Pro-Inflammatory Implications of 2-Hydroxypropyl-β-cyclodextrin Treatment

Lifestyle- and genetically induced disorders related to disturbances in cholesterol metabolism have shown the detrimental impact of excessive cholesterol levels on a plethora of pathological processes such as inflammation. In this context, two-hydroxypropyl-β-cyclodextrin (CD) is increasingly considered as a novel pharmacological compound to decrease cellular cholesterol levels due to its ability to increase cholesterol solubility. However, recent findings have reported contra-indicating events after the use of CD questioning the clinical applicability of this compound. Given its potential as a therapeutic compound in metabolic inflammatory diseases, in this study, we evaluated the inflammatory effects of CD administration in the context of cholesterol-induced metabolic inflammation in vivo and in vitro. The inflammatory and cholesterol-depleting effects of CD were first investigated in low-density lipoprotein receptor knockout (Ldlr-/) mice that were transplanted with Npc1nih or Npc1wt bone marrow and were fed either regular chow or a high-fat, high-cholesterol (HFC) diet for 12 weeks, thereby creating an extreme model of lysosomal cholesterol-induced metabolic inflammation. In the final three weeks, these mice received daily injections of either control (saline) or CD subcutaneously. Subsequently, the inflammatory properties of CD were investigated in vitro in two macrophage cell lines and in murine bone marrow-derived macrophages (BMDMs). While CD administration improved cholesterol mobilization outside lysosomes in BMDMs, an overall pro-inflammatory profile was observed after CD treatment, evidenced by increased hepatic inflammation in vivo and a strong increase in cytokine release and inflammatory gene expression in vitro in murine BMDMs and macrophages cell lines. Nevertheless, this CD-induced pro-inflammatory profile was time-dependent, as short term exposure to CD did not result in a pro-inflammatory response in BMDM. While CD exerts desired cholesterol-depleting effects, its inflammatory effect is dependent on the exposure time. As such, using CD in the clinic, especially in a metabolic inflammatory context, should be closely monitored as it may lead to undesired, pro-inflammatory side effects.

Lifestyle-and genetically induced disorders related to disturbances in cholesterol metabolism have shown the detrimental impact of excessive cholesterol levels on a plethora of pathological processes such as inflammation. In this context, twohydroxypropyl-b-cyclodextrin (CD) is increasingly considered as a novel pharmacological compound to decrease cellular cholesterol levels due to its ability to increase cholesterol solubility. However, recent findings have reported contra-indicating events after the use of CD questioning the clinical applicability of this compound. Given its potential as a therapeutic compound in metabolic inflammatory diseases, in this study, we evaluated the inflammatory effects of CD administration in the context of cholesterolinduced metabolic inflammation in vivo and in vitro. The inflammatory and cholesteroldepleting effects of CD were first investigated in low-density lipoprotein receptor knockout (Ldlr -/ ) mice that were transplanted with Npc1 nih or Npc1 wt bone marrow and were fed either regular chow or a high-fat, high-cholesterol (HFC) diet for 12 weeks, thereby creating an extreme model of lysosomal cholesterol-induced metabolic inflammation. In the final three weeks, these mice received daily injections of either control (saline) or CD subcutaneously. Subsequently, the inflammatory properties of CD were investigated in vitro in two macrophage cell lines and in murine bone marrow-derived macrophages (BMDMs). While CD administration improved cholesterol mobilization outside lysosomes in BMDMs, an overall pro-inflammatory profile was observed after CD treatment, evidenced by increased hepatic inflammation in vivo and a strong increase in cytokine release and inflammatory gene expression in vitro in murine BMDMs and macrophages

INTRODUCTION
