Chemotaxis ability declines with age
We first characterized the chemotaxis behavior of young adult (day 1 of adulthood, D1) animals cultivated under our laboratory conditions toward three volatile odorants: diacetyl, isoamyl alcohol, and benzaldehyde. To quantitate chemotaxis ability, we performed population chemotaxis assays and defined the following indices (Fig. 1A) 5: 1) the complete chemotaxis (CTXc) index, which reflects the ability to sense odorants and complete migration toward the odorant source, 2) incomplete chemotaxis (CTXi) index, which reflects the ability to sense odorants and can be high even when animals cannot reach the attractant, 3) the dispersion index, which indicates the number of animals that reached either attractant or control area, and thus reflects the animals’ overall ability to move. We examined chemotaxis behaviors toward three odorants across a wide range of concentrations (10− 5 ~ 102%) and calculated dose-response curves (Fig. 1B-1J). The parameters of the dose-response curves are summarized in Supplementary Table 1. As previously shown 5, young animals showed chemotaxis behavior toward diacetyl (Figs. 1B and 1C, black) and isoamyl alcohol (Figs. 1E and 1F, black) in a dose-dependent manner up to 10%; animals were attracted to benzaldehyde up to 1% but were repulsed at 100% (Figs. 1H and 1I, black). The CTXc and CTXi indices were similar at D1, suggesting that young adult animals can sense and complete the locomotion towards an attractive odorant source. At D1, the dispersion indices were relatively high even at the low concentration of odorants (Figs. 1D, 1G, and 1J, black), suggesting that D1 animals are generally exploratory, even when they do not sense any attractants.
We next characterized the effect of aging on the chemotaxis ability. To prepare aged animals, we treated animals with 5-fluorodeoxyuridine (FUdR), which prevents their offspring from hatching 15. When fed with the standard laboratory diet, E. coli, aged (day 5 of adulthood, D5) animals showed a rightward shift of the CTXi dose-response curves for all three odorants, suggesting the less sensitivity to odorants (Fig. 1C, 1F, and 1I, magenta). EC50 values of the CTXi index drastically increased for diacetyl (113.5 times), and only mildly increased for both isoamyl alcohol (8.4 times) and benzaldehyde (6.3 times) (Supplementary Table 1). For both isoamyl alcohol and benzaldehyde, aged animals were still able to chemotax toward higher concentrations of odorant, suggesting that D5 animals did not lose the locomotory ability at this age (Fig. 1E and 1H, magenta). Nevertheless, aged animals showed a lower CTXc index for diacetyl even when the odorant concentrations were high (Fig. 1B, magenta). Similarly, the dispersion indices were high at high concentrations of isoamyl alcohol and benzaldehyde but low at high concentrations of diacetyl (Figs. 1D, 1G, and 1J, magenta). The time-course experiment of chemotaxis toward diacetyl showed that animals experienced a marked reduction of CTXc and CTXi indices at D3 of adulthood, while the dispersion index remained relatively high at D3 (Supplementary Fig. 1). Interestingly, the timing of the chemotaxis decline was correlated with the end of self-reproduction (Supplementary Fig. 1). Collectively, we concluded that although aged animals have defects in the sensation of all three odorants, the severity of the sensory loss is odorant-dependent.
Feeding with lactic acid bacteria maintains diacetyl chemotaxis in aged animals
To address the effects of diet on the age-dependent chemotaxis decline, we used Lactobacillus reuteri (L. reuteri) as an alternative diet because we have previously shown that switching animals’ diet from E. coli to L. reuteri during adulthood prevents age-dependent thermotaxis decline12. To avoid the dietary effects on the development, animals were fed with E. coli until D1 and then with E. coli or L. reuteri until day 5 of adulthood (D5). Compared to E. coli-fed aged animals, L. reuteri-fed aged animals showed better sensation (CTXi index) for diacetyl (Fig. 1C, light blue) and isoamyl alcohol (Fig. 1F, light blue) but not for benzaldehyde (Fig. 1I, light blue). Moreover, L. reuteri-fed aged animals maintained a higher CTXc index toward diacetyl. These results suggest that similar to what we previously saw in the age-dependent loss of odorant sensing, maintenance of chemotaxis behavior through diet acts in an odorant-specific manner.
