Microbiota and Diapause-Induced Neuroprotection Share a Dependency on Calcium But Differ in Their Effects on Mitochondrial Morphology

Abstract The balance between the degeneration and regeneration of damaged neurons depends on intrinsic and environmental variables. In nematodes, neuronal degeneration can be reversed by intestinal GABA and lactate-producing bacteria, or by hibernation driven by food deprivation. However, it is not known whether these neuroprotective interventions share common pathways to drive regenerative outcomes. Using a well established neuronal degeneration model in the touch circuit of the bacterivore nematode Caenorhabditis elegans, we investigate the mechanistic commonalities between neuroprotection offered by the gut microbiota and hunger-induced diapause. Using transcriptomics approaches coupled to reverse genetics, we identify genes that are necessary for neuroprotection conferred by the microbiota. Some of these genes establish links between the microbiota and calcium homeostasis, diapause entry, and neuronal function and development. We find that extracellular calcium as well as mitochondrial MCU-1 and reticular SCA-1 calcium transporters are needed for neuroprotection by bacteria and by diapause entry. While the benefits exerted by neuroprotective bacteria require mitochondrial function, the diet itself does not affect mitochondrial size. In contrast, diapause increases both the number and length of mitochondria. These results suggest that metabolically induced neuronal protection may occur via multiple mechanisms.


Introduction
Axonal degeneration underlies many neuropathies and neurodegenerative diseases (Coleman and Perry, 2002) that involve a controlled dismantling of neuronal morphologic extensions (Saxena and Caroni, 2007). This process in the nematode Caenorhabditis elegans can be counteracted by developmental arrest induced by food deprivation (Calixto et al., 2012;Caneo et al., 2019) as well as by exposure to specific intestinal microbiota (Urrutia et al., 2020). The mechanisms that promote neuronal protection in both of these conditions are unclear. Is neuroprotection achieved actively as neuronal repair or merely the result of diminished neurodegenerative processes? Moreover, it is unknown whether ingested microbiota and metabolic state act similarly in this context.
Given the pivotal role of gene expression in the link between environment and cellular phenotype (Davidson and Levin, 2005;van Kesteren et al., 2011;Chen et al., 2015;Lu et al., 2022), identifying transcriptional changes at various stages of recovery may hold the key to understanding novel mechanisms and pathways for neuroprotection.
Oxidative damage is central to neurodegeneration, and mitochondria are key to the neutralization of oxidative stress (Kim et al., 2015;Singh et al., 2019;Ballard and Towarnicki, 2020). The effects associated with failure in the neutralization of ROS include impairment of synthesis and transport of lipids, and Ca 21 transport (Vance, 2014). Loss of mitochondrial homeostasis has been linked to degenerative processes in different animal and cell models (Pathak et al., 2013;Błaszczyk, 2020). In mice, intracellular calcium transporters, such as the mitochondrial calcium uniporter, have been shown to contribute to degeneration (Calvo-Rodriguez et al., 2020;Calvo-Rodriguez and Bacskai, 2021). However, mitochondria maintain cellular homeostasis by also modulating energy production and capture of calcium (Beal, 1996;Shvartsman et al., 2007;Ferrer, 2009;Picard and McEwen, 2014). Calcium signaling in microdomains is important in determining whether neuronal regeneration or degeneration occurs (Xu et al., 2001;Berridge, 2006;Yan and Jin, 2012) and involves the participation of the endoplasmic reticulum (ER) and mitochondria.
To gain mechanistic insight on the commonalities between microbiota and diapause induced neuroregeneration, we use a genetically encoded insult to the touch circuit of C. elegans (Driscoll and Chalfie, 1991;Calixto et al., 2012), whereby a mutation triggers a loss of selectivity of the MEC-4 channel (Shi et al., 2018), leading to an increase in cytoplasmic calcium and the subsequent energetic collapse of the neuron (Driscoll and Chalfie, 1991;Xu et al., 2001;Bianchi et al., 2004;Shi et al., 2018). Using this model, we investigate here the contribution of gene expression, mitochondria morphology, and the need for calcium in these two interventions.

C. elegans growth and maintenance
Wild-type, transgenics, and mutant C. elegans were maintained at 20°C, as reported previously (Brenner, 1974). The following nematode strains were used: wild type (N2); TU2773 [uIs31(Pmec-17mec-17 All animals were maintained in Escherichia coli OP50 before using or feeding with other bacteria.

Bacterial growth
E. coli OP50 and E. coli HT115 bacteria were grown overnight on Luria-Bertani (LB) plates at 37°C from glycerol stocks kept at À80°C. The next morning, a large amount of the bacterial lawn was inoculated in LB broth and grown for 6 h with agitation at 200-220 rpm at 37°C. Three hundred microliters of bacterial culture was seeded onto 60 mm nematode growth media (NGM) plates and allowed to dry overnight.

