skn-1 is required for interneuron sensory integration and foraging behavior in Caenorhabditis elegans

Nrf2/skn-1, a transcription factor known to mediate adaptive responses of cells to stress, also regulates energy metabolism in response to changes in nutrient availability. The ability to locate food sources depends upon chemosensation. Here we show that Nrf2/skn-1 is expressed in olfactory interneurons, and is required for proper integration of multiple food-related sensory cues in Caenorhabditis elegans. Compared to wild type worms, skn-1 mutants fail to perceive that food density is limiting, and display altered chemo- and thermotactic responses. These behavioral deficits are associated with aberrant AIY interneuron morphology and migration in skn-1 mutants. Both skn-1-dependent AIY autonomous and non-autonomous mechanisms regulate the neural circuitry underlying multisensory integration of environmental cues related to energy acquisition.


Introduction
The nuclear factor erythroid 2-related factor 2 (Nrf2) is a member of the Cap'n'Collar basic leucine zipper family known to play a key role in the cellular antioxidant and detoxification responses to stress [1]. In mammals under homeostatic conditions Nrf2 is maintained inactive via interaction with kelch-like ECH associated protein 1 (Keap1) and Cullin3. Cullin3 ubiquitinates and targets Nrf2 for continuous proteasomal degradation. In response to oxidative or electrophilic agents the interaction of Nrf2 with Keap1 is disrupted preventing Cullin 3-dependent degradation. The stabilized Nrf2 protein translocates into the nucleus where it dimerizes with small musculoaponeurotic fibrosarcoma (Maf) proteins to drive the transcription of cytoprotective genes [1]. Nrf2 and its invertebrate orthologs skn-1 (Caenorhabditis elegans) and CncC (Drosophila melanogaster) have also been shown to promote health span and longevity. In C. elegans and D. melanogaster the genetic activation of the Nrf2 signaling pathway leads to PLOS  enhanced longevity [2,3]. The comparison of Nrf2 activity in rodents with different maximum lifespan potential revealed a positive correlation between Nrf2 activation status and longevity [4]. Interestingly, despite showing high constitutive levels of active Nrf2 [4], the naturally long lived rodent naked mole rat has very low antioxidant enzyme expression [5]. This suggests that Nrf2 may exert pro-longevity benefits beyond its well characterize role as an antioxidant modulator. One of the best studied examples of biological conditions able to promote longevity in a wide range of experimental models is decreased nutrient availability [6,7]. Caloric restriction and signaling pathways involved in metabolic homeostasis are recognized as universal effectors of longevity phenotypes [6][7][8]. Optimal metabolic homeostasis depends upon proper integration of intrinsic and extrinsic environmental signals by the nervous system. The impact of sensory integration (i.e. gustatory, olfactory, thermosensory) in health span and aging phenotypes has been extensively studied in invertebrates models [9]. For example, experimental impairment of olfaction extends lifespan in flies [10]. In C. elegans, mutations or manipulations impairing sensory neuronal function extend lifespan [11,12], and the longevity phenotype induced by caloric restriction depends upon the activity of Nrf2/skn-1 in the ASI chemosensory neurons [13]. Activation of Nrf2 protects neurons in experimental models of stroke [14][15][16], Parkinson's disease [17,18], and Huntington's disease [19]. However, the contribution of the Nrf2 pathway to neuronal functions under physiological conditions is still largely unknown. In the present study we provide evidence that Nrf2 ortholog skn-1 plays a fundamental role in foraging and food-related sensory integration in C. elegans.

