Cyclodextrin mediates rapid changes in lipid balance in Npc1-/- mice without carrying cholesterol through the bloodstream.

An injection of 2-hydroxypropyl-β-cyclodextrin (HP-β-CD) to mice lacking Niemann Pick type C (NPC) protein results in delayed neurodegeneration, decreased inflammation, and prolonged lifespan. Changes in sterol balance observed in Npc1(-/-) mice 24 h after HP-β-CD administration suggest that HP-β-CD facilitates the release of accumulated lysosomal cholesterol, the molecular hallmark of this genetic disorder. Current studies were performed to evaluate the time course of HP-β-CD effects. Within 3 h after HP-β-CD injection, decreases in cholesterol synthesis rates and increases in cholesteryl ester levels were detected in tissues of Npc1(-/-) mice. The levels of RNAs for target genes of sterol-sensing transcription factors were altered by 6 h in liver, spleen, and ileum. Despite the cholesterol-binding capacity of HP-β-CD, there was no evidence of increased cholesterol in plasma or urine of treated Npc1(-/-) mice, suggesting that HP-β-CD does not carry sterol from the lysosome into the bloodstream for ultimate urinary excretion. Similar changes in sterol balance were observed in cultured cells from Npc1(-/-) mice using HP-β-CD and sulfobutylether-β-CD, a variant that can interact with sterol but not facilitate its solubilization. Taken together, our results demonstrate that HP-β-CD works in cells of Npc1(-/-) mice by rapidly liberating lysosomal cholesterol for normal sterol processing within the cytosolic compartment.


Animals and treatments
Heterozygous ( Npc1 +/ -) mice on a BALB/c background were bred to generate wild-type ( Npc1 +/+ ) and homozygous-null ( Npc1 Ϫ / Ϫ ) littermates ( 19 ). Mice were group-housed in plastic cages containing wood-chip bedding in an animal facility with temperature-controlled rooms (23 ± 1°C) and a maintained light cycle (12 h light on/12 h light off). The mice were allowed ad libitum access to water and a standard rodent diet containing 0.02% w/w cholesterol (7001; Harlan Teklad, Madison, WI). Mice were genotyped upon weaning between 19 and 21 days of age ( 19 ). For all in vivo studies, 49-day-old mice were given a single subcutaneous injection of HP-␤ -CD (H107, Sigma) at 4,000 mpk or vehicle [saline at equivalent volume of 20 l/g body weight (bw)]. When mice were used as a source of cells for culture, macrophages were obtained from 2-month-old mice, and hippocampal neurons were harvested from 1-day-old pups. As no differences in response to HP-␤ -CD were evident between male and female mice in our previous reports ( 5,9,12,13,17,18 ), these studies utilized either equal numbers of males and females per group ( Figs. 1, 4D, 5-7 ) or only females ( Figs. 2, 3 , 4A-C ). All animal research was conducted in conformity with the Public Health Service Policy on Humane Care and Use of Laboratory Animals, and all experiments were performed with prior approval from the Institutional Animal Care and Use Committee of the University of Texas Southwestern Medical Center.

