Niemann–Pick C1 (NPC1) disease is a rare, neurodegenerative lysosomal cholesterol storage disorder, typified by progressive cognitive and motor function impairment. Affected individuals usually succumb to the disease in adolescence. 2-Hydroxypropyl-β-cyclodextrin (HP-β-CD) has emerged as a promising intervention that reduces lipid storage and prolongs survival in NPC1 disease animal models. A barrier to the development of HP-β-CD and other treatments for NPC disease has been the lack of validated biochemical measures to evaluate efficacy. Here we explored whether cholesterol homeostatic responses resulting from HP-β-CD-mediated redistribution of sequestered lysosomal cholesterol could provide biomarkers to monitor treatment. Upon direct CNS delivery of HP-β-CD, we found increases in plasma 24(S)-HC in two independent NPC1 disease animal models, findings that were confirmed in human NPC1 subjects receiving HP-β-CD. Since circulating 24(S)-HC is almost exclusively CNS-derived, the increase in plasma 24(S)-HC provides a peripheral, non-invasive measure of the CNS effect of HP-β-CD. Our findings suggest that plasma 24(S)-HC, along with the other cholesterol-derived markers examined in this study, can serve as biomarkers that will accelerate development of therapeutics for NPC1 disease.
Niemann–Pick C (NPC) disease is a rare, neurodegenerative disease caused by mutations in either the NPC1 or NPC2 genes. The majority of NPC cases are caused by mutations in NPC1, a late endosomal/lysosomal protein that participates in the export of lipoprotein-derived cholesterol from that compartment (1). NPC1 disease is characterized by accumulation of cholesterol and other lipids in the viscera and central nervous system (CNS) (2,3). Individuals with this disorder often present as neonates with hepatosplenomegaly and jaundice, or in early childhood with ataxia and progressive cognitive and motor function impairment, and usually succumb to the disease in adolescence (3,4). There are currently no FDA-approved therapies for NPC disease, though miglustat, an inhibitor of glycosphingolipid synthesis, is approved for use outside the USA and provides limited benefit (4–6), there are currently no FDA-approved therapies for NPC disease.
NPC1 disease is principally a disorder of cellular cholesterol regulation. Cells harboring inactivating mutations in NPC1 exhibit marked impairment of low-density lipoprotein (LDL) cholesterol esterification and mobilization of newly hydrolyzed LDL cholesterol to the plasma membrane (7–9). As a consequence of these trafficking defects, NPC1 mutant cells demonstrate lysosomal sequestration of LDL cholesterol, delayed down-regulation of the LDL receptor and de novo cholesterol biosynthesis, and impaired ABCA1-mediated cholesterol efflux (10–14). These observations have established the NPC1 protein as a key participant in intracellular sterol trafficking.
Many of the prominent neuropathological features of human NPC disease (e.g. neuronal lipid storage and progressive loss of Purkinje cells) are recapitulated in the BALB/c NPCnih (Npc1−/−) mouse, a spontaneously occurring inbred model that harbors a retroposon insertion in the Npc1 gene (15,16). In the Npc1−/− mice, accumulation of unesterified cholesterol and gangliosides occurs in morphologically normal neurons at birth and thus precedes neuronal injury and cell loss (2,17). Concomitant with the lipid accumulation, brains of Npc1−/− mice exhibit activation and infiltration of microglia and expression of pro-inflammatory mediators (17–19). The clinical, neuropathological and biochemical abnormalities present in juvenile-onset patients are similarly recapitulated in a naturally occurring feline model of NPC1 disease. The NPC1 cat model is an inbred line with a missense mutation in the NPC1 gene (p.C955S) that is evolutionarily conserved and in a cysteine-rich region of the NPC1 protein commonly mutated in juvenile-onset patients, and faithfully models disease progression in this subset of human patients (20–22). Both the murine and the feline NPC1 disease models have provided critical tools for preclinical assessment of potential NPC1 disease therapeutics (23–27).
Treatment with 2-hydroxypropyl-β-cyclodextrin (HP-β-CD), a hydroxyalkyl derivative of the heptameric cyclic sugar β-cyclodextrin, has been shown to reduce both cholesterol and sphingolipid storage and prolong survival in Npc1−/− mice (23,28–30) and NPC1 cats (C. Vite, unpublished data). The mechanism of action for HP-β-CD is uncertain. While there is some evidence that HP-β-CD may expel lysosomal contents through regulated exocytosis (31), at HP-β-CD concentrations (∼100 μm) shown to be effective in vivo, there is no net reduction in cellular cholesterol. Rather, treatment with HP-β-CD rapidly down-regulates CNS expression of cholesterol synthetic genes, suggesting that HP-β-CD triggers release of unesterified cholesterol from lysosomes, allowing for trafficking of this cholesterol to the endoplasmic reticulum (ER) where it can be detected by the cellular sterol-sensing machinery (29,32). Promising pre-clinical studies have led to administration of HP-β-CD in five NPC1 patients through investigator-sponsored single-patient Investigational New Drugs (INDs), and laid the groundwork for a Phase 1 study of intracerebroventricular (ICV) delivery of HP-β-CD at the National Institutes of Health (NIH) (33).
