TAF-4 is required for the life extension of isp-1, clk-1 and tpk-1 Mit mutants

Introduction

In humans, many mutations that disrupt the mitochondrial electron transport chain (ETC) result in pathogenesis; several are the cause of debilitating childhood illnesses [1]. Altered mitochondrial function has also been implicated in various late-onset diseases, including Alzheimer's, Parkinson's and Huntington's Diseases [2]. Unexpectedly, mutations that reduce mitochondrial ETC function can sometimes extend life rather than shorten it. In the nematode Caenorhabditis elegans, for example, when any of several components of the ETC are disrupted, either by mutation or RNAi-treatment, lifespan is increased anywhere from 10% to 3-fold over wild-type [3]. RNAi knockdown of subunits in complex I, III, IV or V can extend the lifespan of Drosophila melanogaster as well [4]. In the yeast Saccharomyces cerevisiae, several mutations that reduce mitochondrial respiration are also associated with increased replicative lifespan, a surrogate measure of mother cell longevity [5]. Finally, in Mus musculus, removal of Surf1 (which encodes a cytochrome c oxidase assembly factor [6]), or hemizygous removal of Mclk1 (which encodes demethoxyubiquinone monoxygenase, a gene essential for ubiquinone biosynthesis[7]), results in longer-lived animals. Thus, multiple organisms from phyla separated by millions of years of evolution, all respond to some forms of mitochondrial ETC dysfunction with increased lifespan.

Electron transport chain disruptants of C. elegans that have extended lifespan are collectively called Mit (Mitochondrial) mutants [8]. While most of the 40+ Mit mutants have been identified via large-scale RNAi screening [9], a handful of bona fide mutants have been identified in chemical screens, including the missense or nonsense alleles of isp-1(qm150), clk-1(e2519 and qm30), and tpk-1(qm162). These genes encode the Rieske iron-sulfur protein [10], the worm ortholog of DMQ monoxygenase [11], and thiamine pyrophosphokinase (TPK-1) [12], respectively. Loss of TPK-1 disrupts cellular thiamine levels, which in turn disrupts the α-ketoacid dehydrogenases that operate the mitochondrial tricarboxylic acid cycle that feeds electrons into the ETC [12, 13]. Recently, we presented evidence suggesting that inhibition of the α-ketoacid dehydrogenases (namely pyruvate dehydrogenase, α-ketoglutarate dehydrogenase and branched-chain α-ketoacid dehydrogenase) may be the proximal mechanism leading to the characteristic Mit phenotype [13]. The downstream molecular players that permit Mit mutants and their defective mitochondria to increase lifespan remain, however, largely unresolved.

In humans, it is well established that many mutations that disrupt the mitochondrial ETC manifest as threshold effect disorders [8]. That is, the phenotypic appearance of the diseased state is not linearly-dependent upon the degree of ETC inhibition caused by the mutation. Compensatory mechanisms, known as retrograde responses, act to delay, or in extreme cases even prevent, disease manifestation. The same is true in the C. elegans Mit mutants [3, 14]. A novel retrograde-like response was recently delineated in C. elegans by Haynes and colleagues. Under normal conditions, the transcription factor ATFS-1 is imported into mitochondria and degraded. Following mitochondrial dysfunction, cytoplasmically-localized ATFS-1 is instead trafficked to the nucleus whereupon it directly activates a suite of some 391 genes [15]. Loss of ATFS-1 is lethal to both isp-1(qm150) and clk-1(qm30) Mit mutants. Five additional transcription factors are known to be required for Mit mutant life extension. These are p53/cep-1[14], hypoxia factor-1 (HIF-1) [16], the C.elegans-specific homeodomain protein CEH-23 [17], the E-twenty-six transcription factor ETS-9 [17], and the nuclear hormone receptor, NHR-25 [17]. A sixth transcription factor, CREB homolog-1 (CRH-1), is almost certainly required as well, given that aak-2/AMP kinase (AMPK) is essential for the life extension of isp-1(qm150) and clk-1(qm30) Mit mutants [18] and for over-expressed AMPK to extend lifespan it must phosphorylate CRH-1 [19]. It is clear, from all these studies, that the life extending mechanisms that are invoked following mitochondrial ETC dysfunction in C. elegans involves coordination between multiple pathways. In this study, we now extend these findings by undertaking a screen of worm transcription factors. We establish TAF-4 as a novel transcription factor that mediates the lifespan extension in three distinct Mit mutants: isp-1, tpk-1 and clk-1.

Results

The Mit phenotype is defined by the concordant appearance of several traits that co-segregate with life extension; these include slowed larval development, reduced adult size, and diminished fertility [3]. We and others have also observed that many stress-response genes, including hsp-6 [15] and gst-4 [21] (markers of UPR^mt activation and oxidative stress, respectively), are constitutively activated in Mit mutants. We constructed an isp-1(qm150); Pgst-4::GFP transcriptional reporter strain and observed that GFP expression was robustly upregulated in the hypodermis and neurons of these animals at the L4/YA boundary – a time when mitochondria are rapidly expanding [23]. GFP expression slowly subsided in these animals as they aged until it reached a level indistinguishable from that in control Pgst-4::GFP animals, which contain a faint, albeit constitutive and SKN-1 independent, background hypodermal signal [21].

