Research ArticleNIEMANN-PICK TYPE C DISEASE

Chronic administration of an HDAC inhibitor treats both neurological and systemic Niemann-Pick type C disease in a mouse model

See allHide authors and affiliations

Science Translational Medicine  17 Feb 2016:
Vol. 8, Issue 326, pp. 326ra23
DOI: 10.1126/scitranslmed.aad9407

Crossing the Rubicon

Histone deacetylase inhibitors (HDACi) are a class of compounds that have shown promise for treating certain types of cancer. Because HDACi do not cross the blood-brain barrier (BBB), they cannot enter the brain and therefore are of limited value for treating neurological diseases. Now, Alam et al. have developed a new formulation that enables the FDA-approved HDACi vorinostat to cross the BBB and enter the brain tissue of mice with Niemann-Pick type C disease. The authors show that long-term treatment with the formulation improves brain function and prolongs mouse survival, suggesting that this new HDACi formulation will be of benefit for treating Niemann-Pick type C disease and perhaps other types of neurological diseases as well.

Abstract

Histone deacetylase inhibitors (HDACi) are approved for treating rare cancers and are of interest as potential therapies for neurodegenerative disorders. We evaluated a triple combination formulation (TCF) comprising the pan-HDACi vorinostat, the caging agent 2-hydroxypropyl-β-cyclodextrin (HPBCD), and polyethylene glycol (PEG) for treating a mouse model (the Npc1nmf164 mouse) of Niemann-Pick type C (NPC) disease, a difficult-to-treat cerebellar disorder. Vorinostat alone showed activity in cultured primary cells derived from Npc1nmf164 mice but did not improve animal survival. However, low-dose, once-weekly intraperitoneal injections of the TCF containing vorinostat increased histone acetylation in the mouse brain, preserved neurites and Purkinje cells, delayed symptoms of neurodegeneration, and extended mouse life span from 4 to almost 9 months. We demonstrate that the TCF boosted the ability of HDACi to cross the blood-brain barrier and was not toxic even when used long term. Further, the TCF enabled dose reduction, which has been a major challenge in HDACi therapy. TCF simultaneously treats neurodegenerative and systemic symptoms of Niemann-Pick type C disease in a mouse model.

INTRODUCTION

Histone deacetylase inhibitors (HDACi) are an important class of emerging therapeutics, approved for treating three rare cancers (14). They elicit complex cellular responses by blocking HDAC enzymes, thus promoting acetylation of both histones and nonhistone proteins (5, 6). In selected genetic disorders, HDACi-induced histone modifications not only can result in increased or decreased transcriptional expression of mutated genes (7) but also confer indirect benefits through acetylation of nonhistone proteins (such as transcription factors and heat shock proteins) that modulate chaperones and proteostatic networks (79). Because of their broad effects on transcription, maximizing efficacy while limiting the dose is a major challenge in HDACi therapy. We were interested in developing and validating a therapeutic strategy that would enable lowering of the HDACi dose while still effectively treating both systemic and cerebral Niemann-Pick type C (NPC) disease. This disease presents additional challenges because it requires effective HDACi penetration across the blood-brain barrier (BBB) but nonetheless must allow HDACs in the brain to function. In particular, HDAC3 activity is essential for Purkinje cell function (10). Here, we have explored a triple combination formulation (TCF) of the HDACi vorinostat in a murine model of NPC with both cerebral and systemic defects that closely mimic the human disease (11).

NPC is a rare autosomal recessive neurodegenerative disease caused by a defect in either NPC1 or NPC2 genes (12), but 95% of cases are due to a defect in NPC1. Both NPC1 and NPC2 proteins have physiological function in the transport of cellular cholesterol (1315). Cells with defects in these genes accumulate cholesterol primarily in the late endolysosomal system because of a block in cholesterol transport from the lysosome to the endoplasmic reticulum (16, 17). Insertion of a point mutation in the Npc1 gene that blocks cholesterol transport in cells confers a similar neurodegenerative disease phenotype in a mouse model, providing definitive molecular evidence that aberrant NPC1 protein function causes the disease (18). Recent studies suggest that in the central nervous system (CNS), NPC1 protein is essential for myelination (19). In clinical disease, progressive neurodegeneration is a hallmark of NPC (20). Disease progression can be heterogeneous and neurodegenerative decline may span one to two decades, but once initiated is fatal (12). Systemic symptoms such as splenomegaly and hepatomegaly usually appear first followed by neurocognitive and neuromuscular degeneration (20).

