Differential roles for DNAJ isoforms in HTT-polyQ and FUS aggregation modulation revealed by chaperone screens

Protein aggregation is a hallmark of neurodegeneration. Here, we find that Huntington’s disease-related HTT-polyQ aggregation induces a cellular proteotoxic stress response, while ALS-related mutant FUS (mutFUS) aggregation leads to deteriorated proteostasis. Further exploring chaperone function as potential modifiers of pathological aggregation in these contexts, we reveal divergent effects of naturally-occurring chaperone isoforms on different aggregate types. We identify a complex of the full-length (FL) DNAJB14 and DNAJB12, that substantially protects from mutFUS aggregation, in an HSP70-dependent manner. Their naturally-occurring short isoforms, however, do not form a complex, and lose their ability to preclude mutFUS aggregation. In contrast, DNAJB12-short alleviates, while DNAJB12-FL aggravates, HTT-polyQ aggregation. DNAJB14-FL expression increases the mobility of mutFUS aggregates, and restores the deteriorated proteostasis in mutFUS aggregate-containing cells and primary neurons. Our results highlight a maladaptive cellular response to pathological aggregation, and reveal a layer of chaperone network complexity conferred by DNAJ isoforms, in regulation of different aggregate types.


Supplementary Figure 3: HTT-polyQ aggregation modulation screen for chaperone modifiers.
(a-b) DsRed co-expressing cells show lower HTT-134Q-GFP aggregation baseline compared to the coexpression of luciferase (a) or HTT-134Q alone (b), and were thus chosen as a baseline control for the coexpression screen. Data are presented as mean values +/-SEM for n=4/3 for DsRed/luciferase respectively. (a) and n=2 (b) biologically independent samples. Source data are provided as a Source Data file. (c) PulSA analysis of HEK293T cells, 48h after transfection with HTT-17Q-GFP or HTT-134Q-GFP. A distinctive group of cells containing HTT-134Q-GFP aggregated (AGG+) is shown on the left, and is missing in cells expressing the wild-type protein (HTT-17Q-GFP). AGG+ cell population was around 34% in control DsRed co-expressing cells and reduced to about 13% when DNAJB8 was co-expressed. In contrast, coexpression of HSP90AB1 increased AGG+ fraction to about 43%. (d) Fluorescence microscopy images of HEK293T cells 48h after co-expression of HTT-134Q-GFP with either DsRed control, DNAJB8 or HSP90AB1. DNAJB8 aggregation alleviation is qualitatively observed compared to DsRed control or HSP90AB1, which aggravated the aggregation phenotype. Shown are representative fields from n=3 biologically independent experiments. (e) HTT-134Q Aggregation modulation scores (x-axis) are not explained by the degree of chaperone overexpression (y-axis), as measured using sandwich ELISA assay (log2), in HEK293T cells transfected with HTT-134Q-GFP for 48h. Correlation p-value is indicated. (f) HTT-134Q Aggregation modulation scores (x-axis) do not correlate with basal chaperone expression levels (y-axis), as measured by RNA-seq from HEK293T cells expressing HTT-134Q-GFP for 48h, which have undergone sorting for aggregate containing cells (AGG+). Expression levels represented by log10 TPM values. Correlation p-value is indicated. (g) HTT-134Q Aggregation modulation scores (x-axis) do not correlate with fold changes of chaperones following HTT-134Q aggregation. Fold changes (presented as log2, y-axis) were calculated based on RNA-seq data TPM values as in (f), which were divided by the RNA-seq TPM values of the chaperones in the HTT-134Q-GFP non-aggregating cells (denoted AGG-). Correlation p-value is indicated.

