Detection of Phospholipidosis Induction: A Cell-Based Assay in High-Throughput and High–Content Format

Abstract

Drug-induced phospholipidosis is characterized by the accumulation of intracellular phospholipids in cells exposed to cationic amphiphilic drugs. The appearance of unicentric or multicentric multi-lamellar bodies viewed under electron microscope (EM) is the morphological hallmark of phospholipidosis. Although the EM method is the gold standard for detecting cellular phospholipidosis, this method has its drawbacks, including low throughput, high cost, and unsuitability for screening a large chemical library. In this study, a cell-based phospholipidosis assay has been developed using the LipidTOX Red reagent in HepG2 cells and miniaturized into a 1536-well plate format. In order to validate this assay for high throughput screening, the LOPAC library of 1280 compounds was screened on a quantitative high throughput screening platform. A group of known phospholipidosis inducers, such as amiodarone, propranolol, chlorpromazine, desipramine, promazine, clomipramine, and amitriptyline, was identified by the screen, consistent with previous reports. Several novel phospholipidosis inducers including NAN-190, ebastine, GR127935 and cis-(Z)-flupenthixol were identified in this study and confirmed using the EM method. These results demonstrate that this assay can be used to evaluate and profile large numbers of chemicals for drug-induced phospholipidosis.

Introduction

Phospholipidosis (PLD) is characterized by excessive intracellular accumulation of phospholipids, with a distinct morphological appearance of lamellar bodies visible under an electron microscope (EM).¹ Drugs, including many antibiotics, antidepressants, and antihistamines, induce PLD in various tissues such as lung,² heart,³ liver,⁴ and kidney.⁵ These compounds share the common feature of a cationic amphiphilic structure, and so are commonly called cationic amphiphilic drugs (CADs). Specifically, CADS contain a hydrophobic aromatic or aliphatic ring structure and a hydrophilic side chain containing a cationic amine group which is charged at physiological pH.¹ There are several hypothesized mechanisms for PLD including: direct inhibition of the enzymatic activity of the phospholipases or other intralysosomal catabolic pathways by CADs,^{6; 7} or the binding of CADs to phospholipids resulting in indigestible drug-lipid complexes.^{6; 8}

At present, there is still no clear evidence that PLD has significant toxicological effects in animals or humans.⁹ However, several research groups found that gentamycin induced PLD in kidney tissues and also caused renal tubular toxicity in the affected tissue.^{5; 10} Amiodarone, an antiarrhythmic agent, has been reported to induce liver cirrhosis accompanying phospholipidosis in a patient with long-term administration of a daily low dose of amiodarone. Reasor et al¹¹ also reported that amiodarone-induced pulmonary toxicity in rats may be linked to the induction of phospholipidosis, suggesting that drug-induced PLD may have potential adverse effects. Although the toxicological significance of PLD is still under investigation, additional information on the PLD potential of drugs is often requested by the U.S. Food and Drug Administration (FDA) in the drug development process.¹²

To date, the determination of PLD relies on transmission electron microscopy (EM), a gold standard method for the detection of PLD. However, the EM method is time-consuming, labor intensive, and unsuitable for screening a large number of compounds. Currently, various fluorescent dyes including NBD-PE,^{13; 14} NBD-PC,¹⁵ and Nile red¹⁶ have been developed to measure PLD in cells. More recently, Nioi et al¹⁷ also developed a PLD assay in 96-well plate format using LipidTOX Red, a fluorescent lipophilic dye. However, these assays have complicated assay protocols requiring several wash steps, which limit throughput. In this study, we have validated a HepG2 cell-based PLD assay in 1536-well plate format, and screened 1280 compounds in the LOPAC (Library of Pharmacologically Active Compounds) collection in an established quantitative high-throughput screening (qHTS) platform¹⁸. The PLD inducers identified from the primary screen were confirmed using the EM method. Thus, the results suggest that this assay is robust and can be used to screen large compound collections to identify novel PLD inducers.

Materials and Methods

Cell Culture

Human HepG2 (Hepatocellular carcinoma) cell line was purchased from the American Type Culture Collection (ATCC, Manassas, VA, U.S.A.). HepG2 cells were cultured in Minimum essential (Eagle) medium (ATCC, Manassas, VA, U.S.A.) supplemented with 10% FBS (Hyclone Laboratories, Logan, UT, U.S.A.) and 100U/ml penicillin and 100µg/ml streptomycin (Invitrogen, CA, U.S.A.). The cells were maintained at 37°C under a humidified atmosphere and 5% CO₂.

