Summary
An organisms’ ability to adapt to heat can be key to its survival. Cells adapt to temperature shifts by adjusting lipid desaturation levels and the fluidity of membranes in a process that is thought to be controlled cell autonomously. We have discovered that subtle, step-wise increments in ambient temperature can lead to the conserved heat shock response being activated in head neurons of C. elegans. This response is exactly opposite to the expression of the lipid desaturase FAT-7 in the worm’s gut. We find that the over-expression of the master regulator of this response, Hsf-1, in head neurons, causes extensive fat remodeling to occur across tissues. These changes include a decrease in FAT-7 expression and a shift in the levels of unsaturated fatty acids in the plasma membrane. These shifts are in line with membrane fluidity requirements to survive in warmer temperatures. We have identified that the cGMP receptor, TAX-2/TAX-4, as well as TGF-β/BMP signaling, as key players in the transmission of neuronal stress to peripheral tissues. This is the first study to suggest that a thermostat-based mechanism can centrally coordinate membrane fluidity in response to warm temperatures across tissues in multicellular animals.
Introduction
The ability of an organism to adapt to environmental change can be key to its survival. The model organism C. elegans survives and reproduces optimally over an environmental temperature range of 12°C to 25°C (Sulston JE 1988). Temperatures beyond this range cause C. elegans severe physiological stress(van Oosten-Hawle and Morimoto 2014). When exposed to heat stress, C. elegans activates a highly conserved stress response, called the Heat Shock Response (HSR), during which the Heat Shock Factor 1 (HSF-1) transcription factor rapidly induces the expression of heat shock proteins (HSPs) (Lindquist 1986; Åkerfelt, Morimoto, and Sistonen 2010), which then clear and refold heat-damaged proteins (Morimoto 1997).
In addition to such cell autonomous responses, centrally controlled strategies also help to integrate external and internal cues across tissues to regain homeostasis. Rats, like other mammals, are endotherms and can control their body temperature. They have Transient receptor potential (TRP) channels in their skin and abdomen, which detect ambient temperature. This information is integrated in the hypothalamus, which serves as a thermostat (Madden and Morrison 2019). The efferent output from the hypothalamus provides negative feedback by controlling heat dissipation when it is warm, and heat conservation and thermogenesis when it is cold (Madden and Morrison 2019). Invertebrates, such as drosophila and C. elegans, are ectotherms and do not internally regulate temperature. However, they possess well-described thermostat-mediated escape responses that enable it to avoid noxious stimuli, for example, the heat-dependent activation of the bilateral thermosensory AFD neurons, which coordinate a thermotaxis response (Kimura et al. 2004; Hedgecock and Russell, 1975)
Although thermostat-based behaviours help ectotherms to escape noxious stimuli like heat, the longer-term survival of ectotherms when environmental temperatures change depends on their ability to remodel lipids within the plasma membrane, which is highly sensitive to external temperatures. Homeoviscous adaptation (HVA) is a mechanism that regulates the viscosity and permeability of membranes to ensure the robustness of biochemical reactions(Sinensky 1974; Cossins and Prosser 1978). In homeoviscous cold adaptation, membrane bilayers undergo a reversible change of state from a non-fluid (ordered) to a fluid (disordered) structure whereby the membrane’s phospholipids (PLs) fatty acyl (FA) chains become increasingly unsaturated (Mendoza 2014; Ernst, Ejsing, and Antonny 2016). In C. elegans, three Δ9-acyl desaturase enzymes, FAT-5, FAT-6 and FAT-7, de novo synthesize FAs, and in particular monounsaturated FAs (MUFAs)(Watts and Ristow 2017). These FAT enzymes are orthologs of human stearoyl-coA desaturases (SCDs) (Brock et al, 2006a). Of these three fat genes, fat-7 is upregulated at 16°C and is essential for maintaining the fluidity of membranes at cold temperatures (Murray et al. 2007; Savory et al, 2011a; Taylor et al., n.d.; Shi et al. 2013). In warm temperatures, the opposite occurs and fat-6/7 are negatively regulated (Lee et al. 2019a) concomitant with increased FA saturation in PLs (Tanaka et al. 1996; Murray et al, 2007). Indeed, mutants that abnormally upregulate fat-7 cannot survive at 25°C (Ma et al, 2015).
HVA was first discovered in single-celled organisms in which membrane sensors control membrane fluidity(Ernst et al, 2016). In C. elegans, two temperature-controlled cell autonomous sensors adjust FAT-7 expression: the transmembrane cool-sensor PAQR-2/AdipoR2 (Svensson et al. 2011; Svensk et al. 2013) and the heat-induced Acyl-CoA dehydrogenase, ACDH-11(Ma et al. 2015). PAQR2 reportedly also has a cell non-autonomous role in modulating HVA (Svensk et al. 2013) but no direct input from the brain has yet been linked to this adaptive response.
Increasing evidence indicates that across metazoans, neurotransmitters and neurohormonal signals modulate fat metabolism across tissues (Mattila and Hietakangas 2017; Dietrich and Horvath 2012; Cornejo et al. 2016; Srinivasan et al, 2015). In C. elegans, many of these pathways converge on the regulation of fat stores in the gut by controlling either the expression of lipases (Noble, et al 2013) or the Fat-7-dependent de novo synthesis of FAs (Yu et al. 2017; Clark et al. 2018; Horikawa and Sakamoto 2010) . Notably, ligands from the TGF-β/BMP signalling pathway provide a good example of this regulation; these ligands are secreted from neuronal cells and lead to peripheral fat remodeling in fat store cells(Yu et al. 2017). However, this well-described molecular pathway, which is also conserved in mammals (Tan et al. 2012), has not been studied in the context of HVA.
Here, we explore the antagonistic relationship between the Hsf-1-mediated activation of the HSR and the regulation of unsaturated lipids in C. elegans. Previous experiments have shown that while HSP levels increase in ambient temperatures of 25°C compared to 15°C, SCD enzymes show reduced expression at this temperature (Savory eta al 2011a; Lee et al. 2019a) . In addition, both the regulators of SCD enzymes-such as MDT-15- and SCD enzymes themselves are known to negatively regulate HSP expression at 15°C19(Savory et al 2011a) . However, it is not known whether Hsf-1 dependent cellular responses can modulate SCD expression and MUFA levels. Here, we show that hsp transcripts are expressed primarily in the head neurons of C. elegans at 20°C and show a clear temperature-sensitive expression pattern. We used the neuronal overexpression of HSF-1 (nhsf-1) as a tool to study the consequences of ectopically activating HSR in neurons at 20°C. We find that HSF-1 overexpression, via nhsf-1, in addition to its previously described role in controlling peripheral stress responses and longevity (Douglas et al. 2015), remodels lipid metabolism. This function is performed by decreasing the expression of fat-7 in the intestine, whilst activating the expression of catabolic lysosomal lipases. We identified the cGMP receptors TAX-2/TAX-4 and TGF-β/BMP signaling to be essential for the transmission of neuronal heat stress information to the intestine. We also show that nhsf-1 can turn on a similar fat remodeling program as that used by nematodes raised at 25°C, resulting in lower oleic acid levels in membrane phospholipids and a general decrease in FA saturation. This is the first study to report that ectotherms might use thermostat-based mechanisms to centrally coordinate complex adaptive responses to warming temperatures, as occurs in endotherms.
