Atorvastatin administration results in ciliopathy phenotypes

Recently, cilia have emerged as the central organelle controlling heart development^16,20. Interestingly, atorvastatin as well as mevastatin and rosuvastatin produced zebrafish embryos with a gross morphology reminiscent of cilia defects: embryos exhibited a body curvature, which worsened with increasing concentrations of atorvastatin (Fig. 2a, b, Supplementary Figure 2e). Treatment with statins furthermore disturbed normal otolith seeding, which is controlled by different types of cilia in the otic placode²¹ (Fig. 2c, d, Supplementary Figure 2f). Body curvature and otolith defects of atorvastatin-treated embryos were ameliorated by reconstitution with cholesterol (Supplementary Figure 3). We also observed a thickening of the yolk extension (Fig. 2a), pinheads and smaller eyes, which have been similarly reported for a mutant of HMG-CoA-R²² indicating that the ciliopathy phenotypes are likely due to HMG-CoA-R inhibition and no off-target effects. To further validate this finding and investigate the mechanism underlying the heart defect under statins more closely, we assessed left–right (LR) asymmetry development. LR asymmetry is a prerequisite of oriented and faithful heart development. As a consequence, the majority of patients devoid of proper LR asymmetry establishment develop congenital heart defects as heart looping becomes randomized²³. In addition, other abdominal organs may be erroneously distributed^18,24. Indeed, zebrafish embryos treated with increasing concentrations of atorvastatin displayed heart looping defects with substantial fractions of all embryos analysed having either inversely looped hearts or such that completely fail to undergo looping (Fig. 2e). Analysis of pancreas placement as detected by in situ hybridization for the endocrine pancreas marker insulin revealed further that LR asymmetry was also disturbed on the level of abdominal organs (Fig. 2f). Similarly, early leftward genes such as lefty2, which locates to the left cardiac mesoderm, and pitx2, which is expressed in the left lateral plate mesoderm, were randomly expressed on both sides of the midline (Fig. 2g, h). The same LR asymmetry defects could be observed upon treatment with other statins, too (Supplementary Figure 4). In contrast to atorvastatin, however, there was an increased incidence of embryos developing a cardia bifida, which had similarly been shown for HMG-CoA-R mutants²². Thus, statin treatment produces phenotypes resembling those arising from cilia defects, which includes aberrant cardiac looping.

Open in a separate window

Fig. 2

Atorvastatin provokes ciliopathy phenotypes. a Atorvastatin treatment results in embryos with curved bodies. Scale bar: 200µm. b Body curvature worsens with increasing atorvastatin concentrations; n=16 (0, DMSO), 6 (2µM), 8 (5µM), 11 (10µM), 9 (20µM), 9 (50µM); **p<0.01, ****p<0.0001 (atorvastatin versus DMSO control treated embryos), Kruskal–Wallis test with Dunn’s multiple comparison test. c Live images of otic placodes at 48 hpf showing defects in otolith formation in atorvastatin-treated embryos. Scale bar: 50µm. d Graph displaying the dose-dependent increase in otolith defects upon atorvastatin treatment; n=15 (0, DMSO), 6 (2µM), 8 (5µM), 10 (10µM), 9 (20µM), 9 (50µM); *p<0.05, ****p<0.0001 (atorvastatin versus control treated embryos), Kruskal–Wallis test with Dunn’s multiple comparison test. e Atorvastatin produces random heart looping. Images of cmlc2 in situ hybridization to visualize the whole heart at 48 hpf. Scale bar: 100µm; n=9 (0, DMSO), 6 (2, 5, 10, 20 and 50µM); p=0.0065 (2µM), 0.0073 (5µM), 0.0005 (10µM),<0.0001 (20 and 50µM). f Pancreas placement becomes randomized upon atorvastatin treatment. Dorsal views of embryos processed by in situ hybridization for the endocrine pancreas marker insulin (ins). Dashed line indicates midline of embryo. Scale bar 200µm. Stacked bar graph summarises 9 (DMSO) and 6 (atorv. concentrations) experiments; p=0.0492 (2µM), 0.0493 (5µM), 0.0002 (10µM),<0.0001 (20 and 50µM). g The leftward gene lefty2 is ambiguously expressed after atorvastatin treatment. L expression on left side, R right side, B both sides of midline (dashed), A absent expression. Scale bar: 200µm; **p=0.0072. h pitx2 is randomly expressed in the left and/or right lateral plate mesoderm after atorvastatin treatment. L expression on left side, R right side, B both sides of midline (dashed), A absent expression. Scale bar: 200µm; *p=0.0155, **p=0.0071. b, d–h Number of individual embryos given below or next to graphs. e–h Two-tailed Fisher’s exact tests (atorvastatin versus control treated embryos)

