ISGylation of DRP1 closely balances other post-translational modifications to mediate mitochondrial fission

Abstract

Dynamin related protein 1 (DRP1), a pivotal mitochondrial fission protein, is post-translationally modified by multiple mechanisms. Here we identify a new post-translational modification of DRP1 by the ubiquitin-like protein, interferon-stimulated gene 15 (ISG15). DRP1 ISGylation is mediated by ISG15 E3 ligase, HERC5; this promotes mitochondrial fission. DeISGylation of DRP1 however leads to hyperfusion. Heterologous expression of SARS-CoV2 PLpro, a deISGylating enzyme, results in similar mitochondrial filamentation, significant decrease in total DRP1 protein levels and efflux of mtDNA. We report that deISGylated DRP1 gets ubiquitylated and degraded by TRIM25, instead of PARKIN and MITOL. While the cytosolic pool of DRP1 is primarily ISGylated, both mitochondrial and cytosolic fractions may be ubiquitylated. It is known that phosphorylation of DRP1 at S616 residue regulates its mitochondrial localisation; we show that ISGylation of phospho-DRP1 (S616) renders fission competence at mitochondria. This is significant because DRP1 ISGylation affects its functionality and mitochondrial dynamics in Alzheimer's disease pathophysiology.

Fig. 1. ISGylation of DRP1 by HERC5 and effect on mitochondrial dynamics.

A A549 cell lysates were immunoprecipitated (IP) with anti-ISG15 antibody. Western blot (IB) analysis of extract with anti-DRP1 antibody shows co-immunoprecipitation (IP) of ISG15-modified DRP1. Note a shift in band size detected between input and immunoprecipitated samples. The proportion of lysate loaded as input and used for immunoprecipitation is denoted in brackets by ‘X’. endogenous DRP1, ISGylated endogenous DRP1. B Cell lysates co-immunoprecipitated to verify interaction between HERC5 with DRP1. The proportion of lysate loaded as input and used for immunoprecipitation is denoted in brackets by ‘X’. endogenous DRP1, ISGylated endogenous DRP1. C Mock or HERC5 siRNAs-treated cell lysates were immunoprecipitated (IP) with anti-ISG15 antibody (left panel). Note Western blot (IB) analysis with anti-DRP1 antibody shows decrease in ISGylated-DRP1 in HERC5-depleted samples. Reverse co-IP on the right corroborates the same. HERC5 and VINCULIN confirm knockdown and loading efficiencies. D A549 cells treated with 10 ng/ml of human IFNα1 (hIFNα1) for 48 h were lyzed and checked for ISGylation of DRP1 by co-IP between ISG15 and DRP1. Note that treatment with hIFNα1 induces enhanced ISGylation of DRP1. Similarly hIFNα1 treatment results in higher ISG15 expression, VINCULIN served as loading control. The proportion of lysates and immunoprecipitates denoted by ‘X’ in brackets. endogenous DRP1, ISGylated endogenous DRP1. E Cells treated with the hIFNα1 as described in panel D and treated with MitoTracker Red FM were imaged under live-cell conditions. Scale bar, 20μm. Box plots showing quantification of mitochondrial length for the experiment. ~150 cells from 3 independent experiments were analysed. The central line and the plus (+) symbol in each box show the median and mean value, respectively. ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. F A549 cells were transfected with indicated mCherry-tagged constructs and co-immunoprecipitated against ISG15 and DRP1. Note that K532R mutation in DRP1 compromises its ISGylation; ISGylated DRP1 band being more prominent in the dark exposure. ISG15 and VINCULIN served as loading control. ISGylated endogenous DRP1, ◄ ISG15-modified mCherry-tagged DRP1, endogenous DRP1, ← mCherry-tagged DRP1. G Cells were transfected with indicated mCherry-tagged constructs and treated with MitoTracker Green FM were imaged under live-cell conditions. Scale bar, 20μm. Box plots showing quantification of mitochondrial length for the experiment. ~200 cells from 3 independent experiments were analysed. The central line and the plus (+) symbol in each box show the median and mean value, respectively. ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. H Experimental hypothesis regarding differential effects of post-translational modifications (ISGylation and ubiquitylation) of DRP1 on mitochondrial dynamics.