We sought to determine that L. reuteri-induced maintenance of chemotaxis in aged animals is due to long-term dietary effects and not to short-term olfactory adaptation, which occurs at a timescale of hours 16. Because both age and diet dramatically affected the chemotaxis behavior toward diacetyl, we used 0.1% concentration of diacetyl in all subsequent experiments. We sequentially switched the animals’ diet from E. coli to L. reuteri at different timings and found that a minimum of four consecutive days of feeding with L. reuteri was required to fully maintain high CTXc and CTXi indices of D5 animals (Fig. 2). On the other hand, switching diets from L. reuteri to E. coli one day prior to the assay was sufficient to decrease both CTXc and CTXi indices. To rule out the possibility that the age-dependent chemotaxis decline is due to olfactory adaptation, we tested a mutant strain of egl-4, which encodes a GMP-dependent protein kinase required for olfactory adaptation of benzaldehyde chemotaxis 16. Since egl-4 mutants displayed lower chemotaxis ability toward 0.1% diacetyl and benzaldehyde even at D1 (Supplementary Fig. 2), we used 1% of odorant instead. At this concentration, egl-4 mutants showed age-dependent chemotaxis decline (Supplementary Fig. 2). These results imply that age- and dietary modulation of chemotaxis is unlikely attributed to olfactory adaptation.
Bacterial diets modulate the expression of odr-10 receptors via its promoter
We asked whether L. reuteri-fed aged animals are able to achieve chemotaxis through the same molecular and cellular mechanisms as young animals. In young animals, low concentrations (0.1%) of diacetyl are sensed by the odorant receptor ODR-10 that is specifically expressed in the AWA olfactory neurons 10. At a higher concentration of diacetyl (10%), both AWA and AWC olfactory neurons redundantly regulate chemotaxis toward diacetyl (Bargmann et al. 1993; Colbert and Bargmann 1995). To determine whether AWA via odr-10 or AWC neurons are mediating the chemotaxis in the L. reuteri-fed aged animals, we compared diacetyl chemotaxis in odr-10 mutants and AWC-ablated animals. AWC was ablated by overexpressing a split caspase under the control of an AWC-specific promoter 17. Consistent with previous results 10, chemotaxis of young animals toward diacetyl required odr-10 and was independent of AWC neurons (Fig. 3A, D1). Similar to that of D1 animals, the chemotaxis of L. reuteri-fed aged animals was dependent on odr-10 and independent of AWC neurons (Fig. 3A, D5). These results indicate that L. reuteri feeding maintains diacetyl chemotaxis through the odorant receptors in the AWA neurons instead of alternative recruitment of the AWC neurons.
Because odr-10 was required for diacetyl chemotaxis in L. reuteri-fed aged animals, we asked whether dietary changes during aging influenced odr-10 gene expression. The expression of the odr-10 transcripts decreased during aging when animals were fed E. coli, while its levels were maintained in L. reuteri-fed aged animals (Fig. 3B, wt). To determine whether odr-10 expression is regulated by its promoter, we measured the fluorescence of a single-copy odr-10 transcriptional reporter (knjSi26[odr-10p::GFP::3’UTR (odr-10)]) in the AWA cell body. The fluorescence was higher in L. reuteri-fed aged animals than in E. coli-fed aged animals (Figs. 3C and 3D). Another multi-copy transcriptional reporter (kyIs37[odr-10p::GFP::3’UTR (unc-54)]) with a different 3’UTR showed the same trend of increased GFP signal in L. reuteri-fed aged animals (Supplementary Fig. 3), suggesting that diet affects odr-10 expression through its promoter. In both knjSi26 (Figs. 3C and 3D) and kyIs37 (Supplementary Fig. 3) transgenes, the GFP fluorescence was increased from D1 to D5 in E. coli-fed animals, unlike the trend we observed in the endogenous odr-10 transcripts (Fig. 3B). The apparent increase in GFP fluorescence in aged animals is likely due to the accumulation of stable GFP proteins during aging because the GFP transcripts in knjSi26[odr-10p::GFP::3’UTR (odr-10)] showed no alteration during aging (Fig. 3E). Given that the endogenous odr-10 transcript, but not GFP transcript under the control of odr-10 promoter and 3’UTR, decreased during aging, the odr-10 transcript itself might undergo post-transcriptional regulation. Collectively, aging and bacterial diet appear to affect the expression of odr-10 transcripts via post-transcriptional regulation and promoter, respectively.