Measures of cholesterol balance
Npc1 +/+ and Npc1 Ϫ / Ϫ mice (n = 4-6 mice/group) were given a subcutaneous injection of HP-␤ -CD or saline. At time points ranging from 0 to 12 h after the injection, the mice were weighed and euthanized (exsanguinated under deep anesthesia) to harvest tissues for the analyses described below.
Sterol synthesis rate. Exactly 1 h prior to euthanasia and tissue harvest, one cohort of mice was given an intraperitoneal injection of 1.76 mCi of [ 3 H] water/g bw. Aliquots of liver and spleen were quickly removed, weighed, and saponifi ed in alcoholic KOH. Organ contents of [ 3 H]labeled digitonin-precipitable sterols (DPS) were determined as previously described ( 20 ). The rate of sterol synthesis is reported as the nanomoles of [ Tissue cholesterol concentration. The aliquots of liver and spleen in KOH used for the synthesis determination were also used to measure total cholesterol content. Briefl y, the cholesterol was extracted and quantifi ed by gas chromatography (GC) with a stigmastanol internal standard as previously described ( 21 ). The total cholesterol concentration of each organ is expressed as the milligrams of cholesterol per gram of tissue.
Unesterifi ed/esterifi ed cholesterol determination. From a different cohort of mice, livers were fl ash frozen and were crushed into a powder, which was weighed and solubilized in a 2:1 chloroform to methanol solution for lipid extraction. Free and esterifi ed cholesterol were separated using a silica column (Sep-Pak Vac RC; Waters Corp., Milford, MA) and were quantifi ed by GC as previously described ( 22 ). The cholesterol concentration of each organ is reported as the milligrams of unesterifi ed cholesterol or cholesteryl ester per gram of tissue.
Plasma lipoprotein profi le. Blood samples were collected using EDTA-coated syringes from the ascending vena cava of anesthetized mice. Injection times for HP-␤ -CD were staggered so that all in the Npc1 Ϫ / Ϫ mouse ( 9 ). As there are no FDA-approved treatments to extend lifespan for NPC patients, the Npc1 Ϫ / Ϫ mouse has been used as a model to test potential therapies. Therefore when HP-␤ -CD was shown to extend the lifespan of Npc1 Ϫ / Ϫ mice to 118 days with a single injection to 7-day-old pups ( 5 ), it quickly became a major focus of the NPC research community.
Since the initial studies in mice, HP-␤ -CD therapy has been shown to be effective in the Npc1 Ϫ / Ϫ cat model ( 10 ), and a NIH-sponsored clinical trial to test HP-␤ -CD efficacy in humans is being organized (http://nctt.nih. gov/2754421). Additionally, much research has been undertaken to further understand how CD is working in Npc1 Ϫ / Ϫ animal models. Notably, serial injections of HP-␤ -CD were shown to double the lifespan of the Npc1 Ϫ / Ϫ mouse ( 11,12 ), and continuous intrathecal administration was found to dramatically rescue Purkinje cell loss within the cerebellum ( 13 ). HP-␤ -CD does not require a functional NPC1 or NPC2 protein to work as suggested by studies using cell culture models (14)(15)(16) and confi rmed in mice when HP-␤ -CD was shown to signifi cantly decrease hepatic cholesterol synthesis rates in Npc1 Ϫ / Ϫ , Npc2 Ϫ / Ϫ , and rare ( 17 ). Studies performed within a variety of tissues harvested 24 h after a single HP-␤ -CD injection to 7-or 49-day-old Npc1 Ϫ / Ϫ mice have shown dramatic changes in sterol dynamics, including a decrease in cholesterol synthesis rates, an increase in the ratio of esterifi ed/unesterifi ed cholesterol, downregulation of the sterol regulatory element-binding protein 2 (SREBP2) and its target genes, and increased expression of liver X receptor (LXR) target genes ( 5,18 ). These measurable shifts in sterol homeostasis serve as surrogate markers to indicate that the cell is responding to a state of cytosolic cholesterol excess. Because these changes were observed after HP-␤ -CD administration to Npc1 Ϫ / Ϫ mice only and not seen for wild-type mice, they further suggest that the excess unesterifi ed cholesterol is being released from the LE/L compartment in Npc1 Ϫ / Ϫ mice. Overall, these data suggest that HP-␤ -CD is able to facilitate the release of the lysosomal pool of unesterifi ed cholesterol from NPC-defi cient cells within 24 h. Although virtually all studies thus far have been performed 24 h after HP-␤ -CD, a single time course experiment using [ 14 C]radiolabeled HP-␤ -CD in 49-day-old mice demonstrated that it is cleared from plasma within 3 h and from the whole body within 6 h after a subcutaneous dose of 4,000 mg/kg body weight (mpk) ( 18 ). Therefore, we performed a series of studies in Npc1 Ϫ / Ϫ mice to evaluate the time course of changes elicited by HP-␤ -CD from 1 to 12 h after administration to further understand its mechanism of action. The experiments were designed to i ) measure the rate of change in lipid balance and corresponding mRNA levels; ii ) determine whether a CD/cholesterol complex is evident at any time in plasma or urine; iii ) evaluate how rapidly infl ammation is reduced; and iv ) assess the direct effects of HP-␤ -CD and SBE-␤ -CD on primary cells (neurons and macrophages) in culture. Together, these studies provide novel insights into the mechanism by which HP-␤ -CD frees trapped unesterifi ed cholesterol from the LE/L of cells lacking NPC1. in PBS to yield fi nal concentrations of 0.3, 0.75, and 1.5 mM. For experiments to compare HP-␤ -CD and SBE-␤ -CD (Captisol; CYDEX Pharmaceuticals, Inc.), cells were treated for 16 h with 0.3 mM of either compound from a 250 mM stock solution made in PBS. After treatment, culture medium was removed from each well, and RNA-STAT60 was used to lyse cells and obtain RNA for analysis by qPCR (as described above for tissues).
Primary hippocampal neurons. At birth, pups from Npc1 +/ Ϫ breeding pairs were genotyped and sacrifi ced one hour later by decapitation. Under a dissecting scope, the hippocampus was removed, minced, and incubated at 37°C in 10% trypsin (Sigma) for 10 min. Cells were collected by centrifugation for 5 min at 300 g , and the pellet was resuspended in media [high-glucose DMEM (Invitrogen) containing 10% heat-inactivated FBS (Atlanta Biologicals) and 1% neurobasal B27 supplement (Invitrogen)]. The primary hippocampal cells were counted and plated at 2.6 × 10 5 cells/cm 2 on poly-L-lysine-coated plates. Once plated, the cells were maintained in a humid incubator at 37°C with 5% CO 2 . The next day, cells were exposed to 5 µM Ara-C (to reduce astrocyte contamination; Sigma) in DMEM supplemented with 2 mM L-glutamine. On culture day 8, cells were provided fresh media containing vehicle or treatment and then harvested for RNA using the experimental designs described above for macrophage cultures.

