A PIKfyve modulator combined with an integrated stress response inhibitor to treat lysosomal storage diseases

Significance

Lysosomal storage diseases (LSDs) are a category of rare genetic disorders resulting from a dysfunctional mutant lysosomal protein. The most common LSD, Gaucher’s disease (GD), is caused by low functional levels of mutant glucocerebrosidase. Low glucocerebrosidase activity is genetically linked to Parkinson’s disease risk. There are few treatment options for GD and other LSDs, especially for the neuropathic forms of these diseases. Here, we show that small-molecule PIKfyve modulators and an integrated stress response inhibitor can independently and together enhance mutant lysosomal enzyme function, including mutant lysosomal glucocerebrosidase function, by targeting cellular proteostasis pathways.

Abstract

Lysosomal degradation pathways coordinate the clearance of superfluous and damaged cellular components. Compromised lysosomal degradation is a hallmark of many degenerative diseases, including lysosomal storage diseases (LSDs), which are caused by loss-of-function mutations within both alleles of a lysosomal hydrolase, leading to lysosomal substrate accumulation. Gaucher’s disease, characterized by <15% of normal glucocerebrosidase function, is the most common LSD and is a prominent risk factor for developing Parkinson’s disease. Here, we show that either of two structurally distinct small molecules that modulate PIKfyve activity, identified in a high-throughput cellular lipid droplet clearance screen, can improve glucocerebrosidase function in Gaucher patient–derived fibroblasts through an MiT/TFE transcription factor that promotes lysosomal gene translation. An integrated stress response (ISR) antagonist used in combination with a PIKfyve modulator further improves cellular glucocerebrosidase activity, likely because ISR signaling appears to also be slightly activated by treatment by either small molecule at the higher doses employed. This strategy of combining a PIKfyve modulator with an ISR inhibitor improves mutant lysosomal hydrolase function in cellular models of additional LSD.

Gaucher’s disease (GD) is an autosomal recessive disorder and is the most prevalent lysosomal storage disease (LSD). Symptoms seem to be caused by the accumulation of harmful quantities of certain lipids, especially within the bone marrow. This leads to skeletal abnormalities; low levels of platelets, which can manifest in bleeding disorders; and low levels of circulating red blood cells, which are associated with anemia and fatigue (1). GD is characterized by abnormally low activity levels of the enzyme glucocerebrosidase (GCase) (<15%), a lysosomal hydrolase encoded by GBA1 that cleaves glucosylceramide into glucose and ceramide, as well as hydrolyzing similar substrates. Reduced GCase enzyme activity results in accumulation of its substrates in the lysosome, particularly in professional degradatory cells such as in the macrophages of the spleen and liver, often leading to enlargement of these organs (1). Common genetic scenarios in GD include a null mutation (84GG) on one allele and partial loss-of-function missense mutation (N370S or L444P) on the other, or partial loss-of-function mutations on both alleles. A missense mutation causes a majority of GCase to fold inefficiently in the neutral pH environment of the endoplasmic reticulum (ER), resulting in excessive degradation and compromising subsequent trafficking to the lysosome (2). Despite their instability in the ER, most mutant GCases have suitable stability and partial activity in the acidic environment of the lysosome, meaning the enzyme function can be rescued with the aid of pharmacological chaperones (3). This approach has been approved by the FDA for treating a different LSD, Fabry’s disease (4). The respiratory drug Ambroxol is currently being tested in clinical trials as a pharmacologic chaperone for GCase (https://www.clinicaltrials.gov/study/NCT03950050). The majority of pharmacological chaperone drug candidates for GCase are active site binders harboring a basic amine, which promotes their accumulation in the lysosome, resulting in partial enzyme inhibition of the additionally trafficked mutant GCase (3).

Interest in the development of additional therapeutic strategies for ameliorating GD has skyrocketed recently owing to the compelling genetic evidence linking mutations in GBA1 to an increased risk of developing Parkinson’s disease (PD) or Lewy body dementia (5–8). Biallelic pathogenic mutations on GBA1 result in a 10- to 20-fold higher risk of developing PD, while monoallelic pathogenic mutations result in a fivefold higher risk (9, 10), even though monoallelic GBA1 mutations alone do not cause GD. Therefore, developing treatments for GD that center around enhancing GCase activity could also be useful for treating PD (11–15).

Of the regulatory agency–approved treatment options for GD, enzyme replacement therapy (ERT) is the most successful, but recombinant lysosomal enzymes exhibit poor blood–brain barrier (BBB) penetrance (16, 17). Thus, ERT is limited to treating nonneuronopathic GD and probably would be ineffective for treating PD because the recombinant enzyme cannot effectively reach the central nervous system (CNS). Another approach approved for treating GD is substrate reduction therapy (SRT), which blocks production of the glucosylceramides that are hydrolyzed by GCase in the lysosome. While SRT causes more side effects than ERT (18–21), second-generation SRT drugs exhibit improved BBB penetrance and neuropathic efficacy (22–25). The potential of SRT for treating PD is unclear, as there is no evidence of any GCase substrate accumulation in the brains of PD patients who are heterozygous carriers of GBA1 mutations—raising questions about whether substrate accumulation is the root of the problem (26, 27). Given the drawbacks of ERT and SRT for treating neuropathic GD, there is still room for improvement in developing pharmacological strategies that will improve CNS GCase activity.

We previously demonstrated the potential of treating GD by altering cellular proteostasis capacity to improve the folding of GCase in the ER, enabling enhanced trafficking to the lysosome (28–32). Moreover, we demonstrated that pharmacological improvement of the folding capacity of the ER has the versatility to improve mutant hydrolase function in multiple LSDs (28, 29), implying that a single drug could be useful for treating multiple LSDs. In addition, previous studies have investigated whether autophagy activation could be a useful therapeutic strategy for treating LSDs, since both autophagic and lysosomal genes are regulated by transcription factor EB (TFEB) (33–36). TFEB is a member of the MiT/TFE family of transcription factors, which bind to E-box regulatory elements. One type of E-box, dubbed the CLEAR element, is ubiquitously found among the promoters of lysosomal biogenesis genes (36). Overexpression of TFEB increases mutant GCase levels in Gaucher patient–derived cells (33), suggesting activation of lysosomal biogenesis could be useful for treating GD. Another MiT/TFE family member, the transcription factor TFE3, has also been shown to regulate lysosomal genes and functions similarly to TFEB by also binding to the CLEAR element (35); therefore, focusing on TFEB as the sole activator of GCase in the context of autophagy could understate the involvement of other MiT/TFE family members. Thus, we hypothesize that activation of the CLEAR pathway is responsible for the upregulation of lysosomal genes like glucocerebrosidase via MiT/TFE family members. Since activation of the CLEAR pathway upregulates numerous lysosomal genes, discovering a small molecule that could induce this pathway could also be useful for treating multiple LSDs.