Atorvastatin alters LR organizer morphology

LR asymmetry in zebrafish is determined long before a heart tube develops. It starts with the formation of the Kupffer’s vesicle (KV), the temporal organ of laterality in zebrafish. The KV is a fluid-filled vesicle, which is formed by a monociliated epithelium and is the corresponding structure to the mouse node. Cilia within and at the edge of this KV both propel the vesicular liquid and sense the flow. This results in the asymmetric expression of genes, which at later stages distinguish the left from the right body half and determine the asymmetric placement of inner organs^18,25. Importantly, the KV ceases to exist after the 18 somite stage when leftward gene expression has robustly been induced²⁶. Hence, timed administration of pharmacologically active compounds can be used to block cholesterol synthesis precisely during stages, when the KV and its cilia form and actively determine LR asymmetry. We thus treated zebrafish embryos from tailbud stage to 20 somites with atorvastatin, which coincides with stages, when the ciliated node in the mouse would form and function (embryonic day 7.5–8.5²⁷). Treatment during this period was sufficient to generate embryos with the same impairments in cardiac and cilia function (Supplementary Figure 5). Therefore, we concluded that cholesterol may be required for proper functioning of the KV. Neither cilia length nor cilia number per KV, however, were altered upon atorvastatin treatment (Fig. 3a–c). The area of the KV, however, was significantly larger (Fig. 3d). Therefore, we analysed whether all cells of KV extended cilia. We found that atorvastatin increased the number of cells per KV, which failed to form a cilium (Fig. 3e). As this potentially impacts on the flow within the node we assessed expression of the flow sensor charon. Charon is initially expressed around the whole KV, but becomes downregulated on the left side and asymmetrically expressed when the counter-clockwise, nodal flow is properly established. In the case of irregular or clockwise flow, Charon can be found equally expressed on both sides of the KV or more pronounced on the left side²⁸. Atorvastatin-treated embryos did in fact show ambiguous expression of charon (Fig. 3f), suggesting that disturbed flow may be the underlying cause of the observed LR asymmetry defects. In summary, atorvastatin causes heart defects likely due to a malformation of the temporal organizer of laterality.

Open in a separate window

Fig. 3

Atorvastatin impacts on the development of the temporal organ of laterality. a Confocal stacks of 6–8 somite stage (ss) Kupffer’s vesicle (KV). Cilia were stained with an antibody against acetylated tubulin (acetTub, green). Apical cell membranes to visualize the KV outline were stained with an antibody against atypical PKC (PKCζ, magenta). Scale bar: 20µm. b Bar graph displaying the length of KV cilia in control and atorvastatin-treated embryos; n=1231 (DMSO) and 1424 (50µM atorv.) cilia; p=0.6357, two-tailed Mann–Whitney test. c Cilia number per KV was not changed in the presence of atorvastatin; p=0.6861, two-tailed t-test with Welch’s correction. d KV area was larger upon atorvastatin treatment. Each circle is one KV and embryo. ** p=0.0034. Two-tailed Mann–Whitney test. e Atorvastatin-treated embryos have more unciliated cells in the KV than control embryos. Cilia were stained using anti-acetylated antibody and KV cells were counted using the Sox17-GFP line. KV cells express high levels of GFP and can be distinguished from surrounding cells outside the KV. Each circle symbolizes one embryo; *p=0.0375, Kruskal–Wallis test with Dunn’s multiple comparison test. f Atorvastatin disturbs correct spatial expression of charon. Representative images of in situ hybridization for charon at 10 ss with correct strong expression on right side and aberrant strong expression on both sides or left side (expression indicated by arrowheads; absent expression not shown). Scale bar: 50µm. Stacked bar graphs summarises three experiments. Fisher’s exact test; n=57 (DMSO), 51 (10µM atorvastatin) and 60 (50µM atorvastatin); ****p<0.0001 for DMSO vs 10µM as well as vs 50µM. b–f Numbers of embryos analysed indicated below or next to graphs

