Depletion of LONP2 unmasks differential requirements for peroxisomal function between cell types and in cholesterol metabolism

Abstract

Peroxisomes play a central role in tuning metabolic and signaling programs in a tissue- and cell-type-specific manner. However, the mechanisms by which the status of peroxisomes is communicated and integrated into cellular signaling pathways are not yet understood. Herein, we report the cellular responses to peroxisomal proteotoxic stress upon silencing the peroxisomal protease/chaperone LONP2. Depletion of LONP2 triggered the accumulation of its substrate TYSND1 protease, while the overall expression of peroxisomal proteins, as well as TYSND1-dependent ACOX1 processing appeared normal, reflecting early stages of peroxisomal proteotoxic stress. Consequently, the alteration of peroxisome size and numbers, and luminal protein import failure was coupled with induction of cell-specific cellular stress responses. Specific to COS-7 cells was a strong activation of the integrated stress response (ISR) and upregulation of ribosomal biogenesis gene expression levels. Common changes between COS-7 and U2OS cell lines included repression of the retinoic acid signaling pathway and upregulation of sphingolipids. Cholesterol accumulated in the endomembrane compartments in both cell lines, consistent with evidence that peroxisomes are required for cholesterol flux out of late endosomes. These unexpected consequences of peroxisomal stress provide an important insight into our understanding of the tissue-specific responses seen in peroxisomal disorders.

Fig. 1

LONP2 knockdown triggered the accumulation of its substrate TYSND1, alterations in peroxisome size and numbers, and luminal protein import failure. a–d Immunoblotting (a, c) and representative confocal images and quantification of peroxisome size and numbers (b, d) in non-targeting (NT)- and LONP2-silenced COS-7 cells (a, b) and U2OS cells (c, d) after 144 h. TYSND1-F and TYSND1-C indicate full-length TYSND1 and the self-cleaved C-terminal region of TYSND1, respectively. Scale bar, 5 µm. e, f Representative confocal images (e) and quantification (f) of the colocalization between peroxisomes (PMP70) and transiently expressed peroxisome luminal protein (CFP-SKL) in NT- and LONP2-silenced COS-7 cells after 144 h. Scale bar, 2 µm. g Representative confocal images showing the colocalization between peroxisomes (PMP70) and transiently expressed peroxisome membrane protein (PEX3-YFP) in NT- and LONP2-silenced COS-7 cells after 144 h. Scale bar, 5 µm. h Immunoblotting for autophagy-related proteins, vinculin was used as a loading control. i Representative confocal images of COS-7 cells grown in HBSS for 24 h (left) and quantification (right) showing the number of LC3-P62 puncta on peroxisomes of NT- and LONP2-silenced COS-7 cells and U2OS cells after 144 h. Cells were grown in either DMEM or HBSS for 24 h and LC3-P62 puncta on peroxisome was visualized with transiently expressed GFP-LC3 and the immunofluorescence of p62 and PMP70. Scale bar, 2 µm. j Representative confocal images and quantification showing the size and mean branch length of mitochondria in NT- and LONP2-silenced COS-7 cells after 144 h. Scale bar, 5 µm. On all plots, dots represent individual cells from n = 3 technically independent experiments depicted in different colours, and triangles represent the mean within an independent experiment. The mean values were used to calculate the average (horizontal bar), s.d. (error bars), and P-values (using a two-tailed unpaired Student’s t-test)

Fig. 2

Cell-specific ISR activation upon silencing of LONP2. RNA sequencing (n = 3 independent experiments per condition), LONP2 silencing for 7 days. a Principal component (PC) score plot. b Volcano plots showing all identified transcripts. c Transcripts encoding integrated stress response (ISR) factors. d Immunoblotting for ISR-related proteins, vinculin was used as a loading control. e Translation assay: cells were incubated with puromycin for 10 min., and incorporation of puromycin into newly synthesized proteins was revealed with anti-puromycin antibodies. GAPDH was used as a loading control. f Transcripts encoding ribosomal biogenesis-related rRNA, snoRNA and mRNA genes. g Immunoblot of ribosomal biogenesis regulator RRS1 in NT- and LONP2-silenced COS-7 cells and U2OS cells