Fifteen transcription factors control multiple aspects of the Mit phenotype in isp-1(qm150) mutants

To expedite discovery of transcription factors required for Mit mutant life extension we monitored several surrogate phenotypes that co-segregate with increased lifespan in isp-1(qm150); Pgst-4::GFP mutants. Phenotypes that were scored include (i) rate of larval development, (ii) viability, (iii) GFP reporter-gene fluorescence, (iv) gross morphology, (v) fertility and (vi) fecundity. We screened over 400 bacterial feeding RNAi clones targeting approximately 40% of the known and predicted transcription factors in C. elegans [24] (Supplementary Table II). After three rounds of screening, twenty-four RNAi clones were isolated that differentially altered at least one longevity co-segregating phenotype in isp-1(qm150); Pgst-4::GFP mutants relative to control Pgst-4::GFP animals. To address possible differences in the efficacy of RNAi between the two strains, and to be certain that the qualitative phenotypes that we used to identify the set of twenty-four RNAi clones were also robust across investigator, we tested if the F₁ progeny from each strain continued to show the same differential responses as their parents when further cultured on RNAi lawns. Fifteen RNAi clones passed an additional three rounds of F₁ screening when scored by an independent investigator. The differential effects caused by each RNAi on the two strains are summarized in Table I (refer also to Supplementary Figure S1).

Figure 1. Flow Chart Describing Screening Methodology.

Feeding RNAi clones corresponding to more than 400 transcription factors were retrieved from the Ahringer RNAi library. isp-1; Pgst-4::GFP double mutants (TJ564) and a Pgst-4::GFP control strain (CL2166), both containing GFP under the control of an oxidative stress-sensitive promoter, were screened in parallel for differential changes in reporter expression, development, adult size, fecundity and fertility. Fifteen transcription factors were identified that reproducibly and differentially altered some aspect of the isp-1 phenotype relative to the control strain (Table I). Seven of the identified RNAi clones significantly reduced isp-1 lifespan. Of these only two clones, targeting taf-4 and hif-1, significantly and differentially reduced the lifespan of isp-1, clk-1 and tpk-1 Mit mutants relative to control strains.

Table 1. Transciption Factor RNAi Clones That Differentially Effect Lifespan Co-segregating Phenotypes of isp-1(qm150)

RNAi	CL2166 Phenotype (vs vector control*)	TJ564 Phenotype (vs vector control)
aha-1	Most vector-like, some slow to develop	Most vector-like, some slow to develop and show weak GFP induction
ceh-18	Severe developmental asynchrony, fewer F1	Severe developmental asynchrony, almost sterile
ceh-24	Vector-like	Medium GFP induction
daf-2**	Vector-like	Thin adults, almost infertile, many abandoned plate
dpr-1	Vector-like	Almost infertile, GFP weakly induced
hif-1	GFP weakly induced	GFP strongly induced
jun-1	Vector-like	Severe developmental asynchrony, sterile adults, some L4 arrest, GFP variably induced
elt-3	GFP-suppressed in head hypodermis***	Reduced fertility, medium/strong GFP suppression
nhr-105	GFP-suppressed in head hypodermis***	Medium GFP suppression
nhr-125	Vector-like	Mild developmental asynchrony, adults weakly induce GFP
nhr-214	Vector-like	Mild developmental asynchrony (potential L3/L4 arrest)
nhr-27	Vector-like	Reduced fertility (mild), weak/medium GFP induction, F1 pale
nhr-49	Reduced fecundity, GFP dim	Reduced fertility, worms small and thin, GFP suppressed
taf-4	Delayed egg production, medium GFP induction	Delayed development, sterile, no GFP induction
tbx-2	Vector-like	Reduced fertility, weak/medium GFP suppression
Y62E10A.17	Vector-like	Delayed development, delayed egg production
* Vector control = pL4440-containing HT115 bacteria cultured on RNAi plates (see Materials & Methods for details).
** daf-2 is not a transcription factor. It was included in the screen as a blind control.
*** Both elt-3 and nhr-105 showed an unusual suppression of background GFP fluorescence in head hypodermis. Background staining in remaining hypodermal cells appeared normal.

Among the set of identified clones, probability analysis using a hypergeometric distribution revealed that RNAi targeting genes encoding transcription factors belonging to the basic helix-loop-helix (HIF-1, AHA-1) and basic leucine zipper domain (JUN-1) sub-families occurred more frequently than expected by chance (p = 0.0079 and p = 0.039, respectively; Figure 2A). RNAi clones targeting six nuclear hormone receptor genes were also retrieved: dpr-1, nhr-27, nhr-49, nhr-105, nhr-125 and nhr-214. Given the nuclear hormone receptor (NHR) bias in our transcription factors sub-library, this represented neither an enrichment nor de-enrichment of these genes (p = 0.09). Of particular note, phylogenetic analyses revealed that two of these genes, nhr-27 and nhr-214, encode closely related NHR homologs (Figure 2B,C). The most recent ancestor of nhr-214 is nhr-272 but the RNAi targeting this gene was not retrieved in our screen. The six additional RNAi clones that we identified targeted genes encoding two homeodomain-containing proteins (CEH-18 and CEH-24), the GATA-recognizing zinc-finger protein ELT-3, the T-box containing transcription factor TBX-2, TAF-4 which is a component of the TFIID complex that forms part of the general RNA Pol II transcriptional machinery, and finally, an unnamed transcription factor (Y62E10A.17) belonging to the activating protein 2 (AP-2) family of transcription factors. We have called this gene udi-1 for ‘Unruly Development with isp-1’ in reference to the opposite phenotypes observed in isp-1(qm130) Po and F₁ generation animals when fed RNAi against this gene (compare Table I with Supplementary Table I).

