Temporal pattern of neuronal insulin release during Caenorhabditis elegans aging: Role of redox homeostasis

Abstract The insulin‐IGF‐1/DAF‐2 pathway has a central role in the determination of aging and longevity in Caenorhabditis elegans and other organisms. In this paper, we measured neuronal insulin secretion (using INS‐22::Venus) during C. elegans lifespan and monitored how this secretion is modified by redox homeostasis. We showed that INS‐22::Venus secretion fluctuates during the organism lifetime reaching maximum levels in the active reproductive stage. We also demonstrate that long‐lived daf‐2 insulin receptor mutants show remarkable low levels of INS‐22::Venus secretion. In contrast, we found that short‐lived mutant worms that lack the oxidation repair enzyme MSRA‐1 show increased levels of INS‐22::Venus secretion, specifically during the reproductive stage. MSRA‐1 is a target of the insulin‐IGF‐1/DAF‐2 pathway, and the expression of this antioxidant enzyme exclusively in the nervous system rescues the mutant insulin release phenotype and longevity. The msra‐1 mutant phenotype can also be reverted by antioxidant treatment during the active reproductive stage. We showed for the first time that there is a pattern of neuronal insulin release with a noticeable increment during the peak of reproduction. Our results suggest that redox homeostasis can modulate longevity through the regulation of insulin secretion, and that the insulin‐IGF‐1/DAF‐2 pathway could be regulated, at least in part, by a feedback loop. These findings highlight the importance of timing for therapeutic interventions aimed at improving health span.


