skn-1-Dependent and -Independent Regulation of aip-1 Expression following Metabolic Stress in Caenorhabditis elegans

Abstract

Maintenance of a stable, properly folded, and catalytically active proteome is a major challenge to organisms in the face of multiple internal and external stresses which damage proteins and lead to protein misfolding. Here we show that internal metabolic stress produced by reactive intermediates resulting from tyrosine degradation triggers the expression of the aip-1 gene, which is critical in responses to the environmental toxin arsenic and the clearance of unstable polyglutamine and Aβ proteins. aip-1 acts via binding to the proteosome and enhancing proteosomal function. We find that full induction of aip-1 depends on the oxidative-stress-responsive skn-1 transcription factor but significant induction still occurs without skn-1. Importantly, activation of skn-1 with wdr-23(RNAi), which dramatically induces the expression of other skn-1 target genes, produces a minimal increase in aip-1 expression. This suggests that the previously demonstrated specificity in aip-1/AIRAP induction could reflect the actions of multiple synergistic activators, such as the heat shock factor homolog hsf-1, which we also find is required for full induction. These may be triggered by proteosome dysfunction, as we find that this event links the multiple inducers of aip-1. Together, our results show that cell stress triggers aip-1 expression by both skn-1-dependent and -independent pathways.

In all eukaryotes, tyrosine is converted into energy via a five-step metabolic pathway (Fig. 1) (44). Impaired tyrosine degradation is observed in hereditary type I tyrosinemia, where mutations affecting fumarylacetoacetate hydrolase (FAH), which catalyzes the final step in the pathway, produce elevations in highly reactive tyrosine metabolites such as fumarylacetoacetate and maleylacetoacetate, which damage proteins and DNA (4, 6, 7, 13, 15, 17, 18, 21, 27-29, 34, 36, 39, 52, 59). We have shown that hereditary type I tyrosinemia can be modeled in Caenorhabditis elegans through treatment of worms with interfering RNA against the FAH homolog fah-1 (18). The resulting impairment in tyrosine degradation has dramatic effects on treated worms, including a shortened life span, destruction of the intestine, and decreased fertility. Further, RNA interference (RNAi) treatment leads to the activation of several cell stress responses, including those for oxidative and endoplasmic reticulum (ER) stress, in worms (18). Impaired tyrosine degradation also leads to accelerated aggregation of an aggregation-prone polyglutamine repeat protein in worms, but the mechanism leading to this increase is unclear (49).

FIG. 1.

Tyrosine is degraded via a five-step metabolic pathway. Shown are the chemical structures of the intermediates and enzymes involved in catalyzing each step. Also indicated are the locations of the tatn-1(baf1) and hpd-1(ok1955) mutations and fah-1(RNAi) used to manipulate levels of the reactive metabolites maleylacetoacetate and fumarylacetoacetate.

The aip-1/AIRAP (arsenic-inducible RNA-associated protein) proteins were initially identified as part of the response to the environmental toxin arsenic (56). Arsenic produces multiple types of cell stress, including ER stress, oxidative stress, and protein misfolding (56). Following exposure to arsenic, the aip-1/AIRAP proteins are induced at the transcriptional level (56). These proteins bind to the 19S regulatory cap of the proteosome and augment its ability to clear damaged proteins by enhancing substrate access to the catalytic core (57). In addition, aip-1 also is constitutively expressed in worms at a low level, and this expression is essential in C. elegans to maintain proteostasis in animals expressing an aggregation-prone polyglutamine repeat protein or Aβ peptide and to ensure normal worm longevity (23, 64). A related mammalian protein, AIRAPL, also shows constitutive expression, is likely prenylated via a C-terminal CAAX motif, and is associated with the ER in the absence of cell stress (64). In contrast, AIRAP is diffusely expressed in the cytoplasm following induction (56, 64). Together, this suggests that AIRAP is primarily regulated at the transcriptional level whereas AIRAPL is regulated at the posttranslational level by stress (64). The aip-1 gene shares properties of both in that it is both constitutively expressed and induced by cell stress, and additionally, aip-1 has a C-terminal CAAX motif which is required for function (64). Important unanswered questions with regard to aip-1/AIRAP include identifying the spectrum of inducers, learning whether stressors other than unstable proteins or xenobiotics can stimulate expression, and determining how inducers are identified by cells and activate aip-1/AIRAP expression.

Known inducers of aip-1/AIRAP expression include arsenic, overexpression of an aggregation prone Aβ peptide in worm muscle, tert-butyl peroxide, juglone, and heat shock (23, 25, 50, 53, 56). Arsenic triggers the expression of aip-1 in worms via the activation of the skn-1 transcription factor (57). skn-1 is a bZIP protein which coordinates responses to oxidative stress in worms (2, 3, 11, 26, 30). Since arsenic produces oxidative stress, it is not unexpected that skn-1 is involved in the control of aip-1, but it is not clear if skn-1 is the sole transcription factor involved in the control of aip-1 expression. It is possible that misfolded proteins could also act through skn-1, because proteosomal dysfunction has been shown to be an activator of skn-1 nuclear translocation and target gene expression (30). Alternatively, aip-1 could be induced independently of skn-1, as suggested in a recent microarray study using tert-butyl peroxide (50). Both hsf-1 and the zinc finger protein-encoding gene slr-2 have been shown to be required for aip-1 expression following heat shock, so hsf-1 or slr-2 could also be involved in the effects of stressors on aip-1 expression (25, 35).

In this work, we explore the in vivo effects of tyrosine metabolites on cell stress responses. The overlap between the effects produced by arsenic exposure and impaired tyrosine degradation with regard to cell stresses led to the hypothesis that tyrosine metabolites represent endogenous activators of aip-1 in C. elegans. We test this hypothesis and find that these metabolites are endogenous triggers for aip-1 expression. Hence, while aip-1 can respond to external stressors, these internal stressors may represent a more common trigger for the pathway. We further find that a pathway involving the oxidative-stress-sensitive transcription factor skn-1 is required for the full expression of aip-1. But in contrast to other skn-1 target genes studied, the activation of aip-1 can still occur in the absence of skn-1, which suggests a degree of redundancy in the regulation mechanism (2, 3, 26, 30). Further, the activation of skn-1 alone is unable to induce aip-1 expression, which is consistent with a necessary role for additional regulators. One of these regulators could be the worm heat shock factor homolog hsf-1, which has been shown to regulate aip-1 expression following heat shock and we find is also required for full expression of aip-1 in response to metabolic stress (25). Finally, while the exact stimulus for aip-1 expression is unknown, we present evidence that impaired proteosome function could be a common inducer of the pathway, as we find that aip-1 is induced by multiple stimuli, including RNAi against heat shock proteins and proteosomal subunits or the chemical juglone, which would be expected to increase cellular levels of damaged proteins. Consistently, we observe enhanced polyglutamine repeat protein aggregation in worms treated with fah-1(RNAi), as well as a dramatic increase in polyubiquitinated proteins. Together, our results demonstrate that both skn-1-dependent and -independent pathways contribute to the control of aip-1 expression following metabolic stress.

MATERIALS AND METHODS

Strains.