What lies upstream of odr-10 expression under the aged, L. reuteri-fed conditions? In young animals, odr-10 expression is regulated by the nuclear hormone receptor, ODR-7 10,18. The expression of odr-10 transcripts in both D1 and D5 animals was dependent on odr-7 irrespective of diet (Fig. 3B). However, odr-7 transcripts in aged L. reuteri-fed animals did not differ from that of E. coli-fed aged animals (Fig. 3F). Our result implies that bacterial diet may affect the activity, rather than the expression, of ODR-7 in regulating odr-10 expression.
We next asked whether increasing the odr-10 expression is sufficient to confer high chemotaxis ability in aged animals. Thus, we generated a strain overexpressing odr-10 under the control of the odr-7 promoter and tested its chemotaxis in aged animals. The odr-10 overexpression restored the CTXi index, but the CTXc index was not comparable to young animals (Supplementary Fig. 4). This result suggests that increasing odr-10 expression is sufficient to restore an aged animal’s sensitivity toward diacetyl but insufficient to restore its ability to complete chemotaxis behavior.
Food deprivation maintains high chemotaxis ability in aged animals
C. elegans is shown to ingest both E. coli and L. reuteri 19. We sought to understand whether the maintenance of chemotaxis ability in L. reuteri-fed aged animals is due to the presence of a beneficial effect provided by L. reuteri or the absence of a detrimental effect derived from E. coli. To distinguish these possibilities, we assessed whether aged animals under food deprivation would mimic either behavioral phenotypes presented by L. reuteri-fed or E. coli-fed conditions. We found that aged animals under food deprivation maintained a higher chemotaxis ability toward diacetyl than animals fed with E. coli (Fig. 4A, magenta vs. white box plots), suggesting that feeding E. coli decreases the animals’ chemotaxis ability during aging. When tested for chemotaxis toward isoamyl alcohol and benzaldehyde, aged animals under food deprivation did not show recovery in chemotaxis ability (Supplementary Fig. 5). We further characterized the time-course relationship between food deprivation and age-dependent chemotaxis decline. Similar to L. reuteri-fed condition, food deprivation for at least three days was required to observe high chemotaxis ability at D5 (Fig. 4A). We also asked whether re-feeding food-deprived animals with E. coli could, in turn, reduce their chemotaxis performance, and examined re-feeding 6 or 24 hours prior to the assay at D5. Although 6 hours of re-feeding slightly reduced the CTXi index, further reduction of CTXi and CTXc indices were observed by 24 hours of re-feeding (Fig. 4B). The time scale of the effect and the requirement of food consumption is in agreement with the former statement that the modulation of chemotaxis behavior is not due to chemosensory adaptation. Consistent with this notion, exposure to the E. coli odor did not decrease the chemotaxis ability of aged animals cultured without food (Fig. 4C).
