Ubiquinone is necessary for Caenorhabditis elegans development at mitochondrial and non-mitochondrial sites.

Ubiquinone (UQ) is a lipid co-factor that is involved in numerous enzymatic processes and is present in most cellular membranes. In particular, UQ is a crucial electron carrier in the mitochondrial respiratory chain. Recently, it was shown that clk-1 mutants of the nematode worm Caenorhabditis elegans do not synthesize UQ(9) but instead accumulate demethoxyubiquinone (DMQ(9)), a biosynthetic precursor of UQ(9) (the subscript refers to the length of the isoprenoid side chain). DMQ(9) is capable of carrying out the function of UQ(9) in the respiratory chain, as demonstrated by the functional competence of mitochondria isolated from clk-1 mutants, and the ability of DMQ(9) to act as a co-factor for respiratory enzymes in vitro. However, despite the presence of functional mitochondria, clk-1 mutant worms fail to complete development when feeding on bacteria that do not produce UQ(8). Here we show that clk-1 mutants cannot grow on bacteria producing only DMQ(8) and that worm coq-3 mutants, which produce neither UQ(9) nor DMQ(9), arrest development even on bacteria producing UQ(8). These results indicate that UQ is required for nematode development at mitochondrial and non-mitochondrial sites and that DMQ cannot functionally replace UQ at those non-mitochondrial sites.

Ubiquinone (UQ) 1 is a prenylated benzoquinone that is an essential co-factor in the mitochondrial respiratory chain, where its function is best characterized. UQ is also found in many other locations in the cell, such as the lysosome and Golgi membranes, as well as in nuclear and plasma membranes (1). The exact role of UQ at these extramitochondrial sites is being actively explored (e.g. Refs. 2 and 3).
The gene clk-1 of the nematode Caenorhabditis elegans affects many physiological rates, including embryonic and postembryonic development, rhythmic behaviors, reproduction, and life span (4). clk-1 encodes a 187-amino acid protein that is localized in mitochondria (5) and that is homologous to the yeast protein Coq7p, which has been shown to be required for UQ biosynthesis (6). clk-1 has also been shown to be necessary for UQ biosynthesis in worms (7,8) and in the mouse (9). Indeed, UQ 9 is entirely absent from mitochondria purified from worm and mouse clk-1 mutants (8,9) (the subscript refers to the length of the isoprenoid side chain). Instead, these mitochondria accumulate demethoxyubiquinone (DMQ 9 ), which is an intermediate in the synthesis of UQ 9 (8,9). Consistently, recent evidence suggests that clk-1 encodes a DMQ hydroxylase (10), which converts DMQ to ubiquinol. In Escherichia coli, DMQ 8 is able to sustain respiration in isolated membranes although at a lower rate than Q 8 (11). Similarly, DMQ 9 is capable of sustaining electron transport in eukaryotic mitochondria, as the function of purified mitochondria (5), and mitochondrial enzymes (8), from clk-1 worm mutants appears to be almost intact compared with the wild type. In addition, synthetic DMQ 2 has been shown to function in vitro as a co-factor for electron transport from worm complex I and, albeit more poorly, from complex II (8). Finally, it was found that in the absence of any exogenous UQ the oxygen consumption of mouse embryonic stem cells with a deleted mclk1 gene is only reduced by 35% (9).
Recently, it has been found that clk-1 mutants are unable to grow on a UQ-deficient bacterial strain despite the presence and the activity of DMQ 9 (7). Although, dietary UQ is generally not capable of reaching mitochondria (reviewed in Ref. 1), this has been interpreted to suggest that DMQ 9 is insufficient for normal mitochondrial function and that dietary bacterial UQ 8 can reach the mitochondria and function there in trace amounts (7).
To resolve these issues, we have generated a strain carrying a knockout mutation in the nematode gene coq-3, which encodes a methyltransferase required for UQ synthesis, and whose inactivation does not lead to DMQ accumulation in yeast (12). We find that coq-3 worms are not able to complete development even on bacteria that contain UQ 8 . These results indicate that 1) dietary UQ cannot complement a UQ deficiency in the absence of DMQ, 2) the growth impairment of clk-1 mutants without a dietary supply of UQ is due to a non-mitochondrial requirement for UQ, and 3) DMQ cannot functionally replace UQ at the non-mitochondrial sites.

