ULP-2 SUMO protease regulates UPR^mt and mitochondrial homeostasis in Caenorhabditis elegans

Abstract

Mitochondria are the powerhouses of cells, responsible for energy production and regulation of cellular homeostasis. When mitochondrial function is impaired, a stress response termed mitochondrial unfolded protein response (UPRmt) is initiated to restore mitochondrial function. Since mitochondria and UPRmt are implicated in many diseases, it is important to understand UPRmt regulation. In this study, we show that the SUMO protease ULP-2 has a key role in regulating mitochondrial function and UPRmt. Specifically, down-regulation of ulp-2 suppresses UPRmt and reduces mitochondrial membrane potential without significantly affecting cellular ROS. Mitochondrial networks are expanded in ulp-2 null mutants with larger mitochondrial area and increased branching. Moreover, the amount of mitochondrial DNA is increased in ulp-2 mutants. Downregulation of ULP-2 also leads to alterations in expression levels of mitochondrial genes involved in protein import and mtDNA replication, however, mitophagy remains unaltered. In summary, this study demonstrates that ULP-2 is required for mitochondrial homeostasis and the UPRmt.

Graphical abstract

Highlights

• •Mitochondrial unfolded protein response (UPRmt) is initiated upon mitochondrial stress.
• •ULP-2 SUMO protease is required for UPRmt induction.
• •Down-regulation of ulp-2 reduces mitochondrial membrane potential.
• •In ulp-2 null animals mitochondrial networks are expanded and mitochondrial DNA levels increase.

1. Introduction

The mitochondrial unfolded protein response (UPRmt) is a mitochondria-to-nucleus retrograde signaling pathway that helps to maintain the quality of mitochondria by ensuring the functional integrity of the mitochondrial proteome. It has been implicated in various diseases including cancer and neurodegeneration, highlighting the importance of understanding its regulation [[1], [2], [3]].

Studies in C. elegans identified ATFS-1 as a key transcription factor that mediates the UPRmt [4]. Under basal conditions, it is imported into mitochondria and degraded [5]. However mitochondrial dysfunction reduces mitochondrial import, resulting in ATFS-1 trafficking to the nucleus, where it activates the transcription of genes encoding chaperones (e.g., hsp-6, hsp-60) and multiple stress-related factors to re-establish mitochondrial homeostasis and promotes mitochondrial biogenesis [6]. Constitutive activation of the UPRmt results in expanded mitochondrial networks with increased mtDNA [7].

Growing evidence supports a regulatory role for the Small Ubiquitin-like Modifier (SUMO) post-translational modification in mitochondrial function and UPRmt [[8], [9], [10], [11]]. SUMOylation can affect protein function, interactions, and localization and can be reversed by Sentrin/SUMO-specific proteases (SENPs) [[12], [13], [14]]. SUMOylation is highly conserved in C. elegans, where the SENP family consists of four SUMO proteases termed ubiquitin-like proteases (ULPs): ULP-1, ULP-2, ULP-4 and ULP-5 [15,45]. In mammalian systems, SUMOylation has been shown to regulate mitochondrial dynamics by modifying DRP1 and MFN1/2 [16,17]. In C. elegans, SUMO has been shown to modify the key transcription factors regulating UPRmt activation, ATFS-1 and DVE-1 [8]. DeSUMOylation of these transcription factors by ULP-4 promotes their nuclear localization and enhances activation of the UPRmt. In addition, SUMO has been shown to affect UPRmt by SUMOylation of SKN-1 and DAF-16 [8,10].

In a genome-wide RNAi screen for UPRmt regulators in C. elegans, we have identified that RNAi of ulp-2 suppressed UPRmt induced by impaired mitochondrial dynamics [18]. ULP-2 is a nuclear/cytosolic SUMO protease essential for embryonic development[45]. Here, we showed that ulp-2 is a positive regulator of UPRmt and additional mitochondrial properties that are essential for cellular function.

