Defective Cytochrome P450-Catalysed Drug Metabolism in Niemann-Pick Type C Disease

Niemann-Pick type C (NPC) disease is a neurodegenerative lysosomal storage disease caused by mutations in either the NPC1 or NPC2 gene. NPC is characterised by storage of multiple lipids in the late endosomal/lysosomal compartment, resulting in cellular and organ system dysfunction. The underlying molecular mechanisms that lead to the range of clinical presentations in NPC are not fully understood. While evaluating potential small molecule therapies in Npc1-/- mice, we observed a consistent pattern of toxicity associated with drugs metabolised by the cytochrome P450 system, suggesting a potential drug metabolism defect in NPC1 disease. Investigation of the P450 system in the context of NPC1 dysfunction revealed significant changes in the gene expression of many P450 associated genes across the full lifespan of Npc1-/- mice, decreased activity of cytochrome P450 reductase, and a global decrease of multiple cytochrome P450 catalysed dealkylation reactions. In vivo drug metabolism studies using a prototypic P450 metabolised drug, midazolam, confirmed dysfunction in drug clearance in the Npc1-/- mouse. Expression of the Phase II enzyme uridinediphosphate-glucuronosyltransferase (UGT) was also significantly reduced in Npc1-/- mice. Interestingly, reduced activity within the P450 system was also observed in heterozygous Npc1+/- mice. The reduced activity of P450 enzymes may be the result of bile acid deficiency/imbalance in Npc1-/- mice, as bile acid treatment significantly rescued P450 enzyme activity in Npc1-/- mice and has the potential to be an adjunctive therapy for NPC disease patients. The dysfunction in the cytochrome P450 system were recapitulated in the NPC1 feline model. Additionally, we present the first evidence that there are alterations in the P450 system in NPC1 patients.


Introduction Results
Our laboratory has tested numerous small molecule therapies as possible therapeutic interventions in Npc1 -/mice [12]. During these trials it was observed that compounds that are primarily subject to first-pass metabolism in the liver were associated with reduced tolerability and toxicity [12] (Table 1). When lower doses were used positive therapeutic benefits were observed [12] ( Table 1). The reduced doses required for ibuprofen, vinpocetine and curcumin in Npc1 -/mice are summarised along with the normal dose used in other mouse strains (Table 1). In contrast, drugs such as miglustat, which are renally excreted intact, could be used at doses comparable to those reported in other murine models and strains [12,35].
To test the hypothesis that the increased drug toxicity observed in Npc1 -/mice is the result of perturbations in the cytochrome P450 system, we investigated the enzymatic activities of the methoxyresorufin-O-dealkylation (MROD), ethoxyresorufin-O-dealkylation (EROD), pentoxyresorufin-O-dealkylation (PROD) and, benzoxyresorufin-O-dealkylation (BROD) in Npc1 mutant, heterozygous and control littermates mice at 3, 6, and 9 weeks of age. Significant reductions in all four enzymatic reactions were noted in Npc1 -/mice at all time points tested (p < 0.0001) (Fig 1A-1C and p values and enzymatic activities are provided in Table 2). Heterozygous Npc1 mice demonstrated no statistical differences in any of the four-enzymatic activities compared with control littermates at 3 weeks of age ( Fig 1A and Table 2). However, Npc1 +/mice had significant reductions in MROD, EROD, PROD and BROD by 6 weeks of age (Fig 1B and Table 2), and remained significantly reduced at 9 weeks of age (Fig 1C and  Table 2). In agreement with these findings, cytochrome C reductase activity was significantly reduced at 9 weeks of age in the Npc1 -/and Npc1 +/mice compared to control littermate mice ((4.9 ± 0.2 vs. 11.6 ± 1.2 pmol/min/mg p = 0.032) (4.0 ± 0.3 vs. 11.6 ± 1.2 pmol/min/mg p = 0.008) respectively) (S1 Fig). No differences were noted between the sexes of any genotype at any time point.
Previously, microarray analysis using mRNA isolated from female livers of Npc1 -/and control littermate mice over the mutant mouse's life span revealed 62 of 101 known genes of the cytochrome P450 family as having modified expression [9] (S1 Table). In total, 44 of these 62 genes were down regulated at a minimum of one time point within the life span of the Npc1 -/mouse, while 14 of the 44 were significantly down regulated at all time points. Genes belonging to subfamilies 1 to 3 (42 genes in all), the products of which are primarily responsible for drug metabolism, were particularly affected, with the vast majority being down-regulated. We therefore performed qPCR on mRNA isolated from 9-week-old female livers to validate the changes in the expression array data for 11 of these genes (Fig 2): Cyp1a2, Cyp2a4, Cyp2b10, Cyp2b13, Cyp2c37, Cyp2c40, Cyp2c50, Cyp2c54, Cyp3a16, Cyp3a41 and Cyp3a44. A subset of 9 genes was assayed across all time points and compared directly with the microarray data (S2 Fig). All assayed genes showed high concordance with the microarray data. The expression of most Table 1. Examples of therapeutic drugs requiring lower dosing in Npc1 -/mice.
One potential mechanism explaining the reduced expression of P450-related genes is a bile acid deficiency or imbalance, as bile acid levels are part of the multifactorial regulation of the P450 system [41]. We therefore treated Npc1 -/-, Npc1 +/and control littermate (Npc1 +/+ ) mice of both genders with UDCA (0.5% of diet) from three weeks of age and measured P450 enzyme activity at 6 and 9 weeks of age. UDCA therapy resulted in the complete normalization or overexpression of P450 enzyme activities (Fig 4A and 4B and Tables 3 and 4). Interestingly, this normalization was consistently greater at 6 weeks of age, while the absolute percentage increase was reduced at 9 weeks compared to the 6-week values. This is not an unexpected result as the disease is continuing to progress over time leading to a fibrotic liver [42]. As such these findings likely implicates reduced bile acid levels or imbalances in bile acid species in NPC1 as an important mechanism underlying the drug metabolism defect.
