Reduced bone morphogenic protein signaling along the gut–neuron axis by heat shock factor promotes longevity

Abstract Aging is a complex and highly regulated process of interwoven signaling mechanisms. As an ancient transcriptional regulator of thermal adaptation and protein homeostasis, the Heat Shock Factor, HSF‐1, has evolved functions within the nervous system to control age progression; however, the molecular details and signaling dynamics by which HSF‐1 modulates age across tissues remain unclear. Herein, we report a nonautonomous mode of age regulation by HSF‐1 in the Caenorhabditis elegans nervous system that works through the bone morphogenic protein, BMP, signaling pathway to modulate membrane trafficking in peripheral tissues. In particular, HSF‐1 represses the expression of the neuron‐specific BMP ligand, DBL‐1, and initiates a complementary negative feedback loop within the intestine. By reducing receipt of DBL‐1 in the periphery, the SMAD transcriptional coactivator, SMA‐3, represses the expression of critical membrane trafficking regulators including Rab GTPases involved in early (RAB‐5), late (RAB‐7), and recycling (RAB‐11.1) endosomal dynamics and the BMP receptor binding protein, SMA‐10. This reduces cell surface residency and steady‐state levels of the type I BMP receptor, SMA‐6, in the intestine and further dampens signal transmission to the periphery. Thus, the ability of HSF‐1 to coordinate BMP signaling along the gut–brain axis is an important determinate in age progression.

system has long been appreciated as the hub of information transfer and communication between the body's other organ systems (Miller et al., 2020). Modulating critical regulators of cellular adaption to stress in the nervous system initiates systemic alterations with cytoprotective and age-defying potential (Douglas et al., 2015;Durieux et al., 2011;Taylor & Dillin, 2013). Receipt of these neural-born signals in the peripheral tissues is essential for the transmission and initiation of distinct cellular responses. Communication between the brain and intestine has become increasingly appreciated in the context of human health and can occur through direct synaptic innervation of the tissues, neuroendocrine means, systemic fluctuations in metabolism, or the immune response (Ambrosini et al., 2019;Boehme et al., 2020;Carabotti et al., 2015;Jena et al., 2020). To date, the molecular and cellular details underlying these integrated, trans-organ signaling mechanisms and their role in age regulation remain unclear.
The heat shock transcription factor, HSF-1, is an ancient regulator of cellular adaption to thermal stress with evolved roles as an aging regulator in the nervous system of the nematode, Caenorhabditis elegans (Morimoto, 1998;Morley & Morimoto, 2004). While HSF-1 activates the expression of protein folding enzymes to ensure proteome integrity amid thermal destabilization (Akerfelt et al., 2010;Morley & Morimoto, 2004), it has also developed as an important aging regulator in part through its ability to cooperate with the FOXO transcription factor, DAF-16 (Hsu et al., 2003). Elevating the expression of HSF-1 exclusively within the nervous system extends animal lifespan and requires the activity of enteric DAF-16 (Douglas et al., 2015;Morley & Morimoto, 2004). Neural hsf-1 was recently shown to reduce signaling through the bone morphogenic protein (BMP) pathway for adaptation to elevated, nonpermissive temperatures (Chauve et al., 2021). As a member of the greater transforming growth factor β (TGFβ) signaling pathway in C. elegans, DBL-1 is a neural-produced BMP ligand (Morita et al., 1999;Suzuki et al., 1999) that signals to peripheral tissues through its binding to the heterodimeric receptor complex composed of the type 1 receptor, SMA-6, and type 2 receptor, DAF-4 (Gumienny & Savage-Dunn, 2013;Roberts et al., 2010;Wang et al., 2002). In addition to regulating body size and mail tail development (Morita et al., 1999;Suzuki et al., 1999), DBL-1 has been linked with olfactory learning, reproductive aging, lipid metabolism, and the innate immune response Luo et al., 2010;Zhang & Zhang, 2012;Zugasti & Ewbank, 2009). DBL-1 signal transduction in peripheral tissues acts through the nucleocytoplasmic SMAD proteins (SMA-2, SMA-3, and SMA-4) (Gumienny & Savage-Dunn, 2013). SMA-3, in particular, cooccupies some of the same DNA regulator elements as DAF-16 and jointly regulates the expression of select genes (Qi et al., 2017). Since DAF-16 is required for many DBL-1 functions including larval development (Kaplan et al., 2015), response to environmental nanopolystyrene (Liu et al., 2020), and lipid accumulation (Clark et al., , 2021Liu et al., 2020), we sought to determine whether DBL-1 signaling might play an important role in HSF-mediated age regulation.
Herein we report that the nonautonomous mode of age regulation by neural HSF-1 acts through this BMP/DBL-1 signaling pathway ( Figure 1a). Both the neuron-specific ligand, DBL-1, and its dedicated receptor in the intestine, SMA-6, were required for lifespan extension by neural HSF-1. We find that neural HSF-1 represses dbl-1 transcription thereby reducing BMP signal transduction in the intestine via a F I G U R E 1 BMP signaling across tissues is required for lifespan extension by neural hsf-1. (a) Schematic representing the BMP/DBL-1 signaling pathway in Caenorhabditis elegans. (b,c) Lifespan analysis of neural hsf-1 transgenic worms at 20°C in either the dbl-1(nk3), b, or sma-6(wk7), c, mutant background. See Table S1.  (Watterson et al., 2022). In turn, this reduces steadystate protein levels of the SMA-6 receptor and further dampens signal transduction. Our studies are consistent with another report that impaired DBL-1 signaling extends C. elegans lifespan (Luo et al., 2009). Moreover, the ability of HSF-1 to coordinate BMP/DBL-1 signaling across tissues resides upstream of enteric DAF-16. While previous reports have demonstrated that neural hsf-1 utilizes a distinct transcellular signaling mechanism involving chaperone activation to protect against acute heat stress (Douglas et al., 2015), thermal adaption by neural hsf-1 under mild yet chronic heat stress conditions at 25°C appears to utilize a similar DBL-1 dependent mechanism (Chauve et al., 2021). Consistent with these previous studies, we identify the neuron-specific BMP ligand, DBL-1, as a critical signaling molecule required for neural HSF-1 to extend lifespan and report how its transmission to the intestine modulates surface residency of the SMA-6 receptor through a SMAD negative feedback loop.