Main text
Fat desaturase expression negatively correlates with heat shock protein expression at increased temperatures
Hsp genes are expressed more highly at 25℃ relative to their expression levels at 20℃ or 16℃ in C. elegans (Gomez-Orte et al, 2018; Lee et al, 2009; Lee et al, 2019). Nevertheless, their overall expression levels are low even at 25℃, precluding the traditional in vivo visualization of multi-copy Hsp transgenic lines (Guisbert et al. 2013). To circumvent this technical issue, we took advantage of cGAL, a temperature-robust GAL4-UAS binary expression system, which is a more robust gene expression system than traditional transcriptional reporters (Wang et al. 2017). As shown in Figure 1A and B, When GAL4 is driven by an hsf-1 dependent promoter (hsp16.41) in C. elegans, GFP expression is primarily restricted to cells that have axonal projections in the head region. Surprisingly, GFP is only expressed in neuronal cells, and not in any other somatic tissues (Figure 1B), indicating that head neurons are exquisitely sensitive to small, step increments in temperature. Single-cell RNA sequencing confirmed that the only tissue with detectable hsp levels are head neurons (Taylor et al., n.d.). We observed that the number of GFP-positive neurons increased almost 16-fold from 15℃ to 25℃, when gfp was driven by hsp16.41(Figure 1C, Supp Table 1A). To rule out that this might be due to an indirect effect of temperature on the expression capacity of the GAL4-UAS binary system, we looked at the output of unc-47, a promoter that is restricted to GABAergic neurons but that does not have HSF-1 binding sites. As shown in Supp. Figure 1A and Table S1A, and as expected for a temperature-robust cGAL system (Wang et al. 2017), we observed that when gfp is driven by unc-47, there is no significant temperature-dependent shift in the expression of gfp. A positive interaction measured by a two-way ANOVA analysis indicates that the temperature-sensitive behavior of the hsf-1 responsive promoter is not driven by the effect of temperature on the binary expression system but reflects the activity of hsp16.41.
The striking temperature-dependent increase in hsp expression in anterior neurons is opposite to that of fat-7. SCD enzymes work by inserting a double bond into the 9th carbon of either palmitic acid (FAT-5) or stearic acid (FAT-6 and FAT-7) (Watts and Ristow, 2017). The FAT-5 desaturase is specific for the synthesis of palmitic acid (16:0), whereas the FAT-6 and FAT-7 desaturases, which share 86% homology at the nucleotide level, mainly desaturate stearic acid (18:0), producing a C18:1 MUFA called oleic acid (C18:1(9z), OA) (Brock et al 2006b; Watts and Browse, 2000) . Consistent with previous reports (Murray et al. 2007; Savory et al, 2011b; Gómez-Orte et al., 2018), we observe that there is a 38-40% decrease in transcript levels of fat-6 and fat-7 as the temperature increases from 15℃ to 20℃. The levels of fat-6 do not decrease further at 25℃ (Figure S1C), though fat-7 declines to 6% of its levels at 15℃ (Figure S1B, Supp. Table S1B). Together, these results indicate that the activation of the HSR in anterior neurons has an opposite expression pattern to fat-6/7 expression in the gut. These observations prompted us to study the potential relationship between neuronal HSR and the expression of enzymes that are known to modulate membrane fluidity.
Neuronal stress causes reduced oleic acid and fat stores
To explore the role of heat stress responses in neurons, we used transgenic C. elegans lines where hsf-1 expression is driven by the promoter rab-3, which is expressed only in neurons and depleted from other tissues (nhsf-1) (Douglas et al, 2015; Nonet et al, 1997). We used two integrated nhsf-1 lines (nhsf-1(1) and nhsf-1(2) that overexpress hsf-1 at different levels in head neurons (Figure S1D, Supp. Table S1C). Others have previously reported that these animals have increased longevity and heat stress resilience(Douglas et al. 2015) but in addition to these phenotypes, we observed a clear intestine, a phenotype usually associated with decreased fat stores (McKay et al. 2003).This observation indicates that the ectopic expression of hsf-1in neurons is potentially linked to fat metabolism in peripheral tissues. C.elegans does not have dedicated adipocytes but rather stores fats in organelles called lipid droplets (LDs) in the gut and hypodermis (Watts and Ristow 2017). Electron microscopy and lipidomic analyses have shown that the core of LDs is composed of triglycerides (TAGs), enclosed by a monolayer of phospholipids (PLs) and protein (Zhang et al. 2010; Vrablik and Watts 2013).
Lipid droplets (LDs) can be readily quantified using the fluorescent dye Bodipy (493/503). When Bodipy’s nonpolar structure binds to the neutral lipid components of LDs, it emits a green fluorescence signal with a narrow wavelength (Ashrafi et al, 2005). In young adult nematodes with increasing levels of hsf-1 in their neurons, we observed a dose-dependent reduction in Bodipy fluorescence of 30% in nhsf-1(1) and 40% in nhsf-1(2), relative to wild type animals (Figure 1 D-F, Supp.Table 1B). LDs are a vital energy source during periods of low food availability(Watts and Ristow 2017). A potential cause, therefore, for decreased LD levels in nhsf-1 worms is that the ectopic expression of nhsf-1 causes feeding to cease, resulting in the depletion of fat stores. However, the expression of mir-80, a micro-RNA that is upregulated in starved animals(Vora et al. 2013), is not different in nhsf-1 young adult animals compared to age-matched, wild type controls (Figure S1E, Table S1F). In addition, pharyngeal pumping rates, which when slowed can reduce or preclude a worm’s feeding ability, were also not significantly different in nhsf-1 nematodes relative to wild type, age-matched controls (Figure S1F, Supp. Table S1D). These observations indicate that LD depletion in nhsf-1 is not due to starvation.
One of the FA components of LDs are C18 MUFAs (Zhang et al. 2010; Vrablik and Watts 2013). Because the fat-6; fat-7 double mutant show a dramatic decrease in the levels of oleic acid (OA) and LDs (Shi et al, 2013), we assessed the expression levels of Δ9 desaturases in nhsf-1 worms. We found that an in vivo transcriptional fat-7 reporter (Walker et al. 2011) was reduced in a dose-dependent fashion in the two nhsf-1 lines by 28 % in nhsf-1(1) and by 38% in nhsf-1(2) relative to wild type (Figure 1G, H). At the transcript level, fat-6 is also decreased by almost 7-fold in nhsf-1 worms (Figure 1I Supp.Table 1E).