SJ4003 (aip-1p:GFP) was a gift of David Ron (NYU, New York, NY) (56). AM140 (rmIs132), CL2166 {dvIs19[pAF15(gst-4::GFP::NLS)]}, DP38 [unc-119(ed3)], FK171 [mek-1(ks54)], KU4 [sek-1(km4)], KU25 [pmk-1(km25)], TJ1060 [spe-9(hc88), fer-15(b26)], and VC1539 [hpd-1(ok1955)] were obtained from the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources. LG335 {skn-1(zu135)/nT1[qIs51]} was a gift of Nicholas Bishop and Leonard Guarente (8). LG335 was crossed with SJ4003 to generate ALF127 {skn-1(zu135)/nT1[qIs51] aip-1p:GFP}. VC1539 was outcrossed three times with N2 to generate ALF114. ALF103 [tatn-1(baf1)] was described previously (18). Strains carrying aip-1p:GFP or Q35:YFP and hpd-1(ok1955), tatn-1(baf1), mek-1(ks54), sek-1(km4), and pmk-1(km25) were created by standard genetic crosses. The presence of hpd-1(ok1955), mek-1(ks54), sek-1(km4), and pmk-1(km25) was verified by single-worm PCR using deletion-specific oligonucleotides (sequences are available upon request). tatn-1(baf1) was verified by single-worm PCR, followed by restriction digestion with PflMI to detect a restriction fragment length polymorphism introduced by the mutation (18). The presence of skn-1(zu135) was verified by scoring for maternal-effect lethality.

Generation of transgenic animals.

The 500-bp and 1-kb aip-1p:GFP reporters were generated via PCR using genomic DNA from the SJ4003 strain as a template (oligonucleotide sequences are available upon request). The resulting PCR products were subcloned into pPD95.75 (Addgene Inc., Cambridge, MA) digested with BamHI and XhoI as a BglII-and-XhoI fragment. The resulting plasmids were then sequenced to confirm the promoter sequence. These plasmids were prepared for bombardment by insertion of the unc-119 marker gene into the ampicillin resistance gene on the vector backbone via homologous recombination using punc-119c (16). The resulting plasmid was used to bombard DP38 [unc-119(ed3)] as previously described (5, 66). Transgenic strains were identified by rescue of the unc-119 mutant phenotype. Strains ALF115 [unc-119(ed3) bafIs115] and ALF116 [unc-119(ed3) bafIs116] are described here.

The 1-kb aip-1p:GFP reporter lacking both skn-1 binding sites was prepared by site-specific mutagenesis using the QuikChange Lightning multisite-directed mutagenesis kit (Stratagene Inc., La Jolla, CA). Both WWTRTCAT skn-1 binding sites were changed to WWTCTGCAG, which has been shown to abolish skn-1 binding and can be recognized by the restriction enzyme PstI (oligonucleotide sequences are available upon request) (2). The resulting plasmid was sequenced to verify its sequence and then used for bombardment as described above. Strains ALF117 [unc-119(ed3) bafEx117], ALF118 [unc-119(ed3) bafEx118], ALF119 [unc-119(ed3) bafEx119], ALF120 [unc-119(ed3) bafEx120], ALF121 [unc-119(ed3) bafEx121], ALF122 [unc-119(ed3) bafEx122], ALF123 [unc-119(ed3) bafEx123], ALF124 [unc-119(ed3) bafEx124], ALF125 [unc-119(ed3) bafEx125], and ALF126 [unc-119(ed3) bafEx126] were obtained and used for further analysis.

RNAi treatment.

The fah-1 RNAi vector was previously described (18). Vectors for aip-1, hsf-1, hsp-1, hsp-3, hsp-6, pas-5, pas-6, pbs-3, pbs-6, pbs-7, rpn-1, rpn-8, rpn-11, and skn-1 were drawn from the Ahringer library and confirmed by sequencing (31). RNAi was delivered by feeding as previously described, with the exception that the concentration of carbenicillin was increased to 50 μg/ml and the isopropyl β-d-thiogalactopyranoside (IPTG) concentration was decreased to 1 mM (18). RNAi mixing experiments were performed by growing separate overnight cultures of each clone and mixing equal volumes immediately before spotting onto plates. We have previously shown that diluting fah-1(RNAi) has no effect on the resulting phenotype (18).

Q35 aggregate assay.

Eggs were extracted using hypochlorite treatment, placed on control or fah-1(RNAi) plates, and grown at 23°C for 3 days. Aggregates were scored as described previously (45). We found that placing the worm plates on ice for 5 min facilitated the scoring of worms by decreasing movement. Digital photographs were captured using an Olympus BX51 upright microscope and DP70 camera as previously described (18). All of the images and counts shown in Fig. 11 were obtained in parallel on the same day to facilitate comparison.

FIG. 11.

Tyrosine metabolites produce proteosome dysfunction. (A) Images showing effects of control and fah-1(RNAi) treatments of AM140 (Q35:YFP) transgenic worms or tatn-1 or hpd-1 mutants carrying the Q35:YFP transgene. The arrows indicate aggregates in a fah-1(RNAi)-treated worm. (B) Graphical representation of aggregate counts from worms as shown in panel A [n = 20 for all groups; P < 0.0001 for wild-type (WT) fah-1 versus WT control interfering RNA, P = 0.595 for hpd-1(ok1955) fah-1 versus hpd-1(ok1955) control interfering RNA, P = 0.868 for tatn-1(baf1) fah-1 versus tatn-1(baf1) control interfering RNA]. (C) Graph showing aggregate counts from Q35:YFP wild-type or tatn-1 mutant transgenic worms treated with the tyrosine metabolite SA [n = 20 for all groups; P < 0.0001 for WT SA versus control treatment and for tatn-1(baf1) SA treatment versus control treatment]. (D) Total protein from fah-1- or control interfering RNA-treated TJ1060 worms was separated by SDS-PAGE and transferred to nitrocellulose. Blots were probed with an antiubiquitin antibody (left panel) or an antiactin antibody to demonstrate equal loading (right panel). (E) Total protein from N2, ALF103 [tatn-1(baf1)], and ALF114 [hpd-19(ok1955)] worms was separated by SDS-PAGE and transferred to nitrocellulose. Blots were probed with an antiubiquitin antibody (top) or an antiactin antibody to demonstrate equal loading (bottom).

aip-1:GFP expression.

Eggs were isolated from SJ4003 worms or other mutants carrying the aip-1p:GFP transgene via hypochlorite treatment and placed on plates spotted with interfering RNA. Effects on aip-1:GFP expression were assessed by digital photography using the same microscope and camera as described above in day 1 adult worms (18). Treatment of adult worms with RNAi was also performed by placing day 1 adult worms on RNAi and assessing the effects 48 h later. ImageJ (NIH) was used to quantify the relative fluorescence of images, and the image data were subsequently analyzed in Prism5 (GraphPad Software, San Diego, CA) (1).

gst-4:GFP expression.

Eggs were isolated from CL2166 worms via hypochlorite treatment and placed on plates spotted with interfering RNA. Effects on gst-4:GFP expression were assessed by digital photography using the same microscope and camera used as described above for day 1 adult worms (18). All of the images in a panel were captured on the same day using identical camera settings to facilitate comparison. ImageJ (NIH) was used to quantify the relative fluorescence of images, and the image data were subsequently analyzed in Prism5 (GraphPad Software, San Diego, CA) (1).

