RNAi procedures:

RNAi of genes on chromosome I was carried out by sequential feeding of an arrayed library of genomic fragments obtained from the United Kingdom Human Genome Mapping Project Resource Center (Cambridge, UK) as described (Fraser et al. 2000). Isopropyl-β-d-thiogalactopyranoside (IPTG, 1 mm) was added to the bacterial growth media to induce transcription of the double-stranded RNA. Mock RNAi feeding was conducted with Escherichia coli transformed with the same plasmid, but lacking an insert; they, too, were cultured in the presence of 1 mm IPTG.

cDNA-based RNAi-feeding plasmids directed toward C. elegans and C. briggsae ubl-5 were constructed by reverse transcriptase PCR amplification of the coding sequence from whole-animal RNA. The C. elegans primers used were F46F11.Kpn.1S (5′-GGCCGGTACCATGATTGAAATCACAGTAAACGATC-3′) and F46F11.2AS (5′-GAAATGAATTCTGATGAATCATTGG-3′) and the C. briggsae cDNA was amplified using the same UBL-5.brig.1S and UBL-5.brig.2AS primers described above. The inserts were ligated into the pPD129.36 plasmid.

The screen for genes whose inactivation by RNAi interferes with hsp-60gfp expression was conducted by placing four SJ52 L4 larvae of the zc32 II; hsp-60gfp(zcIs9) V genotype on 60-mm plates seeded with E. coli strains carrying an RNAi-feeding plasmid and allowing their offspring to develop at the permissive temperature for 72 hr and then switching the plates to the nonpermissive temperature of 25° and observing, 24 hr later, the effects of the RNAi on growth, viability, fertility, and the expression of the hsp-60gfp. At this point, the plate was populated with both F₁ and developing F₂ progeny of the original four animals, allowing us to score the phenotype across three generations of RNAi exposure.

The timing of exposure to RNAi was varied to manipulate the level of gene inactivation, which enabled us to overcome lethal or growth-suppressive phenotypes associated with complete inactivation. Placing embryonated eggs purified from gravid animals on RNAi plates restricts exposure to the RNAi to the later stages of embryonic development and often bypasses the lethality associated with earlier inactivation of some genes. The phenotype was scored in the F₀ animal. This design was used to study the interactions of zc32 with RNAi of mitochondrial chaperones (Figure 2) to measure the effect of ubl-5(RNAi) on induction of the mitochondrial chaperones (Figure 3, C and D), on mitochondrial morphology in myo-3gfp^mt animals (Figure 4A), and on the assembly of protein complexes in the mitochondrial matrix (Figure 5B) and to study the effect of RNAi of mitochondrial chaperones on UBL-5GFP localization (Figure 6). In other experiments we measured the phenotype by assessing the effect of gene inactivation on the F₁ progeny that develop on RNAi-feeding plates (Figure 3, A and B; Figure 4, B and C).

Open in a separate window

Figure 2.—

The zc32 mutation compromises the ability to cope with further mitochondrial stress. Shown are photomicrographs of plates with wild-type (N2) and zc32 mutant animals that developed at the permissive (20°) and nonpermissive temperature (25°) from embryos exposed to mock RNAi or RNAi of genes that induce further mitochondrial stress (spg-7, phb-2, and hsp-60) or ER stress (ire-1). Photographs were taken 84 hr after seeding the RNAi plates with embryonated eggs. The fraction of animals that reached developmental stage ≥L4 (mean ±SEM) is indicated below the images.