Aging research in Caenorhabditis elegans and other organisms has
shown that lifespan is genetically and environmentally determined.
The C. elegans genome encodes a single insulin/IGF-1-like receptor (DAF-2); however, it carries several genes that encode for insulin-like peptides (ILPs). Forty members of the insulin family have been found through genetic and bioinformatic analyses (Li & Kim, 2008), and several ILPs were shown to regulate longevity and developmental processes (Fernandes de Abreu et al., 2014). Many ILPs show neuronal expression or are expressed in specific subsets of neurons, while a few are expressed in the intestine. Both types of peptides regulate longevity through DAF-16 (Li & Kim, 2008;Murphy, Lee, & Kenyon, 2007). Evidence gathered in C. elegans indicates that DAF-16 influences lifespan cell non-autonomously by regulating the insulin pathway in several tissues (Libina, Berman, & Kenyon, 2003). The strongest evidence showing the role of this pathway in neurons comes from experiments in which the expression of the DAF-2 receptor exclusively in the nervous system is sufficient to abolish lifespan extension of daf-2 mutants (Dillin, Crawford, & Kenyon, 2002;Wolkow, Kimura, Lee, & Ruvkun, 2000). In mammals, there is also evidence of the importance of the insulin/IGF signaling in the central nervous system and its relationship with aging (Broughton & Partridge, 2009). Irs2 (insulin receptor substrate 2) knockout mice are diabetic and as a consequence they have a shorter lifespan (Selman, Partridge, & Withers, 2011); however, if the deletion of the Irs2 gene is brain-specific, the mice are long lived even though they have a diabetic phenotype (Taguchi, Wartschow, & White, 2007).
It is possible that insulin from the nervous system is not only transcriptionally regulated (Berendzen et al., 2016;Libina et al., 2003;Murphy et al., 2007) but it may also be controlled at the secretion level throughout the animal's life. Does insulin release from neurons remain constant during C. elegans lifespan or does it change over time? Does the pattern of insulin release influence the aging process? Current research has focused on identifying genes that regulate secretion of insulin/IGFs from neurons in C. elegans. However, there are no studies on whether there is a temporal course of insulin release during C. elegans lifespan. Some genes have been described to increase insulin release when mutated, such as goa-1 (a subunit of a trimeric G protein) and tom-1 (a syntaxin binding protein). Interestingly, these mutants also show a reduction in the animals lifespan (Ch'ng, Sieburth, & Kaplan, 2008). Others, such as gon-1, decrease the release of some insulin peptides (ins-7 and daf-28) (Yoshina & Mitani, 2015). Consequently, mutations that reduce insulin secretion cause an increase in the animal's lifespan (Yoshina & Mitani, 2015), while mutations that exacerbate insulin secretion decrease lifespan (Ch'ng et al., 2008).
Given the importance of the insulin-IGF-1/DAF-2 pathway in the determination of lifespan and the aging process across species, we focused our studies on characterizing neuronal insulin secretion during C. elegans lifetime to find out whether there is a specific temporal pattern of secretion. We also analyzed if this pattern could be modulated by the insulin-IGF-1/DAF-2 pathway itself.
Additionally, we evaluated if any of the DAF-16 target genes could modulate the pathway at the level of insulin secretion from neurons. As mentioned before, DAF-16 induces in part the expression of the cell´s antioxidant machinery (Murphy, 2006;Sun, Chen, & Wang, 2017). Our previous work shows that the DAF-16 upregulated target MSRA-1, an oxidation repair enzyme ortholog of the Drosophila and human MsrA genes, is necessary to maintain wildtype (Wt) lifespan (Lee et al., 2005;Minniti et al., 2009). Unlike other antioxidant enzymes such as SOD-1 and SOD-3 (Doonan et al., 2008;Van Raamsdonk & Hekimi, 2012), the absence of the single C. elegans MsrA gene (msra-1) causes a 30% decrease in lifespan (Minniti et al., 2009). This function is conserved from yeast to rodents (Chung et al., 2010;Koc, Gasch, Rutherford, Kim, & Gladyshev, 2004;Moskovitz et al., 2001). We also showed that MSRA-1 is expressed in the nervous system of Wt and daf-2 mutants (Minniti et al., 2009). In Drosophila, this oxidation repair enzyme also activates FOXO (DAF-16) increasing its nuclear localization (Chung et al., 2010). However, the mechanism involved is unknown.
In this paper, we show that there is a consistent pattern of  (Ch'ng et al., 2008). This in vivo secretion assay allows us to discriminate between neuropeptide expression and its release from neurons since its synthesis is maintained constant during lifespan due to the promoter selected. INS-22 is one of many C. elegans neuropeptides, and it is endogenously expressed in most neurons, including DA and DB neurons (Baugh, Kurhanewicz, & Sternberg, 2011;Li & Kim, 2008). We estimated the secretion of INS-22::Venus by comparing the fluorescence present in neurons: Venus fluorescence in axonal processes (Diagram in Figure 1Aa) vs. secreted insulin: Venus fluorescence accumulated in the coelomocytes (Figure 1Ca). We expect a negative correlation between puncta number and coelomocyte fluorescence. However, there could be a time delay between secretion and coelomocyte uptake.
First, we analyzed neuronal puncta in DA6 and DB6 neurons. As expected, we can detect INS-22::Venus in the axons of these neurons ( Figure 1A). We analyzed fluorescence images from adult individuals: 1-, 2-, and 3-day-old worms (we consider this period as the active reproductive stage: ARS), 4-day-old worms (declining reproductive stage: DRS), and 7-day-old worms (post-reproductive stage: PRS) ( Figure 1A,B). In order to compare larval with adult INS-22::Venus secretion, we also evaluated the L4 stage ( Figure 1). In our model, we found that the neurons are producing INS-22::Venus in all the developmental stages analyzed, including the L4 stage.
Even though there is a decrease in the number of fluorescent puncta from the L4 stage to day 1 of adulthood, by day 2, the number of puncta is equivalent to that of the L4 (Figure 1b In order to explore if this pattern of INS-22::Venus secretion is specific for this peptide or a common behavior of other neuropeptides, we evaluated two other neuropeptides: ANF::GFP and NLP-21::Venus. We found a similar temporal secretion pattern for these two peptides (Supporting Information Figure S1).