A major barrier to the development of treatments for NPC disease has been the lack of validated biochemical measures (‘biomarkers’) to evaluate efficacy of therapy in clinical trials. Previous studies using the Npc1−/− mouse model have identified plasma inflammatory markers (e.g. galectin-3, cathepsin D, cathepsin S and lysozyme) associated with disease progression, but none specifically monitor neurodegeneration in NPC disease (34,35). Our work with a carefully phenotyped cohort of human NPC1 subjects enrolled an NIH-sponsored natural history study recently identified robust cholesterol-derived NPC1 disease biomarkers: cholestane-3β,5α,6β-triol (‘triol’), a cholesterol oxidation product that is elevated 10-fold in the plasma of NPC1 subjects, and 24(S)-hydroxycholesterol (24(S)-HC), an enzymatically generated oxygenated cholesterol that is reduced in the plasma of NPC1 subjects (36). We hypothesized that HP-β-CD-mediated redistribution of sequestered lysosomal cholesterol in the CNS will promote a cascade of sterol homeostatic responses. HP-β-CD-mediated reduction of lysosomal cholesterol would be expected to increase synthesis of 24(S)-HC—which occurs exclusively in large neurons in the CNS and is rapidly exported into the plasma; stimulate esterification of lysosomal cholesterol redistributed to the ER, as monitored by cholesteryl ester (CE) content of CSF lipoproteins; and decrease non-enzymatic formation of triol and release into the CSF, reflecting reduction in lysosomal cholesterol storage. The present study was designed to test these hypotheses by monitoring these sterol homeostatic markers following direct CNS delivery of HP-β-CD in NPC1 mouse and cat models, and by measuring these markers in NPC1 patient cohorts receiving IV or ICV HP-β-CD.
Peripherally administered HP-β-CD enters the CNS and promotes sterol homeostatic responses
Subcutaneous (SC) administration of HP-β-CD has been shown previously to provide neuroprotection and to prolong lifespan in Npc1−/− mice (23,28,30). Since HP-β-CD has been reported not to cross the blood–brain barrier (BBB) in Npc1−/− mice (37,38), the mechanism through which peripheral delivery of the drug provides CNS benefit has been uncertain. To assess whether HP-β-CD enters the CNS, we administered a single SC dose of 4000 mg/kg to 4-week-old Npc1−/− mice and measured drug levels by liquid chromatography–tandem mass spectrometry (LC-MS/MS). Peak drug levels in the plasma and brain tissue were 550 and 8.7 μm, respectively, and occurred at 90 min post-injection (Fig. 1A). Drug concentration in the liver was intermediate at 137 μm. No differences in the distribution of substituted HP-β-CD isoforms were observed in the plasma and tissue profiles, as compared with the injected material (not shown). The brain to plasma ratio was 1.5%, which approximates the CSF to plasma ratio reported for human albumin (39,40).
We hypothesized that at the low micromolar concentrations of HP-β-CD achieved in brain tissue, HP-β-CD exerted its effects, not through bulk extraction of cellular cholesterol, but by release of lysosomal unesterified cholesterol and redistribution to sites of regulation of cellular cholesterol homeostasis. Cholesterol arriving at the ER would be expected to suppress sterol regulated element binding-protein (SREBP) gene expression, stimulate cholesterol esterification via acyl-CoA:cholesterol acyltransferase (ACAT), and be enzymatically oxidized to 24(S)-HC. 24(S)-HC is synthesized almost exclusively in large neurons in the CNS—the neuronal population most affected by NPC1 disease—and is rapidly exported into the plasma (41). Since the CNS is responsible for nearly all of the 24(S)-HC in the circulation, we explored whether peripheral monitoring of plasma 24(S)-HC could serve as a biomarker for the central effects of HP-β-CD. Following administration of a single SC dose of 4000 mg/kg to 4–5 week-old Npc1−/−mice, plasma 24(S)-HC concentration increased 1.9-fold 24 h after injection as compared with vehicle alone (Fig. 1B). No increase in plasma 24(S)-HC occurred in control mice in response to HP-β-CD treatment (not shown). No significant changes in SREBP target gene expression or CE formation were observed in brain tissue (Fig. 1C and D). By contrast, in liver tissue exposed to 15-fold greater HP-β-CD concentration, SREBP target gene expression was profoundly suppressed and CE molecular species significantly increased more than 10-fold, as compared with vehicle-treated mice (Fig. 1E and F). Thus, at an HP-β-CD dose previously shown to provide neuroprotection in the Npc1−/− mice, the low micromolar drug concentration achieved in brain tissue is sufficient to release lysosomal cholesterol and augment plasma 24(S)-HC concentration.