Measurement of plasma cytokine concentrations
For the 0 h time point, 60 µl of blood was harvested from the tail vein of Npc1 +/+ or Npc1 Ϫ / Ϫ mice. The mice were allowed to recover for 1 day, and then mice were injected with HP-␤ -CD or saline. Then 3, 6, 12, and 24 h later, another 60 µl aliquot of blood was harvested from the tail. Each tail-blood isolation was collected into a tube with 3 µl of 0.5 M EDTA on ice. Blood was spun down at 1 × 10 4 g for 10 min at 10°C, and plasma was moved to a fresh tube, fl ash frozen in liquid nitrogen, and stored at Ϫ 80°C until analyses for cytokine levels were performed. Ten microliters of each sample was analyzed in duplicate, alongside standard curves, on a 7-plex ultrasensitive proinfl ammatory mouse cytokine plate (#K11012C-1; MesoScale Discovery, Gaithersburg, MD) according to the manufacturer's instructions. The plate was read using a SECTOR® Imager 6000 instrument (MesoScale Discovery). Concentrations for IFN ␥ , interleukin (IL)-1 ␤ , IL-6, IL-10, IL-12p70, KC/GRO, and tumor necrosis factor (TNF) ␣ are reported as picogram (pg) of protein per milliliter of plasma.

Data analysis
All data are presented as the mean ± SEM. Statistically significant differences ( P < 0.05) between the Npc1 +/+ controls and the 0 h CD or saline-treated Npc1 Ϫ / Ϫ group were tested using an unpaired Student t -test and are represented by a double dagger ( ‡ ).

Signifi cant differences between Npc1
Ϫ / Ϫ mice studied 0 h after CD treatment and those studied at later time points were determined using a one-way ANOVA followed by Dunnett's posthoc analysis and are represented by asterisks (* P < 0.05 to *** P < 0.0005). All statistical tests were performed using GraphPad Prism5 software (GraphPad Software, Inc. San Diego, CA).

RESULTS
De novo cholesterol synthesis rates are reduced in Npc1 ؊ / ؊ mice 3 h after HP-␤ -CD injection As HP-␤ -CD is cleared within 3 to 6 h from the plasma and total body of 49-day-old mice ( 18 ), it is likely that HP-␤ -CD exerts its effect(s) on Npc1 Ϫ / Ϫ mice during this animals were exsanguinated at the same point during the light cycle [9 h after lights on or Zeitgeber time (ZT) 9]. Equal volumes of plasma/mouse were pooled for similarly treated animals (n = 2/group) and samples were provided to a core facility where lipoprotein profi les were obtained using fast protein liquid chromatography (FPLC) as described previously ( 23 ). Total cholesterol concentrations in eluted fractions were measured using both an enzyme-based method (Infi nity Cholesterol Reagent, Sigma) and by solvent extraction followed with measurement by GC (see above). No differences were found between the two methods, so only values from the enzyme-based method, expressed as micrograms of total cholesterol per fraction, are reported.
Urine cholesterol concentration. After receiving an injection of saline or HP-␤ -CD before lights out (ZT 10), a separate cohort of mice were individually housed in metabolic chambers in a maintained animal facility to collect urine for the following 48 h. While in the chamber, mice were allowed ad libitum access to food and water. Urine collections were spun at 1 × 10 4 g for 5 min at 4°C, and the supernantants were stored at 4°C. Later, the samples were dried down and then saponifi ed in alcoholic KOH. Total cholesterol was extracted and measured by GC as described above. The cholesterol content of each urine sample is reported as the micrograms of total cholesterol per 24 h.