C. elegans mec-4d transcriptomic analysis
Total RNA was isolated from synchronized C. elegans mec-4d populations feeding on E. coli OP50, or E. coli HT115, at 12, 24, and 48 h after hatching using RNA-Solv Reagent (Omega Biotek). Spectrophotometric quantification and integrity of RNA were determined on a bioanalyzer (model 2100, Agilent Technologies). mRNA libraries were prepared with the TruSeq RNA Sample Prep Kit (Illumina) according to the manufacturer protocol. The quality and size distribution of the libraries were evaluated with the model 2100 Bioanalyzer using a DNA 1000 chip (Agilent Technologies) and were quantified using the KAPA Library Quantification Kit for Illumina Platforms (Kapa Biosystems), on the Step One Plus Real-Time PCR System (Applied Biosystems). The libraries were sequenced using the HiSeq paired-end protocol (2Â 100 bp). High-quality reads were selected using Trimmomatic version 0.36 and mapped to the C. elegans reference genome (WormBase release ws235) using Tophat version 2.0.9. (Trapnell et al., 2012). The resulting bam files were transformed for visualization of mapping results in the University of California, Santa Cruz, genome browser (https://genome.ucsc.edu). The mapping results and HTSeq-count version 0.6.0 were applied with the intersection nonempty argument to get the more permissive count. Based on the counts, a differential expression (DE) analysis in R version 3.3.2 was performed, between conditions at the different time points using both DESeq2 (Love et al., 2014) and edgeR (Robinson et al., 2010). The threshold for DE analysis cutoffs was defined as log10|FC| . 1, and adjusted (adj) p value , 0.05. Data from results for the differentially expressed genes are displayed in heatmaps in Extended Data Figure 1-1, and data-mining scripts and processing is displayed in a repository in the following link: https://github.com/ ArlesUrrutia/mec4dvsOPvsHT115. Results for adj p value and Log fold change (Log2FC) were cleaned for unmeasurable genes and then analyzed in Python version 3.9 for Volcano plot observations of both DE analyses (Extended Data Fig. 1-1). Data and data-mining results for the differentially expressed genes are displayed in Extended Data Table 1-2.

Enrichment analysis
Gene ontology (GO) analysis was performed using the enrichment tool in WormBase (Angeles-Albores et al., 2016. GExplore (Hutter et al., 2009;Boeck et al., 2016) was used to find, among our set of RNA interference (RNAi)-positive genes, those that are highly expressed in the dauer stage.

Cultures without calcium
To obtain calcium-free media, CaCl 2 was not added to the NGM media. Bacteria used to feed animals was grown over a day with the addition of EGTA 500 mM (Winkler) in LB broth (Calixto et al., 2012).

Dauer synchronization
After the detection of a few dauers by direct observation, plates were washed using 1% SDS, and the liquid containing mixed-stage animals was placed on an Eppendorf tube. The tube was centrifuged for 2 min at 2000 rpm, and the supernatant was removed. The pellet of nematodes was washed for 15 min with a solution of SDS 1% with 2,5 mg/ml carbenicillin (PhytoTechnology Laboratories) and 25 mg/ml amphotericin B (Fungizone, Thermo Fisher Scientific). Afterward, animals were washed three more times with sterile distilled water and antibiotics. The pellet was placed on NGM plates, and after 1 h the portion of agar where the pellet was placed was cut off using a stainless steel spatula then washed with the same mixture of sterile distilled water and antibiotics. Animals were centrifuged for 2 min at 4500 rpm and placed in cell culture plates (TrueLine) with distilled water and antibiotics, as mentioned before. The media were replaced every 2 weeks.

RNA interference by feeding
Bacterial clones from Ahringer Library were taken from glycerol stocks and grown overnight on LB plates containing tetracycline (12.5 mg/ml; PhytoTechnology Laboratories). The next morning, a chunk of bacterial lawn was grown on liquid LB containing carbenicillin (50 mg/ml; PhytoTechnology Laboratories) for 8 h. NGM plates were prepared to add 1 mM IPTG (isopropyl b -D-1-thiogalactopyranoside), and 400 ml of bacterial growth was seeded. As a control for RNAi, unc-22 dsRNA was used, which renders animals uncoordinated (Unc).