2. Materials and Methods

2.1. Worm strains

C. elegans nematodes were cultured at 20 °C on NGM plates seeded with E. coli OP50 as previously described [19]. The following worm strains and alleles were used: Bristol N2 Wild Type (WT) strain, JV2 [jrIs2(rpl-17p::Grx-1-roGFP2) [20], fzo-1(tm1133) (National BioResource Project), drp-1(tm1108) (National BioResource Project), bcSi9 [hsp-6p::gfp::unc-54 3’UTR] [21], SJ4100 zcIs13[hsp-6p::gfp] [22]. MD3721 fzo-1(tm1133);bcSi9[hsp-6p::gfp::unc-54 3’UTR], MD4271 drp-1(tm1108);bcSi9[hsp-6p::gfp::unc-54 3’UTR], MD3674 fzo-1(tm1133);zcIs13[hsp-6p::gfp], NX399 ulp-2(tv380)/mln-1[mIs14 dpy-10(e128)], NX548 ulp-2(tv380)/mln-1[mIs14 dpy-10(e128)];bcSi9[hsp-6p::gfp::unc-54 3’UTR] (this study). IR2539 unc-119(ed3);Ex[myo-3p::TOMM-20::Rosella;unc-119(+)] [10]. All experiments using ulp-2(tv380)/mln-1[mIs14 dpy-10(e128)][45] strains were conducted on second generation (F2) ulp-2(tv380) homozygous worms. The first generation of ulp-2(tv380) homozygous worms show a high degree of embryonic lethality[45]. To quantify the effect of ULP-2 knockout, the second generation survivors expressing minimal residual maternal ULP-2 product were analyzed.

2.2. Real-time QPCR primers

Gene	Left Primer 5’-3’	Right Primer 5’-3’	Reference
ulp-2	ATTGCACCAAACTGGCATCC	GCGAAGGCATATCCGATTCTT	This study
atfs-1	ACGTGACGCTGGAAGCATG	GCGGCTGCTGACATGTCTG	[23]
hsp-6	CTGGAGATAAGATCATCGCTGTC	AGAGCGTGATCGAAGTCTTCTC	This study
hsp-60	ATGAAGTCACCGGAAACAGC	TGTGCGAACCACCTTAGTTG	This study
pmp-3	GTTCCCGTGTTCATCACTCAT	ACACCGTCGAGAAGCTGTAGA	[24]
atp-6	GTTTATGCTGCTGTAGCGTG	CTGTTAAAGCAAGTGGACGAG	[10]
ama-1	TGGAACTCTGGAGTCACACC	CATCCTCCTTCATTGAACGG
Ant-1.3–1.4	CGGACTCTACAGAGGATTCTTTG	TTGGCAGTGTCGAACATTCC	This study
C47G2.3	GATCCTACGAAACAGCTAACCC	AGTCGCTTTTTGCACGGATC	This study
tin-9.1	AAGAAGAATCGTGCGCCAAC	TTGGCTTGAGCGTTGAGAAG	This study
F53A3.7	ACCATCACATCATCGCTGAC	TCTCCTCGCGTTTTCAGATG	This study
F32B4.2	GGGGAAGCTTATTTGGAGGATG	AAAGTGAGCCGGTGAAATCG	This study
C27H6.9	AAGATGGGTCACCTGCTGAAG	ATTGGCGGATGTCACTTTGC	This study
tomm-23	CGCTAGGACATATCGGATGGG	TGCTTCATCGTTGCGTTCAC	This study
F4611.1	ATCCACTCGTGCTTCCGTATC	GGCGTTCTTCGTTTTTCACAAG	This study
mtx-2	TGCCTTGTCGATGAATGCAG	ACATGAGAGCTTGGTCATTGAG	This study

2.3. RNA-mediated interference

For imaging experiments, RNAi by feeding was performed as previously described [21]. ulp-2(RNAi) and cco-1(RNAi) were selected from the Ahringer library [25] and atfs-1(RNAi) from the Vidal RNAi library [26]. All RNAi clones were sequenced prior to use, and the empty vector L4440 was used as a negative control [referred to as ‘control(RNAi)’]. RNAi clones were cultured overnight in 2 mL of LB Carbenicillin (100 μg/mL) at 37 °C and 200 rpm. The RNAi cultures were adjusted to 0.5 OD and seeded on RNAi plates containing 6 mM IPTG and dried overnight. L4 larvae of the different strains were transferred to the RNAi plates, and L4 larvae of the F1 generation were collected for the imaging experiments. cco-1(RNAi) was diluted 1:3 with L4440(RNAi) in all experiments.