As one of the most important steroid alcohols of the human body, cholesterol carries a number of essential functions ranging from acting as a key structural component of cell membranes (1) to serving as precursor for bile acids, steroid hormones and vitamin D metabolites (2). Due to this key cellular function of cholesterol, a highly coordinated extra-and intracellular transport system maintains cholesterol homeostasis at a whole body as well as at a cellular level (3,4). Evidently, deficiencies in this transport system leading to a disproportionate cellular supply of cholesterol result in severe pathogenic profiles. At one hand, these deficiencies can be induced via mutations in genes encoding for receptors and/or enzymes involved with cholesterol homeostasis. Examples of such disorders include Wolman disease (5) and the Niemann-Pick diseases (6) (both lysosomal storage disorders) as well as familial hypercholesterolemia (7). Though these diseases are rather rare, their pathogenesis is often severe and sometimes results in premature death. Besides genetic predisposition, another and more prevalent manner that influences cholesterol homeostasis is via excessive overnutrition, leading to metabolic syndrome and its associated diseases (8). In this context, excessive cholesterol accumulates at locations such as the vessel wall (atherosclerosis) and the liver (non-alcoholic fatty liver disease (NAFLD)) subsequently triggering resident macrophages to evolve into foam cells characterized by lysosomal cholesterol accumulation that induces an undesired inflammatory response, also referred to as metabolic inflammation (4,9). This inflammatory response is an antecedent for more severe consequences both in the context of atherosclerosis (heart attack, stroke, peripheral vascular disease (10)) as well as NAFLD (advanced liver diseases such as cirrhosis and hepatocellular carcinoma (9)). Taken the prevalence and the severity of these disorders into consideration, efficient therapeutic options aimed at reducing cellular cholesterol levels are needed.
A straightforward and successfully applied approach to reduce cholesterol levels is via pharmacological approaches, which include the use of statins (11), fibrates (12) and more recently antibodies targeting proprotein convertase subtilisinkexin type 9 (PCSK9) (13). However, substantial reduction of plasma cholesterol levels cannot be guaranteed for all patients (11), requiring alternative ways to improve cholesterol homeostasis. Relevantly, while the aforementioned interventions focus on reducing plasma cholesterol levels, a relative new approach aims at mobilizing cholesterol at the site of accumulation. This has led to the proposition to use 2-hydroxypropyl-b-cyclodextrin (CD), a substance initially used to solubilize lipophilic pharmaceutical agents, as a compound to treat cholesterol accumulation. Indeed, CD mobilized cholesterol from foam cells (14), promotes atherosclerosis regression (15), improved hepatic cholesterol metabolism (16), reduced lysosomal size of neural stem cells derived from Wolman disease patients (17) and is under evaluation in advanced human clinical trials for Niemann-Pick disease, type C1 (NPC1) (18). In contrast to these beneficial effects, CD causes massive damage to all cells of the developing (19) and some cells of the adult auditory system (20). Moreover, while dietary supplementation of b-cyclodextrin in hypercholesterolemic rats improved lipid metabolism, it concomitantly produced hepatotoxic effects characterized by increased plasma aminotransferase levels (21). Together, though the use of CD has clearly shown its benefits, the previous observations question whether this compound can be directly used in the clinic. Given its potential as therapeutic compound in metabolic inflammatory diseases, in this study, we therefore investigated the inflammatory aspects of CD administration in the context of cholesterol-induced metabolic inflammation in vivo and in vitro.
For this purpose, we investigated the inflammatory and cholesterol-depleting properties of CD in vivo and in a series of in vitro experiments for metabolic inflammation. First, we analyzed the effect of CD injection in low-density lipoprotein receptor knockout (Ldlr -/-) mice transplanted with Npc1 nih or Npc1 wt bone marrow on a high-fat, high-cholesterol (HFC) diet. Due to the absence of LDL receptors, and following an HFC diet, Ldlr -/mice are characterized by high plasma LDL levels, mimicking diet-induced dyslipidemia and serving as an excellent animal model for metabolic diseases. Furthermore, Npc1 nih bone marrow-transplanted mice feature a dysfunctional NPC1 protein in their immune cells, due to a deleterious frameshift mutation in the Npc1 allele, leading to lysosomal cholesterol buildup and a pro-inflammatory state. As such, chimeric Npc1 nih transplanted (Npc1 nih -tp) Ldlr -/mice given an HFC diet constitute an extreme state of lysosomal cholesterol-induced metabolic inflammation (22). Next, as CD is able to mobilize lysosomal cholesterol, we assessed the ability of CD to influence lysosomal size of metabolically challenged bone marrow-derived macrophages (BMDM) by employing confocal microscopy. Finally, in a series of in vitro experiments we investigated the inflammatory properties of CD in more detail.

