Mitochondrial Import Efficiency of ATFS-1 Regulates Mitochondrial UPR Activation

See allHide authors and affiliations

Science  03 Aug 2012:
Vol. 337, Issue 6094, pp. 587-590
DOI: 10.1126/science.1223560

Initiating Mitochondrial Repair

The mitochondrial unfolded protein response (UPRmt) mediates the up-regulation of nuclear encoded mitochondrial chaperone genes in response to mitochondrial dysfunction. How mitochondrial dysfunction is communicated to the nucleus is unclear, but requires the transcription factor, ATFS-1. Nargund et al. (p. 587, published online 14 June) found that the key point of regulation in UPRmt signaling is mitochondrial protein import efficiency of ATFS-1. In addition to a nuclear localization sequence (NLS), ATFS-1 has a mitochondrial targeting sequence (MTS) that is necessary for UPRmt repression. ATFS-1 is normally imported efficiently into mitochondria and degraded by the Lon protease. However, in the presence of stress, some ATFS-1 fails to be imported into mitochondria and is trafficked to the nucleus. The juxtaposition of a C-terminal NLS to an N-terminal MTS in a transcriptional activator thus couples unfolded protein load in the mitochondrial matrix to a rectifying transcriptional response in the nucleus.


To better understand the response to mitochondrial dysfunction, we examined the mechanism by which ATFS-1 (activating transcription factor associated with stress–1) senses mitochondrial stress and communicates with the nucleus during the mitochondrial unfolded protein response (UPRmt) in Caenorhabditis elegans. We found that the key point of regulation is the mitochondrial import efficiency of ATFS-1. In addition to a nuclear localization sequence, ATFS-1 has an N-terminal mitochondrial targeting sequence that is essential for UPRmt repression. Normally, ATFS-1 is imported into mitochondria and degraded. However, during mitochondrial stress, we found that import efficiency was reduced, allowing a percentage of ATFS-1 to accumulate in the cytosol and traffic to the nucleus. Our results show that cells monitor mitochondrial import efficiency via ATFS-1 to coordinate the level of mitochondrial dysfunction with the protective transcriptional response.

Mitochondria import ~99% of their proteome through the TOM (translocase of the outer membrane) and TIM (translocase of the inner membrane) complexes (1, 2). The mitochondrial protein-folding environment is maintained by mitochondrial molecular chaperones whose expression levels are coupled to the state of mitochondrial protein homeostasis by a mitochondria-to-nuclear signaling pathway termed the mitochondrial unfolded protein response (UPRmt) (3, 4). Evidence in C. elegans implicates the mitochondrial inner membrane peptide transporter HAF-1 and the bZip transcription factor ATFS-1 (activating transcription factor associated with stress–1) in UPRmt signaling (5).

During mitochondrial stress, ATFS-1 accumulates in the nucleus as the result of a nuclear localization signal (NLS). A protein sequence prediction algorithm, Mitoprot II, predicted the presence of an N-terminal mitochondrial targeting sequence (MTS) as well (6) (Fig. 1A). Indeed, amino acids 1 to 100 of ATFS-1 were sufficient to target green fluorescent protein (GFP) to HeLa cell mitochondria (Fig. 1B and fig. S1). Consistent with cleavage of the MTS, the mitochondrial-enriched form of ATFS-11–100::GFP was smaller than the unprocessed form found in the postmitochondrial supernatant (Fig. 1C).

Fig. 1

In the absence of stress, ATFS-1 is imported into mitochondria and degraded. (A) ATFS-1 schematic. NES, nuclear export signal; abbreviations for amino acids: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; Y, Tyr. (B) Photomicrographs of HeLa cells expressing ATFS-11–100::GFP or GFP stained with MitoTracker. Scale bar, 0.25 mm. (C) Immunoblots of HeLa cells expressing GFP or ATFS-11–100::GFP after fractionation into total lysate (T), postmitochondrial supernatant (S), and mitochondrial pellet (M). Longer exposure of the ATFS-11–100::GFP panel was required because of toxicity and weak expression. (D) Immunoblots of atfs-1pr::atfs-1::gfp, wild-type (WT), or atfs-1(tm4525) worms raised on control(RNAi) or lon(RNAi). ATFS-1 (arrowhead) and ATFS-1::GFP (asterisk) are marked. (E) Immunoblots of wild-type worms fed control(RNAi) or lon(RNAi) after cellular fractionation. Endogenous ATFS-1 is indicated with an arrowhead. Endogenous NDUFS3 serves as a mitochondrial marker and α-tubulin (Tub) as a cytosolic marker. (F) Photomicrographs of hsp-60pr::gfp transgenic worms raised on control(RNAi), lon(RNAi), or spg-7(RNAi). Scale bar, 0.5 mm.