EXPERIMENTAL PROCEDURES
Nematode and Bacterial Strains-We used the wild type N2 (Bristol) strain. Mutant strains analyzed include daf-2(e1370), eat-2(ad465), dpy-9(e12), and mau-2(qm160). None of these strains had any difficulty in growing on bacteria that do not produce UQ. A detailed analysis was performed using clk-1(qm30), clk-1(e2519), and clk-1(qm51). Standard procedures were used for bacterial and worm cultures (13), except that the nematode growth medium plates contained 0.5% glucose, to minimize the reversion of UQ-deficient strains. The genotypes of the bacterial strains used are described in Table I.
Production of the coq-3 Knockout Mutation-A null mutation in the C. elegans gene coq-3 was generated according to the protocol previously established by Molder and Barstead (snmc01.omrf.uokhsc.edu/revgen/ RevGen.html). A detailed description of the manipulations performed, including the primer sequences used for PCR, can be obtained upon request. See Fig. 2 for a schematic representation of the coq-3 gene (which is part of an operon), the extent of the deletion in coq-3 (qm188) * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  mutants, the positions of the primers used to screen for the mutation, and the positions of the primers used to amplify a rescuing fragment of the gene. To verify the genotype of coq-3, we performed PCR analysis and used sets of primers whose priming regions are either outside of the coq-3 gene or inside the region corresponding to the qm188 deletion mutation.
Phenotypic Characterization of coq-3-We analyzed the gonads of the coq-3 mutants using DIC microscopy. Quantitative measurements of the growth rate and the brood size were performed as described previously (14). For brood size measurements, we counted the entire progeny of 10 worms. The experiments were performed twice.
Transformation Rescue-We followed the usual micro-injection procedure to generate standard extrachromosomal arrays (15). A PCR fragment (50 ng/l) comprising the coq-3 genomic sequence was injected to assay for rescue. We used the pRF4 plasmid (120 ng/l), which contains a dominant mutation rol-6 (su1006d) as a co-injection marker to screen for transgenic worms. coq3/dpy-4 heterozygous worms were utilized for injection, since coq-3 is homozygous lethal. We selected the homozygous coq-3 rescued lines by picking the lines in which the Dpy phenotype was absent in the progeny and confirmed the genotype by PCR analysis. The primer sequences can be obtained upon request (see also Fig. 2).
Growth and Brood Size Analysis-To evaluate the development on various bacterial strains, we first placed adult worms in a drop of bleach on a plate containing the test bacteria. This step ensures that no OP50 bacteria is present in the test plate. L1 larvae that hatched from the bleached eggs were transfered to a fresh test plate, and the growth of the worms was examined. For brood size measurements, we counted the entire progeny of 10 worms. The experiments were performed twice for N2 and four times for the clk mutants.

Ubiquinone Is Necessary for C. elegans Development and
Fertility-It was recently reported that clk-1 mutants are incapable of completing development when fed on a ubiG E. coli mutant strain (7). The ubiG gene product is required at two steps of the UQ biosynthesis pathway, and ubiG mutants do not produce any UQ (16). We tested whether this growth phenotype resulted from a specific toxicity of the ubiG strain (GD1) for clk-1 mutants or from the absence of UQ. For this purpose, we systematically analyzed the growth of clk-1 mutant worms on a variety of E. coli mutants that are defective for UQ biosynthesis (ubi mutants). Nine E. coli enzymes have been described as participating in UQ biosynthesis (see Fig. 1) (16). They are all membrane-bound, except the first one, ubiC, which is a soluble chorismate lyase (17). The next enzyme in the pathway is the prenyltransferase ubiA that attaches the isoprenoid side chain to the quinone ring (eight subunits in E. coli) (18). The other enzymes are grouped in three categories: decarboxylases (ubiD and ubiX), monooxygenases (ubiB, ubiH, ubiF), and methyltransferases (ubiG, ubiE) (19 -24). We examined the growth of the three clk-1 mutant strains on strains of bacteria mutant for each of these genes, except ubiC (Table I). The three clk-1 mutant alleles are qm30 and qm51, which are putative nulls, and e2519, a point mutation that results in a milder phenotype (14,25).
