Asymmetric Arginine Dimethylation Modulates Mitochondrial Energy Metabolism and Homeostasis in Caenorhabditis elegans

ABSTRACT

Protein arginine methyltransferase 1 (PRMT-1) catalyzes asymmetric arginine dimethylation on cellular proteins and modulates various aspects of biological processes, such as signal transduction, DNA repair, and transcriptional regulation. We have previously reported that the null mutant of prmt-1 in Caenorhabditis elegans exhibits a slightly shortened life span, but the physiological significance of PRMT-1 remains largely unclear. Here we explored the role of PRMT-1 in mitochondrial function as hinted by a two-dimensional Western blot-based proteomic study. Subcellular fractionation followed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis showed that PRMT-1 is almost entirely responsible for asymmetric arginine dimethylation on mitochondrial proteins. Importantly, isolated mitochondria from prmt-1 mutants represent compromised ATP synthesis in vitro, and whole-worm respiration in prmt-1 mutants is decreased in vivo. Transgenic rescue experiments demonstrate that PRMT-1-dependent asymmetric arginine dimethylation is required to prevent mitochondrial reactive oxygen species (ROS) production, which consequently causes the activation of the mitochondrial unfolded-protein response. Furthermore, the loss of enzymatic activity of prmt-1 induces food avoidance behavior due to mitochondrial dysfunction, but treatment with the antioxidant N-acetylcysteine significantly ameliorates this phenotype. These findings add a new layer of complexity to the posttranslational regulation of mitochondrial function and provide clues for understanding the physiological roles of PRMT-1 in multicellular organisms.

INTRODUCTION

Arginine methylation has drawn attention as a widespread posttranslational modification that often occurs in RNA-binding proteins, signaling molecules, DNA repair machinery, and transcriptional factors (1–5). This modification is classified according to the number and position of the methyl groups on a guanidino nitrogen of arginine residue, namely, monomethylarginine (MMA), asymmetric dimethylarginine (ADMA), or symmetric dimethylarginine (SDMA). A family of protein arginine methyltransferases (PRMT) is responsible for catalyzing first the formation of an MMA as an intermediate, and subsequently type I enzymes, including PRMT-1, -2, -3, -4, -6, and -8, further catalyze the generation of ADMA, while type II enzymes, including PRMT-5, PRMT-7, and FBXO11, catalyze the generation of SDMA (5). Among them, PRMT-1 is known to be the predominant type I enzyme in mammalian cells, whereas its physiological significance at the whole-body level has remained unclear due to embryonic lethality in the null mutants of mouse and Drosophila (6–8). Meanwhile, we have previously reported that the loss-of-function mutants of nematode PRMT-1 in Caenorhabditis elegans are viable (9). This finding led us to use C. elegans as an ideal model for further investigating the physiological functions of asymmetric arginine dimethylation in vivo.

Mitochondria play essential roles in cellular metabolism, whereby the tricarboxylic acid (TCA) cycle and subsequent oxidative phosphorylation (OXPHOS) convert energy from distinct electron donors to ATP. Organisms have evolved sophisticated systems to monitor mitochondrial functions and maintain homeostasis through communication between mitochondria and the nucleus (10). One of these mitochondrial systems is a conserved protein quality control termed mitochondrial unfolded protein response (UPR^mt), in which the accumulation of unfolded protein within the mitochondrial matrix leads to the transcriptional activation of nuclear genes encoding mitochondrial chaperones and proteases to help protein folding and assembly, as well as the clearance of defective proteins in stressed mitochondria (11–13). In C. elegans, disruption of mitochondrial function activates not only UPR^mt but also the xenobiotic-detoxification and pathogen response pathways, suggesting that animals interpret an impaired mitochondrial activity as xenobiotic exposure or pathogen attack (14–16). Actually, C. elegans displays bacterial avoidance behavior when mitochondria are inhibited by RNA interference (RNAi) or an antibiotic such as antimycin even in the absence of pathogen attack (15, 16). This intrinsic protective behavior, termed food avoidance, is also found in Drosophila melanogaster (17) and is reminiscent of mammalian gastrointestinal responses to enteric pathogens that trigger nausea symptoms (18).

Accumulating evidence has demonstrated that mitochondrial proteins undergo several posttranslational modifications, such as acetylation and phosphorylation (19, 20). For example, an estimated 35% of all mitochondrial proteins are acetylated, with the highest enrichment in OXPHOS, and tightly regulated by reversible acetylation (21). Interestingly, arginine methylation has recently emerged as a new regulatory posttranslational modification in mitochondrial proteins; namely, NDUFS2, a subunit of mitochondrial complex I, is symmetrically dimethylated at arginine 85 by a predicted methyltransferase, NDUFAF7, in the matrix of mitochondria, and this methylation is required for assembly of complex I (22). In addition, it has been also reported that NDUFAF7-induced symmetric dimethylation of NDUFS2 is essential for normal vertebrate development (23). On the other hand, the responsible enzyme for asymmetric arginine dimethylation on mitochondrial proteins and the biological roles of this major arginine methylation (5, 24) in mitochondrial functions remain largely unknown.