Fig. 2. Reduced DRP1 levels and mtDNA release in presence of CoV-2 PLpro.

A Cells transfected with the indicated GFP-tagged constructs and treated with MitoTracker Red FM were imaged under live-cell conditions. Scale bar, 5μm. B Box plot showing quantification of mitochondrial length for the experiment described in A. ~90 cells from 3 independent experiments were analysed. The central line and the plus (+) symbol in each box show the median and mean value, respectively. ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. C A549 cells transiently transfected with the indicated GFP-tagged constructs were immunoblotted against the indicated antibodies. D Graphs plot changes in expression of proteins analysed in panel C. Data represent the mean ± SEM of three independent experiments. ns, not significant (p>0.1), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. E Cells were transfected with indicated GFP-tagged constructs and immunostained with antibodies against TOMM20 and TFAM. Z-stacks (0.15μm slices) were taken. Representative images show 3-D projections. Enlarged views of the areas within the white boxes shown (insets). White arrowheads mark TFAM puncta outside TOMM20 boundary. Scale bar, 5μm. F Graph represents data from ∼175 cells from 3 independent experiments. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. Error bars,±SEM. G A549 cells transfected similarly as in panel E and were immunoblotted against TOMM20 and TFAM antibodies, β-ACTIN was used as loading control. Graphs plot show changes in expression of proteins analysed. Data represent the mean ± SEM of 3 independent experiments. ns, not significant (p > 0.6), *p ≤ 0.05 using unpaired 2-tailed Student’s t-test. H Cytosolic and membrane fractions obtained from digitonin permeabilised A549 cells were immunoblotted with TFAM antibody; inputs indicate total protein in cell lysates. HSP90 and RTN4 served as controls for cytosolic and membrane fractions, respectively. I Graph with data from panel H shows results of 3 independent experiments. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. Error bars, ±SEM. J DNA isolated from whole cell and cytosolic extracts was subjected to SYBR Green-based qPCR to quantify nuclear (GAPDH) and mitochondrial (COXII) DNA using specific primers. Plots show abundance of total (left) and cytosolic (right) cellular mtDNA. ns, not significant (p = 0.7), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. Error bars, ±SEM.

Fig. 3. mtDNA release in presence of CoV2 PLpro phenocopy DRP1 and HERC5 depletion.