Figure 2. Transcription Factor Sub-Families are Enriched Within Set of Identified RNAi Clones.

(A) Hypergeometric probability analysis was performed on the 15 transcription factors identified in our phenotypic screen. RNAi targeting genes encoding transcription factors belonging to the basic helix-loop-helix (HIF-1, AHA-1) and basic leucine zipper domain (JUN-1) sub-families occurred more frequently than expected by chance. (B) Locus surrounding nhr-27 and nhr-214 on the X chromosome of C. elegans (Wormbase, Version: WS236). Secondary RNAi targets are sites where RNAi clones are predicted to cross-react with unintended gene loci (JA: refers to the Julie Ahringer RNAi library clones used in this study). (C) Phylogenetic relationship between nhr-27, nhr-214 and other nuclear hormone receptors. (Tree was constructed by www.treefam.org). Numbers indicate bootstrap values (percentage of times branch pattern was maintained).

Figure 3 shows the effect each of the fifteen RNAi clones had on the expression of the Pgst-4::GFP transgene in the isp-1(qm150)and control backgrounds. RNAi targeting hif-1 markedly and robustly induced the Pgst-4::GFP reporter in isp-1 mutants. Knockdown of aha-1, which encodes the heterodimeric partner of HIF-1, also induced Pgst-4::GFP in these animals, but less strongly than hif-1. Loss of ceh-24, dpr-1, nhr-27 and nhr-125 all resulted in weak GFP reporter induction in isp-1 mutants while elt-3, nhr-49, nhr-105, taf-4 and tbx-2 all suppressed GFP induction in isp-1 mutants below background levels. In control animals, a weak induction of GFP was observed when animals were exposed to RNAi targeting hif-1 while nhr-49 RNAi resulted in a slight reduction compared to background levels. Finally, elt-3 and nhr-105 reproducibly and specifically decreased the background staining of GFP in hypodermal cells of the head only (Figure 3 and Supplementary Figure S1 A).

Figure 3. Pgst-4::GFP Reporter Expression in isp-1 Mit Mutants is Modulated by Multiple Transcription Factors.

isp-1(qm150) Mit mutants constitutively activate the Pgst-4::GFP reporter from the L4 stage to mid-adulthood. Shown are the effects of different transcription factor feeding RNAi treatments on the expression of GFP in one-day adult isp-1(qm150); Pgst-4::GFP worms versus control one-day old adult Pgst-4::GFP worms. All worms were placed on RNAi lawns as one day arrested L1 and consequently daf-2 RNAi resulted in isp-1(qm150); Pgst-4::GFP dauer arrest (compare Supplementary Figure S2). The Pgst-4::GFP reporter is constitutively activated in dauer worms (not shown).

During our screening procedure, we made an unexpected finding: the method by which Po animals were prepared altered the final effect of some RNAi clones (Supplementary Figure S1). This effect was most evident in isp-1(qm150); Pgst-4::GFP animals fed daf-2 RNAi. In C. elegans, daf-2 encodes the sole insulin/IGF-like receptor tyrosine kinase and RNAi targeting this gene was originally included in our screen as a control. When isp-1(qm150); Pgst-4::GFP animals were placed on daf-2 RNAi as 24 hour-arrested L1 larvae, a synthetic dauer arrest phenotype was observed (Table I, Figure 3). If animals were instead placed on the lawns as eggs, so that larvae hatched directly onto their RNAi food source, these animals developed into fertile adults (Supplementary Figure S1 B). A previous study reported daf-2(e1370); isp-1(qm130) double mutants form fertile adults with a lifespan extension roughly additive of the two single mutants' [10]. That study also reported that isp-1(qm150) mutants remain long-lived, even in the presence of a daf-16 loss of function mutation, which is epistatic to daf-2. These findings have been used to argue against a role for insulin/IGF-1 like signaling in Mit mutant longevity. Our current findings now show that early during larval development, the Daf and Mit longevity pathways do, in fact, interact.