Open in a separate window

Figure 3.—

Inactivation of ubl-5 compromises signaling in the UPR^mt. (A) Fluorescent photomicrographs of wild-type (N2) and zc32 mutant animals transgenic for the UPR^mt reporters hsp-6gfp and hsp-60gfp or for the UPR^er reporter hsp-4gfp. Animals developed from founders placed on mock RNAi or ubl-5(RNAi) plates at 20° and were shifted, where indicated, to the nonpermissive temperature (25°) for 24 hr before analysis. The animals in the left panels were exposed to genomic-based RNAi whereas those on the right were exposed to cDNA-based RNAi constructs derived from C. elegans mRNA (Ce) or C. briggsae mRNA (Cb). Where indicated, the hsp-4gfp reporter was activated by exposure to the ER-stress-inducing drug tunicamycin [note that ubl-5(RNAi) does not affect the induction of hsp-4gfp]. (B) Photomicrographs of progeny of wild-type (N2) or zc32 founder animals that developed on mock or ubl-5(RNAi) plates maintained for 16 hr at the permissive temperature of 20° (to allow egg laying by the founder) and shifted to the nonpermissive temperature of 25° for 3 days. The number of F₁ progeny that reached developmental stage ≥L4 (mean ±SEM)/F₀ hermaphrodite is indicated below the images. (C) Immunoblot of GFP reporter in hsp-60gfp transgenic wild-type, zc32 mutant, or animals exposed to the indicated mitochondrial stress-inducing RNAi in the presence or absence of ubl-5(RNAi). The anti-HDEL blot, which detects C. elegans BiP, serves as a loading control. (D) Northern blot of endogenous hsp-60 and hsp-6 RNA in animals that developed on mock RNAi, spg-7(RNAi) (to induce mitochondrial stress), and a combination of spg-7(RNAi) and ubl-5(RNAi). Samples were processed 72 and 96 hr after seeding the RNAi plates with eggs. Ribosomal RNAs stained with ethidium bromide serve as a loading control. (E) Immunoblot of GFP, as in C. The GFP protein encoded by the hsp-60gfp(zcIs9) transgenic UPR^mt reporter (hsp-60GFP) and the UBL-5CbGFP fusion protein encoded by the rescuing transgenic allele of ubl-5^Cbgfp(zcIs22) are indicated. The asterisk marks an immunoreactive protein fragment found in ubl-5^Cbgfp(zcIs22) animals, which is likely an in vitro proteolytic fragment of the fusion protein. The anti-HDEL blot, which detects C. elegans BiP, serves as a loading control.

Open in a separate window

Figure 4.—

Inactivation of ubl-5 selectively compromises an animal's ability to cope with mitochondrial stress. (A) High-magnification fluorescent photomicrographs of mitochondria labeled with a GFP with a cleavable mitochondrial import signal (GFP^mt) expressed from a myo-3gfp^mt transgene. Animals developed to adulthood from embryos exposed to the indicated RNAi. Note the irregular morphology of the mitochondria in animals exposed to RNAi that induces mitochondrial stress (spg-7, hsp-60, phb-2) or inactivates ubl-5 and the normal morphology of the mock and ire-1(RNAi) animals. (B) Photomicrographs of plates populated by progeny (F₁) of myo-3gfp^mt or myo-3gfp^cyt animals subjected to the indicated RNAi. The F₀ animals had been removed from the plate. The number of F₁ progeny that reached a developmental stage ≥L4 (mean ±SEM)/F₀ hermaphrodite (n = 4) is indicated below the images. The panels at the far right are fluorescent photomicrographs of transgenic animals. (C) Photomicrographs of progeny produced by the F₁ generation of mock and ubl-5(RNAi) transgenic animals expressing GFP in mitochondria (ges-1gfp^mt) or cytoplasm (ges-1gfp^cyt) of the intestinal cells. (D) Immunoblot of GFP and the anti-HDEL loading control in test and control pairs of myo-3gfp^mt and myo-3gfp^cyt or ges-1gfp^mt and ges-1gfp^cyt transgenic animals.

Open in a separate window

Figure 5.—

Inactivation of ubl-5 perturbs formation of protein complexes in the mitochondria. (A) Detection of an ~75-kDa biotinylated protein(s) in fractionated worm extracts by strepavidin–HRP ligand blotting. Total worm extract (T), postmitochondrial supernatant (SN), and mitochondrial pellet (MT) are shown. GFP^cyt expressed from the myo-3 promoter serves as a marker for a cytoplasmic protein. (B) Distribution of an ~75-kDa biotinylated mitochondrial protein(s) in fractions of glycerol gradient prepared from animals exposed to mock RNAi, hsp-60(RNAi), and ubl-5(RNAi). (Top) Unstressed animals. (Bottom) Mitochondrially stressed myo-3gfp^mt animals. The migration of complexes of the indicated size in this gradient is shown. The final fraction (14) also contains the pellet.

