Report

Four glial cells regulate ER stress resistance and longevity via neuropeptide signaling in C. elegans

See allHide authors and affiliations

Science  24 Jan 2020:
Vol. 367, Issue 6476, pp. 436-440
DOI: 10.1126/science.aaz6896

Taking the stress out of life

In the model organism Caenorhabditis elegans, a roundworm, it has been shown that neurons can communicate proteostasis to the periphery to affect aging. Frakes et al. have now identified astrocytelike glial cells that also act as central regulators of systemic protein homeostasis and aging (see the Perspective by Miklas and Brunet). They found that the life span of C. elegans can be extended by expression of a constitutively active version of the transcription factor XBP-1s, which mediates the unfolded protein response of the endoplasmic reticulum (UPRER), in a specific subset of glial cells. Glial XBP-1s initiates induction of the UPRER in distal intestinal cells, which makes the worms more resistant to chronic ER stress. Neuropeptide signaling was required for glial-mediated longevity and induction of the peripheral UPRER, suggesting a distinct mechanism from that initiated by neuronal XBP-1s. Thus, in this animal model of aging, a mere four cells can control organismal physiology and aging

Science, this issue p. 436; see also p. 365

Abstract

The ability of the nervous system to sense cellular stress and coordinate protein homeostasis is essential for organismal health. Unfortunately, stress responses that mitigate disturbances in proteostasis, such as the unfolded protein response of the endoplasmic reticulum (UPRER), become defunct with age. In this work, we expressed the constitutively active UPRER transcription factor, XBP-1s, in a subset of astrocyte-like glia, which extended the life span in Caenorhabditis elegans. Glial XBP-1s initiated a robust cell nonautonomous activation of the UPRER in distal cells and rendered animals more resistant to protein aggregation and chronic ER stress. Mutants deficient in neuropeptide processing and secretion suppressed glial cell nonautonomous induction of the UPRER and life-span extension. Thus, astrocyte-like glial cells play a role in regulating organismal ER stress resistance and longevity.

During aging, there is an organism-wide loss of protein homeostasis, exacerbated by the inability to mount an effective unfolded protein response of the endoplasmic reticulum (UPRER), which likely contributes to tissue damage and increased susceptibility to disease (13). The age-dependent decline in the ability to induce the UPRER can be prevented by the selective overexpression of constitutively active xbp-1s in neurons. Neuronal XBP-1s leads to cell nonautonomous activation of the UPRER in distal intestinal cells, which is sufficient to confer ER stress resistance and prolong life span (2). To date, cell nonautonomous stress signaling has been ascribed only to neurons (2, 47). However, glial cells—the gatekeepers and guardians of the central nervous system—may also play a role in regulating organismal stress resistance and longevity (8).

To determine whether glia play a role in regulating protein homeostasis and longevity, we generated Caenorhabditis elegans strains overexpressing xbp-1s under a glial-specific promoter, ptr-10, which is expressed in most glia except amphid sheath glia (fig. S1A) (9, 10). Animals expressing xbp-1s in most glia (ptr-10p::xbp-1s) exhibited a marked increase in survival compared with control (N2) animals (Fig. 1A). To identify which glial cells were mediating XBP-1s–dependent longevity, we expressed xbp-1s within select subtypes of the 56 C. elegans glial cells (11). Expression of xbp-1s specifically in two amphid and two phasmid sheath glia (AMsh and PHsh) using the fig-1 promoter did not extend life span beyond that of control animals (Fig. 1B and fig. S1B) (12). However, expression of xbp-1s in the four cephalic astrocyte-like sheath glia (CEPsh) using the hlh-17 promoter resulted in an extension of life span (Fig. 1C and fig. S1C) (13, 14).

Fig. 1 Glial xbp-1s extends life span and induces cell nonautonomous UPRER.