We next asked whether food deprivation maintains chemotaxis in aged animals through caloric restriction. To assess this possibility, we tested the mutants of eat-2 that encodes an acetylcholine receptor expressed in pharyngeal muscles whose defects cause caloric restriction by reducing the animal’s feeding efficiency20. We found that eat-2 mutants did not show high chemotaxis ability at D5 (Supplementary Fig. 6), suggesting that the effect of food deprivation tested under our conditions is distinct from the effect of caloric restriction.
Because dietary deprivation is known to extend lifespan21, we measured the lifespan of food-deprived and E. coli-fed animals. We found that food-deprived animals showed 30.6% longer lifespan compared to E. coli-fed animals (Supplementary Fig. 7, Supplementary Table 2). This increase in lifespan does not sufficiently explain the higher chemotaxis ability of food-deprived aged animals because E. coli fed D3 animals already displayed significantly less chemotaxis ability (Supplementary Fig. 1) compared to food-deprived D5 animals. Thus, the high chemotaxis performance of food-deprived aged animals does not appear to be a secondary effect of longer lifespan.
daf-16 is required for the chemotaxis of aged animals under food deprivation
We next investigated which genes are responsible for the food deprivation-dependent high chemotaxis ability in aged animals. We focused on the FOXO transcription factor DAF-16 because we have previously shown that DAF-16 is involved in the dietary effect on thermotaxis behavior in C. elegans 12. The young D1 daf-16(mu86) null mutants showed marginal defects in chemotaxis (Figs. 5A and 5B). On the other hand, the high chemotaxis ability of food-deprived aged animals was completely dependent on daf-16, as daf-16(mu86) mutants performed poorly in both CTXc and CTXi indices (Fig. 5B). Since daf-2 encoding insulin/IGF-1 receptor negatively regulates DAF-16 13, we tested whether daf-2 mutants phenocopied food deprivation. Indeed, daf-2 aged animals showed higher chemotaxis ability than the wild-type aged animals, even under E. coli-fed conditions (Supplementary Fig. 6).
We examined the tissue and isoform specificity of daf-16. The daf-16 genomic locus encodes several different isoforms with different N-termini transcribed from unique promoters (Fig. 5A) 22–24. These isoforms are expressed in different tissues: daf-16a and daf-16f are expressed in the hypodermis, muscle, neurons, and intestines, while daf-16b is expressed primarily in the neurons, pharynx, and muscles 22,24–27. We tested knockout lines carrying isoform-specific mutations of daf-16: 1) daf-16(mg54) introduced with a premature stop codon to daf-16a, d, f; 2) daf-16(knj36) introduced 7-bp deletion to daf-16b isoform (Fig. 5A). Neither daf-16(mg54) nor daf-16(knj36) mutants showed substantial defects in chemotaxis (Fig. 5B), suggesting that the existence of either isoform is sufficient to maintain high chemotaxis of aged animals under food deprivation. We further validated that the daf-16b isoform is sufficient to induce high chemotaxis ability by demonstrating that a transgenic line consisting of a genomic DNA of the daf-16b isoform, including its endogenous promoter (knjSi17[daf-16bp::daf-16b]) fully rescued the phenotype of daf-16(mu86) mutant (Fig. 5C, daf-16bp).
To determine in which tissue daf-16 functions, we performed rescue experiments for the tissue where daf-16b is expressed. The expression of daf-16 in muscles and neurons partially rescued the full-body knockout of daf-16, while the expression in the pharynx did not (Fig. 5C), suggesting that daf-16 might be functioning in neurons and/or muscle to elicit chemotaxis under food deprivation.
As DAF-16 is a transcription factor that functions at least partially in neurons, we asked whether daf-16 might regulate the expression of odr-10. Like L. reuteri-fed conditions, food deprivation increased the endogenous odr-10 transcripts and odr-10p::GFP reporter expression in the wild-type animals (Figs. 6A-C, wt). The higher expression of odr-10 in the food-deprived condition did not depend on daf-16 (Figs. 6A-C, daf-16), suggesting that daf-16 was acting downstream or independently of the odr-10 expression to affect the chemotaxis behavior of food-deprived aged animals.