Open in a separate window

Figure 6.—

The stress of protein misfolding in the mitochondrial matrix promotes nuclear localization of UBL-5GFP. (A) Fluorescent photomicrographs of ubl-5gfp transgenic animals that developed from larvae while exposed to the indicated RNAi. Shown are low magnification (top) and high-magnification (bottom) fluorescent micrographs of the GFP channel and the H33258 nuclear stain. Note that the large polyploid intestinal nuclei stain brightly with H33258 in all samples and with UBL-5GFP in the stressed samples. (B) The relative intensity of the cytosolic and nuclear UBL–GFP signal in animals subjected to the indicated genetic manipulations. The mean ±SEM ratio of signal from the nuclear and cytosolic regions of the intestinal cell is plotted (N = 20, * P < 0.05, unpaired two-tailed Student's t-test, compared with mock sample).

The potential confounding effects of dilution on the outcome of double RNAi experiments were controlled by similar dilution of single RNAi-expressing bacteria with bacteria transformed with the empty RNAi expression plasmid pPD129.36.

Pharmacological treatment:

Ethidium bromide (Sigma, St. Louis) was dissolved in water at 10 mg/ml and added to agar plates to a final concentration of 50 μg/ml and tunicamycin was used at 1 μg/ml as previously described (Calfon et al. 2002; Yoneda et al. 2004).

Microscopy and image analysis:

Low-resolution images of live animals were obtained with an inverted dissecting microscope equipped with epifluorescent illumination (Zeiss Stemi SV 11 Apo) as described (Yoneda et al. 2004). Higher-resolution images were obtained by mounting animals on slides using a digital CCD camera attached to a Zeiss Axiophot 100 microscope. Where indicated, the samples were fixed (3.7% paraformaldehyde, 10 min at room temperature followed by 100% methanol, 5 min at −20°) and exposed to a 1:1000 dilution of H33258 dye to stain the nuclei. Images were constituted from composites of digital photographs obtained under identical exposure conditions.

Image analysis (Figure 6) was conducted by calculating the mean pixel intensity over the intestinal nuclei and an equal surface of cytoplasm using Volocity software for Mac OS X (Improvision, Coventry, UK). The measurements are reported as the mean nuclear/cytoplasmic ratio ±SEM of 20 nuclei from five animals of each genotype. Statistical analysis was performed by the unpaired two-tailed Student's t-test.

Isolating a mutation that conditionally activates the UPR^mt:

The hsp-60gfp reporter has a faint basal activity but is robustly and specifically induced by manipulations that activate the UPR^mt (Yoneda et al. 2004) (Figure 1B). We screened F₂ progeny of EMS-mutagenized hsp-60gfp(zcIs9) transgenic worms for individuals with constitutive activity of the reporter. Such individuals were readily identified (250 of 7500 haploid genomes screened), an observation consistent with the large number of (likely nuclear) genes required for mitochondrial homeostasis (Table I in Yoneda et al. 2004). However, none of the constitutive UPR^mt-active animals could be propagated as a mutant line. The latter observation is consistent with the requirement for mitochondrial integrity in development of the C. elegans germline (Tsang et al. 2001; Artal-Sanz et al. 2003).

A screen for temperature-sensitive (ts) mutations that conditionally activated the UPR^mt yielded a line, zc32, which robustly activated hsp-60gfp and hsp-6gfp at the nonpermissive temperature (25°) and weakly activated the reporters at the permissive temperature (20°) (Figure 1B). Larvae that hatched from mutant eggs placed at the nonpermissive temperature developed into slightly uncoordinated, hsp-60gfp-active adults with a compromised germline and very few progeny (Figure 1C); the latter is consistent with the detrimental role of mitochondrial stress on germline development. The ts mutation of zc32 segregated as simple Mendelian recessive mapping to a small region on chromosome II flanked by the markers R12C12 (−0.63 cM) and C56E9 (−0.17 cM) (see supplemental data at http://www.genetics.org/supplemental/). The phenotype was also uncovered by placing the mutant allele in trans to a deficiency (mnDf96, CGC strain SP788), suggesting that the mutation has significant loss-of-function features. The interval in question contains three genes whose products are predicted to function in mitochondria; however, the coding sequence of all three was identical in zc32 and N2 animals. Furthermore, RNAi inactivation of none of the 116 predicted genes in the interval phenocopied the mutation in wild-type animals nor did such RNAi affect the mutant phenotype of zc32 animals.