The only available treatment for NPC is miglustat, an imino sugar marketed under the trade name Zavesca (21). It was developed to treat Gaucher’s disease, another lysosomal disorder that arises from the accumulation of glycosphingolipids (22). Zavesca is approved for NPC treatment in Europe, Canada, and Japan but was denied Food and Drug Administration (FDA) approval in the United States. Zavesca is, however, prescribed off-label in the United States. It may confer a mild improvement in clinical neurological symptoms but fails to prevent disease progression (23, 24). 2-Hydroxypropyl-β-cyclodextrin (HPBCD) is being studied as an emerging therapy. It chelates cholesterol and has therefore been proposed as a potential therapy for NPC (25), but it does not cross the BBB (26). Therefore, systemic delivery primarily benefits the liver and other organ systems of the body cavity (25, 27), whereas direct delivery into the CNS is needed for substantial neurological improvement (28, 29). However, CNS delivery of HPBCD increases the procedural risk for lifelong therapy, is associated with hearing loss (30), and provides little or no benefit for systemic disease. Here, we test a therapeutic approach to integrate treatment of both cerebral and systemic defects in a mouse model of NPC.

RESULTS

Vorinostat is active in fibroblasts from Npc1nmf164 mice but fails to improve animal survival

Cell-based studies (31, 32) have previously shown that HDACi reduces cellular cholesterol in cultured fibroblasts from NPC patients with a concomitant increase in NPC1 expression, but the effects of HDACi in animal models have not been investigated. Here, we investigated the effects of HDACi in the Npc1nmf164 mouse on a BALB/c background (33), which was derived from the Npc1nmf164 mouse strain on a C57BL/6J background (11). The Npc1nmf164 mouse carries an aspartate-to-glycine change at position 1005 (D1005G) of NPC1. This mutation destabilizes NPC1, consistent with protein misfolding and proteostatic degradation, and thus provides a powerful model for assessing therapies targeting NPC1 mutations that compromise the stability of the protein. Disease progression in this model (monitored over ~120 days) is manifested independently of mouse genetic background, closely mimicking the human disease where neurodegeneration is the principal cause of death (11, 33).

Skin fibroblasts from mutant Npc1nmf164 mice expressed less NPC1 protein compared to their heterozygous or wild-type counterparts (Fig. 1A). Mouse mutant NPC fibroblasts accumulated high levels of cholesterol that were reduced by treatment with vorinostat (Fig. 1B), confirming their cellular responsiveness to HDACi therapy. To treat the animals, we selected a conservative dose of vorinostat (50 mg/kg), lower (by 100 to 200%) than the dose used to treat murine models of cancer (3436). Allometric scaling translated 50 mg/kg in mice to 150 mg/m2 in mice, which was well below the total weekly human intravenous pediatric dose of 396 mg/m2 (37). Because we expected to monitor survival over several months, injection frequency was limited to once weekly. This also enabled a desired (weekly) rest period because continuous HDAC inhibition is expected to be detrimental to neurological (especially cerebellar) function (10). Vorinostat in polyethylene glycol (PEG) 400 was first administered at day 21 after weaning and was maintained once weekly through the animal’s life span, with no significant benefit found for animal survival (Fig. 1C). Increasing the dose by twofold to 100 mg/kg in mice also conveyed no survival benefit (Fig. 1C), suggesting that despite vorinostat’s activity in cultured cells, weekly doses approaching those in pediatric patients were insufficient to reduce neurological disease in mice even after 4 months of treatment.

Fig. 1. Effects on Npc1nmf164 mutant mice of vorinostat alone or in TCF.

(A) Western blots of NPC1 protein in skin fibroblasts from heterozygous Npc1+/nmf164 control mice and Npc1nmf164 (Npc) mutant mice. Loading control, α-tubulin. Molecular weight markers in kilodaltons. (B) Filipin-stained skin fibroblasts from Npc1nmf164 mice treated with 5 μM vorinostat (Vo) for 48 hours showing a decrease in cholesterol. n = 3 replicates. Scale bar, 40 μm. (C) Kaplan-Meier survival curves. Npc1nmf164 mice (number of mice shown in parentheses) were given once-weekly intraperitoneal injections of vorinostat in PEG at 50 mg/kg (Vo, 1×) or 100 mg/kg (Vo, 2×), PEG alone, or were left untreated. The median survival of each group is indicated in days (d). (D) Schematic of the TCF. (E) Western blots show acetylation (Ac) of histones 3 and 4 (H3 and H4) in the brain of Npc1nmf164 mice within 60 min after administration of TCF, vorinostat, HPBCD, or vehicle control. n = 3 replicates (each with 3 mice). Coomassie brilliant blue (CBB)–stained gel (blue) confirmed equal sample loading. (F) Quantitation of the blots shown in (E). Fold change relative to vorinostat (set at 1). *P < 0.05, Student’s t test.