Direct CNS administration of HP-β-CD increases 24(S)-HC and CE formation
Central administration of HP-β-CD has previously been shown to confer neuroprotection in the mouse and cat NPC1 disease models, while avoiding the pulmonary toxicity and limiting the ototoxicity associated with the massive peripheral dose (25) (C. Vite, unpublished data). To determine whether the effects of direct CNS delivery of a low dose HP-β-CD could be monitored by sampling peripheral 24(S)-HC levels, 4–5-week-old Npc1−/− mice were stereotactically administered either 6 mg/kg HP-β-CD or artificial CSF (aCSF) vehicle into the left lateral ventricle and plasma 24(S)-HC plasma concentrations determined over a 7-day time course. As compared with vehicle-treated animals, 24(S)-HC plasma concentrations were significantly elevated 2, 3 and 7 days post-treatment (P < 0.01), with the maximal concentration occurring 3 days post HP-β-CD treatment (Fig. 2A). The area under the curve (AUC) for the 7-day time course was significantly increased for mice treated with HP-β-CD versus aCSF (Fig. 2B). The 24(S)-HC response was dose-dependent with HP-β-CD ED50 ∼2 mg/kg (Fig. 2C), which is in close agreement with the previously reported ED50 for suppression of cholesterol synthesis following intracerebroventricular (ICV) treatment (42). The dose-dependent increase plasma 24(S)-HC concentrations was accompanied by a similar increase in brain CE synthesis (Fig. 2D), with CE16:0, CE18:0 and CE18:1 serving as the most sensitive indicators of HP-β-CD-mediated lysosomal cholesterol release.
24(S)-HC and CE responses to ICV HP-β-CD attenuate upon repeat dosing
The 24(S)-HC timecourse indicated a rapid and sustained release of the sequestered unesterified lysosomal cholesterol in response to ICV HP-β-CD. We hypothesized that if the time required for lysosomal cholesterol reaccumulation exceeded the dosing interval, then the 24(S)-HC response would then be expected to attenuate upon successive HP-β-CD doses. To test this, 4–5-week-old Npc1−/− mice were treated with either 6 mg/kg ICV HP-β-CD or aCSF vehicle, followed by repeat administration 14 days later with either 6 mg/kg ICV HP-β-CD or aCSF and harvested tissue 3 days following the second dose (Fig. 3A). The sterol homeostatic responses were analyzed for each of the four treatment groups and results normalized to values obtained in animals that received two injections of aCSF. Plasma and brain tissue 24(S)-HC were significantly elevated only in group D, even though groups B and C also received HP-β-CD injections (Fig. 3B and C). Group D differed from groups B and C, however, in that this group only received a single dose and the plasma harvest was proximate to the HP-β-CD injection. These findings indicate that the elevated plasma 24(S)-HC—and, by implication, rates of conversion of cholesterol to 24(S)-HC in the CNS—returns to pre-treatment levels within 17 days and that lysosomal cholesterol in the CNS does not reaccumulate to pre-treatment levels within a 2-week period. SREBP target gene expression was similarly significantly suppressed only in group D mice as compared with control treatment (Fig. 3D; P < 0.05 for HMG-CoA synthase and P < 0.01 for HMG-CoA reductase), confirming the lack of re-accumulation within the dosing interval and the transient nature of the cholesterol redistribution. Quantification of CE formation in brain tissue proved to be the most sensitive indicator of HP-β-CD-mediated lysosomal cholesterol release with 10–25-fold elevations of CE molecular species (16:0, 18:0 and 18:1) in group D mice (Fig. 3E). The 3–6.7-fold elevation of CE in groups B and C mice suggest the signature of HP-β-CD-mediated cholesterol redistribution is still detectable in brain tissue 17 days post treatment. Finally, concentrations of cholestane-3β,5α,6β-triol (triol)—a non-enzymatic oxidation product of unesterified cholesterol and NPC1 disease biomarker (36,43)—were reduced in brain tissue (groups B and C) and plasma (group C), only in mice that were treated with HP-β-CD more than 2 weeks prior to tissue harvest. This finding suggests that levels of the triol marker, in contrast to 24(S)-HC, may serve as a chronic measure of cellular cholesterol storage.
Intrathecal (IT) HP-β-CD administration promotes 24(S)-HC and CE pharmacodynamic responses in the NPC1 cat model
The utility of these biochemical markers to monitor response to HP-β-CD was further examined in the NPC1 cat model, the only large animal model of NPC1 disease. Use of the cat model offered substantial advantages over the NPC1 mice because animals could be dosed repeatedly and CSF and blood could be easily sampled in longitudinal studies of individual animals. For these experiments, 3-week-old treatment-naïve NPC1 cats were treated biweekly (until 11 weeks of age) with 3, 30 or 120 mg HP-β-CD or saline via IT injection at the cerebellomedullary cistern. As compared with saline control, cats receiving HP-β-CD IT exhibited a significant dose-dependent increase (2.6-fold for 30 mg and 2.9-fold for 120 mg, P < 0.05) increase in 24(S)-HC in the CSF 3 days following the initial dose (P < 0.01), but CSF 24(S)-HC was not elevated in response to subsequent doses (Fig. 4A). The plasma 24(S)-HC concentration was likewise significantly increased in all three dosing groups for AUC0−3 (0–3 days post dose) and AUC0−7 (0–7 days post dose) as compared with saline control, but only following the initial dose (Fig. 4B and C). Attenuation of the 24(S)-HC pharmacodynamic response was similarly observed in cats receiving monthly IT HP-β-CD (not shown). Consistent with the 24(S)-HC response, CSF CE species (3 mg: 16:0 and 20:4; 30 mg: 16:0, 18:1, 18:2 and 20:4; 120 mg: 16:0 and 18:2) were elevated 1.3–1.6-fold (P ≤ 0.05) in the treatment groups (Fig. 4D–F). Plasma triol concentrations, which primarily reflect production of this oxysterol in peripheral tissues (36), increased linearly with age (i.e., disease-severity) in the control animals (Fig. 4G). By contrast, the expected increase in plasma triol concentrations was prevented animals treated with 120 mg, but not 3 or 30 mg, IT-treated cats, likely due to the presence of higher plasma HP-β-CD concentrations during the clearance phase in the higher dose group (C. Vite, unpublished data).
