Rag1 immunodeficiency‐induced early aging and senescence in zebrafish are dependent on chronic inflammation and oxidative stress

Abstract In mammals, recombination activating gene 1 (RAG1) plays a crucial role in adaptive immunity, generating a vast range of immunoglobulins. Rag1 −/− zebrafish (Danio rerio) are viable and reach adulthood without obvious signs of infectious disease in standard nonsterile conditions, suggesting that innate immunity could be enhanced to compensate for the lack of adaptive immunity. By using microarray analysis, we confirmed that the expression of immunity‐ and apoptosis‐related genes was increased in the rag1 −/− fish. This tool also allows us to notice alterations of the DNA repair and cell cycle mechanisms in rag1 −/− zebrafish. Several senescence and aging markers were analyzed. In addition to the lower lifespan of rag1 −/− zebrafish compared to their wild‐type (wt) siblings, rag1 −/− showed a higher incidence of cell cycle arrest and apoptosis, a greater amount of phosphorylated histone H2AX, oxidative stress and decline of the antioxidant mechanisms, an upregulated expression and activity of senescence‐related genes and senescence‐associated β‐galactosidase, respectively, diminished telomere length, and abnormal self‐renewal and repair capacities in the retina and liver. Metabolomic analysis also demonstrated clear differences between wt and rag1 −/− fish, as was the deficiency of the antioxidant metabolite l‐acetylcarnitine (ALCAR) in rag1 −/− fish. Therefore, Rag1 activity does not seem to be limited to V(D)J recombination but is also involved in senescence and aging. Furthermore, we confirmed the senolytic effect of ABT‐263, a known senolytic compound and, for the first time, the potential in vivo senolytic activity of the antioxidant agent ALCAR, suggesting that this metabolite is essential to avoid premature aging.

As in other vertebrates, only one functional rag1 locus exists in zebrafish (Willett, Cherry, & Steiner, 1997). Therefore, loss of function at this locus completely blocks Ig gene assembly and presumably removes the adaptive immune system (Wienholds, Schulte-Merker, Walderich, & Plasterk, 2002). Rag1 mutants (rag1 −/− ) present with a point mutation that causes a premature stop codon in the rag1 catalytic domain (Wienholds et al., 2002). In this sense, rag1 −/− zebrafish are comparable to the Rag1 −/− severe combined immunodeficiency (SCID) mouse model, where mutations in multiple genes (including Rag1) create a deficiency in the functional adaptive immune system and a loss of B and T lymphocytes (Mombaerts et al., 1992).
However, similar to humans with the same condition, Rag1 −/− SCID mice present with an increased susceptibility to infections and survive in pathogen-free conditions only, but rag1 −/− zebrafish reach adulthood and become fertile without obvious signs of infectious disease in standard, nonsterile aquarium facilities. In fact, it was recently reported that rag1 −/− zebrafish respond more rapidly to a viral infection and have increased survival (García-Valtanen et al., 2017).
However, we noticed that rag1 −/− fish develop earlier signs of aging and present a decreased lifespan compared to wild-type (wt) fish. This may be due to the enhanced innate immune response needed to compensate for the lack of a full adaptive immune system (Jima et al., 2009) and the associated energy cost (McDade, Georgiev, & Kuzawa, 2016), or the mutation itself may lead to premature aging due to the accumulation of DNA damage over time and the involvement of chronic inflammation and oxidative stress. On the other hand, the absence of mature lymphocytes in rag1 −/− fish could generate a higher accumulation of senescent cells.
It was reported that the RAG family genes, in addition to their role in V(D)J recombination, may be implicated in "controlled" DNA damage for regulating gene expression, cell development, and cell fate in natural killer (NK) cells (Karo, Schatz, & Sun, 2014). Indeed, in mice, RAG-deficient NK cells are highly cytolytic, present with an increased likelihood for apoptosis and have high levels of phosphorylated histone H2AX (γ-H2AX), which is indicative of genomic instability; more importantly, these cells are less efficient at repairing genomic breaks during DNA damage (Karo et al., 2014).
In the present work, we used the zebrafish model and the normal development of rag1 −/− fish in nonsterile aquarium conditions to investigate the effect of this mutation on the whole organism.
The zebrafish model is emerging in biomedical research and presents many advantages, including the ability to easily maintain large stocks of fish, fast embryonic development ex utero, high transparency at the early developmental stages, and conserved signaling pathways. All these characteristics make it an excellent model for several human diseases (Forn-Cuní, Varela, Pereiro, Novoa, & Figueras, 2017;Lieschke & Currie, 2007;Santoriello & Zon, 2012). The aging phenotype of zebrafish has been well described using telomerase-deficient fish (tert −/− ), and this included sustained decrease in cell proliferation, acute apoptotic response, accumulation of DNA damage responses (DDRs), senescence, and, finally, tissue atrophy (Henriques, Carneiro, Tenente, Jacinto, & Ferreira, 2013). We assessed the premature aging in rag1 −/− zebrafish using transcriptomic and metabolomic analyses, telomere length measurements, histological procedures, and cell cycle analysis, as well as other parameters. At the functional level, we validated rag1 −/− zebrafish as a model of senescence by determining the effects of senolytic and antioxidant drugs on senescence regulation. Our results confirm the positive effects of senolytic drugs, such as  and, for the first time, the potential in vivo senolytic activity of L-acetylcarnitine (ALCAR).