We find that on all the bacterial ubi -(mutant) strains tested, L1 larvae from the wild type strain N2 are capable of completing development to adulthood and that these adults have a brood size of around 320, which is similar to their brood size on ubi ϩ bacteria (OP50) ( Table II). This indicates that endogenously synthesized Q is sufficient to maintain a wild type phenotype, without a requirement for dietary UQ. We have also examined a number of worm mutants that are not known to be involved in UQ synthesis (dpy-9, eat-2, mau-2), including long lived mutants (daf-2 and a number of strains that show a clk-1-like phenotype that have not yet been fully characterized). In no case was the growth of the mutants impaired on ubi Ϫ bacteria. In contrast, all three clk-1 mutants behave identically on most ubi Ϫ bacterial strains tested: they develop very slowly, or not at all, and produce no progeny (Table II). How-ever, the clk-1 mutants can develop and produce some progeny on ubiD, ubiX, and ubiH mutant strains, which are point mutants that produce residual amounts of ubiquinone (around 15% of the wild type) (16). It appears therefore that relatively low levels of bacterial UQ 8 are sufficient to allow for the growth of clk-1 mutants.
Endogenous Ubiquinone Is Necessary for C. elegans Development and Fertility-To test whether dietary UQ is sufficient for C. elegans development, we produced a knockout mutation of the worm gene coq-3. coq-3 encodes a methyltransferase whose homologues (Coq3p and UbiG) have been extensively characterized in S. cerevisiae and in E. coli, respectively. The enzyme acts at two different steps of Q synthesis, and neither UQ nor DMQ are produced in the yeast and bacterial mutants (24,26). 1. The ubiquinone biosynthesis pathway. The pathway of ubiquinone biosynthesis is schematically depicted and corresponds to the synthesis steps that are common to prokaryotes and eukaryotes. The metabolites that lead to ubiquinone are shown in boxes. They are derived from chorismate, and they contain a polyprenyl chain (eight isoprenoid units in E. coli and nine in C. elegans), added by the prenyltransferase ubiA to the 4-hydroxybenzoate. The prokaryotic genes that encode the enzymes involved in the pathway are on the left of the arrows (ubi genes). The C. elegans genes described in this study (clk-1 and coq-3) are on the right of the arrows. The prokaryotic enzymes can be grouped as following, considering their described enzymatic activities: chorismate lyase (ubiC), prenyltransferase (ubiA), decarboxylases (ubiD and ubiX), monooxygenases (ubiB, ubiH, ubiF), and methyltransferases (ubiG, ubiE). All these enzymes are membrane-bound, except ubiC, which is a soluble enzyme. The ubiquinone precursor accumulated in C. elegans clk-1 mutants is DMQ (depicted in bold). The worm COQ-3 protein is 29% identical to S. cerevisiae Coq3p and 28% to E. coli UbiG. We used a method of random mutagenesis and PCR-based screening to identify a deletion in coq-3 (see "Experimental Procedures" and Fig. 2). The deletion qm188 removes exons 3 and 4, such that no functional COQ-3 protein can be produced ( Fig. 2A). Self-fertilizing coq-3(qm188)/ϩ hermaphrodites produce one-quarter of homozygous coq-3(qm188)/coq-3(qm188) progeny, as verified by PCR (see "Experimental Procedures" and Fig. 2B). These coq-3 homozygotes develop slowly and appear substantially smaller than wild type worms. Most are sterile (Fig. 2C), but ϳ25% (n ϭ 31) produce some progeny (5-10 eggs) that arrest at the L1 stage and die quickly thereafter. These observations are consistent with a partial maternal effect from the heterozygous mothers of coq-3 homozygotes, as the phenotype of the first homozygous generation (slow development to adulthood) is less severe than that of the second homozygous generation (arrest at the L1 stage). The maternal effect could be due to UQ provided to the embryo by the mother or to maternal deposits of coq-3 mRNA or protein.