Validation of sterol homeostatic responses in human subjects administered HP-β-CD
To validate the utility of the 24(S)-HC pharmacodynamics response to monitor HP-β-CD treatment, we measured serial plasma 24(S)-HC concentration in samples obtained from human NPC1 subjects enrolled in an NIH-sponsored Phase 1 trial of ICV HP-β-CD. Plasma samples were obtained from three NPC1 subjects administered an ICV saline infusion, followed by 50 mg ICV HP-β-CD infusion. Plasma 24(S)-HC AUC (8–72 h post dose) were calculated for each subject. When compared with baseline saline infusion, significant increases in the plasma 24(S)-HC AUC were observed in two of three subjects (Fig. 5A; P < 0.05), indicative of a biochemical response to the HP-β-CD treatment. We additionally obtained a set of plasma samples from two NPC1 subjects initially enrolled in an observational study at the U.S. National Institutes of Health Clinical Center. Plasma triol concentrations were measured over 4 years, during which period IV HP-β-CD treatment was initiated under an individual use Investigation New Drug application (Fig. 5B). Following initiation of treatment, plasma triol concentrations were decreased by 46 and 70% in Subjects 1 and 2 after 25 and 24 months of treatment (P < 0.05), respectively, and was even lowered into the normal reference range (<24.5 ng/ml) in Subject 2.
Over the past decade, advances in our understanding of the pathogenesis of NPC disease, coupled with progress in deciphering the function of the NPC1 and NPC2 proteins, have led to a proliferation of new drug targets and candidate therapeutic compounds. The most promising new therapeutic for NPC1 disease, HP-β-CD, has recently entered Phase 1 testing at NIH. The goal of the present study was to develop biochemical outcome measures to assess therapeutic efficacy of HP-β-CD in NPC1 patients. Based on our understanding of the pathophysiological alterations in cholesterol metabolism in NPC1 disease and the in vivo mechanism of action of HP-β-CD, we examined the impact of the HP-β-CD-mediated redistribution of the entrapped lysosomal cholesterol on sterol homeostatic responses. Following direct CNS delivery of HP-β-CD, we observed significant increases in plasma 24(S)-HC AUC in two independent animal models of NPC1 disease. These findings were confirmed in human NPC1 subjects receiving ICV HP-β-CD. Since the enzyme responsible for the 24-hydroxylation of cholesterol is expressed predominantly in large neurons in the CNS and the circulating 24(S)-HC is almost exclusively CNS-derived, the increase in plasma 24(S)-HC provides a peripheral, non-invasive measure of the pharmacodynamic effect of HP-β-CD in the CNS. We propose that plasma 24(S)-HC, along with the other plasma and CSF cholesterol-derived markers examined in this study, will accelerate development of HP-β-CD as a therapeutic for NPC1 disease, by providing novel biochemical metrics to assess drug efficacy, rather than needing to rely solely upon long-term, difficult to quantify clinical observations.
HP-β-CD has profound effects on sterol metabolism in NPC1 disease models. While the mechanism of action is not completely understood, HP-β-CD has been shown to gain access to lysosomes in NPC1 cells through uptake via fluid-phase endocytosis (44). It seems plausible that within the lysosome, HP-β-CD, which has been shown to facilitate transfer between membranes (45), promotes cholesterol transfer between LBPA-rich internal membranes—where lipoprotein cholesterol is liberated by the lysosomal acid lipase—and the limiting membrane. Whereas at millimolar concentrations HP-β-CD efficiently extracts membrane cholesterol in cell-based models, at submillimolar concentrations HP-β-CD does not reduce total cellular cholesterol (46). Rather, at these concentrations HP-β-CD elicits a rapid release of lysosomal cholesterol that is distributed to other membrane pools. This is supported by in vitro studies demonstrating suppression of SREBP2 target gene expression and increased rates of cholesterol esterification in response to HP-β-CD (32,44), and in vivo studies finding similar downregulation of SREBP2 target genes, upregulation of LXR target genes, and inhibition of cholesterol biosynthesis and ester formation following HP-β-CD treatment (30,42,44).
