Global mitochondrial protein import proteomics reveal distinct regulation by translation and translocation machinery

Abstract

Most mitochondrial proteins are translated in the cytosol and imported into mitochondria. Mutations in the mitochondrial protein import machinery cause human pathologies. However, a lack of suitable tools to measure protein uptake across the mitochondrial proteome has prevented the identification of specific proteins affected by import perturbation. Here, we introduce mePROD^mt, a pulsed-SILAC based proteomics approach that includes a booster signal to increase the sensitivity for mitochondrial proteins selectively, enabling global dynamic analysis of endogenous mitochondrial protein uptake in cells. We applied mePROD^mt to determine protein uptake kinetics and examined how inhibitors of mitochondrial import machineries affect protein uptake. Monitoring changes in translation and uptake upon mitochondrial membrane depolarization revealed that protein uptake was extensively modulated by the import and translation machineries via activation of the integrated stress response. Strikingly, uptake changes were not uniform, with subsets of proteins being unaffected or decreased due to changes in translation or import capacity.

Keywords: SILAC; TMT; disease; integrated stress response; mitochondria; protein translocation; proteomics; proteostasis; respiratory chain complexes; translation.

Graphical abstract

Figure 1

Compartment-specific signal boosting of cell-wide and organelle-selective pulsed-SILAC experiments (A) Workflow of cell-wide (left, blue) and mitochondria-selective (right, magenta) pulsed-SILAC proteomics for measuring translation or uptake of mitochondrial proteins, respectively. A non-labeled baseline sample and samples of cells pulse labeled for 2 h were subjected to whole-cell extraction (left) for translation measurement or mitochondrial isolation (right) for measuring protein uptake. Equally, cells of a fully SILAC-labeled booster sample were subjected to whole-cell extraction or mitochondrial isolation to yield a whole-cell or mitochondrial booster. Baseline and pulse-labeled samples for translation or protein uptake measurements were complemented with a whole-cell or mitochondrial booster to improve the sensitivity for all or mitochondrial proteins. Proteins were digested, labeled with tandem mass tag (TMT)11, pooled, and measured by LC-MS/MS with targeted mass difference (TMD). (B–D) Shown are numbers of heavy SILAC-labeled mitochondrial peptides (B), proteins (C), or mitochondrially encoded proteins (D) dependent on the addition of no booster or booster signals derived from whole-cell or mitochondrial extracts (n = 3 in 1 multiplex). (E and F) Number of heavy SILAC-labeled mitochondrial peptides (E) or proteins (F) from purified mitochondria identified in combination with no booster or booster signals derived from whole-cell or mitochondrial extracts (n = 3 in 1 multiplex). See also Figure S1.

Figure 2

Characterization of protein uptake kinetics across the mitochondrial proteome (A) Experimental scheme of a time course experiment to determine mitochondrial protein uptake kinetics. Cells were pulsed-SILAC-labeled for 0–360 min in duplicate and then subjected to mitochondrial isolation, TMT labeling, and measurement by LC-MS/MS with targeted mass difference selection. (B) Plot of heavy-to-total protein ratios of proteins over time (n = 2). (C) Coefficients of determination (R²) of the uptake slope determined for each identified mitochondrial protein, shown as individual data points and half-violin plot. (D) Plot of individual uptake slopes of all identified mitochondrial proteins. Color of data points represents R² value of the respective uptake slope. Data points are scattered along the vertical axis to prevent excessive overlapping. a.u., arbitrary unit. (E) Pearson correlation of protein uptake slopes [log₂] and protein copy numbers [log₂]. R, Pearson coefficient. (F) Normalized protein uptake over time of mitochondrial proteins showing R² ≥ 0.95 grouped into quartiles (Q1–Q4) based on their uptake slope (n = 2, see D). Data points show mean values of the quartiles over time. (G) Representation of Reactome pathway network for proteins within Q1 (from F). Network was obtained with the Cytoscape plug-in ClueGO. Pathways highlighted with colors and gray boxes were significantly enriched. p value corrected with Benjamini-Hochberg procedure. The full Gene Ontology (GO) term list is provided in Table S2. See also Figure S2 and Tables S1 and S2.

Figure 3

Mitochondrial uptake proteomics reveal complex protein uptake patterns upon perturbation (A) Illustration of the effects of MitoBloCK-6 (MB6) and CCCP on the uptake of proteins into the intermembrane space and matrix, respectively (top). Fold changes of mitochondrial protein uptake upon MitoBloCK-6 (left) or CCCP (right) treatment compared to DMSO-treated control cells, shown as density plots (upper) and volcano plots (lower) plotted against the adjusted p value (n = 3). Dashed lines indicate median values of the distributions. Data points of targets with significant changes (fold change [log₂] ≤ −0.7 or ≥0.7, and adjusted p ≤ 0.05 for MitoBloCK-6; fold change [log₂] ≤ −1 or ≥1, and adjusted p ≤ 0.05 for CCCP) are shown in blue. Adjusted p values > 10 were set to p = 10 for plotting; original adjusted p values are given in Tables S3 and S4. Kolmogorov-Smirnov test against normal distribution ∼0, with standard deviation obtained for the tested distribution. (B) Density plots of mitochondrial protein uptake fold changes upon MitoBloCK-6 (left) or CCCP (right) treatment compared to DMSO-treated control cells, based on suborganellar location (n = 3). Dashed lines indicate median values of the distributions. Dots at the bottom of the distributions indicate number and location of data points. OMM, outer mitochondrial membrane; IMS, mitochondrial intermembrane space; IMM, inner mitochondrial membrane; Kolmogorov-Smirnov test against normal distribution ∼0, with standard deviation obtained for the tested distribution. (C) Rank plot showing fold changes of mitochondrial protein uptake upon CCCP treatment compared to DMSO-treated control cells (n = 3). Unchanged targets (−0.35 ≤ fold change [log₂] ≤ 0.35) and the top 50 targets showing the most severe uptake defects highlighted in blue and red, respectively. (D and E) Reactome pathway networks of targets with unchanged (−0.35 ≤ fold change [log₂] ≤ 0.35) (D) or top 50 reduced (E) mitochondrial uptake upon CCCP treatment compared to DMSO-treated control cells. Networks were prepared with the Cytoscape plug-in ClueGO. Pathways highlighted with colors and gray boxes were significantly enriched. p value corrected with Benjamini-Hochberg procedure. Full GO term lists are provided in Table S2. See also Figure S3 and Tables S2, S3, and S4.