Q-PCR quantification of GFP and aip-1 transcript levels.

Eggs were extracted from SJ4003 worms by hypochlorite treatment and grown on plates spotted with either fah-1(RNAi) or control interfering RNA for 3 days. Worms were collected by floatation, rinsed in water, and snap-frozen. Worm lysis and RNA extraction were done with the miRNeasy Mini kit (Qiagen Inc., Valencia, CA). RNA concentration was measured using a Nanodrop spectrophotometer, and 500 ng of RNA was reverse transcribed using Moloney murine leukemia virus reverse transcriptase (Promega, Madison, WI). Quantitative PCR (Q-PCR) was preformed using the PowerSYBR green PCR Master Mix (Applied Biosystems, Foster City, CA) with a Stratagene MX3000P thermocycler (Stratagene Products, La Jolla, CA). Amplification was performed using oligonucleotides designed with the PerlPrimer program for green fluorescent protein (GFP) and aip-1 (42). These oligonucleotides cross intron-exon borders and amplify 100- to 200-bp regions (oligonucleotide sequences are available upon request). The effectiveness of the oligonucleotides was validated via obtaining a single correct-size PCR product on an agarose gel, via melting analysis on the Q-PCR machine, and via Q-PCR on serially diluted cDNA as described previously (41). Primers for ama-1 have been previously described (55). The relative expression of GFP and aip-1 normalized to ama-1 expression was determined via the 2^−ΔΔCt method (41).

SA treatment.

Succinylacetone (SA) was added to molten nematode growth agar (NGA) medium, and the plates were used within 1 week of pouring and spotting with OP50 because SA appears to be unstable in aqueous solution. To assess the effects of SA on Q35:YFP aggregation, AM140 or tatn-1(baf1) rmIs132 eggs were extracted using hypochlorite treatment, arrested overnight in S-basal (0.1 M Nacl, 0.05 M K₃PO₄, pH 6.0), and placed on NGA medium plates containing 1 mg/ml SA (Sigma-Aldrich, St. Louis, MO). Yellow fluorescent protein (YFP) aggregates were counted 2 days later following incubation at 23°C. To assess the effects of SA on aip-1p:GFP expression, SJ4003 eggs were isolated using hypochlorite treatment, arrested overnight in S-basal, and placed on NGA medium plates containing 0.5 or 1.0 mg/ml SA. Effects on GFP expression were assessed by digital photography 3 days later following incubation at 20°C. Worms treated with SA fail to reach adulthood, as previously shown (18).

Xenobiotic treatment.

For juglone, day 1 adult worms were treated with 38 μM juglone (Sigma-Aldrich, St. Louis, MO) in M9 (6 g Na₂HPO₄, 3 g KH₂PO₄, 5 g NaCl, 0.25 g MgSO₄ · 7H₂O per liter) for 1 h before being washed in M9 and being transferred to plates at 20°C (11). Juglone powder was dissolved in 100% ethanol at a 100× concentration fresh before each experiment. A control consisting of worms treated with a similar volume of 100% ethanol was performed in parallel and used for reference. Effects on GFP expression were measured 24 h later.

For tunicamycin, day 1 adult worms were treated with 5 μg/ml tunicamycin on spotted NGA medium plates for 5 h before being washed in M9 and being transferred to plates at 20°C (9). Tunicamycin was dissolved in dimethyl sulfoxide (DMSO). A control consisting of worms treated with a similar volume of DMSO was performed in parallel and used for reference. Effects on GFP expression were measured 24 h later.

For NaCl, day 1 adult worms were transferred to NGA medium plates containing 51, 200, or 400 mM NaCl and incubated at 20°C as previously described (12). Effects on GFP expression were assessed 24 h later.

Western blotting.

TJ1060 eggs were put on control or fah-1(RNAi) plates, grown at 20°C for 2.5 days, and then shifted to 25°C to sterilize the animals. Alternatively, N2, ALF103, or ALF114 eggs were put on control or fah-1(RNAi) plates and grown at 20°C for 3 days. Protein was extracted from worms by boiling in LDS loading buffer (Invitrogen, Carlsbad, CA) supplemented with 10 μM N-ethylmaleimide. A 4.5-μg sample of protein, as determined by Bradford assay (Pierce, Rockford, IL), was subjected to 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and blotted onto nitrocellulose. Blots were probed with antiubiquitin antibody (P4D1; Santa Cruz Biotechnologies, Santa Cruz, CA) at 1:100 or antiactin antibody (Cell Signaling, Danvers, MA) at 1:1,000, followed by IRDye 680 goat anti-rabbit antibody for actin and IRDye 800 donkey anti-mouse antibody for ubiquitin as recommended by the manufacturer (Li-Cor Biosciences, Lincoln, NE). Blots were imaged on a Li-Cor Odyssey system.

RESULTS

Impaired tyrosine metabolism activates AIRAP/aip-1 expression.

The amino acid tyrosine is degraded to energy via a five-step enzymatic pathway in worms and other eukaryotes (Fig. 1). We have recently identified and characterized the genes in the worm metabolic pathway (18). Treatment of worms with fah-1(RNAi), which affects the last enzyme in the pathway, produces a phenotype consisting of small body size, reduced fertility, destruction of the intestine, and reduced survival (18). This phenotype depends on the buildup of the reactive intermediates fumarylacetoacetate and maleylacetoacetate, as RNAi or genetic mutations affecting the upstream enzymes tatn-1, hpd-1, and hgo-1 suppress the effects of fah-1(RNAi) (18). The tatn-1(baf1) mutant was isolated in a genetic screen due to resistance to the effects of fah-1(RNAi) on worm morphology, as shown by the larger body size and preserved intestinal morphology (arrows, Fig. 2E) (18). This allele has a P224S mutation which affects a highly conserved proline found in the tyrosine aminotransferase genes from worms, flies, mice, and humans (18). The hpd-1(ok1955) mutant has a 41-bp insertion followed by a 1,363-bp deletion which removes part of the first and third exons along with the entire second exon (WormBase; L. Nash, A. A. Ferguson, and A. L. Fisher, unpublished data). This mutant is also resistant to the effects fah-1(RNAi) (Fig. 2E).

FIG. 2.

Tyrosine metabolites induce aip-1 expression. (A) Digital images showing the effects of control or fah-1(RNAi) treatment on SJ4003 (aip-1p:GFP) worms or tatn-1(baf1) and hpd-1(ok1955) mutants carrying the aip-1p:GFP transgene. Worms were treated with fah-1(RNAi) from egg hatching, and day 1 adult worms were photographed to examine GFP expression. The intestine (arrowhead) and excretory cell (arrow) are indicated. (B) Graph of fluorescence measured from digital images [n = 15 for all; P < 0.0001 for the wild-type (WT) control versus fah-1(RNAi), P < 0.0001 for the tatn-1(baf1) control versus fah-1(RNAi), P = 0.44 for the hpd-1(ok1955) control versus fah-1(RNAi), P = 0.0006 for WT fah-1(RNAi) versus tatn-1(baf1) fah-1(RNAi), and P < 0.0001 for hpd-1(ok1955) fah-1(RNAi)]. (C) Graph showing relative expression of GFP and the endogenous aip-1 gene in SJ4003 worms treated with fah-1(RNAi) [n = 5 biological replicates; P = 0.0024 for the aip-1 control versus fah-1(RNAi) and P = 0.0004 for GFP]. (D) Graph of fluorescence measured from digital images in panel F showing the effects of treatment of SJ4003 with SA from egg hatching (control versus SA, P < 0.0001 for both by t test; n = 15 for all). (E) Digital images showing morphological effects of control or fah-1(RNAi) treatment of N2, tatn-1(baf1), and hpd-1(ok1955). (Arrow = intestine) (F) Images showing the effect of 0.5 or 1.0 mg/ml SA on 3-day-old worms. SA treatment impairs the development of worms into reproductive adults.