Atorvastatin interferes with cilia formation and signalling

Since residual sterols from the yolk could ameliorate the degree of cholesterol reduction and thus mask a cilia length phenotype in early zebrafish embryos, we analysed cilia also at 24 hpf when more of the yolk had been used up. At this stage, we observed a reduction in cilia length in the neural tube (Supplementary Figure 6) suggesting that atorvastatin does not only interfere with ciliation per se, but potentially with cilium extension, too. In addition, we detected malformations of the pronephric tubes suggesting an additional impact on renal cilia. Embryos treated with atorvastatin developed kidney cysts and showed impaired glomerular fusion and thickening of the distal pronephric tubule (Supplementary Figure 7). Moreover, Tetrahymena thermophila, a multiciliated protozoan, had shorter cilia when treated with atorvastatin (Supplementary Figure 8). To investigate atorvastatin’s effects on cilia more closely, we turned to a heterologous cell system, in which lipid abundance can be better controlled than in a whole organism. We used hTert immortalized human fibroblasts and induced primary cilium formation by serum starvation for 72h with concomitant atorvastatin administration or DMSO application as control (Fig. 4a). Consistent with our observation in the KV of zebrafish embryos, fewer cells extended cilia when atorvastatin was applied during ciliogenesis (Fig. 4b, c). In addition, cilia length gradually decreased with increasing concentrations of atorvastatin (Fig. 4d). To test whether this was due to cholesterol depletion we administered cholesterol to the cells for a short time during starvation. This was sufficient to restore the percentage of ciliated cells as well as the length of cilia (Fig. 4e, f). We next excluded that the shortening of cilia was caused by an overall shrinking of the plasma membrane by measuring the cell size upon atorvastatin treatment. The size of trypsinized fibroblasts was not changed by atorvastatin (Supplementary Figure 9).

Open in a separate window

Fig. 4

Atorvastatin disrupts primary cilia formation and function in mammalian cells. a Cell culture conditions to induce cilium formation by starvation in the presence of atorvastatin. b Atorvastatin administration to human fibroblasts results in fewer and shorter cilia compared to DMSO-treated control cells. acetTub (acetylated tubulin, magenta): cilia; gTub (γtubulin, green): basal bodies. Scale bar: 10µm. c Atorvastatin dose-dependently prevents primary cilia formation in serum-starved fibroblasts. Number of cilia/individual treatments: 412/8 (DMSO), 212/4 (1µM atorv.), 287/6 (5µM atorv.), 155/3 (10µM atorv.), 287/4 (20µM atorv.), 250/4 (50µM atorv.); ****p<0.0001 (atorvastatin versus control treated cells). d Fibroblast cilia length gradually decreases upon increasing atorvastatin concentrations. Number of cilia/individual treatments: 320/7 (DMSO), 214/4 (1µM atorv.), 269/6 (5µM atorv.), 97/4 (10µM atorv.), 147/4 (20µM atorv.), 47/2 (50µM atorv.); ****p=0.0001. e Cholesterol supplementation restores cilia formation in fibroblasts; n=3 treatments with 248–254 cilia; **p=0.0034 (DMSO vs. 10µM atorv.), p=0.0073 (10µM atorv. vs. 10µM atorv.+cholesterol). f Cilia length of human fibroblasts is rescued by cholesterol; n=81–112 cilia from 3 treatments. Circles: individual cilia; **p=0.0027, ****p<0.0001. g Experimental outline to induce ciliary signalling through SAG or PDGF-AA stimulation. h SAG-induced Hedgehog signalling is attenuated by atorvastatin. p values (stimulated vs. non-stimulated): 0.0148 (DMSO), 0.6375 (10µM atorv.), 0.9594 (50µM atorv.). i The response to PDGF-AA stimulation (as measured by AKT phosphorylation at serine 473) is blunted upon atorvastatin treatment compared to DMSO-treated human fibroblasts; n=4. Means±SEM; *p=0.0375, two-tailed paired t-test after log-transformation of values to obtain normal distribution. j mIMCD3 cells cultured as described in (a) develop fewer and shorter cilia in the presence of atorvastatin. Representative confocal z-stacks; cilia: magenta; basal bodies: green. k Atorvastatin results in significantly fewer ciliated mIMCD3 cells. Means±SEM; n=4. ***p=0.0002, two-tailed t-test with Welch’s correction. l Cilia length is reduced under atorvastatin in mIMCD3 cells; n=355 (DMSO) and 188 (atorv.) cilia; ****p<0.0001, two-tailed Mann–Whitney test. c, e, h One-way ANOVA with Sidak’s post-test. d, f Kruskal–Wallis test with Dunn’s multiple comparison test

