DDI2 protease controls embryonic development and inflammation via TCF11/NRF1

Abstract

DDI2 is an aspartic protease that cleaves polyubiquitinated substrates. Upon proteotoxic stress, DDI2 activates the transcription factor TCF11/NRF1 (NFE2L1), crucial for maintaining proteostasis in mammalian cells, enabling the expression of rescue factors, including proteasome subunits. Here, we describe the consequences of DDI2 ablation in vivo and in cells. DDI2 knock-out (KO) in mice caused embryonic lethality at E12.5 with severe developmental failure. Molecular characterization of embryos showed insufficient proteasome expression with proteotoxic stress, accumulation of high molecular weight ubiquitin conjugates and induction of the unfolded protein response (UPR) and cell death pathways. In DDI2 surrogate KO cells, proteotoxic stress activated the integrated stress response (ISR) and induced a type I interferon (IFN) signature and IFN-induced proliferative signaling, possibly ensuring survival. These results indicate an important role for DDI2 in the cell-tissue proteostasis network and in maintaining a balanced immune response.

Keywords: Biological sciences; Developmental biology; Immune respons.

Graphical abstract

Figure 1

DDI2 dysfunction in mice results in mid-gestation embryonic lethality (A) Scheme of Ddi2 gene alternations: left—Ddi2^KO carries a lacZ reporter gene-tagged allele lacking the critical exon 2 representing complete knock-out model, right—Ddi2^ex6 carries deletion of exon 6 generated by TALEN-mediated excision resulting in protease inactivation. Immunoblot of DDI2 protein expression in E10.5 embryo lysates: left—Ddi2^KO; right—Ddi2^ex6 (lane 1 and 2—DDI2^WT and DDI2^ex6 recombinant proteins). β-actin used as a loading control. (B) Lethality screening of Ddi2^KO and Ddi2^ex6 mouse strains, the plot shows percentage of living homozygous embryos. Ddi2^ex6−/- embryos exhibit a narrower window of lethality compared to Ddi2^−/− embryos. Detailed distributions of genotypes per stage and per group of living and dead embryos are shown in Figures S1A and S1B. (C) No differences were observed between Ddi2^ex6−/- (top) and Ddi2^ex6+/+ (bottom) littermate embryos at developmental stage E9.5. The number of somites is stated in each image. (D) Phenotyping of stage E10.5 embryos. Comparison of μCT scans of Ddi2^ex6−/- (left), Ddi2^−/− (middle), and control C57BL/6NCrl (right) embryos using surface rendering representation. The 2D images of the 3D μCT scans are shown from both sides. (E) Phenotyping of stage E11.5 embryos. Whole mount image of Ddi2^ex6−/- embryo (top left) compared to its wild-type littermate (bottom left). Surface rendering of Ddi2^−/− and C57BL/6NCrl control embryo μCT scans (middle and right, respectively). The 2D images of the 3D μCT scans are shown from both sides. WTC—wild-type control. (F) Boxplots showing relative expression of Ddi2^WT (top) and truncated Ddi2^ex6 (bottom) forms of mRNA expressed in Ddi2^ex6 E10.5 stage embryos (Ddi2^ex6+/+—red, Ddi2^ex6+/−—yellow, Ddi2^ex6−/-—blue; n = 7). Relative expression was normalized to the housekeeping genes Tbp and H2afz with applied ANOVA statistical analysis (∗∗). The error bars denote SD. (G) Ddi2 expression analysis using RNA in situ hybridization of C57BL/6NCrl embryos. Three developmental stages at the beginning of the onset of developmental failure in the Ddi2 knock-out model strains are shown.

