CLN8 is an endoplasmic reticulum cargo receptor that regulates lysosome biogenesis

Abstract

Organelle biogenesis requires proper transport of proteins from their site of synthesis to their target subcellular compartment^1-3. Lysosomal enzymes are synthesized in the endoplasmic reticulum (ER) and traffic through the Golgi complex before being transferred to the endolysosomal system^4-6, but how they are transferred from the ER to the Golgi is unknown. Here, we show that ER-to-Golgi transfer of lysosomal enzymes requires CLN8, an ER-associated membrane protein whose loss of function leads to the lysosomal storage disorder, neuronal ceroid lipofuscinosis 8 (a type of Batten disease)⁷. ER-to-Golgi trafficking of CLN8 requires interaction with the COPII and COPI machineries via specific export and retrieval signals localized in the cytosolic carboxy terminus of CLN8. CLN8 deficiency leads to depletion of soluble enzymes in the lysosome, thus impairing lysosome biogenesis. Binding to lysosomal enzymes requires the second luminal loop of CLN8 and is abolished by some disease-causing mutations within this region. Our data establish an unanticipated example of an ER receptor serving the biogenesis of an organelle and indicate that impaired transport of lysosomal enzymes underlies Batten disease caused by mutations in CLN8.

Fig. 1.

CLN8 interacts with lysosomal enzymes. a, Co-expression analysis of ER and lysosomal genes. Shown is a heatmap representing the extent of pairwise co-expression between 620 ER genes (x-axis) and 60 lysosomal genes (y-axis). Among ER genes that are significantly co-expressed with lysosomal genes (P < 10⁻⁴, two-tailed Kolmogorov-Smirnov test; vertical dotted line), CLN8, TMED4, LMF1, and TMED9 encode candidate cargo receptors. b, DQ-Red BSA degradation assay to measure the proteolytic activity of lysosomes upon transfection of CLN8, TMED4, TMED9 and LMF1 plasmids. CTCF, Corrected Total Cell Fluorescence. Data are means ± SEM (n = 3 independent experiments, n = 10 independent images quantified). Scale bar: 20 μm. c, Representative live imaging of reconstituted BiFC fluorescence between Y2-tagged, full-length candidate cargo receptors and pools of Y1-tagged lysosomal enzymes. Green fluorescence shows reconstitution of YFP as an indicator of protein-protein interaction. Control experiments (CTRL) were performed by co-transfecting Y2-tagged candidates with pools of Y2-tagged lysosomal enzymes. Scale bar: 200 μm. d, Pairwise interaction between Y2-CLN8 and lysosomal enzymes evaluated by BiFC followed by flow cytometry. The non-lysosomal proteins, AGN, IGF1 and TGFB1, and Y2-tagged CLN8 were used as negative controls. Data are means ± SEM (n = 3 independent experiments, *P < 0.05, **P < 0.01, two-tailed Student’s t-test). Data are corrected for multiple comparison by using the Bayesian SGoF procedure. e, Co-IP analysis of CLN8 and lysosomal enzymes. Proteins were transiently expressed in HeLa cells, and immunoprecipitates were analyzed by immunoblotting with the indicated antibodies. HEXB was used as a negative control. Input represents 10% of the total cell extract used for IP. Images shown in c and e are representative of n = 3 independent experiments.

Fig. 2.

CLN8 deficiency leads to depletion of lysosomal enzymes. a, Plot of relative protein signals from LC-MS/MS analysis of lysosome-enriched liver fractions from WT and CLN8-deficient mice (KO). Blue dots, lysosomal soluble proteins; red dots, lysosomal membrane proteins; grey dots, other proteins detected by LC-MS/MS. Data are relative to n = 2 independent proteomic analyses conducted on pools of three livers each. b, GSEA of lysosomal soluble protein changes in CLN8-deficient mice compared with WT mice. The upper graph shows the distribution of lysosomal soluble proteins (vertical blue bars) along the LC-MS/MS-detected proteins ranked according to their enrichment in CLN8-deficient vs. WT mice (left: increased in CLN8-deficient mice, red; right: decreased in CLN8-deficient mice, blue). The lower graph is the enrichment score plot showing that lysosomal soluble proteins rank mostly amongst proteins that are depleted in the CLN8-deficient mice (Enrichment Score = −0.92; P < 10⁻⁴). c, Immunoblot analysis of lysosome-enriched fractions confirming depletion of lysosomal enzymes in CLN8-deficient mice (KO) compared to WT mice. CTSD_IP, immature precursor; CTSD_MH, mature heavy form; CTSD_ML, mature light form. Band intensities were quantified and normalized to LAMP1. d, Confocal microscopy of CTSD (red) and LAMP1 (green) on cortical and cerebellar sections of CLN8-deficient and WT mice. Pearson correlation showed a significant reduction of the CTSD/LAMP1 signal overlap in CLN8-deficient mice compared to WT mice. Inserts show four-fold magnification. e, Confocal microscopy of patient-derived CLN8-deficient skin fibroblasts. Skin fibroblasts from two healthy subjects were used as controls. Insets show four-fold magnified images. Quantification is provided in Supplementary Fig. 4b. In c data are means ± SEM (n = 3 independent experiments, *P < 0.05, **P < 0.01, two-tailed Student’s t-test). In d data are means ± SEM (n = 3 independent experiments, n = 10 independent images quantified, ***P < 0.001, two-tailed Student’s t-test). Scale bar: 20 μm.

Fig. 3.