Material and methods
Mice C57BL/6 mice were breed and housed at the National Institute on Aging facility. Male (3 months old) were euthanized by carbon dioxide inhalation, and brains were removed and either dissected for immunoblot analysis, or processed for immunohistochemistry. All procedures were approved by the Institutional Care and Use Committee (IACUC) of the National institute on Aging (ASP 290-LNS-2019), and complied with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Immunohistochemistry
Brains were fixed in 4% paraformaldehyde in PBS for 24 hours, then transferred to 30% sucrose in PBS at 4˚C. Following antigen retrieval, sagittal sections were incubated for 1 h in blocking solution and then incubated overnight at 4˚C with Nrf2 and glial fibrillary acidic protein (GFAP) antibodies. After extensive washing, the primary antibodies were detected using Alexa Flour 488-and 633-conjugated secondary antibody. Nonspecific labeling was determined by omission of the primary antibody. The cell nuclei were counterstained with DAPI (4',6'-diamidino-2-phenylindole) dye. Confocal images were acquired using a Zeiss 510 LSM microscope with 10X and 40X objective lenses.

Behavioral assays
Chemotaxis assays were performed on 10 cm plastic petri dishes containing 10 ml assay medium (1.6% agar, 1mM CaCl 2 , 1mM MgSO 4 , 4mM NH 4 Cl, 25mM KH 2 PO 4 pH 6.0), and analyzed as previously described [23]. Trials were performed in triplicate, with 25 animals per plate, unless otherwise indicated. Briefly, worms were transferred from culture plates to an empty assay plate and allowed to crawl freely for 10 minutes to remove residual bacteria. Worms were transferred to the center 1 cm of the assay plate (origin) and incubated at room temperature for 1 hour, at which time their positions were scored immediately. For sodium chloride assays, 5 μl of 5M NaCl was spotted onto one edge of the assay plate and left for 16 hours at room temperature to form a gradient prior to use [24]. Chemotaxis index was calculated as (O-S)/(T) where O was the number of animals at the odorant location and S the number of animals at the solvent control location after 1 hour and T the total number of worms. Thermotaxis assays were performed on radial thermal gradients on chemotaxis assay plates [25]. For dwelling assays, a 10 μl spot of diluted OP50 was spotted onto the center of the assay plate and allowed to dry for 1 hour. OP50 bacteria were grown overnight at 37˚C with shaking, then washed twice with M9 buffer and diluted to an absorbance at A600 nm of 1.5 (1x concentration).
with differential interference contrast. For quantification, identical areas of the worms were selected, and the total fluorescence was calculated as the mean fluorescence per μm 2 .

Gene expression
Semi-quantitative rtPCR was performed on a Chromo4 system with Opticon 3 software (BioRad). Fold change was calculated using the ΔΔCt model [26], normalized to actin (act-1). Primers are listed in S1 Table.

Results and discussion
In order to gain insights into the potential roles for Nrf2 in brain physiology, we first analyzed the levels of Nrf2 in five different regions of the mouse brain (olfactory bulb, cerebral cortex, hippocampus, cerebellum, and medulla). Immunoblot analysis revealed that Nrf2 protein levels in the olfactory bulb were four-to ten-fold greater than in any other brain regions evaluated ( Fig 1A). Histological analysis of sagittal brain sections revealed high levels of Nrf2 expression in the rostral migratory stream (RSM), as well as in the mitral cell layer (MCL), and granule cell layer (GCL) of the olfactory bulb ( Fig 1B and 1C). The RMS is a specialized pathway through which neuronal precursor cells originating in the subventricular zone migrate into the olfactory bulb ( Fig 1B). Our findings of high Nrf2 expression in RMS cells substantiates recent evidence showing that in mice Nrf2 ablation and age-dependent alterations of the Nrf2 pathway, impair subventricular zone neuronal progenitor cell proliferation and survival [27][28][29]. Once the newly generated cells reach the olfactory bulb they differentiate into γ-aminobutyric acid (GABA) interneurons that integrate into the granule cell layer (GCL), or replace periglomerular cells in the glomerular layer (GL). In the olfactory bulb in addition to specific regional distribution we also observed distinct subcellular patterns of expression. In mitral cells Nrf2 is mainly located in the cytoplasm (Fig 1C), but in granule cell interneurons it is mostly concentrated in the nucleus, suggesting a constitutively active status in these cells (Fig 1C).