TCF enhances vorinostat’s ability to cross the BBB

Vorinostat is poorly soluble in aqueous solution and therefore classified as a Biopharmaceutical Classification System (BCS) class 4 drug (www.accessdata.fda.gov/scripts/cder/drugsatfda/). We reasoned that its rapid clearance from plasma could limit effective penetration through the BBB. We therefore developed a formulation where vorinostat (50 mg/kg) in PEG was complexed with HPBCD (2000 mg/kg) (in a final molar ratio of 0.13 vorinostat/HPBCD) to create a TCF (Fig. 1D). We selected HPBCD because it complexes with hydrophobic compounds to enhance their solubility and bioavailability (38, 39). In addition, when delivered systemically, although it does not cross the BBB, HPBCD reduces the liver lipid burden and, at high concentrations of 4000 mg/kg, also partially reduces markers of neuroinflammation (25, 27, 40, 41). HPBCD is therefore expected to benefit the liver and other organs through indirect mechanisms and to complement vorinostat’s direct effects. PEG was retained to facilitate the release of vorinostat from HPBCD and improve bioavailability. PEG is known to reduce BBB inflammation in cerebral injury (42) and may also provide benefit as part of the TCF, but the mechanism is unknown.

The TCF containing vorinostat, HPBCD, and PEG was designed to be administered systemically to treat both neurological and systemic disease. Within an hour of administration intraperitoneally, neither vehicle control [PEG + dimethyl sulfoxide (DMSO)] nor HPBCD stimulated acetylation of either histone H3 or H4 in the mouse brain (Fig. 1, E and F). Vorinostat (50 mg/kg) in PEG conferred low levels of acetylation, but upon administration of TCF, acetylation was stimulated to two- to threefold (P < 0.05) for histone H3 and five- to ninefold (P < 0.05) for H4 (Fig. 1, E and F). The basal levels of acetylated histones H3 and H4 in the mouse brain were the same for Npc1nmf164 mice and healthy heterozygous mice (fig. S1). These data establish that the TCF stimulated vorinostat’s acetylation activity in the mouse brain.

TCF promotes animal survival

To compare the effects of long-term treatment, mice were given a once-weekly intraperitoneal dose of TCF or one of the following solutions: HPBCD (2000 mg/kg), vorinostat (50 mg/kg) in PEG, or vehicle control (PEG + DMSO). Comparative analyses of animal tissues were undertaken at day 100 of age because prior studies suggested that this was a period of symptomatic disease (untreated Npc1nmf164 mice died by ~125 days) (11). In the mouse brain, TCF stimulated increased expression of calbindin, a marker of both Purkinje cell bodies and Purkinje cell neurites that extend to the cerebellar molecular layer (Fig. 2A). Vorinostat alone or HPBCD treatment resulted in a decrease or no change in expression of calbindin transcripts. As shown by indirect immunofluorescence assays, TCF treatment preserved 25 to 30% of Purkinje cell expression of calbindin (P < 0.001) compared to untreated mice or mice treated with vorinostat alone (Fig. 2B). Nissl staining, which labels neurons, confirmed that 25 to 30% of Purkinje cells remained intact in the TCF-treated mice but not in the untreated mice or mice treated with vorinostat (fig. S2). HPBCD confers a small amount of protection in Purkinje cells, the mechanism for which is unknown (25, 27, 41). Analyses of three inflammatory markers, Gfap (glial fibrillary acidic protein), Mip1α (macrophage inflammatory protein 1α), and Cd68, suggested that HPBCD reduced their expression comparable to that by TCF and better than vorinostat alone in the mouse brain (fig. S3). This is consistent with prior studies showing that HPBCD is partially effective in reducing neuroinflammation (25, 27).

Fig. 2. TCF preserves cerebellar Purkinje cells and promotes animal survival.

(A) Calbindin1 transcripts in Npc1nmf164 mouse brain at 100 days of age in different treatment groups expressed as a fraction of the levels seen in Npc1+/nmf164 control healthy mice (control, set at 100%). Each group contained four to five mice. (B) Fluorescence microscopy detection of calbindin (green) in Purkinje cells and neurites in the cerebellum of Npc1+/nmf164 mice (healthy) or Npc1nmf164 (Npc) mutant mice in different treatment groups at 100 days of age. Purkinje cells (white arrows), neurites in the molecular cell layer (mcl), and the granule cell layer (gcl) are indicated. Blue, 4′,6-diamidino-2-phenylindole (DAPI). Scale bar, 40 μm. Micrographs are representative images of cerebellar lobule IX. Quantification of Purkinje cells across all cerebellar lobules from four mice (four sections per mouse) is shown in the bar graph (represented as a fraction of Purkinje cells relative to control healthy mice.). (C to E) Kaplan-Meier survival curves of untreated and treated Npc1nmf164 mutant mice (both males and females) (C), male Npc1nmf164 mutant mice (D), and female Npc1nmf164 mutant mice (E). Median survival (in days) is indicated. Log-rank test: *P < 0.001, TCF versus 2× HPBCD; **P < 0.05, TCF versus 2× HPBCD. Un, untreated.