RNAi in developing animals
Thirty to sixty newly hatched (0-2 h posthatching) L1s were placed on plates containing dsRNA expressing E. coli HT115 bacteria. Larvae and adults were removed with M9 from plates full of laid embryos. Embryos remained attached to the plates and were allowed to hatch for 2 h. L1 larvae were picked 0-2 h posthatching with a mouth pipette in M9 and placed on experimental NGM plates seeded with the desired bacterial clones. Silencing was tested on the TU2773 strain (systemic, non-neuronal RNAi), and WCH6 [touch receptor neuron (TRN)-specific RNAi; Calixto et al., 2010Calixto et al., , 2012. The morphology of the Anterior Lateral Microtubule (AVM) touch neuron was scored at 72 h posthatching.

RNAi in dauers
To avoid contamination, 30-90 animals were seeded on five plates per RNA condition after hypochlorite treatment of gravid hermaphrodites. By direct observation, we detected the first day of dauers on plates. Nematodes were synchronized on day 2 or 3 depending on the number of dauers required for observation. We followed the protocol described above as dauer synchronization. After some animals crawled out of the drop, 25 individuals were picked for observation under the microscope.

Scoring of neuronal integrity
For morphologic evaluation, worms were mounted on 2% agarose pads. Dauers were mounted and paralyzed with 20 mM levamisole, and developing animals with 1 mM levamisole. Morphologic categories were assigned using the same criteria as in the study by Urrutia et al. (2020). Neurons with full-length axons, as well as those with anterior processes that passed the point of bifurcation to the nerve ring, were classified as AxW. Axons with only a process connected to the nerve ring were classified as AxL, and those that did not reach the bifurcation to the nerve ring were classified as AxT. Lack of axon and soma only was classified as AxØ, and the total absence was indicated as AxØ-S. For simplicity, all graphs show only the AxW category.

Microscopy and photography
Images were taken using a Remote Pro DSLR (Breeze Systems), a camera (Rebel T3i, Canon), and a fluorescence microscope (Eclipse Ni-U, Nikon). Configuration was set up at 1/10 exposure time and ISO correction at 200 for fluorescence images.

Quantitative measurements of mitochondrial morphology
Thirty to sixty newly hatched js609 or WCH42 animals were placed on plates seeded with E. coli OP50 or E. coli HT115, and images of the mitochondria of AVM axons of L2 animals were taken after 24 h. Dauers were obtained 1 week later from the same plate by 1% SDS treatment, placed on sterile NGM plates devoid from bacteria to allow dauers to crawl away from the 1% SDS drop. Living dauers were mounted on 2% agarose pads, and photographs were taken at different focal points to register all mitochondria in each AVM. Each set was measured separately using ImageJ, and each mitochondrion was registered independently. First, we used a Neubauer chamber to standardize the observed length in pixels from the image. Then, using the line tool from ImageJ, we measured the number of pixels that each mitochondrion had in the image. With a scale entered in ImageJ, we obtained the length in micrometers. Each value was recorded in Excel, and, from that, the average value and the number of mitochondria in each axon were calculated.
Mitochondria ,2 mm were classified as fragmented, between 2 and 4 mm as intermediate, and .4 mm mitochondria as filamentous (Neve et al., 2020).

Experimental design and statistical analyses Functional validation by RNAi
The functional validation of the genes identified as upregulated was tested by feeding animals with dsRNA (RNAi). RNAi experiments used unc-22 dsRNA as control, which produces twitching in nematodes. When the plate has .70% of Unc animals, the experiment is considered valid. Statistical analysis is performed using an E. coli HT115 that does not express dsRNA as a control (oneway ANOVA).

Calcium depletion
Calcium chelators such as EGTA can reduce the degeneration rate in mec-4d animals (Calixto et al., 2012). Animals cultured in absence of environmental calcium were compared in a two-way ANOVA test to animals grown in NGM (CaCl 2 , 1 mM) at 72 h after hatching, given the same bacterial diet. Similar comparisons are performed for 1-to 2-week-old dauers in the absence of calcium using the same type of test. The role of intracellular calcium transporters like sca-1 and mcu-1 in neuroprotection was tested using RNAi and compared with an E. coli HT115 control for developing (72 h posthatching) and dauer animals. For developing animals, one-way ANOVA was used, and for dauers, a two-way ANOVA was used.

Mitochondrial measurements
Analysis of mitochondrial number and length was performed using one-way ANOVA comparisons in wild-type and mec-4d genetic backgrounds. Specific comparisons included evaluating animals at the same developmental stage (L2 or dauer) in different bacterial diets and comparing L2 with dauers within the same diet. For the functional testing of mitochondrial genes, RNAi was performed in TRN-specific and systemic strains carrying the mec-4d mutation and the scoring was performed at 72 h posthatching. The E. coli HT115 diet served as the control for each gene tested. One-way ANOVA was used to assess both the treatment and control groups. Pearson's correlation analysis was used to examine the relationship between mitochondrial number and length with the percentage of wild-type axons (AxW) observed in mec-4d animals L2 and dauers.
All experiments were performed at least three times (three biological replicas, started on different days and from different parental nematodes). Each biological replica contained a triplicate (three technical replicas). Statistical evaluation was performed by a one-way or two-way ANOVA with post hoc analyses.
Criteria for data exclusion were as follows: we excluded experimental replicas when there was contamination with unwanted bacteria or fungi on the nematode plates or when bacteria had been almost or completely consumed.