Whether HP-β-CD crosses the BBB has been the subject of controversy. Earlier BBB permeation studies using 14C-labeled HP-β-CD had failed to show brain uptake following an intravenous injection in Npc1−/− mice (37). A limitation of these studies was the use of older mice, in which the Cmax and AUC for HP-β-CD are blunted (29), and an HP-β-CD dose 8-fold lower than that subsequently shown to be required for neuroprotection. More recently, Begley and colleagues reported no significant BBB penetration of HP-β-CD into the brain following intraperitoneal administration, concluding that HP-β-CD may be modulating neuronal function through interaction at the surface of the endothelium of the cerebral vasculature (38). However, the finding that CNS delivery of HP-β-CD in NPC1 animal models is strikingly more effective than peripheral administration, even at >1000-fold reduction in dose (42,47), strongly supports direct CNS effects of HP-β-CD. Here, we addressed this question using sensitive tandem mass spectrometry methods to quantify HP-β-CD concentration in brain tissue in Npc1−/− mice peripherally administered HP-β-CD. In agreement with the previous studies we found that HP-β-CD is largely excluded from the brain with a brain to plasma ratio of 1.5%, which is comparable to the brain to plasma ratio for serum albumin determined in human NPC1 subjects enrolled in the NIH-sponsored observational study (n = 57, mean 0.44, range 0.15–1.4). This suggests that the drug may be crossing the BBB by fluid-phase transcytosis, in a manner similar to serum albumin (39,40). Nonetheless, in the setting of a 4000 mg/kg SC dose, low micromolar HP-β-CD concentration is achieved in the brain tissue, simply by mass action. Comparable HP-β-CD concentrations were measured in the CSF of NPC1 cats after SC drug delivery (C. Vite, unpublished data). These brain concentrations approximate that achieved with ICV delivery of 0.5 mg/kg HP-β-CD, the ED50 for suppression of brain cholesterol synthesis in Npc1−/− mice (42,47). Together, these findings provide support for the emerging consensus that HP-β-CD requires access to the CNS to exert its neuroprotective effects.
In the present study, we explored whether candidate cholesterol metabolites in plasma and CSF could serve as biomarkers to monitor the pharmacodynamic effects of CNS delivery of HP-β-CD on brain cholesterol metabolism. Both 24(S)-HC and CE were acutely elevated in response to HP-β-CD treatment in NPC1 animal models (24(S)-HC and CE) and in human subjects (24(S)-HC). The pharmacodynamics of these metabolites were comparable, with a similar dose-dependent rise detectable at 24 h, reflecting the predicted rapid release of trapped lysosomal cholesterol. The elevation in plasma 24(S)-HC concentration was evident up to a week post-treatment, despite rapid clearance of HP-β-CD from the CSF with a t1/2 of 3.9 h (C. Vite, unpublished data). While it is possible that HP-β-CD may have altered the post-lysosomal cholesterol trafficking machinery in some manner, it seems more likely that alleviation of the lipid ‘traffic jam’ in lysosomes corrected cholesterol flux through this compartment and, as a consequence, normalized 24(S)-HC production. In concert with the elevation in 24(S)-HC and CE, there was a concomitant reduction in brain triol, providing further evidence for reduced lysosomal cholesterol storage.
24(S)-HC is an attractive candidate for a biomarker to monitor acute intervention with HP-β-CD in NPC1 disease. Conversion of unesterified cholesterol to 24(S)-HC is the principal route for excretion of excess cholesterol from the CNS, and this pathway is impaired in NPC1 disease (36). Thus, measurement of 24(S)-HC production provides a readily quantifiable, biochemical tool to assess the effects of drug intervention on the release of the sequestered lysosomal cholesterol. Since the ER-localized sterol-24-hydroxylase is predominantly expressed in large neurons in the brain, the circulating 24(S)-HC is almost exclusively CNS-derived (41). Sampling of plasma 24(S)-HC, therefore, offers a non-invasive metric of the pharmacodynamic effects of HP-β-CD in NPC1 disease. Based on the pre-clinical studies in the NPC1 animal models in this report, the plasma 24(S)-HC AUC (0–3 days post treatment) has been selected as the primary outcome measure in the on-going Phase 1 dose-escalation trial of ICV HP-β-CD in NPC1 disease at NIH (33). In the first cohort, which received the lowest HP-β-CD dose (50 mg ICV), we detected a significant increase in the plasma 24(S)-HC AUC in two of three subjects, providing the first evidence for CNS efficacy of HP-β-CD in human disease.
With respect to monitoring NPC1 disease, the value of plasma 24(S)-HC is principally as a pharmacodynamic marker to assess drug engagement. In contrast to the rapid rise in the plasma 24(S)-HC AUC following HP-β-CD intervention compared with saline control, steady-state plasma 24(S)-HC concentration is reduced on average by only 20% in human NPC1 subjects—likely reflecting loss or dysfunction of large neurons in the CNS (36)—and by itself does not appear to be a sensitive indicator in humans of disease progression. Similarly, in the NPC1 animal models, steady-state plasma 24(S)-HC concentration must be interpreted with caution. The plasma 24(S)-HC concentration is elevated Npc1−/− mice [(34) and our unpublished data], despite a 22% reduction in 24(S)-HC flux in the brains of these animals (48). This seemingly paradoxical observation can be attributed to impaired 24(S)-HC clearance by the liver due to the significant hepatocellular injury and fibrosis in the mouse model, and is further supported by the reduced bile acid secretion previously reported in Npc1−/− mice (49). By contrast, in human NPC1 disease, once beyond the neonatal period the liver disease is more modest and subclinical.