Figure 2

DDI2 dysfunction in mouse embryos leads to systematic breakdown causing premature death (A) RT-qPCR analysis of UPR involved genes in Ddi2^ex6 embryos among stages prior to the onset of lethality. Legend: Ddi2^ex6+/+ —red, Ddi2^ex6+/−—yellow, Ddi2^ex6−/-—blue; E9.5 (n = 5), E10.5 (n = 7), E11.5 (n = 5). The relative expression was normalized to the housekeeping genes Tbp and H2afz. Statistical significance was calculated for each gene throughout the three stages of embryonal development using ANOVA analysis (∗∗) or using a linear mixed-effects model (LMM) for the comparison of gene expression between wild-type and homozygous embryos at each stage of development (∗). Both analyses were subjected to Bonferroni correction; boxplots with SD. (B) Representative immunoblots of key markers of the UPR and ISR pathways (NRF1, ATF4, P-PERK, PERK, DDI2, P-eIF2α, eIF2α, P-PKR, and PKR) in tissue lysates of Ddi2^ex6+/+ (n = 6), Ddi2^ex6−/− (n = 10), and Ddi2^ex6+/− (n = 3) embryos. Tubulin and actin were used as loading controls. The black asterisks denote nonspecific bands. (C) Representative immunoblots of DNA-damage markers (yH2AX), cell cycle (P-Rb, cyclin E1, p21) and apoptosis markers (cleaved caspase 3, caspase 3) in tissue lysates of Ddi2^ex6+/+ (n = 6), Ddi2^ex6−/− (n = 10) and Ddi2^ex6+/− (n = 3) embryos. Tubulin and actin were used as loading controls. (D) Western blot analysis of the insoluble fraction of Ddi2^ex6+/+ (n = 6, red), Ddi2^ex6−/− (n = 10, blue) and Ddi2^ex6+/− (n = 2, yellow) embryo tissue lysates show accumulation of polyubiquitinated proteins of higher molecular weight. The membrane was probed with an anti-K48 linked polyubiquitin antibody, and the expression of ubiquitin conjugates with a molecular weight above 250 kDa was densitometrically quantified and normalized to the Amido Black loading control. The outlier in lane 7 was excluded from the calculation. Statistical significance was determined between Ddi2^ex6+/+ and Ddi2^ex6−/− using Mann-Whitney-test (∗p value ≤0.05; bar with mean ± SD).

Figure 3

DDI2 dysfunction in mice and endothelial cells alters UPS gene expression and abundance of proteins involved in proteostasis (A) Volcano plot of differentially abundant proteins in DDI2 KO cells compared to parental cells. Member proteins of over-represented pathways based on reactome analysis are highlighted. Quantitative analysis based on 3 biological replicates. Pathways enrichment analysis was performed using a list of 282 differentially abundant proteins with the Reactome database vs.79. (B) Heatmap showing differences in the protein metabolism network from the top identified and differentially abundant proteins with more than 1.5-fold difference between DDI2 KO cells (clone 17) and the parental cell line (n = 3). (C) RT-qPCR analysis of mRNA of NFE2L1 (TCF11/NRF1) and proteasomal subunits PSMA2 (α2), PSMB6 (β1), PSMC4 (RPT3), and PSME2 (PA28β) in EAhy926 parental cells (red) and DDI2 KO (blue) cells. Cells were treated with 50 nM BTZ for 8 h. Messenger RNA levels were normalized to RPLP0. two-way ANOVA was used for statistical calculation (n = 4; ∗p value ≤0.05; box and whiskers with min to max). (D) qRT-PCR analysis of NRF1-regulated genes in the UPS pathway in Ddi2^ex6 embryos in stages prior to the onset of lethality. Legend: Ddi2^ex6+/+ —red, Ddi2^ex6+/−—yellow, Ddi2^ex6−/-—blue; E9.5 (n = 5), E10.5 (n = 7), E11.5 (n = 5). Relative expression of genes was normalized to Tbp and H2afz housekeeping genes. Statistical significance was calculated for each gene at all three developmental stages with application of ANOVA statistical analysis (∗∗) or using a linear mixed-effects model (LMM) for comparison of gene expression between wild-type and homozygous embryos at each developmental stage (∗p value ≤0.05; boxplots with SD). Both analyses were subjected to Bonferroni correction.