If the observed decrease in the transcriptional output of these enzymes in nhsf-1 worms was accompanied by reduced enzymatic output, then we would expect to observe changes in vaccenic acid and OA levels. To quantify FA composition in young adult nhsf-1 worms relative to age-matched, wild type controls, we used chromatography coupled to mass spectrometry (GC-MS) to measure levels of total FAs. This analysis showed that although the composition of most free FAs remained stable in nhsf-1 lines, the levels of C18:1 OA was significantly reduced by 27% with respect to wild type in the presence of neural hsf-1 over-expression (Figure 1J, Supp. Table 1D and Supp.Table 5J). These results indicate that ectopic neuronal stress causes the remodeling of LDs, at least in part by compromising the enzymatic output of the Δ9 SCD, FAT-7. It has previously been shown that the loss of a transcriptional activator of fat-7, MDT-15 and the loss of fat-7 itself, can downregulate hsps at 15°C (Lee et al. 2019b). To test if reduced fat-7 levels feedback to further augment stress in neurons at 20°C, we partially knocked down the function of fat-7and fat-6 using RNA interference (RNAi) and tested the effect on the output of UAS/GAL4 driven by the promoter of hsp16.41. As shown in Figure S1G-H, Supp. Table S1E, reduced Δ9 SCD levels did not change neuronal stress, indicating that above 15°C, Δ9 SCD activity functions only downstream of neuronal stress.
LDs are catabolized by lysosomal lipases, which then recycle the FAs that coat LDs back into the cytosol, where they can be broken down by β-oxidation or recycled to membranes (Watts and Ristow 2017). C. elegans contains at least five lysosomal lipases, LIPL-1 to LIPL-5 (O’Rourke and Ruvkun, 2013; Lapierre et al. 2011; Folick, et al. 2015; Ramachandran et al. 2019) which are activated by the acidic environment of the lysosome. Two fasting-induced lipases, LIPL-1 and LIPL-3, belong to the family of Adipocyte-Triglyceride Lipase (ATG)-like Patatin-domain containing lipases. These two lipases localize to the intestine, and their loss is accompanied by an increase in LDs and in overall body fat (O’Rourke and Ruvkun 2013). We observed that the transcripts encoding lipl-1, lipl-2, lipl-3 and lipl-5 are upregulated in nhsf-1 (Highlighted in Figure 1K and Figure S3E and Supp. Table 1F and S3D), indicating that their upregulation might contribute to the remodeling of LDs in nhsf-1.
2. Neuronal stress controls fat metabolism by dampening TGF-β/Sma/Mab signaling
Together our data indicate that neuronal stress remotely controls fat remodeling and that that the presence or absence of a molecular signal must allow two distant tissues to communicate. When performing an in-depth phenotypic characterization of nhsf-1, we observed two phenotypes that shed light upon the potential nature of such a signal. First, animals with an extrachromosomal array that drives neuronal hsf-1 (Ex-nhsf-1) expression were 20% smaller in size relative to age-matched wild type animals (Figure S2A, Supp. Table S2A). Second, nhsf-1(2) animals slowed down germline senescence (Luo et al. 2010). When nhsf-1(2) animals were cultured in the absence of males, they self-reproduced as expected, but produced significantly fewer progeny than wild types controls in the first three days of reproduction. However, they then continue to produce progeny at a significantly higher rate than wild type animals and for a longer period of time (Figure S2B-C, Supp. Table S2B), similarly to mutants that extend germline senescence (Luo et al. 2010)
Based on these phenotypes, we investigated whether nhsf-1 animals might phenocopy the loss of TGF-β/Sma/Mab signaling pathway for three reasons. First, the loss of DBL-1, the sole Sma/Mab pathway ligand in C. elegans, which is related to vertebrate Bone Morphogenetic Protein 10 (BMP) (Morita, Chow, and Ueno 1999),(Suzuki 1998), causes a small (SMA) phenotype. Second, TGF-β/Sma/Mab signalling regulates reproductive span, in parallel to the Insulin/IGF-1 signaling (IIS) and Dietary Restriction pathways (Luo et al. 2009; 2010). And third, mutations in the TGF-β/Sma/Mab pathway cause a lean phenotype, similar to that of the nhsf-1 phenotype. In addition, mutations in TGF-β/Sma/Mab pathway genes decrease the abundance of LDs (Clark et al. 2018; Yu et al. 2017) as well as the expression of the SCD desaturase genes, fat-6 and fat-7 (Yu et al. 2017; Taylor et al., n.d.; Luo et al. 2009)
To test the hypothesis that neuronal stress dampens BMP signaling, we took advantage of the RAD-SMAD reporter, in which multiple, high-affinity SMAD-binding sites are placed upstream of GFP. Others have previously shown that the RAD-SMAD reporter directly and positively responds to BMP signaling (Tian et al. 2010). As shown in Figure 2A, wild type worms grown at the standard temperature of 20°C expressed GFP in both hypodermal and intestinal cells during the L4.8 stage. However, in the presence of nhsf-1(2), the reporter’s GFP signal is visibly reduced at the same developmental stage (see Figure 2B, which shows reduced reporter expression in individual nhsf-1(2) worms). Two nuclei types showed a positive RAD-SMAD signal: small (< 8µm) and large (>8µm) nuclei, which most likely correspond to hypodermal and intestinal nuclei, respectively. As shown in Figure 2C, the RAD-SMAD signal decreases by 28% in small hypodermal nuclei in the presence of nhsf-1(2) compared to wild type levels (Supp. Table2A). A similar pattern was observed in intestinal nuclei, where there was a 53% decrease in the RAD-SMAD signal, relative to wild type levels (Figure 2D, Supp.Table 2A). These results indicate that nhsf-1 partially reduces the activity of the TGF-β/Sma/Mab pathway.