Relative mRNA levels
Tissues. Npc1 +/+ or Npc1 Ϫ / Ϫ mice (n = 4 mice/group) were given a single subcutaneous injection of HP-␤ -CD (4,000 mpk) or saline and were euthanized 0 to 12 h later. Injection times were staggered so that all tissues were harvested at the same point during the light cycle (ZT 9). Mice were exsanguinated; then liver, spleen, and ileum (defi ned as the mucosae from the distal third of the small intestine) were collected and fl ash-frozen in liquid nitrogen. Tissues were stored at Ϫ 80°C until total RNA was isolated using RNA STAT-60 (Tel-Test, Inc.). RNA concentrations were determined by absorbance at 260 nm with a Thermo Scientifi c Nanodrop 100 Spectrophotometer. RNA was treated with RNase-free DNase (Roche) and converted to cDNA using Super-Script II reverse transcriptase (Invitrogen) as previously described ( 23,24 ). Quantitative real-time PCR (qPCR) was performed using an Applied Biosystems 7900HT sequence detection system. Each qPCR was analyzed in duplicate and contained in a fi nal volume of 10 µl: 25 ng of cDNA, each primer at 150 nM, and 5 µl of 2× SYBR Green PCR Master Mix (Applied Biosystems). The nucleotide sequences of the primers used in these analyses are listed in supplementary Table I. Results were evaluated by the comparative cycle number at threshold (C T ) method ( 25 ) using cyclophilin as the invariant housekeeping gene ( 26,27 ), and values were arithmetically adjusted to yield a unit of 1 for the control group.
Primary macrophages. Npc1 +/+ or Npc1 Ϫ / Ϫ mice (n = 4-6 mice/group) received a 1 ml intraperitoneal injection of 3% thioglycollate (autoclaved and aged for 3 months; Becton Dickinson) to elicit macrophages. Three days later, mice were euthanized, and macrophages were obtained from the peritoneal cavity by sterile lavage using ice-cold PBS. Collections from mice of the same genotype were pooled, and cells were collected by centrifugation for 5 min at 150 g at 4°C. Cells were resuspended in media [high-glucose DMEM (Invitrogen) containing 10% heatinactivated FBS (Atlanta Biologicals) and 1% Pen/Strep (Invitrogen)], counted and plated at 1 × 10 5 cells/cm 2 . Once plated, cells were maintained in a humid incubator at 37°C with 5% CO 2 . After 6 h, cells were washed with PBS and exposed to fresh media containing either vehicle (PBS) or treatment. For the CD doseresponse experiments, cells were treated for 4 h with varying amounts of a 250 mM solution of HP-␤ -CD (H107; Sigma) made timeframe. To test this, 49-day-old Npc1 Ϫ / Ϫ mice were given a single dose of HP-␤ -CD, and indices of sterol balance were characterized 0 to 12 h later. Signifi cant increases in organ weights, tissue total cholesterol content, and cholesterol synthesis rates in Npc1 Ϫ / Ϫ mice compared with their wild-type littermates were evident at the outset (0 h) of this study ( Fig. 1 ), which are consistent with previous reports ( 28 ). Over the course of these 12 h studies, the single dose of HP-␤ -CD caused no differences in organ weights or total cholesterol concentrations in liver and spleen of Npc1 Ϫ / Ϫ mice ( Fig. 1A-D ). However, a significant decrease in cholesterol synthesis rates in these tissues was evident within 3 h of the HP-␤ -CD injection ( Fig. 1E, F ). By 6 h after HP-␤ -CD administration, hepatic cholesterol synthesis rates in Npc1 Ϫ / Ϫ mice leveled off at a rate of 197 nmol/g/h and then remained steady for the duration of the 12 h study ( Fig. 1E ). This was a reduction of ‫ف‬ 80% compared with the initial rate ( Npc1 Ϫ / Ϫ 0 h, 1,010 nmol/ g/h) in liver and was even less than that measured in Npc1 +/+ mice (580 nmol/g/h). A similar timeframe of change was observed in sterol synthesis rates from spleen of Npc1 Ϫ / Ϫ mice after HP-␤ -CD administration, although the absolute rates of cholesterol synthesis were much lower compared with liver ( Fig. 1F ).

Hepatic cholesteryl ester content is increased in Npc1 ؊ / ؊ mice 3 h after HP-␤ -CD injection
Although there was no measurable change in total cholesterol content within the fi rst 12 h after HP-␤ -CD injection in Npc1 Ϫ / Ϫ mice ( Fig. 1C ), the relative quantities of unesterifi ed cholesterol (largely representing cholesterol sequestered within the LE/L in the Npc1 Ϫ / Ϫ model) and cholesteryl ester were measured in liver. A modest, albeit not statistically signifi cant, decrease in hepatic unesterifi ed cholesterol from 20.5 mg/g at 0 h to 17.1 mg/g ( Fig. 2A ) was observed 12 h after HP-␤ -CD injection in Npc1 Ϫ / Ϫ mice. In contrast, hepatic cholesteryl ester concentrations in Npc1 Ϫ / Ϫ mice were signifi cantly increased at 3 h after HP-␤ -CD injection and reached a plateau by 6 h ( Fig. 2B ). The rate of change observed in cholesteryl ester content after HP-␤ -CD administration is consistent with that observed for cholesterol synthesis ( Fig. 1E, F ).