Figure 4

DDI2 dysfunction alters proteasome composition and activity, leading to accumulation of high molecular weight polyubiquitinated proteins Proteasomal complexes analyzed is schematically illustrated in the legend. Amido Black staining served as loading control. Vertical dashed line depicts border between two areas from one western blot membrane. Statistical calculation of the p value was performed using two-way ANOVA with Šídák’s multiple comparisons test except 4D (unpaired t-test). Summarized data were given in Table S7. The native tissue lysates of Ddi2ex6^+/− embryos were given exemplary and not quantified. (A) Immunoblots after native page analyzing expression of proteasomal subunits α6 and RPN5 in native cell lysates of parental EAhy926 and DDI2 KO cells (clones #4 and #17) treated with 50 nM BTZ for 8 h compared to non-treated controls (n = 3, 20 μg of total protein/lane; quantification given by bar with mean ± SD). (B) Immunoblots of TCF11, DDI2, PA28β, and β5i/LMP7 in whole cell extracts of EAhy926 parental and DDI2 KO cells (clones #4 and #17) treated with 50 nM BTZ for 8 h compared to non-treated controls (n = 3, 25 μg of total protein/lane). (C) Chymotrypsin-like activity of native cell lysates of EAhy926 parental and DDI2 KO cells treated with 50 nM BTZ for 8 h compared to non-treated controls (n = 3; quantification given by bar with mean ± SD) was measured in a gel-based assay with the fluorogenic substrate Suc-LLVY-AMC. Loading control was the same as for α6 immunoblot. (D) Immunoblots for K-48 linked polyubiquitin of whole cell extracts of EAhy926 parental (red) and DDI2 KO clone #17 (yellow) and clone #4 (blue) cells treated with 50 nM BTZ for 8 h compared to non-treated controls. The statistical significance was calculated using the unpaired two-sided t-test (∗p value ≤0.05, n = 3; bar with mean ± SD). (E and F) Representative immunoblots of native tissue lysates (15 μg of total protein/lane; quantification given by bar with mean ± SD) of Ddi2^ex6+/+ (n = 3), Ddi2^ex6−/− (n = 5) and Ddi2^ex6+/− (n = 1) embryos probed for proteasomal subunits α4, RPT6 and PA28α. (G) Chymotrypsin-like activity of native tissue lysates of Ddi2^ex6+/+ (n = 3), Ddi2^ex6−/− (n = 5), and Ddi2^ex6+/− (n = 1) embryos was measured in gels based on the hydrolysis of the fluorogenic substrate Suc-LLVY-AMC. Loading control was the same as for α4 and RPT6; quantification given by bar with mean ± SD.

Figure 5

The NRF2 pathway is activated upon loss of DDI2 (A) Heatmap showing changes in the stress response pathway of the top identified and differentially abundant proteins with fold difference more than 1.5-fold between DDI2 KO cells (clone 17) and the parental cell line (n = 3). (B) Immunoblots of NRF2 expression in whole cell extracts of the parental EAhy926 (red) and DDI2 KO clone 17 (blue) treated with 50 nM BTZ for 2 and 4 h (NRF2 detection) in comparison to non-treated controls. Protein levels were quantified after normalization to the tubulin control signal. Statistical significance was determined using Mann-Whitney-test (n = 4; ∗p value ≤0.05; bar with mean ± SD). (C) Immunoblots of HO-1 level in whole cell extracts of EAhy926 parental (red) and DDI2 KO (blue) treated with 50 nM BTZ for 8 h in comparison to controls without treatment. Protein levels (n = 4) were quantified and statistically analyzed as in (B). (D) RT-qPCR analysis of NRF1 and NRF2-regulated genes in Ddi2^ex6 embryos in stages prior to the onset of lethality. Legend: Ddi2^ex6+/+—red, Ddi2^ex6+/−—yellow, Ddi2^ex6−/-—blue; E9.5 (n = 5), E10.5 (n = 7), E11.5 (n = 5). Relative expression of genes was normalized to Tbp and H2afz housekeeping genes; outliers were omitted based on the Grubbs’ test. Statistical significance was calculated either for each gene throughout all three developmental stages with application of ANOVA statistical analysis (∗∗) or using a linear mixed-effects model (LMM) for comparison of gene expression between wild-type and homozygous embryos at each developmental stage (∗). Both analyses were subjected to Bonferroni correction; boxplots with SD.