A Cells treated with mock or DRP1 siRNAs were immunostained with anti-TOMM20 antibody and imaged. Scale bar, 10μm. Knockdown efficiency was confirmed by immunoblotting against DRP1, cell lysates were immunoblotted against TOMM20 and TFAM, β-TUBULIN was used as loading control. Graph quantifies expression of indicated proteins. Data represent the mean ± SEM of 3 independent experiments. ns, not significant (p > 0.2) using unpaired 2-tailed Student’s t-test. Box plot with data from image panel; the central line and the plus (+) symbol in each box show the median and mean value, respectively. ~75 cells from 3 independent experiments were analysed. ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. B Representative images of cells similarly treated as in panel A were immunostained against TOMM20 and TFAM. Z-stacks (0.15μm slices) were taken. Images show 3-D projections. Enlarged views of the areas within the white boxes shown (insets). White arrowheads mark TFAM puncta outside TOMM20 boundary. Scale bar, 10μm. Graph represents data from ∼100 cells from 3 independent experiments. ***p ≤ 0.001, using unpaired 2-tailed Student’s t-test. Error bars, ±SEM. C Cells similarly treated as in panel A were stained against LIVE ORANGE mito and PicoGreen to visualise cristae and dsDNA, respectively, and imaged in the slice 3D-SIM live mode. Enlarged views of the areas within the white boxes shown (insets). White arrowheads mark PicoGreen puncta outside LIVE ORANGE mito boundary. Scale bar, 10 μm. D Cytosolic and membrane fractions obtained from digitonin-permeabilised A549 cells were immunoblotted with indicated antibodies; inputs indicate total protein in cell lysates. Note increased TFAM protein levels in cytosolic fraction and decrease in membrane fraction upon DRP1 depletion, more prominent in the blot with darker exposure. HSP90 and RTN4 served as controls for cytosolic and membrane fractions, respectively. DRP1 levels indicate knockdown efficiency. E DNA isolated from whole cell and cytosolic extracts was analysed by qPCR. Nuclear (GAPDH) and mitochondrial (COXII) DNA was quantified using specific primers. Plots show abundance of total (left) and cytosolic (right) cellular mtDNA. ns, not significant (p = 0.07), **p ≤ 0.01 using unpaired 2-tailed Student’s t-test. Error bars, ±SEM. F Mock or DRP1 siRNAs-treated cells were immunostained for ISG15. DRP1 and β-TUBULIN confirm knockdown efficiency and equal sample loading. Graph depicts changes in ISG15 levels. Data represents 3 independent experiments. ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. Error bars, ±SEM. G Representative images of cells were treated with indicated siRNAs and immunostained with antibodies against TOMM20 and TFAM. Z-stacks (0.15 μm slices) were taken. Images show 3-D projections. Insets shown. White arrowheads mark TFAM puncta outside TOMM20 boundary. Scale bar, 10μm. Graph represents data from ∼75 cells from 3 independent experiments. ***p ≤ 0.001, using unpaired 2-tailed Student’s t-test. Error bars, ±SEM. Imunoblot of lysates post-imaging confirm HERC5 depletion, VINCULIN was used as loading control. H Lysates similarly generated as in panel G were immunostained against DRP1, ISG15, HERC5 and VINCULIN. Graphs show the changes in the expression of DRP1 and ISG15. Data represents 3 independent experiments. ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. Error bars, ±SEM. I Cytosolic and membrane fractions obtained from digitonin permeabilised A549 cells were immunoblotted with TFAM antibody; inputs indicate total protein in cell lysates. Note higher levels of TFAM present in cytosolic fraction of HERC5-depleted samples. Blots representative of 3 independent experiments. HSP90 and RTN4 served as controls for cytosolic and membrane fractions, respectively. HERC5 levels confirm knockdown efficiency. J DNA isolated from whole cell and cytosolic extracts was analysed by qPCR. Nuclear (GAPDH) and mitochondrial (COXII) DNA was quantified using specific primers. Plots show the abundance of total (top) and cytosolic (bottom) cellular mtDNA. ns, not significant (p = 0.06), **p ≤ 0.01, using unpaired 2-tailed Student’s t-test. Error bars, ±SEM.

Fig. 4. Ubiquitylation of DRP1 by TRIM25.

A Cell transfected with the indicated GFP-tagged constructs were lysed, immunoprecipitated and immunoblotted. Note differential ISGylation profiles of DRP1 amongst the samples. Reverse co-IP confirms the same result. The input levels of DRP1, ISG15 and VINCULIN in the total lysates served as loading controls. B Cells transfected as in panel A were immunoprecipitated (IP) with anti-DRP1 antibody. Western blot (IB) analysis against HA-tagged Ub reveals increased DRP1 ubiquitylation in CoV2 PLpro WT samples. DRP1 and VINCULIN levels in the total lysates served as loading controls. C Mock or HERC5 siRNAs-treated cell lysates were immunoprecipitated (IP) with anti-DRP1 antibody. Ex vivo ubiquitylation of DRP1 was analysed by immunoblotting against Ub. Lysates were also checked for the levels of DRP1, HERC5 and VINCULIN. D Cells transfected with indicated GFP-tagged constructs were treated MG132 or left untreated. Note increased DRP1 levels upon drug treatment. Immunoblots were analysed against DRP1 and VINCULIN. Graph shows changes in protein levels of DRP1. Data represents 3 independent experiments. ns, not significant (p > 0.07), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. E Cells were transfected with indicated mCherry-tagged constructs, immunoprecipitated against DRP1 and immunoblotted with Ub antibody. Note that K532R mutation in DRP1 compromises its ubiquitylation. endogenous DRP1, ← mCherry-tagged DRP1. VINCULIN was used as loading control. F Cells transfected with CoV2 PLpro WT were treated with the indicated siRNAs and immunoblotted against DRP1. Knockdown efficiency was confirmed by immunoblotting against PARKIN, TRIM25 and MITOL. VINCULIN was used as loading control. Graph shows the changes in the protein levels of DRP1. Data represents 3 independent experiments. ns, not significant (p > 0.07), ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. G Mock or TRIM25 siRNAs treated cells transfected with indicated GFP-tagged constructs were immunoblotted with anti-DRP1 antibody. Note increased DRP1 levels upon TRIM25 depletion. Input levels of TRIM25 and VINCULIN served as loading controls. Graph plots protein levels of DRP1. Data represents 3 independent experiments. ns, not significant (p > 0.08), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. H A549 cell lysates were immunoprecipitated against DRP1 and immunoblotted with TRIM25 antibody. The proportion of lysate loaded as input and used for immunoprecipitation is denoted in brackets by ‘X’. I Mock or TRIM25 siRNAs treated cells were co-immunoprecipitated against ISG15 and DRP1. Note similar ISGylated DRP1 levels across samples. TRIM25 levels in the lysates confirm knockdown efficiency, VINCULIN served as loading control. DRP1 levels were also checked in cell lysates. J Cells expressing the indicated GFP-tagged constructs were depleted off TRIM25 and analysed for co-IP between DRP1 and Ub. Note significantly reduced DRP1 ubiquitylation in TRIM25 depleted cells expressing CoV2 PLpro WT. Lysates were immunoblotted against DRP1, TRIM25 and VINCULIN.