Our group recently completed a cell-based high-content imaging screen to identify small molecules that hasten lipid droplet clearance (37). This million-molecule screen has the potential to identify small-molecule autophagy activators, since autophagy activation is one mechanism of lipid droplet clearance (38). After multiple counterscreens, we narrowed down the number of autophagy activator candidates to 24. We hypothesized that one mechanism of autophagy activation involving significant lysosomal biogenesis (e.g., through activation of the CLEAR pathway) might be useful for treating LSDs. While the majority of molecules identified by our screen did not improve cellular mutant lysosomal hydrolase function, we found two structurally distinct, not previously reported PIKfyve modulators that do improve LSD–associated enzyme deficiencies. These PIKfyve modulators lead to the nuclear localization of the MiT/TFE transcription factors, particularly TFE3, resulting in lysosomal biogenesis and production of more mutant lysosomal hydrolases to be folded in the ER and trafficked to the lysosome, yielding a higher mutant lysosomal hydrolase concentration and activity. Since this strategy produces more difficult-to-fold mutant lysosomal enzymes in the ER, we posit that modest activation of the integrated stress response (ISR) occurs. This explains why an ISR antagonist can relieve partial translational attenuation, affording a further boost in lysosomal mutant enzyme levels, particularly for the most difficult-to-fold mutant lysosomal hydrolases.

Results

Small Molecules from a Lipid Droplet Clearance Screen Improve GCase Activity.

We first probed whether any of the 24 small molecules emerging from our lipid droplet clearance screen had the potential to ameliorate lysosomal storage disorders, initially focusing on compounds that could improve deficient mutant GCase activity thought to cause GD. Using the classical fluorogenic 4-methylumbelliferyl D-glucopyranoside (4-MUG) substrate in a GCase lysed cell assay (3, 29), we investigated whether any of these molecules could increase GCase activity in adult human dermal fibroblasts harboring two copies of wild type (WT) GCase. Two of these hits, A16 and A18 (Fig. 1A), significantly and appreciably improved WT GCase activity after a 72 h treatment period (Fig. 1B) with shorter treatment periods having a reduced effect (SI Appendix, Fig. S1). We suspected that a TFEB transcriptional response might be involved, so we tested whether treatment by either A16 or A18 promoted TFEB nuclear localization, which leads to upregulation of lysosomal genes including GBA1. We used indirect immunofluorescence to quantify the localization of endogenous TFEB in HeLa cells that were also costained with the nuclear dye DAPI (Fig. 1 C and D). Treatment of HeLa cells with A16 or A18 for 20 h increased the percentage of TFEB localized to the nucleus (i.e. colocalized with the DAPI signal; Fig. 1 C and D), indicating that both molecules induced TFEB nuclear localization. We hypothesized that we had identified two small molecules that improved GCase activity via a mechanism mediated by TFEB and/or another MiT/TFE transcription factor(s). The effect of A16 or A18 on GCase activity was not altered by shRNA knockdown of TFEB in fibroblasts; however, shRNA knockdown of TFE3, another MiT/TFE transcription factor, did mitigate the effect of A16 (Fig. 1E and SI Appendix, Fig. S2). In addition, TFE3 localizes to the nucleus in fibroblasts treated with A16 (Fig. 1 F and G). The lack of effect from TFEB knockdown may be due to the modest knockdown of TFEB (58%, vs 86% knockdown of TFE3; SI Appendix, Fig. S2a). Alternatively, TFE3 may have a more substantial role than TFEB in fibroblasts due to differences in abundance. Because shRNA overexpression appears to affect A16 and A18 efficacy, even with nontargeting shRNA, attempts to further knockdown TFEB would require other approaches.

Fig. 1.

Small molecules identified in previous HTS increase GCase activity. (A) Chemical structures of A16 and A18. (B) 72 h treatment of A16 or A18 increased GCase activity in non-Gaucher patient–derived fibroblasts, as measured by 4-MUG substrate hydrolysis in cell lysate. (C) Immunofluorescence confocal microscopy of HeLa cells after 20 h treatment of A16 and A18 showed TFEB nuclear localization; (D) quantification. (Scale bar, 40 µm.) (E) Non-Gaucher patient fibroblasts that were transduced with shRNA targeting TFEB, TFE3, or nontargeting shRNA were treated with A16 and A18 for 48 h. (F) Immunofluorescence confocal microscopy of fibroblasts after 24 h treatment of A16 showed TFE3 nuclear localization; (G) quantification. (Scale bar, 60 µm.) Bar graphs show mean ± 95% CI, and statistics were performed using the Kruskal–Wallis test with multiple comparisons (D and E) or Mann–Whitney T test (G). *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. Experiments were done with at least three replicates (N ≥ 3).

A16 and A18 Improve GCase Activity in Gaucher Patient Fibroblasts.

We next probed whether A16 and A18 could increase GCase activity in Gaucher patient–derived fibroblasts harboring common GCase mutations. We treated both L444P/L444P homozygous and N370S/84GG compound heterozygous Gaucher fibroblasts with increasing concentrations of either A16 or A18 for up to 72 h. Both molecules dose-dependently and time-dependently improved mutant GCase activity, with A16 showing higher potency in both genotypes (Fig. 2A and SI Appendix, Fig. S3). Even higher concentrations of A16 (1 µM) and A18 (10 µM) showed a decrease in the improvement of GCase activity (SI Appendix, Fig. S4 A and B), but higher concentrations (2.5 µM A16 and 10 µM A18) only caused minimal cell toxicity, as assessed by Hoechst and Sytox^TM Orange staining (SI Appendix, Fig. S4 C and D), which suggests that the reduction in GCase activity at high concentrations was not due to cell toxicity. Interestingly, treatment with A16 or A18 increased “cell viability” levels above vehicle treatment as measured by the CellTiter-Glo® ATP-based viability assay (SI Appendix, Fig. S4E), but this increased viability was not reflected in the Hoechst or Sytox^TM Orange staining.

Fig. 2.

A16 and A18 improve GCase activity in Gaucher patient–derived fibroblasts. (A) 72 h treatment of A16 or A18 improved GCase activity in Gaucher fibroblasts with L444P/L444P or N370S/84GG mutations on GBA1, as measured by 4-MUG hydrolysis in cell lysate. (B) Treatment of L444P/L444P Gaucher fibroblasts with A16 or A18 for 24 h increased GCase mRNA levels, as determined by qPCR. (C) N370S/84GG Gaucher fibroblasts that were treated with A16 or A18 for 72 h had their lysate subjected to EndoH digestion. Western blots of lysate showed increased EndoH-resistant GCase bands in small molecule–treated groups. See SI Appendix, Fig. S5 for quantification of EndoH-sensitive GCase. (D) L444P/L444P Gaucher fibroblasts that had been treated with A16 or A18 for 72 h were labeled in situ with fluorescent GCase activity-based probe EW644. Lysates were run on SDS-PAGE, showing higher amounts of ABP-conjugated GCase in small molecule–treated groups. (E) Lysates conjugated with A16 and A18 showed shifted band density to higher molecular weights on ABP-conjugated GCase. Bar graphs show mean ± 95% CI, and statistics were performed using the Kruskal–Wallis test with multiple comparisons. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. Experiments were done with at least three replicates (N ≥ 3).