In this study, a group of mitochondrial proteins were identified as asymmetrically arginine-methylated proteins in vivo through a two-dimensional (2-D) Western blot-based proteomic analysis. Subcellular fractionation of C. elegans proteins uncovered that PRMT-1 is responsible for the formation of almost all asymmetric arginine dimethylation in mitochondria, thus allowing us to investigate the physiological significance of the organelle-scaled asymmetric arginine dimethylation on the mitochondrial function. In vitro ATP synthesis assays demonstrated that the loss of asymmetric arginine dimethylation results in decreased ATP synthesis, probably due to a compromised OXPHOS function, which is supported by a decreased oxygen consumption rate of prmt-1 mutants in vivo. This resulted in increased mitochondrial ROS generation in prmt-1 null mutants, accompanied with activation of stress responses, including UPR^mt. Furthermore, the mitochondrial dysfunction induced by the loss of asymmetric arginine dimethylation caused food avoidance behavior when worms were fed on nonpathogenic bacteria. Our findings provide evidence that asymmetric arginine dimethylation of mitochondrial proteins is required for normal ATP production and mitochondrial homeostasis.

RESULTS

Identification of in vivo substrates of PRMT-1 in C. elegans.

To further explore the physiological roles of asymmetric arginine dimethylation, we attempted to systematically identify in vivo substrates of PRMT-1. As we reported previously (9), nematode C. elegans PRMT-1 is the predominant type I arginine methyltransferase, and the prmt-1(ok2710) null mutant is viable, unlike lethal phenotypes of PRMT-1-deficient mouse and Drosophila (7, 8). We therefore employed C. elegans as an ideal animal model for screening in vivo substrates of PRMT-1 with the strategy of a two-dimensional (2-D) proteomic analysis (Fig. 1A) combining Sypro Ruby protein stain and Western blotting against two distinct asymmetric dimethylarginine antibodies, ASYM24 (9, 25, 26) and ASYM26.

FIG 1.

Two-dimensional (2-D) proteomic analysis combining Sypro Ruby protein stain and Western blotting with two distinct anti-ADMA antibodies. (A) Schematic workflow of 2-D Western blotting for identification of PRMT-1 substrate in C. elegans. Briefly, whole-cell extracts from wild-type N2 or prmt-1 mutant worms were subjected to 2-D PAGE and blotted with anti-ADMA antibodies. N2-specific spots were picked out from Sypro Ruby-stained gels and identified by matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS). (B) Whole-cell extracts from N2 or prmt-1 mutant worms on adult day 1 were separated by 2-D PAGE, followed by Western blotting with ASYM24 (top panels) or ASYM26 (middle panels) antibodies. After merging with Sypro Ruby-stained gels (bottom panels), N2-specific spots detected with anti-ADMA antibodies (indicated by arrows and numbers) were manually picked out for MALDI-TOF MS analysis. MW, molecular weights (in thousands).

Two-dimensional electrophoresis (2-DE) followed by Western blotting against ASYM24 (Fig. 1B, top panels) or ASYM26 (Fig. 1B, middle panels) revealed 10 spots of proteins containing ADMA in N2 samples compared to those of prmt-1 mutants as a negative control. Among them, spots 1, 2, 3, and 4 were detected by both antibodies, while the others were specific to either ASYM24 or ASYM26 antibody. This result appears to be due to the difference in the immunogens of the two antibodies, which in turn broadens the spectrum of proteins allowed to be detected. Next, to identify the proteins detected by Western blotting, the same samples separated by 2-DE were stained with Sypro Ruby fluorescent dye (Fig. 1B, bottom panels), and then these images were overlapped with the images of the Western blot. We found that 9 of 10 ADMA spots have their corresponding spots on the Sypro Ruby staining and also show no significant change in their intensity between N2 and prmt-1 mutant worms. These data suggest that the nine spots are bona fide candidates of ADMA modification in C. elegans.

The results of protein identification by matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) are shown in Table 1. Surprisingly, the subcellular localizations of the nine proteins span from extracellular space, cytoplasm, and nucleus to cellular organelles, including endoplasmic reticulum and mitochondria, indicating that asymmetric arginine dimethylation may be involved in a broad spectrum of cellular processes in vivo. Typically, the identification of three mitochondrial proteins, HSP-60, ATP-2, and Y38F1A.6, indicates that mitochondrial proteins may be representative of a large group of PRMT-1 substrates in vivo.

TABLE 1.

Identification of PRMT-1 in vivo substrates in C. elegans^a

Spot no.	Protein	Homolog	Fragment coverage (%)	MASCOT probability score	UniProtKB entry	Protein mass (kDa)	Subcellular localization
1	F41H10.3, isoform d	Mucin-2	35	59	V6CLM1	86	Exc
2	NHR-12, isoform c	HNF4γ	41	60	I2HAH0	52	Nu
3	RPL-4	RPL4	66	100	O02056	39	Rb
4	RPS-4	RPS4X	72	94	Q9N3X2	29	Rb
5	HSP-60	HSPD1	53	88	P50140	60	Mt
6	ACT-4, isoform c	ACTB	61	105	Q6A8K1	40	Cy
7	CRT-1	CALR	58	121	P27798	46	ER
8	ATP-2	ATP5B	43	157	P46561	58	Mt
9	Y38F1A.6	HOT	55	75	Q9U2M4	51	Mt

^aPRMT-1 substrate proteins revealed by 2-D WB were identified by mass spectrometry analysis of tryptic fragments from gel spots. MASCOT probability scores were generated from tryptic fragment coverage from MALDI-TOF MS of the identified proteins. A probability score of >57 was considered a significant match (P < 0.05). Exc, extracellular; Nu, nucleus; Rb, ribosome; Mt, mitochondrion; Cy, cytoplasm; ER, endoplasmic reticulum.

We have previously reported that human PRMT-1 regulates the mitochondrial apoptosis signaling by methylating BAD protein (27), suggesting potential roles of PRMT-1 in mitochondrion-related cellular functions. However, there was no direct evidence regarding the functional significance of PRMT-1-mediated asymmetric arginine dimethylation in mitochondria. Therefore, the results of proteomic analysis (Table 1) prompted us to further investigate the biological roles of this modification on mitochondrial function.