Fig. 5. Differential post-translational modifications alter DRP1 localisation.

A Cytosolic and mitochondrial fractions obtained from semi-permeabilised cells, transfected with indicated constructs, were immunoblotted using antibody against DRP1. GAPDH and VDAC1 served as controls for cytosolic and mitochondrial fractions, respectively. Note lower levels of DRP1 were observed in samples with CoV2 PLpro WT. Unmodified DRP1, ← and likely to be post-translationally modified forms of DRP1. Note reduced DRP1 levels in input (exposure 1), mitochondrial (exposure 2) and cytosolic (exposure 3) fractions in CoV2 PLpro WT samples. (B) Total RNA isolated from A549 cells transfected with the indicated GFP-tagged constructs was subjected to qRT-PCR using primers against GAPDH (control) and DRP1. Samples were present in triplicate. 2^-ΔΔCt values were plotted. Graph shows results from 3 independent experiments. ns, not significant (p > 0.1) using unpaired 2-tailed Student’s t-test. Error bars, ±SEM. C Cytosolic and mitochondrial fractions obtained from semi-permeabilised cells were analysed for co-IP between DRP1 and ISG15. Note that cytosolic DRP1 pool is primarily ISGylated. VINCULIN and VDAC1 served as controls for cytosolic and mitochondrial fractions, respectively. D Cells fractionated as in panel C were probed for co-IP between DRP1and Ub. Note ubiquitylated DRP1 present in both the fractions. VDAC1 and VINCULIN levels in the total lysates served as loading controls. E Cells transfected with the indicated GFP-tagged constructs, fractionated as in panel C were divided into two. One part was immunoprecipitated with anti-ISG15 antibody and immunoblotted against DRP1. Note lower levels of ISGylated DRP1 in samples with CoV2 PLpro WT than the mutant. VINCULIN and VDAC1 served as controls. F Second part generated in panel E was analysed for co-IP between DRP1 and Ub. VDAC1 and VINCULIN levels confirm fractionation efficiency. G Cytosolic and mitochondrial fractions obtained from semi-permeabilised cells were analysed for TRIM25. VINCULIN and VDAC1 served as controls.

Fig. 6. DRP1 ISGylation affects its function.