In contrast to the 24(S)-HC and CE markers that were acutely elevated in response to HP-β-CD intervention, triol concentrations exhibited a gradual decline. In previous studies we found that triol, a non-enzymatic cholesterol oxidation product generated in the setting of excess unesterified cholesterol and oxidative stress, was elevated in the brain tissue of NPC1 animal models and in the plasma and CSF of human NPC1 subjects (36). A clinical assay based on this highly sensitive and specific marker is rapidly being adopted as the new diagnostic standard for NPC disease (43). In the present study, we examined this marker in both NPC1 disease models and human NPC1 subjects receiving treatment with HP-β-CD. In Npc1−/− mice, HP-β-CD administration lowered triol in brain tissue, but this effect was not evident until 2 weeks post HP-β-CD administration, possibly reflecting a delay in alleviating the cellular oxidative stress burden. While treatment with HP-β-CD blunted the rise in plasma triol in NPC1 cats and led to a marked decline of the marker in human NPC1 subjects, these changes likely reflect the effect of the drug on peripheral organs, rather than the CNS, because the bulk of circulating triol is of liver origin. Ideally, measurement of CSF triol would provide a more direct measure of mitigation of CNS cholesterol storage following intervention with HP-β-CD. However, triol concentration in CSF is ∼1000-fold less than in plasma and a sufficiently sensitive and specific tandem MS assay is not yet available.
An intriguing finding was the brisk attenuation of the cholesterol homeostatic responses following repetitive HP-β-CD dosing. For both the NPC1 mouse and cat models, the increased 24(S)-HC and CE production, as well as suppression of SREBP-2 target gene expression, rapidly attenuated in response to successive biweekly HP-β-CD doses. Attenuation of these sterol homeostatic responses was similarly observed in cats dosed at monthly intervals (Supplementary Material, Fig. S1). While a sensitive pharmacodynamics marker of initial drug engagement, plasma 24(S)-HC will likely not prove useful for monitoring either continued responsiveness or effect on the rate of disease progression. Other biochemical or imaging markers are clearly needed for this purpose and may emerge from analysis of clinical samples from the current Phase 1 trial. These findings also indicate that a single, centrally delivered HP-β-CD dose is remarkably efficient in releasing the accumulated lysosomal cholesterol. The acute depletion of neuronal cholesterol stores is supported by marked reduction of filipin staining in cortical sections of 8- and 56-day-old Npc1−/− mice within 24 h following administration of a single subcutaneous HP-β-CD dose (S. Walkley, personal communication). Moreover, the effect of HP-β-CD is sustained, suggesting that dosing intervals up to 4 weeks may be sufficient for chronic central administration. The sustained effect of HP-β-CD on the cholesterol homeostatic responses observed here is consistent with findings from earlier studies that quantified the effect of HP-β-CD on whole-body cholesterol flow in tissues of Npc1−/− mice (29,30). Liu and colleagues found that a single dose of HP-β-CD at postnatal Day 7 led to suppression of cholesterol synthesis in brain tissue that persisted for at least 3 weeks, and altered whole-body sterol balance, which did not return to levels in untreated mice until six weeks post-treatment (30). The sustained effect of HP-β-CD on cholesterol homeostatic responses has important implications for design of clinical trials to study the efficacy of HP-β-CD in NPC1 patients. Preclinical studies indicate that brain cholesterol flux returns to untreated levels in ∼6 weeks (29); however, the washout period for the drug's pharmacodynamics effects in humans has not yet been studied. This implies that only HP-β-CD-naïve patients may be eligible for future clinical trials. Since there are a limited number of NPC1 patients worldwide, some of whom have already gained access to the drug through expanded use programs, it is imperative that investigators work closely with NPC family organizations and regulators to ensure that the pool of trial eligible subjects is not sufficiently diminished to interfere with the drug approval process.
The development of HP-β-CD as treatment for NPC1 disease has been supported by the NIH Therapies for Rare and Neglected Diseases program (50). Through this program, the required additional pre-clinical studies, assay development and biomarker validation were performed in order to support filing of an IND and initiation of a Phase 1 of ICV HP-β-CD clinical trial at the NIH Clinical Trial within a two-year time frame. The cholesterol homeostatic biomarkers developed through these efforts and reported here serve a critical role by providing primary (plasma 24(S)-HC)) and secondary outcome measures (CSF CE and plasma triol) for the current Phase 1 trial. These biochemical biomarkers, along with other lipidomic and proteomic markers (51–54), will be powerful tools to assess the acute and chronic effects of HP-β-CD. We anticipate that these markers will not only accelerate clinical evaluation of HP-β-CD for delaying NPC1 disease progression, but also have the potential to serve as clinically-validated surrogate outcome measures in future clinical trials.
MATERIALS AND METHODS
2-Hydroxypropyl-β-cyclodextrin (Kleptose HPB, HP-β-CD) was obtained from Roquette Pharma (Lestrem, France). Mass spectrometry reagents were obtained from commercial sources, with the exception of d7-cholesteryl oleate that was synthesized in house.
BALB/c Npcnih (Npc1−/−) mice were obtained from Jackson Laboratories (Bar Harbor, ME) and have been maintained in a breeding colony at Washington University. NPC1 cats were raised in the animal colony of the School of Veterinary Medicine at the University of Pennsylvania under NIH and USDA guidelines for the care and use of animals in research (25). Experimental procedures were approved by the Washington University and University of Pennsylvania Animal Studies Committees and were conducted in accordance with the USDA Animal Welfare Act and the Public Health Service Policy for the Humane Care and Use of Laboratory Animals.