To ascertain whether the observed phenotypes are solely due to the mutation in the coq-3 gene, we introduced the genomic fragment corresponding to the coq-3 gene into coq-3/ϩ heterozygotes using the dominant rol-6 (su1006d) transformation marker by germ line transformation. Homozygous coq-3 transgenic animals (displaying the marker phenotype, Rol) develop normally and are fertile, indicating that the phenotype we observe is indeed due to the coq-3 deletion. However, the extrachromosomal array carrying the coq-3 and rol-6 sequences is incapable of producing a strong maternal effect. Indeed, homozygous animals without the array (phenotypically non-Rol) issued directly from mothers carrying the array (phenotypically Rol) did not develop beyond the L2 stage. The expression of genes from extrachromosomal arrays is sometimes silenced and is generally poor in the C. elegans germline (27). The observation of a maternal effect suggests that the mother deposits an essential product in the oocytes, which here could be UQ or coq-3 mRNA. In either case, proper expression of coq-3 in the germ line appears to be necessary for the effect.
We also observed that the brood size of heterozygous coq-3/ dpy-4 worms (243 Ϯ 38; n ϭ 10) was similar to that of dpy-4/ϩ worms (248 Ϯ 23; n ϭ 10) (Fig. 2D), suggesting that the expression level of coq-3 in the heterozygotes might not be limit-ing for UQ biosynthesis. The lethal phenotype of coq-3 mutants on ubi ϩ indicates that dietary UQ is not sufficient for the growth and development of worms. This is consistent with findings in other systems that indicate that dietary UQ cannot reach the mitochondrial compartment or only in extremely small amounts (1). The possibility that dietary UQ could be sufficient for worms was proposed to account for the viable phenotype of clk-1 mutants grown on ubi ϩ bacteria and their lethal phenotype when grown on ubimutant bacteria. However, the phenotype of coq-3 mutants indicates clearly that even in the presence of dietary bacterial UQ 8 , a total absence of endogenous UQ 9 and DMQ 9 (as in coq-3 mutants) is not equivalent to the replacement of endogenous UQ 9 by endogenous DMQ 9 (as in clk-1 mutants).
In this context, it is of particular interest that clk-1 mutants cannot thrive by feeding on ubiF mutants (see above). Indeed, UQ biosynthesis in ubiF mutants is blocked at the same level as in clk-1 mutants, and ubiF bacteria thus produce DMQ 8 (22). As DMQ 9 performs efficiently in the mitochondrial respiratory chain (8), our findings indicate that neither endogenous nor dietary DMQ can replace UQ at non-mitochondrial sites of UQ requirement. DISCUSSION Our results suggest that UQ is necessary for C. elegans growth and development at different subcellular locations, in particular it appears to be necessary at sites distinct from the mitochondrial respiratory chain (Fig. 3). Indeed, for its respiratory function in the mitochondria, endogenous DMQ 9 can functionally replace endogenous UQ 9 , as indicated by the observation that clk-1 mutant mitochondria do not appear to contain UQ 9 but are functionally competent (8). On the other hand, coq-3 mutants, in which a failure to manufacture UQ 9 and DMQ 9 is expected, display a much more severe phenotype than clk-1 mutants. Thus, at some still unknown site or sites, distinct from the respiratory chain, endogenous DMQ 9 , or dietary DMQ 8 , cannot functionally replace endogenous UQ 9 , while dietary UQ 8 can. In fact, clk-1 mutants, which have functional mitochondria and make DMQ 9 , cannot develop and grow without dietary UQ 8 , even in the presence of dietary DMQ 8 from ubiF bacteria.