2.4. Imaging

For all imaging experiments, worms were mounted on 3 % agarose pads and immobilized in M9 buffer containing 10 mM tetramisole (Sigma L9756). Nikon eclipse 80i. Images were acquired using the NIS-Elements software in magnifications detailed for each experiment below. Leica (M205FA) fluorescence stereo microscope. Images (Fig. S1 A and S2 B–C) were acquired at 100X using the software Leica Application Suite (3.2.0.9652). SP8 TCS HyD confocal microscope. Images were acquired using a 10X or a 63X (oil) objectives and lasers as indicated for each experiment below.

2.5. mtUPR imaging and quantification

Hsp-6p::gfp expressing worms were mounted on slides and imaged with Nikon eclipse 80i (unless otherwise mentioned) using a 10X objective. For image analysis, Regions of Interest (ROI) were selected around each worm and another fixed size background region. Fluorescence was quantified for each individual ROI using Fiji [27] and background fluorescence was subtracted from the worm measurements.

2.6. TMRE staining and quantification

TMRE staining and quantification were performed as previously described [21] with the following modifications: C. elegans larvae were transferred onto plates containing 0.1 μM TMRE (Thermo Life Sciences T669) 24 h before imaging, and imaged at L4 stage using 60x (oil) objective of Nikon eclipse 80i microscope. Images were cropped to a fixed region between the posterior pharynx and vulva, and a segmentation algorithm for mitochondria and the worm body respectively were trained using Ilastik [28]. The worm body segmentation was subsequently used in a custom made Fiji [27] (supplementary file 3) to identify the worm in the original image and remove non-specific signal in the field of view. Within the identified worm, the signal intensity of TMRE was measured on the original image in the previously segmented mitochondria. Schematic representation of the analysis process is described in Fig. S2.

2.7. Mitochondrial morphology

Adult worms were transferred onto NGM plates containing 0.4 mM MitoView™ Green (Biotium 70054) and F1 L4 larvae were used for imaging. Z-stacks were acquired in SP8 confocal microscope using a 63X (oli) objective and 488 nm laser in a fixed region between the posterior pharynx and vulva. Then, Z-projections of three Z-slices for either muscle or hypodermis sections were produced using the Fiji ‘Z project’ function. The body wall muscle was identified by the typical organization of four quadrants. A segmentation algorithm for mitochondria and the worm body respectively were trained using Ilastik [28]. The worm body segmentation was subsequently used in a custom-made Fiji [27] (supplementary file 4) to identify the worm in the original image and remove non-specific signal in the field of view. Within the worm, the segmented mitochondria were “skeletonized” using the “Analyze Skeleton” plugin in Fiji [29] and the following morphology parameters were measured: ‘Total Mitochondrial Area’ is the total mitochondrial area measured in the image. Dividing this measurement by worm area, yields ‘Total Mitochondrial Area Normalized to Worm Area’. ‘Area’, ‘Perimeter’, ‘Junctions’ and ‘Branches’ describe the average measurement per mitochondrion within each worm.

2.8. ROS measurements

Oxidative/reduced cytoplasmic state of the Grx-1-roGFP2 biosensor was conducted as previously described [20]. Briefly, L4 larvae of WT or ulp-2(tv380) genetic background expressing rpl-17p::Grx1-roGFP2 were imaged using a 10X objective in a confocal SP8 microscope. Worms were sequentially excited with 405 nm and 488 nm lasers set at a minimal intensity to avoid ROS induction and bleaching, and emission was detected through a 525/50 nm band pass filter. Transmitted light was captured simultaneously to obtain a bright-field image. The 405 nm excitation image was used to monitor the oxidized Grx1-roGFP2 state, and the 488 nm excitation image was used to monitor the reduced Grx1-roGFP2 state. Oxidative/reduced ratio of Grx-1-roGFP2 was calculated for both WT and mutant strains.

2.9. Mitophagy

Myo-3p::tomm-20::Rosella expressing worms were mounted on slides at L4 stage and imaged in a Nikon eclipse 80i microscope using a 20X objective. Each worm was imaged for GFP and DsRed fluorescence. DsRed was used to normalize for expression levels, and the pH sensitive GFP signal (GFP_SEP) was used to monitor mitophagy. ROI were selected manually using the Fiji ‘Threshold’ function [27], and the GFP_SEP/DsRed ratio was calculated for each individual ROI as previously described [30]. To avoid intestinal autofluorescence, images were cropped to a fixed region that included the worm body wall muscle cells of the head region until after the posterior pharynx as previously described [31].