| Neural hsf-1 requires both the DBL-1 ligand and its intestinal SMA-6 receptor for lifespan extension
Mutations of dbl-1 and its dedicated type I transmembrane serine/ threonine kinase receptor, SMA-6, are involved in age modulation (Luo et al., 2009). Moreover, neural hsf-1 was recently shown to reduce DBL-1 signaling at the nonpermissive temperature of 25°C (Chauve et al., 2021). We hypothesized that this particular BMP signaling mechanism is required for lifespan extension by neural hsf-1.
Indeed, the dbl-1(nk3) null mutation abolished lifespan extension by neural hsf-1 (Figure 1a,b, Figure S1a, Table S1). While dbl-1 expression was necessary for lifespan extension by neural hsf-1, we next examined its sufficiency. Under our experimental conditions, elevating the expression or activity of dbl-1 through transgenic dbl-1::GFP overexpression (Schultz et al., 2014) or RNAi of its negative regulator, lon-2 (Gumienny et al., 2007;Taneja-Bageshwar & Gumienny, 2013) did not impact lifespan and does not appear to be sufficient to promote lifespan extension (Figure S1b-d and Table S1). Due to the requirement of this BMP signaling ligand to extend lifespan, we next examined the necessity of its dedicated type I receptor, SMA-6, in the peripheral tissues ( Figure 1a). Consistent with DBL-1 being required for lifespan extension, preventing receipt of this neural-born DBL-1 ligand via sma-6(wk7) mutation also abolished lifespan extension by neural hsf-1 (Figure 1c, Figure S1e, and Table S1). Previous studies have highlighted the importance of the intestine for neural hsf-1 to extend animal lifespan (Douglas et al., 2015). Rescuing sma-6::GFP expression exclusively in the intestine was sufficient for the neural hsf-1 to extend the lifespan in the sma-6(wk7) mutant background ( Figure S1f and Table S1). Thus, neural hsf-1 requires both the neural-born DBL-1 ligand and its dedicated BMP receptor, SMA-6, in the intestine to modulate the aging process. Yet how neural hsf-1 modulates this transcellular signaling paradigm was unclear.