Reagents

The LipidTOX Red dye and Hoechst 33342 were obtained from Invitrogen. A library of pharmacologically active compounds (LOPAC), containing 1280 compounds with known pharmacological actives, as well as other compounds used in this study were purchased from Sigma-Aldrich (St. Louis, MO, USA).

LipidTOX PLD assay and qHTS

The PLD assay can quantify the PLD changes in cells. A LipidTOX Red dye is used to fluorescently label intracellular phospholipids to detect PLD formation. The assay protocol in a 1536-well plate format is described in Supplemental Figure S1. Briefly, HepG2 cells were dispensed at 900 cells/3µL/well, using a Multidrop Combi 8 channel dispenser (Thermo Fisher, Waltham, MA, USA), into Collagen I-coated (BioMedTech Laboratories, FL, USA) 1536-well black wall/clear bottom assay plates (Greiner Bio-One North America, NC, USA). The assay plates were incubated at 37°C for 4 h to allow the cells to attach to wells, followed by addition of 3 µL of 1:500 LipidTOX Red dye using Multidrop Combi 8 channel dispenser, and then 23 nL compounds via the pin tool (Kalypsys, San Diego, CA, USA). The final concentration of the compounds is ranging from 12 nM to 38.3 µM. Amiodarone (19.16 µM final), a known PLD inducer,¹⁹ was used as a positive control and DMSO was used as a negative control in the screening. The assay plates were then incubated at 37°C for 24 h, 4µL of 8% paraformaldehyde (3.2% final) fixative solution containing 10 µg/ml Hoechst 33342 (4 µg/mL final) in Dulbecco’s Phosphate-Buffered Saline (DPBS) was added to each well using a Kalypsys Washer/Dispenser. After incubation at room temperature (RT) for 15–30 min, the assay plates were washed once with DPBS using a Kalypsys Washer/Dispenser. The assay plates were sealed and stored at 4°C before imaging. The fluorescence intensities (595 nm excitation, 615 nm emission for LipidTOX Red; 352 nm excitation, 461 nm emission for Hoechst 33342) were measured using an ImageXpress Micro Wide field High Content Screening System (Molecular Devices, Sunnyvale, CA, USA) with a 20× Plan Fluor objective. Images were acquired for one site (6.3% of a single well area in a 1536-well plate) in each well and analyzed with the MetaXpress® software (MDC) using the Transfluor module. Several algorithmic outputs, such as Pit Count per Well, Pit Count per Cell, Pit Integrated Intensity per Well (the sum of the intensities for all detected pits in a well) and Pit Integrated Intensity per Cell (the sum of the intensities for all detected pits in a well/number of the cells in the well), were used for quantitative image analysis for PLD.

Cell Viability

Cell viability after compound treatment was determined using a CellTiter-Glo® viability assay (Promega, Madison, WI, USA) by measuring intracellular ATP content. The cells were dispensed at 900 cells/3 µL/well in 1,536-well white wall/solid bottom assay plates. After assay plates were incubated for 4 h at 37°C, 3 µL of 1:500 LipidTOX Red dye using Multidrop Combi 8 channel dispenser and 23 nL compounds via the pin tool were added into each well. The assay plates were incubated at 37°C for 24 h, followed by the addition of 4 µL/well of CellTiter-Glo® reagent using a Flying Reagent Dispenser (FRD). After 30 min incubation at RT, the luminescence intensity of the plates was measured using a ViewLux plate reader (PerkinElmer, Shelton, CT, USA).