As the identity of the gene(s) responsible for the phenotype of zc32 remained obscure, we sought further evidence that the mutation selectively compromises mitochondrial function. Therefore, we compared the ability of zc32 animals to cope with mitochondrial stress or ER stress (as a reference) by inactivating genes whose products are important for either mitochondrial or ER homeostasis. At the permissive temperature, both wild-type and zc32 larvae developed into adult animals, whether exposed to spg-7(RNAi) (a mitochondrial protease), hsp-60(RNAi) (a mitochondrial chaperone), phb-2(RNAi) (a mitochondrial membrane protein assembly factor), or ire-1(RNAi) (a component of the UPR^er, whose inactivation causes ER stress). At the nonpermissive temperature, zc32 mutant animals were selectively impaired in their ability to develop under RNAi that induces further mitochondrial stress, but their ability to cope with ER stress was comparable to the wild type (Figure 2). The above observations present a genetic argument that zc32 animals experience mitochondrial stress at the nonpermissive temperature and that they might be useful tools in identifying genes required for signaling the UPR^mt.

ubl-5, a gene that functions in the UPR^mt:

Using an arrayed RNAi library (Kamath et al. 2003), we sequentially inactivated annotated genes in the C. elegans genome in zc32 animals, searching for genes whose RNAi inhibited the hsp-60gfp reporter transgene at the nonpermissive temperature. To minimize the predicted negative affect of UPR^mt inactivation on development and fertility of zc32 animals, we allowed embryonic development of the RNAi animals to proceed at the permissive temperature and switched the plates to the nonpermissive temperature for the final 24 hr before observing the animals. To identify the nonspecific effects of a given RNAi clone on animal health or reporter gene activity, we monitored, in parallel, the effects of RNAi on animals with a mutation, upr-1(zc6) X, that activates the UPR^er reporter hsp-4gfp (Calfon et al. 2002). We sought genes whose RNAi selectively impaired the health of zc32 animals while selectively diminishing hsp-60gfp activity.

A systematic RNAi survey of all (2445) available genes on chromosome I identified numerous genes whose inactivation selectively impaired the health of zc32 animals. All but one of these were associated with activation of the hsp-60gfp reporter. As expected, most such genes encode known mitochondrial proteins whose inactivation promotes further mitochondrial misfolded protein stress (i.e., they encode chaperones, proteases, subunits of protein complexes, or components of the mitochondrial transcription or translational apparatus) (Yoneda et al. 2004; Hamilton et al. 2005). The selective negative effect of their inactivation on the health of zc32 animals is readily explained by inability of the mutant animals to cope with further mitochondrial misfolded protein stress (Figure 2). RNAi of one gene, lin-35, reproducibly attenuated hsp-60gfp expression but did not selectively compromise the health of zc32 animals, suggesting that the encoded protein, a C. elegans homolog of the retinoblastoma gene product, is not implicated in the endogenous UPR^mt.

Inactivation of one gene, ubl-5, was associated both with decreased expression of the UPR^mt reporter genes (Figure 3A) and with impaired development of zc32 animals (Figure 3B). The RNAi clone used in this screen contained a relatively large genomic fragment; therefore, to confirm that the phenotype observed was due to inactivation of ubl-5 (and not other genes on the same DNA fragment), we constructed a cDNA-based RNAi-feeding plasmid that contained only the coding region of ubl-5, which produced the same phenotype as the original genomic RNAi clone (Figure 3B, left). We also confirmed by Northern blotting the degradation of endogenous ubl-5 mRNA in the ubl-5(RNAi) animals (data not shown).

To examine the scope of ubl-5's role in the UPR^mt, we compared the activity of hsp-60gfp in wild-type and ubl-5(RNAi) animals in which mitochondrial stress had been induced by inactivation of the mitochondrial chaperone HSP-60, protein assembly factor PHB-2, or the mitochondrial protease SPG-7. In each case, ubl-5(RNAi) diminished the expression of the hsp-60gfp reporter, as measured by GFP immunoblot (Figure 3C). Furthermore, ubl-5(RNAi) also lowered levels of stress-induced expression of endogenous hsp-6 and hsp-60 mRNA, an effect that was evident at both 72 and 96 hr after hatching in animals developing on ubl-5(RNAi) plates (Figure 3D).