Seven transcription factors modulate isp-1(qm150) life extension

Having identified a set of transcription factors that alter various aspects of the Mit phenotype in isp-1 mutants, we next sought to determine if any of these proteins also play a role in the extended lifespan of these animals. We performed lifespan analyses on over six thousand isp-1(qm150); Pgst-4::GFP and control Pgst-4::GFP animals that encompassed multiple replicates of each strain cultured on the fifteen different transcription factor RNAi clones. Replicate experiments (≥3) were performed by multiple investigators (Figure 4, and Supplementary Figures S3-S5) so that only the most robust effects were likely to be observed. We identified seven transcription factors that are required for isp-1 life extension. Knockdown of aha-1, ceh-18, hif-1, jun-1, nhr-27, nhr-49 and taf-4 all resulted in a greater reduction of isp-1 lifespan than wild-type (descriptive statistics and significance values are provided in Table II and Supplementary File 3). Of these RNAi treatments, jun-1 and taf-4 knockdown completely abolished the increased lifespan of isp-1(qm150);Pgst-4::GFP mutants.

Figure 4. isp-1 Life Extension is Under the Control of Multiple Transcription Factors.

Seven transcription factors significantly shortened the lifespan of isp-1; Pgst-4::GFP Mit mutants relative to Pgst-4::GFP worms when tested across multiple, independent replicates. (A-G) Survival curves are presented as average of multiple replicates (≥3 replicates per condition, ±S.E., N=60 worms/replicate) for each strain on each of the seven RNAi clones (strain and RNAi clone are marked) as well as vector control. Complete lifespan data are provided in Supplementary File 3 and summarized in Table II. RNAi-mediated knock-down of nhr-49 significantly shortened the lifespan of both strains. daf-2 RNAi was included as a positive control.

Table II

Combined Survival Statistics (nTOTAL = 3135)	Stratified* Cox Hazard Model Estimates
RNAi	Number of Worms (censored)	Mean Expected Lifespan (Days)	Standard Error	Log Rank (vs.Vector)**	Hazard Ratio (RNAi/Vector)	95% C.I. (Lower)	95% C.I. (Upper)	p Value***
aha-1	180 (27)	17.68	0.31	4.9E-02	0.777	0.619	0.975	0.03
ceh-18	181 (101)	18.37	0.48	3.4E-02	0.64	0.486	0.841	0.00
ceh-24	190 (50)	18.94	0.32	2.0E-03	0.639	0.506	0.806	<0.001
daf-2	180 (3)	29.03	0.49	5.2E-63	0.113	0.088	0.146	<0.001
dpr-1	181 (30)	19.26	0.34	1.6E-05	0.574	0.457	0.722	<0.001
elt-3	182 (32)	18.32	0.37	8.9E-03	0.634	0.504	0.797	<0.001
hif-1	180 (38)	17.93	0.29	5.1E-01	0.938	0.745	1.181	0.59
jun-B	180 (51)	17.27	0.26	6.8E-01	1.052	0.83	1.333	0.67
nhr-105	180 (28)	17.38	0.34	6.7E-01	0.848	0.674	1.066	0.16
nhr-125	180 (24)	19.11	0.33	1.5E-04	0.591	0.471	0.742	<0.001
nhr-214	180 (16)	18.99	0.29	2.8E-04	0.686	0.549	0.857	<0.001
nhr-27	180 (38)	18.83	0.36	2.3E-04	0.61	0.484	0.769	<0.001
nhr-49	180 (10)	13.48	0.16	6.4E-38	4.738	3.768	5.958	<0.001
taf-4	181 (13)	18.25	0.31	6.8E-03	0.746	0.597	0.932	0.01
tbx-2	180 (52)	17.37	0.34	9.3E-01	0.923	0.728	1.17	0.51
Y62E10A.17	180 (16)	18.09	0.29	9.8E-02	0.848	0.679	1.06	0.15
Vector	241 (41)	17.71	0.25	-	1	-	-	-

Combined Survival Statistics (nTOTAL = 3256)	Stratified* Cox Hazard Model Estimates
RNAi	Number of Worms (censored)	Mean Expected Lifespan (Days)	Standard Error	Log Rank (vs.Vector)**	Hazard Ratio (RNAi/Vector)	95% C.I. (Lower)	95% C.I. (Upper)	p Value***
aha-1	189 (29)	21.24	0.72	5.7E-11	1.634	1.312	2.035	<0.001
ceh-18	185 (10)	21.81	0.49	5.5E-31	2.303	1.853	2.862	<0.001
ceh-24	180 (9)	27.47	0.61	5.5E-06	0.983	0.792	1.22	0.88
daf-2	152 (47)	40.15	1.25	2.4E-21	0.216	0.161	0.289	<0.001
dpr-1	184 (23)	29.53	0.57	2.5E-02	0.849	0.682	1.056	0.14
elt-3	185 (19)	29.75	0.58	2.2E-02	0.84	0.677	1.043	0.12
hif-1	180 (9)	22.25	0.55	3.3E-23	2.35	1.894	2.916	<0.001
jun-B	183 (24)	14.59	0.38	4.5E-87	10.205	8.101	12.855	<0.001
nhr-105	180 (12)	27.75	0.64	3.1E-04	0.936	0.753	1.164	0.55
nhr-125	180 (18)	28.31	0.67	5.4E-03	0.95	0.764	1.182	0.65
nhr-214	183 (35)	27.73	0.65	3.0E-03	1.123	0.9	1.402	0.30
nhr-27	182 (18)	23.13	0.66	1.3E-13	1.752	1.409	2.178	<0.001
nhr-49	195 (2)	17.76	0.36	3.7E-69	5.027	4.045	6.246	<0.001
taf-4	180 (41)	19.24	0.33	7.3E-53	4.356	3.446	5.508	<0.001
tbx-2	180 (20)	28.74	0.66	1.8E-02	0.912	0.732	1.136	0.41
Y62E10A.17	180 (11)	28.66	0.57	3.8E-06	1.151	0.929	1.427	0.20
Vector	358 (21)	30.21	0.50	-	1	-	-	-