Since GBA1 is a direct transcriptional target of TFEB via the CLEAR element (34) and both A16 and A18 enhance MiT/TFE family member nuclear localization, we hypothesized that their treatment would increase GCase expression. We performed qPCR on L444P/L444P fibroblasts that had been treated with increasing concentrations of either molecule and found that both molecules increased GBA1 mRNA levels (Fig. 2B). The dose-dependent increase in GCase enzyme activity seems to derive at least in part from the increased transcription of mutant GCase due to CLEAR pathway activation, although we didn’t assess the possibility that the turnover of GBA1 mRNA is slowed down by A16 or A18 treatment.

We posited that A16 or A18 treatment could also increase mutant GCase folding and trafficking by upregulating proteostasis network components. The CLEAR pathway is known to upregulate genes involved in folding and trafficking GCase, like M6PR and CALR (33). To monitor folding and trafficking of GCase after A16 or A18 treatment, cell lysate was subjected to EndoH glycosidase administration, which cleaves the N-glycan installed in the ER, indicating ER-retained GCase associated with improper folding (“EndoH-sensitive” GCase). EndoH glycosidase does not cleave the N-glycans off GCase that have been remodeled by Golgi glycan remodeling enzymes (“EndoH-resistant” GCase), reflecting properly folded GCase. Notably, both small molecules increased the amount of EndoH-resistant (properly folded) mutant GCase enzyme without increasing the amount of EndoH-sensitive GCase (Fig. 2C and SI Appendix, Fig. S5), indicating that more GCase had been folded in the ER and trafficked to the Golgi, since GCase must fold properly in the ER in order to bind to receptors for vesicular trafficking to the Golgi.

While in our experience the 4-MUG assay in lysed cells is the most reliable assay to demonstrate increased mutant GCase activity upon pharmacologic adaptation, we also wanted to demonstrate that mutant GCase activity was increasing in intact cells. To do so, we employed a fluorescently tagged cyclophellitol activity–based protein profiling (ABPP) probe (39) to covalently label active GCase in living cells. After probe treatment and conjugate formation, the cells were lysed, and the lysate was separated via SDS-PAGE to visualize the fluorescently labeled conjugate between GCase and the cyclophellitol probe. Since only active GCase can react with the probe, as demonstrated by the lack of EndoH-sensitive GCase conjugate formation (SI Appendix, Fig. S6A), this approach allowed us to quantify the amount of active GCase in intact cells. Cells treated with either A16 or A18 exhibited an increase in the amount of labeled GCase in both L444P/L444P and N370S/84GG GCase fibroblasts (Fig. 2D and SI Appendix, Fig. S6B, respectively). The fold improvement in GCase activity as determined by ABPP labeling appeared to be less than that determined by the lysed 4-MUG assay, especially in the N370S/84GG mutant line (Fig. 2 A and D and SI Appendix, Fig. S6B). However, prolonged treatment of A16-treated N370S/84GG fibroblasts with activity-based probes shows a much more substantial increase in labeled enzyme that is on par with the 4-MUG assay results (compare SI Appendix, Fig. S6C to Fig. 2A). This, in combination with the upregulation of GCase mRNA levels, suggests that the effect of A16 and A18 on total GCase activity is most likely due to an increase in total active GCase rather than a change in the specific activity of the GCase enzyme. Treatment by either A16 or A18 appeared to shift the GCase ABPP conjugate band to a higher molecular weight, consistent with distinct N-glycan enzymology in the Golgi compartment relative to untreated (Fig. 2E and SI Appendix, Fig. S6D). This shift occurred not only in N370S/84GG and L444P/L444P Gaucher fibroblasts, but also in WT GCase fibroblasts (SI Appendix, Fig. S6E). Treatment of probe-labeled lysate with PNGaseF, which cleaves all forms of N-glycosylation, abolished all molecular weight changes that occurred from A16 or A18 treatment (SI Appendix, Fig. S6F), confirming that A16 or A18 treatment altered the N-glycosylation of active GCase. Since N-glycosylation generally improves structural stability and has been shown to be important in the organization of the GCase catalytic site and the enzyme’s catalytic activity (40, 41), the observed increase in GCase glycosylation may improve its catalytic activity and stability, although further studies on GCase glycosylation and stability would have to be performed to substantiate this hypothesis.

A16 and A18 Appear to Increase GCase Activity Via Modulation of PIKFyve.

Cells treated with high concentrations of A16 or A18 exhibited a vacuolization phenotype (SI Appendix, Fig. S7A) that was reminiscent of cells treated with PIKfyve inhibitors like Apilimod (42); therefore, we suspected that both molecules may operate through PIKfyve modulation. Additionally, A18 shares the same N-heterocyclic–morpholine substructure (shown in red) that comprises many PIKfyve inhibitors (e.g., Apilimod, YM201636; Fig. 3A) (43). However, A16 is structurally distinct from these molecules (Fig. 1A). Using the DiscoveRX KinomeScan assay (44), we tested whether A16 and A18 bind to isolated recombinant PIKfyve, and found that both molecules did so, with a K_D of <0.17 nM and 11 nM, respectively (Fig. 3B). These results suggest that modulation of the PIKfyve protein complex is most likely the mechanism of action for the observed increase in GCase activity. Since the PIKfyve kinase, the complementary phosphatase (FIG4), and a scaffold protein (VAC14) all bind to one another in the cell, we prefer to call recombinant PIKfyve binders “modulators” until we know what A18 and A16 do to all the activities of this complex in a cellular context.

Fig. 3.

A16 and A18 operate through PIKfyve modulation in order to improve GCase activation. (A) Structures of A18 with established PIKfyve modulators Apilimod and YM201636, highlighting structural similarities in red. (B) A16, A18, and Apilimod bound PIKfyve as shown by DiscoveRX KINOMEScan for PIKfyve. (C and D) Treatment of Gaucher fibroblasts with (C) Apilimod and (D) YM201636 for 72 h improved GCase activity, as measured by 4-MUG hydrolysis in cell lysate. (E) L444P/L444P Gaucher fibroblasts that were transduced with nontargeting and PIKfyve-targeting shRNA, followed by 72 h treatment with A16 or A18. A16 showed reduced efficacy in improving GCase for the shPIKfyve-transduced cells than the shNonTargeting transduced cells, as measured by 4-MUG hydrolysis in cell lysate. Bar graphs show mean ± 95% CI, and statistics were performed using the Kruskal–Wallis test with multiple comparisons. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. Experiments were done with at least three replicates (N ≥ 3).

To support the hypothesis that PIKfyve modulation improves GCase activity, we treated mutant Gaucher fibroblasts with established PIKfyve inhibitors Apilimod and YM201636 for 72 h (Fig. 3 C and D and SI Appendix, Fig. S7D), and found that both PIKfyve inhibitors improved GCase activity. Additionally, we performed shRNA studies in L444P/L444P fibroblasts and found that knockdown of PIKfyve (SI Appendix, Fig. S7C) reduced GCase improvement from A16 treatment (Fig. 3E), confirming that PIKfyve is most likely the target, although attenuation of enhanced L444P/L444P GCase function was not shown by treatment with the less potent compound A18 upon PIKfyve knockdown.