2.10. mtDNA relative quantification

MtDNA relative quantification was conducted as previously described [32]. Briefly, five L4 worms of each genotype were picked and transferred into a 200 μl PCR tube containing 25 μl of worm lysis buffer (20 mM Tris pH 7.5, 50 mM EDTA, 50 mM EDTA, 200 mM NaCl, 0.5 % SDS) supplemented with proteinase K (1.25 mg/mL), and frozen in −80 °C. Immediately before qPCR run, samples were incubated at 65 °C for 1 h, followed by a 15 min incubation at 95 °C, and 70 μl of H₂O were added to each sample. Subsequently, 1 μl of each sample was used as a DNA template for RT-qPCR (see below), with ama-1 and atp-6 primers to quantify nuclear DNA (nDNA) and mtDNA content, respectively (see primer list). The level of mtDNA was calculated using a delta Ct (ΔCt) of average Ct of mtDNA and nDNA (ΔCt = Ct mtDNA-Ct nDNA) as an exponent of 2 (2ΔCt) and compared between samples.

2.11. Real-time qPCR

Each individual experiment included technical triplicates and experiments were independently repeated at least three times. To confirm gene knock-down using RNAi, plates containing mixed worm population were washed from the RNAi food using M9 buffer at least 4 times in 15 mL falcon tubes with a 40 min incubation time in M9 after the second wash. Then, worms were transferred to an eppendorf tube, spun down and 600 μl Trizol was added to the worm pellet, which was subsequently vortexed and frozen in −80 °C. RNA was extracted from the Trizol samples using the Direct-zol RNA miniprep Plus kit (Zymo Research). To compare endogenous mRNA levels between WT and mutant animals, the same procedure was followed with 300 L4 larvae from each sample. cDNA was synthesized using the Quantabio cDNA synthesis kit. Gene expression is shown as fold change compared to control and normalized to the ubiquitously expressed gene pmp-3. All real-time qPCR experiments were conducted with fast SYBR Green Master Mix (cat. no. 4385612, Applied Biosystems) on a StepOne Plus Real Time PCR System (cat. No. 4376600, Applied Biosystems).

2.12. Image processing

The Fiji [27] ‘Brightness/Contrast’ function was used to increase the brightness of 8-bit images for display where appropriate and was always applied with the same range for all fluorescent images in the same panel. In fluorescent images acquired in the Nikon microscope using the 20X and 60X objectives, fixed-size cropping of the region relevant to our measurements was performed as indicated for each experiment, and measurements were made on the cropped images.

2.13. Statistical analysis

Datasets were analyzed using GraphPad Prism (9.4.1) and were first tested for normality using Shapiro-Wilk normality test, and for equal variance using the F-test. Normally distributed datasets were analyzed using the unpaired t-test (two groups) or One-way ANOVA with Dunnett's post hoc test (multiple groups). Experiments where not all datasets followed a normal distribution were analyzed using the Mann-Whitney U test (two groups) or Kruskal-Wallis test with Dunn's post hoc test for multiple comparisons. Excluded outliers (ROUT, Q = 5 %) are labeled in blue in the supplementary tables which include the complete raw data for each experiment. n represents the number of worms tested.

3. Results

In a genome-wide RNAi screen for regulators of UPRmt, we found that ulp-2 is required for this stress response induced by defects in mitochondrial dynamics [18]. Specifically, ulp-2(RNAi) suppressed UPRmt induced by loss of function mutations in the mitochondrial fusion and fission genes [33] fzo-1(tm1133) and drp-1(tm1108) (Fig. 1A and B). To examine the extent by which ulp-2(RNAi) suppresses the UPRmt, we quantified GFP fluorescence of the hsp-6p::gfp (zcIs13) reporter used in the screen (Fig. S1; Table S1). We also quantified these mutants with the hsp-6p::gfp (bcSi9) reporter (Fig. 1A and B), this reporter has the advantage of a uniform fluorescence distribution that better reflects endogenous hsp-6 expression and smaller inter-individual variability [21]. Using both UPRmt reporters, ulp-2(RNAi) suppressed fzo-1(tm1133) induced UPRmt by approximately 30% compared to control RNAi (Fig. 1A and B; Table S2). Similar suppression of hsp-6p::gfp (bcSi9) was also observed in a drp-1(tm1108) mutant background (Fig. 1A and B). RNAi knockdown was verified using real-time quantitative PCR (RT-qPCR) (Fig. 1C; Tables S3 and S4). Interestingly, ulp-2(RNAi) suppressed hsp-6p::gfp (bcSi9) fluorescence even in the WT genetic background (Fig. 1A and B upper panel), suggesting that ULP-2 regulates hsp-6 expression also under conditions of WT, fzo-1 and drp-1 genetic background.