Concomitant with the loss of cilia we also detected a failure to induce ciliary signalling. Human fibroblasts were again serum starved and simultaneously treated with atorvastatin. In the last 24h of starvation, Hedgehog signalling was stimulated using the Smoothened agonist SAG (Fig. 4g). Alternatively, platelet-derived growth factor (PDGF) receptor-α signalling was induced using a short stimulation (3min) with PDGF-AA (Fig. 4g). While direct stimulation of the Hedgehog signal transducer Smoothened using SAG produced robust target gene expression in control cells, we found target gene transcription completely abrogated in atorvastatin-treated cells (Fig. 4h). This result could also be transferred to zebrafish, which equally displayed reduced Hedgehog signalling upon atorvastatin treatment (Supplementary Figure 10). In addition, we assessed PDGF-AA signalling, which is similarly executed via the primary cilium²⁹. PDGF-AA reproducibly triggered AKT phosphorylation in serum-starved control cells, but to a significantly lower extent in atorvastatin-treated fibroblasts (Fig. 4i). To test whether primary cilia of other mammalian cells reacted the same to statin treatment, we treated murine inner medulla and collecting duct (mIMCD3) kidney cells with atorvastatin according to the protocol outlined in Fig. 4a. We saw a robust reduction in ciliation efficiency and shortened cilia in the presence of atorvastatin (Fig. 4j, l). Finally, we assessed the impact of atorvastatin on multiciliated cells. Human airway epithelia cells were cultured at air–liquid interface (ALI) configuration. When atorvastatin was administered to the medium at the day when ALI was established and remained in the medium during the entire differentiation and maturation time, cilia failed to form (Supplementary Figure 11a). In contrast, DMSO-treated control cells had fully polarized as shown by ZO-1 staining and extended the first cilia (Supplementary Figure 11a). When atorvastatin was added on day 14 of ALI culture after ciliation had already started, we did not see changes in the number of ciliated cells, but lesions within the epithelium had formed (Supplementary Figure 11b-d). We hence conclude that atorvastatin treatment disturbs cilium formation and function and that this function likely is conserved from the evolutionarily low one-cell ciliate Tetrahymena over vertebrate zebrafish embryos to mammalian cells.

Atorvastatin-induced defects are likely due to cholesterol depletion

Cholesterol is one of several end-products of the mevalonate pathway, which is inhibited by statins at a very early step³⁰ (Fig. 5a). To test whether the observed effects originated from a loss in cellular cholesterol, we applied specific inhibitors, which block either farnesyltransferase, the enzyme catalysing the formation of prenylated proteins, or squalene synthase that is required for the generation of cholesterol. In our hands, only inhibition of squalene synthase via YM-53601 treatment significantly reduced cilia length in human fibroblasts (Fig. 5b). We further corroborated this finding by depleting human fibroblasts from free cholesterol using the scavenger methyl-β-cyclodextrin (MβCD) (Fig. 5c). MβCD is commonly used to efficiently deplete cells acutely from free cholesterol³¹. Both, cholesterol depletion by MβCD right before inducing ciliogenesis by starvation and persistent cholesterol depletion during starvation resulted in shorter and fewer cilia (Fig. 5d–f), indicating that cholesterol is not only required to modify Hedgehog signalling molecules³², but also to allow for correct ciliogenesis. Both MβCD treatment regiments also abrogated Hedgehog signalling in human fibroblasts (Fig. 5g) and decreased the percentage of ciliated cells and the length of cilia in mIMCD3 cells (Fig. 5h–j). Finally, we measured cilia length in cells derived from a SLOS patient who has been identified as a compound heterozygote harbouring a G to T transversion in the DHCR7 gene in one allele (c.413G>T, resulting in an amino acid change: Gly138Val) and a C>T transition at nucleotide 1213 (c.1213C>T) resulting in the substitution of tyrosine for histidine at position 405 (His405Tyr). Besides partial syndactyly this patient had been diagnosed with failure to thrive, developmental delay, hypotonia, microcephaly and micrognathia. Fibroblasts of this patient produce less cholesterol and display an accumulation of 7-dehydrocholesterol as well as oxidized products thereof³³. We found that cilia in cells of this patient are significantly shorter than in control cells (Fig. 5k–m), which could be mimicked by pharmacological inhibition of 7-dehydrocholesterol reductase, the enzyme defective in the SLOS cells used in here³⁴, using the compound AY9944 (Fig. 5l). Addition of external cholesterol restored cilium length in SLOS patient fibroblasts (Fig. 5m). Last, but not least, just like in atorvastatin-treated cells, PDGF-AA signalling was blunted in SLOS cells compared to healthy cells (Fig. 5n). These results strongly indicate that cholesterol is required to allow for cilium formation and function.