Electron Microscopy Analysis

Cells were seeded at 500,000 cells/well into 6-well plates (Corning-Costar, Tewksbury, MA, USA). After overnight incubation, the assay plates were treated with five PLD compounds at approximately the EC₅₀: 6 µM amiodarone, 3 µM NAN-190, 1 µM ebastine, 3 µM GR127935, 6 µM cis-(Z)-flupenthixol, or dimethyl sulfoxide vehicle (DMSO; Fisher Scientific, Pittsburgh, PA) control for 24 h. The assay plates were then washed once with DPBS and fixed in Electron Microscopy (EM) fixative solution (2% glutaraldehyde in 0.1M cacodylate buffer, pH 7.2) for 1 h at RT, and then stored at 4° C. For further EM processing and embedding, the cells were post fixed in 1% osmium tetroxide in the same buffer for 1 h and en bloc stained with 0.5% uranyl acetate in 0.1 M acetate buffer, pH 4.2. The cells were then dehydrated in graded ethanol solutions (35%, 50%, 70%, 95%, and 100%) and infiltrated overnight in epoxy resin (Poly/Bed 812, Polysciences). After adding fresh pure resin, the cell plates were cured for 72 h in 55°C. After removing the polystyrene plates, suitable areas for thin sectioning were selected, cut out with a jewelry saw and glued onto empty resin stubs. About 70 nm thin sections were cut on an ultramicrotome (Leica EM UC6) and mounted on naked copper grids. The thin sections were double stained (uranyl acetate and lead citrate), and examined in a Hitachi H-7650 transmission electron microscope, and images were taken using an AMT CCD camera.

Data Analysis

Primary data analysis was performed as previously described.¹⁸ Briefly, raw plate reads for each titration point were first normalized to the amiodarone control (19.16 µM, 100%) and DMSO only wells (basal, 0%), and then corrected by applying a pattern correction algorithm using compound-free control plates (DMSO plates). The standard deviation (SD) of the corrected activities in the DMSO control plates and the lowest concentration plate in each titration series was calculated to assess the extent of random variations in the assay and used as an indicator of assay performance. A good assay is expected to have SD <10%.²⁰ Concentration-response titration points for each compound were fitted to the Hill equation yielding concentrations of half-maximal induction (EC₅₀) and maximal response (activity at the highest test concentration, efficacy) values.

Compounds considered to be active in the cell-based PLD assay showed stimulation in the Pit Integrated Intensity per Well and Pit Integrated Intensity per Cell, and had an efficacy ≥ 30% in both of the above readings. These compounds were selected based on the potency, efficacy, novelty and structure-activity relationship, and then ordered from commercial vendors for confirmation and follow up studies.

Results

Optimization and Validation of the Cell-Based PLD Assay in 1536-Well Plate Format

To identify compounds that induce intracellular phospholipidosis, we used a high-content imaging method to optimize and validate the cell-based PLD assay. The assay was miniaturized into 1536 well plate format with a final assay volume of 6 µL. After HepG2 cells were treated with various concentrations of amiodarone for 24 h in the presence of a PLD dye (LipidTOX Red), the assay plate was fixed with formaldehyde and stained with Hoechst 33342 dye to identify nuclei. Pit Integrated Intensity per Cell was used for quantitative image analysis of puncta formation of PLD in each well. As shown in Fig 1A, amiodarone concentration dependently induced phospholipidosis in HepG2 cells with an EC₅₀ of 4.96 µM; which is consistent with a previous study conducted by Bhandari et al.¹⁹ High-content imaging revealed PLD induction by amiodarone in a concentration dependent manner as shown in Fig 1B.

Figure 1.

Effect of amiodarone on PLD induction in HepG2 cells. (A) Concentration response curve of amiodarone after 24 h treatment. Each value represents the mean ± SD of eight independent experiments. (B) Representative images of intracellular staining with LipidTOX red dye and nuclear staining with Hoechst dye in the absence or in the presence of amiodarone. Images acquired in ImageXpress microsystem using a 20× objective. Nuclei were stained blue and PLD were stained red.

To evaluate measurement parameters and optimize assay performance, a plate was tested using DMSO as negative control and amiodarone as the positive control. After 24 h treatment of amiodarone (19.16 µM final, 16 wells) or DMSO (32 wells), several algorithmic outputs, Pit Count per Well, Pit Count per Cell, Pit Integrated Intensity per Well and Pit Integrated Intensity per Cell, were used for quantitative image analysis of PLD formation. One and two sites were measured in each well. In the one site measurement, the mean signal-to-background (S/B) ratios were 3.6 ± 0.2, 4.1 ± 0.4, 12.6 ± 3.1, and 14.3 ± 4.2 fold in Pit Count per Well, Pit Count per Cell, Pit Integrated Intensity per Well and Pit Integrated Intensity per Cell, respectively. In the two site measurement, the S/B ratios were 3.5 ± 2.1, 4.4 ± 1.6, 15.7 ± 9.0, 17.5 ± 10.6 fold in Pit Count per Well, Pit Count per Cell, Pit Integrated Intensity per Well and Pit Integrated Intensity per Cell, respectively. The assay standard deviations were calculated among replicates of amiodarone and replicates of DMSO. The SDs from one site measurement for Pit Count per Well, Pit Count per Cell, Pit Integrated Intensity per Well and Pit Integrated Intensity per Cell were 9.0, 8.7, 2.9, and 3.2, respectively. The SDs from two site measurement for Pit Count per Well, Pit Count per Cell, Pit Integrated Intensity per Well and Pit Integrated Intensity per Cell were 8.0, 9.0, 2.4, and 3.4, respectively. The assay SD was used to decide upon the best readout parameter to pursue for further screening. The further experiments including primary screening and confirmation studies were performed by imaging one site per well and analyzing data using Pit Integrated Intensity per Cell.