Combined Survival Statistics (nTOTAL = 1085)	Stratified* Cox Hazard Model Estimates
RNAi	Number of Worms (censored)	Mean Expected Lifespan (Days)	Standard Error	Log Rank (vs.Vector)**	Hazard Ratio (RNAi/Vector)	95% C.I. (Lower)	95% C.I. (Upper)	p Value***
aha-1	120 (13)	28.86	0.76	4.2E-04	1.351	1.037	1.76	0.026
ceh-18	120 (15)	30.03	0.76	2.4E-04	1.353	1.037	1.763	0.026
hif-1	121 (11)	29.26	0.68	1.3E-04	1.458	1.121	1.895	0.005
jun-1	120 (6)	20.95	0.48	4.7E-41	5.684	4.294	7.523	<0.001
nhr-105	120 (4)	27.94	0.71	4.8E-09	1.946	1.498	2.527	<0.001
nhr-27	121 (5)	33.17	0.74	8.3E-01	0.885	0.683	1.147	0.356
taf-4	120 (2)	21.36	0.44	4.5E-41	5.739	4.342	7.585	<0.001
Vector	243 (16)	32.15	0.56	-	1	-	-	-

Combined Survival Statistics (nTOTAL = 540)	Stratified* Cox Hazard Model Estimates
RNAi	Number of Worms (censored)	Mean Expected Lifespan (Days)	Standard Error	Log Rank (vs. Vector)**	Hazard Ratio (RNAi/Vector)	95% C.I. (Lower)	95% C.I. (Upper)	p Value***
aha-1	60 (1)	22.69	1.49	4.3E-01	1.223	0.85	1.759	0.277
ceh-18	60 (2)	23.54	1.17	5.0E-01	1.19	0.824	1.72	0.353
hif-1	60 (2)	20.36	1.12	1.7E-03	1.834	1.268	2.653	0.001
jun-1	60 (2)	26.17	0.96	6.2E-01	1.128	0.779	1.633	0.525
nhr-105	60 (1)	22.55	1.55	3.6E-01	1.213	0.845	1.742	0.294
nhr-27	60 (1)	24.57	1.15	8.3E-01	1.048	0.727	1.51	0.803
taf-4	60 (2)	19.01	0.73	7.6E-04	1.954	1.339	2.85	<0.001
Vector	120 (2)	24.06	1.04	-	1	-	-	-

Combined Survival Statistics (nTOTAL = 545)	Stratified* Cox Hazard Model Estimates
RNAi	Number of Worms (censored)	Mean Expected Lifespan (Days)	Standard Error	Log Rank (vs.Vector)**	Hazard Ratio (RNAi/Vector)	95% C.I. (Lower)	95% C.I. (Upper)	p Value***
aha-1	59 (10)	22.43	0.76	5.6E-01	1.043	0.716	1.521	0.825
ceh-18	61 (7)	30.44	1.44	2.8E-09	0.258	0.17	0.391	<0.001
hif-1	61 (5)	27.20	0.99	1.8E-03	0.535	0.369	0.775	<0.001
jun-1	60 (5)	23.05	0.78	6.4E-01	0.904	0.627	1.303	0.588
nhr-105	61 (8)	27.03	0.95	6.1E-03	0.557	0.384	0.809	0.002
nhr-27	60 (19)	25.45	0.96	2.0E-02	0.656	0.441	0.977	0.038
taf-4	60 (4)	19.55	0.44	2.4E-04	1.805	1.24	2.627	0.002
Vector	122 (6)	23.18	0.56	-	1	-	-	-

*

Stratified by experiment,

**

Bold italic indicates significant after controlling for multiple comparisons,

***

Bold italic indicates significant (p<0.05).

In particular, jun-1 RNAi pathologically shortened lifespan. Both RNAi treatments were without effect on Pgst-4::GFPcontrol animals. RNAi knockdown of nhr-49 significantly reduced the lifespan of both strains while, as expected, daf-2 knockdown significantly increased the lifespan of both strains (in isp-1 animals that did not experience dauer arrest). The lifespan increase by daf-2 was additive with isp-1. Interestingly, RNAi targeting dpr-1, nhr-27, nhr-125 and nhr-214 all marginally (~6%), but significantly (p<0.0003), increased the lifespan of Pgst-4::GFP animals (Table II and Supplementary Figure S5).