We next assessed whether improvement in cerebral pathology could be correlated with improved survival. As shown in Fig. 2C, the median life span of mice treated with the TCF was ~200% longer than that of animals treated with vorinostat in PEG (254 versus 134 days; P < 0.001). HPBCD supplemented with DMSO and PEG (fig. S4) or even twice the dose of HPBCD (4000 mg/kg, double the dose used in TCF) were not as effective as TCF [TCF versus HPBCD + vehicle, 254 versus 192 days (P < 0.001); TCF versus 2× HPBCD, 254 versus 180 days (P < 0.001)] (Fig. 2C and fig. S4). Treatment with vorinostat or vehicle alone showed no significant survival benefit. The TCF was equally effective in both male and female mice, with comparable median survival for males (249 days) and females (258 days) (Fig. 2, D and E). Mice treated with the TCF survived until 9 or 10 months of age, which represents advanced adulthood.

TCF delays the onset and progression of neurobehavioral disease in Npc1nmf164 mice

Clinically, NPC disease is defined by major and minor symptomatic domains, whose severity has been scored to monitor the natural history of the disease using at least three different scales (4346). Plasma biomarkers are being investigated (33, 47, 48), but symptom scoring continues to be an important index of disease progression. We extended a previously described murine neurobehavioral symptomatic score (49) to create a disease severity scale for murine NPC (table S1). Each of the six symptomatic parameters (tremor, body position, gait, grooming, limb tone, and weight) in the mouse was assigned to a major patient disease domain (ambulation, cognition, motor control, and dysphagia) and scored for severity in an indicated range. The sum of the individual scores provided the cumulative disease score, with a maximum possible disease score of 13 (Fig. 3A).

Fig. 3. TCF delays the onset and progression of neurodegeneration in Npc1nmf164 mutant mice.

(A) Mouse disease symptoms equivalent to major human NPC1 disease domains (ambulation, cognition, motor control, and swallowing) were quantitatively scored as described in table S1. Cumulative symptom scores are shown as disease progression curves for Npc1nmf164 mutant mice with a score range of 0 to 13 in different treatment groups. The mice were assessed once every 2 weeks starting at 21 days of age and up to euthanasia (that is, at 30% weight loss). *P < 0.05, treated versus untreated; **P < 0.05, TCF versus 2× HPBCD (Student’s t test). (B) Bar diagrams show onset and progression of individual symptoms in untreated and drug-treated mice (time in days on y axis). *P < 0.05, TCF versus 2× HPBCD; **P < 0.01, TCF versus 2× HPBCD (Student’s t test).

Validation of scoring by independent blinded operators in both diseased and healthy animals is shown in fig. S5. A cumulative score of 3 or higher was found to reliably flag onset of symptomatic disease. A threshold of 3 was encountered because older healthy animals often displayed poor grooming (particularly males) and slight impairment in limb tone (from 100 to 140 days). It was nonetheless considered acceptable because these symptoms arise much earlier in diseased animals. Using these criteria, an early cumulative disease score of 4 to 5 reliably detected the onset of symptomatic disease in untreated animals at 77 to 84 days (Fig. 3A). TCF treatment appeared to delay disease onset by about 4 weeks, reaching scores of 4 to 5 at 105 to 112 days (Fig. 3A). At this time (105 to 112 days), mice treated with vehicle or vorinostat alone showed cumulative scores of 9 to 11, whereas those treated with 2× HPBCD yielded intermediate cumulative scores of 6 to 8 (Fig. 3A). Analysis of individual symptomatic domains revealed that worsening in gait, grooming, limb tone, and weight were all delayed in animals treated with the TCF (Fig. 3B). Worsening in gait, grooming, and weight were also delayed by 2× HPBCD, but less than that by the TCF. Vorinostat alone provided no consistent advantage in any symptomatic readout in the context of life span. These data suggested that TCF administration afforded functional benefit to ambulation, cognition, motor control, and dysphagia, major symptomatic domains in neurological disease. In particular, the animals maintained their weight even at terminal stages of the disease and seemed to retain their ability to drink water (in murine NPC, the terminal disease is marked by dehydration seen as weight loss).

TCF reduces liver and spleen inflammation and is not toxic

We also examined the consequences of treatment on the liver (Fig. 4) as an example of an organ outside of the BBB and also because liver disease is prominent in murine NPC. Histological analyses at 100 days of age suggested that TCF reduced macrophage recruitment in the liver to the same degree as HPBCD (Fig. 4A). Vorinostat alone also showed this effect, but it was less marked than for TCF or HPBCD. TCF reduced the expression of the inflammatory markers Cd68, Itgax (integrin αX), Mip1α, and Ctsd (cathepsin D) associated with inflammatory cells, comparable to HPBCD (although the presence of vorinostat in TCF appears to further reduce Cd68 transcripts) (Fig. 4, B to E). Vorinostat alone also had anti-inflammatory activity. Similar data were obtained in the spleen (Fig. 4, F to H), a second organ involved in human NPC disease. Moreover, long-term administration (200 to 300 days) of TCF did not result in toxicity in key organs such as liver and kidney of healthy mice (Table 1 and fig. S6). These data suggest that vorinostat needs to be administered in the TCF to simultaneously treat neurological and systemic disease and improve animal survival.