Mice, Bone Marrow Transplant, Diet and Injections
Throughout the study, mice were housed under standard conditions and had unlimited access to food and water, unless explicitly mentioned otherwise. For one week prior to and up to four weeks after bone marrow transplantation, Ldlr -/mice were housed in filter-top cages and received antibiotics diluted in drinking water to prevent infections following immunosuppression (Neomycin, 100 mg/l, Gibco, Breda, the Netherlands; 6*10 4 U/L polymycin B sulfate). Six weeks old bone marrow donors Npc1 nih and Npc1 wt mice were derived from heterozygous founders of a C57BL/6 genetic background. Genotype of Npc1 nih and Npc1 wt mice was determined as previously described (23). On the day of the bone marrow transplant, Npc1 nih and Npc1 wt littermates were sacrificed via CO 2 inhalation and their bone marrows were isolated. One day before and on the day of the bone marrow transplant, Ldlr -/mice were subjected to six Gray of g-radiation, thus having received 12 Gray of g-radiation before receiving 1*10 7 bone marrow cells collected from Npc1 wt or Npc1 nih mice via intravenous injection. After ten weeks of recovery, transplanted mice were placed on a high-fat, high-cholesterol (HFC) diet (24) for twelve weeks. In the final three weeks (based on (16,25)), mice received daily, subcutaneous injections of control (saline) or 2-hydroxypropylb-cyclodextrin (1800 mg/kg body weight; CD, Sigma-Aldrich), creating four experimental groups: (1) Npc1 wt -tp NaCl (n = 6), (2) Npc1 wt -tp CD (n = 9), (3) Npc1 nih -tp NaCl (n = 6) and (4) Npc1 nih -tp CD (n = 9). A schematic overview of the experimental set-up is provided in Supplementary Figure S1. All experiments were performed according to Dutch laws and approved by the Animal Experiment Committee of Maastricht University.
Upon sacrifice, all tissues were isolated and snap-frozen in liquid nitrogen and stored at -80°C or fixed in 4% formaldehyde/ PBS. The collection of blood and tissue specimens, biochemical determination of lipids in plasma, RNA isolation, cDNA synthesis and qPCR were determined as described previously (26,27). Primer sequences for genes are listed in Supplementary  Table 1. Hepatic sterol content was determined by gas-liquid chromatography-mass spectroscopy, as described elsewhere (28).
Pictures were taken with a Nikon digital camera DMX1200 and ACT-1 v2.63 software (Nikon Instruments Europe, Amstelveen, The Netherlands). Infiltrated macrophages and neutrophil cells (Mac-1 + ) and neutrophils (NIMP) were counted by two blinded researchers in six microscopical views (original magnification, 200x) and were indicated as number of cells per square millimeter (cells/mm 2 ). Hepatic macrophages (CD68) were counted in six microscopical views (original magnification, 200x) and indicated as the percentage of CD68 positive area (Adobe Photoshop CS2 v.9.0.).

Lysosomal Quantification
Step 1: Configuring and adapting the lysosomal pictures For optimal quantification, one cell remained visible per purview. For this purpose, the original photo was cropped, and then configured into a TIFF file for further processing.