Npc1 -/mice treated with UDCA had a moderate increase in MROD activity (expressed as percentage of age-matched controls) at 6 weeks of age compared with untreated Npc1 -/mice. MROD activity is not increased significantly in heterozygous Npc1 +/mice treated with UDCA, at either 6 or 9 weeks of age relative to untreated Npc1 +/mice. Both Npc1 +/+ and Npc1 +/mice treated with UDCA have reduced MROD activity at 6 weeks of age compared with untreated Npc1 +/+ (control) mice, while no reduction was noted by 9 weeks in the Npc1 +/mice treated with UDCA ( Fig 4A and 4B and Tables 3 and 4).
EROD, PROD, and BROD enzymatic activities were all significantly increased at 6 weeks in Npc1 -/mice treated with UDCA, but this effect was diminished at 9 weeks when compared with the untreated Npc1 -/mice. The same trend was apparent in the Npc1 +/treated mice where at 6 weeks EROD, PROD, and BROD were significantly increased (Fig 4A and 4B and Tables 3 and 4). However, in late stage disease (9 weeks of age) no significant increase in PROD activity was detected, with a moderate increase in both EROD and BROD enzymatic activity in the Npc1 +/mice when compared to untreated Npc1 +/mice, similar to what was observed at 9 weeks in the treated Npc1 -/mice.
Effects of UDCA on body weight, rearing and tremor of Npc1 -/mice Monitoring body weight revealed a similar pattern of weight gain and loss in Npc1 -/mice regardless of treatment. However Npc1 -/mice treated with UDCA had a generalized reduction . Data are presented as mean ± SEM, n = 6 (3 males and 3 females), * p-value < 0.05, ** p -value < 0.01, *** p-value < 0.001, **** p-value < 0.0001 calculated using an unpaired nonparametric test Mann-Whitney test. (C) Relative protein quantitation of Cyp3a isolated from liver of 6-week old male mice found a significant decrease in Npc1 +/and Npc1 -/untreated mice compared Npc1 +/+ in weight that was significant at 5, 6, 9, and 11 weeks compared with the untreated Npc1 -/mice (p value = 0.006, p value = 0.032, p value = 0.030 and p value = 0.002 respectively). This lower body weigh in the UDCA treated mice was also reflected in the Npc1 +/+ mice with significant differences noted at 5, 11, and 12 weeks (p value = 0.021, p value = 0.007 and p value = 0.024 respectively). This was an expected result and in agreement with known reduction in body weight observed in other studies with UDCA [43], no significant difference between genders were noted so genders were combined ( Fig 5A).
Rearing (a measure of motor coordination) increased in the Npc1 -/mice treated with UDCA compared to the untreated Npc1 -/mice. While a high degree of variability was seen for all groups, UDCA-treated Npc1 -/mice showed marked improvements, which was significant at 5 weeks of age (p value = 0.0055) and from 7 to 12 weeks (p value = 0.050, p value = 0.002, p value = 0.013, p value = 0.001, p value = 0.0007 and p value = 0.002 respectively) ( Fig 5B). No differences were detected between the Npc1 +/+ treated or untreated mice ( Fig 5B). Npc1 -/mice on UDCA also retained motor function longer than untreated Npc1 -/mice as demonstrated by their ability to rear at 11 weeks, an age at which all untreated Npc1 -/mice are incapable of rearing (p value = 0.0007) ( Fig 5B).
Tremor is a disease phenotype observed in the Npc1 -/mouse model from approximately 7 weeks of age [44]. An age-dependent increase in high frequency tremor (32-53Hz) has been observed in Npc1 mice, with 43Hz used as an arbitrary point of comparison between treatment groups [12]. Npc1 -/mice treated with UDCA presented with significantly lower tremor amplitudes than was observed in the untreated Npc1 -/mice at 9 and 11 weeks of age ( Fig 5C-5F). As previously reported [12], untreated Npc1 -/mice showed significantly higher 43 Hz tremor when compared with untreated Npc1 +/+ mice both at 9 and 11 weeks of age (p value = 0.0025 and p value < 0.00001 (Fig 5E and 5F). Tremor amplitude at 43 Hz was significantly lower at 9 and 11 weeks of age for Npc1 -/mice treated with UDCA compared with untreated Npc1 -/mice (p value = 0.014 and p value = 0.0002) (Fig 5E and 5F). No changes were detected between untreated Npc1 +/+ mice and Npc1 +/+ mice treated with UDCA at either time point (Fig 5E and 5F).
Gene expression analysis of a subset of the Cyp genes, Cyp1a2, Cyp2b10, Cyp2b13, and Cyp2a4 highlight the dynamic nature of these genes and the challenge of fully discerning the exact functional role of each Cyp gene. Basal expression levels of the same genotype and age varied between male and female animals and the manner in which the genders responded to UDCA varied (Fig 6A-6H). Cyp1a2 gene expression showed little change in the UDCA treated males and females; whilst some values proved to be statistically significant the total fold change was small (Fig 6A and 6B). Cyp2b10 and Cyp2a4 expression patterns mirrored the changes in enzymatic activity, with UDCA causing a significant increase in transcript levels ( Fig 6C, 6D, 6G and 6H). Cyp2b13 behaved in the opposite manner based on gender, the females on UDCA exhibiting significantly decreased gene expression, while UDCA-treated males showed significantly increased gene expression (Fig 6E and 6F).