Administration of HP-β-CD
For SC administration of HP-β-CD in mice, postnatal Day 28 mice were injected with 4000 mg/kg in a 20% (w/v) solution in water or water alone as the vehicle control. The volume administered was based on body weight (20 μl/g mouse), such that a 5 g mouse would receive 0.1cc HP-β-CD solution. Mice were then sacrificed after 24 h and perfused with phosphate-buffered saline (PBS) before tissue samples were collected. For pharmacokinetic studies of HP-β-CD, postnatal Day 28 mice were sacrificed after 90 min and perfused with PBS before harvesting tissue samples. Stereotaxic ICV injections of HP-β-CD (2–20 mg/kg) or aCSF were performed as described (42). Postnatal Day 28–33 mice received a single ICV injection in the left lateral ventricle and tissues and plasma harvested at the indicated time points. For dual ICV injection experiments, the mice were injected with of HP-β-CD (6 mg/kg) or aCSF ICV into the left lateral ventricle and the second injection administered 2 weeks later into the right lateral ventricle. Mice were sacrificed 72 h after the second injection and tissue and plasma samples were collected.
For IT delivery of HP-β-CD to NPC1 cats Kleptose® HPB (D.S. 0.4, average M.W. 1400), was used and was provided by Janssen Research and Development (Beerse, Belgium). All HP-β-CD was administered in a 20% (weight/volume) solution dissolved in 0.9% saline (Hospira Inc., Lake Forest, IL) except when administered as a 3% solution dissolved in saline due to the small volume of administration (3 mg). An equal volume of saline was injected intrathecally to control cats. All IT dosing was performed in cats anesthetized with propofol (up to 6 mg/kg intravenously; Abbott Laboratories, Chicago, IL). Cats received HP-β-CD at the cerebellomedullary cistern every 14 days beginning at 3 weeks of age. Approximately 1 ml CSF was collected from the cerebellomedullary cistern and 1 ml of blood was collected from the jugular vein prior to each dosing. Plasma samples for 24(S)-HC measurements were obtained at 0, 1, 2, 3 and 7 days post therapy to determine Cmax and time-activity AUC for 24(S)-HC. CSF was obtained at 0, 2, 4, 6 and 8 weeks from cats to determine triol and CE concentrations. Serum and CSF were frozen at −80°C immediately following collection.
Administration of ICV HP-β-CD in NPC1 patients
Three NPC1 patients were evaluated at the National Institutes of Health Clinical Center in Bethesda, Maryland between January and April 2013 in an open trial administering HP-β-CD into the lateral intraventricular space via an Ommaya reservoir. (NCT01747135; http://clinicaltrials.gov/ct2/show/NCT01747135?term=Niemann-Pick&rank=6). This study was approved by the Institutional Review Board of the Eunice Kennedy Shriver National Institute of Child Health and Human Development. Written consent, and when feasible assent, were obtained. Enrollment included 2 males and 1 female who ranged in age from 10 to 20 with NIH severity scores (55) ranging from 16 to 22 points. Vitamins, supplements and medications (except for miglustat) being used by guardians with the intent to treat NPC1 disease were discontinued. All three patients were on off-label miglustat therapy. After placement of Ommaya reservoirs, patients received saline and HP-β-CD via ICV infusion. HP-β-CD was supplied by Janssen Pharmaceuticals as a 200 mg/ml solution. For administration the HP-β-CD was diluted to a final concentration of 10 mg/ml in normal saline. Fifty milligrams were administered over 5 min followed by a 2 ml flush of the Ommaya Reservoir. Serial blood samples were collected for 24(S)-HC analysis at the following time points: pre-dose and 15, 30 and 60 min post-dose, and 3, 6, 24, 36, 48 and 72 h post-dose (for both saline and study drug). Plasma was isolated and frozen at −80° until analysis. 24(S)-HC was quantified in the plasma samples as described below.
Real-time quantitative RT-PCR
RNA isolation, cDNA synthesis and real-time PCR using SYBR Green Master Mix with template-specific primers was performed as described previously (24). Fold changes in gene expression were determined by normalization to 36B4 expression. The primer sequences were as follows: HMG-CoA Reductase (F, 5′-TGTGGTTTGTGAAGCCGTCAT-3′; R, 5′-TCAACCATAGCTTCCGTAGTTGTC-3′), HMG-CoA Synthase (F, 5′-GGGCCAAACGCTCCTCTAAT-3′; R, 5′-AGTCATAGGCATGCTGCATGTG-3′).
LC–MS/MS measurement of oxysterols
For quantification of 24(S)-hydroxycholesterol (24(S)-HC)) and triol, deuterated internal standards were added—d7–24(S)-HC and d7-triol—and oxysterols extracted either by the method of Bligh and Dyer (56) or by methanol precipitation (43). Derivatization was performed with N,N,-dimethylglycine as described (43), samples dried and reconstituted in 1:1 (v/v) methanol/water. Sample were injected onto an on-line trapping LC-MS/MS system, coupled to a CTC Pal autosampler, Agilent 1100 binary pumps, Shimadzu 20 A binary pumps, a valco valve, and an API-4000 mass spectrometer. A betasil C18 trapping column (10 × 2 mm, 5 µm particle size) and an Agilent Eclipse C18 HPLC column (3.5 × 100 mm, 3.5 µMm particle size) were used for the on-line trapping systems. 1.5% (v/v) formic acid in water (A: solvent) and acetonitrile (B: solvent) were used for Agilent pumps to trap all derivatized sterols and 2%(v/v) formic acid in water (A: solvent) and 2% (v/v) formic acid in 1:1 (v/v) acetonitrile and methanol were used for analytical LC separation of each derivatized sterol by benefit of programmed solvent gradients. Assay time was 12 min for each sample. Quantification of the oxysterols was performed using calibration curves with deuterated oxysterols as internal standards. Detection was performed using positive ion multiple reaction monitoring (MRM) mode Q1/Q3 ions: m/z 573.4/470.4 and 580.4/477.4 for 24(S)-HC and d7–24(S)-HC, respectively, and m/z 591.4/104.1 and 598.4/104.1 for triol and d7-triol, respectively.