Fig. 1.

Downregulation of ulp-2 suppresses markers of the UPRmt. (A) hsp-6p::gfp (bcSi9) fluorescence in WT(+/+), drp-1(tm1108) and fzo-1(tm1133) genetic backgrounds, each following control(L4440), ulp-2 or atfs-1 RNAi treatments as indicated [atfs-1(RNAi) was used as a positive control]. (B) Quantification of hsp-6p::gfp (bcSi9) fluorescence presented in A (One-way ANOVA, ****P < 0.0001, n is indicated for each group). (C) Real-time qPCR quantification of ulp-2 or atfs-1 mRNA in the different genetic backgrounds following the RNAi treatments. (t-test was used to analyze all data except for hsp-6p::gfp following atfs-1(RNAi) and hsp-6p::gfp;drp-1(tm1108) and hsp-6p::gfp;fzo-1(tm1133) both following ulp-2(RNAi) where U test was used, ****P < 0.0001). (D) hsp-6p::gfp (bcSi9) fluorescence in WT(+/+) and ulp-2(tv380) genetic backgrounds following control(L4440) or cco-1 RNAi treatment. (E) Quantification of hsp-6p::gfp (bcSi9) fluorescence presented in D (One-way ANOVA, **P < 0.01; ****P < 0.0001, n is indicated for each group). (F) Real-time qPCR quantification of hsp-6 (U test, ****P < 0.0001) or hsp-60 (t-test, ****P < 0.0001) mRNA in WT(+/+) or ulp-2(tv380) genetic backgrounds. Data is from three biological repeats shown as mean ± SEM. Scale bar 100 μM.

In order to examine if the induction of UPRmt is also impaired in a ulp-2 deletion mutant, we crossed a ulp-2(tv380) null mutant allele with the hsp-6p::gfp (bcSi9) reporter. ulp-2(tv380) bears a 22bp deletion in exon 3, resulting in a premature stop codon[45]. We measured hsp-6p::gfp (bcSi9) fluorescence in WT and in the ulp-2(tv380) mutants following control or cco-1(RNAi) induced stress conditions. cco-1 encodes a predicted subunit of the mitochondrial electron transport chain (ETC), and its knockdown using RNAi induces UPRmt [34]. Similarly to ulp-2(RNAi), in ulp-2(tv380) homozygous worms, hsp-6p::gfp (bcSi9) expression was suppressed compared to WT worms in control conditions. While cco-1(RNAi) induced UPRmt in ulp-2(tv380) mutants, hsp-6p::gfp (bcSi9) expression remained lower compared to WT worms tested under the same conditions (Fig. 1D and E; Table S5). Finally, we measured the mRNA levels of endogenous hsp-6 and hsp-60 in WT and ulp-2(tv380) L4 staged worms using RT-qPCR and found both mitochondrial chaperones to be downregulated in the ulp-2 null mutant (Fig. 1F; Tables S6 and S7). Together this analysis demonstrated that ULP-2 is required for the induction of the key chaperones of the UPRmt system.

To gain insight into the functionality of the mitochondria in the absence of ULP-2, we measured mitochondrial membrane potential. We stained WT and ulp-2(tv380) worms with tetramethylrhodamine ethyl ester (TMRE) [35] under standard conditions and two stress conditions, cco-1(RNAi) and drp-1(tm1108). As expected, cco-1(RNAi) and drp-1(tm1108) exhibited decreased TMRE fluorescence, indicating lower mitochondrial membrane potential (Fig. 2A–D). Interestingly, ulp-2(tv380) also showed decreased TMRE fluorescence under standard culture conditions without further induction of stress, and this was unchanged or further decreased in cco-1(RNAi) and ulp-2(tv380);drp-1(tm1108), respectively (Fig. 2 A-D; Tables S8–S9).

Fig. 2.