As HPBCD treatment is increasingly being used as a therapeutic modality in the NPC1 patient community we tested the ability of HPBCD to correct the P450 system deficit noted in Fig 4. Npc1 -/mice treated with HPBCD had a significant increases in all 4 enzyme activity levels MROD, EROD, PROD, and BROD compared to the Npc1 -/untreated mice (P450 activities untreated mice. UDCA treatment restored Npc1 -/-Cyp3a levels to that of Npc1 +/+ mice while have no effect on the Npc1 +/or Npc1 +/+ Cyp3a values. (D) Relative protein quantitation of Cyp1a2 isolated from liver of 6-week old male mice found a significant decrease in Npc1 +/and Npc1 -/untreated mice compared Npc1 +/+ untreated mice. UDCA treatment restored Npc1 -/-Cyp1a2 levels to that of Npc1 +/+ mice while decreasing the levels for Npc1 +/and Npc1 +/+ Cyp1a2 values. Data are presented as mean ± SD, n = 3. * p-value < 0.05, ** p-value < 0.01, *** p-value < 0.001 calculated using two-tailed, unpaired t test. (F) Representative western blot analysis of Cyp3a and Cyp1a2 in the mouse model of Npc1 plus or minus UDCA treatment.
To determine if the murine dysfunction in the cytochrome P450 system was species specific or a more global NPC finding we interrogated the NPC1 feline model. In the NPC1 feline model, although there was no significant difference in total cytochrome C reductase activity (data not shown), the activities of the specific P450 catalysed dealkylation reactions were significantly lower in NPC1 +/and NPC1 -/cats compared to controls (Fig 8A-8C). In liver tissue from NPC1 -/cats, MROD, EROD, and PROD activity levels (expressed as percentage of age-matched controls) were 19.4 ± 3.9 (p = 0.006), 18.7 ± 5.6 (p = 0.0008), and 40.6 ± 3.1 (p = 0.003) respectively (Fig 8A-8C). However, there was no significant reduction in BROD activity in the NPC1 -/cats (36.5 ± 6.6, p = 0.38 vs NPC1 +/+ ) (Fig 8D). Whilst activity was reduced to 36.5% of the control mean this did not constitute a significant difference due to high variability in the controls. Consistent with the murine model, decreased activity was also found for some P450 isozymes in heterozygous NPC1 +/cats MROD, EROD, and PROD activity levels (expressed as percentage of age-matched controls) were 39.2 ± 10.4 (p = 0.026 vs NPC1 +/+ ), 56.9 ± 12.1 (p = 0.035 vs NPC1 +/+ ), and 41.7 ± 10.4 (p = 0.003 vs NPC1 +/+ ) respectively (Fig 8A-8C). BROD showed little or no reduction in NPC1 +/cats compared to littermate controls ( Fig 8D).
Treatment with HPBCD (8000 mg/kg administered subcutaneously every 7 days) did not correct enzyme activity of MROD, EROD, PROD, or BROD in NPC1 -/cat liver (Fig 8A-8D). Interestingly, treatment with HPBCD did result in the normalization of CYP3A protein expression levels (relative protein quantitation NPC1 -/-0.15 ± 0.04 vs NPC1 -/-HPBCD treated 0.83 ± 0.006 as a function of age-matched controls, p = 0.003) (Fig 8E and 8F). NPC1 -/cats treated with HPBCD showed a significant increase in the expression of CYP3A, although this increase in protein did not appear to translate into increased enzyme activity (Fig 8A-8D). Percent change of enzyme activity rates for MROD, EROD, PROD and BROD at 6 and 9 weeks of age in Npc1 +/and Npc1 -/mice treated with 0.5% UDCA as compared with Npc1 +/and Npc1 -/mice respectively. Data are presented as mean ± SEM, n = 6, calculated using two-tailed unpaired nonparametric Mann-Whitney test. doi:10.1371/journal.pone.0152007.t004 Defective Cytochrome P450-Catalysed Drug Metabolism in Niemann-Pick Type C Disease We further extended these studies to NPC1 patients, utilizing post-mortem liver samples, isolated from NPC1 patients and age and gender matched controls. We observed lower cytochrome P450 reductase activity in all three NPC1 individuals (Fig 9A), as well as a significant reduction when combining the data in aggregate (10.5 ± 0.4 (controls) vs 2.6 ± 1.3 (NPC1 patients) pmol/min/mg, p = 0.005) (Fig 9B). We also observed decreased MROD and EROD activities, although due to high individual variation these differences were not statistically significant (Fig 9C). No apparent differences were noted in PROD or BROD activities (Fig 9C). Relative protein quantitation analysis of CYP3A revealed no differences in aggregate, however there was an increase in CYP3A in the paediatric NPC1 sample, and a large decrease in the adult NPC1 patient compared to their respective controls (Fig 9D-9E). Absolute quantitation of the CYP1A2, CYP2B6 and CYP3A4 isozyme mRNA isolated from the human livers revealed a notable reduction in expression of all three genes in the NPC1 patients compared to healthy age and gender matched controls (Fig 10).