PRMT-1 is the predominant type I PRMT for mitochondrial proteins.

To evaluate the contribution of PRMT-1 to asymmetric arginine dimethylation of overall mitochondrial protein in vivo, we performed subcellular fractionation of mitochondria from C. elegans and quantified the amounts of ADMA in proteins by using liquid chromatography-tandem mass spectrometry (LC-MS/MS). Efficient isolation of light and heavy mitochondria (LM and HM) from total extracts of wild-type and prmt-1 mutants was verified by Western blotting with organelle markers (Fig. 2A). Notably, LC-MS/MS analysis demonstrated that prmt-1 mutants have extremely low levels of ADMA in both mitochondrial fractions compared with the level in N2 fractions (Fig. 2B, left). In contrast, levels of symmetric dimethylarginine (SDMA) were increased in light but not heavy mitochondrial fractions (Fig. 2B, right), which is partially consistent with previous observations in mouse models (24, 28). Taken together, these results indicate that PRMT-1 serves as the major enzyme catalyzing asymmetric arginine dimethylation in mitochondria, and the loss of PRMT-1 alters the dynamics of SDMA formation especially in the light mitochondrial fraction. More importantly, the prmt-1 null mutant worms serve as an ideal model that allows us to investigate the physiological significance of the organelle-scaled asymmetric arginine dimethylation on mitochondrial function.

FIG 2.

PRMT-1 is required for asymmetric arginine dimethylation of mitochondrial proteins in C. elegans. (A) Mitochondrial fractionation of whole-cell extracts from N2 or prmt-1 mutant worms. Each fraction was blotted with antibodies against ATP-2 (mitochondrial marker), H3 (nuclear marker), tubulin (cytoplasmic marker), or PRMT-1. TL, total lysate; LM, light mitochondria; HM, heavy mitochondria; PMS, postmitochondrial supernatant. (B) Determination of ADMA and SDMA levels in mitochondrial fractions. Fifty micrograms of light or heavy mitochondrial proteins from N2 or prmt-1 mutant worms were subjected to acidic hydrolysis and then analyzed with LC-MS/MS.

Loss of prmt-1 causes impaired ATP synthesis in mitochondria.

It has been shown that mammalian PRMT-1 facilitates mitochondrial biogenesis through methylation of peroxisome proliferator-activated receptor γ coactivator 1α (PGC-1α) (29). In C. elegans, however, we did not observe a significant reduction in either mitochondrial mass or mRNA levels of mitochondrial genes, including cyc-1, cco-1, and atp-5 in prmt-1 mutant (see Fig. S1 in the supplemental material). A plausible explanation for this difference may be that PGC-1α is not conserved in C. elegans. Next, since mitochondria play central roles in cellular metabolism, we assessed the metabolites of N2 and prmt-1 mutant worms using the capillary electrophoresis-time of flight mass spectrometry (CE-TOFMS)-based metabolomic analysis. Based on a comparison of the identified 208 metabolites shown in a heatmap, we discovered that a global metabolic suppression, including glycolysis and the TCA cycle, occurs in prmt-1 mutants (see Fig. S2A to C in the supplemental material). In particular, the loss of prmt-1 resulted in a decreased ATP level to a greater extent than that of the ADP level (Fig. S2D), suggesting a possible compromise in the energy-producing function of mitochondria, presumably due to a loss of asymmetric arginine dimethylation on a large number of mitochondrial proteins.

To directly investigate the functional significance of asymmetric arginine dimethylation in mitochondria, we attempted to assess ATP synthesis activity using isolated mitochondria fractions. The same amounts of mitochondrial fractions were prepared from N2 and prmt-1 mutant worms and then subjected to in vitro ATP assays with a supply of intermediate metabolites, including pyruvate, succinate, glutamate, palmitoyl-carnitine, and malate as the substrates (Fig. 3A and B). Compared with the N2 control, mitochondria from prmt-1 mutant worms exhibited a decrease in ATP production even when any substrate was supplied and at any time point (Fig. 3C to F). Given that each substrate should donate electrons to the electron transfer chain via different routes but were all converted to ATP through oxidative phosphorylation (OXPHOS), our data imply that the decreased ATP production of prmt-1 mutants may be attributed to the defects in OXPHOS activity, likely due to the loss of asymmetric arginine dimethylation on mitochondrial proteins.

FIG 3.

Loss of prmt-1 causes impaired ATP synthesis in mitochondria. (A) Isolation of mitochondrial fractions (mito.) for in vitro ATP assay. Isolated mitochondrial fractions from N2 or prmt-1 mutant worms were blotted with anti-H28O16.1 (ATP synthase alpha subunit) or anti-ATP-2 antibodies. (B) Supply of different substrate couples to oxidative phosphorylation. The substrates used in in vitro ATP assays are highlighted in gray. Glu, glutamate; succi, succinate; α-KG, α-ketoglutarate; fuma, fumarate; C I to C V, complexes I to V; ETF, electron transfer flavoprotein; cyc c, cytochrome c. (C to F) ATP synthesis of isolated mitochondria is compromised in prmt-1 null mutants (filled bars) compared to N2 (open bars). Different combinations of substrates were supplied, and ATP synthesis was measured by in vitro ATP assay. Error bars indicate ±SEM for three independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Loss of prmt-1 decreases OXPHOS activity in mitochondria.