(A) Survival of animals expressing xbp-1s in most glia [ptr-10p::xbp-1s, line 1 (dark blue), line 2 (light blue)] compared with control N2 animals (black). (B) Survival of animals expressing xbp-1s in four amphid and phasmid sheath glia [fig-1p::xbp-1s, line 1 (dark blue), line 2 (light blue)] compared with control N2 animals (black). (C) Survival of animals expressing xbp-1s in four cephalic sheath glia [hlh-17p::xbp-1s, line 1 (dark blue), line 2 (light blue)] compared with control N2 animals (black).(D and E) Fluorescent micrograph (D) and quantification (E) of UPRER reporter worms (hsp-4p::GFP) expressing hlh-17p::xbp-1s (left). hlh-17p::GFP reporter worms, pseudo-colored red (right), are shown. Data in (D) are representative of n > 10. Scale bars, 250 μm. Quantification of hsp-4p::GFP fluorescence using COPAS biosorter was normalized to time of flight (length) and extinction (thickness) of animals. Results are shown relative to hsp-4p::GFP alone (control) with error bars representing means ± SD. One-way analysis of variance (ANOVA) Tukey’s post hoc test, n = 2, ****P < 0.0001. Life spans are representative of n = 3. See table S1 for life-span statistics.

We hypothesized that glial xbp-1s was inducing a beneficial UPRER, leading to life-span extension. To characterize the localization and extent of UPRER activation, we generated animals expressing xbp-1s in glia with the UPRER reporter strain, hsp-4p::GFP (15). At day 1 of adulthood, expression of xbp-1s in most glia (ptr-10p::xbp1s) or all glia (mir-228p::xbp-1s) induced hsp-4::GFP in glial cells and in the distal intestine (fig. S2, A to C). Animals overexpressing xbp-1s in AMsh and PHsh glia (fig-1p::xbp1s) exhibited robust hsp-4::GFP induction in fig-1–expressing glial cells and in the distal intestine. This expression pattern was distinct from that of fig-1p::tdTomato reporter animals, in which tdTomato fluorescence was restricted to AMsh and PHsh glia (fig. S1B and fig. S3, A and B). Animals overexpressing xbp-1s in the four CEPsh glia (hlh-17p::xbp-1s) showed induction of hsp-4p::GFP in the CEPsh glia and in the distal intestine and pharynx (Fig. 1, D and E). Notably, green fluorescent protein (GFP) expression was limited to CEPsh glial cells in hlh-17p::GFP reporter animals (fig. S1C) (9, 10, 13, 14, 1618). These data suggest that xbp-1s expression in glial cells can induce cell nonautonomous UPRER in distal intestinal cells and that CEPsh glia have a unique role in regulating xbp-1s–mediated longevity.

To elucidate how CEPsh glia promote longevity via xbp-1s, we first tested whether life-span extension and cell nonautonomous activation of the UPRER from CEPsh glia was dependent on the known signaling components of the UPRER branches, PERK, ATF6, and XBP1, encoded by pek-1, atf-6, and xbp-1, respectively, in C. elegans. No difference was observed in hsp-4::GFP induction with pek-1 or atf-6 RNA interference (RNAi)–mediated knockdown in hlh-17p::xbp-1s animals (Fig. 2A and fig. S4). However, knockdown of xbp-1 reduced GFP fluorescence of hlh-17p::xbp-1s; hsp-4p::GFP animals and abolished the life-span extension of hlh-17p::xbp-1s animals (Fig. 2, A and B, and fig. S4). Whole-worm RNA sequencing (RNA-seq) of hlh-17p::xbp-1s animals revealed 115 differentially expressed genes (adjusted P value <0.05), including a significant increase in xbp-1s–dependent transcripts (Fig. 2, C and D, and table S2) (19). Gene ontology analysis showed enrichment of genes involved in the immune response, stress response, and, as expected, response to ER stress (table S3).

Fig. 2 Cell nonautonomous induction of the UPRER is dependent on xbp-1, but not atf-6 or pek-1.

(A) Fluorescent micrographs of day 1 hsp-4p::GFP; hlh-17p::xbp1-s animals grown on control empty vector (EV), atf-6, pek-1, or xbp-1 RNAi from hatch. Scale bar, 250 µm; n = 3. (B) Survival of control (N2) and hlh-17p::xbp-1s animals grown on EV control RNAi or RNAi-targeting xbp-1. See table S1 for life-span statistics; n = 2. (C) Volcano plot of whole-animal transcriptional profiling from hlh-17p::xbp-1s animals compared with wild type (N2). xbp-1 is highlighted in red. Note that aex-5 (gray) was detected as highly overexpressed because of a small aex-5 promoter and exon fragment present in the 3′ untranslated region in the backbone plasmid used for all constructs. All aex-5 reads aligned to this short fragment. (D) The UPRER is activated in animals expressing xbp-1s in CEPsh glia compared with N2, shown by fold change of two gene groups: UPRER (GO: 0030968 and 1900103) or xbp-1 targets (19). The line inside the box represents the median change of the gene group. ***P < 0.001. GO enrichment analysis for genes with a fold change P value <0.05 for terms with a false discovery rate Q value <0.05 can be found in table S3.