Step 2: Splitting the channels for nucleus and lysosomal quantification In this study we made use of two fluorescent stainings either for the nucleus (DAPI) or lysosome (LAMP-1). To determine the nucleus to lysosomal distance, the color channels were bifurcated for better differentiation, creating a nucleus channel and lysosomal channel.
Step 3: Coordinates of the nucleus.
To determine the lysosome to nucleus distance we first determined the location of the nucleus in the 3D picture. Therefore, we analyzed the nucleus channel with the 3D ObjectCounter.
Step 4: Creating a mask of lysosome channel To obtain the coordinates of the lysosomes, a mask was created by using the ObjectCounter. The mask was created by calculating the geometrical centers of the lysosomes (30). Threshold was determined before the picture was analyzed.
Step 5: Watershed To discriminate between separated or combined lysosomes, the watershed method was used. The watershed method uses the mask to determine the centers of the masses (Parra et al.).
Step 6: Coordinates of the lysosomes After the watershed, the Regions of interest (ROIs) were added to the 3D ROI Manager, where the volume and coordinates were calculated. To quantify the pictures in an efficient manner, a macro was created (Supplementary Methods).
Step 7: Calculating the distance After the coordinates and volumes were obtained, the distance for each lysosome to the nucleus was calculated. For this we used the Pythagoras equation.
Next, we created a plugin-Macro for ImageJ to identify the amount and volume of lysosomes. Finally, the number of lysosomes was quantified and categorized them by size (< 0.1 μm 3 ; 0.1-1 μm 3 ; > 1 μm 3 ) and expressed the number of lysosomes per size relative to the total amount of lysosomes present inside the BMDM. Representative videos were created using Z-stack images played in series and recorded as.AVI files which were trimmed using quick time movie player. DAPI (blue) represents nuclei and LAMP-1 (green) represents lysosomes.

Statistical Analysis
Data were statistically analyzed by performing the unpaired t-test or the two-way ANOVA and Tukey's post hoc test using GraphPad Prism software (version 6 for Windows, GraphPad Software Inc, San Diego, CA, U.S; www.graphpad.com). Data were expressed as the group mean and standard error of the mean.

RESULTS
Administration of 2-Hydroxypropyl-bcyclodextrin Mobilizes Hepatic Cholesterol in Npc1 nih -tp Ldlr -/-Mice Npc1 nih bone marrow-transplanted mice feature a dysfunctional NPC1 protein in their immune cells, due to a deleterious frameshift mutation in the Npc1 allele, leading to lysosomal cholesterol buildup and a pro-inflammatory state. As such, chimeric Npc1 nih transplanted (Npc1 nih -tp) Ldlr -/mice given an HFC diet constitute an extreme state of lysosomal cholesterolinduced metabolic inflammation (22).
To confirm successful injection of 2-hydroxypropyl-bcyclodextrin (CD) into the mice in the final three weeks of the experiments, plasma and hepatic lipid metabolism were profiled. CD-treated Npc1 wt -tp Ldlr -/mice showed reduced plasma triglycerides ( Table 1), though this reduction was already apparent before the start of CD treatment ( Table 1). While no effects were observed on plasma ( Table 1) and hepatic ( Figure 1) cholesterol levels, the ratios of the cholesterol oxidation products 7a-hydroxycholesterol to cholesterol and 27-hydroxycholesterol to cholesterol were both significantly increased in the livers of CD-treated Npc1 nih -and Npc1 wt -tp Ldlr -/mice, suggesting increased mobilization of cholesterol ( Figures 1B, C). Hepatic gene expression analysis of cytochrome P450 7A1 (Cyp7a1), the enzyme responsible for the conversion of cholesterol into 7ahydroxycholesterol, was concomitantly increased ( Figure 1D), while sterol 27-hydroxylase (Cyp27a1) hepatic gene expression levels remained unaffected upon CD-treatment ( Figure 1E). Concerning other non-cholesterol sterols, CD-treatment reduced hepatic cholestanol and desmosterol levels in CDtreated Npc1 nih -tp Ldlr -/mice ( Table 2). No effects were observed on liver and spleen weight ( Supplementary  Figures 2A, B). Body weight over time seemed to be lower in Npc1 nih -tp Ldlr -/mice compared to Npc1 wt -tp Ldlr -/mice, but this difference was not significant (Supplementary Figure 2C). Overall, these findings indicate that CD-treatment mobilizes hepatic cholesterol, confirming successful administration of CD.