We next tested whether the mitochondrial OXPHOS function is actually compromised in prmt-1 mutants. Since mitochondria use oxygen as the final acceptor in the electron transfer chain during aerobic respiration, the oxygen consumption rate (OCR) provides important information for OXPHOS function. We therefore compared the OCRs of N2 and prmt-1 mutants in living worms by measuring the extracellular oxygen flux and found that prmt-1 mutants exhibited a significant decrease in the steady-state OCR, which reflects the basal respiration (Fig. 4A). Furthermore, we also evaluated respiratory capacity by treatment with the mitochondrial uncoupler FCCP (carbonilcyanide p-trifluoromethoxyphenylhydrazone), which uncouples the electron transfer and ATP synthesis, thus maximizing the proton pumping. As shown in Fig. 4B, a similar decrease was observed in prmt-1 mutants, indicating a compromised function of electron transfer chain. Together, the reduction in both basal respiration and respiratory capacity provides direct evidence for a suboptimal OXPHOS function in prmt-1 mutants.

FIG 4.

Loss of prmt-1 decreases oxygen consumption rate in C. elegans. Basal (A) and maximum (B) oxygen consumption rates were measured in N2 and prmt-1 mutant worms. Error bars indicate ±SEM for four independent experiments. *, P < 0.05; **, P < 0.01.

Loss of prmt-1 results in increased ROS production in mitochondria.

Abnormal OXPHOS often couples with the decreased mitochondrial membrane potential and subsequent electron leak, which in turn leads to reactive oxygen species (ROS) production (30–32). To ascertain whether prmt-1 mutants increase the ROS production in mitochondria, we used the mitochondrial superoxide indicator MitoSOX for in vivo imaging. As shown in Fig. 5, mitochondrial ROS was selectively detected in the prmt-1 mutant. In agreement with this result, the metabolomic analysis revealed a decrease in glutathione (GSH) and an increase in glutathione disulfide (GSSG) levels, indicating a disturbed redox balance in prmt-1 mutants (Fig. S2E). Importantly, the revertant prmt-1; Ex71[prmt-1(+)] strain, which fully restores the amount of ADMA (9), exhibited an elimination of mitochondrial ROS, whereas another revertant strain, harboring enzymatically inactive prmt-1G70A, the prmt-1; Ex84[prmt-1G70A(+)] strain, failed to attenuate the increase of ROS production in the prmt-1 null background. In contrast to PRMT-1, depletion of PRMT-5, the predominant type II arginine methyltransferase in C. elegans (33), had no effect on mitochondrial ROS levels (see Fig. S3 in the supplemental material). Considering the results of in vitro ATP assays (Fig. 3) and in vivo OCR (Fig. 4), these findings suggest that asymmetric arginine dimethylation by PRMT-1 contributes to normal OXPHOS activity in C. elegans.

FIG 5.

PRMT-1 prevents mitochondrial ROS production in an enzymatic activity-dependent manner. N2, prmt-1, prmt-1; Ex71[prmt-1(+)], or prmt-1; Ex84[prmt-1G70A(+)] worms on adult day 1 were stained with the mitochondrial superoxide indicator MitoSOX and observed with a fluorescence microscope. Representative microscopy images (original magnification, ×400) are shown. DIC, different interference contrast.

The mitochondrial unfolded protein response is elicited in prmt-1 mutants.

Elevated ROS levels damage macromolecules and perturb the homeostasis of the internal environment of mitochondria, thus triggering the mitochondrial stress. Mitochondrial unfolded protein response (UPR^mt) is an evolutionarily conserved system that confers resistance to mitochondrial stresses by upregulating mitochondrion-specific chaperones, such as heat shock protein 6 (HSP-6) or HSP-60 (16, 34). To investigate whether loss of asymmetric arginine dimethylation on mitochondrial proteins causes stress in mitochondria, we employed a highly sensitive reporter strain, the zcIS13[Phsp6::gfp] strain, that induces hsp-6 promoter-driven green fluorescent protein (GFP) expression specifically responding to mitochondrial stresses (16, 35). As shown in Fig. 6A, knockdown of prmt-1 led to an induction of GFP in the pharynx and tail compared with control knockdown. The activation of the hsp-6 promoter by prmt-1 knockdown was confirmed by Western blotting with GFP antibody (Fig. 6B). Importantly, pretreating with the antioxidant N-acetylcysteine (NAC) inhibited the mitochondrial ROS production in prmt-1 mutant as shown by MitoSOX staining (Fig. 6C) and markedly attenuated GFP expression in the pharynx of the zcIS13 reporter strain when prmt-1 was depleted (Fig. 6D). These observations clearly show that loss of PRMT-1 induces mitochondrial stress, at least partially as a result of disruption of redox homeostasis in C. elegans.

FIG 6.

Loss of prmt-1 induces an unfolded protein response in mitochondria. (A and B) Knockdown of prmt-1 induces GFP expression in the mitochondrial stress reporter zcIS13[Phsp6::gfp] strain. Feeding RNAi was carried out with control L4440 or prmt-1 RNAi vectors from the L4 stage. GFP expression was observed in the F1 generation with a fluorescence microscope (A) and confirmed by Western blotting with anti-GFP antibody (B). (C) Mitochondrial ROS in prmt-1 mutants was reduced by treatment with the antioxidant. prmt-1 mutants grown on H₂O- or NAC-supplemented plates were stained with MitoSOX and observed with a fluorescence microscope. (D) NAC treatment attenuates UPR^mt induced by prmt-1 knockdown. prmt-1 knockdown was performed on H₂O- and NAC-supplemented RNAi plates. GFP expression was observed with a fluorescence microscope. Arrowheads indicate the pharynx bulbs.

In addition, gene ontology analysis of DNA microarray of N2 and prmt-1 mutants revealed a global upregulation of genes, especially those associated with metabolic processes and mitochondrial functions (see Fig. S4 in the supplemental material), suggesting that prmt-1 mutants exhibit the adaptive responses to mitochondrial dysfunction.