Screening Performance and Statistics in qHTS

To evaluate the screening performance of the PLD assay in the qHTS format, we screened the LOPAC collection for PLD induction in HepG2 cells. The screen was carried out over 5 concentrations ranging from 12 nM to 38 µM and Pit Integrated Intensity per Cell was measured for 1 site per well. Amiodarone (19.16 µM final) was included as a positive control and DMSO was used as a negative control in each plate to monitor assay performance and plate-to-plate variations during the screen. The average S/B ratio was 15.1 ± 3.0 for Pit Integrated Intensity per Cell. The assay SD was 2.1 ± 1.6.

Identification of PLD Inducers

In the LOPAC primary screening, 187 compounds were identified with efficacy values > 30%. Among these compounds, 109 were cherry-picked and retested in the PLD assay. The activities for these 109 compounds are listed in Supplemental Table S1, resulting in a 100% concordance rate between the primary screening and confirmation testing. These 109 compounds were further clustered by structural similarity resulting in 29 structural clusters and 40 singletons. Among these compounds, 29 compounds are CADS that are known PLD inducers, shown in Table 1. The EC₅₀ values of these known PLD inducers are comparable with previously reported literature values except for amantadine and pentamidine, which had low efficacies in our assay (Table 1). Paroxetine,²¹ quinacrine,²² and trimipranine⁸ were found to induce PLD in HepG2 cells and the induction of PLD by these compounds was also reported in rat tissues including lung and brain or in biochemical studies. A group of tricyclic dibenzazepine antidepressants (cluster 12 and 14) including imipramine, desipramine, clomipramine, trimipramine, amitriptyline and doxepin are known PLD inducers; whereas opipramol, cyclobenzaprine and cyproheptadine were identified as novel PLD inducers in the present study (supplemental Table S1). A group of neuroleptic drugs with phenothiazine structures (cluster 16), and dibenzoxazepine structures (cluster 13), showed PLD induction in HepG2 cells. While chlorpromazine, thioridazine, promazine, and clozapine are known PLD inducers, the other positive compounds from clusters 13 and 16 were identified in the present study including promethazine, trifluoperazine, fluphenazine, and lozapine (supplemental Table S1).

Table 1.

Potencies and Efficacies of Known Phospholipidosis (PLD) Inducers Identified from PrimaryQuantiative High-Throughput Screening and Confirmed in the Follow-up Assays