Administration of 2-Hydroxypropyl-bcyclodextrin Increases Hepatic Inflammation and Fibrotic Markers in Npc1 nih -tp Ldlr -/-Mice After we confirmed successful administration of CD, we further evaluated the impact of three-week CD treatment on hepatic inflammation by staining hepatic cryosections for the inflammatory markers Mac-1 (infiltrated macrophages and neutrophils; directed against Cd11b), NIMP (neutrophils) and CD68 (macrophages) (Figures 2A-C). While CD treatment only induced a minor non-significant increase for these inflammatory markers in Npc1 wt -tp Ldlr -/mice, this increase was more pronounced in Npc1 nih -tp Ldlr -/mice (Figures 2A, D).  This pro-inflammatory effect of CD was confirmed by histological scoring of HE-staining of livers of the Npc1 nih -tp Ldlr -/mice (Supplementary Figure S3). To confirm these histological data, hepatic gene expression for the inflammatory markers tumor necrosis factor alpha (Tnfa), chemokine (C-C motif) ligand 2 (Ccl2), chemokine (C-C motif) ligand 5 (Ccl5) and serum amyloid A1 (Saa1) was analyzed. In line with the histological data, CD treatment increased the expression of these hepatic inflammatory markers in the Npc1 wt -tp or in the Npc1 nihtp group (Figures 2E-H). Moreover, the hepatic fibrotic markers matrix metallopeptidase 9 (Mmp9) and plasminogen activator inhibitor-1 (Pai-1) were concomitantly increased in CD-treated Npc1 nih -tp Ldlr -/mice, while no significant changes were observed in transforming growth factor beta (Tgfb) and tissue inhibitor of metalloproteinase (Timp1) (Supplementary Figure S4).
Together, these findings indicate that three-week treatment with CD increased hepatic inflammation and mild elevations in fibrosis in our experimental model.

Treatment With 2-Hydroxypropyl-bcyclodextrin Normalizes Lysosomal Size of Metabolically Challenged Bone Marrow-Derived Macrophages
As CD treatment mobilized cholesterol, but increased inflammation in our in vivo model for lysosomal cholesterolinduced inflammation, we opted to investigate the effect of CD on lysosomal size in metabolically challenged BMDM via confocal microscopy. For this purpose, we first compared lysosomal size between wildtype and Npc1 nih BMDM. While Npc1 nih BMDM showed a reduced number of small lysosomes Indicates p < 0.05 and p < 0.01 respectively compared to NaCl-injected Npc1 nih -tp mice. All values are given as the ratio sterol to cholesterol.  Table 2). As internalization of oxidized cholesterol-rich low-density lipoprotein (oxLDL) in BMDM also leads to lysosomal accumulation, we next compared lysosomal size of oxLDL (25μg/ml; 72hr)-incubated wildtype BMDM to controlincubated wildtype BMDM. Incubating wildtype BMDM with oxLDL similarly reduced the number of small lysosomes and increased the number of large lysosomes, confirming the effect of oxLDL on lysosomal size (Figure 3  Next, to assess the impact of CD treatment, we compared lysosomal size of oxLDL-incubated wildtype BMDM that were treated with or without CD (1.95 mM; 4hr). As expected, CD treatment increased the number of small lysosomes and reduced the level of large lysosomes, thereby reversing the oxLDLinduced effects on lysosomal size (Figure 3

Incubation With 2-Hydroxypropyl-bcyclodextrin Exerts Pro-Inflammatory Effects in Metabolically Challenged Bone Marrow Derived Macrophages