| The absence of MSRA-1 increases INS-22::
Venus release during the active reproductive stage Current evidence shows that not all the components that influence the organism redox homeostasis have the same consequences in the determination of lifespan in C. elegans (Doonan et al., 2008). When the insulin pathway is downregulated, (daf-2 mutants) the antioxidant machinery is activated and longevity is extended.
We previously showed that the insulin-IGF-1/DAF-2 pathway regulates the antioxidant enzyme MSRA-1 in C. elegans and the deletion of the msra-1 gene shortens the worms' lifespan (Minniti et al., 2009). Supporting Information Figure S2 shows that by day 5 of adulthood msra-1 mutants have increased levels of proteins that are oxidized in methionines. There are other antioxidant enzymes controlled by this pathway such as the Mn-superoxide dismutase (Honda & Honda, 1999). However, they do not cause a decrease in lifespan when absent (Doonan et al., 2008). This evidence along with  Figure 3a shows that this is indeed the case. Considering that by day 10 of adulthood 50% of msra-1 mutant worms are already dead, all further assays were performed up to day 7 to avoid selecting for the fittest worms.  The graph in Figure 4B shows the quantification of INS-22::Venus Using an RNAi strategy, we further tested the hypothesis that msra-1 is particularly relevant in neurons in terms of INS-22::Venus secretion. In this experimental approach, we used a Wt background strain, which is resistant to RNAi in neurons. We have already shown that using this strategy MSRA-1 expression is maintained only in neurons (Minniti et al., 2015). shows puncta number in msra-1 rescued strains. Scale bar, 5 μm. Data are means ± SE from at least three independent assays. Student's t test was used for statistical analysis. ***p < 0.0001 when comparing Wt with msra-1 mutants. +++p < 0.0001, when comparing Wt L4 with Wt 1-day-old. &&& p < 0.0001, when comparing L4 msra-1 mutants with 1-dayold msra-1 mutants. At least 25 animals per strain were tested for each time point Figure S3a,b). This approach allowed us to detect and compare only the insulin pro-peptide fused to Venus. However, we were not able to analyze the secreted form using this technique because Venus and GFP (used as co-transformation marker in our strains) are of similar size. The data show that the synthesis of this pro-peptide is fairly equivalent in all strains since the differences are not statistically significant.
Subsequently, we tested the hypothesis that msra-1 mutant coelomocytes present exacerbated endocytosis. We used a strain that expresses a secreted form of GFP in muscle cells (strain GS1912, (Fares & Greenwald, 2001)). We compared GFP accumulated in the coelomocytes of hermaphrodites of this strain (Wt background) with GFP accumulated in the coelomocytes of msra-1 mutants. We did not find increased accumulation of secreted GFP in msra-1 mutant coelomocytes (Supporting Information Figure S3c,d).
Therefore, the accumulation of INS-22:Venus in the coelomocytes of msra-1 worms is neither due to increased expression nor to exacerbated endocytosis in these cells.
An unexpected result is that the absence of MSRA-1 (msra-1; daf-2 double mutant) does not seem to significantly affect INS-22:: Venus secretion levels in the daf-2 mutant (Supporting Information Figure S4).   (Figure 5a). Worms exposed to NAC 5 mM show the Wt lifespan phenotype (orange line). A higher NAC concentration (10 mM) shows a negative effect on longevity. This effect is also observed in treated Wt worms (Figure 5b). Figure 5c shows that 5 mM NAC treatment during the ARS restores longevity of msra-1 mutants to near Wt levels (red line). Figure