Supplementary Figure 4: HTT-134Q aggregation is modulated by different chaperones
(a-d) Confocal microscopy images of HEK293T cells following 48h transfection with HTT-134Q-GFP or HTT-17Q-GFP (green) co-expressed with FLAG tagged constructs (FLAG in pink), of (a) DsRed control (b) DNAJB8 (c) HSPB7 (d) HSP90AB1. DAPI (blue) marks nuclei. Shown are representative fields from n=2 biologically independent experiments. (e) Immunofluorescence imaging of HTT-134Q-GFP or HTT-17Q-GFP (green) together with FLAG-tagged DsRed, DNAJB12-FL or DNAJB12-short (FLAG in pink). Aggregates are marked with white arrows. Shown are representative fields from n=2 biologically independent experiments. Figure 5: mutFUS aggregation modulation screen reveals chaperone modifiers (a) DsRed plasmid co-expression together with FUS-R521H-YFP showed lower aggregation baseline compared to luciferase (left) or FUS-R521H alone (right) and was therefore chosen as a baseline control for the co-expression screen. Data are presented as mean values +/-SEM for n=6/8 (DsRed/luciferase, left panel) and n=6 (DsRed/FUS-R521H alone, right panel) biologically independent samples. Source data are provided as a Source Data file. (b) Rescue of FUS-R521H-YFP aggregation by different chaperones is robust to choice of FACS parameter settings. We applied different gates for identification of aggregate positive cell populations -three additional gates are illustrated for DsRed (top row) and DNAJB14-FL expressing cells (middle). We changed the gate settings for all chaperones listed in the table below, which significantly rescued aggregation, as well as for the DsRed controls, and re-calculated the Aggregation modulation score, which is presented in the table (bottom), average of N=4 replicates. All scores remain significant. (c) FUS-R521H-YFP Aggregation modulation scores (x-axis) are not explained by the degree of chaperone overexpression (y-axis, log2), as measured using sandwich ELISA assays, in HEK293T cells transfected with FUS-R521H-YFP for 48h. Correlation p-value is indicated. (d) FUS-R521H-YFP Aggregation modulation scores (x-axis) do not correlate with basal chaperone expression levels (y-axis), as measured by RNA-seq from HEK293T cells expressing FUS-R521H-YFP for 48h, which have undergone sorting for aggregate containing cells (AGG+). Expression levels represented as log10 TPM values. Correlation p-value is indicated. (e) FUS-R521H-YFP Aggregation modulation scores (x-axis) do not correlate with fold changes of chaperones following FUS-R521H-YFP aggregation. Fold changes (presented as log2, y-axis) were calculated based on RNA-seq data TPM values as in (d), which were divided by the RNA-seq TPM values of the chaperones in the FUS-R521H-YFP non-aggregating cells (denoted AGG-). Correlation p-value is indicated. (f) Aggregation modulation score for HTT-134Q plotted against those of FUS-R521H, dashed lines represent the 95% confidence intervals corresponding to each of the aggregating proteins. Correlation between the scores is minimal and rescue chaperones do not overlap. (g-h) Cell viability percentages of top significant chaperones, assayed by DAPI exclusion using FACS. Data are presented as mean values +/-SEM of n=4 (for DNAJC5, DNAJB12-FL DNAJC5B, DNAJC30) and n=6 (for DNAJB14-FL, BAG3, HSF1) biologically independent samples in FUS-WT (g) and n=4 (for DNAJC5, DNAJC5B, BAG3, HSF1, DNAJC30,DNAJB5,HSP90AA1) n=3 (for DNAJB12-FL) and n=14 (DNAJB14-FL) biologically independent samples in FUS-R521H (h) expressing cells. Source data are provided as a Source Data file. (i) Fluorescence microscopy images of HEK293T cells 48h after co-expression of FUS-R521H-YFP with either DsRed, DNAJB14-FL, BAG3 or HSF1. Shown are representative fields from n=3 biologically independent experiments. White arrows mark mutFUS aggregates. (j) Microscopy images of HEK293T cells expressing FUS-R518K-YFP co-transfected with DsRed or DNAJB14-FL respectively. Shown are representative fields from n=2 biologically independent experiments. (k) Aggregation modulation scores for FUS-R518K-YFP co-transfected with DNAJB14-FL, BAG3 or HSF1. Data are presented as mean values +/-SEM n=7/3/3 biologically independent experiments for DNAJB14-FL/BAG3/HSF1 respectively. Dashed lines represent 95% CI as in Fig. 3a. *** -p<0.003, empirical p-value (see Methods). Source data are provided as a Source Data file.