We hypothesized that the increased activation of the UPRER in hlh-17p::xbp-1s animals would render these animals more resistant to age-dependent protein aggregation and chronic ER stress. Expression of xbp-1s in CEPsh glia notably reduced aggregation of yellow fluorescent protein (YFP)–tagged, Huntington-like polyglutamine protein in the intestine (with age) compared with controls (Fig. 3A). Furthermore, animals expressing xbp-1s in CEPsh glia exhibited an increase in survival when chronically exposed to tunicamycin, a chemical inducer of ER stress (Fig. 3B). Perturbing CEPsh glial development, using a partially penetrant reconstituted caspase (recCasp), abrogated the ER stress resistance of hlh-17p::xbp-1s animals grown on tunicamycin-containing plates and decreased the median life span of hlh-17p::xbp-1s animals grown on control plates (Fig. 3B and fig. S5, A and B). Moreover, distal UPRER was reduced in hlh-17p::xbp-1s animals harboring hlh-17::recCasp (Fig. 3C). Consistent with these findings, hsp-4p::GFP induction was suppressed in hlh-17p::xbp-1s animals harboring a loss-of-function mutation in vab-3, a Pax6/7-related gene required for CEPsh glial cell development (fig. S5, C and D) (10). In contrast to other model organisms, ablation of glial cells does not lead to neuronal cell death in C. elegans (20).

Fig. 3 Expression of xbp-1s in glial cells protects animals against protein aggregation and chronic ER stress.

(A) Fluorescent micrograph and quantification of age-dependent accumulation of polyQ44-YFP aggregates in control animals or animals expressing hlh-17p::xbp-1s. Control animals average 3.5 puncta per animal, compared with 1.3 in hlh-17p::xbp-1s animals (P < 0.0001). Scale bar, 250 µm; n = 2. (B) Survival of animals transferred to tunicamycin-containing plates at day 1 of adulthood. CEPsh glial ablation via hlh-17p::recCasp suppresses hlh-17p::xbp-1s ER stress resistance. n = 2. (C) Fluorescent micrograph of hsp-4p::GFP reporter worms expressing hlh-17p::xbp-1s and hlh-17p::xbp-1s; hlh-17p::recCasp. GFP puncta in hlh-17p::recCasp strain represent co-injection (coinj.) marker for hlh-17p::recCasp transgene, which is expressed in coelomocytes. White bracket marks distal intestine, where induction of cell nonautonomous UPRER is reduced in animals expressing hlh-17p::recCasp. Scale bar, 250 µm; n = 3.

Next, we assessed whether overexpression of xbp-1s in CEPsh glia induces other stress responses known to affect protein homeostasis and longevity, such as the mitochondrial UPR (UPRMT), the heat shock response (HSR), or reduced insulin and insulin-like growth factor 1 (IGF-1) signaling (5, 21, 22). We did not observe induction of the UPRMT reporter, hsp-6::GFP, the HSR reporter, hsp-16.2::GFP, or the sod-3p::GFP reporter with hlh-17p::xbp1s expression. However, hlh-17p::xbp-1s animals were still capable of activating these responses (fig. S6). Taken together, these data indicate that expression of xbp-1s in CEPsh glia specifically induces the UPRER, which protects animals from age-dependent protein aggregation and chronic ER stress.

Previously, our laboratory had found that cell nonautonomous activation of the UPRER by neuronal xbp-1s is dependent on the release of small clear synaptic vesicles (SCVs) containing neurotransmitters (2). To determine if glial xbp-1s signals through a mechanism similar to that of neuronal xbp-1s, we generated hlh-17p::xbp-1s; hsp-4p::GFP animals containing an unc-13 mutation, which are deficient in SCV exocytosis (23). Notably, cell nonautonomous signaling remained intact in hlh-17p::xbp-1s animals harboring either unc-13(e51) or unc-13(s69) mutations (Fig. 4A and fig. S7, A to D). Therefore, glia do not transmit UPRER to distal tissues via a SCV-dependent mechanism like neurons.