Fig. 4. Efficacy of TCF in reducing liver and spleen inflammation.

(A) Fluorescence detection of macrophages (red; indicated by white arrows) at 100 days of age in the liver of Npc1+/nmf164 (healthy control) and Npc1nmf164 (Npc) mutant mice treated as indicated. Liver sections (4 to 5 μm) were probed with antibodies to cathepsin S and tetramethyl rhodamine isothiocyanate–conjugated secondary antibodies. Foamy macrophages were abundant and often clustered in untreated Npc mice. Treatment with vorinostat reduced macrophage clustering. Macrophages were barely detected in HPBCD-treated and TCF-treated Npc mice. Red, cathepsin S; blue, DAPI. Scale bar, 40 μm. (B to H) Quantitative polymerase chain reaction (qPCR) analysis of Cd68 (macrophage marker) and other inflammatory markers at 100 days of age in the liver (B to E) and spleen (F to H) of Npc1nmf164 mutant mice treated as indicated. Fold change shown is relative to the levels of transcripts detected in untreated Npc1+/nmf164 (healthy control) mice (set at 1). Each group contained four to five mice.

Table 1. Plasma markers of liver and kidney toxicity.

Healthy control Npc1+/nmf164 mice were treated with TCF, and various plasma markers of liver and kidney toxicity were analyzed. The ages of the mice are shown. ALT, alanine aminotransferase; AST, aspartate aminotransferase; TP, total protein; ALB, albumin; BUN, blood urea nitrogen; Na, sodium ions; Cl, chloride ions; wk, weeks; d, days.

View this table:

TCF acts by target-based mechanisms in the brain

To investigate a mechanistic basis for the observed effects of TCF, we compared vorinostat concentrations in plasma and brain. Within 1 hour of intraperitoneal injection, TCF treatment resulted in about two- to threefold (P < 0.05) greater plasma levels of drug than that seen in animals treated with vorinostat alone (Fig. 5A). Pharmacokinetic analyses revealed greater exposure of mouse brain tissue to vorinostat for up to 2 hours in animals treated with TCF versus vorinostat alone (Fig. 5B).

Fig. 5. TCF increases vorinostat concentrations in mouse plasma and brain.

(A) Npc1+/nmf164 healthy control mice were injected intraperitoneally with vorinostat alone or TCF; plasma was drawn after 1 hour and analyzed for vorinostat concentrations. The data represent mean ± SEM from two independent experiments (five mice per group in each experiment). *P < 0.05, TCF versus vorinostat (Student’s t test). (B) Pharmacokinetics of vorinostat in mouse brain. Npc1+/nmf164 healthy control mice were injected intraperitoneally with vorinostat or TCF. Animals were sacrificed at the indicated time points, perfused with phosphate-buffered saline, and their brains were harvested. n = 10 mice at 0.5 hour; n = 5 mice for the remaining time points. *P < 0.05, TCF versus vorinostat (Student’s t test). The concentration of vorinostat was determined by mass spectrometry.

Because disease progression extends over 120 days (in the absence of treatment), we examined whether drug action was sustained up to 100 days of age (time of late-stage disease in untreated animals). Mice treated with TCF (from the onset of weaning at 21 days) showed elevated Npc1 transcripts in the liver and brain at 100 days of age (Fig. 6, A and B). As expected, HPBCD alone had no effect on the expression of target Npc1 transcripts in either tissue, and vorinostat in PEG had no measurable effect in the brain. Therefore, although vorinostat alone may stimulate low levels of histone acetylation in the brain (Fig. 1, E and F), this was insufficient for activation of transcriptional expression of Npc1 that would be needed for long-term benefit. Because a deleterious effect of reduced HDAC activity on Purkinje cells and cerebellar function has been reported in the literature (10), we examined NPC1 protein in the mouse brain and its expression in the cerebellum of TCF-treated mice (Fig. 6, C and D). Western blots (Fig. 6C) and immunostaining (Fig. 6D) showed that mouse brain NPC1 protein expression increased about eightfold (Fig. 6C), with about a fivefold increase in NPC1 in cerebellar Purkinje cells (Fig. 6D) in TCF-treated mice compared to untreated animals at 100 days of age. The amount of NPC1 protein was 25% of that found in healthy control mice.

Fig. 6. Mechanism of TCF action.