| Rag1 −/− zebrafish have increased expression of immune-related genes but decreased transcription of DNA repair genes
Using microarrays, we analyzed the transcriptome of immune-enriched kidney marrow, which is the bone marrow equivalent in fish.
A total of 1,135 genes were differently expressed in the rag1 −/− zebrafish compared with the wt zebrafish; among these genes, 571 were overexpressed and 564 were under-expressed (Dataset S1).
Gene Set Enrichment Analysis (GSEA) (Subramanian et al., 2005) was used to study the processes altered in the transcriptome V(D)J recombination but is also involved in senescence and aging. Furthermore, we confirmed the senolytic effect of ABT-263, a known senolytic compound and, for the first time, the potential in vivo senolytic activity of the antioxidant agent ALCAR, suggesting that this metabolite is essential to avoid premature aging.

K E Y W O R D S
aging, DNA damage, inflammation, oxidative stress, rag1, senescence of the rag1 −/− fish compared with the wt fish (Table 1). The expression of genes related to the inflammation-, interferon alpha-, and interferon gamma-mediated immune responses was enriched in rag1 −/− . By contrast, genes associated with sensing DNA damage and repair were underrepresented in rag1 −/− . A heatmap showing the expression of DNA damage-related genes revealed a decrease in the number of DNA damage response mediators and DNA repair systems (Figure 1a). For example, the genes rad51 and rad52, which encode proteins that repair DSBs (Shen, Cloud, Chen, & Park, 1996), or ddb1 and ddb2, which encode a complex that repairs UV-damaged DNA (Scrima et al., 2008), were expressed at lower levels in rag1 −/− zebrafish. By contrast, some genes, such as tp53, involved in DNA repair were overexpressed in rag1 −/− zebrafish. Tp53 facilitates DNA repair, impacts the activity of various DNA repair systems, and interrupts the cell cycle to avoid the proliferation of genome-damaged cells (Williams & Schumacher, 2016). This induction of the cell cycle arrest is reflected by the decreased transcription of genes related to cell proliferation (E2F targets, G2M checkpoint, MYC targets) in rag1 −/− (Table 1). Other processes related to senescence, such as oxidative phosphorylation (Stöckl, Hütter, Zwerschke, & Jansen-Dürr, 2006), unfolded protein response (Pluquet, Pourtier, & Abbadie, 2015), and fatty acid metabolism (Quijano et al., 2012), are also altered in the transcriptome of the rag1 −/− zebrafish compared with that of the wt fish (Table 1).
Regarding the Toll-like receptors (TLRs) differentially expressed in both zebrafish lines, those directly involved in the recognition of viruses (tlr8b and tlr22) were higher transcribed in rag1 mutants (Dataset S1). Different apoptosis-related genes such as fas ligand (faslg), fas, calpain 1 (capn1), and caspase recruitment domain family member 6 and 9 (card6 and card9) were also affected by the mutation, as also occurs with one of the main pro-inflammatory cytokines, the interleukin 12ba (il12ba) (Dataset S1).
Although no symptoms of infection or outbreaks were observed in rag1 −/− zebrafish that may explain the upregulation of antiviral and pro-inflammatory genes, bacterial load was assayed as the transcript level of 16S rRNA in wt and rag1 −/− control samples (total visceral mass in 4-month-old fish) by qPCR using universal bacterial primers. No significant differences were found between both lines ( Figure S1). TA B L E 1 Gene sets from the Hallmark database were differentially enriched in rag1 −/− and wild-type (wt) zebrafish using Gene Set Enrichment Analysis (GSEA). In the upper part, the gene sets are enriched in rag1 −/− compared with wt fish, and the lower gene sets are enriched in wt compared with rag1 −/− fish