Open in a separate window

Fig. 5

Cholesterol depletion by different means impairs ciliogenesis. a Cartoon of cholesterol biosynthesis including administered inhibitors and the cholesterol scavenger methyl-β-cyclodextrin (MβCD). L744,832, farnesyltransferase inhibitor. YM-53601, squalene synthase inhibitor. b 5µM YM-53601 reduces cilia length, 20µM L744,832 does not; n=75–187 cilia from 4–5 treatments; ****p<0.0001, ns p=0.4689. c Cells were treated either with 10mM MβCD for a short time during the onset of ciliogenesis or throughout the whole starvation period using 0.5mM. d Confocal z-stacks of MβCD-treated human fibroblasts. Scale bar: 10µm. e MβCD treatments result in shorter cilia in fibroblasts; n=128 (DMSO), 108 (short), 135 cilia (long); **p=0.0026, ****p<0.0001. f MβCD treatment reduces ciliation in fibroblasts. Means±SEM; n=3; *p=0.0158, **p=0.0083. g SAG-induced Smoothened signalling is impaired upon MβCD treatment. Cells were treated as depicted in (c) and stimulated as in Fig. 4h; n=6; *p=0.023, **p=0.0074, ns p=0.0572. h Confocal z-stacks of mIMCD3 cells treated with MβCD as described in (c). Cilia in magenta (acetTub), basal bodies in green (gTub). Scale bar: 10µm. i MβCD treatment reduces ciliation in mIMCD3 cells (same DMSO control as in Fig. 4k); n=4; *p=0.0265, **p=0.0013, One-way ANOVA with Dunnett’s multiple comparison test. j MβCD shortens cilia in mIMCD3 cells (same DMSO control as in Fig. 4l); n=350 (DMSO), 303 (short) and 275 (long) cilia; ****p<0.0001. k In Smith–Lemli–Opitz syndrome (SLOS) patients, DHCR7 is defective resulting in impaired conversion of 7-dehydrocholesterol to cholesterol. l Patient-derived fibroblasts with a genetic variant of DCH7 (G138V; H405Y) extend shorter cilia than control fibroblasts. Treatment of healthy fibroblasts with 5µM of the DHCR7 inhibitor AY9944 shortens cilia; n=282 (DMSO), 161 (SLOS) and 65 cilia (AY9944); ****p<0.0001 (compared to DMSO). m Cilia length of SLOS patient cells is restored after cholesterol supplementation; n=64–77 cilia; ****p<0.0001. n Blunted response to PDGF-AA in SLOS fibroblasts; n=4. Means±SEM; *p=0.0201, two-tailed paired t-test after log-transformation of values to achieve normal distribution. b, e, j, l, m Kruskal–Wallis test with Dunn’s multiple comparison test. f, g One-way ANOVA with Holm–Sidak’s multiple comparison test

Zebrafish husbandry and manipulation of fertilized eggs

Zebrafish were kept in a circulating water system at 27.5–29°C with a 14h day and 10h night cycle. Fish were fed three times per day. Fertilized eggs were obtained by natural mating of adult zebrafish (4–24 months of age) and incubated at 28.5°C until the desired stage. Dead or unfertilized eggs were sorted out. Zebrafish lines used in this study were AB wild-types, cmlc2-nucDsRed⁶², Wt1b-GFP⁶³ and Sox17-GFP⁶⁴. Drug treatments were performed in 24-well plates in 1ml egg water supplemented with 0.003% phenyl-thiourea and 1% DMSO as vehicle or drugs dissolved in DMSO as indicated from tailbud stage on. The size of experimental groups depended on the number of fertilized eggs per clutch. Clutches were randomly and equally divided into treatment groups and raised until further analysed. Otherwise no statistical methods have been applied to pre-determine sample size. As there are always embryos also in control groups, which do not survive, the size of the finally analysed groups could not be held equal. All zebrafish experiments were done at least three times, each time with groups of at least 10 embryos and analysed, whenever possible, in a blinded fashion. The numbers of the individual embryos analysed are given within the graphs. Maintenance and manipulation of zebrafish were in compliance with ethical regulations and approved by the Veterinary Care Unit at Ulm University (“Tierschutzbeauftragte”) as well as the animal welfare commissioner of the regional board for scientific animal experiments in Tübingen, Germany. Experiments were performed according to the European Union Directive 86/609/EEC for the protection of animals, which are intended for experimental and other scientific purposes.

Zebrafish morphology, heart function and kidney formation

Live zebrafish were analysed at the 48 hpf stage and examined for body curvature, pericardiac oedema and otolith defects. In addition, heart rates were counted using an upright microscope with brightfield illumination (see “Imaging”). In detail, zebrafish embryos were placed into a Petri dish with pre-warmed embryo water (28°C) and the number of heart beats was manually counted for 10s. Then, beats per minute were calculated. Kidney formation was assessed with the help of the Wt1b-GFP transgenic line at 48 and 72 hpf.