(A and B) Npc1 transcripts in the (A) liver and (B) brain of Npc1nmf164 mutant mice at 100 days of age in different treatment groups expressed as fold change relative to Npc1+/nmf164 healthy control mice (control set at 1). n = 4 to 5 mice. *P < 0.05, TCF versus HPBCD (Student’s t test). (C) Western blots of NPC1 protein in the brains of Npc1+/nmf164 healthy control mice or Npc1nmf164 (Npc) mutant mice treated with TCF or untreated at 100 days of age, quantified in the bar graph. **P < 0.005, Student’s t test. Loading control, α-tubulin (α-tub). Molecular weight markers in kilodaltons. n = 3 replicates. (D) NPC1 protein (green) in Purkinje cells of Npc1+/nmf164 healthy control mice or Npc1nmf164 (Npc) mutant mice at 100 days of age, either untreated or treated with TCF. White arrows indicate Purkinje cells. Micrographs are representative images of cerebellar lobule IX. Blue, DAPI. Scale bar, 15 μm. Bar graph shows the quantitative fold change in NPC1 protein detected in 16 sections from 4 TCF-treated mice and 8 sections from 2 untreated mice. Healthy control Npc1+/nmf164 set at 100%. ***P < 0.0001, Student’s t test. (E) Microglial cells (green) in the hippocampus of Npc1+/nmf164 healthy control mice or Npc1nmf164 (Npc) mutant mice treated with TCF or untreated. White arrows indicate microglial cells. Blue, DAPI. Scale bar, 40 μm. Bar graph shows quantitative fold change in four sagittal sections from two mice (two sections per mouse) per group. **P < 0.005, Student’s t test. Npc1+/nmf164 healthy control mice set at 100%.

Hippocampal inflammation was also reduced (Fig. 6E), suggesting additional broader benefits in the brain. These data suggested that there was a modest (about twofold) increase in transcriptional activation, but NPC1 protein expression was amplified about eightfold in mouse brain tissue. This was likely due to vorinostat’s broad action through acetylation of both histones and nonhistone proteins (such as heat shock proteins) (79), which may indirectly have enhanced NPC1 protein expression.

Because TCF treatment failed to reveal toxicity even after long-term administration (Table 1 and fig. S6), we further investigated its potential for activating transcription of additional genes, missense mutations of which are known to cause neurological disease. TCF administration stimulated transcription of genes encoding glucocerebrosidase (Gba), galactosylceramidase (Galc), and the lysosomal acid galactosidase β (Glb1) in mouse brain tissue; defects in these genes cause Gaucher’s disease (50), Krabbe’s disease (51), and GM1 gangliosidosis, respectively (52) (Fig. 7, A to C). As seen for Npc1, there was a modest ~1.3- to 1.8-fold increase in Gba, Galc, and Glb1 transcript levels in response to TCF, which was not found with vorinostat alone.

Fig. 7. TCF elevates brain transcripts of genes mutated in other neurodegenerative disorders.

qPCR analyses of fold changes in transcripts in the brains of Npc1+/nmf164 healthy control mice, either untreated or treated with either vorinostat alone or TCF. (A to C) Gba gene encoding glucocerebrosidase (Gaucher’s disease) (A), Galc gene encoding galactosylceramidase (Krabbe’s disease) (B), and Glb1 gene encoding galactosidase β1 (gangliosidosis) (C). Brains were analyzed at 1 hour after intraperitoneal injection. n = 5 mice for each group. *P < 0.05, TCF versus untreated (Student’s t test).

DISCUSSION

The results described here provide evidence for a model (fig. S7) in which systemic delivery of TCF increases plasma and brain concentrations of the HDACi vorinostat, which, together with HPBCD, improves cerebral and systemic disease in a mouse model of NPC.

A major drawback to the use of HDACi is their intrinsic toxicity because they may influence transcription of a large number of genes and hence many cellular pathways. This also holds for inhibitors designed to be specific for a given HDAC because even a single HDAC can regulate hundreds of genes (and hence the value of synthesizing selective HDACi has been debated). Our data show that the pan-HDACi vorinostat, through a new formulation that improves its access to the brain (fig. S7), coupled with a rest period (when no drug was injected), enhanced NPC1 protein expression, restored cerebellar function, reduced cerebral inflammation, and delayed most symptoms of NPC disease in a mouse model. Although HDACs may be essential global regulators, our study suggests that a low intermittent dose of HDACi may result in a modest change in histone acetylation in the brain without metabolic toxicity. Indeed, small transient increases in promiscuous transcription appear to be well tolerated even when induced repeatedly. When these increases are induced in a monogenetic disorder, they result in increased transcription of the mutant gene and the protein it encodes, thereby helping to ameliorate disease.