To determine if the inhibitory effects of ubl-5(RNAi) on the UPR^mt are a consequence of diminished expression of functional UBL-5, we sought to rescue the RNAi phenotype using a transgene resistant to the RNAi effects of C. elegans ubl-5. To this end, we exploited the fact that the homologous gene from C. briggsae encodes an identical protein; however, sequence divergence at the nucleic acid level was predicted to negate the effects of heterologous RNAi. This notion was supported by the observation that RNAi of C. briggsae ubl-5 had no phenotype in C. elegans (Figure 3, A and B, right). A UBL-5GFP-expressing transgene was constructed using the C. briggsae ubl-5 coding sequence (designated ubl-5^Cbgfp). The C. briggsae transgene was immune to the C. elegans coding sequence ubl-5(RNAi), whereas expression of the synonymous C. elegans-based transgene ubl-5^Cegfp was inhibited in tissues susceptible to RNAi effects (supplemental Figure S1A at http://www.genetics.org/supplemental/).

To determine if UBL-5^Cb-GFP is a functional protein, we examined its ability to rescue the phenotype associated with loss of endogenous UBL-5 expression; the marked decrease in fertility of ubl-5(RNAi) observed in hypersensitive eri-1 (mg366) IV mutant animals maintained at 20° was corrected by the ubl-5^Cbgfp transgene (supplemental Figure 1B at http://www.genetics.org/supplemental/). Next we determined if UBL-5^CbGFP can rescue the defect in the UPR^mt imposed by C. elegans ubl-5(RNAi). We bred the rescuing transgene into a background containing the zc32 mutation (to activate the UPR^mt) and the hsp-60gfp transgene (to report on the activation). The GFP protein expressed by the reporter transgene is smaller than the UBL-5-GFP fusion protein expressed from the rescuing transgene and the two are readily distinguished by their mobility on SDS–PAGE. C. elegans ubl-5(RNAi) markedly diminished the expression of the reporter in N2 animals, as noted above (Figure 3C); however, the inhibitory effect of ubl-5(RNAi) on the UPR^mt was rescued by the UBL-5^Cb-GFP transgene (Figure 3E). These experiments implicate UBL-5 protein in the UPR^mt.

ubl-5(RNAi) adversely affects the protein-folding environment in the mitochondria:

While the observations noted above indicate that ubl-5(RNAi) attenuates signaling in the UPR^mt, they do not identify the step in the process at which the gene acts. If UBL-5 is active in the UPR^mt afferent limb, ubl-5(RNAi) is predicted to enhance mitochondrial misfolded protein stress and to promote intrinsic defects in mitochondrial function and structure. However, if ubl-5(RNAi) suppresses the UPR^mt upstream by reducing the load of unfolded/misfolded proteins in the mitochondrial matrix, the mitochondria of ubl-5(RNAi) animals would not be expected to exhibit intrinsic defects in function or structure (see Figure 1A).

To address this issue, we began by examining mitochondrial morphology, making use of myo-3gfp^mt transgenic animals whose body-wall muscles express GFP tagged with a mitochondrial import signal (GFP^mt). In wild-type myo-3gfp^mt transgenic animals, the GFP^mt signal presents a pattern of neatly stacked arrays of mitochondria aligned with the muscle fibers (Labrousse et al. 1999) (Figure 4A). The mitochondria of spg-7(RNAi), hsp-60(RNAi), and phb-2(RNAi) animals are thicker and often fragmented and have brighter GFP^mt signals than animals exposed to mock RNAi treatment or to ire-1(RNAi) (which, by causing ER stress, serves as a control for nonspecific stressful effects of mitochondrial perturbations). These findings are consistent with the hypothesis that inactivation of genes required for protein folding and assembly in the mitochondrial matrix indirectly alters organelle morphology (although a direct role for hsp-60, spg-7, and phb-2 in maintaining mitochondrial morphology cannot be excluded). Importantly, animals raised on ubl-5(RNAi) have a GFP^mt pattern that resembles mitochondrially stressed animals (Figure 4A). The latter observation is consistent with compromised mitochondrial homeostasis in the ubl-5(RNAi) animals and does not support upstream suppression of the UPR^mt by ubl-5 inactivation.