A Cytosolic and mitochondrial fractions obtained from semi-permeabilised cells, transfected with indicated constructs, were immunoblotted against phospho-DRP1 (S616) and total DRP1; VINCULIN and VDAC1 served as controls for cytosolic and mitochondrial fractions, respectively. Note lower levels of phospho-DRP1 (S616) and DRP1 in DRP1^K532R mitochondrial fractions. B Graph represents data from 3 independent experiments. **p ≤ 0.01, ***p ≤ 0.001, using unpaired 2-tailed Student’s t-test. Error bars,±SEM. C Cells transfected with indicated mCherry-tagged constructs and treated with MitoTracker Green FM were imaged in the slice 3D-SIM live mode. White boxes show insets. White arrowheads mark the mCherry and mitochondria localisation. Scale bar, 10μm. D Graph represents number of DRP1 puncta / 2μm length of mitochondria from 3 independent experiments. ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. Error bars,±SEM. ~50 mitochondria were analysed. E Cells transfected with indicated mCherry-tagged constructs and treated with MitoTracker Green FM were imaged under live-cell conditions. Scale bar, 20μm. F Box plot showing quantification of mitochondrial length; ~100 cells from 3 independent experiments were analysed. The central line and the plus (+) symbol in each box show the median and mean value, respectively. ns, not significant (p = 0.6), *p ≤ 0.05, ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. G Cytosolic and mitochondrial fractions generated as in panel A were immunoblotted against phospho-DRP1 (S616) and total DRP1; VINCULIN and VDAC1 served as controls for cytosolic and mitochondrial fractions, respectively. H Graph represents data from 3 independent experiments. ns, not significant (p = 0.7), **p ≤ 0.01, ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. Error bars,±SEM.

Fig. 7. Altered DRP1 ISGylation in Alzheimer’s disease.

A Co-immunoprecipitation against ISG15 and DRP1 was done in human brain whole tissue lysate. Note differential ISGylation profiles of DRP1 between the samples. The input levels of DRP1, ISG15, HERC5, TRIM25 and VINCULIN in the total lysates served as loading controls. B Brain lysates from 5XFAD mice were co-immunoprecipitated and analysed as in panel A. The input levels of DRP1, ISG15, and VINCULIN served as loading controls. C SHSY5Y cells transfected with indicated constructs were treated with DMSO and 0.5 μM Aβ for 48 h. Note reduced presence of ISGylated DRP1 in samples with AICD and Aβ, as detected by co-immunoprecipitation. The input levels of DRP1, ISG15, HERC5, TRIM25 and VINCULIN were analysed. D SHSY5Y cells treated with DMSO or 1 μM Aβ for 48 h were verified for DRP1 ISGylation as in panel C. E Cytosolic and mitochondrial fractions obtained from semi-permeabilised cells, transfected with indicated constructs and treated with DMSO, 0.5 μM Aβ or 1 μM Aβ for 48 h, were immunoblotted against phospho-DRP1 (S616) and total DRP1; VINCULIN and VDAC1 served as controls for cytosolic and mitochondrial fractions, respectively. F Graph represents data from 3 independent experiments. **p ≤ 0.01 using unpaired 2-tailed Student’s t-test. Error bars,±SEM. G Cells transfected with indicated GFP-tagged constructs and treated with DMSO or 0.5 μM Aβ over indicated time periods were imaged with MitoTracker Red FM under live-cell conditions. Scale bar, 20μm. H Box plot showing quantification of mitochondrial length; ~125 cells from 3 independent experiments were analysed. The central line and the plus (+) symbol in each box show the median and mean value, respectively. ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test. I Cells treated with DMSO or 1 μM Aβ for indicated time periods were imaged as in panel G. Scale bar, 20μm. J Mitochondrial length from ~120 cells were analysed and plotted from 3 independent experiments. The central line and the plus (+) symbol in each box show the median and mean value, respectively. ***p ≤ 0.001 using unpaired 2-tailed Student’s t-test.

Fig. 8. Schematic diagram summarising the results.

ISGylation of DRP1 [more specifically phopho-DRP1 (S616)] renders it fission competent. Mitochondrial filamentation occurs due to lack of this post-transaltional modification (as in Alzheimer’s disease, AD scenario) or in the presence of SARS-CoV2 PLpro (a deISGylating enzyme); USP18 could also impart deISGylation in cells. DRP1 is primarily ISGylated by HERC5 leading to fission; Ubiquitin E3 ligases, like PARKIN, MITOL and TRIM25 ubiquitylate and degrade DRP1 to promote fusion. In AD reduced expression of HERC5 and TRIM25 results in compromised DRP1 ISGylation and activity.