Transcriptomic profiling of diet induced neuroprotection in developing C. elegans
Intestinal bacteria and their metabolites can define the degeneration rate of TRNs of developing C. elegans expressing the mec-4d degenerin ( Fig. 1A; Urrutia et al., 2020). Specifically, GABA and lactate-producing E. coli HT115 are protective in contrast with E. coli OP50, which does not produce the metabolites (Fig. 1B). To identify the nematode genes that may underlie diet-induced neuroprotection, we performed RNAseq of mec-4d animals feeding on E. coli HT115 and E. coli OP50 at 12, 24, and 48 h posthatching. Differential expression analysis was performed comparing animals fed on E. coli HT115 with those fed on E. coli OP50 at each time point. Ninety-three genes were differentially expressed, most of which were found in earlier development (12 and 24 h; Extended Data Fig. 1-1, Extended Data Table 1-1). Forty genes were upregulated in animals fed E. coli HT115 (Fig. 1C), while 53 were downregulated (Fig. 1D). Upregulated genes were phenotypically enriched (Angeles-Albores et al., 2016) in neuronal development, axonal pathfinding, and axonal outgrowth, as well as in diapause formation (Extended Data Fig. 1-2), and were enriched in the GO terms calcium binding, neuronal development, the unfolded protein response, and biotic interactions, among other (Extended Data Fig. 1-2, Extended Data Table 1-2). This shows that the E. coli HT115 diet increases expression of genes involved in neuronal processes that can be detected even when the pool of RNAs used for the analysis is from the whole nematode and not a neuron-specific transcriptome.  contrast with our analysis performed in mec-4d at three different larval stages (L1, L2-L3, and L4). Moreover, GO and phenotype enrichment of previous data did not contain any neuronally enriched functions (Extended Data  Table 1-3), suggesting that the protective diet increased the expression of ad hoc genes required for repair in mec-4d background. On the other hand, downregulated genes in E. coli HT115 were associated with body morphology phenotypes (Extended Data Fig. 1-2) and with GO terms such as ATP synthesis and other functions that reside in the mitochondria, in addition to cuticle development (Extended Data Fig. 1-2). Interestingly, previous transcriptomics coincide with enrichment in mitochondrial processes, although the genes are not shared (MacNeil et al., 2013;Revtovich et al., 2019), suggesting that E. coli HT115 has a role over mitochondrial function regardless of genetic background.
We tested whether genes that increase their expression in mec-4d fed on E. coli HT115 contribute to neuroprotection in vivo. This can be functionally tested by targeted silencing using RNAi interference (Fraser et al., 2000;Kamath et al., 2003). We used E. coli HT115 expressing dsRNA for 31 of the 40 upregulated genes to study their contribution to neuroprotection in mec-4d animals. The remaining nine clones were not available in the feeding RNAi library or caused embryonic lethality (Fraser et al., 2000;Kamath et al., 2003). To dissect whether the function of these genes is required in the touch neurons or systemically, we used strains with TRN-specific and systemic RNAi (Calixto et al., 2010). Fifteen dsRNAs caused a decrease in neuroprotection when silenced in non-neuronal tissues (Fig. 1E,G). Of those, seven were needed for protection specifically in the TRNs (Fig. 1F,G). A large proportion of genes necessary for neuroprotection in mec-4d animals are neuronally expressed, including the TRNs (Table 1; Taylor et al., 2021), and show functional clustering (Angeles-Albores et al., 2016) in neuronal phenotypes (Fig. 1H). We compared the functional enrichment provided by genes only required systemically with those required in the TRNs and found a large overlap in neuronal phenotypes despite having come from different gene pools (Fig. 1I). Clustering by GO of RNAi-positive genes also showed enrichment in neuronal categories (Extended Data Fig. 1-3). GO enrichment showed TRNspecific axon guidance and axon projection terms, while system-specific categories were enriched in stress responses to biotic stimulus and to incorrectly folded proteins. The systemic and TRN enrichment in GO terms overlapped in neuronal development and immune defenses to microbes (Extended Data Fig. 1-3). This is coherent with a systemic contribution to lowering cellular stress while neuronal autonomous processes are clustered in categories related to neuronal repair.   Diapause entry is also strongly neuroprotective (Caneo et al., 2019). We compared the transcriptomic results shown above to available data on dauer gene expression obtained by others (Hutter et al., 2009;Boeck et al., 2016). Four of the fifteen genes required for neuroprotection are also upregulated in dauers sax-2, hsp-16.11, T22B7.3, and T12D8.5 (Boeck et al., 2016), suggesting that there is a common neuroprotective gene pool. These genes cluster in categories mainly related to neurogenesis and the unfolded protein response, probably to regulate cellular homeostasis after the early stress response (Extended Data Fig. 1-4). Genes that are not shared with dauers include categories mostly related to axonal guidance, development and body morphology development, and interspecies communication (Extended Data Fig. 1-4), suggesting that there are functional distinction between the two gene pools.