Compound Names	Primary Screen EC50, µM (Efficacy, %)	Confirmation EC50, µM (Efficacy, %)	Literature- Reported EC50, µM	Testing System
(±)-Propranolol	12.6 (65.4)	15.8±4.4 (172.9±107.90)	12.9	I13-35 cells,14
Amantadine	0.9 (4.8)	3.9±4.5 (4.4±11.0)	239	I13-35 cells,14
Amiodarone	2.8 (65.2)	4.2±0.8 (139.8±17.9)	5.6	HepG2 cells,19
Amitriptyline	10.0 (72.9)	4.6±0.5 (166.3±15.9)	15	HepG2 cells,19
Chloroquine	28.2 (24.1)	23.6±5.3 (157.8±24.5)	11.8	I13-35 cells,14
Chlorpromazine	10.0 (83.6)	5.4±1.9 (134.4±46.8)	4.1	I13-35 cells,14
Citalopram	10.0 (22.5)	27.5±3.1 (168.9±37.5)	38	I13-35 cells,14
Clomipramine	14.1 (17.3)	8.7±1 (211.2±72.6)	13.8	I13-35 cells,14
Clozapine	12.6 (20.0)	18.2±4.8 (134.8±46.7)	Active at 16	HepG2 cells,17
Desipramine	11.2 (66.3)	5.1±1.1 (190.5±38.0)	9.1	I13-35 cells,14
Doxepin	10.0 (37.1)	10.1±1.9 (178.3±33.9)	Active at 25	HepG2 cells,17
Fluoxetine	12.6 (120.0)	3.4±0.5 (164.9±29.4)	7.2	HepG2 cells,19
Haloperidol	11.2 (95.7)	4.9±0.7 (105.0±14.6)	17	HepG2 cells,19
Imipramine	10.0 (45.8)	20.8±6.9 (174.8±53.2)	21	I13-35 cells,14
Ketoconazole	2.8 (25.6)	5.4±2.0 (109.3±4.0)	Active at 8.3	HepG2 cells,17
Labetalol	12.6 (14.3)	24.8±4.9 (143.2±81.8)	Active at 12.5 to 25	HepG2 cells,35
Loratadine	14.1 (127.0)	5.5±1.0 (68.2±10.0)	Active at 16	HepG2 cells,17
Maprotiline	4.0 (121.4)	3.1±1.0 (237.1±73.3)	4.9	I13-35 cells,14
Memantine	12.6 (13.1)	29.0±1.7 (70.9±39.8)	93.7	I13-35 cells,14
Paroxetine	5.6 (11.3)	3.0±0.7 (116.9±46.7)	Active at 5	H4 cells,21
Pentamidine	1.4 (9.7)	2.8±2.4 (10.0±1.5)	8.3	HepG2 cells,25
Promazine	22.4 (63.3)	5.7±1.9 (168.1±27.9)	5.3	I13-35 cells,14
Quinacrine	5.0 (65.7)	1.0±0.2 (120.2±14.2)	Active at 60 mg/kg	Rat brain,22
Quinidine	15.8 (53.6)	13.4±2.0 (149.7±19.0)	25	HepG2 cells,25
Tamoxifen	3.2 (54.3)	8.0±0.7 (177.0±39.5)	6.5	HepG2 cells,19
Thioridazine	3.2 (24.5)	5.3±1.2 (108.4±47.8)	8.3	HepG2 cells,19
Trimipramine	3.5 (40.0)	8.7±1.2 (135.6±28.7)	Active at 1 mM	Binding study,8
Zimelidine	14.1 (26.0)	27.4±1.5 (112.7±49.7)	30.4	I13-35 cells,14
Sertraline	3.5 (97.4)	1.4±0.4 (151.9±26.0)	5.8	I13-35 cells,14

Each value of potency (EC₅₀, µM) and efficacy (compared with PLD induction by 19.16 µM amiodarone as a % of positive control) is the mean ± SD of the results from four experiments, with each concentration tested in duplicate, in the confirmation assay. The efficacy is expressed in parenthesis. The literature reported data were determined mostly using LipidTox, NBD-PE, NBD-PC, and Nile Red dyes in various cell systems.

Among the 29 clusters identified from the primary screen, 12 clusters have not been previously reported as PLD inducers. For example, cluster 2 including ebastine, clemastine and aminobenzotropine; cluster 10 including NAN-190, p-MPPF, and GR 127935; and cluster 11 including cis-(Z)-Flupenthixol and chlorprothixene, significantly induced PLD formation in both the primary screen and cherry pick confirmation (supplemental Table S1).

Confirmation of PLD Inducers

To further confirm the PLD induction potential of compounds, 24 compounds (Table 2 and Fig 2) were selected based on the cherry-pick confirmation data and novelty of compound structure, and then purchased from commercial vendors. None of these compounds have previously been reported as PLD inducers except for amiodarone and propranolol positive controls. These compounds were selected from 12 clusters identified from the screening. The activities of all 24 compounds were confirmed in the PLD assay with a 100% confirmation rate. The EC₅₀ values for these 24 compounds in the confirmation studies correlated well with those in the primary screen (R² = 0.71) and cherry-pick confirmation (R² = 0.76). The most potent compound among the 24 confirmed compounds was ebastine, which had average EC₅₀ values of 0.36 µM in the primary qHTS and 0.20 µM in the confirmation study. The next most potent compounds were GR 127935 (average EC₅₀ = 0.8 µM), BIX 01294 (average EC₅₀ = 1.6 µM), and NAN-190 (average EC₅₀ = 2.0 µM); EC₅₀ values provided were from the confirmation study (Table 2).

Table 2.

Potencies and Efficacies of Novel PLD Inducers in Primary qHTS, Confirmation (PLD induction), and Cell Viability Assays.