Vorinostat received FDA exemption for an exploratory phase 1 study for NPC, and the National Institutes of Health is currently accruing NPC patients 18 years and older (https://clinicaltrials.gov/ct2/show/NCT02124083). Vorinostat alone does not permeate the mouse brain sufficiently to stimulate (either directly or indirectly) Npc1 transcription (or other gene transcription) or protein expression in the brain and hence does not affect Purkinje cell function in the cerebellum. Stimulation of transcription through acetylation activity in the brain requires parenteral administration of vorinostat in a formulation that will enhance systemic exposure rather than oral administration of the drug. Testing of systemic and brain tissue exposure to vorinostat required a robust animal model wherein disease progression was not compromised by the immunological background of the mouse (53).

HPBCD injected into the CNS is also being evaluated as a therapy for NPC. In phase 1 studies, Ommaya reservoirs implanted in the brain to directly deliver the drug were discontinued (www.nnpdf.org/cyclodextrin.html) and replaced with a lumbar puncture (making it difficult to estimate the concentration of drug that reaches the brain). CNS delivery of HPBCD is associated with hearing loss (30), does not treat systemic disease (29) or restore NPC1 protein, and requires highly specialized care providers and facilities, suggesting that this strategy may limit comprehensive treatment. In contrast, the TCF does not have to be delivered into the CNS. The TCF increases NPC1 protein expression and treats both neurological and systemic disease. Because the HDACi vorinostat in TCF can directly stimulate transcription and also (indirectly) increase NPC1 protein expression, it may have greater restorative efficacy in treating neurological disease compared to indirect chaperone therapies (54).

HPBCD is designated as “GRAS” (generally regarded as safe) by the FDA, and no adverse effects have been reported so far about its use in a limited number of NPC patients (55). PEG is also well tolerated. Our current dose of vorinostat of 150 mg/m2 is substantially below the daily adult dose and frequency [600 to 900 mg/m2 daily for 3 to 5 days for 3 weeks for hematological and solid tumors (56)]. The vorinostat dose in TCF is within the weekly pediatric dose but is ~20% higher than currently proposed for the daily pediatric dose (which converts to an intravenous dose of 123 mg/m2 daily for five consecutive days every 28 days for cancer treatment). It is expected that dose modulations will be required for pediatric treatment; the relevant pediatric dose may be accommodated by two consecutive days of half-dose TCF administration followed by a suitable rest period. Preferred routes of treatment may also influence dose and need to be determined, but it is expected that the TCF could be used for treating both adult and pediatric disease in conjunction with other emerging NPC therapies.

There may be applications for TCF treatment for neurological diseases beyond monogenetic disorders. Intraperitoneal administration of HPBCD alone has been shown to be beneficial in a murine model of Alzheimer’s disease (57), but the mechanism is unknown. Improving the functional activity of glucocerebrosidase in the lysosome (the target in Gaucher’s disease, which we show as transcriptionally activated by TCF) may be a therapeutic approach for Parkinson’s disease (by reducing α-synuclein oligomers) and other synucleinopathies (58). Finally, because it increases the amount of vorinostat in plasma, TCF and formulations derived from it could be applied to lower the dose of other HDACi (including panobinostat, which recently received FDA approval). Because injecting twice as much vorinostat alone (Fig. 1) showed none of the benefits of a two- to threefold increase in TCF vorinostat in plasma (Figs. 2 to 4), it is likely that this formulation confers improved tissue penetration on vorinostat, which is important because HDACi dose reduction remains a major challenge in disease (including tumor) therapy.

The development of TCF is an example of a target-based strategy facilitating drug access across the BBB, thereby enabling long-term utilization of safe amounts of an HDACi to simultaneously improve neurological and systemic disease in NPC. Although the studies described herein are highly predictive of a desirable outcome in humans, elucidation of the full potential of TCF as a drug needs data from human studies. In addition, major symptomatic domains in human NPC disease, such as learning disability, vertical supranuclear gaze palsy, seizures, and dysarthria, are not manifest or are difficult to measure using our murine scoring system. Nonetheless, because TCF improved symptoms corresponding to major domains of ambulation, ataxia, cognition, motor function, and swallowing in our murine model and according to our scoring system, our data suggest that TCF may have potential for treating NPC patients.