Analysis of cardiomyocyte number in zebrafish

Zebrafish expressing DsRed under the control of the cmlc2 promotor⁶² were treated with atorvastatin or DMSO from tailbud stage on and fixed at 48 hpf with 1% formaldehyde for 50min at room temperature (RT). After washing with PBS, embryos were incubated for 1h in blocking solution containing 10% normal goat serum (Vectorlabs, UK), 2mg/ml bovine serum albumin (BSA) and 0.2% saponin in phosphate-buffered saline (PBS). Primary antibodies diluted in PBS with 0.2% saponin were added to the embryos and incubated overnight at 4°C. An antibody against DsRed was used to enhance the fluorescence of the DsRed. Atrial myosin was stained with the S46 antibody to distinguish atrial from ventricular cardiomyocytes. Next day, samples were washed several times for 15min with PBS supplemented with 0.2% saponin and subsequently incubated in secondary antibodies (diluted in blocking solution) for 3h at RT, followed by several washing steps with PBS. After mounting the embryos with Vectastain containing 4′,6-diamidino-2-phenylindole (DAPI; Vectorlabs) between coverslips, samples were stored at 4°C until imaging via confocal microscopy.

In situ hybridization

Whole mount in situ hybridization was performed according to standard protocols. The SP6 or T7 polymerase (NEB, Germany) along with the DIG RNA labelling system (Roche, Germany) were used to in vitro transcribe probes from linearised plasmids. To detect charon transcripts a 583bp fragment of the complementary DNA (cDNA; NM_212969) was cloned by TOPO TA cloning into pCRII, from which a DIG-labelled probe was transcribed after linearization. All other probes used in this project have been described before: cardiac myosin light chain 2 (cmlc2), chondroitin sulphate proteoglycane 2a (cspg2a), insulin (ins), lefty2, notch1b, nkx2.5, pitx2⁶⁵ and tbx2b^66,67.

Cell culture

Human hTert immortalized 1BR3 wild-type fibroblasts (Genome Damage and Stability Centre, University of Sussex, UK) were cultured in minimum essential medium (MEM) Alpha supplemented with 1% penicillin/streptomycin (Pen/Strep) and 10% foetal calf serum (FCS, all from Life Technologies, Germany) at 37°C in a humidified atmosphere containing 5% CO₂. Primary SLOS patient fibroblasts (clone GM05788) from a compound heterozygous patient with mutations in 7DHC reductase were obtained from the NIGMS Human Genetic Cell Repository at the Coriell Institute for Medical Research. As control, non-immortalised 1BR3 cells (Genome Damage and Stability Centre, University of Sussex, UK) were used. Primary cells were cultured in the presence of 15% FCS. mIMCD3 were cultured in Dulbecco's modified Eagle's medium (DMEM):F12 medium completed with 1% Pen/Strep and 10% FCS. All cells have been regularly checked for mycoplasma contamination and have been tested free of mycoplasma. The day before starting an experiment, 150,000 cells/well were seeded in a 6-well plate. On the following day, cells were changed from full to starvation medium (MEM Alpha or DMEM:F12 with 1% Pen/Strep and 0.1% FCS) supplemented with the pharmacologically active compounds or DMSO for 72h. To assess Hedgehog pathway activity, serum-starved cells were treated with 500nM SAG during the last 24h of starvation. To monitor PDGF pathway activity, cells were stimulated using 50nM PDGF-AA for 3min at the end of the 3 days of starvation.

