRN levels in the brain and plasma. As controls, additional mice were treated with normal drinking water (Vehicle, n = 6; 3 M/3 F) or 2 % sucrose water (Sucrose, n = 7; 3 M/4 F). Similar treatment paradigms in mouse models have shown that trehalose can enhance autophagy [24, 38] and reduce pathologies [24, 39, 40], without adverse effects [41, 42]. In our study, mice treated with trehalose or sucrose consumed slightly more water than vehicle mice (Additional file 1: Figure S5a), although changes in body weight were not significant (Additional file 1: Figure S5b–c). Strikingly, PGRN levels in brain tissue were significantly increased in only the trehalose treated mice by both immunoblot (Fig. 7a‐b) and ELISA (Fig. 7c), with both measures positively correlating (Fig. 7d). We also measured PGRN in plasma by ELISA but found no significant differences across groups (Additional file 1: Figure S5d). This discordance could be due to the fact that circulating plasma PGRN levels do not reflect brain/CSF PGRN levels [43, 44]. Finally, LC3-II was significantly increased in brain tissue of the trehalose treated mice (Fig. 7a, e), which was positively correlated with PGRN by immunoblot (Fig. 7f). Taken together, our results demonstrate that oral administration of trehalose increases brain PGRN levels and autophagic markers in vivo.