| The locomotor impairments of msra-1 mutants can be reversed by treatment with the antioxidant NAC during the active reproductive stage
The concept of healthspan has gained importance since it is the period of time that the organism remains healthy and not merely alive (Herndon et al., 2002;Podshivalova, Kerr, & Kenyon, 2017). Therefore, we also wanted to assess the effects of NAC on the aging associated phenotypes of the short-lived msra-1 mutant. These     et al., 2009). Therefore, we decided to test whether antioxidant treatment during the active reproductive stage can restore msra-1 motility to Wt levels. We analyzed swimming capacity through two parameters: turns/min and trajectory length. Figure 6a (left panel) shows representative images of 1-and 10-day-old worms of different genotypes: Wt, msra-1 mutants and rescued msra-1 mutants (native expression). Figure 6b shows that msra-1 mutants (black circles) move slower than Wt worms (white circles) during the adult stage (turns per minute during swimming). However, at the end of the active reproductive stage (3-day-old adults), the mutants' motility worsens abruptly. This is not the case for the Wt, which maintains its motor abilities at least until day 5. The same is true for the rescued mutants. In fact, the expression of msra-1 exclusively in the nervous system is sufficient to rescue motor capacity to Wt levels.
In the same way, track length during swimming is significantly reduced in the mutants during all stages and the defects can be rescued when msra-1 is expressed in the nervous system ( Figure 6c).
Additionally, we compared the turn angle during swimming. 1-day- Even though there is evidence that the DAF-2 insulin receptor regulates aging during the adult stage (Dillin et al., 2002), in this work, we report that the insulin-IGF-1/DAF-2 pathway could be regulated temporally by the release of the insulin ligand during the ARS (or perhaps even just during the first hours of adulthood). This suggests that insulin levels during the C. elegans early active reproductive period could determine later metabolic processes that influence aging.
One of the main consequences of the inhibition of the insulin pathway that leads to longer lifespan is the activation of the cell´s antioxidant machinery (Tullet, 2015). It has been reported that not all the antioxidant enzymes and redox compounds upregulated by inhibiting the insulin pathway result in a shorter lifespan when mutated (Doonan et al., 2008;Partridge & Gems, 2002). This evidence suggests that there might be genes that are critical for maintaining redox homeostasis that influence aging. MSRA-1 is an antioxidant enzyme that when absent decreases longevity in C. elegans (Minniti et al., 2009) and other organisms from yeast to mammals (Koc et al., 2004;Moskovitz et al., 2001). Moreover, it was reported that this enzyme favors DAF-16 nuclear localization in Drosophila, which suggests it may have a central role in lifespan determination by regulating the insulin pathway (Chung et al., 2010).
Therefore, we investigated if MSRA-1 could regulate the insulin pathway upstream of DAF-16 in C. elegans. We found that indeed, Data are means ± SE from at least three independent assays. Student's t test was used for statistical analysis to compare Wt and mutants. ***p < 0.0001, ** p < 0.001, *p < 0.01. At least 30 animals were tested per strain at each time point longer lifespan. Our results suggest that neurons are not only a main site of insulin expression (Baugh et al., 2011;Ritter et al., 2013), but also a primary tissue where insulin secretion is tightly regulated. The mechanism that governs the temporal regulation of neuropeptide release is unknown; however, there is C. elegans data on a few molecules (goa-1, tom-1, unc-64, and unc-31) involved in the process of dense core vesicle fusion and neuropeptide secretion that support our evidence that increased insulin release is linked to aging (Ch'ng et al., 2008). In mammals, growth hormone, which plays an important role in the regulation of the insulin/insulin-like growth factor 1 signaling, also affects lifespan (Bartke, Westbrook, Sun, & Ratajczak, 2013). Moreover, human subjects enriched for familial longevity show decreased secretion of growth hormone (van Heemst et al., 2005).

MSRA
In this way, during Wt aging, MSRA-1 may participate in a feedback loop that maintains the insulin pathway downregulated by lowering the availability of insulin ligand/s (see model in the Figure 7). It is known that oxidation on methionines can control the activity of several proteins, such as calmodulin, CaM kinase II, and Aß peptide among many others (Minniti et al., 2015;Oien & Moskovitz, 2008).
Therefore, MSRA-1 could be acting directly on the regulation of proteins involved in dense core vesicle fusion machinery, on proteins that control intracellular calcium levels such as Calmodulin (Lim, Kim, & Levine, 2013), or on proteins that bind Calmodulin (CaM kinase II) (Erickson, He, Grumbach, & Anderson, 2008).
It is possible that other antioxidant molecules that regulate redox homeostasis may also participate in the regulation of insulin secretion.
In order to further test the idea that intracellular redox balance regulates the insulin pathway during adulthood, we exposed msra-1 mutants to the antioxidant NAC. As expected, restoration of redox balance through antioxidant treatment of the msra-1 mutants only restored health span (measured as locomotor abilities), insulin release, and lifespan to levels similar to Wt when it was administered during the worms´active reproductive period. After this period, antioxidant treatments are ineffective. It is well known that the organism response to antioxidants displays hormesis, since low levels of these compounds may be beneficial while higher concentrations tend to be detrimental (Schieber & Chandel, 2014 (Klotz et al., 2015).
Our evidence supports a mechanism by which redox balance may negatively regulate insulin/neuropeptide secretion from neurons during the ARS (or part of the ARS) in C. elegans. In this way, antioxidant enzymes such as MSRA-1 could participate in a feedback loop that ensures the pathway regulation during the organism´s lifespan.
Ultimately, these findings suggest that the insulin-IGF-1/DAF-2 pathway could be regulated only in a narrow window of time during the lifetime of the organism.