Fig. 4 Neuropeptides are required for glial cell nonautonomous activation of the UPRER and longevity.

(A and B) Fluorescent micrographs of control (hsp-4p::GFP) and hsp-4p::GFP; hlh-17p::xpb-1s (line 1) animals, with and without the unc-13(e51) or unc-31(e928) loss-of-function mutations, which render animals deficient in SCV or DCV release, respectively. Scale bars, 250 μm; n = 3. (C) COPAS quantification of animals in (B), n = 3. (D and E) Fluorescent micrographs (D) and COPAS quantification (E) of control and hsp-4p::GFP animals expressing intestinal xbp-1s (vha-6p::xbp-1s), with and without the unc-31(e928) mutation. Scale bar, 250 μm; n = 2. n.s., not significant. (F and G) Fluorescent micrographs (F) and COPAS quantification (G) of control and hsp-4p::GFP; hlh-17p::xpb-1s (line 1), with and without the egl-3(ok979) mutation, which renders animals unable to cleave pro-neuropeptides. Scale bar, 250 μm; n = 3. (H and I) Fluorescent micrographs (H) and COPAS quantification (I) of control and hsp-4p::GFP animals expressing xbp-1s in all neurons (rgef-1p::xbp-1s), with and without the egl-3(ok979) mutation. Scale bar, 250 μm; n = 2. (J) Survival of control (N2) animals (black), hlh-17p::xbp-1s (dark blue), neuronal(rgef-1p)::xbp-1s (gray), and hlh-17p::xbp-1s; neuronal(rgef-1p)::xbp-1s (green). n = 2. (K) Survival of control (N2) animals (black), hlh-17p::xbp-1s (dark blue), egl-3(ok979) (gray), and hlh-17p::xbp-1s; egl-3(ok979) (orange). n = 3. See table S1 for life-span statistics. COPAS results are shown relative to hsp-4p::GFP alone (control), with means ± SD. One-way ANOVA Tukey’s post hoc test, ****P < 0.0001.

CEPsh glia reside nearly 300 µm from where we observed robust distal activation of the UPRER. Therefore, we hypothesized that this transcellular signaling mechanism is dependent on neuropeptides, which are packaged into dense core vesicles (DCVs); can be secreted from neurons, glia, or neuroendocrine cells; and can function as long-range signaling hormones. We crossed hlh-17p::xbp-1s animals with an unc-31 loss-of-function mutant in which DCV exocytosis is disrupted. The unc-31(e928) mutation suppressed cell nonautonomous activation of the UPRER, with GFP fluorescence nearly equal to levels observed in hsp-4::GFP controls (Fig. 4, B and C, and fig. S7, E and F) (24). The unc-31(e928) mutation had no effect on cell autonomous activation of the UPRER in intestinal cells or neuronal cell nonautonomous activation of the UPRER (Fig. 4, D and E) (2). Furthermore, we tested a loss-of-function mutation in the proprotein convertase, egl-3, which is deficient in neuropeptide processing, and found that induction of the cell nonautonomous UPRER by CEPsh glia was suppressed (Fig. 4, F and G, and fig. S8, A and B) (25). Blocking neuropeptide processing had no effect on cell autonomous hsp-4p::GFP induction in intestinal cells or cell nonautonomous activation of the UPRER in animals expressing neuronal xbp-1s (fig. S9, A to D, and Fig. 4, H and I). Thus, glial-mediated cell nonautonomous induction of the UPRER is dependent on neuropeptides, which is an entirely distinct mechanism to that initiated by neurons expressing xbp-1s.

As an additional measure of the separation between neuronal and glial induction of peripheral UPRER, we removed CEPsh glial cells in animals expressing xbp-1s solely in neurons, and cell nonautonomous activation of the UPRER remained intact (fig. S10). Thus, neuronal activation of the peripheral UPRER via xbp-1s is independent of CEPsh glia. Next, we investigated whether combinatorial xbp-1s overexpression in both neurons and CEPsh glia would result in an additive increase in activation of the UPRER and life-span extension. Animals overexpressing xbp-1s in both neurons and CEPsh glia induced hsp-4p::GFP and extended life span to a greater degree than animals expressing xbp-1s only within CEPsh glia or neurons (fig. S11, A and B, and Fig. 4J).