Discussion
We hypothesized that Npc1 -/mice have defective hepatic drug metabolism based on evidence that (i) liver metabolised compounds caused toxicity when administered at doses well tolerated in other mouse models/strains [12] (Table 1) and (ii) no evidence of toxicity has been noted for renally excreted compounds such as miglustat [45] in Npc1 -/mice [35]. Since cytochrome P450 enzymes facilitate the clearance of the drugs previously shown to have increased toxicity in Npc1 mutant mice, this superfamily of enzymes was investigated further in NPC1 animal models. Cross species comparisons are challenging due to interspecies variability, namely the number of isozymes and their specific activities [46]. For example, in human and mouse hepatocytes, P450 isoforms show selectivity for specific substrates [47,48]. In humans EROD and MROD are mainly catalysed by CYP1A, whilst BROD is catalysed predominantly by CYP3A (CYP1A and CYP2A to lesser extent) and PROD is catalysed by CYP2B [47]. One confounding factor is that the clusters containing the genes encoding Cyp2a, Cyp2b and Cyp3a isozymes are greatly expanded by gene duplication in the mouse compared to human [46,49] and this close similarity of sequences of Cyp paralogs makes the identification of orthologs difficult. A second confounding issue is there are significant differences between the male and female gender in the rate of drug metabolism and the gene expression of members of the cytochrome P450 family [50], as was observed in Fig 6. We first investigated the P450 system in the murine model of NPC1 disease. In agreement with our empirical observations, the rates of the individual cytochrome P450-catalysed O-dealkylation reactions were reduced in Npc1 -/mice at all ages evaluated, consistent with microarray results indicating a global down regulation of the genes in this pathway as early as 1 week of age (S2 Fig) [9]. Notably, impaired P450 activity occurred prior to the onset of clinical manifestations. This reduction in the O-dealkylation reactions was also present in heterozygous Npc1 +/mice from 6 weeks of age. Consistent with these findings, Npc1 -/and Npc1 +/mice showed a significant reduction in cytochrome C reductase activity at 9 weeks of age compared to age-matched control mice (S1 Fig). While Npc1 -/mice consistently presented with decreased activity and gene expression the Npc1 +/mice were more variable in nature. Npc1 +/-(C-F) Effects of UDCA on tremor in Npc1 -/mice. Tremor was monitored between 0 and 63 Hz in panels (C) and (D), and quantified at 43 Hz in (E) early symptomatic (9 weeks), (F) late symptomatic (11 weeks) time points. 43 Hz is an arbitrary point selected in the high frequency tremor range to allow a simple comparison to be made between treatment groups. The low frequency tremor range overlaps at the low end with grooming activity frequencies and so was not analysed to avoid this as a potential confounder. In both early symptomatic and late symptomatic groups, mice treated with 0.5% UDCA had significantly less tremor than untreated Npc1 -/-. Data are presented as mean ± SD, n = 7-19 animals/group, * p-value < 0.05, ** p-value < 0.01, *** p-value < 0.001, **** p-value < 0.0001 calculated using one-way Anova with Bonferroni's multiple comparison test.
doi:10.1371/journal.pone.0152007.g005  . Data are presented as mean ± SEM, n = 4 (males), * p-value < 0.05, *** p-value < 0.001, **** p-value < 0.0001 calculated using an unpaired nonparametric test Mann-Whitney test. (E) Relative protein quantitation of Cyp3a found a significant decrease in Npc1 -/untreated mice compared Npc1 +/+ untreated mice similar to what was seen in Fig 4C. HPBCD treatment increased Cyp3a protein levels in Npc1 -/mice, both in comparison to the untreated Npc1 +/+ or Npc1 -/mice. (F) Relative protein mice presented with decreased activity while gene expression was only decreased in 2 of the genes assayed. Determination of the protein levels of Cyp3a and Cyp1a2 in the Npc1 +/mice revealed significant decreases consistent with the enzyme activity assays but divergent of the gene expression data. In addition, it is possible that a greater portion of the 101 known Cyp genes are down regulated, but were not assayed here or that the increased expression of a subset of genes is not sufficient to correct the decrease expression of other Cyp genes.
To validate the importance of the observed CYP deficiencies in vivo, we administered the CYP3A metabolised drug, midazolam, to Npc1 -/mice and their wild-type counterparts. Plasma levels of both parent midazolam and CYP3A-mediated midazolam metabolites were increased compared to plasma levels observed in control animals. Increased plasma midazolam levels are consistent with decreased CYP3A4 activity; however, the increase in midazolam metabolites suggests that phase II conjugation of midazolam metabolites (by UGTs) is also impaired in Npc1 mutant mice. Consistent with this hypothesis, UGT expression was globally decreased similar to decreases in expression of CYPs. Therefore, midazolam metabolism occurs more slowly, but increased exposure to metabolites occurs as a function of a reduction in Phase II conjugating enzymes. As a consequence of impaired drug metabolism, the therapeutic range of prodrugs requiring metabolism to be activated and drugs undergoing hepatic clearance may be altered in NPC1.
To determine if the cytochrome P450 defects are ubiquitous to NPC1 disease as a whole or if this is a species-specific manifestation in the Npc1 -/mouse model we investigated the feline model of NPC1. Significant dysfunction was noted with reduced activity of O-dealkylation catalyzing enzymes. Protein levels of CYP3A, responsible for about fifty percent of drug metabolism, were decreased in liver from affected cats. HPBCD treatment of the NPC1 cats partially corrected the P450 system by restoring the activities of the O-dealkylation catalyzing enzymes and increased expression of CYP3A. Therefore, it appears HPBCD may partially correct hepatic metabolism deficiencies in addition to improving neurological presentations of NPC1 disease. The mechanism underpinning this selective correction remains unclear, however what is clear is the deregulation and dysfunction of the cytochrome P450 system in NPC1 extends beyond the murine model.