Signifi cant changes in mRNA levels are observed 6 h after HP-␤ -CD injection
To identify molecular mechanisms that could account for the changes in cholesterol balance measured in Npc1 Ϫ / Ϫ mice just 3 h after HP-␤ -CD administration, mRNA levels for cholesterol-related genes were measured. Previously, signifi cant increases in the expression of target genes for the oxysterol-responsive nuclear hormone receptor LXR and decreases in the mRNA levels of genes regulated by the sterol-sensing transcription factor SREBP2 were reported 24 h after HP-␤ -CD treatment in Npc1 Ϫ / Ϫ mice ( 5 ). These observed mRNA changes are consistent with an increase of intracellular sterol, which would reduce de novo cholesterol synthesis via SREBP2 and enhance cholesterol effl ux and catabolism via LXR. As shown in Fig. 2 , these changes are observed in the livers of Npc1 Ϫ / Ϫ mice as early as 6 h after HP-␤ -CD treatment. The mRNA levels of the SREBP2 target genes, Hmg CoA Synthase ( Hmg CoA Syn ) and Hmg CoA Reductase ( Hmg CoA Red ), are significantly reduced 6 h after HP-␤ -CD by >85% compared with Npc1 Ϫ / Ϫ mice at 0 h to levels even lower than those measured in Npc1 +/+ mice ( Fig. 2C, E ). A decrease in the mRNA levels of Srebp2 and its target genes is also seen in spleen and ileum as early as 3 h after HP-␤ -CD treatment in Npc1 Ϫ / Ϫ mice ( Fig. 3A, B ), although the reduction observed in these tissues was not as profound as in liver. It is noteworthy that initial changes in sterol synthesis rates ( Fig. 1 ) occurred at or before the changes in mRNA levels of the rate-limiting enzyme Hmg CoA Red ( Fig. 2 ), which is consistent with rapid sterol-sensing mechanisms involving posttranslational modifi cations that affect HMG-CoA Red protein stability and activity (reviewed in Ref. 29 ).
Cholesterol 7 ␣ -hydroxylase ( Cyp7a1 ), the LXR target gene in mice, is a rate-limiting enzyme for conversion of cholesterol to bile acids in liver. It should be noted that hepatic Cyp7a1 mRNA levels exhibit a pronounced circadian rhythm ( 30 ); thus, our study design utilized staggered HP-␤ -CD injection times to limit tissue harvesting to a single time of day (ZT 9). Cyp7a1 mRNA levels were reduced by 50% in Npc1 Ϫ / Ϫ mice compared with Npc1 +/+ littermates. In Npc1 Ϫ / Ϫ mice, HP-␤ -CD administration resulted in increased Cyp7a1 mRNA levels with a trend observed as early as 6 h and a signifi cant increase (3-fold) evident at 12 h ( Fig. 2D ). The mRNA level of ATP-binding cassette transporter a1 ( Abca1 ), another LXR target gene, was found to be only transiently increased in livers of Npc1 Ϫ / Ϫ mice at 6 h after HP-␤ -CD ( Fig. 2F ). These results from liver are in contrast to Abca1 mRNA levels in spleen ( Fig. 3C ), where Abca1 expression was dramatically increased by 3 h (2.8fold) after HP-␤ -CD in Npc1 Ϫ / Ϫ mice and remained at this level throughout the 12 h study. This Abca1 expression difference between tissues can likely be explained by the fact that Abca1 is not an LXR target gene in hepatocytes, where a distinct ABCA1 promoter is utilized ( 31 ). Thus, the increase in Abca1 mRNA observed in liver is likely to result from LXR activation of other cells types, such as macrophages or resident Kupffer cells, which have very high levels of ABCA1. The decrease measured at 12 h in liver ABCA1 could be accounted for by a diminished number of macrophages. Within the fi rst 12 h following HP-␤ -CD treatment, no changes in mRNA levels were observed for the LXR target gene Abcg8 in either liver (data not shown) or ileum ( Fig. 3D ).
No change in plasma lipoprotein cholesterol levels or evidence of a CD/cholesterol complex in plasma or urine was observed within 12 h after HP-␤ -CD in Npc1 ؊ / ؊ mice As cholesterol balance changes in Npc1 Ϫ / Ϫ mice occur as early as 3 h after HP-␤ -CD administration during the 6 h timeframe when radiolabeled CD was shown to clear from mature mice ( 18 ), one might hypothesize that HP-␤ -CD binds and transports the excess cholesterol from Npc1 Ϫ / Ϫ cells out of the body. An alternative hypothesis is that

Cyclodextrin mediates rapid changes in Npc1
Ϫ / Ϫ mice 2335 HP-␤ -CD is excreted intact into urine ( 3 ), thus urine was collected for 48 h after HP-␤ -CD administration to determine whether cholesterol was present as a complex with excreted CD ( Fig. 4D ). A signifi cantly higher quantity of cholesterol was excreted by Npc1 Ϫ / Ϫ mice (11 g/day) compared with Npc1 +/+ mice (6 g/day). This difference was not accounted for by differences in body weight, as Npc1 Ϫ / Ϫ mice excreted urinary cholesterol at a rate of 0.72 g/day/g bw while this rate in wild-type littermates was 0.33 g/day/g bw. This observation may be a refl ection of the increased levels of cellular unesterifi ed cholesterol in Npc1 Ϫ / Ϫ mice. Trace amounts of cell lysate from epithelial cells that were shed from the urinary tract wall may have contributed to the increased amount of sterol observed in urine. Importantly however, the injection of HP-␤ -CD did not result in a change in urinary cholesterol excretion in any mice tested. It should be noted that the amount of cholesterol excreted in urine of mice is insignifi cant compared with the amount of sterol secreted from the body by the feces, which was 126 g/day/g bw in Npc1 +/+ mice and 199 g/day/g bw in Npc1 Ϫ / Ϫ mice ( 18 ).