| Neural HSF-1 represses expression of the BMP ligand, DBL-1
As a well-established regulator of worm body length, DBL-1 expression levels directly correlate with animal body length (Morita et al., 1999;Suzuki et al., 1999;Wang et al., 2002). Transgenic animals expressing hsf-1 in the nervous system demonstrated smaller body size throughout development and a 10% reduction in body length as egg-laying Day 1 adults ( Figure S2a,b). Since neural hsf-1 was reported to modulate DBL-1 activity (Chauve et al., 2021), we examined whether these morphological differences in body size correspond with dbl-1 expression levels. Consistent with this prior study (Chauve et al., 2021), analysis of transcriptomic datasets confirmed a 2-fold reduction in dbl-1 expression in neural hsf-1 transgenic animals ( Figure 2a). By contrast, transcript levels for a different TGFβ signaling ligand, DAF-7, were not changed in neural hsf-1 transgenic animals, further highlighting the selectivity of HSF-1 in the BMP branch of the greater TGFβ signaling pathway ( Figure S2c). Three predicted consensus heat shock elements were identified within the 5′ untranslated region of DBL-1, suggesting a direct mode of transcriptional regulation (Figure 2b). Analysis of previously reported ChIP-seq datasets (Labbadia & Morimoto, 2015) Table S2). Moreover, analysis of transcripts differentially regulated in neural hsf-1 animals identified TGFβ signaling (cel04350) as the 4th most significantly enriched KEGG pathway (Figure 2f). Consistent with HSF-1 and its dependence on the DAF-16/FOXO transcription factor for age regulation (Douglas et al., 2015;Hsu et al., 2003), both FOXO signaling (cel04068) and insulin resistance (cel04931) were significantly enriched in KEGG pathway analysis ( Figure 2f). Collectively, these data suggest that neural hsf-1 binds the DBL-1 promoter and mediates its transcriptional repression resulting in physiological defects regarding body length.

| Neural hsf-1 reduces peripheral DBL-1 signaling
Since HSF-1 represses expression of the neural-produced DBL-1 ligand, we examined how these transcriptional changes in the nervous system elicited a complementary response in the animal's peripheral tissues (Wang et al., 2002). Worm morphology suggested that neural hsf-1 initiated a periphery response as indicated by shortened body length of adults ( Figure S2a,b). Monitoring whole-body fluorescence from two distinct DBL-1 pathway transcriptional reporters, spp-9p::GFP (Roberts et al., 2010) and RAD/SMADp::GFP::NLS (Tian et al., 2010), enabled the spatial characterization and quantification of DBL-1 signaling to peripheral tissues. In keeping with spp-9 expression being inversely proportional to DBL-1 levels (Madhu et al., 2020;Roberts et al., 2010), we observed enhanced We then asked whether DAF-16/FOXO, which is required for lifespan extension by neural hsf-1 (Douglas et al., 2015), is also needed for DBL-1 signaling. Reduced expression of daf-16 via RNAi treatments did not impact spp-9p::GFP fluorescence in the animal's peripheral tissues (Figure 3b), suggesting it may act downstream of, or in parallel with, DBL-1 signaling to impact age regulation. In further support, activation of spp-9 transcription by neural hsf-1 still occurred in daf-16(mu86) null mutant animals ( Figure 3c). To examine the combined impact of FOXO and BMP signaling on lifespan, coadministering daf-16 and dbl-1 RNAi significantly reduced lifespan when compared to dbl-1 alone ( Figure 3d and Table S1), highlighting that both BMP and FOXO signaling might serve distinct yet complementary roles with respect to lifespan determination by neural hsf-1.