ALI cultures

ALI cultures were essentially performed as previously described⁶⁸. In detail, cryo-preserved primary human bronchial epithelial cells from healthy donors (Epithelix, EP51AB, Genève Switzerland) were thawn and suspended in 75cm² culture flasks with 20ml airway epithelial cell basal medium supplemented with airway epithelial cell growth medium supplement pack (catalogue nos. C-21260 and C-39160, PromoCell, Germany) at 37°C, 5% CO₂ and 95% humidity. Then, 10ml of the medium was changed every second day until 90% confluency was reached. Cells were detached using the DetachKit of Promocell (cat. no. C-41200). To establish ALI epithelia, 30–40×10³ hTEpC were seeded onto collagen-coated (cat. no. 04902, StemCell Technology, Germany) transwell filters (cat. no. 3470, Thermo Fisher). Filters were placed in 24-well plates and wells were filled with 600µl each of airway epithelial cell basal medium supplemented containing airway epithelial cell growth medium supplement pack. Cells were cultivated on filters under submerged conditions until confluence was reached (usually for 4 days) at 37°C, 5% CO₂ and 95% humidity. Then, apical medium was aspirated and basolateral medium was replaced by ALI medium containing DMEM-H and LHC basal medium (both from Gibco via Thermo Fisher, Germany) at a 1:1 ratio and was supplemented with: 0.87µM insulin, 0.21µM hydrocortisone, 0.5ng/ml human epidermal growth factor, 0.01µM triiodothyronine, 0.125µM transferrin, 2.7µM epinephrine, 10µg/ml bovine pituitary extract, 0.5mg/ml bovine serum albumin (all from Promocell, Germany), 0.5µM phosphorylethanolamine, 0.5µM ethanolamine, 3µM zinc sulphate, 0.05µM retinoic acid, 1.5µM ferrous sulphate, 0.11mM calcium chloride, 0.6mM magnesium chloride, 30µM sodium selenite, 1µM manganese chloride, 0.5nM sodium silicate, 1µM ammonium molybdate tetrahydrate, 5µM ammonium metavanadate, 1µM nickel sulphate, 0.5µM tin chloride (all from Sigma-Aldrich GmbH, Germany), 100U/ml penicillin and 100µg/ml streptomycin (both from Gibco via Thermo Fisher, Germany). Basolateral medium was changed every second day. From day 14 on, additional PBS washing steps were performed two times per week to avoid mucus accumulation. To address the impact of atorvastatin on ciliogenesis, cells were either treated from day 1 of epithelial differentiation by addition of atorvastatin to the basolateral medium and fixed on day 14 or treated from day 14 until day 28. Cells were then fixed with 4% paraformaldehyde (PFA) followed by a second fixation with MeOH.

MβCD assay

Two different approaches were used to deplete cholesterol from 1BR3 fibroblasts or mIMCD3 cells with MβCD. Cells were treated with 10mM MβCD for 30min prior to starvation or incubated in 0.5mM MβCD during the whole course of starvation. In brief, cells cultured in full medium were washed once with PBS and changed to serum free medium supplemented with MβCD to a final concentration of 10mM. After 30min of incubation, cells were washed again with PBS and changed to serum starvation medium. For long-term treatments, MβCD was added to cells changed to starvation medium at a concentration of 0.5mM and incubated until fixation.

Cholesterol replenishment

For rescue experiments, water-soluble cholesterol was used and titrated to the maximal concentration, which did not induce cell death. SLOS patient and control cells were incubated with 5mM water-soluble cholesterol 1h prior to starvation. After washing with PBS, cells were starved for 72h.

As additional cholesterol proved highly toxic to healthy control cells, a different treatment strategy was used for 1BR3 cells treated with atorvastatin: at the onset of starvation, 10µM atorvastatin was added. After 48h, replenishment was performed by co-treatment with 5mM cholesterol for 1h. Afterwards, starvation and statin treatment was continued for another 24h until fixation. In zebrafish, 10µM water-soluble cholesterol was added along with 10µM atorvastatin from tailbud until the 48 hpf stage.

Immunofluorescence

Zebrafish embryos were treated from tailbud stage until fixation at either 6–8 somite stage for KV analyses or at 24 hpf for neural tube analysis. Fixation was done at 4°C overnight using 4% buffered PFA. Permeabilization and immunofluorescence staining of zebrafish embryos were performed as previously published⁶⁹.

Cells grown on coverslips were fixed with 100% chilled methanol for 5min or 4% PFA at room temperature for 10min. ALI cultures were fixed using 4% PFA followed by a second fixation with MeOH. After fixation, cells and filters were washed with PBS, permeabilized for 10min (cells) or 15min (filters) with 1% Triton X-100 in PBS, blocked for at least 1h in PBS supplemented with 10% normal goat serum (Vectorlabs, UK) and incubated in primary antibody diluted in blocking buffer. Next morning, the samples were washed three times for 10min using PBS and incubated at room temperature in the dark with the secondary antibody diluted in blocking buffer (1:1000). Finally, coverslips and ALI filters were mounted with Vectastain containing DAPI (Vectorlabs, UK) and sealed with nail polish. For cilia length analysis in human fibroblasts and mIMCD3 cells, at least 30 cilia per treatment and experiment were imaged. To assess the percentage of ciliated cells, at least 100 cells per treatment condition and experiment were counted. Each experiment was performed at least 3 times.

Analysis of area and cell number of the KV

In confocal stacks, protein kinase C ζ (PKCζ) staining was used to visualize the outline of the KV, which was then traced with ImageJ to calculate the KV area. The number of cells per KV was counted in Sox17-GFP embryos, in which brightly fluorescing cells represent KV cells.