Figure 6

DDI2 KO cells show reduced TCF11/NRF1 activation and increased downstream stress markers induction upon proteotoxic stress (A) Analysis of TCF11/NRF1 and ATF4 levels in individual cellular fractions (nuclear, non-nuclear, and chromatin associated) of parental and DDI2 KO EAhy926 cells (treatment with 50 nM BTZ for a 4-h time course, n = 5). Calnexin, tubulin, CREB, and histone H3 served as markers of individual fractions. u = unprocessed; p = processed. (B) Immunoblots and quantification of TCF11/NRF1 protein abundance in whole cell extracts of parental EAhy926 (red) and DDI2 KO (blue) treated with 50 nM BTZ for 8 and 16 h or without treatment. Quantification was performed upon normalization to an Amido Black loading control. Statistical calculation of the p value was performed using two-way ANOVA with Šídák’s multiple comparisons test (n = 3; bar with mean ± SD). Summarized data were given in Table S7. For complete time course experiment see Figure S6C. (C) Immunoblots of P-PKR, PKR, eIF2, and P-eIF2α in whole cell extracts from parental EAhy926 cells and clones #4 and #17 of DDI2 KO cells, treated with 50 nM BTZ over a time course of 8 h (n = 3, 25 μg of total protein/lane). Vertical dashed lines depict border between two areas from one western blot membrane. The respective tubulin serves as loading control. (D) Immunoblots of ATF4 and CHOP in whole cell extracts from parental EAhy926 cells and two clones of DDI2 KO cells treated with 50 nM BTZ for 8 h (n = 3, 25 μg of total protein/lane). Vertical dashed lines depict border between two areas from one western blot membrane. Tubulin serves as loading control. (E) Immunoblots of apoptosis and senescence markers in whole cell extracts from parental EAhy926 cells and two clones of DDI2 KO cells (n = 3, 25 μg of total protein/lane). Vertical dashed lines depict border between two areas from one western blot membrane. The respective tubulin serves as loading control.

Figure 7

Depletion of DDI2 affects type I interferon signaling (A) Venn diagram showing identification of an individual interferon regulated gene signature based on proteome data. The interferon assignment was divided into type I, type II, and type III. The numbers represent varying proteins, which are assigned differently to the types. (B) Simplified scheme of one of the major type I interferon (IFN) production pathways and downstream signaling. The activation of various pattern recognition receptors (PRRs) initiates the activity of IκB kinase- (IKKε) and TANK-binding kinase-1 (TBK1) leading to DEAD box protein 3 (DDX3) mediated phosphorylation of the transcription factors IFN regulatory factor 3 (IRF3) and 7 (IRF7), serving as transcriptional activators of type I IFN. Upon binding of IFNα/β to the interferon alpha/beta receptor 1 and 2 (IFNAR1/2) the two Janus kinases (JAK and TYK2) bound to the receptor chains are activated. The signal transducer and activator of transcription 1 and 2 (STAT1/STAT2) are subsequently phosphorylated and released. These transcription factors form heterodimers and interact with IRF9. This complex migrates to the nucleus and induces transcription of IFN stimulated genes (ISGs). The scheme was created with BioRender.com. (C) Representative immunoblots of differences in protein expression of P-TBK1, TBK1, P-STAT1, and STAT1 in Ddi2^ex6+/+ (n = 6), Ddi2^ex6−/− (n = 10) and Ddi2^ex6+/− (n = 3) embryos. Tubulin was used as a loading control. Vertical dashed lines depict border between two areas from one western blot membrane. (D) Immunoblots of the protein expression of P-TBK1, TBK1, P-IKKε, IKKε, P-IRF3, IRF3, P-STAT1, STAT1, P-STAT3, and STAT3 in response to proteotoxic stress and under normal conditions. Expression was monitored in whole cell extracts of human EAhy926 parental cells and DDI2 KO clone #17 cells treated with 50 nM BTZ for 8 h (n = 4, also see Figure S7F). Tubulin was used as a loading control. (E) RT-qPCR expression analysis of key interferon stimulated genes ISG15, IFNB, IFI44, and IFI44L in human EAhy926 parental (red) and DDI2 KO (blue, clone #17) cells (also see Figures S7C–S7E). Messenger RNA amounts were normalized to RPLP0. Statistical significance was calculated using Mann-Whitney test (n = 4, ∗p value ≤0.05; box and whiskers with min to max).