Mit mutant life extension is mediated by a general set of transcription factors

We next tested the generality of aha-1, ceh-18, hif-1, jun-1, nhr-27, andtaf-4 in the life extension of two other Mit mutants: tpk-1(qm162) andclk-1(qm30). We included RNAi targeting nhr-105 in this analysisalso, since knock-down of this transcription factor caused a marginally significant reduction in isp-1 lifespan (Table II). Survival curves for tpk-1(qm162)and clk-1(qm30) are presented in Figures 5 and 6, respectively. Summary statistics and significance testing are described in Table II. All RNAi clones except nhr-27 significantly reduced the lifespan of tpk-1 mutants. As was observed for isp-1(qm150), RNAi targeting jun-1 and taf-4 both caused potent life shortening in tpk-1(qm162) mutants.

Figure 5. Six Transcription Factors Required for isp-1(qm150) Life Extension are also Required for tpk-1(qm162) Life Extension.

Replicate survival plots for tpk-1(qm162) worms cultured on RNAi targeting transcription factors that differentially shortened the lifespan of isp-1 mutants. Knock-down of six transcription factors, taf-4, jun-1, hif-1, aha-1, ceh-18, as well as nhr-105, each significantly reduced tpk-1 lifespan. nhr-27 RNAi was without effect (N=60 worms/trace). Vector-only RNAi is included in each plot as the relevant control and the single curve represents combined data from four independent assays (error bars: +/−SEM, N=262 worms total). Complete lifespan data are provided in Supplementary File 3 and summarized in Table II.

Figure 6. TAF-4 and HIF-1 are Required for clk-1(qm30) Life Extension.

(a) Survival curves for clk-1(qm30); Pgst-4::GFP transgenic worms cultured on RNAi targeting each of seven transcription factors required for isp-1(qm150) life extension. For this panel, as well (b) and (c), a summary of lifespan data and significance testing is provided in Supplementary File 3 and Table II. (b) In the absence of the Pgst-4::GFP reporter, clk-1(qm30) remains surprisingly recalcitrant to any life-shortening effect of the seven RNAi clones. Only taf-4 and hif-1 RNAi significantly shorten lifespan. (c) Survival analysis of clk-1(qm30), clk-1(e2519) and N2 wild-type animals cultured on standard NGM/OP50 plates. Mean lifespan of each strain is 29.5, 21.3, 18.8 days, respectively. Both qm30 and e2519 are significantly longer-lived than N2 (p=0.049 and 1.3E-09, respectively. (N=60 worms/trace, for all panels). (D) Model depicting role of TAF-4 in Mit mutant life extension. TAF-4 is a component of the general transcription machinery leading to recruitment of RNA Pol II to basal promoters. CREB contains binding sites for TAF-4, CRTC-1 and CBP. AMPK activation inhibits the interaction of CBP and CRTC-1 with CREB. Loss of TAF-4 blocks Mit mutant life extension, so too does removal of AMPK.

The situation for clk-1(qm30) mutants was considerably different. In clk-1(qm30); Pgst-4::GFP animals, only knockdown of taf-4 significantly shortened lifespan (Figure 6A, Table II). Unexpectedly, knockdown of ceh-18, hif-1, nhr-105 and nhr-27 all resulted in a significant extension of lifespan. For nhr-27, this effect was similar to that seen for control Pgst-4::GFP animals (Figure 4). When we repeated this experiment using clk-1(qm30) mutants without the Pgst-4::GFP reporter in the background, again taf-4 RNAi significantly shortened lifespan, but so too did RNAi targeting hif-1 (Figure 6B, Table II); none of the other RNAi clones altered lifespan. Surprised by the shape of the survival curves for the clk-1(qm30) strain (which is suggestive of two sub-populations [25]), and by the overall weakness of the life extension observed in both strains when cultured on vector-only RNAi, we decided to test the lifespan of the clk-1(qm30) strain when grown on standard NGM/OP50 plates. Under these conditions robust extension of life was observed (Figure 6C), as reported previously [11]. Given that clk-1(qm30) acquires its ubiquinone from its bacterial food source [26], the most likely explanation for these unexpected results is that the HT115 bacteria used for feeding RNAi purposes more effectively rescues the ubiquinone deficit of clk-1(qm30) animals than does OP50 bacteria, in effect, making them only weak Mit mutants. Consistent with this notion, we found that the loss of function qm30 allele grown on HT115 had survival kinetics more akin to the hypomorphic e2519 allele cultured on OP50 (Figure 6C). Based on our current results, we conclude that at least two transcription factors, TAF-4 and HIF-1, appear to be generally required for Mit mutant life extension. Four other transcription factors, AHA-1, CEH-18, NHR-105 and JUN-1, are also likely to be important mediators of lifespan in this class of long-lived C. elegans mutants.

Discussion

Since the discovery of the first Mit mutant almost twenty years ago [11], significant effort has been expended in trying to determine why this group of worms are long-lived. These animals met early resistance as a relevant model to study aging since, at the time, it was almost unfathomable that a mutant defined by mitochondrial ETC dysfunction could also be long-lived. With time it became clear that life extension in the face of ETC disruption was not peculiar to C. elegans [4, 6], that manifestation of the Mit phenotype is a threshold effect phenomenon [3], and that multiple processes are activated in these animals to mitigate the detrimental effects of ETC loss with the net consequence being an extension of lifespan [3, 8, 14-18]. Our present studies now extend considerably our understanding of the transcription factor network underlying appearance of the Mit phenotype. Through a focused RNAi screen, we have uncovered a trove of DNA-binding proteins that affect various aspects of the Mit phenotype in isp-1(qm150) mutants. We have identified fifteen transcription factors in total, seven of which are required for isp-1 life-extension, and two of these, namely TAF-4 and HIF-1, are also required to extend the lifespan of two other Mit mutants, tpk-1(qm162) and clk-1(qm30).