Whereas adult N2 animals that developed from larvae raised on ubl-5(RNAi) had a roughly normal brood size, zc32 embryos raised to adulthood at the nonpermissive temperature on ubl-5(RNAi) plates had very few progeny (Figure 3B). To determine if these negative interactions of ubl-5(RNAi) extend to other conditions associated with enhanced mitochondrial stress, we took advantage of the observation that high-level expression of mitochondrially targeted GFP (myo-3gfp^mt) markedly attenuates an animal's ability to cope with further mitochondrial stress whereas expression of even higher levels of cytoplasmic GFP (myo-3gfp^cyt) is less perturbing (Figure 4B). These observations are consistent with the fact that GFP is slow to fold and requires significant assistance by chaperones (Wang et al. 2002). Thus, evidence of a selective negative interaction with a transgene expressing GFP^mt suggests that a given genetic lesion is associated with mitochondrial stress. We found that ubl-5(RNAi) adversely affected growth and development of myo-3gfp^mt animals but had no such negative effect on myo-3gfp^cyt animals expressing GFP in the cytosol (Figure 4, B and D). Furthermore, ubl-5(RNAi) compromised fertility of ges-1gfp^mt animals expressing GFP^mt in their intestinal cells, but had no such negative effect on ges-1gfp^cyt animals expressing higher levels of cytosolic GFP (Figure 4, C and D).

To further examine the effect of ubl-5 inactivation on the protein-folding environment in the mitochondrial matrix, we followed the fate of marker mitochondrial proteins that normally assemble into multi-subunit complexes in the mitochondrial matrix. As antisera to C. elegans mitochondrial proteins were not available, we exploited the fact that some abundant mitochondrial proteins are naturally tagged with a covalently bound biotin moiety and can be detected by their reactivity to enzyme-linked streptavidin following denaturing SDS–PAGE and ligand blotting. In mammalian mitochondria, the α-subunits of 3-methylcrotonyl-CoA carboxylase (αMCC) and propionyl-CoA carboxylase (αPCC) account for most of the biotinylated proteins detected by this ligand-blot assay (Baumgartner et al. 2001). In C. elegans, too, a ligand blot using HRP-linked strepavidin yielded a major labeled species of ~75 kDa, consistent with the predicted size of αPCC (F27D9.5, predicted 79.7 KDa), αMCC (F32B6.2, predicted 73.7 kDa), or both. The biotinylated species were enriched in the mitochondrial fraction (Figure 5A) and were distinct in size from the major E. coli biotinylated proteins detected by the same assay (data not shown) (as worms are fed E. coli, it was important to deal with the potentially confounding effects of contaminating bacterial biotinylated proteins on this assay).

Both αPCC and αMCC assemble into high-molecular-weight complexes in the mitochondrial matrix (Chloupkova et al. 2000), and velocity gradient centrifugation showed a peak of biotinylated proteins with a predicted mass of ~800 kDa, consistent with the expected mass of a dodecamer of α- and β-subunits of MCC or PCC (Figure 5B, top). The peak of biotinylated protein was shifted toward lighter fractions in animals in which mitochondrial protein folding was compromised by hsp-60(RNAi) and a similar shift was also observed in myo-3gfp^mt animals, consistent with impaired complex formation or turnover in stressed animals. In unstressed animals, ubl-5(RNAi) did not markedly affect the distribution of biotinylated proteins in the gradient (although a trend toward a lighter peak was noted; data not shown). However, compounding the stress-inducing myo-3gfp^mt transgene with ubl-5(RNAi) shifted the peak to lighter fractions (Figure 5B, bottom-most panel), indicating that loss of UBL-5 adversely affects the protein-folding environment in mitochondrially stressed worms. These experiments support the hypothesis that ubl-5(RNAi) leads to a state of enhanced mitochondrial misfolded protein stress by compromising the counteractive response to such stress.

We thank Art Horwich and Wayne Fenton (Yale University); Frank Frerman (University of Colorado); Catherine Clarke and Tanya Jonassen (University of California at Los Angeles); Jodi Nunnari (University of California at Davis); Janet Shaw (University of Utah); Terry McNally (Abbott Labs); Chi Yun, Scott Clark, and Gabriele Campi (Skirball Institute) for advice; and the C. elegans Genetics Center for strains. This work was supported by the Ellison Medical Foundation and National Institutes of Health grants ES08681 and DK47119 to D.R. and F32-NS050901 to C.M.H.