Loss of enzymatic activity of prmt-1 induces food avoidance behavior.

In general, C. elegans worms are raised on bacterial lawns on agar plates and stay there for feeding. It is known that, however, this normal feeding behavior is disrupted by impairment in mitochondrial respiration, because C. elegans interprets mitochondrial dysfunction as xenobiotic exposure or pathogen attack and thereby tends to avoid food intake even if the food does not contain pathogenic bacteria (15, 16). Indeed, quantitative reverse transcription (RT)-PCR demonstrated an increase in stress response gene expression, including xenobiotic detoxification and pathogen response in prmt-1 mutants (see Fig. S5 in the supplemental material). Thus, we investigated the food avoidance phenotype of the prmt-1 mutant for evaluating mitochondrial function. As shown in Fig. 7A and B, prmt-1 mutants exhibited a dispersed feeding behavior in contrast to wild-type N2. It should be noted that this phenotype was similar to that caused by depletion of isp-1 (Fig. 7A and B), which encodes a component of the mitochondrial complex III in the electron transport chain, and a loss-of-function mutant of isp-1 has been shown to increase ROS production and activate UPR^mt (16, 30, 32).

FIG 7.

Loss of enzymatic activity of prmt-1 induces food avoidance behavior. (A) prmt-1 mutant and isp-1 knockdown worms exhibit food avoidance behavior in contrast to controls. Representative images of N2 or prmt-1 mutants (left) and N2 fed on control L4440 or isp-1 RNAi plates (right). (C) The food avoidance phenotype exhibited in prmt-1 mutant is rescued by enzymatically active, but not inactive, prmt-1 transgenic revertant. Representative images of N2, prmt-1, prmt-1; Ex71[prmt-1(+)], or prmt-1; Ex84[prmt-1G70A(+)] worms. Arrowheads indicate the off-lawn worms. (E) The food avoidance phenotype exhibited in prmt-1 mutants is suppressed by silencing of surveillance genes for mitochondrial stress. Representative images of N2 or prmt-1 mutant worms fed on control L4440, inx-17, or thoc-2 RNAi plates are shown. Arrowheads indicate the off-lawn worms. (B, D, F) Quantified results of the food avoidance phenotype illustrated in panels A, C, E, respectively. In panel F, N2 and the prmt-1 mutant are represented by open bars and filled bars, respectively. Error bars indicate ±SEM for three independent experiments. n.s., P > 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Next, we tested whether the food avoidance behavior results from the loss of asymmetric arginine dimethylation in mitochondria by using the two transgenic strains, the prmt-1; Ex71[prmt-1(+)] and prmt-1; Ex84[prmt-1G70A(+)] strains. As shown in Fig. 7C and D, the food avoidance phenotype of prmt-1 mutants was counteracted in the wild-type worms but not in those with the enzymatic inactive revertant of prmt-1, indicating that the asymmetric arginine dimethylation in mitochondria is required for the normal feeding behavior.

To further determine whether the food avoidance phenotype of prmt-1 mutant is indeed due to mitochondrial dysfunction, we performed epistasis analysis with RNAi against two surveillance genes, inx-17 and thoc-2, that have been identified to be involved in food avoidance downstream of mitochondrial dysfunction (16). As expected, silencing inx-17 or thoc-2 in prmt-1 mutants dramatically suppressed the food avoidance phenotype (Fig. 7E and F). These findings suggest that the food avoidance behavior of prmt-1 null mutants can be attributed to mitochondrial dysfunction.

NAC treatment ameliorates the food avoidance behavior of prmt-1 mutant.

Given that the loss of prmt-1 activity leads to an increase in mitochondrial ROS production and thereby perturbs the homeostasis of the internal environment of mitochondria (Fig. 5 and 6), we finally asked whether the food avoidance phenotype exhibited by prmt-1 null mutants could be cancelled by a ROS scavenger. We therefore performed food avoidance assays in the presence or absence of NAC. As expected, we found that NAC treatment significantly ameliorates the dispersed feeding behavior of prmt-1 mutants (Fig. 8). This result indicates that the excessive ROS production caused by loss of prmt-1 is largely involved in the food avoidance phenotype in C. elegans.

FIG 8.

NAC treatment cancels the food avoidance behavior of prmt-1 mutant. (A) Antioxidant treatment partially rescued the food avoidance phenotype in prmt-1 mutants. Representative images of N2 or prmt-1 mutant animals raised on water or NAC-supplemented assay plates are shown. Arrowheads indicate the off-lawn worms. (B) Quantified results of the food avoidance phenotype illustrated in panel A. N2 and the prmt-1 mutant are represented by open bars and filled bars, respectively. Error bars indicate ±SEM for three independent experiments. **, P < 0.01.

DISCUSSION

In this study, we used C. elegans as a suitable multicellular model for screening in vivo substrates of PRMT-1 with the strategy of 2-D proteomic analysis based on Western blotting against two distinct ADMA antibodies (Fig. 1A). We identified nine proteins modified by asymmetric arginine dimethylation in vivo (Fig. 1B and Table 1). Of nine substrates, a nuclear receptor NHR-12 isoform c is the ortholog of mammalian HNF4γ, which has been known to be regulated by PRMT-1 in a methylation-dependent manner (36). Additionally, the identification of ribosomal RPL-4 and RPS-4 leads us to assume a similarity to recent findings that arginine methylation of mammalian RPS2 and RPS3 is involved in ribosomal assembly and biogenesis (37–39). Thus, our present results raise the possibility that asymmetric arginine dimethylation plays indispensable roles in transcriptional and translational regulation spanning a wide range of organisms.