We have shown that tyrosine metabolites produce both oxidative and ER stress in worms (18). The AIRAP/aip-1 proteins are induced by environmental stressors, such as arsenic, that produce both oxidative and ER stress (56, 57). Hence, we hypothesized that the tyrosine metabolites may activate aip-1 expression. We examined whether aip-1 is activated following exposure to tyrosine metabolites produced by fah-1(RNAi) and found that fah-1(RNAi) was a robust activator of an aip-1p:GFP reporter gene in transgenic worms (Fig. 2A and B). To confirm that the increase in GFP expression reflected an increase in the expression of the endogenous aip-1 gene, we used Q-PCR to quantify both the GFP transcript and the aip-1 transcript in control or fah-1(RNAi)-treated animals (Fig. 2C). We found that fah-1(RNAi) significantly increased the expression of both, with aip-1 showing 3.8-fold induction and GFP showing 11.1-fold induction. Hence, changes in reporter expression reflect changes in the expression of the endogenous aip-1 gene. Notably, in treated worms, the expression of GFP is seen both in the intestine (Fig. 2A, arrowhead), which expresses fah-1 and presumably has direct exposure to these reactive metabolites, and also in the excretory cell, which does not express fah-1 (Fig. 2A, arrow) (18).

To verify that tyrosine metabolites were directly responsible for the induction of aip-1p:GFP, we created transgenic worms carrying the transgene and the tatn-1(baf1) and hpd-1(ok1955) mutations (Fig. 1). When the worms were treated with fah-1(RNAi), both mutations suppressed the morphological effects of RNAi treatment due to the decreased accumulation of fumarylacetoacetate and maleylacetoacetate (Fig. 1 and 2E). Consistent with the reduced accumulation of metabolites, both the hpd-1(ok1955) and tatn-1(baf1) mutants showed reduced GFP expression following RNAi treatment (Fig. 2A and B). This finding is consistent with a direct effect of tyrosine metabolites on the expression of aip-1.

Notably, the hpd-1(ok1955) mutation completely suppressed the increase in GFP expression following fah-1(RNAi) treatment, while the tatn-1(baf1) mutant showed an intermediate phenotype with retained induction in the excretory cell (Fig. 2A and B). This may reflect the differing strengths of the hpd-1(ok1955) mutation, which is a 1,363-nucleotide deletion, versus tatn-1(baf1), which is a point mutation, in terms of the flux through the tyrosine degradation pathway (18). We suspect that tatn-1(baf1) is a hypomorphic allele, as the tatn-1(ok3323) allele, which is a 602-nucleotide deletion in exon 2, is lethal (WormBase; A. L. Fisher, unpublished data). However, the similarity between hpd-1(ok1955) and tatn-1(baf1) with regard to resistance to the adverse effects of fah-1(RNAi) on morphology (Fig. 2E) or Q35:YFP aggregation (see below) is consistent with a protective effect of aip-1 induction in the tatn-1(baf1) mutants.

We also treated wild-type worms carrying the aip-1p:GFP transgene with SA. SA is produced in vivo from the tyrosine metabolites maleylacetoacetate and fumarylacetoacetate via nonenzymatic isomerization and decarboxylation, and SA is the only commercially available metabolite (39, 43). We have previously shown that treatment of worms with SA mimics the phenotype produced by fah-1(RNAi) (18). We found that the treatment of worms with 0.5 or 1.0 mg/ml SA induced the expression of aip-1 (Fig. 2D and F). These findings demonstrate that tyrosine degradation products are the first identified endogenous activator of the aip-1/AIRAP pathway.

The skn-1 transcription factor regulates aip-1 expression.

The oxidative-stress-sensitive transcription factor skn-1 has been previously shown to be involved in the transcriptional activation of aip-1 in arsenic-treated worms (57). Similarly to arsenic, we have previously shown that tyrosine metabolites produce oxidative stress in fah-1(RNAi)-treated worms (18). Consequently, we tested whether the previously demonstrated activation of the oxidative-stress-responsive gst-4 gene by fah-1(RNAi) depends on skn-1 and whether the activation of aip-1 also requires skn-1.

The gst-4 gene was identified as an oxidative-stress-responsive gene, and a gst-4p:GFP reporter gene shows skn-1-dependent activation to multiple sources of oxidative stress (11, 30, 40, 60). Treatment of gst-4p:GFP transgenic worms with a 1:1 mixture of control interfering RNA and fah-1(RNAi) produced a robust activation of the reporter, which is consistent with our prior finding that a similar dilution of fah-1(RNAi) does not reduce the phenotype produced (Fig. 3A and B) (18). In contrast, treatment with a mixture of skn-1(RNAi) and fah-1(RNAi) resulted in no induction of the reporter (Fig. 3A and B). This finding is consistent with the activation of skn-1 following the accumulation of tyrosine metabolites and a subsequent absolute requirement for skn-1 to induce gst-4 expression.

FIG. 3.

skn-1 is activated by tyrosine metabolites and is necessary for full activation of aip-1 expression. (A) Digital images showing the effects of treatment of CL2166 (gst-4p:GFP) worms with control interfering RNA, a 1:1 mixture of fah-1 and control interfering RNA, or a 1:1 mixture of fah-1 and skn-1(RNAi). (B) Graph of fluorescence measured from digital images [n > 20 for all; P < 0.0001 for fah-1 and control interfering RNA versus fah-1 and skn-1(RNAi)]. (C) Digital images showing the effects of treating SJ4003 or ALF127 {skn-1(zu135)/nT1[qIs51] aip-1p:GFP} transgenic worms with fah-1 or control interfering RNA. (D) Images showing the effects of treatment of SJ4003 worms with control interfering RNA, a 1:1 mixture of fah-1 and control interfering RNA, or a 1:1 mixture of fah-1 and skn-1(RNAi). (E) Graph of fluorescence measured from digital images in panel D [n = 15 for all; P < 0.0001 for control interfering RNA versus fah-1 and control interfering RNA or fah-1 and skn-1(RNAi)]. (F) Graph of fluorescence measured from the digital images in panel C [n = 15 for all; P < 0.0001 for the wild-type (WT) control versus fah-1(RNAi) and P = 0.0003 for the skn-1 control versus fah-1(RNAi)].

