Neuronal circuitry regulates the response of Caenorhabditis elegans to misfolded proteins

Thermosensory Neuronal Mutants Suppress Aggregation of Polyglutamine Expansion Proteins Expressed in Intestinal Cells of C. elegans.

The role of neuronal signaling on chronic protein misfolding in C. elegans was examined by using the gcy-8 (30) and ttx-3 (31) loss-of-function mutations that specifically affect the AFD thermosensory neurons and AIY interneurons, respectively. GCY-8 is a guanylyl cyclase expressed only in the two AFD neurons (30), and TTX-3 is an LIM-homeobox transcription factor expressed in the AIY interneurons (31). We monitored misfolding by using the in vivo folding reporter Q44::YFP, which consists of 44 glutamine residues fused to YFP under the control of the intestine-specific vha-6 promoter (32). Q44::YFP has been shown to aggregate in an age-dependent manner, impairing cellular function (32). Our previous data that gcy-8 and ttx-3 mutants are deficient in the HSR (29) led us to expect that mutations in the thermosensory circuitry would exacerbate polyglutamine expansion (polyQ) protein aggregation.

Instead, expression of Q44 in the background of gcy-8 or ttx-3 mutations led to a reduction in the onset and extent of polyQ aggregation in intestinal cells (Fig. 1 A–C and Fig. S1A). Whereas 40% of adult day 2 animals expressing Q44::YFP accumulate immobile polyQ aggregates (Materials and Methods), none of the gcy-8; Q44 animals, and only 25% of the ttx-3; Q44 animals of comparable age, had aggregates (Fig. 1B). This suppression of aggregation persisted on day 3 of adulthood with less than half the gcy-8; Q44 and ttx-3; Q44 animals with aggregates, and the remaining animals containing fewer than 20 aggregates (Fig. 1 A and B and Fig. S1A), compared with 80% of WT Q44 animals with more than 20 aggregates (Fig. 1 A and B and Fig. S1A). The reduction in polyQ aggregation in the gcy-8 and ttx-3 animals was not to caused by decreased expression of polyQ mRNA and protein levels relative to WT animals (Fig. S1 B and C). Fluorescence recovery after photobleaching (FRAP) was used to demonstrate that the aggregates, in all animals, were immobile and not exchangeable with the surrounding diffusible polyQ::YFP (Fig. 1 B and C). In the thermosensory mutant animals that expressed fewer polyQ aggregates, the diffuse polyQ protein in intestinal cells of gcy-8; Q44 and ttx-3; Q44 animals was soluble, with properties similar to YFP alone (Fig. 1C). Thus, disruption of thermosensory neuronal activity suppressed aggregation and increased soluble polyQ in intestinal cells. This effect was not a consequence of general inhibition of sensory neuronal activity, as disruption of the chemosensory neurons ASH, ADL, and ASE in the osm-9 mutation (33) (Fig. 1 A and B and Fig. S1A) or ocr-2 mutation (33) (Fig. S2 A–C) did not affect polyQ aggregation.

Thermosensory Neuronal Mutants Suppress Aggregation and Toxicity of Multiple Folding Reporters in C. elegans.

These results led us to ask whether the beneficial effects of thermosensory mutations on polyQ aggregation was limited to the intestinal cells or extended to metastable proteins expressed in other tissues of C. elegans. To address this, we examined the effect of the gcy-8 and ttx-3 mutations on three additional models for protein misfolding corresponding to polyQ aggregation in muscle cells (14, 34), expression of ALS-associated mutant SOD-1^G93A (35), and an endogenous metastable temperature-sensitive isoform of paramyosin (36).

For each protein, we observed that gcy-8 and ttx-3 mutations suppressed misfolding and associated toxicity (Figs. 2 and 3 and Fig. S3). Q35::YFP expressed in body wall muscle cells exhibits age-dependent aggregation (Fig. 2 A and B), which was markedly suppressed in gcy-8 and ttx-3 animals, as determined by morphological criteria (Fig. 2 A and B) and FRAP analysis (Fig. S3A). Suppression of polyQ aggregation in the gcy-8 and ttx-3 backgrounds was not caused by decreased expression of Q35 mRNA or protein (Fig. S3 B and C) and was specific to disruption of the thermosensory circuitry, as RNAi-mediated down-regulation of other gene products expressed by the AFD neurons also led to the suppression of aggregation (Fig. S3D), whereas the osm-9 (Fig. 2B) and ocr-2 (Fig. S2D) mutations did not.