Medicinal Chemistry Striving to Improve the Potency and Efficacy of A18.

We hoped that a more potent and efficacious A18 analogue(s) could be identified by substituting one of the three components attached to the sp² carbons (namely the morpholine, 4-iodoaniline, and 2-(4-morpholonyl)ethanamine substructures) (SI Appendix, Table S1). We evaluated the effects of A18 analogues on L444P/L444P fibroblast GCase activity (Fig. 4A and SI Appendix, Fig. S8 A–E). A18 analogues replacing the morpholine group mostly eliminated improvements to GCase activity (SI Appendix, Fig. S8A), consistent with significantly reduced PIKfyve binding (SI Appendix, Fig. S9). The loss of GCase activity improvement from replacing the morpholine substructure of A18 supports our previous hypothesis that PIKfyve modulation improves GCase activity (Fig. 3A). Substituting the 4-iodoaniline group with other aromatics did not enhance potency (SI Appendix, Fig. S8B). Iodobenzenes are vulnerable to photolysis (45); more medicinal chemistry will be necessary to develop an analogue of A18 without this functional group, but this should be possible.

Fig. 4.

Medicinal chemistry efforts on A18 gave rise to multiple chemical analogues with different effects on GCase improvement and PIKfyve binding. (A) GCase activity in L444P/L444P Gaucher fibroblasts after 72 h treatment with A18 analogues LM-2-34 and LM-2-27A, as determined by 4-MUG hydrolysis in cell lysate. (B) Structure of A18 analogues LM-2-34 and LM-2-27A, with differences from A18 highlighted in red. (C) A18 analogues bound to PIKfyve as shown by DiscoveRX KINOMEScan for PIKfyve. (D) Western blots of EndoH treated/nontreated cell lysate from Gaucher fibroblasts that have been treated with A18 or LM-2-27A for 72 h. See SI Appendix, Fig. S8F for quantification of EndoH-sensitive GCase. (E) Treatment of Gaucher fibroblasts with LM-2-27A for 24 h increased GCase mRNA levels, as determined by qPCR. The bar graph shows mean ± 95% CI, and statistics were performed using the Kruskal–Wallis test with multiple comparisons. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. Experiments were done with at least three replicates (N ≥ 3).

Substituting the A18 2-(4-morpholonyl)ethanamine substructure with alternatives afforded two improved molecules (Fig. 4 A and B). The first molecule, LM-2-34, substituted the 2-(4-morpholonyl)ethanamine for an indole, also found in A16. The indole by itself is not sufficient for PIKfyve binding, as LM-2-41, which substitutes the morpholine of A18 with an indole, showed no improvement to GCase activity (SI Appendix, Fig. S8A), nor did it show PIKfyve binding (SI Appendix, Fig. S9). Considering that the K_D of LM-2-34 binding to recombinant PIKfyve is not much lower than that of A18 (Fig. 4C), it may be important to address the activities of the PIKfyve complex in cells to understand why LM-2-34 is a more potent compound. The second interesting molecule in this series is LM-2-27A, which substituted the 2-(4-morpholonyl) ethanamine substructure group for a 5-ethoxy-1,3-benzothiazole-2(3H)-thione (Fig. 4B). LM-2-27A is interesting because it exhibited decreased potency and poorer binding to PIKfyve, yet it displayed the highest efficacy at improving GCase activity so far at 10 µM (Fig. 4 A and C and SI Appendix, Fig. S8E), implying that the precise binding mode to the PIKfyve–FIG4–VAC14 cellular complex may matter. Treatment with LM-2-27A (10 µM) also improved GCase folding and trafficking, as shown by EndoH western blots (Fig. 4D and SI Appendix, Fig. S8F), and increased GCase mRNA levels (Fig. 4E). Further medicinal chemistry efforts on LM-2-27A afforded multiple analogues, although none of them improved the efficacy or potency of LM-2-27A (SI Appendix, Fig. S8D).

PIKfyve Modulators Plus an ISR Inhibitor Enhance GCase Activity.

Previously, our group hypothesized that activation of the unfolded protein response (UPR) through the use of the proteasome inhibitor MG-132 improves GCase activity in Gaucher cells (29). We refrained from further exploring this approach, as proteasome inhibition activates a variety of pathways beyond UPR (46, 47) and may not be an ideal approach for treating GD. We turned instead to a small molecule that selectively activates all three arms of the UPR. This molecule, dubbed 132 (not to be confused with MG-132), does not induce other forms of cellular stress (48). We found that 132 could dose-dependently and significantly improve GCase activity (SI Appendix, Fig. S10); however, the magnitude of the effect was modest and did not reflect that achieved by MG-132 treatment. We tested whether activation of the UPR through 132 administration could synergize with A18 and found that cotreatment led to an additive effect, though it was not as substantial as we had hoped for (SI Appendix, Fig. S10).

While investigating the effect of selective UPR modulators and related compounds on GCase activity in L444P/L444P Gaucher fibroblasts, we found that inhibition of the ISR using ISRIB dose-dependently improved GCase activity (Fig. 5A). Although the magnitude of the effect alone was modest, cotreatment of ISRIB with a PIKfyve modulator substantially improved GCase activity (Fig. 5B). Notably, this improvement from cotreatment appeared to be greater than the sum of the individual treatments. ISRIB improves ~10% on its own and A16 improves GCase activity ~70% alone, thus the additive treatment is expected to afford an ~80% increase; however, the observed magnitude of the effect is ~100%, or doubling of GCase activity (Fig. 5B). Ratiometrically, the increase in GCase activity is greater than the individual effect of ISRIB (Fig. 5C). With the combination of a PIKfyve modulator and ISRIB, we were able to achieve ≥2-fold improvement in GCase activity in multiple Gaucher cell lines (SI Appendix, Fig. S11), with the highest improvement (2.6-fold increase) resulting from cotreatment with ISRIB and A18 analogue LM-2-27A (Fig. 5D). Employing activity-based probe labeling of GCase in live L444P/L444P fibroblasts (Fig. 5E), we found that cotreatment with ISRIB and Apilimod further increased the amount of active GCase beyond Apilimod alone. Upon performing qPCR on L444P/L444P Gaucher fibroblasts treated with Apilimod ± ISRIB (Fig. 5F), we found that ISRIB did not transcriptionally contribute to differences in GCase expression, indicating that ISR inhibition was operating by a mechanistically distinct pathway. Using an ATF4-luciferase HEK293T reporter cell line (49) (ATF4 being a downstream transcription factor of the ISR), we confirmed that all the PIKfyve modulators of focus herein caused ISR activation at doses higher than those achieving mutant lysosomal hydrolase activity increases (compare Fig. 5B to SI Appendix, Fig. S12 A–C) and that cotreatment with ISRIB antagonizes ISR activation toward or below basal levels (SI Appendix, Fig. S12 A–C). However, at the PIKfyve modulator concentrations optimal for enhancing cellular GCase activity [A16 (500 nM); A18 (1 µM); Apilimod (50 nM)], after 24 h of treatment, we do not see RNAseq-based transcriptional changes that support the hypothesis that we are activating the ISR in L444P fibroblasts (SI Appendix, Fig. S12 D–G). Upon evaluating this PIKfyve modulator/ISR inhibitor combination strategy in additional Gaucher fibroblast lines, we observed modest enhancements for other genotypes, indicating that the observed enhancement from cotreatment appears to mostly benefit the L444P GCase mutation (SI Appendix, Figs. S11 and S13), which is among the most difficult-to-fold GCase variants. Since PIKfyve modulators produce more mutant lysosomal enzymes to be folded in the ER and trafficked to the lysosome, we postulate that the ISR is partially activated downstream of the UPR stress-responsive signaling pathway due to the increased production of L444P GCase. We posit that incomplete folding of mutant lysosomal hydrolase in the ER leads to weak UPR activation and coupled ISR activation, possibly through PERK. Considering that 24 h treatment with PIKfyve modulators only affords modest upregulation of GCase protein (SI Appendix, Fig. S1 and Fig. 3), we suspect that a 24 h treatment may be too short to cause appreciable buildup of lysosomal proteins and ISR activation (SI Appendix, Fig S12 D–G). Alternatively, we cannot rule out the possibility that the PIKfyve modulators A16, A18, and Apilimod directly activate the ISR, although, to the best of our knowledge, this connection has not been reported in the literature.