Sample Name	CASRN	Structure	Confirmation EC50, µM (Efficacy, %)	Primary Screen EC50, µM (Efficacy, %)	Cell Viability IC50, µM (Efficacy, %)
(+)-Butaclamol	55528-07-9		4.9±1.2 (70.4±19.0)	6.3 (43.1)	47.8±32.2 (47.2±18.3)
(±)-Propranolol*	318-98-9		37.0±23.0 (295.9±123.3)	12.6 (65.4)	Inactive
5-(N,N-hexamethylene) amiloride	1428-95-1		15.2±7.7 (86.6±5.6)	13.0 (52.8)	56±0 (40.7±4.9)
Amiodarone*	19774-82-4		6.1±0.9 (117.1±38.0)	2.8 (65.2)	39±15.4 (63.1±32.1)
BIX 01294	1392399-03-9		1.6±0.4 (109.0±35.1)	3.5 (56.0)	10±1.6 (98.6±3.3)
BW 723C86	160521-72-2		20.0±4.8 (219.6±40.5)	11.2 (62.2)	45.7±14.6 (37.7±11.9)
CGS-12066A	109028-09-3		10.0±4.0 (153.0±56.8)	8.0 (103.7)	45.7±14.6 (56.7±13.3)
cis-(Z)-Flupenthixol	51529-01-2		7.6±4.4 (171.9±122.9)	1.4 (37.3)	62.8±0 (131.7±6.4)
Ebastine	90729-43-4		0.3±0.2 (84.0±50.1)	0.4 (44.5)	29.8±2.4 (78.3±2.5)
Eliprodil	119431-25-3		9.0±5.0 (57.4±33.4)	10.0 (39.7)	26.5±2.2 (24.6±2.1)
Ellipticine	519-23-3		9.0±3.3 (779.1±227.5)	5.0 (48.9)	29.8±2.4 (102.3±4.0)
GR 127935	148642-42-6		0.8±0 (74.2±5.8)	6.3 (110.4)	42.6±10.3 (106.7±13.6)
GW405833	180002-83-9		13.0±6.9 (194.5±135.2)	11.2 (49.4)	66.6±31.3 (70.0±58.7)
Ifenprodil	23210-58-4		12.3±6.7 (80.3±65.6)	7.9 (49.7)	21.1±1.7 (25.5±0.8)
L-765,314	189349-50-6		4.0±1.7 (126.7±92.2)	4.0 (54.1)	70.5±0 (115.2±0.3)
Metergoline	17692-51-2		5.8±1.3 (170.2±18.4)	2.5 (66.8)	62.8±0 (127.9±4.1)
Methiothepin	74611-28-2		6.1±0.4 (163.2±43.0)	3.5 (76.9)	63.2±10.3 (121.8±9.0)
Mibefradil	116666-63-8		10.4±3.3 (88.9±41.4)	4.5 (37.7)	25.2±4.1 (97.0±3.3)
NAN-190	115338-32-4		2.0±0.3 (112.7±49.0)	2.2 (48.0)	Inactive
NNC 55-0396	959517-26-1		7.1±0 (91.9±8.3)	5.0 (91.6)	18.8±1.5 (106.3±4.3)
Octoclothepin	4789-68-8		6.9±0.5 (133.0±47.6)	3.2 (135.0)	62.8±0 (121.0±4.9)
Ouabain	11018-89-6		3.0±1.8 (48.3±9.7)	0.4 (53.6)	0.2±0 (43.6±1.1)
Spiperone	749-02-0		2.5±0.0 (104.8±14.6)	5.0 (79.2)	8.9±12.5 (83.5±7.3)
Vinblastine	143-67-9		9.5±5.5 (87.1±50.8)	12.6 (197.2)	Inactive

^*Known PDL inducers

Each value of potency (EC₅₀ or IC₅₀, µM) and efficacy (compared with PLD induction by 19.16 µM amiodarone as a % of positive control) are the mean ± SD of replicates from three experiments. The efficacy values are expressed in parenthesis. CASRN, Chemical Abstracts Services Registry Number.

Figure 2.

