The Hsp40 co-chaperone DNAJC7 modifies polyglutamine but not polyglycine aggregation

Abstract

Polyglutamine (polyQ) diseases, including Huntington's disease and several spinocerebellar ataxias, are caused by abnormally expanded CAG nucleotide repeats, which encode aggregation-prone polyQ tracts. Substantial prior evidence supports a pathogenic role for polyQ protein misfolding and aggregation, with molecular chaperones showing promise in suppressing disease phenotypes in cellular and animal models. In this study, we developed a FRET-based reporter system that models polyQ aggregation in human cells and used it to perform a high-throughput CRISPR interference screen targeting all known molecular chaperones. This screen identified as a strong suppressor of polyQ aggregation the Hsp40 co-chaperone DNAJC7, which has previously been shown to modify aggregation of other disease proteins (tau and TDP-43) and has mutations causative for amyotrophic lateral sclerosis. We validated this phenotype and further established a physical interaction between DNAJC7 and polyQ-expanded protein. In contrast, DNAJC7 did not modify aggregation of polyglycine (polyG) in a FRET-based model of neuronal intranuclear inclusion disease. In addition to establishing new inducible, scalable cellular models for polyQ and polyG aggregation, this work expands the role of DNAJC7 in regulating folding of disease-associated proteins.

Fig. 1.. An inducible cell-based model of nuclear, detergent-resistant, p62-positive polyglutamine (polyQ) protein aggregates monitored by a FRET-based reporter.

a, Schematic of lentiviral constructs for doxycycline-inducible expression of nuclear-localized fluorescent proteins fused to a C-terminal ataxin-3 with 79 glutamines. HEK293T cells engineered with CRISPRi machinery were transduced with these constructs, and clones expressing both fluorescent proteins were selected to generate the NLS-FRET-Q79 cell line. b, Fluorescence imaging of NLS-FRET-Q79 cells at one day and five days of doxycycline c, Immunofluorescence staining for endogenous p62 in NLS-FRET-Q79 cells after 5 days of doxycycline. d, Fluorescence imaging of NLS-FRET-Q79 cells at 5 days of doxycycline before and after treatment with detergent, in the same field of view. e, Schematic illustrating how polyQ aggregation gives rise to a “FRET-high” population observable by flow cytometry. f, Flow cytometry plots at one day or five days of doxycycline, before and after treatment with detergent. All scale bars are 10 μm.

Fig. 2.. Exogenous polyQ proteins seed aggregation of the polyQ FRET reporter.

a, Workflow for generating homogenates from cultured cells or mouse cortical tissue, followed by transfection into the NLS-FRET-Q79 reporter cell line. Created in Biorender. b-d, Flow cytometry plots showing fraction of FRET-high cells after transfection with homogenates derived from NLS-FRET-Q79 cells (b) and cortical homogenates from Huntington’s disease (HD) transgenic mice versus non-transgenic controls (c), with quantification of technical triplicates (d); bars and error bars represent mean ± standard deviation (sd).

Fig. 3.. A CRISPRi screen of molecular chaperones identifies DNAJC7 as a modifier of polyQ aggregation.

a, Workflow of CRISPRi screen to identify molecular chaperone modifiers of polyQ aggregation in NLS-FRET-Q79 cells. Created with BioRender. b, Volcano plot CRISPRi screen results, highlighting DNAJ protein and proteasomal subunit hits. c, Flow cytometry plots of NLS-FRET-Q79 cells at 4 days of doxycycline with or without proteasomal inhibitor (carfilzomib) d-e, Western blot (d) and quantification (e) of DNAJC7 levels in NLS-FRET-Q79 cells expressing non-targeting (NTC) sgRNAs or sgRNAs targeting DNAJC7; each lane is lysates from separate wells. f, Results of flow cytometry measuring fold-change in the fraction of FRET-high cells after 5 days of doxycycline in sgRNA⁺ cells transduced with NTC or DNAJC7. Bars and error bars represent mean ± sd for n = 3 independent experiments.