Intriguingly, as demonstrated by the subcellular fractionation (Fig. 2A), PRMT-1 is not localized in the mitochondria of C. elegans, and we have also observed a similar result in human cells (data not shown). These results suggest that, unlike other posttranslational modifications such as phosphorylation and acetylation (19, 20), asymmetric arginine dimethylation of mitochondrial proteins does not occur within the mitochondria. Alternatively, given that over 99% of mitochondrial proteins are encoded by the nuclear genome, they appear to be methylated simultaneously with translation in the cytoplasm before mitochondrial import. If so, it would add another layer of regulation to mitochondrial homeostasis, potentially by modulating the transport of mitochondrial proteins in response to the intracellular environment.

Loss of asymmetric arginine dimethylation in mitochondria results in decreased ATP synthesis activity (Fig. 3C to F), indicating an impaired OXPHOS function, which is further supported by the results that prmt-1-deficient worms exhibit a lower basal respiration rate and smaller respiratory capacity than wild-type worms (Fig. 4). Remarkably, a global mass spectrometric analysis of Trypanosoma brucei, an early-branching single-cell eukaryote, has reported that 54 and 13 mitochondrial proteins are asymmetrically and symmetrically dimethylated, respectively, and many of them are implicated in OXPHOS function (26). In addition, it is notable that ATP-2, a key component of ATP synthase, is identified as a substrate of PRMT-1 in the current study. Considering these data together with our current findings, the compromised OXPHOS function of the prmt-1 null mutants may be attributed to the loss of asymmetric arginine dimethylation on overall mitochondrial substrate proteins, including ATP-2. In this meaning, our study for the first time bridged the asymmetric arginine dimethylation of mitochondrial proteins to fundamental mitochondrial function.

We provided evidence that the enzymatic activity of PRMT-1 contributes to suppress mitochondrial ROS generation (Fig. 5), probably by maintaining the OXPHOS function, while knockdown of PRMT-5, the major type II arginine methyltransferase in C. elegans (33), had no effect on that function (see Fig. S3 in the supplemental material). These results imply that symmetric arginine dimethylation, at least by PRMT-5, is not necessary for normal OXPHOS function. However, instead of PRMT-5, the noncanonical methyltransferase NDUFAF7 included in the matrix of mitochondria has been shown to methylate NDUFS2, a subunit of mitochondrial complex I, and thereby serve to assemble complex I (22), suggesting that, unlike the asymmetric one, symmetric arginine dimethylation on mitochondrial proteins is regulated by a previously unidentified type II arginine methyltransferase(s) localized in mitochondria. Further investigations are needed to determine the functional difference between asymmetric and symmetric arginine dimethylation on mitochondrial proteins.

As a behavioral response to the mitochondrial dysfunction of prmt-1 mutants, we observed a food avoidance behavior (Fig. 7), which is thought to be a defensive phenotype to protect core cellular functions, including mitochondrial respiration (15), because C. elegans living in a microbe-rich environment is persistently attacked by pathogenic bacteria, with their mitochondria easily targeted as a valuable source of iron (40, 41). Meanwhile, even when food does not contain pathogenic bacteria, mitochondrial dysfunction leads to food avoidance behavior, possibly because the mitochondrial defects are interpreted as an attack by a pathogen (16). In agreement with this idea, we found an upregulation of genes involved in xenobiotic detoxification and natural immune response in prmt-1 null mutants (Fig. S4). Furthermore, genetic analysis demonstrated that food avoidance of prmt-1 null mutants can be blocked by silencing either thoc-2 or inx-17, both of which have been recently identified as mitochondrial surveillance genes (16). THOC-2 has been shown to be an mRNA-binding protein facilitating the nuclear export of mRNA and required for proper neuronal development (42), while the function of INX-17, which is expressed abundantly in interneurons, is still unknown. Although the detailed mechanisms underlying food avoidance by mitochondrial dysfunction remain unclear, it is possible that the loss of asymmetric arginine dimethylation induces this defensive behavior through the same pathway and neuronal circuits as those induced by inhibition of OXPHOS components (e.g., isp-1), because the stress responses triggered in the two events are similar (30–32). Importantly, the depletion of atp-2 or hsp-60 in C. elegans induces UPR^mt (43, 44) and food avoidance behavior (15), strongly indicating that PRMT-1 serves as a positive regulator of mitochondrial function by methylating its substrates, including ATP-2 and HSP-60.

In conclusion, we identified in vivo substrates of C. elegans PRMT-1 at the whole-body level. This is the first evidence that links asymmetric arginine dimethylation of mitochondrial proteins to ATP synthesis activity and mitochondrial homeostasis. Our study adds new evidence to the diverse regulations of mitochondrial proteins by posttranslational modifications. It is speculated that arginine methylation, like phosphorylation or acetylation, may represent a strategy employed by eukaryotic cells to regulate the mitochondria, originally engulfed as alphaproteobacteria (10), to meet the cellular energy demands.

MATERIALS AND METHODS

C. elegans strains.

All strains were maintained and handled according to standard methods. Briefly, nematodes were maintained on NGM OP50 (nematode growth medium with Escherichia coli strain OP50) plates at 20°C. Bristol N2 wild-type, RB2047 prmt-1(ok2710), and SJ4100 zcIs13[hsp-6::GFP] worms were obtained from the Caenorhabditis Genetics Center, and the prmt-1(ok2710) mutant was outcrossed to N2 three times (43). The extrachromosomal prmt-1 wild-type and G70A mutant-overexpressing lines in the prmt-1(ok2710) background were TKB71 trcEx71[prmt-1(+); Pmyo2::DsRed]; prmt-1(ok2710) and TKB84 trcEx84[prmt-1G70A(+); Pmyo2::DsRed]; prmt-1(ok2710), respectively (9).