Induction of intracellular phospholipidosis (PLD) accumulation labeled with LipidTOX dye in HepG2 cells after 24 h of compound treatment. Representative images of fixed cells stained with LipidTOX (red color) and Hoechst (blue color) are from 24 compounds with the indicated concentration around EC₅₀ values; butaclamol, 4.8 µM; propranalol, 19.16 µM; 5-(N,N-hexamethylene) amiloride, 19.16 µM; amiodarone, 4.8 µM; BIX 01294, 2.39 µM; BW 723C86, 19.16 µM; CGS-12066A, 9.58 µM; cis-(Z)-flupenthixol, 4.79 µM; ebastine, 0.3 µM; eliprodil, 9.58 µM; ellipticine, 9.58 µM; GR 127935, 1.2 µM; GW405833,19.16 µM; ifenprodil, 9.58 µM; L765,314, 4.8 µM; metergoline, 4.79 µM; methiothepin, 4.79 µM; mibefradil, 9.58 µM; NAN-190, 2.4 µM; NNC 55-0396, 9.58 µM; octoclothepin, 9.58 µM; ouabain, 2.39 µM; spiperone, 2.4 µM; vinblastine, 19.16 µM.

In addition, the cytotoxicity of these 24 compounds after 24 h treatment with cells was evaluated in a cell viability assay. Of the 24 compounds, ouabain, spiperone, BIX 01294 and NNC 55-0396 showed some degree of cytotoxicity, with IC50 values of 0.2, 8.9, 10, and 18.8 µM, respectively, and maximum inhibition of cell viability of 42%, 83%, 98%, and 106%, respectively. These compounds need further confirmation for PLD induction using electron microscopy analysis. All the other compounds had either minimal or no cytotoxic effect on the cells (Table 2).

To confirm their ability to induce PLD, NAN-190, ebastine, GR127935 and cis-(Z)-flupenthixol were further examined for their effect on PLD induction by electron microscopy analysis, the gold standard method for characterization of PLD. Amiodarone was used as the positive control and DMSO treatment was used the negative control. HepG2 cells were treated with these compounds at the concentration around their EC₅₀ values. Either unicentric or multicentric multi-lamellar bodies, the hallmark of PLD, were observed in the cytoplasm after the treatment with amiodarone, NAN-190, ebastine, GR127935, or cis-(Z)-flupenthixol, but not with DMSO treatment (Fig 3). The electron microscopy results were in agreement with the data from the high content assay (LipidTOX staining) shown in Fig 4.

Figure 3.

Transmission electron micrographs of PLD induction in HepG2 cells treated with drugs for 24 h. (A) amiodarone (6 µM), (B) NAN-190 (3 µM), (C) ebastine (1 µM), (D) GR127935 (3 µM), (E) cis-(Z)-flupenthixol (6 µM), and (F) DMSO only. Insets from drug treatment show large multicentric myeloid bodies in the cytoplasm. Bar = 2 µm for main panel and bar = 500 nm for inset.

Figure 4.

Phospholipidosis induction in HepG2 cells using LipidTOX dye. After treatment with (A) amiodarone (6 µM), (B) NAN-190 (3 µM), (C) ebastine (1 µM), (D) GR127935 (3 µM), (E) cis-(Z)-flupenthixol (6 µM), and (F) DMSO only for 24 h, the cells were fixed and stained with LipidTOX and Hoechst. Images were acquired in ImageXpress microsystem using a 20× objective.

Discussion

Drug-induced PLD can lead to the impairment of lipid metabolism and phospholipid accumulation in cells, especially in lysosomes.⁶ A number of clinical and experimental drugs have been shown to induce PLD in several cell culture and animal models¹² although a direct link between PLD and tissue/organ toxicity is lacking. Drug-induced PLD is still a concern to both the U.S. FDA and the pharmaceutical industry. In order to identify drug-induced PLD in early drug development, Phospholipidosis Working Group from the FDA Center for Drug Evaluation and Research has developed the guidelines on the regulatory implications of PLD and has developed a database of PLD findings from published literature and drug review submissions.²³ Currently, electron microscopy is still the gold standard method for PLD detection, but this method is time-consuming and labor intensive making it unsuitable for large compound library screening. Measurement of drug-induced PLD has also been conducted in experimental animals, but large quantities of compounds are required, and the induction of PLD may take days to months depending on the affinity of the drugs for susceptible tissues.²⁴ In order to increase the screening throughput, several in vitro cell-based PLD assays have been developed using various fluorescent dyes or probes in order to identify phospholipids that accumulate intracellularly and induce lysosomal concentric lamellar bodies after exposure to PLD inducing agents.^13–16 Recently, Nioi et al¹⁷ has developed a cell-based assay using LipidTOX, a fluorescent lipophilic dye, in 96-well plate format, which is a highly sensitive assay for detecting drug-induced PLD in HepG2 cells. In addition, Nioi et al¹⁷ compared this cell-based assay with gene expression experiments that measured 17 genes identified as potential biomarkers of PLD using real-time PCR²⁵ and found that this cell based assay using LipidTOX was less time consuming, more sensitive, and higher throughput than gene expression analyses.