Supplementary Figure 6: DNAJB14 and DNAJB12 full length isoforms interactions and localization
(a) Aggregation modulation scores for FUS-R521H-YFP cells co-transfected with non-tagged DNAJB14-FL or DNAJB14-short. Data are presented as mean values +/-SEM of n=4 and n=7 biologically independent samples (for DNAJB14-short and -FL respectively). Dashed lines represent 95% CI. *** -p<0.003, empirical p-value (see Methods). Source data are provided as a Source Data file. (b) Confocal microscopy of HEK293T cells co-transfected with DNAJB14-FL or DNAJB14-short (pink) together with FUS-R521H or FUS-WT (green). Some DJANGO structures are indicated by yellow arrows. No DJANGOs are observed in the short isoform. Shown are representative fields from n=4 biologically independent experiments. (c) Image analysis quantification recapitulated the aggregation modulation scores of DNAJB14-FL vs. DNAJB14-short (see Methods). Images were taken using confocal microscopy. Data are presented as mean values +/-SEM of n=2 biologically independent samples of DNAJB14-FL (with 72 and 75 cells), n=3 biologically independent samples of DNAJB14-short (with 128,133 and 38 cells), normalized to n=2 biologically independent samples of DsRed controls (with 90 and 126 cells). Source data are provided as a Source Data file. (d) Aggregation modulation scores for the different FUS mutants (as shown) co-transfected with either DNAJB14-FL or DNAJB14-Short. Data are presented as mean values +/-SEM of n=2 (for R495EfsX527+DNAJB14-short), n=4 (for R495EfsX527+DNAJB14-FL), and n=7 (for R518K+DNAJB14-FL) and n=6 (for all the rest) biologically independent samples for . Dashed lines represent 95% CI. *** -p<0.003, empirical p-value (see Methods). Source data are provided as a Source Data file. (e) Signal intensity quantification of DNAJB14 isoforms binding to FUS-R521H-YFP, each normalized to the input FLAG signal. Data are presented as mean values +/-SEM of n=5 experiments for FL and short, n=4 for Δshort and IgG. Densitometry was performed using Fiji. Source data are provided as a Source Data file. (f-g) Co-IP (f) showed that DNAJB14-FL interacted with FUS-WT-YFP while DNAJB14-short showed a minimal level of interaction. (g) As in (e), for FUS-WT-YFP, data are presented as mean values +/-SEM of n=5 biologically independent experiments. Source data are provided as a Source Data file. (h) Live cell fluorescence microscopy imaging using automated boxed microscope (see Methods). Cells were co-transfected with FUS-R521H-YFP and DNAJB14-FL or DNAJB14-short respectively, then were imaged 24h post transfection until 54hrs post transfection. Times are indicated. In the presence of DNAJB14-FL, aggregates are generated in lower rates, and rarely disappear. Shown are representative fields from n=2 biologically independent movies. (i) ER localization of DNAJB14 (bottom) and DNAJB12 (top), stained with anti-FLAG (pink) -shown via colocalization with the ER marker mCherry-ER-3 (green), containing the signal peptide of CALR (see Methods) in immunofluorescence imaging. Shown are representative cells from n=3 biologically independent samples. (j-k) Same as in (b)  (a) Co-IP of cells co-transfected with either WT or mutant FUS with three different isoforms of FLAG tagged DNAJB14 (FL, short, HPDmut1). Interaction between exogenous DNAJB14 and the endogenous HSP70 is shown in the upper panel. DNAJB14 signal was assayed using anti-FLAG antibody WB in the lower panel. HC-Heavy chain. Source data are provided as a Source Data file. (b) Signal intensity quantification of HSP70 bound to the DNAJB14 different isoforms, all normalized to FLAG signal. Data are presented as mean values +/-SEM of n=3 biologically independent samples shown in panel a. Densitometry was performed using Fiji. Source data are provided as a Source Data file. (c) Percentage of DJANGOs containing cells was calculated using image analysis of IF images, as in Fig. S6m, showing the same data but with additional mutants. Data are presented as mean values +/-SEM for n=2/3/2/3/2 (FUS-R521H+DNAJB14-FL/-short/-HPDmut1/-ΔDUF/DsRed respectively), n=3/3/2/2/3 (FUS-WT+DNAJB14-FL/-short/-HPDmut1/-ΔDUF/DsRed respectively) biologically independent samples. Source data are provided as a Source Data file. (d-e) As in Fig. 4c, for DNAJB14-HPDmut2 with FUS-R521H-YFP, and both HPD mutants together with WT-FUS (d), as well as DNAJB14 ΔDUF with WT-FUS(e). Shown are representative fields from n=4/2 biologically independent experiments (for d and e respectively).