Expression of Q35 in muscle cells causes a 60% reduction in motility of day 4 adult animals (14, 34) relative to age-matched nontransgenic animals (Fig. 2C). However, in gcy-8 and ttx-3 animals expressing Q35, the motility rates were restored close to WT levels (Fig. 2C), whereas the osm-9 mutation showed no effect (Fig. 2C). This protective role of the gcy-8 and ttx-3 mutations also extended to another disease-related aggregation prone protein, mutant SOD-1^G93A (35). Expression of mutant SOD-1^G93A in the background of a WT thermosensory circuitry causes developmental delays such that only 30% of animals develop past the second larval (L2) stage at 25 °C. In the thermosensory mutant background, nearly all SOD-1^G93A–expressing animals developed past L2 stage (Fig. S3E).

In addition to suppressing the misfolding and toxicity of exogenous aggregation-prone proteins, the thermosensory mutations also suppressed the proteotoxicity associated with misfolding of a metastable temperature-sensitive isoform of paramyosin, unc-15(ts) (36). At the restrictive temperature (25 °C), paramyosin (ts) misfolds and aggregates, disrupting muscle structure and function (36). Consequently, only approximately 10% of unc-15(ts) embryos develop beyond the comma stage (Fig. 3A) (37), with adult animals showing approximately 75% reduction in motility relative to WT animals (Fig. 3B and Movie S1). In the presence of gcy-8 and ttx-3 mutations, the toxicity caused by the misfolding of paramyosin unc-15(ts) (Fig. 3) was substantially reduced. Approximately 30% of gcy-8;unc-15(ts) and approximately 40% of ttx-3;unc-15(ts) embryos hatched (Fig. 3A), and motility of adult gcy-8;unc-15(ts) and ttx-3;unc-15(ts) animals was restored to near that of WT animals (Fig. 3B and Movie S1).

These results demonstrate that modulation of thermosensory neuronal activity suppressed protein misfolding and related cellular dysfunction as a result of the expression of multiple metastable proteins in separate tissues. This suggests that the protein folding environment was affected globally by cell-nonautonomous control by the thermosensory neuronal circuitry.

Thermosensory Circuitry Modulates Protein Misfolding Through Dense Core Vesicle-Dependent Neurosecretion.

To investigate how the gcy-8 mutation influenced aggregation and toxicity, we examined whether the loss of downstream signaling pathways would recapitulate the global suppression of protein misfolding seen in the gcy-8 animals. The thermosensory mutations not only suppressed aggregation in muscle cells but also affected intestinal tissue that is not innervated, suggesting that neuronal signaling could occur through secreted molecules. Therefore, we tested whether inhibition of one of the major modes of neurosecretion, the dense core vesicle (DCV) release through calcium-activated protein secretion (CAPS) (38, 39), also suppressed polyQ aggregation in intestinal and muscle cells.

Deletion of the single C. elegans orthologue of CAPS, unc-31 (38, 39), and the subsequent loss of DCV-dependent neurosecretion resulted in suppression of Q44::YFP aggregation in the intestine (Fig. 4 A–C) and Q35::YFP aggregation in muscle cells (Fig. 4H), and the appearance of soluble protein by FRAP analysis (Fig. 4G). Moreover, polyQ aggregation in the gcy-8; unc-31 double mutants was indistinguishable from either of the single gcy-8 or unc-31 mutants alone (Fig. 4 D–F and H), suggesting that gcy-8 and unc-31 are in the same genetic pathway. These results showed that the AFD neurons modulate aggregation through suppression of DCV-dependent neurosecretion and suggested that these pathways may exert a net-inhibitory influence on the protein quality-control machinery of nonneuronal cells.

Suppression of Proteotoxicity in Thermosensory Mutants Caused by HSF-1–Dependent Up-Regulation of Chaperones.

Protein quality control in C. elegans is regulated at the molecular level by HSF-1 (34, 40). Reducing the levels of hsf-1 by RNAi-induced knockdown restored aggregation in the gcy-8 animals to that of WT animals (Fig. 5A and Fig. S4B). Likewise, the suppression of polyQ aggregation in unc-31 mutant animals was HSF-1–dependent (Fig. S4C). RNAi-induced knockdown of DAF-16 did not revert the suppression of polyQ aggregation in the gcy-8 (Fig. 5B and Fig. S4B) or unc-31 animals (Fig. S4C), suggesting that the effects of the neuronal mutations on protein aggregation was dependent on HSF-1.