However, similar RNAi treatment of aip-1p:GFP transgenic worms produced distinct results, as treatment of worms with a 1:1 mixture of fah-1(RNAi) and control interfering RNA produced a strong induction of GFP expression (Fig. 3D and E). But treatment with a mixture of skn-1(RNAi) and fah-1(RNAi) still resulted in an increase in GFP expression (Fig. 3D and E). This increase was roughly half of the response to the fah-1(RNAi) and control interfering RNA mixture but much greater than the increase observed in treated gst-4p:GFP worms (Fig. 3A and B). The response of aip-1 to treatment with arsenite has also been recently shown to be at least partially skn-1 dependent (57).

To ensure that our findings were not due to an incomplete response to skn-1(RNAi), we constructed the ALF127 {skn-1(zu135)/nT1[qIs51] aip-1p:GFP} strain to genetically remove skn-1. The ALF127 progeny lacking skn-1 are identified by the lack of the myo-2:GFP reporter carried on the nT1 balancer chromosome. Consistent with our findings obtained using skn-1(RNAi), treatment of this strain with fah-1(RNAi) still produced an increase in GFP expression in the skn-1 progeny (Fig. 3C and F). As before, this induction is significantly lower than that observed in the nonmutant SJ4003 (aip-1p:GFP) strain treated in parallel (Fig. 3C and F). We also observed increased aip-1 expression in the skn-1(+) progeny of treated ALF127 worms (not shown). These findings suggest that aip-1 expression may be under skn-1-dependent but also skn-1-independent control.

We also found that the skn-1(−) progeny from ALF127 appeared to be more strongly affected by fah-1(RNAi) than either wild-type worms or the skn-1(+) progeny from ALF127. To explore this observation, we examined ×20 images of treated worms and found that the skn-1(−) worms exhibited more damage to the intestine than wild-type fah-1(RNAi)-treated worms (Fig. 4A). Associated with the increase in tissue damage, we also observed an increase in larval arrest in the fah-1(RNAi)-treated skn-1(−) worms (Fig. 4B). Larval arrest is an uncommon phenotype in wild-type fah-1(RNAi)-treated worms, with 85% of worms reaching adulthood after 3 days while only 35% of skn-1(−) fah-1(RNAi)-treated worms reached adulthood. This suggested that skn-1 mutant worms were more sensitive to the effects of tyrosine metabolites. To test this hypothesis, we treated skn-1 mutants, LG335 {skn-1(zu135)/nT1[qIs51]}, with 0.2 mg/ml SA, as we have previously shown that low doses of SA produce a mixture of adult worms and arrested larvae (18). We found that 33% of the skn-1(+) progeny reached adulthood while only 5% of the skn-1(−) progeny reached adulthood (Fig. 4C). To test whether this protective effect of skn-1 involved aip-1, we treated the tatn-1(baf1) mutant with either a 1:1 mixture of fah-1 and control interfering RNA, aip-1 and control interfering RNA, or fah-1 and aip-1(RNAi) to see whether we could block the protective effect of aip-1 suggested by the induction observed above. We found that ∼10% of the fah-1- and aip-1(RNAi)-treated worms lost the protective effect of the tatn-1(baf1) mutation and resembled wild-type worms treated with fah-1(RNAi) (Fig. 4D). Our findings suggest that skn-1 regulates aip-1 as part of a protective response to reactive tyrosine metabolites. However, aip-1 could be either a direct or an indirect target gene of skn-1.

FIG. 4.

skn-1 protects worms against tyrosine metabolites. (A) Digital images of wild-type (WT) and skn-1 mutant worms treated with fah-1(RNAi) showing increased damage to the intestine in skn-1 fah-1(RNAi)-treated worms (arrow = intestine). (B) Graph showing the percentages of adult animals among wild-type, skn-1 heterozygous, and skn-1 homozygous worms treated with control or fah-1(RNAi) (n = 20 for all). (C) Graph showing the percentages of adult animals among skn-1 heterozygous and skn-1 homozygous worms treated with 0.2 mg/ml SA (three replicates of 25 worms each; P = 0.01 by t test). (D) Digital images showing the effects of treatment of ALF103 [tatn-1(baf1)] worms with a 1:1 mixture of fah-1 and control interfering RNA, aip-1 and control interfering RNA, or fah-1 and aip-1(RNAi). The image for fah-1 and aip-1(RNAi) represents ∼10% of the animals and <1% in the other treatments.

aip-1 is a potential skn-1 target gene, as the aip-1 promoter has two WWTRTCAT consensus binding sites for skn-1 (ATTGTCAT at −70 and AATGTCAT at −250 relative to the TATA box) within 500 bp of the transcriptional start site (2, 53). To test whether these sites are functional, we generated a series of aip-1 promoter deletions. We found that 500 nucleotides of promoter sequence was insufficient to respond to fah-1(RNAi), but a 1-kb promoter drove GFP expression in the intestine and to a lesser extent in the excretory cell after exposure to fah-1(RNAi) (Fig. 5A and data not shown). We tested four transgenic lines which demonstrated very similar results (Fig. 5A and B and not shown). Mutation of both skn-1 binding sites, by changing them to WWTCTGCAG, which prevents skn-1 binding in other target genes, in the 1-kb promoter resulted in a reduction in GFP expression following exposure to fah-1(RNAi) (Fig. 5A and B) (2). Mutation of the binding sites also had no detectable effect on the basal GFP expression which was low in both the mutants and the control (data not shown). Of nine transgenic lines tested, none was comparable to the lines generated with the nonmutated promoter (Fig. 5A and B). However, several of the transgenic lines harboring the reporters with mutated binding sites still demonstrated variable GFP induction following RNAi treatment (Fig. 5B). This observation is consistent with aip-1 being a direct target of skn-1 and also being the target of a skn-1-independent pathway(s).

FIG. 5.

aip-1 is regulated by two skn-1 binding sites in the promoter. (A) Digital images of transgenic worms with 1 kb of the aip-1 promoter sequence (bafIs115) treated with a control or fah-1(RNAi) and images of transgenic worms with the 1-kb aip-1 promoter transgene in which both skn-1 binding sites have been mutated (bafEx117 and bafEx121) treated with fah-1(RNAi). (B) Graph of fluorescence measured from digital images from two nonmutated control lines and nine individual transgenic worm lines carrying the mutated promoter (n = 15 for all; P < 0.0001 for mutants versus bafIs115, except P = 0.0007 for bafEx117 and P = 0.0001 for bafEx124; P < 0.0001 for mutants versus bafIs116, except P = 0.0001 for bafEx117).

In worms, the skn-1-dependent responses to oxidative stress involve the actions of several kinases which directly or indirectly regulate the nuclear localization and transcriptional activity of skn-1 (3, 26, 33). Further, tyrosine metabolites have been shown to activate ERK kinases in mammalian cells (28). Given these observations and the involvement of skn-1 in aip-1 regulation, we asked whether the upstream stress-responsive kinase-encoding gene mek-1, sek-1, or pmk-1 is required for the expression of aip-1. We crossed the aip-1p:GFP reporter into worms lacking each of these kinase genes and then used fah-1(RNAi) to investigate the responses to this stressor. We found that loss of sek-1 marginally reduced the expression of aip-1p:GFP, whereas loss of pmk-1 showed a statistically significant but still rather modest reduction in the induction of aip-1 following treatment (Fig. 6). This is consistent with either a mode of skn-1 activation independent of these kinases or compensation by other activators of aip-1 expression.