Because of its prominence in olfactory neurons we asked whether Nrf2 plays roles in chemosensation-related behaviors that enable animals to locate food sources and mates, as well as to avoid potentially harmful substances and predators. As the predominant sensory input, chemosensation is particularly important in the nematode C. elegans, where it regulates several metabolic and behavioral responses including pharyngeal pumping, locomotion, egg laying, life span and dauer formation [30]. In C. elegans the ortholog of Nrf2, skn-1, encodes three major isoforms (a, b and c) which each play distinct functional roles. skn-1a has been recently shown to localize to the mitochondrial membrane and respond to starvation [31]. skn-1b is expressed in neurons and mediates dietary restriction-induced longevity [13]. skn-1c on the other hand is intestinal and has been linked to oxidative stress resistance and longevity [2,32]. To assess the role of skn-1/Nrf2 in chemosensory perception we took advantage of the fact that the C. elegans nervous system and behavioral responses to odorants are well-characterized. We performed standard chemotaxis experiments comparing the ability of wild type N2 and skn-1 (zu135) null worms to respond to known chemoattractants. There were no significant differences in the chemotaxis index for volatile compounds such as benzaldehyde (BA) (Fig 2A and  S2 Table), butanone (BU) and diacetyl (DA) which are recognized by AWC and AWA sensory neurons. However, we did find a small significant decrease in ASE-dependent chemotaxis index toward sodium chloride in skn-1 worms (Fig 2A). During the course of our chemotaxis experiments we noted a tendency of skn-1 worms to remain in close proximity to the origin of the assay plate. When we analyzed this behavior in more detail, we found that a significant fraction of skn-1 worms preferred to remain in the center area of the plate regardless of the type of the compound, soluble or volatile, used as chemoattractant (Fig 2B and 2C). This  behavior was more pronounced when suboptimal concentrations of chemoactractant were used (Fig 2D and S3 Table), did not depend upon the skn-1 background as the skn-1(zu67) strain showed a similar phenotype (S1A Fig and S3 Table), and was not related to impaired motility as demonstrated by identical dispersal indexes (S1B Fig).
In addition to salt chemoattraction, ASE neurons have recently been shown to promote adaptive food-leaving behavior as food becomes limiting [33]. Chemotaxis enables nematodes to locate new food sources, and many chemoattractants are products of bacterial decay. One explanation for the behavioral phenotype we observed is that skn-1 worms can detect very small amounts of carried over bacteria and prefer to dwell around the origin area. To test this possibility we performed 'leaving behavior' assays in which known dilutions of OP50 were placed at the origin. As shown in Fig 2E at concentrations that N2 worms considered indicative of no or negligible levels of food, and led them to roam throughout the plate, skn-1 worms preferred to remain in the area close to the bacterial lawn. Because of this enhanced sensitivity to very small quantities of bacteria, it is possible that for skn-1 worms the chemotaxis experiments performed under standard conditions were actually multi-sensory integration experiments. We thus decided to compare the behavior of N2 and skn-1 worms in conditions where an attractant and a specific concentration of OP50 E. coli placed at the origin were provided as sensory cues. We found that in such experimental settings skn-1 worms preferred to dwell on the OP50 lawn rather than move toward the chemoattractant (Fig 2F and S4 and S5 Tables). These results suggest that skn-1 worm behavior under conditions requiring integration of multiple food-related sensory cues is impaired compared to wild type animals.