| Intestinal SMA-6 protein levels are reduced through repressed membrane trafficking
Canonical DBL-1/BMP signal transduction is composed of multiple transmembrane serine/threonine kinase receptors. Upon ligand binding, the type I BMP receptor, SMA-6, and its complementary type II transmembrane receptor, DAF-4, are recruited into a heterotetrameric complex at the plasma membrane which in turn activates F I G U R E 2 HSF-1 represses transcription of the neural-born BMP ligand, DBL-1. (a) Relative dbl-1 transcript abundance in the respective mutant and transgenic strains determined by RNAseq. Mean ± SEM, ***p = 0.0003 and ** p = 0.0034 by one-way ANOVA with the Dunnett's multiple comparison test. n = 3. (b) Schematic of predicted consensus heat shock element in the 5′ UTR of dbl-1. (c) Enrichment of HSF-1::GFP binding upstream of the DBL-1 transcriptional start sites (TSS) (2.5 kb) determined by previously reported HSF-1::GFP chromatin immunoprecipitation. Red asterisks highlight predicted Heat Shock Elements (HSE). (d, e) Relative dbl-1 transcript abundance after transient heat shock determined by RNAseq, d, and RT-qPCR, e. Mean ± SEM, *p = 0.0391 by twoway ANOVA with the Sidak multiple comparison test (reference Figure S2c) and ****p ˂ 0.0001 by the unpaired t-test. n = 2, d, n = 3, e. (f) KEGG pathway analysis from DAVID Bioinformatic's Resource v6.8. Plot displays the top 10 over-represented terms from genes significantly regulated in neural hsf-1 transgenic animals (p ˂ 0.05). Ubiquitin mediated proteolysis Metabolic pathways FoxO signaling pathway TGF-E signaling pathway Insulin resistance mTOR signaling pathway Dorso-ventral axis formation MAPK signaling pathway Endocytosis DNA replication Enrichment score -log 10 (FDR q-value) the intracellular receptor-regulated r-SMAD proteins, SMA-2 and SMA-3. SMA-2, SMA-3, and co-Smad SMA-4 then form a complex that enters the nucleus and regulates transcription of target genes (Gumienny & Savage-Dunn, 2013) (Figure 1a). While daf-4 transcript levels remained unchanged, we observed DBL-1-dependent transcriptional activation of sma-6 in neural hsf-1 transgenic worms ( Figure S4a). Conversely, the DAF-7 type I receptor, daf-1, which also heterodimerizes with DAF-4, is repressed in a DBL-1-dependent manner ( Figure S4a), suggesting a potential inverse relationship between signaling by DBL-1 versus DAF-7. However, further studies are required to elucidate the interplay between these different TGFβ signaling pathways.
Despite elevated sma-6 transcript abundance upon neural hsf-1 expression, we observed that steady-state protein levels of ectopically expressed SMA-6::GFP in the intestine were reduced by 79% at late L4 larval/Day 1 adult stages as indicated by western blot analysis ( Figure 4a and Figure S4b). Consistent with reducing dbl-1 signaling by neural hsf-1, dbl-1(nk3) mutants showed a similar 72% decrease in enteric SMA-6::GFP protein levels, while pathway activation via lon-2 RNAi did not significantly affect SMA-6::GFP levels ( Figure 4a and Figure S4b). We confirmed that reducing sma-6 expression via RNAi was sufficient to decrease BMP signal transduction in peripheral tissues as evidenced by increased fluorescence of the spp-9p::GFP reporter strain ( Figure S4c). Thus, reducing dbl-1 expression by neural hsf-1 or gene deletion compromises steady-state protein levels of its dedicated SMA-6 receptor, which corresponds with reduced BMP signaling.
Intrigued by the loss in intestinal SMA-6::GFP protein levels by neural hsf-1, we further investigated how reducing BMP signaling by neural hsf-1 might impact steady-state protein levels of the SMA-6 receptor. Several regulators of BMP/TGFβ signaling have the capacity to bind and modulate BMP/DBL-1 receptor dynamics, including trafficking to endosomes for recycling or degradation (Gumienny & Savage-Dunn, 2013). Thus, we examined whether reduced DBL-1 signaling by neural hsf-1 and signal ablation in dbl-1(nk3) mutants elicited a dose-dependent change in the extracellular DBL-1 pathway regulators LON-1, LON-2, SMA-10, CRM-1, DRAG-1, and ADT-2 (Gumienny & Savage-Dunn, 2013). Transcript abundance for lon-1, sma-10, and drag-1 was significantly changed in neural hsf-1 transgenic animals, but sma-10 was the only extracellular regulator whose transcriptional repression corresponded with dbl-1 levels in the cell ( Figure 4b). SMA-10, a LRIG (leucine rich and immunoglobulin domains) homolog physically binds SMA-6 and regulates its activity . Moreover, loss of sma-10 aberrantly affects endocytic trafficking and reduces signal strength of the SMA-6 receptor (Gleason et al., 2017). We hypothesized that reduced dbl-1 signaling by neural hsf-1 represses sma-10 expression in peripheral tissues thereby disrupting surface residency and steady-state expression of SMA-6 in endothelial cells, which ultimately reduces BMP signal transduction.