Fig. 3

LONP2 silencing leads to impaired retinoic acid signaling. a Gene ontology of RNA sequencing data (dataset is described in Fig. 2). Numbers on Venn diagram indicate the number of genes in each category. b Immunoblotting of retinoic acid regulating proteins, vinculin used as a loading control. c COS-7 or U2OS cells were incubated in the presence or absence of 5 µM retinoic acid for 24 h and mRNA transcript levels for LONP2 (left two panels) and TGM2 (right two panels) quantified with qRT-PCR. n = 3 independent experiments, means are depicted in different colours. The mean values were used to calculate the average (horizontal bar), s.d. (error bars) and P-values (using a two-tailed unpaired Student’s t-test)

Fig. 4

Lipid profiles of COS-7 and U2OS cells differ drastically and are differentially affected by LONP2 silencing. Untargeted lipidomic analysis of n = 5 independent cell experiments for each group under the basal condition of after LONP2-silencing for which the final dataset included 2,094 lipid features (Table S2). a Principal component (PC) score plot. b Log₂ (fold changes) of 206 lipid features annotated by MS/MS and data alignment using our in-house database, and grouped by lipid subclass, whereby red and black dots indicate lipids that reach or not our selected significance threshold for the indicated comparison, P_corr-value < 0.1. c Box plots of unique lipids annotated by MS/MS from the 100 most significant lipid features discriminating LONP2-silenced COS-7 cells (top) and U2OS cells (bottom) from NT-silenced cells. The midline represents the median fold change vs. NT-silenced cells, the box represents the interquartile range between the first and third quartile, and the whiskers represent the lowest or highest values. d Immunoblotting of cholesterol sensor INSIG1, vinculin used as a loading control. Abbreviations: sphingolipid (SP), sterol lipid (ST), glycerolipid (GL), glycerophospholipid (GP), fatty acyls (FA), sphingomyelin (SM), ceramide (Cer), glucosylceramide (GlcCer), cholesterol (Chol), cholesterol ester (CE), triacylglycerol (TG), diacylglycerol (DG), 1-alkyl, 2-acylglycerophosphocholine (PCO-), 1-(1Z-alkenyl), 2-acylglycerophosphocholine (PCP-), 1-alkyl, 2-acylglycerophosphoethanolamine (PEO-), 1-(1Z-alkenyl), 2-acylglycerophosphoethanolamine (PEP-), diacylglycerophosphocholine (PC), diacylglycerophosphoethanolamine (PE), diacylglycerophosphoinositol (PI), monoacylglycerophosphocholine (LPC), monoacylglycerophosphoethanolamine (LPE), diacylglycerophosphoglycerol (PG), free fatty acid (FFA) and acylcarnitine (CAR). All lipid features were categorized into subclasses and labeled based on the LIPID MAPS® Structure Database

Fig. 5

LONP2 knockdown induced lipid storage, including aberrant cholesterol accumulation in lysosomes. a Representative confocal images (left) and quantification (right) of lipid droplets (Nile Red) in NT- and LONP2-silenced COS-7 cells after 6 days. Scale bar, 5 µm. b Representative confocal images showing the Filipin-stained intracellular cholesterol (green), Nile Red-stained lipid droplets (magenta) and endolysosomes (blue) visualized with the immunofluorescence of LAMP1. Scale bar, 2 µm. c, d Free cholesterol overload experiment in NT- and LONP2-silenced COS-7 cells (c) and U2OS cells (d) after 7 days. Representative confocal images (left) and quantifications (right) of the colocalization between cholesterol (Filipin) and endolysosomes (LAMP1). Scale bar, 2 µm. e Transmission electron microscopy micrographs of NT- and LONP2-silenced COS-7 cells. Scale bar, 500 nm. f Immunoblotting (left) and image quantification (right) of cholesterol overload experiment in NT- and LONP2-silenced COS-7 cells rescued with transient expression of siRNA-resistant LONP2 cDNA for 48 h. Data from n = 3 technically independent experiments, and their respective means are depicted in different colours. The mean values were used to calculate the average (horizontal bar), s.d. (error bars) and P-values (using a two-tailed unpaired Student’s t-test)