Two-dimensional Western blotting and protein identification.

Whole-body extracts from N2 or prmt-1 mutant worms were obtained by homogenizing 20 ml of wet-packed worms in 2-DE buffer {7 M urea, 2 M thiourea, 30 mM Tris, 4% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), 0.01% Triton X-10, 60 mM dithiothreitol (DTT)} with a protease inhibitor cocktail (Nacalai Tesque) using sonication. After centrifugation, proteins were precipitated with methanol-chloroform and resuspended in 2-DE buffer. Three hundred micrograms of protein was separated on a 75-mm, pH 5 to 8, agarose stick gel, followed by a 10% SDS-PAGE gel. The resulting gels were either stained with Sypro Ruby (BioRad) fluorescent dye or transferred to a polyvinylidene difluoride (PVDF) membrane and blotted with ADMA antibodies (ASYM24 [Merck Millipore] and ASYM26, [EpiCypher]). N2-specific Western blotting spots were picked out from Sypro Ruby gel after merging the two staining results.

Detected PRMT-1 substrates were identified with MALDI-TOF MS according to a protocol previously described (48). Stained protein spots were excised and dehydrated with acetonitrile and then reduced and alkylated with DTT and iodoacetamide before tryptic digestion in situ. Tryptic peptides were desalted by binding and then elution from a C₁₈ ZipTip (Millipore) and then mixed with an α-cyano-4-hydroxycinnamic acid (Wako) matrix and spotted onto a stainless steel plate. The dried spots were analyzed by MALDI-TOF MS (ultrafleXtreme; Bruker). The resulting peptide peaks were manually picked up, and the masses of the peptides were used for protein identification through MASCOT peptide mass fingerprint matching. MASCOT scores larger than 57 are considered significant.

Feeding RNAi.

For RNAi constructs against prmt-1 and isp-1, the 500 bp of each coding sequence was amplified by PCR and cloned into L4440 vector. The following primers were used: prmt-1 RNAi, forward (5′-ACCTTACCATGGCAGCTTCTCCGTGACAA-3′) and reverse (5′-GTATTCTGCAGCTTGGGGGTGTTTCGA-3′); isp-1 RNAi, forward (5′-ACCTTACCATGGCAGCTTCTCCGTGACAA-3′) and reverse (5′-GTATTCTGCAGCTTGGGGGTGTTTCGA-3′). The RNAi clones against prmt-5, inx-17, and thoc-2 were obtained from the Ahringer library. All RNAi constructs were transformed into E. coli HT115 and then seeded onto NGM plates containing 25 μg/ml carbenicillin and 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) for feeding RNAi experiments.

Mitochondrial fractionation.

Worms were allowed to grow until the young adult stage and harvested with M9 buffer. After clearance with sucrose flotation, worms were homogenized by using Dounce homogenizer in 5 volumes of mitochondrial isolation buffer (MIB; 210 mM mannitol, 70 mM sucrose, 0.1 mM EDTA, 5 mM Tris, and protease inhibitor cocktail). The homogenates were centrifuged at 1,000 × g for 10 min. The supernatants were saved, and the pellets were resuspended in the same volume of MIB and homogenized again. After centrifugation at 1,000 × g for 10 min, an aliquot of the combined supernatants was saved as total lysate (TL). The rest of the supernatants were centrifuged at 5,000 × g for 10 min. The pellets were washed twice by resuspension in 1 ml MIB and centrifuged at 2,000 × g for 10 min. After the combined supernatants were centrifuged again at 2,000 × g for 10 min, the heavy mitochondria (HM) were pelleted by centrifuging the supernatants at 15,000 × g for 10 min and resuspended in MIB. The supernatants from the 5,000 × g centrifugation were further centrifuged at 15,000 × g for 10 min, and the resulting supernatants were saved as postmitochondrial supernatants (PMS). The pellets were washed by resuspension in MIB and centrifugation at 5,000 × g for 10 min. The light mitochondria (LM) were pelleted by centrifuging again at 15,000 × g for 10 min and resuspended in MIB.

The successful mitochondrial fractionation was validated by Western blotting with anti-ATPB (ab14730; Abcam), anti-histone H3 (number 4499; Cell Signaling Technology), or anti-α-tubulin (T5168; Sigma-Aldrich). C. elegans PRMT-1 was detected with a rabbit polyclonal PRMT-1 antibody previously described (9). The mitochondria used for in vitro ATP assays were isolated according to a simplified protocol as previously described (45).

Acid hydrolysis and LC-MS/MS.

Acid hydrolysis of worm proteins was performed essentially as previously described (33). Briefly, after mitochondrial fractionation, each fraction of worm proteins (50 mg) was hydrolyzed with 6 N HCl at 110°C for 24 h in glass vials. N-propyl-l-arginine (N-PLA; Sigma-Aldrich) was added as the internal control.

ADMA or SDMA was quantified using a Shimadzu Nexera ultrahigh-pressure liquid chromatography system coupled to an LCMS-8050 triple-quadrupole mass spectrometer (Shimadzu, Kyoto, Japan) as described previously (28). LC separation was performed on a SeQuant ZIC-HILIC column (2.1 by 150 mm; Merck KGaA, Darmstadt, Germany) with a SeQuant ZIC-HILIC guard fitting apparatus (1.0 by 14 mm; Merck KGaA) at 40°C. Acetonitrile-water-formic acid at 1:98:1 (A) and acetonitrile-water-formic acid at 98:1:1 (B) were used as mobile phases; a water-dissolved sample (1 μl) was injected and eluted with the followed gradient elution program: 95% B for 1 min, shifted to 5% in 9 min, and kept at 5% for 4 min. The system was allowed to equilibrate for 7 min at 95% B prior to the next analysis. The flow rate was 0.2 ml/min.