Interaction with COPI and COPII complexes mediates CLN8 trafficking. a, Confocal microscopy showing ER-to-Golgi shift of CLN8 localization upon treatment with CBM. Trace outline is used for line-scan analysis of Relative Fluorescence Intensity (RFI) of CLN8, GM130 (Golgi marker) and KDEL signals. Signal overlap is quantified by Pearson correlation analysis of n = 3 independent experiments, n = 10 independent images quantified. b, Confocal microscopy of HeLa cells showing ER-to-Golgi shift of CLN8 localization upon KKXX signal mutagenesis (CLN8dK). c, BiFC assay of α-COP with CLN8 showing disruption of interaction upon mutagenesis of CLN8’s KKXX signal (CLN8dK). d, Co-IP assay of γ-COP and CLN8. Input represents 10% of the total cell extract. e, Confocal microscopy of BiFC complexes composed of TPP1-Y1 and Y2-CLN8 (with or without CBM) or TPP1-Y1 and Y2-CLN8dk. f, Immunoblot of lysates from HeLa cells transfected with Myc-tagged Sec24 proteins after pull down with the cytosolic C-terminus of CLN8 fused with GST showing disruption of interaction upon mutagenesis of CLN8’s COPII signal. g, Confocal microscopy of HeLa cells showing decreased Golgi localization of CLN8 upon mutagenesis of CLN8 ER export signal (CLN8WdK) using the Golgi-localizing CLN8dK backbone. In b, e and g, trace outline, RFI and Pearson correlation analyses are performed as in a. Images shown in c, d and f are representative of n = 3 independent experiments. In a, b, e and g, data are means ± SEM (n = 3 independent experiments, n = 10 independent images quantified, **P < 0.01, ***P < 0.001, two-tailed Student’s t-test). In a, b, e and g, scale bars are 20 μm. In c, scale bars are 200 μm.

Fig. 4.

Defective maturation of lysosomal enzymes upon CLN8 deficiency. a-c, Metabolic radiolabeling of CLN8^−/− cells and their parental HeLa cells showing defective maturation of CTSD and PPT1 (a), TPP1 (b) and GALNS (c) in the absence of CLN8. CTSD and PPT1 were immunoprecipitated by using antibodies against the endogenous proteins. 3xFlag-tagged TPP1 and GALNS were expressed by using a doxycycline-inducible vector used to transduce CLN8^−/− and control cells. Arrows indicate the mature, lysosome-associated enzyme. The asterisk in c indicates nonspecific signal. d, Quantification of the mature enzymes for metabolic radiolabeling experiment in a-c. e, f, Decreased enzyme stability in CLN8^−/− cells compared to parental HeLa cells. CTSD and PPT1 proteins were monitored at the indicated time points following cycloheximide-mediated blockage of protein synthesis. 3xFlag-tagged TPP1 and GALNS were expressed in cells transduced with a doxycycline-inducible vector and monitored at 0, 4, 8, and 12 hrs upon doxycycline removal to turn off their synthesis. In all experiments, GAPDH was used to normalize the residual protein for quantifications. g, Quantification of the immunoblot experiments in e and f. In d and g, data are means ± SEM (n = 3 independent experiments, *P < 0.05, **P < 0.01, two-tailed Student’s t-test).

Fig. 5.

CLN8 interaction with lysosomal enzymes requires the second luminal loop. a, Schematic representation of CLN8 protein. b, Schematic representation of CLN8ΔL construct. Amino acids are represented by colored circles. Amino acids in yellow indicate the position of clinical mutations. The ER export signal and the ER retrieval signal at the protein C-terminus are indicated. c, Shown is a multi-alignment of CLN8 protein sequences along with a plot of local evolutionary rates. The red lines at the top mark the detected evolutionary constrained regions (ECRs). Transmembrane domains (TM, purple lines), luminal domains (ER, orange lines) and cytosolic domains (Cyt, green lines) are reported. Clinical mutations that fall in the second luminal loop are reported. d, Confocal microscopy analysis showing that Y2-CLN8ΔL co-localizes with both CLN8-Myc and the ER marker, KDEL. Scale bar: 20 μm. e, Confocal microscopy analysis showing that mutagenesis of the KKXX signal of the CLN8ΔL-Y protein determines its localization to the Golgi, indicating that the second luminal loop of CLN8 is not required for CLN8 ER export. f, BiFC assay of Y1-CLN8 with Y2-CLN8ΔL showing a reconstituted GFP signal indistinguishable from that of Y1-CLN8 with Y2-CLN8. g, Comparative BiFC assay of Y1-tagged lysosomal enzymes with either Y2-CLN8 (top panels) or Y2-CLN8ΔL (bottom panels) showing disruption of interaction upon removal of CLN8 second luminal loop. h, Comparative BiFC/flow cytometry analysis of Y1-tagged lysosomal enzymes with Y2-CLN8 or Y2-CLN8ΔL constructs. Values are expressed as a percentage of the enzyme-Y1/Y2-CLN8 interaction. i, j, Comparative BiFC/flow cytometry assay of CLN8 constructs harboring clinical mutations in the second luminal loop. Images shown in d, e, f and g are representative of n = 3 independent experiments. In h, i and j, data are means ± SEM (n = 3 independent experiments, **P < 0.01, two-tailed Student’s t-test). In i and j, the group-level p-values were estimated from the mean z-scores from each individual test. In d and e, scale bars are 20 μm. In f and g, scale bars are 200 μm.