We also measured cytochrome P450 and isozyme activity in post-mortem liver samples from NPC1 patients along with age and gender matched controls. A reduction in activity, consistent with the data from the NPC1 feline and mouse models, was observed. These data suggest that patients with NPC1 may have an increased susceptibility to drug toxicities or drugdrug interactions. A significant limitation of this research is the number of samples available and the manner in which they have been collected. We are only aware of these three NPC1 liver samples in any of the tissue banks worldwide and the variation in the age of the individuals at the time of collection compounds the difficulty in interpreting this data. The NIH has not collected liver biopsies on any NPC1 patients to date. While we are unaware of any direct clinical manifestation or decrease drug tolerances in NPC1 patients, the possibility exists that there could be a subclinical defect in cytochrome P450 system that will need further investigation.
We hypothesised that dysregulation of cytochrome P450 enzymes is linked to the defect in late endosome/lysosome to ER cholesterol efflux. Two pathological processes (S4 Fig) are likely involved: first, an absence of oxysterol activation of LXR, and secondly reduced bile acid quantitation of Cyp1a2 revealed a significant decrease in Npc1 -/untreated mice compared Npc1 +/+ untreated mice. HPBCD treatment increased Npc1 -/mice Cyp1a2 protein levels in comparison to untreated Npc1 -/mice, but did not restore to Npc1 +/+ levels. Data are presented as mean ± SD, n = 4. * pvalue < 0.05, **** p-value < 0.0001 calculated using two-tailed, unpaired t test. (G) Representative western blot analysis of Cyp3a and Cyp1a2 isolated from 7 week old male Npc1 +/+ or Npc1 -/mouse livers plus or minus HPBCD treatment. doi:10.1371/journal.pone.0152007.g007 Defective Cytochrome P450-Catalysed Drug Metabolism in Niemann-Pick Type C Disease synthesis or the synthesis of atypical bile acids as a consequence of cholesterol sequestration in LE/Lys. Enzymatically generated oxysterols are known to be deficient in NPC1 [34]. The reduction in these LXR agonists likely results in the decreased expression of LXR regulated genes, including CYP7A1, which catalyses the first step in bile acid biosynthesis [51][52][53][54]. This could explain the improvement in the O-dealkylation reactions noted in the feline NPC1 -/-HPBCD treated animals, as HPBCD has been shown to increase bioavailability of endolysosomal sequestered cholesterol, leading to increased oxysterol production [16].
Bile acid biosynthesis is also known to be significantly lower in Npc1 -/mice compared to control mice when challenged with a high cholesterol diet, likely due to Npc1 -/mice failing to increase bile acid biosynthesis [55,56]. Bile acids are endogenous ligands for PXR, which regulates the expression of a large number of Phase I and Phase II drug metabolising enzymes [57]. Thus, it is plausible that one mechanism leading to decreased cytochrome P450 mediated drug metabolism in NPC1 is the decreased enzymatic oxysterol production and subsequent bile acid deficiency. For example, a novel 7-oxo and a high frequency of sulphated and additionally conjugated forms of various cholenoic acids in urine have been detected and are indeed potential biomarkers of the disease [31][32][33]. We therefore tested whether bile acid deficiency/imbalance contributes to the P450 defect identified in NPC1 disease. Npc1 mice were supplemented orally from 3 weeks of age with UDCA, which normalized or led to supraphysiological P450 enzyme activities, confirming that bile acid deficiency/imbalance likely participates to the drug metabolism defect in NPC1. Of some concern are the supraphysiological P450 enzyme activity levels reached by all mice receiving UDCA therapy. These elevated activities could lead to increased rates of drug metabolism requiring greater doses to obtain the desired physiological effect or in extreme case may expose patients to greater risk of developing various forms of cancer [58]. However, the dosing used in this mouse model was based on literature values and should be considered to be a proof of concept to probe the mechanism of P450 suppression in NPC1. The dose is approximately 30-40 fold higher than standard clinical dosing in humans. Thus, we are now investigating whether doses more typical of clinical practice can normalize P450 activity in the NPC1 mouse.
Bile acid supplementation with UDCA, in addition to correcting the P450 defect, also reduced tremor and improved motor function in the Npc1 mice. The mechanism underlying these observed neurological benefits is not known. One possible explanation is that UDCA ameliorates the mitochondrial defect present in NPC1 [59,60], resulting in increased adenosine 5'triphospate synthesis and improved neuromuscular function. Recent studies by Mortiboys et al. demonstrated marked improvement in mitochondrial function in both parkin-mutant fibroblasts, and the LRRK2 G2019S fly model of Parkinson disease when treated with UDCA [61,62]. Of note that mitochondrial dysfunction has been reported in NPC1 disease [59].
The P450 superfamily is known to be differentially regulated in the brain [63]. Secondary effects in the brain within this superfamily have been demonstrated in NPC1 mouse, including the reduction in Cyp46a1 [34]. It is unclear what role CYP gene products in the brain play in the metabolism of xenobiotics that cross the blood-brain barrier [64]. However, evidence exists that variation in the activity of the different CYP family members can result in alterations in xenobiotic responses in the brain. For example, Elbaz et al. demonstrated that polymorphisms within CYP2D6 in farmers exposed to pesticides resulted in an increased risk of developing Parkinson's disease [65]. It is therefore possible that some of the neuropathology in NPC results from reduced xenobiotic metabolism due to lower expression of the P450 system in the brain.