To further test for this potential interaction, we performed genetic epistasis experiments. We hypothesized that if nhsf-1 reduces TGF-β/Sma/Mab signaling, then in the absence of DBL-1, nhsf-1 should not be able to further reduce LD accumulation. To test for this, we used dbl-1(nk3), which carries a large deletion in the coding region of dbl-1, and monitored the accumulation of LDs in this mutant via BODIPY staining. As described elsewhere (Clark et al. 2018; Yu et al. 2017) this BMP mutant has 20% lower levels of LD accumulation, relative to wild type animals (Figure 2E, Supp.Table 2B). However, in the double dbl-1(nk3)/nhsf-1(2) mutant, LD accumulation levels are not significantly different relative to those observed in nhsf-1. In addition, and as described elsewhere (Luo et al. 2009; Yu et al. 2017) the loss of dbl-1 function causes a reduction in fat-7 expression, in line with our observations for nhsf-1 (Figure 1). Our findings reveal that relationship exists between nhsf-1 and dbl-1 with regards to fat-7 expression, because the double mutant is similar to the nhsf-1 single mutant (Figure 2 G-H). Together, these results suggest that the loss of LD accumulation in nhsf-1 is caused at least in part, by the loss of TGF-β/Sma/Mab signaling pathway activity. In further support of an epistatic relationship, we observed that the nhsf-1 transgene when crossed onto a dbl-1(nk3) genetic background did not cause the double mutants to be shorter than either single mutant (Figure S2D).
3. Neuronal TAX-2/TAX-4 expression is essential for the neuronal-stress fat-remodeling phenotype
To identify the neurons that transmit the stress-related signal that results in fat remodelling, we performed a suppressor RNAi screen in nhsf-1 animals. Because neurons tend to be refractive to RNAi, we combined nhsf-1 with a loss of function allele in rrf-3, which enhances RNAi sensitivity, including in neurons(Simmer et al. 2002) .Animals were fed with bacteria producing RNAi against seven genes that are required for sensory neuron function, including tax-4, ttx-3, cat-2, tax-6, tbh-1, tph-1, unc-31. Specifically, we screened for the dampening of fat-7p:GFP expression in RNAi-bacteria-fed nhsf-1 worms. Because fat-7 expression levels are highly sensitive to dietary variations, the screen was performed using an extra-chromosomal array, which carried the transgene rab-3P: hsf-1 (henceforth called Ex nhsf-1). In this experimental setup, due to the incomplete transmission of extrachromosomal arrays, siblings of different genotypes were grown side by side under identical conditions. The effect of RNAi treatment on fat-7p:GFP in siblings that inherited the array was compared to those that had not (Figure S3A-B, Supp. Table S3A). As expected, in animals fed with an empty vector (EV), demonstrated a significant 37% reduction in the transcriptional output of the fat-7p: GFP reporter in nhsf-1, compared to wild type siblings. We found that four of the RNAi treatments acts as suppressors of the fat-7 reduction in nhsf-1. However, two of these four were non-specific suppressors because they caused an increase in fat-7pGFP in control siblings (Supp. Table S3A), the exceptions being tax-6 and tax-4. As the normalized data shows, while tax-6 RNAi causes a mild suppression, tax-4 RNAi reverts fat-7 expression to almost wild type levels (Figure S3B). TAX-2 and TAX-4 are respectively, the α and β subunits of a hetero-oligomeric cyclic nucleotide-gated channel that is required for the proper functioning of several sensory neurons and that is also involved in chemosensation, thermotaxis and Dauer formation(Komatsu, Mori, and Rhee 1996; Coburn and Bargmann 1996).
Because RNAi causes only a partial knock-down of gene function and introduces technical variability, we further investigated these results using loss of function mutations. TAX-2/TAX-4 are expressed in 14 sensory neurons in the head, including in the AFD neuron (White, et al 1986; Coburn and Bargmann 1996; Komatsu et al,1996). AFD is the main thermosensory neuron, is required for thermotaxis (Satterlee et al, 2001; Wang et al, 2013; Hobert et al, 1997) and has been shown to key for the nhsf-1 modulation of the peripheral heat shock responses (Prahlad et al, 2008; Douglas et al. 2015). The DBL-1 ligand is also expressed in several neuronal cell types, including the ventral cord neurons and the AFD neuron (Morita et al, 1999). An obvious potential mechanism would be that neuronal stress directly turns off the expression of DBL-1 ligand in the thermosensory AFD neuron. However, it is unclear if DBL-1 is expressed in AFD as others have argued that it is instead expressed in AVA interneurons (Zhang and Zhang, 2012).
In our RNAi screen, we ruled out the involvement of ttx-3; ttx-3 encodes a LIM-homeodomain protein, which is expressed in the AIY neuron, and is required for thermotaxis (Hobert, 1997). To completely rule out the involvement of the AFD neuron in fat metabolism, we used genetic mutations in ttx-3 and in ttx-1, which encodes a homeodomain-containing protein that is necessary and partially sufficient for AFD development (Pereira et al. 2017). Our results indicate that nhsf-1 only partially suppresses the activity of a BMP signaling sensor (see Figure 2); we therefore reasoned that removing a potentially important source of DBL-1 (the AFD) should further increase the phenotypic penetrance of the nhsf-1 transgene with regards to fat-7 expression and body size. However, the tested ttx-1 (p767) and nhsf-1 (Figure 3A, C, Supp.Table 3B) showed a negative interaction in a two-way ANOVA test. These results indicate that the difference between wild type vs nhsf1 is similar to the difference between ttx-1 and ttx-1; nhsf-1. Likewise, a negative two-way ANOVA interaction was found for ttx-3 (ks5) mutants (Figure S3C, Supp. Table S3B). It nevertheless remains possible that if nhsf-1 eliminates dbl-1 expression only from the AFD. If this was the case, then genetically compromising the AFD function would not enhance the nhsf-1 phenotype. However, if this were correct, then eliminating the function of the AFD in a wild type background should alter LD accumulation. Our results suggest that this isn’t the case because none of these mutants on their own shows a significant change in fat-7 promoter-driven GFP expression relative to wild type (Fig 3 A and C, Fig S3 C, Supp.Table 3B, Supp. Table S3B), indicating that the AFD per se does not modulate fat remodeling in worms. These results also rule out any involvement of the thermosensory AFD neuron on the modulation of Δ9 SCD expression. They also indicate that the effect of nhsf-1 on stress resistance, which depends on AFD function (Douglas, et al 2015), is separable from the effect of nhsf-1 on body fat distribution.
To corroborate the suppressive effect of tax-4 RNAi, we used a loss of function allele, tax-4(p674) (Satterlee, et al 2014). As shown (Figure 3 A, B), the presence of nhsf-1 significantly decreases the expression of fat-7p: gfp by 32% compared to wild types. In the absence of tax-4, GFP expression was not significantly altered (Supp. Table 3A). However, the tax-4(p68) mutation partially suppressed the effect of nhsf-1 on fat-7 expression, increasing fat-7p: gfp expression to almost wild type levels (Figure 3A-B, Supp.Table 3A). A two-way ANOVA test shows a positive interaction, indicating that the difference between sibling pairs is higher for the control pair than for animals bearing the tax-4(p678) mutation. Together, these results indicate that eliminating tax-4 function partially suppresses the inter-tissue effect of nhsf-1 with respect to fat-7 expression in the gut. A further analysis indicated that tax-4 does not suppress the small phenotype of nhsf-1 (Figure S3D, Supp. Table S3CD), separating the regulation of body size by nhsf-1 from its regulation of fat metabolism.