To understand UPRmt regulation in C. elegans, we sought to determine the localization of ATFS-1 in the absence of UPRmt activation. We were unable to detect endogenous ATFS-1 or the ATFS-1::GFP fusion protein with ATFS-1– or GFP-specific antibodies (Fig. 1D and fig. S2). Additionally, atfs-1pr::gfp worms expressed GFP strongly in all cells, indicative of an active promoter, whereas the ATFS-1::GFP fusion protein was nearly undetectable (fig. S3A). These data suggest that ATFS-1 is rapidly degraded.

We hypothesized that ATFS-1 was degraded by a mitochondrial matrix protease such as the caseinolytic peptidase ClpP or Lon (7). To reduce Lon or ClpP function via RNA interference (RNAi), we fed worms Escherichia coli that expressed double-stranded RNA targeting lon or clpp transcripts for degradation. Animals fed lon(RNAi), but not clpp(RNAi), accumulated endogenous ATFS-1 as well as ATFS-1::GFP (Fig. 1D and fig. S3, B and C). ATFS-1 was absent in lysates from atfs-1(tm4525) worms (Fig. 1D and fig. S3D), which were unable to activate the UPRmt (fig. S4). lon(RNAi) did not affect atfs-1 transcription (fig. S5A). Furthermore, in lon(RNAi)-treated worms, ATFS-1 cofractionated with a known mitochondrial protein (Fig. 1E and fig. S3E). Unlike spg-7(RNAi), which impairs a mitochondrial protease required for electron transport chain (ETC) quality control and mitochondrial ribosome biogenesis (8), lon(RNAi) did not activate the UPRmt transcriptional reporter hsp-60pr::gfp or impair worm development (5) (Fig. 1F). Therefore, in the absence of UPRmt activation, ATFS-1 is imported into mitochondria and degraded.

During UPRmt activation, ATFS-1::GFP accumulated in nuclei (Fig. 2A) (5). The predominant form of ATFS-1 that accumulated during spg-7(RNAi) or ethidium bromide (EtBr) treatment was of a higher molecular weight than the form detected in mitochondria of worms raised on lon(RNAi) and was enriched in the postmitochondrial supernatant (Fig. 2B and fig. S4C); these findings suggest that during UPRmt activation, a percentage of ATFS-1 remains in the cytosol, thus maintaining its MTS.

Fig. 2

In the presence of mitochondrial stress, unprocessed ATFS-1 accumulates in nuclei. (A) Photomicrographs of two intestinal cells in atfs-1pr::atfs-1::gfp or hsp-16pr::atfs-1∆1–32.myc::gfp transgenic animals raised on control(RNAi), spg-7(RNAi), tim-23(RNAi), or EtBr (100 μg/ml) with the nuclei outlined (right panels). The punctae (arrowheads) are endogenous autofluorescence from intestinal cell lysosomes. Mean percentages (±SEM) of worms with nuclear accumulation of ATFS-1::GFP are indicated (N = 3). Scale bar, 15 μm. (B) Immunoblots of fractionated lysates from wild-type worms raised on control(RNAi), spg-7(RNAi), or EtBr (100 μg/ml). Lanes 1 to 9 are 100 μg from the described fractions; lane 10 (asterisk) is 3 μg from the mitochondrial pellet of worms raised on lon(RNAi) for size comparison. Unprocessed and lon(RNAi)-stabilized (Ls) ATFS-1 are indicated, as are nonspecific bands (dot). (C) Immunoblots of fractionated extracts from wild-type or haf-1(ok705) worms raised on control(RNAi) in the absence or presence of EtBr (30 μg/ml) and expressing hsp-16pr::gfpmt. (D) Immunoblots of fractionated extracts from wild-type or haf-1(ok705) worms raised on lon(RNAi) in the absence or presence of EtBr (30 μg/ml) and expressing hsp-16pr::atfs-1FL.