Calcium depletion and neuroprotection
Transcriptomics analysis (Fig. 1) revealed that several genes related to calcium homeostasis are differentially expressed in neuroprotective conditions. This prompted us to look further into the calcium contribution to neuroprotection induced by diet and by diapause. E. coli HT115fed mec-4d nematodes growing in media lacking calcium showed a significantly lower percentage of wild-type TRN axons (AxW) at 72 h after hatching ( Fig. 2A), suggesting that calcium also contributes to axonal regrowth. In contrast, AxW counts in animals on the standard E. coli OP50 diet did not differ in low calcium. We next asked whether calcium plays a similar role in diapause-induced regeneration. E. coli OP50 and HT115-fed worms were induced to enter diapause by bacterial food exhaustion. Calcium removal in both dauer populations caused a significant decrease in wild-type axons (Fig. 2B). Notably, the extent of axonal regrowth and reduction in control and calciumdepleted conditions, respectively, did not differ between the two groups of dauers, suggesting that feeding on E. coli HT115 did not mask or attenuate diapause benefits. Together, the results support the idea that, although both microbiota-induced and diapause-induced neuroprotective processes require calcium, the underlying mechanisms are not completely overlapping.

Mitochondrial and reticular calcium transporters in TRNs are required for neuroprotection
Given the observed dependency on calcium, we next asked whether calcium redistribution across different subcellular compartments by specific transporters plays a role in neuroprotection. For the E. coli HT115 diet during development and in diapause, we evaluated the requirement for mcu-1, the mitochondrial calcium uniporter, and sca-1, the sarco endoplasmic reticulum calcium ATPase (SERCA; Betzer et al., 2018;Csordás et al., 2018;Sarasija et al., 2018;Wang et al., 2019;Calvo-Rodriguez et al., 2020). These two transporters were silenced by feeding mcu-1 and sca-1 dsRNA-expressing bacteria to synchronized L1 mec-4d worms, and we scored neuronal integrity of the AVM TRN either 72 h later for developing animals (Urrutia et al.,

2020) or after the second generation had become dauers (see detailed protocol in Materials and Methods).
At 72 h post-dsRNA treatment during development, only the simultaneous silencing of mcu-1 and sca-1 caused a reduction in AxW in the TRN-specific RNAi strain (Fig. 2C,D), suggesting that either transporter is sufficient for diet-induced neuroprotection. In dauers, on the other hand, silencing either mcu-1 or sca-1 impaired axonal regeneration with TRN-specific or systemic RNAi (Fig. 2E,F). Together, these results suggest that either calcium transporter can compensate for the loss of the other in E. coli HT115-mediated neuroprotection, whereas both are necessary systemically in dauer neuroprotection.