LC-MS/MS measurement of CEs
For quantification of CEs, either d7-cholesteryl oleate or d7-cholesteryl linoleate was added as an internal standard, and lipid extractions performed as described above. Samples were injected into an on-line trapping system as described above. A betasil C18 trapping column and an Agilent Eclipse C8 HPLC column were used for the on-line trapping systems. Of note, 1.5% formic acid in water and acetonitrile were used for Agilent pumps to trap all CE, and 10 mm ammonium acetate in water and 10 mm ammonium acetate in methanol were used for analytical LC separation of each CE through programmed solvent gradients. Quantification of CE was achieved using calibration curves with d7-cholesterol linoleate as an internal standard. Detection was performed using positive ion MRM in Q1/Q3: m/z 642.5/369.3 for CE(16:0), 670.6/369.3 for CE(18:0), 668.6/369.3 for CE (18:1), 666.6/369.3 for CE(18:2), 690.6/369.3 for CE(20:4) and 673.6/376.4 for CE(d7–18:2).
LC-MS/MS measurement of HP-β-CD
Plasma and homogenized brain tissues were spiked with β-cyclodextrin as an internal standard followed by protein precipitation with methanol. Supernatants were dried under a stream of nitrogen and the sample was reconstituted in water for LC-MS/MS analysis. Samples were injected onto an on-line trapping system as described above using a Targa C18 trapping column and a Varian metasil C18 column. Of note, 1.5% formic acid in water and acetonitrile were used for Agilent pumps to trap all HP-β-CD, and 10 mm ammonium acetate in water and 10 mM ammonium acetate in methanol were used with programmed solvent gradients. For quantification of HP-β-CD, calibration standards were prepared with β-cyclodextrin as an internal standard. Detection was performed using positive ion MRM in Q1/Q3: 1152.4/325.3 β-cyclodextrin; 1210.5/279.2 HP-β-CD 1-isopropyl conjugation;1268.6/279.2 HP-β-CD 2-isopropyl conjugations; 1326.7/279.2 HP-β-CD 3-isopropyl conjugations; 1384.7/279.2 HP-β-CD 4-isopropyl conjugations;1442.8/279.2 HP-β-CD 5-isopropyl conjugations; 1500.8/279.2 HP-β-CD 6-isopropyl conjugations; 1558.9/279.2 HP-β-CD 7-isopropyl conjugations; 1616.9/279.2 HP-β-CD 8-isopropyl conjugations; 1674.9/279.2 HP-β-CD 9-isopropyl conjugations; and 1732.9/279.2 HP-β-CD 10-isopropyl conjugations
Results are expressed as mean ± SEM. For group comparisons, the statistical significance of differences in mean values was determined by a two-tailed Student's t test using Graphpad Prism 6 for Mac OS X (version 6.0b). A P value of 0.05 or less was considered significant. For analysis of 24(S)-HC response in human subjects, individual AUCs were calculated using a trapezoidal rule. Baseline (saline infusion) SD for each individual AUC was obtained using a bootstrapping resampling method. AUCs for drug response >2.3-fold the SD of the baseline AUC were considered significant (57).
This work was supported by the Bench to Bedside Program, the Office of Rare Diseases Research and the Therapies for Rare and Neglected Diseases (TRND) Program in the National Center for Advancing Translational Sciences of the National Institutes of Health, and the intramural research program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development. Additional support was provided by R01 NS081985 to D.S.O. and J.E.S., R01 NS073661 to C.H.V, NCRR02512 and P40–02512, NIH CTSA UL1 TR000448 to D.S.O and C.H.V, a grant from Dana's Angels Research Trust to D.S.O and C.H.V., and a grant from the Ara Parseghian Medical Research Foundation to F.D.P. The content of this publication is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
This work was performed in the Metabolomics Facility at Washington University (P30 DK020579). ICV injections were conducted by the Hope Center Animal Surgery Core at Washington University by Ronaldo Perez. Support for the clinical studies was provided by John McKew (National Center for Advancing Translational Sciences of the NIH); Naomi O'Grady, Nancy Ames, Myra Woolery, Carmen Brewer, Kelly King, Chris Zalewski, George Grimes, Judith Starling, Hope Decederfelt, Andrea Gropman, Cindy Toth, Mary Fedwa and Karimrim Calis (National Institutes of Health Clinical Center); Steven Silber, Mark Kao, Marjo Janssen and Ilona Scott (Johnson & Johnson); and Alan Hubbs (NICHD). Aiyi Liu provided assistance with statistical design. We are grateful to the Hadley Hope Fund for their support.
Conflict of Interest Statement. D.S.O. and F.D.P. were awarded US Patent 8497122, entitled ‘Biomarkers for Niemann–Pick C Disease and Related Disorders,’ for use of oxysterols as biomarkers for Niemann–Pick C disease.