Minimal changes were observed in the expression of cytokines within 12 h of HP-␤ -CD injection
Molecular markers of the proinfl ammatory state have been shown to be elevated in virtually all tissues of Npc1 Ϫ / Ϫ HP-␤ -CD facilitates egress of cholesterol from the LE/L of Npc1 Ϫ / Ϫ cells to allow sterol to be processed by normal cellular mechanisms. To test these hypotheses, the levels of cholesterol in the plasma and urine were measured following HP-␤ -CD injection of Npc1 Ϫ / Ϫ mice at various times consistent with the renal clearance rates of intact HP-␤ -CD ( 3 ). To measure total plasma cholesterol, FPLC gel fi ltration for fractionation of lipoproteins was performed on pooled plasma samples ( Fig. 4 ). First, lipoprotein profi les of Npc1 +/+ and Npc1 Ϫ / Ϫ mice were compared, and it was observed that plasma HDL cholesterol content (represented by fractions 23-35 in Fig. 4A ) is signifi cantly greater in Npc1 Ϫ / Ϫ mice, consistent with a previous report ( 32 ). During the fi rst 3 h after injection when HP-␤ -CD is still present in the plasma, there were no apparent changes in the total plasma lipoprotein cholesterol levels of Npc1 Ϫ / Ϫ mice ( Fig. 4B ). Likewise, at time points (6 and 12 h) when changes in sterol balance are underway and HP-␤ -CD is no longer detectable in the whole animal, there was no evidence of altered total cholesterol content of any lipoprotein fractions in plasma from Npc1 Ϫ / Ϫ mice ( Fig. 4C ). Notably, there was no shift in sterol within any other fractions or in the eluate collected 20 min before and after fractionation (data not shown); therefore, no increase in the level of total cholesterol was measurable in the bloodstream at any time to suggest the presence of a CD/ cholesterol complex. As nearly all cytokine mRNA levels were reduced by 24 h after HP-␤ -CD administration to Npc1 Ϫ / Ϫ mice ( Fig. 5A ), the time course of this HP-␤ -CD effect was evaluated in several tissues, for which a representative set of mRNA levels is provided ( Fig. 5C-F ). There was a trend toward decreased mRNA levels of Tnf ␣ in liver following HP-␤ -CD administration, which reached signifi cance by 12 h ( Fig. 5C ). There was no change in Mip1 ␣ mRNA levels in liver or spleen by 12 h ( Fig. 5D, E ); however, a signifi cant decrease in Mip1 ␣ mRNA in ileum was observed as early as 6 h ( Fig. 5F ). In summary, changes in cytokine expression in tissues of Npc1 Ϫ / Ϫ mice were absent or modest during the fi rst 12 h after HP-␤ -CD despite dramatic reductions in mRNA and protein levels in liver by 24 h, suggesting that the anti-infl ammatory effects of HP-␤ -CD are secondary to changes in sterol homeostasis.

Signifi cant changes in mRNA levels were observed in Npc1 ؊ / ؊ cells treated with HP-␤ -CD and SBE-␤ -CD in culture
In the adult mouse, a single subcutaneous injection of HP-␤ -CD at 4,000 mpk is estimated to provide a maximal mice and to increase with disease progression ( 28,33,34 ). These same markers are reduced when Npc1 Ϫ / Ϫ mice are provided with a therapy, such as HP-␤ -CD, that relieves lysosomal lipid accumulation and extends lifespan ( 5,12,34,35 ). In liver, the mRNA levels of Cd68 (macrosialin, a cell surface marker of monocytes and macrophages) and cytokines Mip1 ␣ , Il-12p40 , Ifn-␥ , Cxcl1 , and Tnf ␣ were dramatically higher in Npc1 Ϫ / Ϫ mice compared with Npc1 +/+ mice ( Fig. 5A ), and similar elevations were observed for selected cytokines in other organs (spleen, Fig. 5E ; and ileum, Fig. 5F ). Twenty-four hours after a single injection of HP-␤ -CD, the mRNA levels of elevated infl ammation markers in Npc1 Ϫ / Ϫ liver were signifi cantly reduced ( Fig. 5A ), and comparable changes were observed for the hepatic protein levels of IFN-␥ , CXCL1, and TNF ␣ (data not shown). These fi ndings were extended to determine whether plasma levels of these cytokines were similarly affected in Npc1 Ϫ / Ϫ mice and whether they could potentially serve as biomarkers to follow NPC disease progression and/or effi cacy of novel therapies. While the mRNA levels of Il-12p40 , Ifn-␥ , Cxcl1 , and Tnf ␣ were much higher in the livers of Npc1 Ϫ / Ϫ mice and reduced with HP-␤ -CD treatment ( Fig. 5A ), these changes were not mirrored by the circulating protein levels of these cytokines in mouse plasma ( Fig. 5B ). A signifi cant increase was only observed for plasma IL-6 Ϫ / Ϫ mice ( Fig. 6 ). As observed in whole animals, administration of HP-␤ -CD had virtually no effect in wild-type cultured cells.   ( Figs. 2, 3, and 6 ). Both forms of ␤ -CD led to nearly equivalent reductions in Hmg CoA Syn and Ldlr mRNA levels in macrophages ( Fig.  7A, C ). In addition, SBE-␤ -CD and HP-␤ -CD exhibited compa rable effi cacy in inducing the LXR target gene Abcg1 and decreasing Ldlr mRNA in neurons ( Fig. 7B, D ).
( Fig. 6D ), were maximally reduced with the minimal dose of 0.3 mM HP-␤ -CD. Although the reduction in Hmg CoA Syn mRNA levels did not reach signifi cance in macrophage, the decrease in Ldlr was statistically signifi cant between the vehicle and HP-␤ -CD-treated Npc1 Ϫ / Ϫ neurons. The ability of various cyclodextrins to interact, adsorb, and ultimately extract unesterifi ed cholesterol from lipid membranes into an aqueous environment has been extensively studied ( 2,16,(36)(37)(38)(39)(40). Single ␤ -CDs can accommodate an unesterifi ed cholesterol in their hydrophobic pocket, but about one-third of the sterol molecule is expected to protrude from the CD ( 36,40 ). In this 1:1 confi guration, all ␤ -CDs appear capable of facilitating the translocation of free cholesterol between apposing membranes (or between donor and acceptor lipid vesicles), thus behaving like a "shuttle." To fully extract unesterifi ed cholesterol from lipid membranes, ␤ -CDs must form