PDGF assay and western blotting

1BR3 wild-type or SLOS patient cells seeded in 12-well plates were serum starved for 3 days and the wild-type cells in addition treated with 10µM atorvastatin. Before starting lysis in 60µl cell lysis buffer (2% SDS, 1× phosphatase and protease inhibitors (both Roche, Germany), 50mM Tris pH 6.8), the cells were stimulated for 3min with 50ng/ml PDGF-AA. DMSO supplementation was used as vehicle control for all treatments. Lysates were supplemented with 0.5µl Pierce universal nuclease (Thermo Fisher), cleared by centrifugation and protein content was analysed with the Bicinchoninic acid (BCA) assay kit (Sigma-Aldrich, Germany). Equal amounts of proteins were separated on a Bolt Bis-Tris Mini gradient gel (4–12 %) in a Bolt Mini Gel tank filled with MES SDS running buffer (all from Life Technologies, Germany) and blotted at 25 V for 1h onto a nitrocellulose membrane (Bio-Rad Laboratories, Germany) in a X Cell II blot module (Life Technologies, USA). Blocking of the membrane was performed with 3% milk powder in Tris-buffered saline containing 0.1% Tween-20 (TBST) for 1h at RT followed by overnight incubation at 4°C with primary antibodies diluted in TBST containing 3% BSA. Next day, blots were washed three times in TBST containing 0.2% BSA and 0.2% NP-40, respectively, and incubated for 1h at RT in IR-Dye-coupled secondary antibodies. Western blot signals were detected, normalized and quantified with a LI-COR Odyssey SA system and software. Antibodies used for pathway analyses were anti-AKT and anti-phospho-AKT, whereas glyceraldehyde 3-phosphate dehydrogenase (GAPDH) served as loading control.

qPCR

Total RNA was isolated from fibroblasts or zebrafish embryos using the Quick-RNA MiniPrep kit (Zymo, USA). cDNA was synthesized by Superscript III reverse transcriptase with oligodTTPs (Life Technologies, Germany) using equal amounts of total RNA as template. qPCR was run in duplicates or triplicates on a LightCycler 480 with the Universal Probe System (both Roche, Germany) and the GoTaq Probe qPCR Master Mix (Promega, USA). The following primers and probes were used: human SDHA: forward (Fw): 5′-GGA CCT GGT TGT CTT TGG TC, reverse (Rev): 5′-CCA GCG TTT GGT TTA ATT GG, Universal Probe 80; human GLI1: Fw: 5′-CCA GCC AGA GAG ACC AAC AG, Rev: 5′-CCC GCT TCT TGG TCA ACT T, Universal Probe 7; zebrafish b2m: Fw: 5′-ACA TCA CTG TAC AGG GGA AAG TC, Rev: 5′-TCC GTT CTT CAG CAG TTC AA, Universal probe 65; zebrafish gli1: Fw: 5′-GGT CTC GAT GCC AGT GGA, Rev: 5′-CAC TGA CGG AGC CAG TCC, Universal probe 5.

Imaging

For cilia analyses z-stacks were acquired using a Leica TCS SP5II confocal microscope with a section distance of 0.3µm. The stacks were merged using the 3D projection tool in the Leica acquisition software of the confocal and imported into ImageJ⁷⁰. With the help of a Wacom tablet individual cilia were traced and the length of the trace was measured in ImageJ. In Tetrahymena, only cilia in the central portion of the cell were measured. A Zeiss Axiophot epifluorescence microscope was used to verify and analyse staining efficiency and to image ALI cultures. All zebrafish live or whole mount in situ hybridization images were acquired with a Leica M125 upright microscope and a Leica IC80 HD camera for brightfield images and a Leica M205 FCA microscope equipped with a Leica DFC9000GT sCMOS camera for fluorescence images.

We thank Sandra Burczyk and Cornelia Donow for excellent help in fish maintenance and the zebrafish community for the generous supply with plasmids and fish lines. The S46 antibody developed by Dr. Stockdale was obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology, Iowa City, IA. This study was funded by the Deutsche Forschungsgemeinschaft (PH144/4-1 to M.P.; Pulmosens GRK 2203 to O.H.W.) and by the Boehringer Ingelheim Ulm University Biocenter (to M.P. and O.H.W.). L.M. is a fellow of the International Graduate School in Molecular Medicine at Ulm University, which is funded by the German Excellence Initiative of the Deutsche Forschungsgemeinschaft. Grant funding by the Boehringer Ingelheim University Biocenter is not coupled to the pharmaceutical company’s interests and has not influenced the current study.