Loss of ULP-2 leads to reduced mitochondrial membrane potential without affecting cellular redox potential. (A) TMRE fluorescence of representative worms of the indicated genotypes. (B) Quantification of TMRE fluorescence shown in A. Fluorescent values were normalized to WT(+/+) (Kruskal-Wallis test with Dunn's post hoc test, **P < 0.01; ***P < 0.001; ****P < 0.0001, n is indicated for each group). (C) TMRE fluorescence of representative WT(+/+) or ulp-2(tv380) worms following cco-1(RNAi) or L4440 empty vector control(RNAi). (D) Quantification of TMRE fluorescence shown in C. Values were normalized to WT(+/+) Control(RNAi). (Kruskal-Wallis test with Dunn's post hoc test, ****P < 0.0001. ns-not significant, n is indicated for each group). (E) Quantification of Grx1-roGFP2 fluorescence ratios of its oxidative to reduced state (λex405, λem580)/(λex490, λem580) in WT(+/+) and ulp-2(tv380) worms (U test, ns-not significant, n is indicated for each condition). Data is from at least three biological repeats, shown as mean ± SEM. Scale bar = 10 μM.

To examine if cellular redox potential is affected by the changes we detected in mitochondrial membrane potential, we crossed the ulp-2(tv380) mutant worms to the rpl-17p::Grx1-roGFP2 (jrIs2) biosensor, which specifically detects glutathione redox potential [20] that correlates with changes in the amount of endogenous ROS [36]. We found that the Grx1-roGFP2 oxidative/reduced state in these mutants did not differ significantly from WT worms (Fig. 2E; Table S10).

To better characterize this mitochondrial phenotype, we next measured additional mitochondrial properties related to mitochondrial function and network structure. We characterized the mitochondrial morphology of ulp-2(tv380) compared to WT worms using the potential independent dye, MitoView Green (Fig. 3; Tables S11–S12). In both hypodermis and muscle, ulp-2(tv380) had a significantly increased mitochondrial area compared to controls: Mitochondrial area of WT worms was 8.7 ± 0.65 μm² in the hypodermis and 10.36 ± 1.05 μm² in muscle, consistent with a previous report [37] while mitochondrial area of ulp-2(tv380) worms was 20.75 ± 2.48 μm² in the hypodermis and 22.47 ± 2.97 μm² in muscle (average ± SEM). Other mitochondrial morphology parameters that we tested, including total mitochondrial area, mitochondrial perimeter, junctions and branching were increased in ulp-2(tv380) worms compared to controls in the two tissues tested (Fig. 3; Tables S11–S12). Together, this analysis showed that ulp-2 knockout caused a significant expansion of mitochondrial networks both in the hypodermis and the body wall muscles.

Fig. 3.

Mitochondrial networks are expanded in the absence of ULP-2. (A–B) Upper panel: Confocal images of the hypodermis and muscle tissue respectively of WT(+/+) or ulp-2(tv380) worms stained with MitoView™ Green. Middle and Lower panel: Segmented and skeletonized images of the original images shown in the upper panel. (C) Insets of the white rectangles in the upper panel of A-B as indicated. (D–E) Quantifications of mitochondrial morphology parameters of the images presented in A-B respectively. Each dot on the graph represents the average measurements from a single worm. Definitions of morphology parameters are described in ‘Materials and Methods’ (all morphology parameters were analyzed using U test except for ‘Total Mitochondrial Area Normalized to Worm Area’ and muscle junctions and branches that were analyzed using t-test; n = 21,18 for WT and ulp-2(tv380) respectively, **P < 0.01; ***P < 0.001; ****P < 0.0001). Data is from at least three biological repeats shown as mean ± SEM. Scale bar = 10 μM.