Npc1
Ϫ / Ϫ mice by greater than 35% and subsequent studies ( 11,12,18 ) showing that repeated injections increased longevity by nearly 2-fold pointed to this agent as a novel therapy for NPC patients. CDs are membrane-impermeant oligosaccharides that have been extensively used by cell biologists to manipulate cholesterol content and traffi cking in cultured cells ( 38 ). In elegant studies by Rothblat and colleagues ( 36,37 ), addition of HP-␤ -CD at concentrations exceeding 1 mM to the media of cultured cells acted as a cholesterol "sink," resulting in the rapid effl ux of cholesterol from these cells. The kinetics of this sterol extraction strongly suggested that HP-␤ -CD was desorbing cholesterol from the plasma membrane of cells into the aqueous media by the formation of a 2:1 complex ( 2 ). This mechanism has now been further studied at atomic resolution using molecular dynamics simulations ( 40 ). Notably, ␤ -CD variants unable to form this 2:1 complex, such as tetradecasulfated-␤cyclodextrin, were unable to extract sterol under these conditions. When provided at concentrations <1 mM in the presence of added phospholipid vesicles, virtually all ␤ -CDs could serve as a "shuttle" to transfer sterol down a concentration gradient from membrane to vesicle acceptor. Thus, the behaviors of CDs as "sinks" and "shuttles" to affect plasma membrane cholesterol have been quite thoroughly characterized.
More recent studies utilizing a methyl-␤ -cyclodextrin modifi ed by the addition of a fl uorescein tag demonstrated that this CD is, in fact, taken up from media into cultured cells by a clathrin-dependent pathway ( 49 ). This fl uorescent CD appeared in endosomal structures of HeLa and BHK cells within minutes in a dose-and time-dependent manner. It should be noted that the cells used for these studies have functional NPC proteins, Neither HP-␤ -CD nor SBE-␤ -CD treatment had any effect of Npc1 +/+ cells. Thus, these cell-based studies demonstrate a cell-autonomous activity of ␤ -CDs and further support a mechanism of action that does not rely on the ability of cyclod ex trins to solubilize cholesterol into an aqueous environment.