In addition to effecting Δ9 SCDs, nhsf-1 also transcriptionally upregulates lipases, including lipl-1 by 11-fold (Figure 3D and Supp.Table 3C) and lipl-3 by 6-fold (Figure S3E and Supp. Table S3B). We therefore used tax-2(p671), a missense mutation in a conserved region of the predicted pore region of the α subunit of the cGMP channel (Coburn and Bargmann, 1996), to test for a potential interaction with lipases. Our findings show that the tax-2(p671) mutation suppressed LIPL levels; in the compound tax-2(p671); nhsf1 mutant, lipl-1 and lipl-3 expression levels returned to near wild-type levels. If transcriptional suppression of fat metabolizing enzymes is also accompanied by a change in FAs, then we would expect the elimination of cGMP function to suppress the nhsf-1 phenotype of depleted LDs. As observed before (Figure 3E-F, Supp.Table 3D), nhsf-1 caused a significant 63% reduction in LDs relative to wild type. tax-2(p671) did not significantly alter LD levels in wild type animals, but rescued LD levels in nhsf-1 (2). Together, our data shows that eliminating the function of TAX2/TAX4 can suppress the expression of enzymes that are required for LD homeostasis with downstream consequences for fat stores.
What might the underlying mechanism for this suppression be? It is possible that cGMP-gated channels help to detect or to transduce the HSF-1 dependent transcriptional response in neuronal cells. To test this hypothesis, we examined the ability of neurons to activate an HSF-1 dependent target, hsp16.41, in the absence of tax-2 function. As shown in Figure S3F-G, in tax-2(p671) animals that lack a functional cGMP channel, some unidentified neurons in the head region of GAL4/UAS animals can still activate hsp16.41. From this observation, we conclude that the cGMP receptor does not participate in detecting or transducing stress responses per se in neuronal cells. We support the view that HSF-1 activation in some of TAX-2/4-expressing neurons might directly or indirectly repress the TGF-β/Sma/Mab pathway. Loss of TAX-2/4 might disrupt the functioning of key neurons, disabling the ability of the HSR to influence TGF-β/Sma/Mab signaling.
TAX-2 and TAX-4 are expressed by a subset of 14 overlapping sensory neurons (White et al, 1986; Coburn et al, 1996; Komatsu et al, 1996). Tax-2(p694) is an allele with a deletion in the promoter region of tax-2 that disrupts its expression only in AQR, ASE, AFD, BAG PQR and URX neurons (Coburn and Bargmann 1996; Bretscher et al. 2011). We find that that this allele does not suppress the fat accumulation phenotype of nhsf-1 (Figure 3E-F, Supp.Table 3D). This indicates that none of these neurons is involved in modulating fat remodeling, and that all or possibly some of the remaining neurons that express the cGMP channel, including ASG, ASJ, ASK, AWB, ASI, AWC must be responsible for control of body fat in response to neuronal stress. Supp. Figure 4 summarizes a model that explains how the activation of HSF-1 in cGMP-expressing sensory neurons either directly or indirectly dampens the TGF-β/Sma/Mab signaling pathway, causing a decrease in fat desaturation and LD stores in peripheral tissues.
Overexpression of hsf-1 expression in neurons is sufficient to fine-tune membrane fluidity
At 25℃, fat-7 expression in the gut is very low, while the number of neurons mounting a stress response is upregulated (Figure 1 and S1). Because the ectopic expression of hsf-1 in neurons is sufficient to reduce fat-7 expression in the gut, we propose a model in which sensory neurons in the head, work as a thermostat that detects small increases in temperature to turn on a program that adjusts the fluidity of membranes to better adapt to warmer temperatures (illustrated in Figure S5). Others have already shown that nhsf-1 causes animals to be more resilient to heat (Douglas et al. 2015). One of the predictions that can be derived from this model is that nhsf-1 animals have a constitutively active thermostat. If this is correct, then at least some of the transcriptional responses that animals mount when raised at warmer temperatures, should be shared by nhsf-1 worms.
We identified differentially expressed (DE) genes between nhsf-1 (2) and wild types by DEseq2 (see materials and methods) and found 2,136 DE genes. Among the gene ontology (GO) groups that characterize the genes upregulated by more than one log2-fold change, include defense response and innate immunity genes, genes required for the formation of ribonuclear proteins, genes involved in reproduction, and genes involved in lysosome function. These data were compared to DE genes from published sources that compared the transcriptomes of animals grown at 15℃ with animals grown at 25℃, under a standard OP50 E coli diet (Gómez-Orte et al. 2018). To determine the probability of overlap between the two datasets, we employed a hypergeometric distribution using different filtering criteria (0, 0.05, 1, 1.5 or 2 log2-fold change, (Table 7). Using any of these, the overlap was found to be significantly different from an overlap expected by chance. The Venn diagram in Figure 4A, shows that the two sets of DE genes overlap by about 10%. Figure 4B shows the gene ontology (GO) categories of the overlapping genes. Among the GOs that are common to animals grown at 15℃ and that are down-regulated in nhsf-1 animals, are: translation control, ribosome formation, amino acid catabolism, and mitochondria. Interestingly, a common GO category that is upregulated both at 25℃ and in nhsf-1 animals includes genes required for lipid degradation and that are present in the lysosome (see Figure 4C), such as lipl-2 and lipl-5. In addition, the fat desaturase, Fat-7, is highly downregulated in animals raised at 25℃ (Figure 4C, blue) and in animals overexpressing nhsf-1 (Figure 1). Together, these data indicate that nhsf-1 might regulate a subset of genes that are required to regulate the synthesis and mobilization of MUFAs.
Lipid metabolism lies at the heart of HVA. To ensure membrane fluidity is maintained at warmer temperatures, PLs in the membranes have lower unsaturation levels in FA chains (Sinensky, et al 1974, Cossins and Prosser, 1978). If nhsf-1 animals were better suited to warmer temperatures, then we would expect the FA composition of glycerophospholipids to follow a similar rule. To test this idea, we performed liquid chromatography-mass spectrometry to quantify the levels of PLs that make up the bulk of the plasma membrane in animals raised at the standard 20℃ and their FA composition. Figure 4D shows a heat map of normalized values of different lipids, quantified in four different genotypes, including in the SCD double mutant fat-6 (tm33)1; fat-7 (wa36); in the BMP mutant, dbl-1(nk3); and in the two nhsf-1lines. We found that the quantity of phosphatidylcholine (PC), phosphatidylinositol (PI), and phosphatidylethanolamine (PE) were significantly changed in all mutants relative to wild type (Supp. Table 8), suggesting that the decrease in LDs is accompanied by an increase in specific membrane PLs.