Fig. 5.

Inhibition of the ISR via ISRIB improves GCase activity and can amplify the effect of PIKfyve modulators, as determined by 4-MUG hydrolysis in cell lysate. (A) 72 h treatment of ISRIB improved GCase activity in L444P/L444P Gaucher fibroblasts. (B) 72 h cotreatment of ISRIB with A16, A18, or Apilimod further improved GCase activity in L444P/L444P Gaucher fibroblasts over individual treatments. (C) Ratiometric GCase activity of (B), comparing the improvement of ISRIB when cotreated with PIKfyve modulators. (D) 72 h cotreatment of ISRIB with LM-2-34, LM-2-27A, or YM201636 further improved GCase activity in L444P/L444P Gaucher fibroblasts over individual treatments. (E) L444P/L444P Gaucher fibroblasts that had been treated with Apilimod and/or ISRIB for 72 h were labeled in situ with fluorescent GCase activity-based probe EW644. Lysates were run on SDS-PAGE, showing higher amounts of ABP-conjugated GCase in Apilimod-treated cells, and further increase with ISRIB cotreatment. (F) 24 h treatment of L444P/L444P Gaucher fibroblasts with Apilimod increased GCase transcription, but ISRIB did not alter GCase mRNA levels, nor did it make any contribution to GCase mRNA levels when cotreated with Apilimod, as determined by qPCR. (A–D) was performed by measuring 4-MUG hydrolysis in cell lysate. Bar graphs show mean ± 95% CI, and statistics were performed using the Kruskal–Wallis test with multiple comparisons. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. Experiments were done with at least three replicates (N ≥ 3).

The PIKfyve/ISR Strategy Appears to be Useful for Treating Multiple LSDs.

Having established that PIKfyve modulators, functioning through the CLEAR transcriptional program, can improve cellular mutant GCase activity, we wanted to probe whether other LSDs could benefit from such a treatment. We also evaluated whether combining a PIKfyve modulator with an ISR inhibitor could contribute an additional increase in lysosomal hydrolase activity for lysosomal enzymes beyond GCase using the analogous 4-MU-based lysed cell assays. Fabry patient fibroblasts hemizygous for R301G on the GLA gene displayed a substantial enhancement of mutant α-galactosidase activity from cotreatment (Fig. 6A and SI Appendix, Fig. S14A). Cotreatment of PIKfyve modulators with ISRIB enhanced acid α-glucosidase activity in Pompe fibroblasts that are heterozygous for an IVS1-13T>G substitution on GAA, which generates a mis-spliced protein (Fig. 6B and SI Appendix, Fig. S14B). The PIKfyve modulator/ISRIB combination also improved β-galactosidase activity in mucopolysaccharidosis IV fibroblasts that are compound heterozygous for W273L/W509C or W273L/R482H GLB1 mutations (Fig. 6 C and D and SI Appendix, Fig. S14 C and D). Finally, cotreatment of PIKfyve modulators and ISRIB improved lysosomal α-mannosidase activity in α-mannosidosis fibroblasts that are compound heterozygous for MAN2B1 Q639X/R750W or homozygous for MAN2B1 R750W/R750W (Fig. 6 E and F and SI Appendix, Fig. S14 E and F). The results from these different assays indicate that treatment with a PIKfyve modulator and an ISR inhibitor exhibits promise for improving multiple lysosomal storage disorders, although the degree of improvement varied among the different mutant lysosomal enzymes. We also monitored the activity change in the corresponding WT hydrolases (i.e., β-galactosidase, α-galactosidase, α-mannosidase, and α-glucosidase) in addition to glucocerebrosidase and acid lipase in healthy dermal fibroblasts (SI Appendix, Figs. S15 and S16). Generally, the three PIKfyve modulators (A16, A18, or Apilimod) exhibited improved WT lysosomal enzyme function, and ISRIB cotreatment provided additional benefit, although the degree of improvement varied among the different WT enzymes, which may be dependent on the extent of coupled UPR-ISR activation caused by WT lysosomal hydrolase overexpression. The cellular hydrolase activity improvement trend among PIKfyve modulators was consistent with the molecules’ PIKfyve binding strength to an isolated PIKfyve kinase (Apilimod > A16 > A18).

Fig. 6.

PIKfyve modulators improve multiple lysosomal enzyme activities in various LSD fibroblasts. (A) Fabry, (B) Pompe, (C and D) mucopolysaccharidosis IV, (E and F) α-mannosidosis fibroblasts that were treated with PIKfyve modulators for 72 h had increased mutant enzyme activity, as measured by respective 4-MU-based fluorescent substrate assays in cell lysate. See SI Appendix, Fig. S14 for corresponding ratiometric data. Bar graphs show mean ± 95% CI, and statistics were performed using the Kruskal–Wallis test with multiple comparisons. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. Experiments were done with at least three replicates (N ≥ 3).

Discussion

In 2008, our laboratory provided evidence that proteasome inhibitor treatment (MG-132) improved mutant L444P glucocerebrosidase function in GD fibroblasts and enhanced G269S hexosaminidase A function in Tay–Sachs disease fibroblasts. We postulated that this benefit resulted from activation of the UPR (29) because we observed UPR activation. However, herein we showed that a fit-for-purpose UPR activator did not substantially increase L444P glucocerebrosidase function in fibroblasts (SI Appendix, Fig. S10). It is now well established that proteosome inhibition activates autophagy (50), and it would be interesting to study if autophagy activation is the basis of mutant lysosomal hydrolase activity improvements mediated by MG-132 treatment that we previously ascribed to UPR activation (29).