To identify the cell type responsible for secreting the peptides mediating cell nonautonomous UPRER, we expressed wild-type unc-31(cDNA) in either neurons or glia in hlh-17p::xbp-1s; unc-31(e928) animals. Neuronal unc-31(cDNA) did not restore activation of the UPRER in the intestine of hlh-17p::xbp-1s; unc-31(e928) animals (fig. S12, A and B). In contrast, expression of unc-31(cDNA) in CEPsh glia or egl-3(cDNA) in CEPsh glia or all glia led to an increase in activation of the UPRER, albeit a modest increase (fig. S12, C and D). These data suggest that the neuropeptides required for glial-mediated cell nonautonomous activation of the UPRER do not originate from neurons but are secreted, in part, by glial cells themselves.

Lastly, we sought to determine whether neuropeptide signaling was mediating longevity in hlh-17p::xbp-1s animals. Loss-of-function egl-3 mutants are inherently long-lived because of reduced insulin and IGF-1 signaling (26). However, we did not observe an additive increase in survival of hlh-17p:;xbp-1s animals harboring the egl-3(ok979) mutation, suggesting that lifespan extension of hlh-17p::xbp-1s animals requires neuropeptides (Fig. 4K).

Previously, cell nonautonomous stress signaling from the brain to the periphery has been ascribed only to neurons. However, our data identify a subtype of astrocyte-like glial cells that coordinate systemic protein homeostasis and aging via neuropeptide signaling—a distinct mechanism from that initiated by neuronal XBP-1s (fig. S13). This suggests there is regional and functional specificity of glial cells to control physiology and aging that evolved as early as the nematode. We speculate that, depending on the physiological cue received by the nervous system, either neurons or glia can signal via XBP-1s to peripheral tissues to coordinate organismal protein homeostasis.

Supplementary Materials

science.sciencemag.org/content/367/6476/436/suppl/DC1

Materials and Methods

Figs. S1 to S13

Tables S1 to S3

References (2733)

References and Notes

Acknowledgments: We thank the Shaham laboratory and the Caenorhabditis Genetics Center (CGC) (P40 OD010440) for several strains used in this study and the Garrison laboratory for several plasmids. Thank you to the Dillin laboratory for comments and discussion throughout the project, R. Higuchi-Sanabria for preparing the samples for RNA-seq, and L. Joe for preparing the RNA-seq libraries. We thank P. Douglas for thoughtful input and discussion at the beginning of this work. Funding: A.E.F. is supported by NIH (F32 AG051355); M.G.M. is supported by NIH (F31 AG060660); R.B.-Z. is supported by the EMBO Long-Term Fellowship (462-2017); H.K.G. is supported by NSF (DGE1752814); and A.D. is supported by the Thomas and Stacey Siebel Foundation, the Howard Hughes Medical Institute, and NIH grants AG042679 and AG059566. Author contributions: A.E.F. conceived the study, generated C. elegans strains, performed experiments (life spans, microscopy, COPAS biosorting, and data analysis), and wrote the manuscript. M.G.M. performed life spans, microscopy, tunicamycin ER stress assays, and worm crosses and provided intellectual input. S.U.T. performed life spans, worm crosses, and COPAS biosorting and prepared artwork for Fig. 1 and fig. S13. R.B.-Z. analyzed RNA-seq data, generated figures, and provided intellectual input. J.D. helped generate and analyze strains, prepared artwork in fig. S7, and provided intellectual input. H.K.G. performed backcrosses and promoter characterization and provided intellectual input. N.K. assisted in generating strains and performed crosses and life spans. S.M. performed life spans and assisted with crosses and preparation of samples for RNA-seq. A.D. provided invaluable feedback throughout the project and toward the manuscript. All authors reviewed and edited the manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the supplementary materials. The RNA-seq dataset supporting the conclusions of this article is available in the National Center for Biotechnology Information Sequence Read Archive repository, under accession number BioProject PRJNA589459. Further information and requests for reagents may be directed to dillinlabmaterials{at}berkeley.edu and will be fulfilled by A.D.

Stay Connected to Science

Navigate This Article