FIG. 6.

Stress-responsive kinases play a minor role in the activation of aip-1 expression. (A) Digital images of wild-type (WT) aip-1:GFP worms or aip-1:GFP worms with mutations in the mek-1-, sek-1-, or pmk-1-encoded kinase treated with fah-1(RNAi). (B) Graph of fluorescence measured from digital images [n > 15 for all; P > 0.0001 for wild-type fah-1(RNAi) versus mek-1 fah-1(RNAi), P = 0.1888 for wild-type fah-1(RNAi) versus sek-1 fah-1(RNAi), and P = 0.0063 for wild-type fah-1(RNAi) versus pmk-1 fah-1(RNAi)].

To further explore the role of skn-1 in the regulation of aip-1, we directly activated skn-1 via the treatment of worms with wdr-23(RNAi). The wdr-23 gene encodes a WD-40 repeat protein which has been recently shown to bind skn-1 and regulate its degradation (11). Both mutation of wdr-23 and wdr-23(RNAi) have been shown to lead to skn-1 nuclear localization and produce strong induction of skn-1 target genes (11). Consistent with these observations, we found that treatment of gst-4p:GFP transgenic worms with wdr-23(RNAi) led to a dramatic increase in GFP expression relative to control interfering RNA (Fig. 7C and D). However, treatment of aip-1p:GFP transgenic worms produced a minimal increase in GFP expression (Fig. 7A and B). Our results suggest that while aip-1 is a direct target gene of skn-1, aip-1 is likely not solely regulated by skn-1, and skn-1 likely requires other transcription factors to induce aip-1 expression.

FIG. 7.

skn-1 is not sufficient for aip-1 expression. (A) Digital images showing the effects of control, fah-1, and wdr-23(RNAi) treatments of SJ4003 (aip-1p:GFP) transgenic worms. (B) Graph of fluorescence measured from digital images (n > 15 for all; P < 0.0001 for wdr-23 versus control interfering RNA, P = 0.0007 for fah-1 versus control interfering RNA). (C) Images showing the effects of control and wdr-23(RNAi) treatments on CL2166 (gst-4p:GFP) transgenic worms. (D) Graph of fluorescence measured from digital images (n = 10 for all; P < 0.0001 for wdr-23 versus control interfering RNA).

aip-1 is also regulated by hsf-1 in response to tyrosine metabolites.

In worms, aip-1 has been shown to be upregulated in response to heat shock, and the induction of aip-1 is dependent on hsf-1, which is the worm homolog of heat shock factor (25). To test whether hsf-1 could be an additional regulator of aip-1 expression in response to fah-1(RNAi), we treated worms with control interfering RNA, control interfering RNA mixed 1:1 with fah-1(RNAi), 1:1 fah-1(RNAi) and hsf-1(RNAi), or 1:1 hsf-1(RNAi) and control interfering RNA. We found that treatment with fah-1 and hsf-1(RNAi) together reduced the activation of aip-1 relative to fah-1 and control interfering RNA (Fig. 8A and B). This finding suggests that the full expression of aip-1 depends on both hsf-1 and skn-1.

FIG. 8.

hsf-1 is required for aip-1 expression produced by tyrosine metabolites. (A) Digital images of SJ4003 (aip-1p:GFP) worms treated with the control, a 1:1 mixture of the control and fah-1(RNAi), a 1:1 mixture of fah-1 and hsf-1(RNAi), and a 1:1 mixture of the control and hsf-1(RNAi). (B) Graph of fluorescence measured from digital images [n > 15 for all; P < 0.0001 for the control versus the control and fah-1(RNAi) and P < 0.0001 for the control and fah-1(RNAi) versus hsf-1 and fah-1(RNAi)].

Misfolded proteins and proteosome dysfunction contribute to aip-1 expression.

Our results suggested that oxidative stress contributes to aip-1 regulation but that an additional effect of tyrosine metabolites is important in triggering expression. To explore the types of cell damage that induce aip-1 expression, we tried several pharmacologic and RNAi treatments.

Recent microarray data suggested that the xenobiotic juglone activates the expression of aip-1, so we tested the effect of this compound on aip-1p:GFP transgenic worms (53). We found that treatment of worms with 38 μM juglone significantly increased GFP expression 24 h following treatment (Fig. 9A and D). Juglone is similar to tyrosine metabolites because juglone produces oxidative stress, ER stress, and direct covalent damage to proteins (61). To explore whether ER stress is sufficient to induce aip-1 expression, we treated aip-1p:GFP transgenic worms with tunicamycin. Treatment of worms with 5 μg/ml tunicamycin produces a minimal increase in GFP expression relative to the control, which is consistent with prior results obtained with cultured cells (56). This dose of tunicamycin robustly activates the ER stress response, as indicated by the hsp-4p:GFP reporter gene (data not shown). Finally, we placed worms under hypertonic conditions via exposure to 200 or 400 mM NaCl. Hypertonic stress has been shown to enhance protein aggregation and lead to an increase in cellular glycerol content (12, 37, 38). This stress fails to induce GFP expression above the control level (Fig. 9B and D). Together, our results suggested that xenobiotics that produce multiple types of cell stress serve to induce aip-1 expression, while ER stress or protein aggregation alone may not be sufficient.

FIG. 9.

Multiple proteotoxic stimuli activate aip-1 expression. (A) Digital photos showing the effects of treatment of aip-1p:GFP transgenic worms with the xenobiotics juglone and tunicamycin or the control, fah-1, hsp-1, hsp-3, or hsp-6 RNAi. (B) Digital photos showing the effect of transferring day 1 adult aip-1p:GFP transgenic worms to NGA medium plates containing 51, 200, or 400 mM NaCl for 24 h. (C) Digital images of adult worms treated with the control, hsp-1, or hsp-3(RNAi) for 48 h. (D) Graph of fluorescence measured from the digital images in panels A and B (n > 10 for all). To facilitate comparison across separate trials, the values were normalized with the mean value of the untreated control images in each trial, and this normalized value is plotted [in the individual trials, P = 0.0002 for juglone versus the control, P = 0.444 for 400 mM NaCl versus 51 mM NaCl, P = 0.255 for 200 mM NaCl versus 51 mM NaCl, P = 0.001 for tunicamycin versus the control, P < 0.0001 for hsp-1 and hsp-3(RNAi) versus control interfering RNA, and P = 0.0002 for hsp-6(RNAi) versus control interfering RNA]. (E) Graph of fluorescence measured from the digital images in panel C [n = 15 for all; P < 0.0001 for hsp-1 and hsp-3(RNAi) versus the control].