Fig. 4.. DNAJC7 physically interacts with mutant HTT exon 1 and suppresses its aggregation.

a, Schematic of a lentiviral construct encoding a doxycycline-inducible eGFP-tagged fragment of Huntingtin exon 1 containing 72 glutamines (GFP-HTTex1-Q72), used to generate a monoclonal HEK293T CRISPRi cell line. b, Fluorescence micrograph of GFP-HTTex1-Q72 cells at 7 days of doxycycline treatment before and after treatment with Triton X-100. c, Results of flow cytometry experiments measuring increases the fraction of total events containing detergent-resistant GFP⁺ cells comparing those stably transduced with sgRNAs that are NTCs or targeting DNAJC7. Data represent mean ± sd from n = 4 independent experiments. d, Fluorescence micrographs of HEK293T cells 48h after transient co-transfected GFP-HTTex1-Q72 and either mTagBFP2 (BFP) alone or BFP-tagged DNAJC7. Arrows indicate cytoplasmic aggregates. e, Flow cytometry plotting GFP-height versus GFP-Width (pulse shape analysis) of HEK293T cells transfected with GFP-HTTex1-Q72 for 48h high. The detergent-resistant population of cells are designated as aggregate-positive (Agg⁺). f, Results of flow cytometry experiments measuring Agg⁺ cells in HEK293T cells 48h after transient co-transfection with GFP-HTTex1-Q72 and either BFP or BFP-DNAJC7. Bars and error bars represent mean ± sd from n = 4 independent biological replicates (indicated by color), with each performed in technical triplicate. g, Western blot of input lysates and α-FLAG immunoprecipitated lysates from HEK293T cells transfected with 3×FLAG-tagged HTTex1 with either 25 or 72 glutamines and immunoblotted for FLAG, DNAJC7, and GAPDH. HMW: High molecular weight. IgG-L: Immunoglobulin light-chain. All scale bars are 10 μm.

Fig. 5.. An inducible model of polyglycine (polyG) aggregation reveals detergent-resistant, p62-positive nuclear inclusions and seeding activity.

a, Lentiviral constructs for generating NLS-FRET-G100 cell line expressing the upstream open reading frame (uORF) of the NOTCH2NLC gene with polyglycine tract of 100 residues. b, Immunofluorescence staining for p62 of NLS-FRET-G100 cells at 5 days of doxycycline. c, Fluorescence imaging of NLS-FRET-Q79 cells at 5 days of doxycycline before and after detergent treatment, in the same field of view. d, Flow cytometry plots at one and five days of doxycycline treatment, and the latter before and after detergent treatment. e, Flow cytometry plots of NLS-FRET-Q79 cells at 4 days of doxycycline with or without proteasomal inhibitor (carfilzomib). f, Flow cytometry results of NLS-FRET-G100 cells at 3 days of doxycycline with and without transfection with NLS-FRET-G100 cell homogenates (collected five days after doxycycline) for 48h. All scale bars are 10 μm.

Fig. 6.. DNAJC7 partially physically interacts with polyG but does not alter its aggregation.

a, Volcano plot showing results of a CRISPRi screen for molecular chaperone modifiers of polyG aggregation in the NLS-FRET-G100 cell line. b, Pairwise comparison of Gene Scores between the polyG screen (from panel a) and the polyQ screen (Fig. 3b). c, Results of flow cytometry experiments measuring fold-change in the fraction of FRET-high cells after 5 days of doxycycline in sgRNA⁺ cells transduced with NTC or DNAJC7. Data represent mean ± sd from n = 3 independent experiments. d, DNAJC7 co-immunoprecipitates with 3xFLAG-tagged G100 to a greater extent than with untransfected lysates or HTTex1-Q25, but to a lesser extent than with HTTex1-Q72, in transiently transfected HEK293T cells. HMW: High molecular weight. IgG-L: Immunoglobulin light-chain e, Quantification of DNAJC7 co-immunoprecipitation from n = 4 independent experiments. Dotted lines connect paired values for Q72 and G100 from the same immunoblot. Bars and error bars represent mean ± sd. *p<0.05 by paired t-test.