Diapause induces increase in mitochondrial number and size in mec-4d animals
Regeneration under a chronic degenerative stimulus can be energetically demanding (Caneo et al., 2019). Specifically, MEC-4d overactivity imposes an ionic imbalance that can only be repaired by active Na 1 extrusion such as would occur through the NaK ATPase (Davis et al., 1995;Calixto, 2015). Because mitochondria are central to cellular energetic control and have a core role in neuronal cell death (Dawson and Dawson, 2017) and in axonal degeneration in the mec-4d model (Calixto et al., 2012), we examined whether diet-induced and diapauseinduced neuroprotection were associated with mitochondrial changes in TRNs. We measured mitochondrial number and length in the AVM neuron of animals expressing a fluorescent mitochondrial marker (jsIs609:Is[Pmec-4::MLS::gfp]) in both wild-type (Fatouros et al., 2012) and mec-4d backgrounds (Fig. 3A). These quantifications were performed in 1-or 2-week-old dauers and in L2 control animals because dauers have an L2d lineage (Karp and Ambros, 2012). Initially, we measured each mitochondrion and recorded each value separated by sample, then we examined mitochondrial values and categorized them into three distinct groups based on their size: filamentous, intermediate, and fragmented (Neve et al., 2020;Fig. 3).
First, we noted that mec-4d L2 animals have significantly less mitochondria in the TRNs than wild types in either diet (Fig. 3B). Second, regardless of genotype, mitochondria numbers are similar between diets (Fig. 3C,D). These two observations indicate that E. coli HT115 does not exert its neuroprotective effects through rescuing mitochondria reductions. However, mec-4d dauers grown on E. coli OP50, but not on E. coli HT115, have significantly more mitochondria than wild-type animals (Fig. 3B). Third, TRNs in wildtype dauers possess fewer mitochondria compared with their L2 controls (Fig. 3C), supporting the idea that diapause has lower metabolic demands under nondegenerative conditions. This decrease was observed regardless of diet in wild types but varied in mec-4d dauers. TRNs in mutant dauers previously fed with E. coli OP50, but not E. coli HT115, have significantly more mitochondria than their respective wild-type dauers (Fig. 3B) and L2 mutants (Fig.  3D), suggesting that mitochondria increase may be a mechanism in dauers to alleviate homeostatic stress in mec-4d TRNs in a nonprotective diet.
Nonfragmented mitochondria are considered to be optimized for function and may protect against neuronal damage (Chen et al., 2007;Kiryu-Seo and Kiyama, 2019;Wang et al., 2021). We wondered whether they play a role in diet-induced or diapause-induced neuroprotection. Each mitochondrion in the AVM axons was classified as nonfragmented (filamentous, intermediate) or fragmented. We represented the percentage of animals with the largest mitochondria (nonfragmented) in each condition (Fig.  3E-G). First, TRNs in L2 mec-4d mutants have smaller mitochondria than wild-type TRNs in either diet, suggesting that diet does not compensate for mitochondrial size changes that are associated with mec-4d in developing animals (Fig. 3E). In dauers, however, the number of nonfragmented mitochondria is similar between genotypes (Fig. 3E). Wild-type L2 and dauers have equal numbers of nonfragmented mitochondria, independent of the diet, suggesting that, in the absence of a prodegenerative stimulus, TRN mitochondria do not change in size (Fig.  3F). Importantly, mec-4d dauers have larger mitochondria than mec-4d L2 controls on E. coli OP50 (Fig. 3G), and there is a similar trend on E. coli HT115 (adj p-value, 0.06), supporting the idea that mitochondrial enlargement may account for diapause-mediated regeneration. We further evaluated the relationship between the percentage of wild-type axons observed and mitochondrial parameters. Linear regression analysis revealed a strong relationship between AxW morphology and mitochondrial length for animals on both diets (Fig. 3H,I), although no clear relationship between mitochondrial numbers and axonal regeneration can be stated on E coli HT115-fed worms (Fig. 3I). Thus far, our results support the idea that diapause-induced neuroprotection, but not diet-induced neuroprotection, is associated with mitochondrial changes in size and number.
To confirm the lack of association between diet and mitochondrial number and size, we further investigated whether HT115-induced protection requires genes involved in mitochondrial fusion (eat-3, fzo-1; Kanazawa et al., 2008)-mitochondrial fission (drp-1; Lu et al., 2011;Navarro-González et al., 2017;Byrne et al., 2019) dynamics using TRN-specific and systemic RNAi strains as before. Consistent with our previous evaluations, there was no significant effect on neuronal protection for drp-1 or eat-3 dsRNA (Fig. 3G,H). However, the silencing of fzo-1, known to be important for external membrane fusion of mitochondria, significantly decreased the number of wildtype axons. Interestingly, fzo-1 is the ortholog of mitofusins, which in addition to mitochondrial fusion, stabilize the connection between mitochondria and ER, to, for example, transport calcium from the ER to the mitochondria (Decuypere et al., 2011;Michel and Kornmann, 2012;Herrera-Cruz and Simmen, 2017;Rodríguez-Arribas et al., 2017). Thus, fzo-1 loss could be affecting axonal regrowth through these processes.
We tested the relevance of two other genes involved in mitochondrial metabolism, cts-1 (citrate synthase, tricarboxylic acid cycle) and icl-1 (isocitrate lyase, glyoxylate cycle), in axonal regrowth. Both of them are required for neuroprotection induced by diet in the TRN-specific RNAi strain (Fig. 3G,H), showing that metabolic function of mitochondria is critical for dietary neuronal protection in a cell-autonomous manner.
Together, our results suggest that diet-induced neuroprotection requires mitochondrial function in metabolism while diapause-induced neuroprotection may increase mitochondrial number and size.