To examine if the increased mitochondrial area is accompanied by increased mtDNA, we measured the relative mtDNA content of L4 staged ulp-2(tv380) and WT worms using RT-qPCR [32]. Indeed, we found that ulp-2(tv380) animals have significantly higher amounts of mtDNA (Fig. 4A; Table S13). We next examined if the increased mtDNA and mitochondrial area arise from increased mitochondrial biogenesis or decreased mitophagy. To assess if loss of ulp-2 increases mitochondrial biogenesis, we tested the expression level of ten mitochondrial biogenesis-related genes using qPCR in WT and ulp-2(tv380) L4 stage worms (Fig. 4B; Table S14). Interestingly, F32B4.2 and C47G2.3 which are predicted to function in protein import, were downregulated in the absence of ulp-2. In contrast, skn-1 and hmg-5, worm orthologs of mammalian NFE2 and TFAM that were previously implicated in mitochondrial biogenesis and mtDNA replication [[38], [39], [40]] and additional genes involved in mtDNA replication and maintenance (Fig. 4B) were upregulated. We then measured mitophagy using myo-3p::tomm-20::Rosella expressed in C. elegans muscle [10,31]. The Rosella is a double fluorophore composed of DsRed followed by SEP, a pH-sensitive GFP fluorophore [41]. GFP_SEP/DsRed ratio of worms treated with ulp-2(RNAi) was similar to control(RNAi), indicating that downregulation of ulp-2 does not affect basal mitophagy (Fig. 4C and D; Table S15). Sodium azide decreased GFP_SEP/DsRed ratio and was used as a control [30] (Fig. 4E and F; Table S16). In summary, this data suggests that ULP-2 deficiency compromises essential mitochondrial properties resulting in an expanded yet dysfunctional mitochondrial network without increase in mitophagy. Overall, we identified this SUMO protease as a major regulator of mitochondrial homeostasis and stress response.

Fig. 4.

Ulp-2 deficiency increases relative mtDNA and modifies mitochondrial biogenesis gene expression without altering basal mitophagy. (A) RT-qPCR of relative mitochondrial DNA (mtDNA) normalized to nuclear DNA (nDNA) of ulp-2(tv380) compared to WT(+/+) (U test, ****P < 0.0001). (B) RT-qPCR measurement for expression level of mitochondrial genes in WT(+/+) or ulp-2(tv380) worms. C. elegans genes are indicated followed by their mammalian orthologues. Genes are grouped according to their known function as indicated in the upper part of the graph but may have additional functions (U test, *P < 0.05, **P < 0.01, ***P < 0.001 ns-not significant). (C) Worms expressing the mitophagy biosensor myo-3p::tomm-20::Rosella in muscle tissue following control L4440(RNAi) or ulp-2(RNAi) as indicated. White rectangles are presented on the right as insets. (E) Positive control to C where worms expressing myo-3p::tomm-20::Rosella were treated with NaN3 or M9 buffer. White rectangles are presented on the right as insets. (D,F) Related to C and E respectively: Quantification of the GFP_SEP/DsRed fluorescence ratio in the indicated experimental conditions (t-test, ****P < 0.0001, ns-not significant, n is indicated for each group). For visualization purposes, images presented as GFP or DsRed were pseudo colored in Green or Magenta respectively. Data is from at least three biological repeats, shown as mean ± SEM. Scale bar = 10 μM.

4. Discussion

In this work, we established that ulp-2, encoding the ULP-2 SUMO protease, is a regulator of the UPRmt and mitochondrial homeostasis in C. elegans. These findings align with the growing evidence supporting a role for SUMOylation in these processes [8,10,42,43]. We demonstrated that downregulation of ulp-2 suppressed the expression of hsp-6p::gfp fluorescence with or without the induction of stress. Endogenous hsp-6 and hsp-60 expression were also downregulated in the ulp-2 mutant strain. A previous study by Gao et al. has identified ulp-4 as a regulator of UPRmt while ulp-2 was not identified as such [8]. This inconsistency may stem from differences in the RNAi protocols, as even low levels of ULP-2 expression result in significant differences in organismal phenotypes [45]. In addition, in Gao et al. study, a yeast two-hybrid candidate approach showed that ULP-4 but not ULP-2 interact with ATFS-1 and DVE-1. This finding together with the partial-overlapping subcellular localization of the two SUMO proteases (ULP-4 is localized to the mitochondria and cytosol while ULP-2 is localized to the cytosol and nucleus) ([[44], [45]]), suggest that ULP-2 regulates the mitochondria through other SUMO targets.

Furthermore, TMRE fluorescence was decreased in the ulp-2(tv380) genetic background under all the conditions tested. Previously, a correlation between reduced mitochondrial membrane potential and induction of UPRmt was established [46]. Here we revealed that downregulation of ULP-2 led to decreased mitochondrial membrane potential and UPRmt. Interestingly, downregulation of ULP-4 and ATFS-1 were also found to suppress both mitochondrial membrane potential and the UPRmt [7,8,44].