Interneurons play a central role in the integration of different sensory modalities. Based on the distinct distribution of Nrf2 in the mammalian olfactory interneurons, the perturbed chemotaxis behavior in skn-1 mutant C. elegans and the known C. elegans neuronal wiring ( Fig  3A), we focused on AIY and AIA as interneurons potentially responsible for the observed skn-1 behavioral phenotype. Previous findings suggest that AIY interneurons are critical for chemotaxis responses to near-threshold levels of odors [34], food choice behavior [35] and dwelling [34][35][36]. We took advantage of the fact that AIY interneurons are essential for C. elegans thermotaxis behavior through AFD sensory neurons (Fig 3A) [37]. We hypothesized that if AIY functions are altered in skn-1 worms, they will likely display abnormal thermotaxis behavior. The ability of N2 and skn-1 worms to migrate toward their cultivation temperature (23˚C) was assessed in population thermotaxis assays using ttx-3 mutant worms in which AIY neurons are dysfunctional [38][39][40]. While N2 animals moved toward the region of the plate closer to their previous cultivation temperature, both skn-1 loss-of-function strains zu135 and zu67 showed a cryophilic phenotype similar to, although not as pronounced as ttx-3 null worms (Fig 3B and S6 Table). Overall our behavioral results indicate that skn-1 is required for proper integratory functions of sensory interneurons. Notably, altered olfactory integration with impaired fine discrimination has also recently been shown in Nrf2 knock out mice [29].
With regard to the C. elegans nervous system, skn-1 has been shown to be expressed in ASI [13] and dopaminergic neurons [42]. Because our data suggest that skn-1 is required for normal AIY function, we analyzed wgIS342;otIS133 double transgenic animals using confocal microscopy. Because of the large genomic DNA sequence used to create the construct, wglS342 drives the expression of all skn-1 isoforms under endogenous coding and regulatory sequences, thus providing a more comprehensive and faithful expression pattern. As shown in Fig 3C,   skn-1::EGFP is highly expressed in neurons in the lateral, ventral and dorsal ganglia, and to a lesser extent in the anterior ganglia. Furthermore, skn-1::EGFP clearly colocalizes with the specific AIY marker ttx-3::RFP, confirming that skn-1 is indeed expressed in AIY interneurons.
The homeodomain transcription factors ceh-10, ttx-3 and ceh-23 coordinate the AIY cell fate specification [43]. We used qRT-PCR to quantify the expression of these transcription factors and some of their known targets in N2 and skn-1 worms. Levels of ceh-10 were comparable amongst the various strains (data not shown), however both ttx-3 and ceh-23 were significantly reduced in skn-1(zu135) and (zu67) worms (Fig 3D). ser-2a and ser-2b, which encode a tyramine receptor, were also significantly reduced in skn-1 worms (Fig 3D). skn-1(zu135), but not skn-1(zu67), showed decreased levels of sra-11 (Fig 3D), suggesting a specific effect of the skn-1b isoform in the regulation of this G protein-coupled receptor implicated in olfactory imprinting [44]. We verified the changes in ser-2 levels by analyzing the expression pattern of GFP driven by ser2::prom1 in N2 and skn-1(zu135) transgenic animals. As shown in Fig 3E a significant decrease for ser-2:: GFP expression was observed in neurons as well as pharyngeal cells and head muscles in skn-1 worms. Surprisingly, levels of hen-1 and kal-1 were increased in both skn-1 strains (Fig 3D). In humans Kallmann syndrome is characterized by hypogonadism, and hyposmia or anosmia. In C. elegans misregulation of the Kallmann syndrome gene ortholog kal-1 has been shown to alter neurite branching and the formation of axon collaterals, as well as axon pathfinding, and establishment of target connections [45,46]. The analysis of AIY morphology in skn-1 worms showed a significant increase of aberrant branching from the axon, as well as of short neurite-like projections from the cell body (Fig 4A). We also observed a less penetrant phenotype of axonal misrouting characterized by the axon of one of the AIY pair running over and around the dorsal midline, and failing to interact with the contralateral axon (Fig 4A). Such defects depended upon levels of skn-1 expression, as heterozygous skn-1 animals displayed an intermediate percentage of abnormalities (normal = 73.7%; abnormal = 26.3%). Furthermore, regardless of the presence of morphological defects we observed that, relative to the pharynx grinder, the position of the AIY cell bodies in skn-1 worms was far more caudal compared to N2 worms (Fig 4B), while the distance of the AIY axon fasciculation at the nerve ring was shorter (Fig 4B). Similar neuroanatomical defects have been described in ttx-3 mutant animals [42,46]. We assessed the possibility that our behavioral and morphological findings result from skn-1 dependent misregulation of AIY by generating transgenic animals in which skn-1b cDNA is specifically and exclusively driven in AIY by a ttx-3 promoter (CY696). Compared to skn-1 mutants, CY696 animals displayed about 54% less AIY morphological defects (normal = 76.3%; abnormal = 23.7%), but identical AIY migratory (Fig 4B), and dwelling behavioral defects (Fig 4C and S7 Table). These findings suggest that skn-1b acts at least partially in a cell autonomous fashion to control AIY morphogenesis, whereas the AIY migratory and behavioral impairments are non-AIY autonomous defects.