To increase our insight into the inflammatory effects of CD, we performed a series of in vitro experiments in BMDM, framed within the context of metabolic inflammation. First, we incubated wildtype and Npc1 nih BMDM with CD for 4hr followed by 4hr lipopolysaccharide (LPS) stimulation. Four-hour CD treatment increased TNFa protein secretion in both wildtype and Npc1 nih BMDM, suggesting an acute pro-inflammatory effect of CD also in wildtype and Npc1 nih BMDM ( Figure 4A; Supplementary Figure  S5A). To provide stronger evidence for this inflammatory effect of CD, Npc1 nih BMDM were challenged with oxLDL, followed by 4hr CD incubation and 4hr LPS stimulation, thereby creating a more severe model of metabolic inflammation. Similar to our previous findings, CD treatment increased TNFa protein secretion in both oxLDL-and control-challenged Npc1 nih BMDM ( Figure 4B; Supplementary Figure S5B). This was confirmed at gene expression level, showing increased levels of proinflammatory markers (Tnfa, iNos/Arg1 ratio) ( Figure 4B Supplementary Figure S5B) and decreased levels of antiinflammatory markers (IL-10; Figure 4B; Supplementary Figure S5B) after CD treatment. These pro-inflammatory findings of CD were also confirmed in a separate BMDM experiment with similar set-up in wildtype BMDM (Supplementary Figure S6A) and without the final LPS stimulus (Supplementary Figure S6B). As another model for metabolic disturbance, we investigated the inflammatory effects of CD on Ldlr -/-BMDM, which is a mutation causing disturbances in cholesterol metabolism. Twenty-four-hour incubation with oxLDL followed by 4hr CD treatment resulted in increased gene expression levels of Tnfa, IL-1b, and IL-18, confirming the pro-inflammatory effects of CD also in Ldlr -/-BMDM ( Figure 4C; Supplementary Figure S4C). Nevertheless, CD treatment increased Lxra and Npc2 expression, confirming the cholesterol-mobilizing properties of CD. To further confirm these findings, a similar experiment was conducted with the addition of a final stimulation with Pam3Cys or LPS for 4hr. Stimulation with Pam3Cys ( Figure 4D; Supplementary Figure S4D) and LPS ( Figure 4E; Supplementary Figure S4E) showed similar results as previously described, showing pro-inflammatory and cholesterol-mobilizing effects of CD treatment in Ldlr -/-BMDM. Together, these findings show that while CD maintains its effect on cholesterol metabolism, pro-inflammatory effects become apparent after 4hr incubation in metabolically challenged BMDM.

Incubation With 2-Hydroxypropyl-bcyclodextrin Exerts Pro-Inflammatory Effects in Macrophage-Derived Cell Lines
To validate the inflammatory effect of CD in non-primary cells, we investigated the impact of CD treatment on the macrophagederived mouse RAW 264.7 and human THP-1 cell lines. Firstly, RAW 264.7 cells were incubated with CD for 4 or 8hr, followed by 4hr LPS stimulation. While 4hr incubation did not show any inflammatory effect, CD treatment for 8hr increased TNFa protein secretion (Figures 5A, B; Supplementary Figures S7A, B). Furthermore, THP-1-derived macrophages were challenged with sphingomyelinase-aggregated LDL (smLDL), native LDL or control for 24hr and incubated with CD for 4hr. Secreted protein levels of TNFa increased in all three experimental groups after CD incubation ( Figures 6A-C). Together, these results indicate that CD also shows pro-inflammatory properties in macrophage-derived cell lines.