We hypothesized that the accumulation of damaged and misfolded proteins could serve as a trigger for aip-1 expression, as the oxidative stress and direct protein damage produced by juglone, arsenic, or tyrosine metabolites could then synergize with the accumulation of ER proteins targeted for ER-associated degradation. To test this hypothesis, we treated worms with RNAi against hsp-1, hsp-3, and hsp-6, which are involved in the maintenance of proteostasis in the cytoplasm, ER, or mitochondria, respectively (32, 46, 62). We found that hsp-1(RNAi) and hsp-3(RNAi) both induce aip-1p:GFP, while hsp-6(RNAi) produced a lesser induction (Fig. 9A and D). As hsp-1(RNAi) produced larval arrest when started at egg hatching, we also repeated the experiments with hsp-1 and hsp-3(RNAi) in adult worms to exclude a developmental effect on aip-1 expression. We found that both RNAi treatments induced GFP expression in adults after 48 h of treatment (Fig. 9C and E), which shows that aip-1 expression is independent of development.

Worms treated with hsp-1 or hsp-3(RNAi) would be expected to accumulate ubiquitinated proteins targeted for degradation via the proteosome. Hence, we hypothesized that inhibiting proteosomal function might mimic the effects of hsp-1 or hsp-3(RNAi) or tyrosine metabolites and lead to aip-1 expression. To test this hypothesis, we treated worms with RNAi against rpn-11 (19S regulatory particle subunit) or pbs-3 (20S β subunit) and assessed the effect on aip-1 expression. We found that this led to strong induction of the aip-1p:GFP reporter gene in treated worms (Fig. 10A and B). Our result suggested that proteosomal dysfunction could serve to trigger aip-1 expression. To further test this hypothesis, we chose additional proteosomal genes and inhibited them via RNAi. We found that all of the genes showed an increase in GFP expression relative to the control (Fig. 10A and B). Recently, these same clones were shown to have differing effects on skn-1 nuclear localization and gst-4 activation, but we found little connection between these reported effects and the effects on aip-1 expression (30). This could be due to the involvement of hsf-1 in aip-1 regulation. To exclude developmental effects of proteosomal subunit RNAi on aip-1 expression, we repeated the experiment with day 1 adults and found that RNAi treatment still increased GFP expression (Fig. 10C and D).

FIG. 10.

Proteosomal dysfunction induces aip-1 expression. (A) Digital images of aip-1p:GFP worms treated with control interfering RNA or interfering RNA against proteosomal subunits starting at egg hatching. Many of these RNAi treatments also result in developmental arrest during larval life. (B) Graph of fluorescence measured from digital images (n = 15 for all; P < 0.0001 for all versus control interfering RNA). (C) Digital images of aip-1p:GFP worms treated with control interfering RNA or interfering RNA against proteosomal subunits starting at adulthood. (D) Graph of fluorescence measured from digital images [n > 14 for all; P < 0.0001 for pbs-3, pas-6, pbs-6, and pbs-7(RNAi) versus control interfering RNA, P = 0.0029 for rpn-1(RNAi) versus the control, P = 0.0007 for rpn-8(RNAi) versus the control, and P = 0.0001 for pas-5 and rpn-11(RNAi) versus the control].

Impaired tyrosine degradation produces proteosomal dysfunction.

We then asked whether worms treated with fah-1(RNAi) show evidence of proteosomal dysfunction. We hypothesized that these metabolites could lead to an increase in misfolded proteins and lead to relative proteosomal dysfunction. We tested this possibility in two ways.

First, fah-1(RNAi) has been shown to increase the aggregation of a synthetic polyglutamine repeat protein in C. elegans (49). This suggests that either fah-1 has a novel role in protein folding or the accumulation of tyrosine metabolites by fah-1(RNAi) leads to the accumulation of aggregates. We tested these possibilities by using mutations in tatn-1(baf1) and hpd-1(ok1955), which block enzymes upstream of fah-1 in the tyrosine degradation pathway and prevent the development of metabolic disease (18). Following treatment with fah-1(RNAi), wild-type worms expressing the Q35:YFP transgene in muscle show a marked increase in the number of YFP aggregates compared to control interfering-RNA-treated worms (Fig. 11A and B). However, worms with mutations in either tatn-1(baf1) or hpd-1(ok1955) are protected from the increase in polyglutamine aggregates following RNAi treatment (Fig. 11A and B). This suggests that the accumulation of tyrosine metabolites is responsible for the observed increase in protein aggregation, perhaps due to impaired proteosomal function.

To directly test if tyrosine metabolites produce aggregation of Q35:YFP, we treated worms expressing Q35:YFP with SA. We found that SA treatment of wild-type Q35:YFP transgenic worms and tatn-1(baf1) Q35:YFP transgenic worms led to an increase in aggregates in both (Fig. 11C). This demonstrates that tyrosine metabolites directly lead to an increase in polyglutamine protein aggregation and that mutations in upstream enzymes like tatn-1 cannot block this effect.

Second, we examined whether worms treated with fah-1(RNAi) show an increase in polyubiquitinated proteins awaiting degradation by the proteosome (54). Adult worms treated with fah-1(RNAi) from egg hatching show a dramatic increase in polyubiquitinated proteins compared to control interfering-RNA-treated animals, as shown by Western blotting (Fig. 11D). This increase depends on an intact tyrosine degradation pathway, as the tatn-1 and hpd-1 mutations block this accumulation (Fig. 11E). The aggregation of Q35:YFP and the increase in polyubiquitinated proteins observed in fah-1(RNAi)-treated worms are consistent with relative proteosome dysfunction.

While fah-1(RNAi) produces proteosomal dysfunction, the regulation of aip-1 by fah-1(RNAi) and hsp-1(RNAi) could occur by different mechanisms. To test these possibilities, we asked whether the proteosomal dysfunction produced by fah-1(RNAi) has a mechanism in common with hsp-1(RNAi). First, we combined fah-1(RNAi) and hsp-1(RNAi) to see whether the combination would produce a further increase in aip-1 expression. We found that the 1:1 combination of fah-1(RNAi) and hsp-1(RNAi) was not different than 1:1 hsp-1(RNAi) and control interfering RNA with regard to aip-1p:GFP expression (Fig. 12A and C). Second, we tested whether the activation of aip-1p:GFP requires skn-1 by treating ALF127 with hsp-1 interfering RNA. Similarly to fah-1(RNAi), we observed a reduction but not elimination of aip-1 expression in the skn-1(−) progeny (Fig. 12B and D). Hence, the proteosomal dysfunction produced by hsp-1(RNAi) and that produced by fah-1(RNAi) likely share a response pathway.

FIG. 12.

Proteosomal dysfunction produced by fah-1(RNAi) and that produced by hsp-1(RNAi) trigger similar pathways. (A) Digital images of aip-1p:GFP worms treated with control interfering RNA, a 1:1 mixture of the control and fah-1(RNAi), a 1:1 mixture of fah-1 and hsp-1(RNAi), or a 1:1 mixture of control interfering RNA and hsp-1(RNAi). hsp-1(RNAi) produces developmental arrest during larval life. (B) Digital photos of SJ4003 or ALF127 {skn-1(zu135)/nT1[qIs51] aip-1p:GFP} worms treated with the control or hsp-1(RNAi) from egg hatching. (C) Graph of fluorescence measured from the digital images in panel A [n > 15 for all; P < 0.0001 for all versus control interfering RNA, P = 0.293 for hsp-1 and fah-1(RNAi) versus hsp-1 and control interfering RNA]. (D) Graph of fluorescence measured from digital images in panel B [n > 15 for all; P < 0.0001 for hsp-1 versus control interfering RNA for wild-type (WT) and skn-1 and P = 0.0034 for wild-type hsp-1(RNAi) versus skn-1 hsp-1(RNAi)].