Mass spectroscopy operating conditions were optimized as follows: interface voltage, 4.00 kV; interface current, 0.55 μV; interface temperature, 300°C; desolvation line temperature, 250°C; heating block temperature, 400°C; drying gas (N₂), 10 liters/min; nebulizing gas (N₂), 3 liters/min; heating gas (air), 10 liters/min; collision-induced dissociation gas (argon), 230 kPa; conversion dynode potential, −10.00 kV; detector potential, −2.20 kV. Multiple-reaction monitoring was used for detection of arginine derivatives with a dwell time to 100 ms. Q1 was set to transmit the parental ions MH⁺ at m/z of 203.15 for ADMA and SDMA and of 324.6 for N-PLA. The daughter ions were monitored in Q3 at m/z 158.05, 171.95, and 70.15 for ADMA, SDMA, and N-PLA, respectively. All analyses and data processing were completed with the LabSolutions Ver. 5.60 software (Shimadzu Scientific Instruments, Inc., Columbia, MD).

In vitro ATP assay.

The protein concentrations of isolated mitochondria were measured by DC protein assay (Bio-Rad) and adjusted to 0.2 mg/ml with buffer B (225 mM sucrose, 44 mM KH₂PO₄, 12.5 mM Mg acetate, and 6 mM EDTA). The equal amounts of mitochondria used for in vitro ATP assay were confirmed by Western blotting with anti-ATP5A (ab14748; Abcam) or anti-ATPB antibody. Commercially purchased ADP (Wako) was incubated with glucose and hexokinase to deplete the residual ATP contamination before use. After reaction, the hexokinase was inactivated by heating to 99°C for 3 min. In vitro ATP assay was conducted in a 96-well plate (Corning), with every sample measured in triplicate. Twenty microliters of different combinations of intermediate metabolites (PM, 5 mM pyruvate plus 5 mM malate; SM, 25 mM succinate plus 5 mM malate; GM, 50 mM glutamate plus 25 mM malate; PCM, 0.25 mM palmitoyl-l-carnitine plus 10 mM malate) and 10 μl of 1 mM ADP were added to the wells, followed by 20 μl of isolated mitochondria or buffer B as background. The produced ATP was measured based on luciferin/luciferase reaction by the ATP Bioluminescence assay kit HS II (Roche) according to the manufacturer's instructions. After 50 μl of luciferase reagent was added to the reaction mixture, the luminescence was read by a Wallac 1420 ARVOsx multilabel counter (Winpact Scientific) at different time points.

Oxygen consumption.

Oxygen consumption was measured using a Seahorse XF24 Extracellular Flux Analyzer at 25°C according to the method described previously with slight modification (44, 46). Briefly, young adult worms were washed off the plates and rinsed with M9 buffer. Following shaking in the buffer for 1 h to allow to completely digest the remaining bacteria, about 30 worms were transferred into each well of a 24-well assay plate in a total volume of 525 μl. Basal respiration was measured for a total of 40 min, in 5-min intervals that included a 2-min mixing period, a 1-min time delay, and a 2-min measurement. To measure respiratory capacity, 15 μM FCCP was injected, and the oxygen consumption was measured for nine consecutive intervals. Worms collected from each plate were considered one technical replicate.

MitoSOX staining and imaging.

MitoSOX staining was performed according to the method previously described (30). Adult worms were wounded by microinjection needles to increase the permeability of MitoSOX Red molecular probe (Invitrogen). After wounding, worms were immediately transferred to a MitoSOX Red staining solution (5 μM in M9, prepared from a 5 mM stock in dimethyl sulfoxide [DMSO] before use) and stained in the dark for 20 min at room temperature. Stained worms were washed three times with M9 before imaging under a fluorescence microscope with DsRed filter.

Food avoidance assay.

The food avoidance assay was performed as previously described (16). Briefly, one milliliter of overnight culture of RNAi E. coli was pelleted and resuspended in 50 μl of LB and then dropped in the center of 3.5-cm dishes containing 1 mM IPTG. For NGM dishes, 50 μl of OP50 was seeded. Dried dishes were kept at room temperature overnight to allow the IPTG induction. Only circular lawns of uniform size and density were used for food avoidance assays. Synchronized L1 worms (about 50 animals per dish) were dropped into the center of each bacterial lawn. Food avoidance phenotypes were scored when the worms grew to the young adult stage. For the transgenic lines TKB71 and TKB84, L4 larvae with the transgenes positively expressed were selectively picked out according to the coinjection marker DsRed onto the assay plates, and the food avoidance phenotype was scored at the young adult stage.

Antioxidant treatment.

Treatment with the antioxidant N-acetylcysteine (NAC) was conducted as previously described (47). Briefly, all the agar plates were prepared from the same batch of NGM agar, whereas treatment plates were supplemented with 10 mM NAC, and control plates were supplemented with water. After plates were poured and dried for about 30 min, they were sealed and stored at 4°C. Freshly prepared E. coli OP50 isolates were spotted onto plates on the previous evening and allowed to dry and grow overnight. Synchronized L1 larvae were fed on the supplemented plates and allowed to grow until the young adult stage.

Statistical analysis.

Results were presented as means ± standard errors of the means (SEM). Statistical significances were determined by two-tailed unpaired Student's t test.