| SMA-3 mediates repression of membrane trafficking
It remained unclear how reducing DBL-1 signaling initiated transcriptional repression of these critical regulators of membrane trafficking and receptor dynamics. As intracellular signal transducers of DBL-1 signaling in the periphery, SMA-2 and SMA-3 represent the receptor-regulated Smads (R-Smads) while SMA-4 is the Co-Smad (Savage et al., 1996) (Figure 1a). These Smad proteins engage in dynamic nucleocytoplasmic translocation with nuclear retention being stimulated by pathway activation (Inman et al., 2002). We confirmed that sma-3 RNAi, and to a lesser extent sma-4 RNAi, reduced BMP  Figure 5b). Furthermore, we observe SMA-3-binding peaks within the first kilobase upstream of the translation start sites for rab-5, rab-7, and rab-11.1 (Figure 5b), all of which were repressed in neural hsf-1 and dbl-1(nk3) mutant backgrounds (Figure 4c and Figure S4d). Predicted binding sites for HSF-1, heat shock elements, were not identified in the promoter regions of these membrane trafficking genes (Table S3). Thus, their transcriptional fluctuations are likely due to DBL-1/BMP signal transduction in the peripheral tissues. Unlike the RAB GTPases that were repressed in neural hsf-1 and dbl-1(nk3) mutant backgrounds, the lipid surveillance responsive transcript, rab-11.2, was activated in the same worm strains without a detectable SMA-3-binding peak (Figure 5b). Overall, reducing receipt of the DBL-1 signal promotes a BMP negative feedback loop within the intestine. Our data suggest that this occurs in part through SMA-3 and its ability to repress critical factors involved in membrane trafficking and receptor dynamics, which ensure steadystate levels and cell surface residency of the SMA-6 receptor.
However, activation of lipid surveillance under these same conditions does not appear to be directly regulated by BMP signaling and SMA-3 but rather appears to be an indirect consequence of reduced RAB GTPase function and membrane trafficking.
We sought to understand whether BMP signaling acts to bridge HSF-1 activity in the nervous system with DAF-16 in the intestine. As previously mentioned, our data indicate that reducing BMP signaling by neural hsf-1 resides upstream of DAF-16 activity in the intestine (Figure 3b,c). Since DAF-16 complexes with SMA-3 to regulate the transcriptional output of target genes (Qi et al., 2017), we hypothesized that DAF-16 activity converges with SMA-3 to mediate transcriptional repression of membrane trafficking regulators. As a regulator of GTPase activity, the Ras homolog enriched in the brain, RHEB-1, was one gene whose promoter region was reported to be co-occupied and its expression mutually regulated by DAF-16 and SMA-3 (Qi et al., 2017). We confirmed SMA-3 binding within a kilobase of the untranslated promoter region upstream of the RHEB-1 translational start site ( Figure S5c) and this corresponded to reduced rheb-1 transcription in both neural hsf-1 and dbl-1(nk3) mutant strains ( Figure S5d). Thus, a previously characterized promoter mutually occupied by both DAF-16 and SMA-3 showed repression of its respective transcript upon reduction and ablation of dbl-1 signaling.
While we expect similar transcriptional signatures between reduced dbl-1 signaling by neural hsf-1 and signal ablation via dbl-1(nk3) mutants, we anticipate select differences, which likely account for their longevity effects. To investigate this, we compared a core set of 37 daf-16 transcriptional targets, which were compiled across several studies (Kumar et al., 2015;McElwee et al., 2003;Murphy et al., 2003;Riedel et al., 2013), and observed significant changes in 62% of these core transcripts in neural hsf-1 animals while dbl-1(nk3) null mutations modulated 54% of these core genes (Figure 5c,d).