DISCUSSION

aip-1 expression is induced by tyrosine metabolites.

The AIRAP/aip-1 genes play critical roles in the response to environmental arsenic exposure and are required for a normal worm life span (56, 57, 64). However, it was unknown if any endogenous sources of stress are recognized by this pathway. We found that metabolites which accumulate in worms with impaired tyrosine degradation due to RNAi against fah-1 serve as a potent activator of aip-1 expression (Fig. 2). Our results indicate that aip-1 may not only be involved in facilitating the degradation of proteins that are either intrinsically unstable or damaged by external stressors but it could also be involved in preventing the accumulation of damaged proteins from chronic internal metabolic stress.

Similarly to arsenic, which is a known inducer of aip-1 expression, the tyrosine metabolites maleylacetoacetate and fumarylacetoacetate produce multiple types of cell stress in exposed cells (7, 13, 15, 18, 21, 27-29, 36, 56, 59). The production of multiple types of cell stress simultaneously may distinguish inducers of aip-1 expression from other exposures, like heat, peroxide, or tunicamycin, which stress cells but do not consistently activate aip-1 expression (25, 56). While the exact mechanisms leading to aip-1 expression are unknown, perhaps either overloading of proteosome capacity or the resulting accumulation of polyubiquitinated proteins may be involved because we find that RNAi against heat shock proteins or proteosomal subunits triggers aip-1 expression.

It is unknown if other metabolic intermediates activate this pathway, but it is reasonable to expect that other similarly reactive compounds would. For example, the neurotransmitter dopamine is derived from tyrosine and has similar chemical properties. In vivo, dopamine produces types of cell damage similar to those produced by the tyrosine metabolites because recent work has shown increases in oxidative stress and activation of the unfolded protein response following exposure (10, 14, 19, 20, 24, 58). Perhaps AIRAP could be activated following dopamine exposure and play a role in the prevention of damage to dopaminergic neurons. This would be consistent with the observation that dopamine and proteosome inhibitors show synergistic effects with regard to cell toxicity and protein aggregation (47, 63, 65, 67).

Beyond other metabolic pathways, our results lead to the important conclusion that a large number of stressors which produce sufficient levels of proteotoxic stress can serve as inducers of aip-1 expression. Hence, aip-1/AIRAP may be important in a significant number of disease states.

aip-1 induction involves skn-1-dependent and -independent pathways.

The expression of aip-1 is complex, as both skn-1 and other transcription factors are required for full induction following exposure to tyrosine metabolites (Fig. 3 and 8). skn-1 is the C. elegans homolog of the gene for Nrf2 and is directly activated by oxidative stress via multiple protein kinases and also via proteosomal dysfunction (2, 3, 26, 30, 33). Both arsenic and tyrosine metabolites could activate skn-1 via either mechanism, as both oxidative stress and proteosomal dysfunction are observed following exposure to either stressor (18, 56, 57). Our results suggest only a modest role for the sek-1 kinase pathway, as mutation of pmk-1 only slightly reduces the induction of aip-1 following exposure to tyrosine metabolites (Fig. 6). Recently, other kinases involved in the regulation of skn-1 have been described, so these could represent an alternative way of stimulating skn-1 activity (33). However, skn-1 is not sufficient to activate the expression of aip-1, as genetic activation of skn-1 with wdr-23(RNAi) results in only a marginal increase in expression (Fig. 7). Our findings are consistent with a model whereby both skn-1 and other proteins, including hsf-1, bind to the aip-1 promoter and direct expression (Fig. 13). Besides hsf-1, we have found evidence for additional transcription factors in the regulation of aip-1 expression, and the recently described slr-2 gene might also be directly involved in expression (data not shown) (35). These proteins may interact with skn-1 either at the level of DNA binding or in the recruitment of coactivators or the basal transcription machinery.

FIG. 13.

Model of control of aip-1 expression. Internal or external stressors such as arsenic, tyrosine metabolites, or juglone lead to the accumulation of damaged proteins due to relative proteosome dysfunction. Either proteosome dysfunction or oxidative stress related to the initial exposure leads to the activation of skn-1, which directly regulates aip-1 expression. Secondary pathways, including hsf-1 and perhaps others, marked X, are also activated and are required for aip-1 expression. The production of AIP-1 protein then facilitates the degradation of the damaged proteins and might lead to termination of the initial trigger for aip-1 induction.

Beyond regulating the aip-1 promoter, skn-1, hsf-1, and perhaps additional transcription factors play an important role in cellular physiology. In addition to the activation of aip-1, RNAi against hsp-1 and multiple proteosomal subunits also activates the expression of gpdh-1, which encodes glycerol-3-phosphate dehydrogenase and synthesizes intracellular glycerol in response to hypertonic stress (38). This suggests that the activated transcription factors could be involved in the regulation of responses to distinct stresses that have related cellular effects, such as an increase in protein aggregation (12). Importantly, while the cell damage and cell stress pathways show some overlap, the outputs are rather distinct, which implies that somewhere in the process specificity with regard to stress and response is generated. For example, we find that exposure to osmotic stress does not have a significant effect on aip-1 expression (Fig. 9). Perhaps the involvement of multiple transcription factors in the transcriptional response to stress allows tuning of the response to the character of a given stressor. For instance, the aip-1 promoter contains binding sites for skn-1, whereas the gpdh-1 promoter does not (Fisher, unpublished). This could suggest that inducers of aip-1, such as juglone, arsenic, and tyrosine metabolites, need to have the ability to produce oxidative stress, whereas perhaps osmotic stress, which induces gpdh-1, does not and activates transcription factors other than skn-1. Further work is necessary to understand the molecular basis of the observed specificity.

Could AIRAP/aip-1 be a target of drug discovery?

The AIRAP/aip-1 pathway holds significant promise as a means to treat diseases associated with misfolded proteins in either the cytoplasm or the ER. This includes multiple neurodegenerative diseases, including those associated with polyglutamine repeat expansion, Lewy bodies or Pick bodies (i.e., Parkinson's disease, Pick's disease, and dementia with Lewy bodies), some familial forms of amyotrophic lateral sclerosis, genetic diseases such as α1-antitrypsin deficiency associated with the accumulation of misfolded ER proteins, and metabolic diseases such as diabetes (54). Diabetes is particularly promising, as ER stress has been shown to lead to disease development in transgenic animals, and reduction of ER stress with chemical chaperones has treated diabetes in obese ob/ob mice (22, 51).

Furthermore, consumption of arsenic in drinking water has been associated with the development of diabetes, which is interesting in light of the production of several types of protein damage and induction of AIRAP/aip-1 by arsenic exposure in worms and cultured cells (48). Finally, in worms, loss of aip-1 has been shown to lead to premature aging (64). Hence, the development of compounds that can activate this cell stress response pathway without the production of cellular damage holds great potential for providing new opportunities to treat or, more importantly, prevent diseases associated with aging. This is consistent with recent work showing that increased aip-1 expression can ameliorate the formation of amyloid in a worm model of Alzheimer's disease involving the expression of an Aβ peptide in muscle cells (23).