HIF-1α is a conserved transcription factor required in many organisms for survival under low oxygen conditions. In worms, HIF-1α (HIF-1) controls the expression of over 60 genes following acute hypoxia [27]. Under normoxic conditions, HIF-1 is rapidly degraded, but in several RNAi-induced Mit mutants, as well as in two genetically-defined Mit mutants – isp-1(qm150) and clk-1(qm30) – HIF-1 is constitutively active [16]. Inhibition of HIF-1 by RNAi in isp-1(qm150) and clk-1(qm30) Mit mutants negated their life extension. Additionally, stabilization of HIF-1 in wild-type animals, by inhibition of either VHL-1 or EGL-9 (which encode the E3 ubiquitin ligase and the prolyl hydroxylase required for HIF-1 degradation, respectively), was sufficient to increase the lifespan of wild-type worms but did not further increase the lifespan of Mit mutants. In this study, we independently identified HIF-1 to be required for the extended lifespan of three different genetic Mit mutants: isp-1(qm150), tpk-1(qm162) and clk-1(qm30). We also examined oxidative stress levels in control and isp-1 mutant worms by utilizing the Pgst-4::GFP reporter that was integrated into these strains. The isp-1(qm150) mutants had a high basal level of GFP expression compared to control worms, suggesting that isp-1 mutants are in a state of heightened oxidative stress. RNAi knockdown of HIF-1 had only a mild effect on GFP expression in control worms, while in isp-1 mutants, knockdown of HIF-1 resulted in a strong GFP signal. This suggests that in isp-1(qm150) Mit mutants, functional HIF-1 may be part of a compensatory response that mitigates some of the oxidative stress created by mitochondrial dysfunction. This finding is consistent with recent suggestions that Mit mutant life extension requires ROS signaling [16].

Some of the other transcription factors that we found to be essential for lifespan extension of isp-1(qm150) mutants are AHA-1, CEH-18, JUN-1, NHR-27 and NHR-49. AHA-1 is the worm ortholog of HIF-1β. Under hypoxic conditions, when HIF-1α is stabilized, both proteins form a complex that binds DNA and leads to transcription of genes under the control of the HIF Response Element (HRE) [27]. A previous study revealed that AHA-1 is required for the lifespan extension of isp-1(qm150) and clk-1(qm30) mutants [16]. Our study corroborates that finding; we show that AHA-1 is required for the lifespan extension of isp-1(qm150) mutants, as well as tpk-1(qm162) mutants. CEH-18 is another transcription factor that we found to be required for the life extension of both isp-1(qm150) and tpk-1(qm162) Mit mutants. This protein is known to have a role in oocyte development, where it acts to prevent oocyte maturation and ovulation in the absence of sperm. [28]. In our experiments, ceh-18 RNAi reduced fertility and affected development of both wild-type and isp-1 mutants (Table I) to differing extents. Previously, Hamilton and colleagues showed that ceh-18 RNAi caused a significant extension in mean lifespan of wild-type worms (from 12.3 to 14.5 days) [9]. We also noted a slight increase of lifespan in wild-type worms treated with ceh-18 RNAi (Table II), but this effect was not significant after controlling for multiple comparisons.

We found JUN-1 to also be required for the life extension of isp-1(qm150) and tpk-1(qm162) Mit mutants. JUN-1 is the worm ortholog of the human transcription factor JunB. In mammalian cells, JunB is important for cell cycle regulation and for differentiation in response to growth factors such as transforming growth factor beta-1 (TGFβ1) [29]. In C.elegans, JUN-1 was recently shown to mediate life extension in response to intermittent fasting [30].

NHR-27 and NHR-49 are two more transcription factors identified by our screen. Very little is known about the role of NHR-27 in worms. Its closest human homolog is NR1H4, which is a nuclear bile acid receptor also known as Farnesoid X receptor (FXR). NR1H4 is activated by intracellular bile acids and, in response, regulates expression of a number of bile acid responsive genes, mainly in liver and intestinal cells [31]. Recent studies have uncovered a bile acid-like biosynthetic pathway in worms that governs longevity through the gonad [32]. With regards to NHR-49, this protein is known to regulate fatty acid metabolism in C.elegans in response to changes in nutrient status [33]. In several RNAi-induced Mit mutants, NHR-49 was shown to mediate a starvation-like response that included induction of genes required for β-oxidation, the glyoxylate cycle, gluconeogenesis and glycolysis [34]. Surprisingly, NHR-49 was not required for the life extension of at least four of these RNAi-induced Mit mutants (nuo-1, ucr-1, cyc-1 and cco-3), but it is worth noting that the specific mutant allele employed in those studies affected only three of four NHR-49 splice variants. The effect of NHR-49 loss on genetically-defined Mit mutants had not been tested until now.