We next examined whether chaperone levels were up-regulated in the thermosensory mutants expressing aggregation-prone proteins. Mutations affecting the thermosensory neurons alone do not alter the basal levels of chaperones (Fig. 5 and Fig. S5). Expression levels of the constitutive HSP70 (hsp-1), the inducible HSP70s (C12C8.1 and F44E5.4), the inducible small HSP (hsp16.2), and HSP90 (daf-21) were similar for gcy-8 and WT animals (Fig. 5 and Fig. S5). In addition, as seen in other model systems the aggregation of polyQ did not up-regulate chaperone expression (Fig. 5C and Fig. S5 A–E). However, the expression of polyQ in the gcy-8 animals but not in other chemosensory mutants resulted in the up-regulation of each of these HSP genes (Fig. 5C and Figs. S2E and S5 A–E).

Similar results were obtained with regard to the misfolding of paramyosin. At the permissive temperature of 15 °C, the basal levels of inducible HSP70s (C12C8.1 and F44E5.4) in WT and gcy-8 animals were similar regardless of whether unc-15(ts) was expressed. The misfolding of paramyosin (ts) at the restrictive temperature of 25 °C in animals with WT thermosensory neurons did not cause the up-regulation of hsp70 (C12C8.1 and F44E5.4) mRNA: chaperone levels in the unc-15(ts) animals were similar to animals that expressed the WT isoform of paramyosin (Fig. 5D and Fig. S5F). In contrast, the misfolding of paramyosin in gcy-8 mutant animals resulted in up-regulation of the inducible cytoplasmic HSP70s (Fig. 5D and Fig. S5F).

We confirmed that the up-regulation of chaperones in gcy-8 mutant animals expressing misfolded proteins was dependent on HSF-1 (Fig. 5E). Moreover, up-regulation of chaperones was responsible for the suppression of aggregation in the thermosensory mutant animals (Fig. 5 F and G). RNAi-induced knockdown of hsp70 (C12C8.1) levels in the gcy-8;Q35 animals prevented the gcy-8–dependent rescue of aggregation (Fig. 5F). We observed no effect in Q35 animals with WT thermosensory function, presumably as hsp70 (C12C8.1) was not induced. Similarly, knockdown of hsp90 (daf-21) in the gcy-8;Q44 animals blocked the suppression of aggregation observed in these animals (Fig. 5G).

Thus, the chronic expression of aggregation-prone proteins in C. elegans tissues does not result in the up-regulation of chaperones. Disruption of neuronal activity restored HSF-1–dependent chaperone up-regulation upon chronic expression of misfolded proteins, and suppressed aggregation and toxicity.

Thermosensory Neuronal Function Allows C. elegans to Distinguish Between Acute Temperature Stress and Chronic Stress of Protein Misfolding.

The up-regulation of chaperones in thermosensory mutant animals and the rescue of aggregation and toxicity presented a conundrum, as we had previously observed that these mutations in the neuronal circuitry dampened the HSR. We reasoned that this may be because a short, acute HS that causes a rapid and transient change in protein folding dynamics is not recognized in the same manner as a prolonged chronic accumulation of misfolded proteins. The ability of the thermosensory neurons of C. elegans to detect and respond to temperature change and regulate proteostasis may allow for distinct organismal responses to these stresses.

As shown previously, aggregation of polyQ in C. elegans does not up-regulate chaperones in animals with a WT neuronal circuitry (Fig. 5 and Fig. S5). However, exposure of Q35 and Q44 animals with WT thermosensory neurons to HS (34 °C for 15 min) induced the expression of HSP genes (Fig. 5 H–J), whereas gcy-8;Q35 and gcy-8; Q44 animals showed no further increase in chaperone expression beyond basal levels (Fig. 5 H–J). This indicates that the expression of aggregation-prone proteins within the cells of C. elegans does not compromise the HSR. Rather, a WT thermosensory neuron compromises the response to chronic protein aggregation. To verify that the activity of thermosensory neurons distinguished between acute and chronic stress, we exposed WT and gcy-8 mutant animals to chronic heat stress. We predicted that gcy-8 animals, although deficient for chaperone induction upon acute HS, would induce chaperone expression when exposed to the chronic stress. This was indeed the case: even through gcy-8 animals did not respond to a transient HS (Fig. S6A), chaperone induction was observed upon exposure to repeated or prolonged HS (Fig. S6 B and C).