| DISCUSS ION
Herein, we provide molecular details underlying a transcellular signaling paradigm involving the heat shock transcription factor, HSF-1, in the nervous system and its ability to modulate animal physiology and age progression across tissues through the TGFβ/BMP signaling pathway. Previous studies have identified that HSF1 activity in cancer-associated fibroblasts promotes malignancy through two stromal-derived signaling molecules; TGFβ and the stromal-derived factor, SDF1 (Scherz-Shouval et al., 2014). More recently in C. elegans, it was shown that hsf-1 overexpression in the nervous system can reduce BMP/DBL-1 signaling in peripheral tissues (Chauve et al., 2021). Our studies highlight BMP/DBL-1 signaling as a means by which neural hsf-1 extends animal lifespan and places this signal transduction mechanism upstream of the FOXO transcription factor, DAF-16. Specifically, we postulate that neural hsf-1 represses dbl-1 expression through binding to consensus heat shock elements within its 5′ promoter region. Decreasing neural expression of dbl-1 reduces BMP signal transduction in intestinal epithelia, in part, through loss of SMA-6 receptor availability at the cell surface, which occurs through SMA-3-mediated repression of sma-10 and several Rab GTPases. Previous studies have demonstrated that downstream nucleocytoplasmic SMAD proteins, SMA-3 and SMA-2, directly interact with DAF-16 (Qi et al., 2017) and provide a likely point of convergence between BMP signaling and DAF-16-mediated age regulation. While previous studies report a modest lifespan extension in dbl-1(nk3) mutants (Luo et al., 2009), removal of dbl-1 alone was not sufficient to extend lifespan in our experimental paradigm but was required for neural hsf-1 to extend lifespan. We hypothesize that in addition to reducing dbl-1 signaling, transcriptional fluctuations observed in neural hsf-1 and not in dbl-1(nk3) null mutants alter critical molecular pathways, which are needed in addition to reduced BMP/DBL-1 signaling to prolong aging. Knowing that DAF-16 is also required for lifespan extension by neural hsf-1 (Douglas et al., 2015) yet its overexpression has been reported to yield modest lifespan extensions (Henderson & Johnson, 2001), perhaps both reduced DBL-1 signaling and DAF-16 activation work synergistically to provide robust lifespan effects. However, additional studies are required to further understand how BMP/DBL-1 signaling impacts the molecular dynamics of FOXO within the nucleus.
BMP signaling plays an important role in growth and development and thus has implications in several human disease states (Wang et al., 2014). Yet reducing DBL-1/BMP activity in C. elegans and subsequently restricting growth through development possesses age-defying properties (Luo et al., 2009). This is reminiscent of another well-established age-extension paradigm in reduced insulin signaling (Kenyon et al., 1993), which also requires DAF-16 expression in the intestine (Libina et al., 2003;Uno et al., 2021).
(d) Venn diagram shows the overlap of genes whose expression levels are significantly different (p ˂ 0.05) between the respective RNAseq datasets (multiple unpaired t-tests). Genes denoted in white are transcriptionally activated in the neural hsf-1 and repressed in the dbl-1(nk3) strains.
restricting organismal growth and development earlier in life as observed in neural hsf-1 transgenics might enable these animals to reinvest their molecular resources and metabolic reserves later in life. Yet the same metabolic genes regulated by neural hsf-1 overexpression (Chauve et al., 2021) were also reported to be similarly modulated upon hsf-1 RNAi treatments (Brunquell et al., 2016).
Thus, further studies are required to understand this inexplicable link between HSF-1, BMP signaling, and lipid metabolism. While our studies focused on the gut-neuron axis, BMP signaling to other tissues yields alternate physiological consequences as demonstrated by BMP/DBL-1 signaling to the hypodermis, which is an important determinant in animal body length (Wang et al., 2002). Our studies show that rescuing SMA-6 expression exclusively in the intestine of the short-lived sma-6(wk7) enables neural hsf-1 to extend lifespan. This is consistent with the requirement of DAF-16 in the intestine for neural hsf-1 to extend animal lifespan (Douglas et al., 2015). Yet in both cases, a lifespan extension typical of neural hsf-1 was not fully achieved by the enteric rescue of either SMA-6 or DAF-16, suggesting that signaling between other tissues like the hypodermis or the germline, likely plays a complementary role in age regulation by the heat shock factor.
Our data suggest that HSF-1 in the nervous system communicates to DAF-16 in the intestine through BMP/DBL-1 signaling.
Previous studies have demonstrated that the homeobox domain protein, CEH-28, activates DBL-1 expression specifically in M4 neurons (Ramakrishnan et al., 2014). Our data provide additional regulatory information regarding the controlled expression of this DBL-1 ligand as HSF-1 appears to bind consensus heat shock elements in the 5′ promoter region of DBL-1. It remains unclear whether HSF-1 acts in these M4 neurons in conjunction with CEH-28 to modulate DBL-1 expression or whether it is acting in other parts of the nervous system to repress dbl-1 expression and promote longevity. Upon receipt of the DBL-1 ligand in peripheral tissues, signal transduction ultimately leads to the nuclear accumulation of SMA-2/SMA-3 (Gumienny & Savage-Dunn, 2013), which can co-occupy some of the same DNA regulator elements as DAF-16 (Qi et al., 2017). We observe the SMA-10 and several RAB GTPases are significantly repressed by neural hsf-1 and in dbl-1 mutants. Consistent with the removal of dbl-1 being sufficient to repress transcription of these genes, we observed SMA-3 binding to the promoter regions of these membrane trafficking genes.
Thus, under times of growth and development, DBL-1 would likely promote the expression of these genes to help establish polarity and maximize nutrient uptake through ensured cell surface residency of nutrient transport machinery. While neural hsf-1 reduces dbl-1 expression and extends animal lifespan, dbl-1 ablation does not extend animal lifespan in our laboratory, yet it still acts to repress transcription of sma-10 and several Rab GTPases, and reduce steady-state levels of intestinal SMA-6::GFP. Thus, further characterization of nuclear interactions between SMA-3 and DAF-16 might help identify additional factors that contribute to lifespan extension through reduced DBL-1 signaling.

| Caenorhabditis elegans lifespan analysis
Worms were age-synchronized using hypochlorite treatment of gravid animals for egg retrieval. Eggs were transferred onto RNAi plates seeded with EV control or different RNAi constructs. 100 μg/ ml 5-fluorouracil-2′-deoxyribose (FUdR) was supplemented to media plates on Day 1 of adulthood to avoid progeny that would disrupt the synchronization of the worm population. Each lifespan was set up to consist of 10 plates per condition, with at least 10 worms per plate, and at least 100 worms per condition. Plates were scored every day or every other day for "censored" or "dead" worms. Worms that were censored were those that were missing from the plate, bagged with progeny, exploded/ruptured, or desiccated. Whole plates were only censored in the event of fungal/bacterial contamination. Worms that showed no response to touch with a platinum wire at the head or tail were scored as dead. To alleviate transgenic silencing observed in neural hsf-1 transgenic animals (AGD1289), worms were thawed from frozen stocks and allowed to recover for 3-4 generations prior to lifespan analysis.