The requirement for TAF-4 in Mit mutant longevity has not been previously reported. TAF-4 is a ubiquitously expressed component of the general transcription factor TFIID that participates in formation of the pre-initiation complex necessary for transcription by RNA polymerase II (pol II) [35]. TFIID is comprised of the TATA-binding protein (TBP) and a group of 14 subunits called TBP-associated factors (TAFs), of which TAF-4 is one. Work in C. elegans suggests that TAF-4 is essential for all RNA Pol II transcription in the early embryo [36]. Presumably, this requirement is relaxed after embryonic development since we have now observed that L1 hatchlings exposed to TAF-4 RNAi throughout their remaining life grow normally and show no overt phenotype beyond reduced fertility. Consistent with this finding, mice null for TAF4 are embryonic lethal, but fibroblasts derived from TAF4 null embryos are viable and grow under normal tissue culture conditions [37]. Wright and colleagues showed that TAF4 plays a significant role in mediating transcription from a TATA-less downstream core promoter element, but in TATA-containing promoters, TAF4 was not essential for transcription [38]. In adult mice, TAF subunits show differential expression in a variety of tissues [39], suggesting that the TAFs may selectively and differentially activate tissue specific transcription. Recent studies also show that TAFs may have specificity toward particular core promoters: Gazit and colleagues showed that the TAFs bind DNA with high affinity and have a weak sequence preference [40]. They also showed that TAF4 is required for core promoter function of not all, but only a subset of genes.

One such subset of genes that require TAF4 as a coactivator are genes that are under the control of the glutamine-rich transcription factor CREB (cyclic AMP response element binding protein) [41]. CREB is activated in response to a multitude of extracellular signals including growth factors, hormones and stress signals [42]. CREB and CREB-regulated transcriptional coactivators (CRTCs) control genes involved in a variety of physiological functions including energy homeostasis, carcinogenesis, ER stress, immune function and neurogenesis. Mengus and colleagues showed that induction of CREB transcriptional activity is strongly diminished in taf4^−/− cells [37]. The role of TAF4 as a coactivator for CREB is important in the context of our findings. CREB is known to be induced by mitochondrial dysfunction [43]. Recently, CREB has been shown to mediate lifespan extension in C. elegans downstream of AMPK (5' AMP mediated protein kinase), which is a master regulator of cellular energy balance [19] and essential for the longevity of isp-1(qm150) and clk-1(qm30) Mit mutants [18]. CRTC-1 binds to and activates CREB, leading to transcription of target genes downstream of CREs (cyclic AMP response elements) [44]. Mair and colleagues showed that AMPK phosphorylates CRTC-1, thereby preventing CRTC-1 translocation to the nucleus [19]. They proposed that this is how AMPK blocks activation of CREB to extend lifespan. Paradoxically, our results show that TAF-4, which activates CREB, is required for extended lifespan in Mit mutants. Considering the diversity of functions of the various CREB target genes, we suggest that CREB activation of certain target genes can lead to reduced lifespan, while CREB activation of alternate target genes can lead to increased lifespan. The fact that in the Mair et al. study CREB knockdown resulted in increased lifespan, while complete loss of CREB did not, supports the idea that CREB may be playing a dual role in modulating longevity. The Mit mutants may be mediating their life extension through pathways that lead to activation of particular CREB-target genes that are different from those controlled by CRTC-1. Since TAF-4 is required for this activation, knockdown of TAF-4 prevents Mit mutant life extension (Figure 6D). Apart from CRTC, CREB has another signal-responsive coactivator, CREB Binding Protein (CBP), which has its own binding site on CREB. A recent study suggests that while CRTC can rescue the activation of some CREB target genes in the absence of CBP, there are some CREB target genes that require CBP for activation [45]. CBP and CRTC therefore have some degree of selectivity towards which CREB target genes they activate in response to various signals. What role CBP plays along with TAF-4 in the lifespan modulation of Mit mutants remains an interesting area of further study.

In summary, using a targeted RNAi screening approach, we have identified multiple transcription factors whose functions are important for the life extension of one or more Mit mutants. These include seven transcription factors essential to isp-1(qm150) mutants, six of which are also important in tpk-1(qm162) mutants. Previously, HIF-1 was shown to be required for longevity in Mit mutants, JUN-1 in intermittent feeding, NHR-49 genes in nutrient status, and NHR-27 in signals from the germline. The fact that all these genes were identified validates our screening approach and emphasizes that it is not one gene or pathway responsible for life extension in Mit mutants, but a network of genes that must be specifically modulated. Nowhere is this more evident than in the case of CREB, which can act either to extend life or shorten it. TAF-4 is one of CREB's regulators and was identified by our screen as being required for Mit mutant longevity. This role of TAF-4 in lifespan determination is novel and our future studies will explore its interactions with other proteins as well as the downstream genes it regulates. Our findings support a retrograde response model of Mit mutant longevity, whereby reduced mitochondrial function leads to activation of various transcription factors that alter gene expression in order to promote survival and increase longevity.

Materials and Methods

Supplementary Materials

Supplemental File

Supplemental File

Supplemental File