Herein we report several small-molecule PIKfyve modulators, which together with an ISR inhibitor are capable of producing a significant improvement in mutant GCase activity in a variety of Gaucher patient fibroblast lines and in the activities of additional mutant lysosomal enzymes associated with four different LSDs in fibroblast disease models. We also show that established PIKfyve modulators like Apilimod and YM201636 are capable of replicating these results, confirming this pathway as being important for mutant lysosomal hydrolase cellular activity increases.

PIKfyve is a kinase responsible for phosphorylating phosphatidylinositol (PI) and phosphatidylinositol 3-phosphate (PI(3)P) into phosphatidylinositol 5-phosphate (PI(5)P) and phosphatidylinositol 3,5-bisphosphate (PI(3,5)P₂), respectively, and is the only known source of PI(3,5)P₂ (51). PIKfyve interacts with VAC14 (a pentameric scaffold) and FIG4 (a phosphatase that converts PI(3,5)P₂ back to PI(3)P) in a 1:5:1 stoichiometric ratio to form a complex, whose interactions and activities appear to be interdependent. While A16 and A18 inhibit the isolated PIKfyve kinase, the interdependence of the kinase and phosphatase activities in the PIKfyve complex inside of fibroblasts makes us hesitant to conclude that PIKfyve inhibition is the mechanism for enhancing mutant lysosomal hydrolase function without further understanding of how A16 and A18 affect the kinase and phosphatase activities in a cellular context (52–54). Despite being more potent for increasing mutant GCase activity in fibroblasts, neither of the A18 analogues we produced through medicinal chemistry efforts appear to exhibit substantially increased binding to the isolated PIKfyve kinase, implying that how these analogues bind to the PIKfyve–FIG4–VAC14 complex and control the concentrations of PI, PI(3)P, PI(5)P, and PI(3,5)P₂ could matter with regard to mutant lysosomal hydrolase functional rescue.

PIKfyve modulation by Apilimod has been evaluated in clinical trials for its inhibition of toll-like receptor activation and IL-12/IL-23 production (42); however, it failed to show any efficacy in treating autoimmune disorders like Crohn’s Disease and Rheumatoid Arthritis in placebo-controlled trials (55, 56). Notably, Apilimod did show favorable safety and pharmacokinetic profiles in both trials. Other cases where PIKfyve modulators have shown therapeutic value are in targeting B cell lymphoma and viral infection (57–59) and in improving motor neuron survival and reducing tau aggregates in neurodegenerative diseases (60–62). Genetic ablation of PIKfyve has been shown to be detrimental toward neurodevelopment (51, 63); however, knockdown of PIKfyve impacts the formation/degradation of the members of the PIKfyve–FIG4–VAC14 complex, so these effects could also be attributed to VAC14, FIG4, or proteins known to interact with the PIKfyve–FIG4–VAC14 complex (63, 64). While we have identified autophagy activators in our original lipid-droplet clearance screen, we are reluctant at this time to conclude that A16 and A18 activate autophagy, because several papers have shown that PIKfyve inhibition leads to reduced autophagic flux and endocytic trafficking (65–67), while other publications show that PIKfyve inhibition leads to increased secretory autophagy (68). Nonetheless, PIKfyve inhibition via Apilimod has been shown to induce TFEB nuclear localization associated with lysosomal biogenesis (69). Given the complicated relationship among the PIKfyve–FIG4–VAC14 components in the complex, predicting what type of PIKfyve allosteric modulation could be beneficial or detrimental as it relates to hastening lysosomal flux and/or enhancing mutant lysosomal hydrolase function will require further experimental effort.

The ISR is a eukaryotic stress-responsive signaling pathway initiated by four kinases, including PERK, that attenuates global protein translation while allowing selective translation of specific mRNAs containing 5′ upstream open reading frames (uORFs), including the ATF4 transcription factor (70). We hypothesize that PIKfyve pharmacological modulation by A16, A18, or Apilimod directs one or more MiT/TFE transcription factors to the nucleus, resulting in the production of more mutant lysosomal enzymes, leading to more enzyme folding in the ER and more mutant hydrolase trafficking to the lysosome—increasing the concentration and activity of the mutant hydrolase enzymes in the lysosome. Since these lysosomal enzymes are difficult to fold, particularly in the case of misfolding-prone hydrolases like L444P GCase, we hypothesize that their increased production results in ISR activation downstream of UPR activation, possibly mediated by PERK. This ISR activation leads to the inhibition of eIF2B activity, which partially inhibits global protein translation, including translation of mutant lysosomal hydrolase enzymes. ISRIB, an ISR inhibitor reported by the Walter laboratory, potently reverses these effects (IC₅₀ = 5 nM) by facilitating the assembly of more eIF2B heterodecamers, thus resuming translation and down-regulating ATF4 target genes (71, 72). We hypothesize that the improvement of mutant lysosomal hydrolase activity from treatment with A16, A18, or Apilimod plus ISRIB is due to the latter relieving global translational attenuation and that the efficacy of this improvement may correlate with the magnitude of ISR activation, which is why the effect is most notable in cell lines harboring the most difficult-to-fold mutant lysosomal hydrolases. This hypothesis is supported by qPCR studies on L444P/L444P Gaucher fibroblasts treated with Apilimod ± ISRIB, revealing that ISRIB did not increase the L444P GCase transcript levels beyond the increase caused by the PIKfyve modulator Apilimod (Fig. 5F). Nonetheless, the amount of L444P GCase enzyme produced and L444P GCase function both increase more with ISRIB added (Fig. 5E), consistent with a reduction in translational attenuation. In other words, treatment with ISRIB in cells harboring lysosomal hydrolases that are difficult to fold antagonizes translational attenuation that results from PIKfyve modulation–associated UPR-ISR signaling, thus permitting PIKfyve modulators to translate lysosomal genes and bypassing the negative translational attenuation effect of ISR activation. Even overproduction of WT lysosomal hydrolases by our PIKfyve modulators may afford modest UPR and ISR activation that may affect lysosomal genes, explaining the alpha-galactosidase activity increase upon ISRIB treatment in WT fibroblasts (SI Appendix, Fig. S15D). Alternatively, the PIKfyve modulators employed herein could directly activate the ISR through an unknown mechanism.

Here, we showed that modulation of PIKfyve activity combined with ISR inhibition significantly improves deficient lysosomal hydrolase levels, folding, and function in fibroblast models of five different lysosomal storage disorders. Moreover, in WT fibroblasts, modulation of the activity of PIKfyve combined with ISR inhibition significantly increased lysosomal acid lipase (LIPA) activity, suggesting that this combination treatment is promising for ameliorating Wolman’s disease (an LSD associated with mutant LIPA). Finally, both Apilimod (https://clinicaltrials.gov/study/NCT05163886) and ISR inhibitors based on ISRIB (https://clinicaltrials.gov/study/NCT04948645) have proven or are proving to be safe in clinical trials for other diseases, which makes them promising candidate drugs for repurposing to treat several LSDs, including GD and potentially PD.