Discussion
Food availability and microbiota composition affect neuronal integrity and survival (Caneo et al., 2019;Liu et al., 2020;Urrutia et al., 2020;Shandilya et al., 2022;Zhang et al., 2022;Urquiza-Zurich et al., 2023). To find commonalities between these two neuroprotective conditions, we study gene expression, calcium contribution, and mitochondrial parameters in the AVM neuron of C. elegans. Transcriptomics analysis reveals that feeding on neuroprotective bacteria induces the expression of genes required for calcium dynamics, neurogenesis and neuronal function, and dauer formation. Removing extracellular calcium affects both diet-induced and diapause-induced neuroprotection. Interestingly, simultaneous silencing of both mcu-1 and sca-1 is necessary to prevent axonal regrowth induced by diet, whereas perturbation of either calcium transporter is sufficient to disrupt diapause-conferred neuroprotection. Moreover, larger mitochondria in the TRNs are promoted by diapause but not by protective microbiota.
A neuroprotective gene pool?
The TRNs are highly regenerative cells capable of regrowth after axotomy (Wu et al., 2007) or under protective treatment in chronic models of damage such as the mec-4d degenerin (Caneo et al., 2019;Urrutia et al., 2020). Microbiota protection occurs early in development and is long-lasting (Urrutia et al., 2020), which is coherent with the early expression of neuroprotective genes found here. But how does the expression of these genes contribute to creating a protective environment and promote neuronal regrowth? We hypothesize that there are two components working in parallel, one systemic and one in the TRNs. Based on our transcriptomics analysis, we think that the systemic component includes direct effects of bacteria in the intestine by the secretion of specific metabolites (MacNeil et al., 2013), and bidirectional communication through immune genes such as clc-1, irg-5, and cnc-4. It also includes the expression of genes that operate outside the TRNs to create a propitious environment for regeneration (crh-2, sax-2), lowering redox stress and alleviating the energetic demand of repairing tissues (hsp-16.11). The cell-autonomous effect relies on TRN expression of genes that promote neuronal growth in situ on favorable conditions in the extracellular milieu (ten-1, hrg-2). Metabolites produced by E. coli HT115, such as GABA and lactate, could improve the stress status of the intestine of nematodes previously reared on E. coli OP50. For example, GABA production and extrusion through the GABA shunt (Feehily et al., 2013) is a mechanism used by bacteria to lower the acidic stress of the intestine of the host where they colonize. E. coli HT115, unlike E. coli OP50, contains all enzymes required for GABA production and export (Urrutia et al., 2020). An interesting class of genes that appeared in our screen and others (MacNeil et al., 2013;Revtovich et al., 2019) are the hrg (heme-responsive genes). Free heme is highly reactive and can intercalate in lipid bilayers (Chen et al., 2012). HRG-2, a heme deficiency-responsive membrane protein, regulates heme homeostasis and detoxification (Chen et al., 2012), alleviating cellular redox stress. This function is required for E. coli HT115-induced neuroprotection systemically and in the TRNs.
In parallel, E. coli HT115 metabolites could directly travel to the TRNs using specific transporters like UNC-47 or SNF-5; or could trigger signaling cascades such as those initiated by the transcription factor DAF-16/FOXO or through specific GABA receptors like GAB-1 or LGC-37, all of which are necessary for E. coli HT115 protection (Urrutia et al., 2020). Others studies have shown that E. coli HT115 lowers cellular stress in C. elegans by counteracting vitamin B12 deficiency and the toxic accumulation of propionate, most of which improves mitochondrial health (Revtovich et al., 2019). Bacterial metabolites and the induction of a nematode gene pool could then directly lower systemic mitochondrial stress (hsp-16.11) and promote specific functions in adjacent tissues to the TRNs. For example, adhesion molecules such as cadherins mediate cell signaling and neuroregeneration (Hirota et al., 2001;Kanemaru et al., 2013;Friedman et al., 2015;Yulis et al., 2018;Punovuori et al., 2021). The teneurins are transmembrane proteins fundamental for the development of the nervous system (Drabikowski et al., 2005;Mörck et al., 2010;Tucker et al., 2012;Topf and Drabikowski, 2019) and also are neuroprotective Trubiani et al., 2007;Tessarin et al., 2019).
A few genes necessary for E. coli HT115-induced protection are also upregulated in dauers (Hutter et al., 2009;Boeck et al., 2016), suggesting that both signaling processes might share a common gene pool. Previous analysis shows that C. elegans diapause induces large transcriptional changes to accommodate periods of long-lasting starvation (Dalley and Golomb, 1992;Jones et al., 2001;Wang and Kim, 2003). Some of these changes include the overexpression of proregenerative genes such as dlk-1 and the repression of antiregenerative genes such as efa-6 (Chen et al., 2011;Calixto, 2015). A similar scenario could be induced by the protective microbiota.