| Quantitative PCR
Quantitative reverse-transcriptase PCR (qPCR) was performed as previously described Egge et al. 2019. Worms were age-synchronized by hypochlorite treatment and grown on EV control or experimental RNAi at 20°C until the L4 stage of development. Total RNA was extracted by three rounds of freeze/thaw using TRIzol (ThermoFisher Scientific), followed by a chloroform and isopropanol precipitation extraction process. RNA pellets were rinsed twice with 75% ethanol, air-dried, and re-suspended in 20-50 μl molecular biology grade water. A DS-11 FX+ spectrometer (DeNovix) was used to measure RNA concentration including 260/280 and 260/230 ratios. Reverse transcription was con-

| Protein extraction and western blotting
All western blots were performed on late L4 larvae/day 1 adult worms. Animals have washed off the plates with liquid M9 buffer, centrifuged at 1000 g for 30 s, and washed twice with M9 before being transferred to 1.5 ml Eppendorf tubes and rapidly flash frozen in liquid nitrogen. Frozen worm pellets were thawed on ice and worm extracts were generated by glass/zirconia bead disruption in nondenaturing lysis buffer (50 mM Hepes pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton, EDTA-free mini-protease inhibitor cocktail [Roche]). Crude lysates were subject to centrifugation at 8000 g at 4°C for 5 min prior to protein determination with a BCA protein quantification kit (Thermo Scientific). Lysates were supplemented with 2× Laemmli sample buffer, boiled at 90°C for 10 min, resolved by SDS-PAGE, transferred to 0.22 μm nitrocellulose membranes, and subject to western blot analysis (Egge et al., 2019).
Electrophoresis was performed at a constant 100 V and protein size was compared with the prestained protein ladder (EZ-RUN, ThermoFisher).
All antibodies were prepared in 5% BSA/PBST. The membrane was incubated with primary antibody overnight at 4°C followed by the secondary antibody for 1 hour at room temperature. Rabbit antiα-Tubulin (Cat. No. ab4074, Abcam) primary antibody was used at 1:15,000, and secondary anti-IgG goat polyclonal antibody horseradish peroxidase (Cat. No. 20403-1ML, Biotium) was used at 1:5000. Both the rabbit anti-GFP antibody (ThermoFisher, A6455) and the anti-IgG goat polyclonal antibody horseradish peroxidase (Cat. No. 20403-1ML, Biotium) were used at 1:5000. Bands were visualized after incubation with Clarity Western ECL Substrate (Bio-Rad, 1705061) and detection on film was performed using a Konica Minolta SRX-101A film processor. Samples were run in triplicate and ratioed band intensities were averaged for each biological replicate.

| Microscopy
For representative fluorescence micrographs of transcriptional reporter strains, five to six worms per condition were aligned on NGM/ carb plates in a drop of M9 supplemented with 100 mM levamisole.
Imaging was performed on a Zeiss Axio Zoom.V16 set to 100× magnification. Images were acquired using transmitted light and standard filter settings for excitation and emission of fluorescence probes and recorded on a CCD camera (Zeiss AxioCam 503 mono). Zeiss ZEN software was used to control acquisition. Exposure settings and additional processing parameters remained consistent among samples in each experiment.
To examine SMA-6::GFP co-localization with FM4-64, approximately 100 days, 1 adult worms were rinsed off plates with M9 buffer, centrifuged (1000 g) for 30 s, and washed twice with M9 buffer. Worms were then incubated in 100 μl 0.4 mM FM4-64 (Fisher Scientific, T3166) for 2 h at 20°C in an Eppendorf Thermomixer C shaker set to 300 rpm. Following incubation, worms were washed three times with M9 and allowed to recover on NGM plates seeded with OP50 for approximately 30 min in the dark. Confocal micrographs were acquired using a Leica SP8 confocal microscope (Leica, Buffalo Grove, IL) and Leica Application Suite X (LAS X) software. Live worms were mounted in M9 supplemented with 100 mM levamisole and samples were imaged at 40× or 63× with oil immersion. All images were acquired in xyz acquisition mode at a speed of 600, and each line was imaged three times and averaged to reduce noise. Lasers 488 and/or 552 were used to excite fluorophores using hybrid detectors (HyD).
Laser power, range, and gain were adjusted according to strain/experiment but remained consistent between independent repeats.