Materials and Methods

Chemical Synthesis.

Synthesis of A18 was adapted from previous literature (73). Detailed synthetic methods are described in SI Appendix. All reactions were carried out under an inert atmosphere (argon or nitrogen) with dry solvents under anhydrous conditions, unless otherwise noted. Anhydrous solvents (acetone, acetonitrile, tetrahydrofuran) were purchased from Acros Organics® (extra dry over molecular sieves, AcroSeal^TM) and used as received. Deuterated solvents were purchased from Cambridge Isotope Laboratories, Inc. Yields refer to chromatographically and spectroscopically (¹H NMR) homogeneous materials, unless otherwise stated. Reagents were purchased from commercially available sources at the highest commercial quality and used without further purification, unless otherwise stated. Reactions were monitored by either liquid chromatography–mass spectrometry (LC–MS) or thin layer chromatography (TLC). LC–MS was performed on an Agilent 1260 Infinity II LC system, equipped with a single quadrupole mass selective detector, using an analytical reverse-phase high-performance liquid chromatography (RP-HPLC) column (Agilent ZORBAX RRHT StableBond C18, ϕ2.1 × 50 mm, 1.8 µm, 80 Å, flow rate = 0.5 mL/min) with a linear gradient of 10% to 95% solvent B over 17 min (solvent A: 0.1% formic acid in water, solvent B: 0.1% formic acid in acetonitrile). TLC was carried out on glass TLC plates (MilliporeSigma Supelco silica gel 60 with fluorescent indicator F₂₅₄, 5 × 20 cm), illuminated with UV light (254 nm). TLC plates are stained using an acidic ethanol solution of p-anisaldehyde, an acidic n-butanol solution of ninhydrin, a basic aqueous solution of potassium permanganate, followed by brief heating on a hot plate. Separation of reaction mixtures was performed by either preparative RP-HPLC or flash column chromatography. Preparative RP-HPLC was performed on an Agilent 1260 Infinity LC system (Phenomenex Gemini NX-C18, ϕ30 × 250 mm, 5 µm, 110 Å, flow rate = 25 mL/min, solvent A: 0.1% NH₄OH in water, solvent B: 0.1% NH₄OH in acetonitrile, unless otherwise stated). Flash column chromatography was performed on a CombiFlash NextGen 300+ system using silica gel flash columns (Luknova SuperSep™ high performance or RediSep Gold® high performance). Yields refer to chromatographically and spectroscopically (¹H NMR) homogeneous materials, unless otherwise stated. ¹H and ¹³C NMR spectra were recorded at 298 K on a Bruker Avance III HD 600 MHz NMR spectrometer equipped with a 5 mm CPDCH CryoProbe and calibrated using residual undeuterated solvent for ¹H NMR [CHCl₃: δ_H = 7.26 ppm; (CHD₂)(CD₃)SO: δ_H = 2.50 ppm], and deuterated solvent for ¹³C NMR [CDCl₃: δ_C = 77.16 ppm; (CD₃)₂SO: δ_C = 39.52 ppm] as an internal reference. The following abbreviations were used to designate multiplicities: app = apparent, br = broad, s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet. High-resolution mass spectrometry (HRMS) analysis was performed on a Waters Xevo G2-XS TOF, calibrated against sodium formate clusters and using a LeuEnk lockmass, by the Automated Synthesis Facility at The Scripps Research Institute using electrospray ionization (ESI) methods.

Cell Culture.

A list of the cells used in this study and their respective sources can be found in SI Appendix, Table S2. All fibroblast lines were cultured in high glucose DMEM (Gibco) supplemented with 10% fetal bovine serum, GlutaMAX (Gibco), penicillin-streptomycin (Gibco), and nonessential amino acids (Gibco). HeLa, U251, Lenti-X 293T, and ATF4-FLuc were cultured in high glucose DMEM (Gibco) supplemented with 10% fetal bovine serum, penicillin-streptomycin (Gibco), and sodium pyruvate (Gibco).

Lentivirus Preparation.

shRNA encoding lentivirus plasmids were purchased from Dharmacon. Lentiviruses were packaged in Lenti-X 293T and were concentrated using the Lenti-X concentrator (Takara). Concentrated lentiviruses were tittered and normalized using the qPCR Lentivirus Titer Kit (Applied Biological Materials). Fibroblasts were transduced overnight with nontargeting shRNA lentivirus or a mixture of three shRNA: TFEB (RHS4430-200289329, RHS4430-200303243, RHS4430-200289350), TFE3 (RHS4430-200186455, RHS4430-200235637, RHS4430-200237825), or PIKfyve (RHS4430-200249616, RHS4430-200241483, RHS4430-200245797). Cells were then selected with puromycin for 4 d before being used for experiments.

Confocal Microscopy.

For TFEB nuclear localization studies, HeLa cells were plated on poly-L-lysine (Sigma-Aldrich) coated glass coverslips in a 24-well at a density of 25,000 cells per well and allowed to attach overnight. The next day, cells were treated with small molecules dissolved in DMSO (final DMSO concentration = 0.1%) for 20 h. For TFE3 nuclear localization studies, fibroblasts were plated at a density of 50,000 cells per well and treated with small molecules for 24 h. After treatment, cells were fixed in 4% PFA (Electron Microscopy Sciences) for 30 min at 37 °C. Cells were permeabilized in 0.1% Triton-X 100 for 5 min at RT and blocked with 10% donkey serum (Sigma-Aldrich) and 3% IgG-free BSA (Jackson Immuno Research) for 30 min at RT. Cells were stained with primary antibody overnight at 4 °C, stained with AlexaFluor 546 secondary antibody for 1 h at RT, and stained with DAPI (0.1 to 1 µg/mL, Life Technologies) for 10 min at RT. See SI Appendix, Table S3 for antibodies and dilutions. Cells were mounted on glass cover slides using ProLong Diamond Antifade Mountant (Life Technologies) for 1 h at RT and sealed with clear nail polish (Electron Microscopy Science). Slides were imaged on a Zeiss LSM 780 Confocal Laser Scanning Microscope or Nikon Confocal A1 Microscope. Nuclear localization was quantified using Fiji (ImageJ).

Brightfield Microscopy.

For observation of PIKfyve-related vacuolization, U251 cells were plated in a 24-well at a density of 25,000 cells per well and allowed to attach overnight. The next day, cells were treated with small molecules dissolved in DMSO (final DMSO concentration = 0.1%) for 24 h, after which they would be imaged on an Olympus CKX53 microscope with a DFK23UX236 camera (Imaging Source).

Cell Assays.