The impaired function of both phase I and II drug metabolism reported in this manuscript has multiple potential preclinical and clinical implications. In preclinical studies, increased toxicity of liver metabolized drugs may occur and drug efficacy may be masked by toxicity due to impaired P450 function. Thus, potentially disease-modifying therapies may have been  discounted in previously conducted animal model studies. Adjusting standard doses proportional to the reduced P450 capacity (Table 1) and concurrent utilization of bile acid therapy may mitigate this problem and should be considered for future proof of concept animal model studies utilising drugs that undergo hepatic metabolism. It is interesting to note that some drugs demonstrating efficacy in the NPC1 mouse model, such as curcumin, inhibit P450 enzymes, suggesting that the positive benefits of this weak SERCA antagonist outweigh drug toxicity effects at certain doses. Heterogeneity in response to this natural product has been reported [66,67]. As the outcome in the Npc1 -/mouse treated with curcumin may depend on the balance between xenobiotic levels in mouse chow [68,69] (variable between suppliers and in different countries) and residual P450 function in response to the stage of the disease when the therapy starts, it is possible that this underlies the variation in response to curcumin reported by different laboratories.
Altered P450 metabolism may also explain an intriguing observation regarding liver disease in NPC1. Cholestatic jaundice and hepatic dysfunction is frequently present in neonates with NPC1. Although this may be lethal, it frequently resolves. The normal maturation of both phase I and II metabolism post birth is well described in healthy individuals [70]. Although additional investigation is necessary, it is plausible that impaired phase I and II metabolism in NPC1 in conjunction with the developmental immaturity of these metabolic pathways contributes to the neonatal hepatic manifestations of NPC1.
In conclusion, we have identified a novel drug metabolism defect in NPC1 disease in multiple species (mouse, and feline) as well as providing initial evidence for disturbance in NPC1 patients. These effects can be ameliorated by treatment with bile acid. Treatment with UDCA also decreased tremor severity and improved motor function in the Npc1 mutant mice suggesting a neurological benefit. These findings suggest that UDCA therapy may be a useful adjunctive therapy in NPC1 disease.

Materials and Methods Animals
Npc1 mouse model. Npc1 mutant (BALB/cNctr-Npc1m1N/J, Npc1 -/-), control (Npc1 +/+ ) and Npc1 heterozygous mice (Npc1 +/-), were generated from heterozygote breeding. Genotyping was performed as described by Loftus et al [10]. All mice were maintained under a standard 12h light/12h dark cycle with water and food available ad libitum. All procedures conducted in Oxford were performed according to the Animals (Scientific Procedures) Act 1986 under a project license (PPL No. 30/2923) from the UK Home Office. All mouse work conducted at the National Institutes of Health conformed to NIH guidelines and was approved by the NICHD Institutional Animal Care and Use Committee.
Feline NPC1 Samples. Cats were raised in the animal colony of the School of Veterinary Medicine, University of Pennsylvania, under the guidelines of the National Institutes of Health and US Department of Agriculture for care and use of animals in research. The University of Pennsylvania Institutional Animal Care and Use Committee approved this study. Liver samples from the feline model of NPC1 disease were from age-matched cats (24 weeks old) (NPC1 +/+ , NPC1 +/and NPC1 -/-) in addition to samples from HPBCD-treated cats (subcutaneous, 8000 mg/kg/ every 7 days beginning at 3 weeks of age), obtained at different ages (21, 38 and 59 weeks old).

Human Samples
Human tissue was obtained from the National Institute of Child Health and Human Development Brain and Tissue Bank for Developmental Disorders at the University of Maryland, Baltimore, MD, http://medschool.umaryland.edu/BTBank/, contract HHSN275200900011C, Ref.
No. N01-HD-9_0011. Liver samples were collected at autopsy from three NPC1 cases patients (UMB# M4002M, UMB# M4003M, UMB# M4004M), and three age-and gendermatched controls (UMB#1500 6 years old had a perforated bowel, haemopericardium, hypoxic ischaemic encephalopathy, secondary to a motor vehicle accident. UMB#662 12 years old died of a traumatic accident with multiple injuries. UMB#1672 died of asphyxia). The National Institute of Child Health and Human Development Institutional Review Board approved human studies.

Cytochrome P450 Activity Assay
Hepatic Microsomal Preparation. Livers were harvested, weighed and minced in ice-cold KCl (1.15%, w/v). Samples were homogenized using a Polytron in ice-cold SET buffer (0.25 M sucrose, 5 mM EDTA and 20 mM Tris-HCl, pH 7.4) in a 1:3 (w/v). Liver homogenates were subjected to differential centrifugation at 4°C as follows: homogenates were first centrifuged at 1,000 g for 20 minutes, the post-nuclear supernatants were spun at 12,000 g for 30 minutes and the resultant post-mitochondrial fractions (supernatants) were centrifuged at 100,000 g for 60 minutes. Microsomal pellets were re-suspended in ice-cold homogenisation (SET) buffer (1:3 w/v). Microsomal protein concentrations were measured using the bicinchoninic acid kit (Sigma-Aldrich, UK) according to the manufacturer's instructions with bovine serum albumin as the standard.