| Large-particle flow cytometry
A COPAS FP-250 flow cytometer (Union Biometrica) was used to analyze age-synchronized worms using the attached sample cup or acquired from a 96-well plate by an LP Sampler (Union Biometrica).
Sample solution was comprised of M9 while the flow sheath solution contained a proprietary recipe, COPAS GP sheath reagent (PN: 300-5070-100, Union Biometrica). Flow data were collected and processed using the FlowPilot software (ver. 2.6.1, Union Biometrica).
Further data processing was performed in Excel (Microsoft) and statistical analysis in Prism 9 (Graphpad). Extinction was detected using the 488 nm laser line with a 1.3 ND filter and the gain of 1.0.
Propidium iodide, mCherry, and DsRed were excited using a 561 nm laser, while YFP and GFP were excited using a 488 nm laser. Gains for fluorescence detection were set to 2.0, and PMT voltage adjusted within the linear range of the instrument was consistent for each experiment (500 for YFP, mCherry, DsRed, and 450 for GFP). WormProfiler (an in-house program written for the MATLAB runtime) was used to standardize worm lengths and calculate fluorescence intensity profiles for individual worms within samples.
Unless otherwise noted, the integral of the fluorescence intensity for each worm within samples was averaged, and mean values were normalized according to time-of-flight (TOF) to account for variation in worm size across RNAi conditions. Mean integral or peak fluorescent values were standardized relative to empty vector control.
Statistical analysis was performed using CLC software (version 9.5, CLC Bio). Gene expression data were compared by the Baggerly's test and genes were considered significantly regulated when absolute fold change ≥1.5 and p-value ˂0.05. Deposited ChIP-seq datasets from the previous publication were analyzed using the CLC genomic workbench software v9.5. Raw data obtained from either Geo Datasets (GSE81523 for HSF-1::GFP ChIP-seq) or EnCode datasets (ENCSR992FVB for SMA-3::GFP ChIP-seq) were trimmed and mapped to the reference C. elegans genome (NCBI accession number NC_002944). With respect to HSF-1 binding within the DBL-1 promoter in a subset of neurons, genomic reads isolated in complex with HSF-1::GFP were divided by the total input reads for the same respective genomic region.
The CLC shape-based peak caller was used for all SMA-3::GFP ChIP. ChIP-enriched DNA was aligned onto input DNA; when the sequence coverage of a genomic region in the enriched DNA exceeded the input DNA, a ChIP peak score was called. ChIP peaks with their respective P values determined by CLC were generated for candidates whose transcript levels were differentially regulated as determined by RNAseq. To identify heat shock element consensus motif presence in the 5′ untranslated promoter region of the DBL-1 gene, we examined 2.5 kilobases upstream of the DBL-1 translation start site using the MEME suite (v4.11.1) (Bailey et al., 2015). Additional analysis for HSEs within the promoters of other genes was determined using TFBIND (https://tfbind.hgc. jp/). For each gene, 1000 bp upstream of the TSS (translational start site) was searched for sites with >90% sequence similarity of the HSF-1 consensus binding site, obtained from the transcription factor matrix ID (TRANSFAC R.3.4), on both the (+) and (−) strand.

| Statistical analysis
Statistical analyses, including t-test, mixed-effects, Chi-squared, and ANOVA with post hoc multiple comparisons analysis, were conducted using Prism (version 9.0, GraphPad) and statistical analysis for large transcriptomic datasets was performed using CLC genomics workbench (version 9.5). For large worm populations, outliers were removed using the ROUT method (Q = 1%) prior to analysis.

ACK N OWLED G M ENTS
We thank Venkat Malladi from the UT Southwestern Bioinformatics Core Facility, funded by the Cancer Prevention and Research Institute of Texas (CPRIT, RP150596), for binding site analysis. We appreciate both the Morimoto and Savage-Dunn lab for making ChIP-seq datasets publicly available. We thank the Caenorhabditis Genetics Center (CGC) and the Dillin, Grant, and Padgett laboratories for strains used in this study. We extend our gratitude to our funding sources: TWU institutional funds to BM and TLG and the American Federation of Aging Research, the Glenn Center for Aging, the Welch Foundation (I-2061-20210327), the NIH (R01AG076529 and R01AG061338 to PMD) and RO1GM097591 to TLG. Schematics created with biore nder.com.

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated and analyzed during this study are included in this article and its extended data are also available from the authors upon reasonable request. Transcriptomic data files that support the findings of this study in C. elegans will be deposited in the NCBI Gene Expression Omnibus (GEO) upon publication.