Mitochondrial toxins such as paraquat, which activated the UPRmt (Fig. 3A), are known to impair mitochondrial import, thereby causing the accumulation of MTS-containing proteins in the cytosol (9). To determine whether a general impairment of import occurs during UPRmt activation, we generated a transgenic strain expressing GFP with a MTS (GFPmt) (10, 11). Because steady-state detection of unprocessed MTS-containing proteins is very difficult (12), we expressed GFPmt via the inducible hsp-16 promoter (fig. S6). Only in the presence of UPRmt-activating stress was unprocessed GFPmt detected in the postmitochondrial supernatant, consistent with impaired import (Fig. 2C). Similarly, when ATFS-1 was expressed via the hsp-16 promoter, unprocessed ATFS-1 was only detectable in the postmitochondrial supernatant during UPRmt-activating stress (Fig. 2D and fig. S6). Import was not completely blocked during UPRmt activation because the processed forms of ATFS-1 [revealed by lon(RNAi)] and GFPmt were detected in mitochondria (Fig. 2, C and D).

Fig. 3

HAF-1 modulates UPRmt signaling by slowing mitochondrial import of ATFS-1. (A) Photomicrographs of wild-type and haf-1(ok705); hsp-60pr::gfp worms raised on control(RNAi), tim-23(RNAi), cco-1(RNAi), EtBr, or paraquat (0.5 mM). Scale bar, 0.5 mm. The images for cco-1(RNAi) and paraquat were exposed longer because of smaller worm size. (B) Immunoblots of wild-type, clk-1(qm30), or clk-1(qm30); haf-1(ok705) worms raised on control(RNAi) or lon(RNAi). (C) Photomicrographs of atfs-1(tm4525); hsp-60pr::gfp worms expressing ATFS-1FL, ATFS-1∆1–32.myc, or ATFS-1∆1–32.myc.∆NLS raised on control(RNAi). The lower panel harbors the haf-1(ok705) allele. Scale bar, 0.5 mm. (D) Schematic illustrating ATFS-1 regulation.

Import into the matrix requires the TOM complex, the TIM23 complex, the ETC, and the matrix-localized molecular chaperone mtHsp70 (1). tim-23(RNAi) caused ATFS-1::GFP to accumulate within nuclei (Fig. 2A) and strongly induced hsp-60pr::gfp expression (Fig. 3A and fig. S7). Furthermore, impairment of mtHsp70 (3) or the ETC by the isp-1(qm150) mutation (13) or cco-1(RNAi) also activated the UPRmt, which suggests that ATFS-1 responds to mitochondrial protein import perturbations (Fig. 3A).

We next considered how HAF-1, the previously identified UPRmt regulator (5), affected ATFS-1. As expected, haf-1(ok705) worms were unable to induce hsp-60pr::gfp expression caused by the clk-1(qm30) mutation (13) (Fig. 3B) or when raised on EtBr (30 μg/ml) (Fig. 3A). However, hsp-60pr::gfp induction caused by mitochondrial stresses that arrest worm development, such as treatment with EtBr (100 μg/ml) or spg-7(RNAi), did not require haf-1 (fig. S8). Additionally, UPRmt activation caused by conditions that directly inhibited mitochondrial import, such as treatment with tomm-40(RNAi) (fig. S8A), tim-23(RNAi), cco-1(RNAi), or paraquat (9), also did not require haf-1 (Fig. 3A). These findings suggest that HAF-1 affects UPRmt signaling by modulating the mitochondrial import of ATFS-1. During mitochondrial stress, steady-state measurements indicated that more ATFS-1 accumulated within mitochondria of haf-1(ok705) worms as revealed by lon(RNAi) (Fig. 3B and fig. S6B). Thus, in the absence of haf-1, ATFS-1 partitions to mitochondria during stress, thereby reducing UPRmt activation.

To more directly examine the effect of haf-1(ok705) on ATFS-1 mitochondrial import efficiency, we expressed ATFS-1 from the inducible hsp-16 promoter (fig. S6). As discussed above, unprocessed ATFS-1 accumulated in the postmitochondrial supernatant during UPRmt-activating stress. However, in haf-1(ok705) worms, much less ATFS-1 was detected in the cytosol (Fig. 2D). Similarly, the slowed import of GFPmt during mitochondrial stress was not observed in haf-1(ok705) animals (Fig. 2C); this finding suggests that HAF-1 is a general attenuator of mitochondrial protein import during stress and that UPRmt signaling is probably modulated by HAF-1 in this manner.