Enzyme Activity. P450 reductase activity was measured spectrophotometrically (Jenway 6305) following the NADPH-dependent reduction of cytochrome C using a commercial cytochrome P450 reductase assay kit (Cytochrome C Reductase (NADPH) Assay Kit, CY0100, Sigma-Aldrich). Microsomal preparations (50 μg protein) were incubated with 950 μL of 36 μM cytochrome C in potassium phosphate buffer (300 mM, pH 7.8) containing 0.1 mM EDTA. The mixture was incubated at 25°C and the reaction initiated by the addition of 100 μL of 0.85 mg/ml NADPH. The rate of the reduction of cytochrome C reaction was followed at 550 nm. Activity was expressed as nmol cytochrome C reduced per min per mg of microsomal protein, using extinction coefficient of 21.1 mM -1 .cm -1 . To investigate the individual subfamilies of P450 system enzymes, the following reactions were measured: ethoxyresorufin O-dealkylation (EROD) for CYP1A1, 1A2, methoxyresorufin O-dealkylation (MROD) for CYP1A2, pentoxyresorufin O-dealkylation (PROD) for CYP2B and benzoxyresorufin O-dealkylation (BROD) for 3A (1A, 2A and 2B to a lesser extent) [47,48,71,72]. All samples were assayed in duplicate. Microsomal preparations were incubated with 50 mM Tris-HCl (pH 7.6), MgCl 2 (25 mM), liver microsomes (0.5 mg protein/ml) and 5μM of the substrate (7-methoxyresorufin, 7-ethoxyresorufin, 7-pentoxyresorufin and 7-benzyloxyresorufin, obtained from Sigma-Aldrich, UK). Reactions were started by adding NADPH (0.25 mM, final concentration) and after 30 minutes of incubation at 37°C, the amount of resorufin formed was spectrofluorimetrically measured with a microplate reader (FLUOstar OPTIMA), using wavelengths for excitation (range 560-570 nm) and emission (range 590-600 nm). The amount of product formed, as an index of enzyme activity, was determined by comparison to a standard curve made with pure resorufin (Sigma-Aldrich) (0-250 pmol) dissolved in DMSO and made up to the final volume with Tris-HCl/MgCl 2 buffer.

Western Blotting
Frozen liver tissue was homogenized in RIPA buffer with protease (Roche) and phosphoprotease inhibitors (sodium fluoride, sodium orthovanadate). Protein lysate concentrations were determined using the DC Protein Assay (Bio-Rad). Proteins lysates (15 to 50ug) were resolved in a 4-12% Bis-Tris NuPage gel using the MES buffer system (Invitrogen). Transfer was carried out using the iBlot System (Invitrogen) onto nitrocellulose membranes (human) or transferred onto Hybond-P polyvinylidene difluoride (PVDF) membrane (GE healthcare, Chalfont St. Giles, UK) using the Xcell II Blot Module (Invitrogen) (feline and murine). Membranes were blocked in 5% BSA prepared in Tris-buffered saline containing 0.1% Tween 20 (PBS-T) buffer at 4°C overnight. Incubation with the CYP3A (mouse, Santa Cruz Technologies) primary antibody at a dilution of 1:500 or CYP1A2 (mouse, Santa Cruz Technologies) primary antibody at a dilution of 1:1000 was carried out at 4°C overnight. Washing with PBS-T was carried out prior to incubation with secondary antibody. Incubation for 2 hours with the anti-GAPDH (1:2000 dilution, rabbit) (Cell Signaling Technologies) or monoclonal anti-β-actin antibody-peroxidase conjugate (Sigma) was used as a loading control (at 1:10,000 dilution). Again the membrane was washed with PBS-T. HRP-conjugated anti-mouse and anti-rabbit secondary antibodies were used for developing (1:10,000 dilution, Sigma) or immediately visualized. The Bio-Rad HRP chemiluminescent kit was used for exposure. Detection of bands and quantification was performed using the Bio-Rad Gel Doc Imaging System.

Gene expression
Microarray hybridization and data analysis. Microarray analysis previously reported [9]. Real-time PCR. Total RNA (10 μg), from the same set of livers used for the microarray analysis, was reverse-transcribed into cDNA using a High-Capacity cDNA archive kit according to the manufacturer's instructions (Applied Biosystems). The following Taqman assays were used for the validation of the micro-array data (Applied Biosystems): Cyp2b10 (Mm00456588_mH), Cyp2b13 (Mm00771172_g1), Cyp2c37 (Mm00833845_m1 Ã ), Cyp2c40-67-68 (Mm04204172_mH, amplifying the three Cyp2c40, Cyp2c67 and Cyp2c68 mRNA), Cyp2c50 (Mm00663066_gH Ã ), Cyp2c54 (Mm02602271_mH Ã ), Cyp3a16 (Mm00655824_m1 Ã ), Cyp3a41a-b (Mm00776855_mH, amplifying both Cyp3a41a and Cyp3a41b mRNA), Cyp3a44 (Mm01703321_mH Ã ), Cyp1a2 (Mm00487227_m1) and Cyp2a4 (Mm00487248_g1). Gapdh gene assay was used as reference (Taqman Rodent GAPDH control reagents, Applied Biosystems). Quantitative real-time PCR was performed in 384-well plates with Applied Biosystems 7900 real-time PCR system, using the same series of samples than for the microarrays. Each sample was analyzed in triplicate, using 50 ng of total cDNA for each reaction. The relative quantification of gene expression was performed with the comparative cycle number measured with the threshold method (C T ), using the 1-week-old control samples as a reference for quantification. A two-tailed unpaired t-test, with Welch's correction when necessary was performed to assess the significance of the difference of means between control and Npc1 -/samples at each age.
The PCR product was then loaded on a 1% agarose gel for purification of each fragment using the QIAquick gel extraction kit (QIAGEN). DNA concentration was determined using a Qubit fluorometer (Invitrogen). Absolute quantitation PCR was performed in 96-well plates with Applied Biosystems 7300 real-time PCR system, using the following Taqman assays (Applied Biosystems): CYP1A2 (Hs01070369), CYP2B6 (Hs03044634), and CYP3A4 (Hs00430021). Each sample and standard was analyzed in triplicate, using 50ng of total cDNA for each sample, and 10-fold serial dilutions of the standard fragment ranging from 10 6 to 10 1 copies. cDNA concentration was then determined for each sample using the standard curve generated by the six standards.