The Pro-Inflammatory Effect of 2-Hydroxypropyl-b-cyclodextrin on Bone Marrow-Derived Macrophages Is Timeand Concentration-Dependent
As increased depletion of cholesterol from the plasma membrane was shown to result in pro-inflammatory responses (31), we first investigated whether incubation time influenced the pro- inflammatory effect of CD. For this purpose, wildtype BMDMs were incubated with CD for 5, 10 and 30 min as well as for 1 and 4hr. Protein secretion of TNFa showed a marked timedependent response after CD treatment ( Figure 7A). This time-dependent, pro-inflammatory effect of CD was confirmed at gene expression levels for the inflammatory markers Tnfa, IL-1b and Ccl2, only showing increased expression after 1 and 4hr incubation ( Figures 7B-D). Lxra showed a similar increase after (B) TNFa protein levels and Tnfa, IL10 and iNOS/Arg1 ratio gene expression data of oxLDL (25µg/ml; 24hr)-exposed Npc1 nih BMDM treated with or without CD (4hr) that were terminally stimulated with LPS (100ng/ml; 4hr). Gene expression data were set relative to control-exposed Npc1 nih BMDM treated with saline and stimulated with LPS. (C-E) Gene expression analysis of Tnfa, IL-1b, IL-18, Lxra and Npc2 of oxLDL (25µg/ml; 24hr)-exposed Ldlr -/and Wt BMDM treated with or without CD (1.95 mM; 4hr) and terminally stimulated with control (C), Pam3Cys (4hr) (D) or LPS (4hr) (E). Gene expression data were set relative to oxLDL-exposed Wt BMDM treated with saline. Colored (red or blue) boxes are compared to the box directly at their left via two-way ANOVA followed by Tukey post-hoc analysis, indicating the effect of CD. Data are the result of 2 or 3 independent experiments.
To further confirm our hypothesis, we also performed a doseresponse curve with CD (4hr incubation) in wildtype BMDMs. As indicated in Figure 8, CD incubation showed a clear concentration-dependent effect on TNFa secretion, adding fuel to our argument that the pro-inflammatory effects of CD are related to cholesterol depletion from the plasma membrane.

DISCUSSION
New perspectives to reduce cholesterol levels aim at improving cellular instead of plasma cholesterol levels, raising the argument to use CD as a pharmacological compound to improve cholesterol homeostasis. Here, we confirm the cholesteroldepleting effects of CD in a metabolic inflammatory context, but concomitantly show a detrimental time-and concentrationdependent inflammatory effect of CD treatment. Therefore, while CD is able to decrease cellular cholesterol levels, our findings demonstrate that its use in the clinic should be closely monitored especially in patients with a metabolic inflammatory background. Our observation that CD treatment promotes cellular cholesterol mobilization, but induces a time-and concentration-dependent inflammatory effect implies the importance of the subcellular distribution of cholesterol and the subsequent impact of cholesterol depletion from these specific locations. Unesterified cholesterol serves a key structural function in the plasma membrane as it is critical for the formation of liquid-ordered rafts, which determine membrane fluidity. The membrane fluidity is therefore coupled to the free cholesterol/phospholipid ratio of the plasma membrane as this ratio determines the formation of the aforementioned rafts (32,33). Under hyperlipidemic conditions, the plasma membrane serves as the first pool to deposit free cholesterol from external sources (34). However, accumulation of free cholesterol in the plasma membrane above an optimal free cholesterol/phospholipid ratio negatively impacts plasma membrane fluidity (33). Indeed, Yvan-Charvet et al. showed that increased accumulation of free cholesterol in the plasma membrane (induced by Abca1 or Abcg1 knockout) directly results in pro-inflammatory responses, supporting our current notion of the limited physiological capacity of the plasma membrane to harbor free cholesterol (31). Therefore, exceeding the plasma membrane's capacity to harbor cholesterol directly results in intracellular cholesterol accumulation (mainly in lysosomes), which is considered a severe pathological phenomenon for mediating inflammatory responses (4).