Simultaneous regulation of multiple targets by ULP-2 may underly why glutathione redox potential was unaffected by ULP-2 deficiency. Since mitochondrial membrane potential, mitochondrial morphology and possibly mitochondrial membrane lipid composition, as detailed bellow, are all affected in ulp-2 null mutants and can also affect mitochondrial ROS generation [[47], [48], [49], [50], [51]], it could be that these effects offset each other, possibly as a compensatory response, so that the overall glutathione redox potential, reflecting the cellular ROS burden, is maintained. Further studies are needed to elucidate the underling mechanism.

Analysis of mitochondrial morphology revealed an increase in mitochondrial area. We also observed an increase in the relative mtDNA content in ulp-2(tv380) worms. However, mitophagy was unaffected following downregulation of ulp-2. Expansion of mitochondrial networks in the muscle and hypodermis suggest an increase in biogenesis or maintenance of the mitochondria in these high energy demand tissues. Indeed we measured increased expression of genes related to mtDNA replication, including increased expression of hmg-5, the worm ortholog of mammalian TFAM supports a notion that the increased mtDNA found in the ulp-2 mutant are actively replicating [52]. This, together with increased expression of skn-1, the worm ortholog of mammalian NFE2, and increased expression of acl-3, the worm ortholog of mammalian TAFAZZIN which modifies cardiolipin, a mitochondrial-specific lipid, to its mature form [53], indicates the existence of mitochondrial biogenesis. However, we measured downregulation of several mitochondrial genes related to the mitochondrial import machinery, suggesting that mitochondrial biogenesis is yet incomplete in this mutant, which could underlie its dysfunction.

Recently, it has been demonstrated that atfs-1 mediates mitochondrial network expansion [7]. A gain of function mutation atfs-1(et18) with a constantly active UPRmt had expanded mitochondria and increased levels of mtDNA, while atfs-1 deficiency resulted in UPRmt suppression and decreased mitochondrial mass and mtDNA. However, both atfs-1(et18) mutant and atfs-1 deficient worms exhibited reduced respiration and increased expression of mtDNA replication genes. It was thus suggested that alternative stress response(s) exist that mediate this increase [7]. We propose that while suppressing UPRmt and expression of several mitochondrial genes, ulp-2(tv380) null mutants maintain an alternative compensation mechanism for their dysfunctional mitochondria by increasing mitochondrial area and mtDNA [54,55]. To summarize, our findings place ULP-2 as an important regulator of many aspects of mitochondrial homeostasis and stress response. Further studies will examine how SUMOylation regulates mtDNA replication and maintenance and its relation to mitochondrial membrane potential and UPRmt.

Data availability

Data is available upon request.

CRediT authorship contribution statement

Lirin Michaeli: Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Visualization, Writing – original draft. Eyal Spector: Formal analysis, Investigation, Methodology. Simon Haeussler: Conceptualization, Methodology. Cátia A. Carvalho: Conceptualization, Methodology. Hanna Grobe: Software. Ulrike Bening Abu-Shach: Investigation. Hen Zinger: Investigation. Barbara Conradt: Conceptualization, Resources, Writing – review & editing. Limor Broday: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing.

Declaration of competing interest

None.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

figs1.

ulp-2(RNAi) suppress UPRmt induced by disruption to mitochondrial fusion. (A) hsp-6::gfp(zcIs13) UPRmt reporter fluorescence in fzo-1(tm1133) genetic background following Control (L4440 empty vector), ulp-2 or atfs-1 RNAi treatments as indicated. (B) Quantification of hsp-6::gfp(zcIs13) fluorescence presented in A. Data is normalized to the control(RNAi) condition (One-way ANOVA, ****P<0.0001, n≥20 in each group). Data is one of three biological repeats, shown as mean ± SEM. Scale bar=100µM.

figs2.

A schematic representation of TMRE quantification. A raw image is converted to 8-bit (A), followed by segmentation of mitochondria using Ilastik (B). An additional segmentation is performed on the worm’s body (C) to avoid measuring signals outside of the worm. The ImageJ plugin integrates the 8-bit image with the segmented images to generate Regions Of Interest (ROI) only around the mitochondria that are found within the worm body while additional noise is removed by filtering out particles below a defined size. The plugin fills holes within the worm body segmentation and considers only the largest area while ignoring smaller isolated areas. The resulting ROIs (D) are measured for intensity. Scale bar: 10 μm.