| Nematode strains and culture
Nematodes were raised at 20°C (unless otherwise indicated) under standard laboratory conditions (Stiernagle, 2006

| Lifespan analysis
Lifespan analyses were conducted at 20°C as described previously (Kenyon et al., 1993;Minniti et al., 2009) with the following modifications. In all experiments between 25 and 30, L4 hermaphrodites were transferred to plates with OP50 bacteria and containing the chemical 2′-fluoro-5-deoxyuridine 100 μM (FUdR; Sigma) to inhibit the production of progeny (day 0). The worms were scored every 2 days, and they were considered dead when they no longer responded to a gentle prodding with a platinum wire. Worms were transferred to new plates every 2 days. Worms that crawled off the plates during the assay were replaced using the backup plates. All lifespan assays were repeated in at least three independent experiments. Lifespan curves were generated and analyzed with the GraphPad Prism version 5.0. The log-rank (Mantel-Cox) test was used for statistical analysis of survival curves.

| NAC treatment
Worm populations were exposed to different NAC (N-acetylcysteine, Sigma) concentrations during different periods of time. NAC (diluted in water) was poured onto ready NGM plates before bacteria were seeded to reach the desired final concentration in the agar.

| Motility assays
Individual adult animals (between 15 and 25 animals per experiment) were placed on a 30 μl drop of M9 buffer (Stiernagle, 2006). After a 2-min recovery period, each individual worm was recorded for 1.0 min and the worms´behavior was analyzed using the WORM-LAB2.0 Software (MBF Bioscience). Turns per minute and track length were automatically examined and quantified as described in (Morales-Zavala et al., 2017).

| Microscopy
Animals were anesthetized in 20 μM NaAzide and mounted on slides. Images were acquired using the same exposure parameters for all experimental conditions, with a 40× objective in an Olympus BX51 microscope (Shinjuku, Tokyo, Japan) equipped with a digital camera Micropublisher 3.3 RTV (JH Technologies, Fremont, CA).

| Image fluorescence quantification
Digital quantification of INS22::Venus puncta number in the dorsal nerve cord (anterior or posterior regions nearest to the vulva) and INS22::Venus fluorescence in the posterior coelomocytes was done using the IMAGEJ software. Data were estimated as integrated density value using the same threshold parameters for control and treatment situations.

| Protein extraction
Collected worms were resuspended in 200 μl lysis buffer in the presence of protease inhibitors (50 mM HEPES pH = 7.5; 6 mM MgCl2; 1 mM EDTA; 75 mM sucrose; 25 mM benzamidine; 1%Triton X-100) and frozen at 80°C. The samples were sonicated three times on ice for 15 s and centrifuged at 12,000 g for 15 min; the supernatants were used for Western blot analysis.

| RNAi experiments
These experiments were performed using standard RNAi feeding protocols (Murphy et al., 2003) with the corresponding Escherichia coli strain from the Ahringer RNAi feeding library (F43E2.5 clone).
The exposure of worms to the msra-1 RNAi clone was done from the embryo stage.

| Data and statistical analysis
The statistical analyses were performed using the GRAPHPAD PRISM5 software.

ACKNOWLEDG MENTS
Some nematode strains used in this work were provided by the