Evidently, under conditions that generate severe accumulation of intracellular cholesterol (exemplified by the NPC1 mutation and LDLR knockout), CD-mediated depletion of cholesterol is highly desirable. In line with this concept, the increased lysosomal size in metabolically challenged macrophages was reverted after incubation with CD treatment. This advantageous cholesterol-depleting effect of CD has been confirmed in multiple previous reports (15,18,31,35). In contrast, in the current study, we consistently show a harmful, pro-inflammatory (and even mild pro-fibrotic) effect of CD under cholesterol-induced inflammatory conditions in vivo and in vitro. This pro-inflammatory effect can be rationalized by the prolonged cellular exposure to CD, which influences plasma membrane cholesterol combined with intracellular cholesterol levels. Indeed, given the key role of cholesterol in the formation of liquid-ordered rafts, excessive depletion of plasma membrane cholesterol (as is induced with CD treatment) is highly undesirable as it disrupts these rafts, affecting membrane fluidity. This rationale is further supported by findings showing that cholesterol extraction of the plasma membrane is a primary location where CD exerts its function (36,37). Therefore, excessive depletion of cholesterol from the plasma membrane by CD might be an explanation for the observed proinflammatory findings in our study as well as the time-and concentration-dependent character of our findings. The timedependent findings are also in line with experiments reported by Pilely et al. and Ding et al., that show that while incubation of different types of cyclodextrin for a short period of time (10 and 30 min) is anti-inflammatory (35), incubation for 24 and 48hr lead to ototoxicity (19). Together, these observations indicate that while CD has potential in advantageously depleting cellular cholesterol, it is essential to monitor the quantity of cellular cholesterol and adjust the therapeutic dose/time accordingly to prevent undesired pro-inflammatory side-effects.
Notably, while neutrophils generally constitute the major immune cell population under inflammatory conditions, in the current study we only focused on the effect of CD on macrophages. This choice is based on previous findings by our group that demonstrate macrophages to be the most important immune cell in the in vivo model here described (as evidenced by changes in hematopoiesis (38) and organspecific inflammation (22)). Nevertheless, as neutrophil aberrations have also been described in NPC1 disease (39), future research should further explore the involvement of neutrophils in CD-induced inflammatory responses in a metabolic inflammatory context. These dichotomic characteristics of CD raise the question whether structural manipulation of cyclodextrins can reduce the harmful pro-inflammatory effects, while maintaining the advantageous cholesterol-depleting effects of CD. Cyclodextrins are composed of cyclic oligosaccharides of 6, 7 or 8 glucose units (referred to as a-, band g-cyclodextrins respectively), providing cyclodextrin a polar and hydrophilic surface combined with a non-polar cavity (40). This dual-property structure of CD grants itself for modification via polymerization, creating so-called polycyclodextrins. Indeed, various modifications have been performed on cyclodextrin, creating more efficient and effective cyclodextrins for a plethora of purposes (40). In line, Kulkarni et al. adapted the structure CD (the type we employed in our study) and showed that their linear degradable, high molecular weight polymer variation improved the pharmacokinetic profile and bioavailability in NPC mice (41). Moreover, using polyrotaxanes enabled specific release of CD inside lysosomes, thereby minimizing the effect on plasma membrane cholesterol (42). Based on these reports, it is anticipated that designing polycyclodextrins is a promising approach that can have a considerable clinical impact due to its ability to reduce the pro-inflammatory properties of CD described in our study.
In conclusion, though we confirm its cholesterol-depleting effects, we here demonstrate a time-and concentrationdependent harmful pro-inflammatory effect of CD under metabolic inflammatory conditions. As such, we suggest that clinical use of CD, in particular in a metabolic inflammatory context, should be closely monitored to prevent undesired side effects.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by Animal Experiment Committee of Maastricht University.

ACKNOWLEDGMENTS
We thank the present and past members of Prof. Sverdlov's group for their technical assistance in the experiments.