These data show that the thermosensory neurons inhibit chaperone up-regulation upon chronic stress and the occurrence of misfolded proteins, but permit HSP up-regulation upon acute HS, allowing the animal to respond differently to acute and chronic proteotoxic conditions.

C. elegans Strains and Growth Conditions.

The C. elegans strains and methods of cultivation are listed in SI Materials and Methods.

Scoring of Aggregates.

Aggregates were scored every 24 h post-L4, using a Leica fluorescent stereomicroscope (MZFLIII) with the EYFP filter set (excitation, 510/20; emission, 560/40). Aggregates were recognized visually, based on experience from FRAP regarding which foci did not recover following photobleaching. Aggregates too close to be resolved under the stereomicroscope, or smaller satellite aggregates scattered near a large aggregate were accounted as one. This system was consistently maintained for all strains. The number of aggregates in each worm was scored blind, by independent investigators and yielded very similar results. To quantify Q44 aggregation in the intestine, animals were binned into pools containing 0, 1 to 20, and greater than 20 aggregates. To quantify Q35 aggregation in the body wall muscle cells, all visible aggregates in approximately 100 animals, corresponding to more than three biological samples, were counted over a period of 4 d.

RNA Extraction and Quantitative RT-PCR.

mRNA was prepared by using TRIzol extraction as previously published and detailed in SI Materials and Methods.

RNAi Experiments.

RNAi experiments were conducted according to standard protocols and the experimental protocol, and RNAi constructs are detailed in SI Materials and Methods. RNAi against the AFD genes was conducted in the sensitized strain rrf-3 strain expressing Q35, which accumulated slightly lower numbers of aggregates.

Embryonic Lethality Experiments.

Toxicity in the SOD-1^G93A animals was measured by transferring 10 L4 WT G93A and gcy-8 G93A larvae from 20 °C to 25 °C, allowing them to lay eggs for 24 to 36 h, counting total embryos, and calculating the percentage of larvae that grew to L3 stage. The experiment was repeated three or four times.

Embryonic lethality in unc-15 (ts) animals was measured by allowing approximately 10 unc-15(ts), gcy-8;unc-15(ts), or ttx-3;unc-15(ts) animals to lay eggs at 15 °C, and 24 h later, transferring 20 to 50 embryos at the comma cell stage onto fresh plates at 25 °C. The number of larvae that developed into L2 larvae after 24 to 48 h were scored. Experiments were repeated at least three times.

Motility Assays.

The motility of N2, gcy-8, Q35, gcy-8;Q35, ttx-3;Q35, unc-15(ts), gcy-8;unc-15(ts), ttx-3;unc-15(ts) was assayed upon transferring 10 animals to a new 6-cm plate seeded uniformly with Op50 equilibrated to room temperature. Movies of crawling animals were recorded with a Leica MZ10 microscope and Hamamatsu C10600-10B (Orca-R2) camera by using SimplePCI software (Leica) at 2 × 2 binning and 5 frames/s, and analyzed by using ImageJ as described in SI Materials and Methods. Each sample of 10 animals was measured only once. At least three samples of 10 animals were considered one biological sample, and each report of motility involved three to five biological samples.

FRAP.

A detailed description of FRAP methodology, the basis for selection of the aggregates, and how many aggregates were analyzed is included in SI Materials and Methods.

SDS/PAGE Gels and Western Blots for Examining Protein Levels.

PolyQ protein expression was measured by boiling 10 adult day 3 animals in SDS sample buffer for 15 min, and conducting a standard Western analysis. Antibodies used are listed in SI Materials and Methods. No cleavage of YFP from the polyQ-expressing polypeptide was observed.

HS.

For acute HS, 10 L4 animals (N2, Q35, Q44, gcy-8, gcy-8;Q35, gcy-8;Q44) were moved onto a fresh plate from a population, allowed to mature into day 3 adults and heat-shocked in a water bath for 15 min at 34 °C according to previously published methods. We ensured that the polyQ-expressing animals had accumulated aggregates before HS. For prolonged chronic HS, 10 L4 animals were exposed to a constant temperature of 28 °C for 24 h. Repeated acute HSs were delivered by three cycles of exposure to 34 °C for 15 min interspersed by 15 min recovery at 20 °C.