DISCUSSION
Niemann-Pick type C disease is a rather unique lysosomal storage disorder because the protein products of mutated genes, NPC1 or NPC2 , do not harbor enzymatic activity. Rather, NPC1, a large integral protein of the lysosomal membrane, and NPC2, a small soluble protein localized to the lumen of the LE/L, work in tandem to translocate LDL-derived cholesterol from the lyosomal compartment into the cytosol of cells ( 42,43 ). Thus, in the absence of NPC1 and/or NPC2 function, unesterifi ed cholesterol, as well as other lipids, accumulates in the LE/L of cells, where it is inaccessible to the sterol-sensing machinery of the cytosol/nucleus. Therefore, while there is an excess of cholesterol trapped in the LE/L in NPC-diseased cells, these cells are in a state of cholesterol deprivation and therefore upregulate de novo cholesterol synthesis.
While enzyme replacement therapies are being attempted for many other lysosomal storage disorders (reviewed in Refs. 44,45 ), this approach is not amenable to NPC disease caused by the loss of NPC1 function ( ‫ف‬ 95% of cases). A limited number of small-molecule therapies have been reported to extend the lifespan of Npc1 Ϫ / Ϫ mice ( 34,35,(46)(47)(48), but none have been shown to extend lifespan more than 10-15% without undesirable side effects. Thus, the initial report ( 5 ) demonstrating that a single injection of HP-␤ -CD extended the lifespan of young Ϫ / Ϫ cells is represented by a double dagger ( ‡ ), and differences in RNA levels resulting from HP-␤ -CD treatment within cells of a given genotype are represented by asterisks (* P < 0.05 to *** P < 0.0005).
were not observed in our studies, were likely the result of longer incubation times and the absence of serum in their culture conditions.
In our subsequent comparison of HP-␤ -CD (which can act as "shuttle" or "sink") against SBE-␤ -CD (which can function only as "shuttle"), similar changes in sterolresponsive gene expression were obtained for these two CDs, but only in Npc1 Ϫ / Ϫ cells ( Fig. 7 ). Thus, it can be concluded that under these culture conditions and CD concentrations, ␤ -CD is not acting as an extracellular "sink" to extract/ solubilize plasma membrane cholesterol, a mechanism that should affect cholesterol dynamics of both control and mutant cells. In addition, the similar behaviors of HP-␤ -CD and SBE-␤ -CD to elicit changes in intracellular sterol revealed by altered gene expression suggest that only the "shuttle" capacity of CD is required. This scenario is consistent with proposals that CD is promoting the translocation of unesterfi ed cholesterol from internal membranes of the LE/L across the limiting lysosomal membrane ( 16,17 ) for egress into the cytosolic compartment of Npc1 Ϫ / Ϫ cells.
A subcutaneous injection of HP-␤ -CD at 4,000 mpk to mice is estimated to result in maximal plasma levels of ‫ف‬ 1.3 mM ( 18 ) with a systemic half-life of ‫ف‬ 1.6 h ( 13 ). Within 4 h, greater than 90% of a CD dose is eliminated intact by urinary excretion from rats ( 51 ). On the basis of these observations, we elected to evaluate the temporal changes in sterol homeostasis and the fates of CD and sterol in Npc1 Ϫ / Ϫ mice injected with HP-␤ -CD (summarized in supplementary Fig. II). The fi rst signifi cant changes observed were a pronounced decrease in sterol synthesis rates in liver and other organs ( Fig. 1 ) and an increase in hepatic cholesteryl ester ( Fig. 2B ). By 3 h after injection, sterol synthesis rates were reduced by half and reached a so confi rmation of this behavior in NPC-defi cient cells is warranted. However, the directed delivery of CD as a conjugate of dextran to the lysosomal compartment of human NPC-mutant fi broblasts results in the loss of cholesterol, as revealed by diminished fi lipin staining of this cellular compartment ( 15 ).
Accumulating evidence suggests that ␤ -CD molecules can enter cells by endocytosis, and in the case of NPCdefi cient cells, promote the egress of lysosomal cholesterol. But do CDs serve as a "sink" (which binds cholesterol at a 2:1 ratio to fully solubilize it) or a "shuttle" (which binds cholesterol at a 1:1 ratio) in their capacity to move sterol from this intracellular compartment, and what are the fates of CD and liberated cholesterol? The studies presented in this work addressed these questions using a complement of in vitro and in vivo models.
Using cell culture studies of neurons and macrophages isolated from wild-type and Npc1 Ϫ / Ϫ mice, a 4 h exposure to increasing concentrations of HP-␤ -CD (ranging 0-1.5 mM) had virtually no effect on wild-type cells but caused altered gene expression consistent with sterol release from the LE/L into the cytosol of cells lacking NPC1 function ( Fig. 6 ). A CD-facilitated egress of lysosomal sterol would be predicted to reduce processing of SREBP2 and decrease expression of its target genes, including Hmg CoA Syn and Ldlr . In addition, this fl ux of sterol from lysosome to cytosol/nucleus would increase the expression of LXR target genes, such as Idol and Abcg1 . These fi ndings are consistent with the recent report of Peake and Vance ( 50 ) demonstrating the reduction of lysosomal sterol and increase of cellular cholesteryl ester in cultured Npc1 Ϫ / Ϫ neurons, astrocytes, and microglia treated with HP-␤ -CD. The cytotoxic effects they observed for high-dose HP-␤ -CD, which plateau at 6 h that was maintained for the duration of these 12 h studies (at a level ‫ف‬ 20% of 0 h Npc1 Ϫ / Ϫ and below that of wild-type mice). At 6 and 12 h after injection, changes in the expression of genes associated with altered cellular sterol balance were evident ( Figs. 2 and 3 ). This sequence of changes is consistent with CD-mediated lysosomal cholesterol release, fi rst to affect posttranslation mechanisms reducing sterol synthesis rates (via regulation of HMG-CoA Red stability) and increasing formation of cholesteryl ester via ACAT activity, and second to modulate transcriptional processes regulated by SREBP2 and LXR. During this timeframe, there was no evidence of increased cellular effl ux of sterol either to alter the cholesterol content of circulating lipoproteins ( Fig. 4B, C ) or as a CD-cholesterol complex in plasma or urine ( Fig. 4B-D ). Our methods were able to accurately detect signifi cant but subtle increases (in the microgram range) in the level of cholesterol within urine and plasma in Npc1 Ϫ / Ϫ mice compared with Npc1 +/+ mice. Therefore, if a change had occurred after HP-␤ -CD to alter the level of cholesterol in plasma or urine, it would have been easily ascertained through our analyses. Thus, the breadth of these studies suggest that HP-␤ -CD acts within Npc1 Ϫ / Ϫ cells as a shuttle to facilitate the egress of lysosomal unesterfi ed cholesterol to intracellular membrane or protein acceptors, without serving as a sink to facilitate cholesterol transport and elimination via plasma and/or urine.
Finally, all agents that have been identifi ed to extend the lifespan of Npc1 Ϫ / Ϫ mice have decreased expression of proinfl ammatory cytokines in virtually every tissue measured ( 5,12,34,35,47 ). Therefore, we included this readout in our time course studies of HP-␤ -CD. Twenty-four hours after injection of HP-␤ -CD, mRNA levels ( Fig. 5 ) for Tnf ␣ , Mip-1 ␣ , and other markers of increased macrophage number and activation were signifi cantly reduced in tissues of Npc1 Ϫ / Ϫ mice. However, these changes were for the most part not observed within 12 h of HP-␤ -CD administration, suggesting that they occur after changes in sterol homeostasis. And importantly, although tissue RNAs and proteins were altered, circulating cytokine levels were neither elevated in the Npc1 Ϫ / Ϫ mice nor affected to a signifi cant degree by CD treatment ( Fig. 5 and supplementary Fig. I).
In conclusion, comprehensive studies have been done to monitor the acute effects of HP-␤ -CD in Npc1 Ϫ / Ϫ mice. Overall, these data suggest that HP-␤ -CD is not acting as a "sink" to bind and carry cholesterol out of cells into plasma and/or out of the body but that it is liberating the trapped unesterifi ed cholesterol from the lysosomes of Npc1 Ϫ / Ϫ cells very rapidly by facilitating the egress of this cholesterol to intracellular sites for normal sterol processing.