Supplementary Figure 8: DNAJB14 -DNAJB12 complex inter-dependence in protection from mutant FUS aggregation
(a) Co-IP of cells co-transfected with either FUS-WT or FUS-R521H together with FLAG tagged DNAJB14 isoforms (FL, short, or ΔDUF). DNAJB14 isoforms were pulled down using anti-FLAG antibody. Interaction between HSP70 and DNAJB14 isoforms was assayed using WB for endogenous HSP70 (upper panel). DNAJB14 signal was assayed by anti-FLAG antibody WB for the different isoforms (lower panel, LC-Light chain) see arrows: Membrane containing DNAJB14-FL-used with secondary antibody LC fragment specific. Membrane containing DNAJB14-short and ΔDUF -used with secondary antibody FC fragment specific. Source data are provided as a Source Data file. (b) Quantification HSP70 interactions, HSP70 bands divided by FLAG band intensities, was performed using Fiji. Data are presented as mean values +/-SEM of n=2 biologically independent samples from panel a. Source data are provided as a Source Data file. (c) Fold changes in mRNA levels of the DNAJB12-FL gene normalized to the GAPDH were assayed using qPCR for DNAJB12 knockdown samples (siDNAJB12) and control samples (siControl). Data are presented as mean values +/-SEM of n=3 biologically independent experiments, each standardized to the siControl sample. Source data are provided as a Source Data file. (d) Aggregation modulation scores for HEK293T cells transfected with FUS-R521H-YFP and DNAJB12-FL, DNAJB14-FL or both. The mutual overexpression of both DNAJB12-FL and DNAJB14-FL shows a similar extent of aggregation protection as that of DNAJB12-FL alone. Data represent mean +/-SEM for n=4/4/3 (for DNAJB12+DNAJB14/DNAJB12/DNAJB14 respectively) biologically independent samples. Dashed lines represent 95% CIs as in Fig. 3a. *** -p<0.003, empirical p-value (see Methods).

Supplementary Figure 9: DNAJB14-FL restores deteriorated proteostasis caused by mutFUS aggregation
(a) Scatter plots show high correlation between the two RNA-seq experiments, of mRNA expression in FUS-WT or mutFUS expressing cells (either aggregate-containing, AGG+, or diffused, AGG-) either expressed alone (as in Fig. 1), or with DsRed co-expression (Fig. 6). (b) Expression of genes belonging to the HSP70 family of chaperones 3 was significantly repressed in cells containing mutFUS aggregates co-expressing DsRed (AGG+, solid blue line), while no such repression was seen in cells with diffused expression of mutFUS (AGG-, dashed blue line). Plots show the cumulative distribution function (CDF, y-axis) of LFC values belonging to the HSP70 chaperone family (using two definitions, see Methods), compared to the background CDF of all expressed mRNAs (background, in grey lines). P-value for the differences of each sample compared to the background distribution was calculated using t-test (twosided), and is indicated in the legend when significant. (c) Clustergram of the correlation matrix of z-score normalized mean TPM values of all differentially expressed genes (DEGs) in the experiment, shows very high correlation between cells co-expressing DNAJB14-short and cells co-expressing DsRed. Data are presented as mean values +/-SEM for n=2 biologically independent experiments. Source data are provided as a Source Data file. (m) HEK293T cells co-transfected with FUS-R521H-YFP and DsRed, and then treated with different heat stress regimens; Heat shock of 44C for 1.5h with additional 18h recovery at 37C, 44C for 2h, or 42C for 8h, followed by PulSA analysis at 48h post transfection and right after the treatment. All three regimens showed no aggregation amelioration. Data are presented as mean values +/-SEM for n=6 biologically independent samples for the recovery experiment, and n=5 for 2h at 44C and 8h at 42C. Source data are provided as a Source Data file.