Functional study
plasma by the addition of 200 μL of 2 mM sodium hydroxide to precipitate plasma proteins, followed by 1 ml of 4000 ng/ml LZP (internal standard; IS) in MTBE. Samples were vortexed, centrifuged, dried down under a stream of desiccated air, and reconstituted with 100μL of (45/ 55, v/v) 10 mM ammonium acetate/methanol. Samples were again vortexed and transferred to a mini-centrifuge tube for further centrifugation, before the supernatant was transferred to a glass Agilent HPLC vial and injected onto an LC/MS system. A gradient elution was used to separate midazolam, the metabolites, and lorazepam, and consisted of an initial mobile phase of (45/55, v/v) 10 mM ammonium acetate/methanol that was increased linearly to 75% methanol over 8 min, then immediately back to 55% B for the remainder of the 10 min run. The column used was a Zorbax SB-C18 (2.1x75mm, 3.5μm), flow rate was 0.35 ml/min, column temperature was 35°C, and the injection volume was 10 μL. Midazolam, metabolites, and IS were detected using single ion monitoring (SIM) by mass spectrometry in the positive electrospray ionization mode, with one SIM for MDZ (m/z 326.1), one for both metabolites (m/z 342.1), and the third for LZP (m/z 321.2). Because both metabolites had the same molecular weight, thus same molecular ion ([M-H] + ), the chromatography was able to separate each metabolite for individual identification; however both peak areas were integrated together as one for simplicity. Other mass spectrometric conditions were: capillary voltage 2500V, fragmentor 100V, drying gas temperature 330°C, and a drying gas flow of 5 L/min. A typical LC-MS chromatogram is depicted in S3 Fig.

Statistical Analysis
A two-tailed, unpaired, nonparametric (Mann-Whitney) or student t-test was performed to compare the residual enzyme activities or protein expression of the different groups. One-way ANOVA with Tukey's post-hoc test was used to compare all sets of data where applicable. A two-tailed, unpaired t test, with Welch's correction for heteroscedasticity when necessary, was performed to determine the significance of the difference in cytochrome P450 gene expression levels between Npc1 +/+ , Npc1 +/and Npc1 -/mice at each age. Behavioural studies were analysed by using one-way/two-way ANOVA with Bonferroni's multiple comparison test or an unpaired nonparametric test (Mann-Whitney) as appropriate. Statistical analysis was performed with GraphPad Prism.
For the midazolam functional study, using Bailer's Method for destructive sampling, exposures (AUC) for both parent midazolam and metabolites, were calculated and compared between the genotypes. The mean (n = 3) plasma concentrations of each analyte (midazolam or metabolites) at each time point were averaged and a mean AUC ± standard deviation was calculated for each genotype (total n = 12; 3 mice at 4 time points). The AUC values were then subjected to both a one-way ANOVA as well as unpaired Student's t-tests (with Welch's correction) for direct comparisons between genotypes.
Supporting Information S1 Fig. Activity of Cytochrome C reductase in mouse model of NPC1 at 9 weeks of age. Data are expressed as pmoles cytochrome C reduced per minutes normalised for microsomal protein content (mg). Data are presented as mean ± SEM, n = 5, Ã p-value <0.05, ÃÃ pvalue < 0.01, calculated using two-tailed unpaired nonparametric Mann-Whitney test. Microarray data and subsequent qPCR validation of 9 cytochrome P450 gene expression performed at 1, 3, 5, 7, 9 and 11 weeks old in the Npc1 mouse model. qPCR (left axis) and array data (right axis) are presented on the same graph for comparison. Control mice samples figure in white, Npc1 -/mice in black, n = 4. Only females were used. A two-tailed, unpaired t test, with Welch's correction when necessary, was performed to determine the significance of the difference in means between control and mutant mice at each age: Ã p-value<0.05; ÃÃ p-value<0.001; ÃÃÃ p-value < 0.0001. Result of the ANOVA test is shown for the array data. In the presence of a functional NPC1 and NPC2 protein, LDL-derived cholesterol is trafficked out of the lysosome where it can be converted to side chain oxysterols. Side chain oxysterols activate LXR target genes, including bile acid synthesis enzymes. Bile acids in turn activate other nuclear receptors FXR, CAR and PXR. These nuclear receptors stimulate the transcription of cytochrome P450 enzymes. However, NPC1/NPC2 dysfunction (changes in red) reduces side chain oxysterols and subsequently down regulates the expression levels of downstream enzymes and nuclear receptors, resulting in low expression levels of the xenobiotic metabolizing enzymes and transporter proteins. (TIF) S1 Table. Micro array analysis of the 62 differentially expressed cytochromes P450 genes. List of the 62 genes encoding cytochromes P450, with a modified expression in 1, 3, 5, 7, 9, and 11-week-old Npc1 -/mice compared to their control littermates. 42 of these genes belong to the subfamilies 1 to 3 mainly responsible for drug metabolism. FC: fold-change, ns: not significant. (DOCX) S2 Table. Micro array analysis of the 15 differentially expressed UGT genes. List of the 15 genes encoding UGT, with a modified expression in 1, 3, 5, 7, 9, and 11-week-old Npc1 -/mice compared to their control littermates. FC: fold-change, ns: not significant. (DOCX)