Because ttx-3 is normally expressed in other head neurons (AIA, ADL, ASI, ADF) [34,43], and the AIY specific expression of skn-1b is not sufficient to rescue the behavioral abnormalities, we tested a ric-19 promoter to drive the pan-neuronal expression of skn-1b (CY691). CY691 transgenic animals had about 41% less AIY anatomical defects (normal = 69.9%; abnormal = 30.1%), and an intermediate phenotype between N2 and skn-1 null worms in terms of AIY migratory defects (Fig 4B), and food-leaving behavior (Fig 4D and S8 Table). This result, together with the behavioral similarities we observed between the skn-1 strains lacking all three skn-1 isoforms (zu135) and skn-1a/c isoforms (zu67), suggest that expression of skn-1b in non-neuronal cell types is required for a full functional recovery.
Nrf2/skn-1 is widely known for its role in adaptive responses to many different types of stress, particularly oxidative stress. Accordingly, much attention has been given to Nrf2/skn-1 target genes that encode antioxidant proteins [47]. Studies of cell culture and animal models have shown that Nrf2/skn-1 plays a major role in the process of hormesis, in which exposure of a cell or organism to a low level of stress (e.g., oxidative, metabolic, xenobiotic) results in resistance to greater levels of the same or different stressor [48]. Moreover, recent findings suggest that Nrf2/skn-1 mediates, in part, beneficial effects of bioenergetics challenges such as food restriction on organ system function, disease resistance, and longevity [49][50][51][52]. Previous studies of C. elegans have shown that phytochemicals that activate Nrf2/skn-1 can significantly increase lifespan [53], and that skn-1 in ASI neurons is required for the lifespan-extending effect of food deprivation [13]. Our data reveal a previously unknown role for Nrf2/skn-1 in regulating the organization and functionality of neuronal networks involved in behavioral responses of worms to food availability, temperature, and specific odorants. We speculate that this 'decision-making' role for skn-1 in AIY neurons evolved to optimize adaptive behavioral responses to a wide range of environmental challenges faced by worms in their normal habitat.

Conclusions
Our findings show that Nrf2/skn-1 is expressed in interneurons involved in chemosensation in mice and C. elegans, and that genetic deletion of skn-1 results in impaired sensory integration. The inability to activate skn-1 signaling causes the skn-1 worms to linger in inadequate food patches rather than migrate toward new potential food sources. Lack of skn-1 results in morphological, migratory and functional alterations of AIY suggesting that skn-1 is important for the proper development of a pair of neurons that play a pivotal role in coordinating behavioral responses to food and temperature, two fundamental environmental factors influencing growth, reproduction and survival.
Supporting information S1 Fig. (A) Percentage of N2 and skn-1(zu67) animals remaining at the origin in presence of the indicated concentrations of benzaldehyde. (B) N2 and skn-1(zu135) worms display identical locomotion abilities. Animals were placed on an empty plate and allowed to crawl freely for 60 min. Their position was scored and the percentage of animals in the center (Zone 1) and periphery (Zone 2) determined. Data are mean and S.E.M. Ã p<0.05 versus N2.