This model is consistent with the findings by numerous studies on UQ uptake and metabolism in other systems, such TABLE II Growth and brood size analysis of wild type and Clk worms on ubi ϩ and ubi Ϫ bacteria Experiments were performed twice, yielding similar results. The sample size is indicated in parentheses. Growth is scored as the capacity to reach adulthood and be fertile. Each ubi Ϫ strain is mutant in a gene implicated in the ubiquinone biosynthesis pathway except RKP1452, which has knockouts in the two first genes in the pathway (ubiC and ubiA). The order of the ubi Ϫ bacteria in the table reflects the ordering of the ubi genes in the ubiquinone biosynthesis pathway.  2. Characterization of the coq-3 deletion mutant. A, genomic structure of the coq-3 gene. coq-3 is located on linkage group IV of C. elegans and is part of an operon, comprising the gdi-1 gene and a gene (Y57G11C.12) encoding a subunit of NADH-ubiquinone oxidoreductase. coq-3 contains five predicted exons. The deletion in coq-3 (qm188) mutants removes 2456 bp, eliminating exons 3 and 4, and consequently no functional protein can be produced. Primers (SHP) used for PCR screening of the deletion library and individual worms are indicated. B, PCR analysis of genomic DNA from N2, coq-3/dpy-4, and coq-3/coq-3 worms. To check the presence of a deletion in the coq-3 gene, PCR analyses were carried out using genomic DNA from single worms. Each DNA preparation was simultaneously tested with primers recognizing sequences either outside the coq-3 gene (SHP 1774 and SHP 1775, lanes 2-4) or inside the obtained deletion (SHP 1840 and SHP 1865, lanes [5][6][7]. See A for the primer localization. When using primers amplifying the whole coq-3 gene, a band of 4.3 kb was obtained from a wild type worm (lane 2). In contrast, a mutant band was amplified at 1.8 kb from a coq-3/coq-3 worm. DNA from heterozygotes only amplified the mutant band (lane 2), probably because the PCR conditions are in favor of the shorter amplicon. When using primers annealing in the deletion region, both wild type and heterozygote worms gave a PCR product of 1.1 kb (lanes 5 and 6), while no band was detected from a coq-3/coq-3 homozygote worm, which confirmed that the mutants were homozygous. C, photographs of adult gonads of N2 and coq-3/coq-3 worms. In contrast to the wild type (N2), no embryos are present in most coq-3/coq-3 adult worms, which underlies their sterile phenotype. D, brood size analysis. Experiments were performed twice, yielding similar results. The sample size is indicated in parentheses. Growth is scored as the capacity to reach adulthood and be fertile. We analyzed coq-3/dpy-4 heterozygotes, maternally rescued coq-3/coq-3 (m ϩ z Ϫ ), coq-3/coq-3 rescued with a transgene carrying the wild type sequence of the coq-3 gene (coq-3/coq-3 rescue), and dpy-4/dpy-4 as well as dpy-4/ϩ as controls. The results show that coq-3/dpy-4 heterozygotes have a brood size comparable with dpy-4/ϩ or dpy-4/dpy-4 animals. coq-3 homozygotes rescued with a wild type version of the coq-3 gene are fertile, but their brood-size is low. as rodents (1). What has been found is that dietary UQ appears to be taken up only poorly (2-3% of the initially ingested ubiquinone), and the majority is then distributed to the plasma membrane, the lysosomes, and the Golgi, with only minute quantities, if any at all, appearing in the mitochondria. Given that every cell endogenously produces UQ, it is possible that no active uptake system exists to assimilate this rather complex lipid.
Our studies clarify the roles of endogenous and dietary UQ in the worm's biology. Also, we demonstrate for the first time the functional importance of UQ at non-mitochondrial locations for an organism's viability. Action of dietary UQ at non-mitochondrial sites could underlie the beneficial effects of dietary UQ for patients with mitochondrial diseases (1). For example, UQ has been found to participate in reactions that regulate the redox state of the cell at the plasma membrane (28). Disease states that arise from deficient mitochondria are often found to increase cellular oxidative stress, and dietary UQ could stimulate a protective function at the plasma membrane (28). In addition, in bacteria, quinones have recently been found to act as the primary signal of the redox state of the cell (29). In E. coli, UQ negatively modulates the phosphorylation status and function of ArcB, an important global regulator of gene expression. The eventual discovery of additional roles for UQ in eukaryotes, and in particular as a signaling cue, will help to better under-stand the pleiotropic effects of mutations in genes that affect UQ, including clk-1.
Finally, we note that the coq-3 and clk-1 mutant strains provide genetic models to identify compounds that could selectively replace ubiquinone at the mitochondria and/or at nonmitochondrial sites. The development of such bio-available ubiquinone mimetics could be of great medical interest. FIG. 3. A model for the subcellular targets of UQ. This model hypothesizes at least two spatially distinct functional requirements for ubiquinone in C. elegans. It distinguishes between UQ at mitochondrial and non-mitochondrial locations based on the phenotype of clk-1 mutants on ubibacteria and that of coq-3 mutants on ubi ϩ bacteria. Endogenous UQ 9 is able to reach all of the locations where UQ is necessary, while dietary UQ 8 only accesses non-mitochondrial sites. On the other hand, endogenous DMQ 9 does not function at those sites, but is active in mitochondrial respiration. Mitochondria are depicted in red, and the undetermined non-mitochondrial locations are represented in blue. Arrows point to locations within the cell where a given quinone is functional.