NBD-PE, NBD-PC, Nile Red and LipidTOX are commonly used PLD dyes for the cell based assays. Compared to the LipidTOX dye that can be easily dissolved in an aqueous solution, NBD-PE^{13; 14} and NBD-PC¹⁵ require an additional 30 min of sonication step after dissolving in solution. Nile Red staining requires cell fixation first, followed by several washings, and finally adding the Nile Red solution.¹⁶ In the present study, we have further evaluated the cell based PLD assay using LipidTOX red by miniaturizing it into a 1536 well plate format. The assay optimized in 1536-well plate format performed well even when the number of washing steps was reduced to a single wash. We also compared the S/B ratios from the reading of one or two sites per well and found similar results. The S/B ratios for Pit Integrated Intensity per Well were 12.6 ± 3.1 for one site per well, and 15.7 ± 9.0 for two site per well, respectively. The S/B ratios for Pit Integrated Intensity per Cell were 14.3 ± 4.2 for one site per well, and 17.5 ± 10.6 for two sites per well, respectively. Therefore, using one site reading per well and the reduction of washing steps greatly increased the speed of image acquisition and screening throughput.

From the screening, we have identified more than one hundred PLD inducers as listed in Supplemental Table S1. Some are known PLD inducers and others are first identified as PLD inducers in this study. From the LOPAC library we found that a group of tricyclic analogs including imipramine, clomipramine, trimipramine and amitriptyline used as antidepressant in the clinic were correctly identified as PLD inducers. Another group of neuroleptic drugs including chlorpromazine and promazine used as antipsychotics in the clinic were also identified as the PLD inducers. These results are consistent with previously findings.²⁶ Several other tricyclic analogs and neuroleptic drugs, such as opipramol, cyclobenzaprine, cyproheptadine, promethazine, trifluoperazine, fluphenazine, and lozapine, which have not previously reported, were also found as PLD inducers in this study.

Among the novel clusters of PLD inducers identified from this study, three interesting clusters (Supplemental Table S1) were identified as novel PLD inducers: antihistamines with the benzohydryloxy moiety (cluster 2), neural active drugs with the 1-(2-methoxyphenyl)-4-piperazine moiety (cluster 10), and the thioxanthene class of antipsychotic drugs (cluster 11). Ebastine from Cluster 2 is a H1 receptor antagonist and is used as a second-generation antihistamine agent for the treatment of allergic disorders.²⁷ Ebastine has also been reported to prolong the QT interval²⁸ by blocking the human ether-a-go-go (HERG) channel.²⁹ In cluster 10, GR127935, p-MPPF and NAN-190 are serotonin receptor 5-HT antagonists. Specifically, GR127935³⁰ acts as a selective antagonist of 5-HT1D, while p-MPPF³¹ and NAN-190³² are selective antagonists of 5-HT1A. Both cis-(Z)-flupenthixol³³ and chlorprothixene³⁴ from cluster 11 are antipsychotic drugs acting by selectively blocking dopamine receptors of the D2 subtype and the GABA_A subset of GABA receptors, respectively. More interestingly, we observed that these compounds in addition to others listed in Table 2 were also found to inhibit the hERG channel (data not shown). Compounds with charged nitrogen moieties are known to disrupt Na+/K+ channels such as the hERG channel. The same structural feature is present in CADs that are PLD inducers. This common structural feature may underlie both hERG inhibition and PLD induction. Further investigation is warranted regarding whether phospholipidosis plays a role in the toxicities of these compounds.

In summary, we have developed and validated a cell-based PLD assay in 1536-well plate format. From the primary screening of the LOPAC library, the dibenzazepine and phenothiazine class of tricyclic antidepressants, some of which are known PLD inducers (e.g. imipramine, desipramine, clomipramine, and amitriptyline), were identified. Additional compounds, such as NAN-190, ebastine, GR127935 and cis-(Z)-flupenthixol, have not been reported as PLD inducers prior to this study. Taken together, our results show that this cell-based PLD assay can be used to quickly and efficiently screen large compound collections to identify compounds that can induce PLD formation in cells.