Calcium contribution
Active maintenance of calcium levels is required to promote regeneration, which directly involves the mitochondrial calcium uniporter mcu-1 and the SERCA pump sca-1 (Sarasija et al., 2018;Wang et al., 2019;Calvo-Rodriguez et al., 2020;Calvo-Rodriguez and Bacskai, 2021). The blockade of mcu-1 prevents cellular neuronal death in the context of Alzheimer's disease (Sarasija et al., 2018;Calvo-Rodriguez and Bacskai, 2021). Here we show that during development the silencing of both mcu-1 and sca-1 is required for a reduction in wild-type axons TRN autonomously, suggesting that they are redundant. It is also possible that an increase in intracellular calcium by itself is not damaging under a protective diet. Intraorganellar stress has been shown to trigger degeneration (Sarasija et al., 2018;Wang et al., 2019).
In diapausing animals, the silencing of mcu-1 or sca-1 halted the regeneration of TRNs. Our results contrast with the simple idea that calcium by itself only damages the cell. Silencing of sca-1 and mcu-1 induces higher cytoplasmic calcium concentration, but only regeneration is impaired, and an increment in neuronal death is not observed when calcium is present in the environment of diapausing animals. Earlier it was proposed that calcium toxicity may be related to an increment in mitochondrial calcium, which promotes oxidant-induced mitochondrial loss of function, ATP depletion, and mitochondrial bursting (Harman and Maxwell, 1995;Calvo-Rodriguez et al., 2020).
The reduction of intracellular calcium transporters does not simply generate an increment of calcium concentrations inside the neuron, but rather alters the subcellular pattern of calcium levels that also prevents mitochondrial bursting (Sarasija et al., 2018;Calvo-Rodriguez et al., 2020;Calvo-Rodriguez and Bacskai, 2021), suggesting that mitochondrial or ER damage is more relevant for the outcome of the cell than the cytoplasmic concentration of calcium.
Regeneration actively requires calcium (Khachaturian, 1989;Chierzi et al., 2005;Ghosh-Roy et al., 2010;Sun et al., 2014). In E. coli HT115 calcium depletion impaired the protection and regrowth of axons, as is observed in dauer-induced regeneration. This suggests that calcium may be crucial during axonal maintenance generally.

Mitochondria and neuroprotection
The role of mitochondrial function in neuronal protection has been linked to the microbiota-metabolites-brain axis (Saint-Georges-Chaumet and Edeas, 2016;Nurrahma et al., 2021). The mitochondria of animals that enter diapause become larger and more numerous in mec-4d animals under prodegenerative pressure. Since energy production is associated with mitochondrial fusion (Skulachev, 2001), longer mitochondria may be key to the correlations found in our work and previous reports (Chen et al., 2007;Knott et al., 2008;Chen and Chan, 2010;Chang et al., 2019;Wang et al., 2019).
Intermittent fasting improves cognitive traits and mitochondrial function (Liu et al., 2020). Dauers have reduced metabolic rates and elevated levels of heat shock proteins, are resistant to oxidative stress, and exhibit a low metabolic rate compared with other larval stages (Anderson, 1982;Burnell, 1989;O'Riordan and Burnell, 1990;Dalley and Golomb, 1992;Penkov et al., 2020). Most of these characteristics are associated with increased mitochondrial functions, such as more ATP availability or a reduction of toxicity (Fariss et al., 2005;Labbadia et al., 2017). mec-4d dauers have longer and more mitochondria, suggesting that these traits help maintain axons (Knott et al., 2008;Chang et al., 2019). Diapause induction affects mitochondrial physiology (Artal-Sanz and Tavernarakis, 2009;Lourenço et al., 2015;Lourenço and Artal-Sanz, 2021), suggesting that the regeneration of dauers might be a consequence of an increased buffering capacity and improved energy management by larger mitochondria.
drp-1 or eat-3 silencing and consequently impaired mitochondrial fission/fusion did not affect neuroprotection by diet, consistent with our observation that animals fed with E. coli HT115 do not increase the number or size of mitochondria. However, the loss of fzo-1/mitofusin-1, which initiates the fusion of the external mitochondrial membrane that caused a dramatic reduction in AxW morphology, suggesting that mitochondrial structure influences the regenerative processes. Unsurprisingly, the silencing of genes required for mitochondrial metabolic function cts-1 and icl-1 affect neuroprotection in a TRN-autonomous manner, which directly impacts ATP production of the cells and their capacity to respond to stress (Zampese et al., 2022). This supports the idea that changes in the metabolism induced by the microbiota can cause the neuronal protection to be independent of an increase in mitochondrial number or size.