All lysosomal enzyme assays were adapted from previous literature (29, 74–78). Experimental conditions were done in triplicate and done in at least three unique cell passages. Fibroblasts were seeded at a density of 10,000 cells per well on a 96-well plate and allowed to attach overnight. The next day, cells were treated with small molecules dissolved in DMSO (final DMSO concentration ≤ 0.2%) for the appropriate time frame. Cells were lysed in-plate with 25 µL of cOmplete Lysis-M (Roche) for 10 min at RT, after which they would be assayed for lysosomal enzyme activity in-plate.

Reagents for lysosomal enzyme assays were as follows. CBE buffer: 200 mM sodium acetate/acetic acid, pH 5; 2 mM CBE (MedChem Express). 4-MUG buffer: 200 mM sodium acetate/acetic acid, pH 5; 0.15% Triton-X 100; 0.15% sodium taurodeoxycholate (Sigma-Aldrich); 2.5 mM 4-methylumbelliferyl-β-D-glucopyranoside (4-MUG, Sigma-Aldrich). Lalistat2 buffer: 200 mM sodium acetate/acetic acid, pH 4; 10 µM Lalistat2 (MedChem Express). 4-MUP buffer: 200 mM sodium acetate/acetic acid, pH 4; 0.5 µg/mL Cardiolipin (Avanti Polar Lipids); 0.00375% Triton-X 100; 1 µM 4-methylumbelliferyl-palmitate (4-MUP, Cayman Chemical). 4-MUGal buffer: McIlvaine buffer, pH 4.5, 100 mM N-acetyl-D-galactosamine (NAG, Biosynth), 5 mM 4-methylumbelliferyl-α-D-galactopyranoside (4-MUGal, Sigma-Aldrich). 4-MUaGlu buffer: McIlvaine buffer. pH 4.3, 10 µM Acarbose (Biosynth); 2.5 mM 4-methylumbelliferyl-α-D-glucopyranoside (4-MUaGlu, Biosynth). 4-MUaMan buffer: McIlvaine buffer, pH 4.3, 5 mM 4-methylumbelliferyl-α-D-mannopyranoside (Sigma-Aldrich). 4-MUbGal buffer: McIlvaine buffer, pH 4.5, 0.8 mM 4-methylumbelliferyl-β-D-galactopyranoside (4-MUbGal, Biosynth).

For measuring GCase/Lysosomal acid lipase activity, nonspecific enzyme activity wells were treated with 25 µL CBE/Lalistat2 buffer (total activity wells were treated with 25 µL vehicle equivalent) and incubated at 37 °C for 30 min. All wells were treated with 50 µL 4-MUG/4-MUP buffer and incubated at 37 °C for 1 to 6 h. Reaction was quenched by addition of 150 µL 2 M glycine-NaOH, and fluorescence was measured using a plate reader at 350/450 nm. Lysosomal enzyme activity was calculated by subtracting average nonspecific enzyme activity from total enzyme activity wells. For measuring α-galactosidase/acid α-glucosidase/lysosomal α-mannosidase/lysosomal β-galactosidase, wells were treated with 25 µL 4-MUGal/4-MUaGlu/4-MUaMan/4-MUbGal buffer and incubated at 37 °C for 1 to 6 h. Reaction was quenched by addition of 75 µL 2 M glycine-NaOH, and fluorescence was measured using a plate reader at 350/450 nm.

ATF4 activity assay was adapted from previous literature (49). Experimental conditions were done in triplicate and done in at least three unique cell passages. HEK293T cells stably expressing an ATF4-Firefly Luciferase reporter were plated at a density of 5,000 cells per 96-well and allowed to attach overnight. The next day, cells were treated with small molecules dissolved in DMSO (final DMSO concentration ≤ 0.2%) for 24 h, after which they were treated with Firefly Luciferase Assay reagent (TargetingSystems) according to the manufacturer’s instructions.

For cell toxicity/viability, fibroblasts were seeded at a density of 10,000 cells per well on a 96-well and allowed to attach overnight. The next day, cells were treated with small molecules dissolved in DMSO (final DMSO concentration = 0.1%) for 72 h. Nuclear stain cell viability was performed by treating cells with Hoechst (Invitrogen) and SYTOX^TM Orange (Invitrogen) according to the manufacturer’s instructions. ATP level cell viability was performed by treating cells with CellTiter-Glo® (Promega) according to the manufacturer’s instructions.

RNA Extraction, RT-PCR, qPCR, and RNAseq.

RNA extraction was performed using the Direct-zol RNA microprep kit (Zymo) according to the manufacturer’s instructions. RT-PCR was performed using ProtoScript II (NEB) according to the manufacturer’s instructions. qPCR was performed using Power SYBR Green PCR Master Mix (ThermoFisher) according to the manufacturer’s instructions. A list of qPCR primers can be found in SI Appendix, Table S4.

Whole transcriptome RNA was prepared and sequenced by BGI Americas on the BGI Proprietary platform, which provided paired-end 50 bp reads at 20 million reads per sample. Each treatment group was performed in triplicate. RNA-seq reads were aligned using DNAstar Lasergene SeqManPro to the Homo_sapiens-GRCh38.p7 human genome reference assembly, and assembly data were imported into ArrayStar 12.2 with QSeq (DNAStar Inc.) to quantify the gene expression levels and normalization to reads per kilobase per million. Differential expression analysis was assessed using DESeq2 in R, which calculated statistical significance of compound-treated cells compared to DMSO-treated cells, using a standard negative binomial fit of the reads per kilobase per million data to generate fold-change quantifications. The complete RNAseq dataset is deposited in gene expression omnibus (GEO) as GSE263252 (79).

GCase Activity-Based Probe and Glycosidase Studies.

Gaucher fibroblasts were seeded at a density of 500,000 cells per well on a 6-well plate and allowed to attach overnight. The next day, cells were treated with small molecules dissolved in DMSO (final DMSO concentration ≤ 0.2%) for 72 h, after which they were collected. If activity-based probes were applied, cells were treated with 10 nM EW644 (generous gift from Dr. Herman Overkleeft) for 2 h prior to being collected, unless specified otherwise. Cell pellets were lysed by sonication in 25 mM potassium-phosphate buffer with 0.1% Triton-X 100 and protease inhibitor cocktail.

For ABP studies, ABP-conjugated lysate was run on a 10% SDS-PAGE and imaged on a ChemiDoc MP (BioRad) using the Cy5 preset. After imaging, the gel was stained for total protein using GelCode Blue Stain Reagent according to the manufacturer’s instructions and imaged on a ChemiDoc MP (BioRad) using the Cy5 or Coomassie preset.

For glycosidase studies, lysate was subjected to EndoH or PNGaseF (NEB) treatment according to the manufacturer’s instructions. Lysate was run on a 10% SDS-PAGE, transferred to PVDF membrane, and blotted for GCase with GAPDH or Vinculin as loading control. See SI Appendix, Table S4 for antibodies and dilutions. Western blots were imaged on a ChemiDoc MP or a LI-COR flatbed imager. Band intensities were quantified using ImageJ.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Data, Materials, and Software Availability

RNASeq data have been deposited in Gene Expression Omnibus (GSE263252) (79).