To examine the acyl chain composition of the glycerophosphoplids by determining the number of double bonds for FAs in position sn-1 and sn-2. Figure 4E shows the specified number of double bonds (from 0 = saturated, to 9 = poly-unsaturated) for one example, specifically: PE 38 (phosphatidylethanolamine with an acyl chain of 38 carbons). All genotypes phenocopy each other in that the number of double bonds decreases with respect to wild type animals. Table 8 show other examples of FAs with a reduced number of double bonds. These results suggest that the membranes of nhsf-1 worms, and those of the other two mutants that have reduced levels of OA production (fat-6; fat7 and dbl-1), are consequently less fluid and can better adjust to warmer temperatures, even though these animals were grown at the permissive temperature of 20℃. Together, our results indicate that the over expression of hsf-1 in neurons coordinates an organismal response to decrease the production and storage of MUFAs. They show that nhsf-1 mutants phenocopy SCD and BMP mutants by decreasing fat stores and by remodeling the composition and fluidity of the plasma membrane.
Discussion
Our results show that the ectopic activation of the heat shock responses in neurons causes LDs to be depleted from fat store cells, and the widespread remodeling of the abundance and composition of membrane PLs, in a manner that would support HVA to high temperatures. We identify TAX2/4-dependent sensory neurons as being important for this response, and a ligand of the BMP family as being the key signal that must be decreased to communicate this response from the brain to the gut. We propose that neuronal stress in key TAX2/4 neurons acts as a thermosensor that enables heat adaptation at the organismal level by regulating peripheral membrane and protein homeostasis.
The biological thermostat is a basic model of thermoregulation that is used by many species Thermostats generally have three components: (1) afferent thermosensation; (2) central integration in the brain; and (3) an effector output that causes negative feedback. Mammals use thermosensing-TRP channels to sense heat (Madden and Morrison, 2019) however, but this TRP subtype is not present in the C. elegans genome (Xiao and Xu, 2009). Nevertheless, C.elegans does have thermo-sensing neurons, such as the AFD and the ASJ neurons, in which heat activates a cGMP pathway that opens the TAX-2/TAX-4 ion channel to increase calcium currents. However, the primary mechanism by which temperature is sensed by TAX-2/TAX-4 expressing neurons remains elusive. Here, we report an alternative potential thermostat that involves the activation of the heat shock response in TAX-2/TAX-4 expressing neurons. The TAX-2/TAX-4 cGMP receptor is expressed in 14 sensory neurons. Our analysis of the allele of tax-2(p694) indicates that the neurons that are potentially important in mediating the effects of neuronal stress on fat metabolism include: ASG, ASJ, ASK, AWB, AWC, and ASI. Of these neurons, the light and pheromone-sensing neuron ASJ is a candidate of particular interest as it is heat-activated (Ohta et al. 2014) (Zhang et al. 2018) and suppresses the protective effects of cold acclimation to cold resistance(Ohta et al. 2014). Indeed, ASJ shortens lifespan at 15°C by inhibiting the metabolic transcription factor DAF-16/FOXO in the intestine, which is also known to regulate fat-7 activity (Zhang et al. 2018). It would be interesting to determine whether suppressing TAX-2/TAX-4 exclusively in the ASJ could rescue fat-7 expression levels.
In addition to optimizing the fluidity of membranes in warmer temperatures, previous work has shown that nhsf-1 animals are better prepared for surviving heat stress as they can potentiate the expression of HSPs across tissues (Douglas et al. 2015). Our findings show that both responses are separable because while loss of the thermosensory circuit disrupts nhsf-1-dependent thermotolerance, but it does not alter the nhsf-1 lean phenotype. Together, these experiments suggest that higher temperatures might turn-on a thermosensor in wild type animals, to coordinate multiple adaptive responses that help animals thrive in a warmer environment.
A cell autonomous sensor, the Acyl-coA dehydrogenase acdh-11, has previously been reported to be upregulated on heat acclimation (at 25°C) and to downregulate cold-responsive fat-7 by sequestering C11/C12 fatty acids that would otherwise bind to and activate NHR-49, a transcriptional regulator of fat-7 (Ma et al. 2015). Our RNAseq analysis does not reveal a change in the expression of acdh-11 in nhsf-1 animals, so it is possible that both systems act independently. Future work should aim to untangle how these two systems interact to ensure HVA.
These findings can also inform our understanding of heat adaptive mechanisms in vertebrates. Recent mammalian studies suggest that FA desaturation patterns correlate with latitude in mammals that inhabit different environments, a trend that is particularly clear for aquatic and semi-aquatic mammals (Guerrero and Rogers, 2019). Therefore, understanding the regulation of membrane fluidity is relevant to understanding how thermoregulation occurs in some endotherms. The control of SCDs is also increasingly recognized as relevant to human pandemics, such as obesity and metabolic syndrome (Sampath and Ntambi, 2011). It is possible that some of the neuronal circuits and molecular players we describe here have been co-opted to modulate fat metabolism in mammals with BMP signaling being a good example of this. And finally, the ER stress response is highly relevant in mammals. Specifically, ER stress in Pomc hypothalamic neurons improves glucose utilization and insulin sensitivity protecting mammals from diet induced obesity (Williams et al. 2014). It is not known if the activation of HSR in neurons can also have systemic consequences for health in mammals, but this would be a worthwhile avenue to explore.
Funding
This work was supported by ERC 638426 and BBSRC [BBS/E/B000C0426].
1-General methods
C. elegans maintenance
Nematodes were grown on NGM plates seeded with Escherichia coli OP50 strain at 20°C unless otherwise stated, according to standard methods (Brenner 1974).
C. elegans strains
The full list of C. elegans strains used and generated for the purpose of this study can be found in supplemental table A. We noticed that the strain AGD1289 would sometimes lose its clear intestine phenotype when grown for more than a month, so we used freshly thawed AGD1289 worms for less than a month.
Worm synchronization
Worms were age-synchronized either by egg-lay in a 2h period or by treatment with alkaline hypochlorite solution, according to standard procedures (Stiernagle 2005) for experiments that required large amounts of synchronized animals (such as the Bodipy staining, RNA-seq and lipidomics experiments). For bulk qRT-PCR experiments and for microscopy-based quantification of fluorescent reporters at L4.8 or L4.9 stages, we used worms synchronized by egg-laying grown in parallel at different temperatures. Worms were harvested 97h after egg-laying at 16°C, 42h after egg-laying at 25°C, and L4.8 or L4.9 worms were picked from a mixed population grown at 20°C. The L4 sub-stages were assessed according to the morphology of the vulva, as described in Mok et al, 2015.