Figure 4

Stress shapes mitochondrial protein uptake via translation and import regulation (A) Density plot of fold changes of translation and mitochondrial uptake of mitochondrial proteins upon CCCP treatment compared to DMSO-treated control cells (n = 3). Dashed lines indicate median values of the distributions. Kolmogorov-Smirnov test against normal distribution ∼0, with standard deviation obtained for the tested distribution. (B) Correlation of CCCP-induced fold changes of protein translation and mitochondrial uptake, compared to DMSO-treated cells. (C) Scheme of integrated stress response (ISR) activation by CCCP and inhibition of the resulting translation attenuation by ISRIB (left). Violin plot of CCCP-induced changes in whole-cell translation with and without co-treatment with ISRIB, compared to DMSO-treated control cells (right, n = 3). Center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; points, outliers. Two-sample Kolmogorov-Smirnov test. (D) Violin plot of CCCP-induced mitochondrial protein uptake changes in the absence and presence of ISRIB, compared to DMSO-treated control cells (n = 3). Center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; points, outliers. Two-sample Kolmogorov-Smirnov test. (E) Targets with significant CCCP-induced uptake defects (fold change [log₂] ≤ −1 and adjusted p ≤ 0.05) segregate into 2 groups with import-driven (308 proteins; fold change [log₂] ≤ −1 and adjusted p ≤ 0.05) or translation-driven (33 proteins; −0.35 ≤ fold change [log₂] ≤ 0.35) uptake defects upon co-treatment with ISRIB. Volcano plots of cells treated with CCCP (center) or CCCP and ISRIB (right). Adjusted p > 10 were set to p = 10 for plotting; original adjusted p values given in Table S5. (F) Density plots of changes in mitochondrial protein uptake upon CCCP (left) or CCCP+ISRIB (right) compared to DMSO-treated cells for import-driven or translation-driven proteins (from E) separated by suborganellar locations. Dashed lines indicate median values of the distributions. Dots at the bottom of the distributions indicate number and location of data points. Areas of targets with import- or translation-driven uptake defects are highlighted in blue and cyan, respectively. Please note that the dataset only contains 2 IMS proteins, which is why their distributions are not included in the figure. (G) Histogram showing the distribution of mitochondrial proteins and proteins with import- or translation-driven import defects. (H) Distribution of mitochondrial proteins with import-driven import defects across different TargetP scores. Increasing TargetP scores negatively correlate with mitochondrial targeting signal (MTS) confidence. (I) Density plot of changes in mitochondrial protein uptake upon treatment with CCCP and ISRIB of proteins with import- or translation-driven uptake defects (left). Proteins were grouped based on their TargetP score. Areas of targets with import- or translation-driven uptake defects were highlighted in blue and cyan, respectively. Dots at the bottom of the distributions indicate number and location of data points. Inequality of the distributions was determined with a 2-sample Kolmogorov-Smirnov test (right). ns, non-significant. (J) Bar graph showing the fraction of targets with import- or translation-driven uptake defects for each TargetP score. See also Figure S4 and Table S5.

Figure 5

Mitochondrial uptake of respiratory complex I and mitochondrial translation machinery components is controlled by their import efficiency (A and B) Reactome pathway networks of targets with translation-driven (A) or import-driven uptake defects (B) upon CCCP treatment, prepared with the Cytoscape plug-in ClueGO. Significantly enriched pathways were highlighted with colors and gray boxes. p value corrected with Benjamini-Hochberg procedure. Full GO term lists are provided in Table S2. (C–G) Volcano plots showing fold changes of mitochondrial protein uptake plotted against the adjusted p value for cells co-treated with CCCP and ISRIB, compared to DMSO-treated control cells. Data points of components of the mitochondrial large ribosomal subunit (C), mitochondrial genome-encoded proteins (D), respiratory complex I (E), TOM complex (F), or TIMM22, TIM23, and PAM complexes (G) with significant CCCP-induced uptake defects (fold change [log₂] ≤ −1 and adjusted p ≤ 0.05) were labeled and colored according to their uptake behavior. Data points of proteins changing significantly (fold change [log₂] ≤ −1 or ≥ 1, and adjusted p ≤ 0.05) are shown in dark gray. Areas of targets with import- or translation-driven uptake defects were highlighted in blue and cyan, respectively. Note that only the fold change [log₂] and not the adjusted p value upon CCCP+ISRIB treatment, was taken into account for classification of import- and translation-driven targets here. See also Tables S2 and S5.