MATERIALS AND METHODS

Study design

The objectives were to evaluate the efficacy of an HDACi, vorinostat (also referred to as Vo), for the treatment of NPC disease in a mouse model, Npc1nmf164, that closely mimics the human disease. We confirmed that in vitro–grown mouse skin fibroblasts were responsive to vorinostat, but the survival of Npc1nmf164 mutant mice was unaffected by vorinostat alone. However, our analyses showed that administration of vorinostat in a newly designed TCF conferred benefit by increasing vorinostat concentration in plasma and in the mouse brain. Npc1nmf164 mice were treated with TCF, and the effects were determined through biochemical and histological analyses of organs, neurobehavioral data, and survival. Mice were randomly selected for drug treatments. Sample size, number, the composition of replicates, and an intermediate end point of 100 days of age were based on prior knowledge of this mouse model (11, 33). The numbers were not altered during the course of the study. The final end point when all data collection was stopped was a weight loss of greater than or equal to 30% in accordance with Institutional Animal Care and Use Committee (IACUC) policies. All rules were predefined in advance. All data were included, and the criteria were established prospectively. Experiments were performed two or three times (as indicated in the figure legends). Results were substantiated by repetition over 3 years at the University of Notre Dame animal facility. Plasma samples were assessed for vorinostat or toxicity markers by blinding, whereas remaining studies could not be subjected to blinding because there were prominent effects on survival and neurobehavioral outcomes.

Animals

Npc1nmf164, BALB/c strain carrying an aspartate-to-glycine mutation at position 1005 (D1005G) in Npc1 gene was used as an NPC disease model. When diseased animals were unable to reach the food provided on the holder, regular food (2019 Teklad diet, Harlan Laboratories) was replaced with DietGel 76A (Clear H2O). The studies were approved by the IACUC of University of Notre Dame.

Drug preparations and injections

Vorinostat (50 and 100 mg/kg), prepared in 5% DMSO and 45% PEG 400, was given once weekly through intraperitoneal injection starting at day 21. TCF is a mixture of vorinostat (50 mg/kg), HPBCD (2000 mg/kg), PEG 400 (45%), and DMSO (5%). Mice were given two intraperitoneal doses of HPBCD (2000 mg/kg) at 7 and 15 days of age. Starting at day 21, mice were given once-weekly intraperitoneal injections of TCF. 2× HPBCD (4000 mg/kg, double the dose used in TCF)– and HPBCD (2000 mg/kg)–treated mice were given once-weekly intraperitoneal injections of the respective HPBCD dose starting at day 7. Vehicle control (5% DMSO and 45% PEG) was given to mice once weekly through intraperitoneal injection starting at day 21. The injection volume across the treatment group was 10 ml/kg body weight. Mice were sacrificed when they lost ≥30% of the maximum weight or otherwise indicated.

Neurobehavioral assessment of mice

A modified version of the previously described method (49) was used for assessing the neurobehavioral functions of mice. Six different parameters, namely, tremor, body position, gait, grooming, limb tone, and weight, were assessed on a scale of 0 to 13. More specific descriptions of the assessments along with the equivalent human symptoms are provided in Fig. 3A and table S1.

Analysis of vorinostat in plasma and brain

Heterozygous mutant (Npc1+/nmf164) mice (ages 6 to 7 weeks) were injected intraperitoneally with vorinostat (50 mg/kg in 45% PEG) or TCF. Plasma analysis was done 1 hour after injection, whereas the analysis in the brain was done at 0.5, 1, 2, and 4 hours after injection, as described further in the Supplementary Materials.

Statistical analysis

Kaplan-Meier survival curves were plotted using GraphPad Prism, and the log-rank test was undertaken to determine the statistical significance. Unless mentioned, results shown are mean ± SEM. Student’s t test was carried out to determine the statistical significance of the data using two-tail analyses. P < 0.05 is considered significant.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/8/326/326ra23/DC1

Materials and Methods

Fig. S1. Baseline acetylation of histones in mouse brain tissue.

Fig. S2. Nissl staining of mouse brain sections.

Fig. S3. Quantitative analyses of brain inflammation.

Fig. S4. Comparative analysis of survival between mice treated with TCF and those treated with HPBCD + vehicle.

Fig. S5. Comparison of disease severity progression curves by two independent operators.

Fig. S6. Histological analysis of liver from mice treated long-term with TCF.

Fig. S7. Generalized model for TCF in treating cerebral and systemic disease.

Table S1. List of parameters used to score the neurobehavioral functions of Npc mice.

REFERENCES AND NOTES

  1. Acknowledgments: We thank J. Shin, M. Cloutier, and Y. Ryan for assistance with the neurobehavioral scoring; B. Cooper for analysis of vorinostat in plasma and brain; W. Claypool for careful reading of the manuscript. Funding: This study is supported in part by the Parsons-Quinn Endowment, University of Notre Dame, and NPC Therapeutics LLC. Author contributions: M.S.A. conceived and designed the study and its experiments, performed the experiments, analyzed the data, and wrote the paper. M.G. maintained mouse colonies and undertook neurobehavioral scoring, injection, and organ harvest in animal experiments. K.H. conceived and designed the study and its experiments, analyzed the data, and wrote the paper. Competing interests: K.H. holds equity interest in NPC Therapeutics LLC. M.S.A., M.G., and K.H. are coinventors on the following patent entitled “Formulation for the treatment of neurological diseases and cerebral injury,” Provisional Application No. 62/011,553; filed 12 June 2014.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article