Experimental design
Each experiment was performed in at least three biological triplicates. In each biological replicate, each condition comprised at least 30 individual animals in microscopy experiments (bodipy quantification, fluorescent reporter quantification), 15-20 animals in confocal experiments (RAD-SMAD reporter quantification), 15-30 animals in qRT-PCR experiments, 10 animals for pharyngeal assay, and 20 animals for fertility assay.
Pharyngeal pumping assay
Synchronized young adult animals from each genotype were singled out and assayed for pharyngeal pumping at the young adult stage. Experiments were done in triplicate with at least 10 worms per condition. Pharyngeal pumping movements were followed for 30s, under a dissection microscope, and each animal was scored at least twice.
Fertility assay
About 20 worms per condition were singled out in 12 well plates seeded with 50µL OP50 at the L4 stage. Each day, all animals were passed onto new 12 well plates. The F1 progeny laid by each individual worm was scored 2-3 days after the P0 parent had been transferred to the well, when the F1s were either L4 or adults. For ease of scoring, the 12 well plate was left on ice for a few minutes, until the animals were immobilized.
DNA lysate preparation and PCR genotyping
Between 10 to100 worms were picked into 10 µL of Worm Lysis Buffer (50mM KCl, 10mM Tris (pH 8.3), 2.5mM MgCl2, 0.45% NP40, 0.45% Tween-20, 0.01% Gelatin). Tubes were freeze-thawed once before 1 µL 01mg/mL of proteinase K was added to each tube. Worms were lysed and their genomic DNA was released by heating tubes to 65°C for 60-90 minutes. Proteinase K was inactivated by heating to 95°C for 15 min. Commonly, 1 µL DNA lysate was added to each PCR reaction. All PCR genotyping reactions were performed with Taq DNA polymerase (New England Biolabs #M0267L) with Thermopol buffer according to manufacturer instructions. The list of PCR primers used for genotyping can be found in supplemental table B.
RNAi
The RNAi suppressor screen was designed to look for increase of fat-7p::GFP levels in animals carrying an overexpression of hsf-1 in neurons. As neurons tend to be refractive to RNAi, the screen was performed in the MOC201 strain carrying a loss of function allele in rrf-3(pk1426), which enhances RNAi sensitivity, including in neurons (Simmer et al. 2002). Because fat-7 expression levels are highly sensitive to dietary variations, the screen was performed using an extra-chromosomal array, which carried the transgene rab-3P: hsf-1 (henceforth called Ex nhsf-1). In this experimental setup, due to the incomplete transmission of extrachromosomal arrays, siblings of different genotypes were grown side by side under identical conditions. The effect of RNAi treatment on fat-7p:GFP in siblings that inherited the array was compared to those that had not. RNAi-mediated knock down of candidate genes involved in neuronal functioning was performed using clones from Dr Julie Ahringer’s RNAi library, including tax-4, ttx-3, cat-2, tax-6, tbh-1, tph-1, unc-31. Bacterial cultures were grown overnight in LB with 100 µg/mL Carbenicillin and induced with 1 mM IPTG for 2 h. RNAi seeded plates were left for 48h to dry at room temperature. Animals were initially grown on OP50 plates and were individually transferred at the L4.8 stage (Mok et al, 2015) onto RNAi plates. About 50 extrachromosomal overexpressing neuronal hsf-1 worms and non extra-chromosomal worms from the same background were transferred onto the same RNAi plate. Animals were transferred onto fresh RNAi plates at day 2 of adulthood. At day 3 of adulthood, the animals were mounted, and fat-7p::GFP fluorescence was monitored. We tried to image about 40 extrachromosomal carrying neuronal hsf-1 overexpression array, and 40 non extra-chromosomal siblings. At least 3 biological replicates of each experiment were performed.
qRT-PCR on bulk worm samples
To monitor steady-state mRNA levels on bulk samples of worms, we used the Power SYBR® Green Cells-to-CT™ kit and hand-picked a pool of about 15-25 animals in 10 µL Lysis buffer. Reverse transcription was performed using the 2X RT buffer from the Power SYBR® Green Cells-to-CT™ kit according to manufacturer instructions and cDNA was diluted either 1:4 or 1:5. Each qRT-PCR reaction contained 1.5 µL of primer mix forward and reverse at 1.6 µM each, 3.5 µL of nuclease free water, 6 µL of 2X Platinum® SYBR® Green qPCR Supermix-UDG with ROX (ref 11744-500) and 1 µL of diluted cDNA. The list of PCR primers used for qRT-PCR can be found in supplemental table B. The PCR efficiency was calculated for each couple of primer by running a standard curve on a dilution series. Validated couples of primers had a PCR efficiency between 90 and 113% with R2>0.97 (supplemental table B).
2-Lipidomics
Worm harvesting for lipidomics
Worms of every genotype were synchronized using hypochlorite treatment according to standard procedures (Stiegerale 2005). Young adult animals were grown at 20°C and harvested at 50-52h post L1 plating for N2, MOC141 and NU3, at 54h for AGD1289, and at 70h for BX156, as they were developmentally delayed. One 9 cm NGM plate containing either 1000 young adults for total fatty acids lipidomics, or 500 worms for phospholipids analysis was harvested for each genotype. Worms were washed with 15 mL of M9 buffer at least 3 times. Most of the supernatant was removed and the pellets with collected worms were frozen at −80°C in low-protein binding Eppendorf tubes, before being processed for lipidomics analysis.
GC-MS Analysis of Fatty acids
16-35 mg of C. elegans were extracted by adding 1 mL chloroform/methanol (2:1, v:v, containing 0.01 % BHT and 10 µg of C9:0 and C13:0, respectively as internal standard) and processed using an ultrasound sonotrode for 30 sec at 40 Hz (type UW 2070, Bandelin). Afterwards, 0.5 mL water was added to each tube and each sample was shaken vigorously for 1 min. Next, the extract was centrifuged for 5 min at 3000 rpm. The chloroform layer was transferred into a new vial and the solvent was removed with a gentle stream of nitrogen. The residue was resolved in 200 µl tetrahydrofuran. 400 µl methanolic base 0.5 M (Acros Organics) was added. After 1 min of rigorous shaking, the sample was heated for 15 min at 80°C. Afterward, 200 µL water and 200 µl hexane was added. After another minute of vigorous shaking, the sample was centrifuged for 1 min at 1500 rpm. The hexane layer was transferred into a new vial. 1 µL sample was injected into the gas chromatograph coupled to the mass spectrometer (GC-MS 40, Shimadzu) with a split of 5 and injection temperature of 260°C. The separation was performed with a Zebron ZB-5MSplus column (30 m x 0.25 mm x 0.25 µm) (Phenomenex) with helium as carrier gas: 35°C was held for 2 min. Then the temperature was raised by 10°C per minute to 140°C, which was held for 10 min. Afterwards, the temperature was raised to 240°C at a rate of 2°C per minute. 240°C was held for 10.5 min. Quantification was performed using the GCMSSolution Version 4.30 (Shimadzu).