To determine whether prevention of ATFS-1 import into mitochondria is sufficient to cause ATFS-1 nuclear accumulation and UPRmt activation, we generated a series of transgenic lines that expressed ATFS-1FL (full-length wild-type ATFS-1), ATFS-1∆1–32.myc (an engineered variant of ATFS-1 unable to be imported into mitochondria), and ATFS-1∆1–32.myc∆NLS (a variant lacking the NLS) (fig. S9). Removal of the MTS was sufficient to cause nuclear accumulation of ATFS-1∆1–32.myc::GFP (Fig. 2A), and expression of ATFS-1∆1–32.myc caused constitutive expression of hsp-60pr::gfp, indicating that impaired mitochondrial import of ATFS-1 is sufficient for UPRmt activation. Activation of hsp-60pr::gfp by ATFS-1∆1–­32.myc did not require haf-1 (Fig. 3C), further supporting a role for HAF-1 in mitochondrial import regulation. Mutating the NLS in ATFS-1 lacking the MTS prevented hsp-60pr::gfp expression, indicating that when ATFS-1 cannot be imported into mitochondria, ATFS-1 requires the NLS for UPRmt activation (Fig. 3, C and D).

To examine the physiological role of ATFS-1 during mitochondrial dysfunction, we raised wild-type, clk-1(qm30), and isp-1(qm150) (13) worms on control(RNAi) or atfs-1(RNAi). Although unstressed worms were unaffected by atfs-1(RNAi), the mitochondrial-stressed worms were unable to develop (Fig. 4A and fig. S13A). Because ATFS-1 is regulated by mitochondrial import efficiency, which is linked to mitochondrial function, we sought to identify the entire ATFS-1–mediated response. Transcripts from wild-type and atfs-1(tm4525) worms raised in the presence and absence of stress were compared. A broad transcriptional response totaling 685 genes was induced during mitochondrial stress (Fig. 4B and table S2), of which 391 required atfs-1 (table S3).

Fig. 4

ATFS-1 mediates a broad and protective transcriptional program. (A) Representative photomicrographs of wild-type or isp-1(qm150) worms raised on control(RNAi) or atfs-1(RNAi). Scale bar, 1mm. (B) Heat map comparing gene expression patterns of wild-type or atfs-1(tm4525) worms raised on control(RNAi) or spg-7(RNAi). (C) Functional categories of the 391 ATFS-1–dependent genes identified by hierarchical clustering. (D to G) Expression levels of dnj-10, skn-1, gpd-2, and tim-23 mRNA in wild-type or atfs-1(tm4525) worms raised on control(RNAi) or spg-7(RNAi) determined by quantitative reverse transcription polymerase chain reaction; N = 3, ± SD, *P < 0.05 (Student's t test).

Included in the ATFS-1 program were many genes with roles in protecting against mitochondrial dysfunction (Fig. 4C), including the mitochondrial chaperone genes dnj-10 (14) (Fig. 4D) and hsp-60 (fig. S13B). Numerous components involved in reactive oxygen species detoxification required ATFS-1 for induction during stress, including the transcription factor skn-1 (15) (Fig. 4E). Several glycolysis genes including gpd-2 (encoding glyceraldehyde-3-phosphate dehydrogenase) were also induced (Fig. 4F), which suggests that the UPRmt may contribute to a shift in adenosine triphosphate production from respiration to glycolysis. ATFS-1 was also required for the induction of tim-23 and tim-17 (Fig. 4G and fig. S13C), core components of the TIM23 complex (1).

The presence of a NLS and a MTS in a single transcriptional activator allows the cell to monitor global mitochondrial import efficiency and determine the level of mitochondrial dysfunction. If mitochondria are functioning properly, the constitutively synthesized ATFS-1 partitions into mitochondria where it is degraded. As mitochondrial dysfunction increases, mitochondrial import efficiency is reduced, favoring translocation of ATFS-1 to the nucleus. Thus, mitochondrial homeostasis is maintained by the stress-dependent partitioning of a transcriptional activator between an inactive state in mitochondria and an active state in the nucleus.

Supplementary Materials

Materials and Methods

Figs. S1 to S13

Tables S1 to S3

References (1622)

References and Notes

  1. Acknowledgments: Supported by the Louis V. Gerstner Jr. Young Investigators Fund, the Alfred W. Bressler Scholar Fund, the Ellison Medical Foundation, and NIH grant R01AG040061. We thank K. Rainbolt for lon(RNAi) advice, the National BioResource Project, and the Caenorhabditis Genetics Center and the Genomics Facility at Memorial Sloan-Kettering Cancer Center. The microarray data have been submitted in MIAME-compliant format to the Gene Expression Omnibus database (accession no. GSE38196).
View Abstract

Navigate This Article