Phospholipids analysis
The worms were homogenized using a Precellys evolution with a cryolys unit to keep the sample frozen during homogenization (Bertin Technologies, France). Precellys bead-beating tubes with reinforced walls for hard tissue (CRK28-R) were used with 3 cycles of 7200 rpm at 45 seconds. The homogenized sample was then transferred to a glass tube containing Chloroform/Methanol for lipid extraction using Folch method (Folch, J. et. al. J. Biol Chem 1957). Phospholipids were separated using a Cogent HPLC column (150 × 2.1 mm, 4 µm particle size) placed on a Shimadzu XR (Shimadzu, Kyoto, Japan) using the conditions described in Zhuang, X. et. al. The phospholipids were then detected using an Orbitrap Elite mass spectrometer in full scan mode with a mass range of 200-1000 m/z at a target resolution of 240,000 (FWHM at m/z 400). Data were analysed using Lipid Data Analyzer (2.6.0–2) software (Hartler, J. et.at Nat. Methods 2017).
3-Fat content analysis using bodipy
Bodipy protocol
We used bodipy 493/503 (Thermo Fisher Scientific D3922) to stain neutral lipids in fixed animals. We adapted the protocol developed by Klapper and colleagues (Klaper et al. J Lipid Res 2011) to fix worms using 60% isopropanol. About 1000 worms per genotype were synchronized by hypochlorite treatment (Stiernagle, 2005). Animals were collected and washed at least 3 times in M9 buffer. Worms were then fixed for 5 minutes in 60% isopropanol in 1.5 mL protein-low bind Eppendorf tubes, with occasional inversion of the tubes. As fixation with isopropanol was sometimes variable, we also tried fixation with cold methanol for 10 minutes, the rest of the protocol remaining identical After fixation, we let the worms settle by gravity and washed them once more with M9. The M9 supernatant was removed, leaving approximately 50 µL. The tubes were frozen and thawed twice and then incubated with 500 µL of bodipy 493/503 (diluted in M9 at 1µg/mL) at room temperature for 1h on a rotator. After 1h, the worms were washed twice with 1 mL M9 solution containing 0.01% triton. We kept the samples at 4°C and imaged them either the same day or the next day but not more than 2-3 days after collection.
Mounting of fixed animals
For mounting, we used a mouth pipette with a glass capillary to remove all liquid in the Eppendorf tube. About 8-10 µL of Vectashield® antifade mounting media without DAPI (Vector Laboratory 94010) was added and worms were mounted for imaging on a 2% agarose pad using a 18×18 mm glass coverslip.
4-Microscopy
All worms imaged were mounted on a 2% agarose pad. For imaging of live worms, animals were paralyzed in 3mM Levamisole diluted in M9. Fluorescence exposure was identical across all conditions of the same experiment. To image intestinal levels of fat-7p::GFP fluorescence in live worms and BODIPY 493/503 fluorescence in fixed worms, we used a Nikon Ti Eclipse fluorescent microscope at objective 20X. For imaging GFP-positive neurons in PS7171 and PS7167, we used a Nikon fluorescent stereomicroscope SMZ18, as it was easier to capture neurons in 2D from live animals. PS7171 and PS7167 were synchronized at L4.8 stage (Mok at al., 2015) in this experiment. LW2436 and MOC229 worms were imaged on a Nikon A1R confocal microscope at 20X objective to determine the fluorescence levels of the RAD-SMAD reporter. The Z-stacks taken were then processed by deconvolution, and stitched together.
Image Analysis
In order to straighten and quantify fluorescence from acquired C. elegans images, we used the FIJI/ImageJ worklow called “Worm-align” that allows to generate single-or multi-channel montage images of aligned worms from selected animals in the raw image. The output of “Worm-align” was then imported and run through a CellProfiler pipeline called “Worm_CP” for fluorescence quantification of the animals of interest selected with “Worm-align”. Both “Worm-align” and “Worm_CP” pipelines are available and described in Okkenhaug et al. JoVE, 2019, in revision.
5-RNA sequencing
RNA-sequencing library preparation
RNA was extracted by standard Trizol extraction techniques. Libraries were made using either NEBNext mRNA Second Strand Synthesis Module (E6111) followed by NEBNext Ultra II DNA Library Prep Kit for Illumina (NEB-E7645) or NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (E7760) with the NEBNext Poly(A) mRNA Magnetic Isolation Module (NE7490) as per manufacturers protocols using half reactions. 13 Cycles of amplification was used for library enrichment; quality and size distribution of the the libraries was ascertained by running on a Bioanalyzer High Sensitivity DNA Chip (Agilent 5067-4626) and concentration was determined using KAPA Library Quantification Kit (KK4824). Libraries were sequenced on an Illumina HiSeq 2500 system by the Babraham Sequencing Facility.
RNA-sequencing analysis
The FASTQ files were quality trimmed with Trim Galore v0.4.4 (https://github.com/FelixKrueger/TrimGalore), used in conjunction with Cutadapt v1.15 (DOI:10.14806/ej.17.1.200), and then mapped to the C. elegans reference genome WBcel235 using HISAT2 v2.1.0 (Kim et al), in either single- or paired-end mode depending on the sequencing protocol followed. To improve the mapping efficiency across splice junctions, HISAT2 took as additional input the list of WBcel235.75 known splice sites. RNA-seq QC and analysis was performed on the mapped reads using the genome browser SeqMonk v1.45.4 (https://www.bioinformatics.babraham.ac.uk/projects/seqmonk/). To identify differentially expressed genes, the number of reads positioned over exons were first calculated for every gene using the program’s “RNA-seq Quantitation Pipeline”. Differentially expressed genes were subsequently called by DESeq2 v1.22.2 (Love et al.,), launched from SeqMonk with default settings. Differentially expressed genes were defined by having an adjusted P value cutoff <0.05 after multiple-testing correction. The gene expression Principal component analysis when comparing nhsf-1 and wild type showed that the transcriptomes formed clusters according to their molecular subtypes, indicating high quality and consistent homogeneity of transcriptomes.
Acknowledgements
We would like to acknowledge Boo Virk, for helping with experiments and editing figures; Cindy Voisine, Michael Witting, Jon Houseley, Len Stephens, Rebeca Aldunate and Jane Alfred for feedback on the manuscript. We are also thankful to Andy Dillin and Amy Walker for reagents; Mario de Bono and Julie Ahringer for helpful discussions. We also acknowledge the help of Babraham Institute Facilities and CGC for providing strains.