| Rag1 −/− zebrafish die prematurely and show signs of senescence and aging
Survival analysis revealed that rag1 −/− zebrafish die significantly early compared to their wt siblings ( Figure 2a). Whereas rag1 −/− showed an average lifespan of approximately 35 weeks, wt zebrafish survived significantly longer (p < 0.0001) and did not show almost mortality events after this period. At the end of the mortality experiment, most of the surviving rag1 −/− zebrafish showed aging-associated deformities, such as cachexia and/or lordosis, whereas these phenotypes were not observed among the wt individuals ( Figure S2).  (Jung, Bader, & Grune, 2007;Kishi et al., 2008).
Tissue dysfunction is usually accompanied by leukocyte infiltration.
As it was previously observed, muscle and skin from uninfected rag1 −/− zebrafish presented with significant amounts of leukocytes, whereas equivalent sections from wt fish revealed a limited leukocyte presence

| Comparative metabolomics reveals oxidative stress and guanine deficiency in rag1 −/− zebrafish
A metabolomic study was conducted with 10-month-old rag1 −/− and wt fish. A total of 16,544 signals were detected and filtered to obtain 4,620 metabolite signals with high confidence. An analysis of variance was performed to identify a total of 1,304 variables that were significantly different (p < 0.05) between the mutant and wt ex-  acid with a characteristic presence in the brain, retina, and lens of poikilothermic vertebrates but without a known function (Baslow & Guilfoyle, 2015). Nevertheless, NAH seems to be equivalent to the Nacetylaspartate (NAA) metabolite in mammals (Baslow, 1997). Studies have suggested that high NAA levels are related to oxidative stress (Surendran & Bhatnagar, 2011). (c) Guanine was less abundant in the rag1 −/− fish. This nucleotide is present in both DNA and RNA and is the most readily oxidizable of the four DNA bases, due to its low redox potential (Steenken & Jovanovic, 1997). was found in liver from rag1 −/− mutants.

| Rag1 −/− zebrafish show an increased number of senescent cells, which is ameliorated by senolytic and antioxidant agents
We conducted a functional in vivo analysis after adding the senolytic agent ABT-263 or the antioxidant ALCAR, which is decreased in rag1 −/− fish, to the water of 3-month-old wt and rag1 −/− zebrafish.
After one month, the gene expression of the senescence markers tp53 and mdm2, and the inhibitors of the cyclin-dependent kinases (CDKs) (both CIP/KIP -p21, p27, p57-and INK4 inhibitors -cdkn2a/b, cdkn2c-), was analyzed by qPCR, and senescence-associated β-galactosidase (SA-β-gal) activity was measured. Without treatment, the transcript levels of mdm2, p21 and p27 were higher in that rag1 −/− fish than in their wt siblings (Figure 5a). In the wt zebrafish, ABT-263 reduced the mRNA levels of tp53, mdm2, and p21, whereas ALCAR reduced those of tp53, p27, p57, and cdkn2c ( Figure 5b). In rag1 −/− fish, however, ABT-263 affected the transcript levels of all the tested genes, while ALCAR significantly reduced the tp53, mdm2, p27, cdkn2a/b, and cdkn2c mRNA levels ( Figure 5c). The administration of these treatments to rag1 mutants was able to reduce the expression of the analyzed genes to lower levels than that observed in wt fish ( Figure S3).
These results were further analyzed using one of the most widely used biomarkers for replicative senescence detection: Cells were histochemically stained with the artificial substrate X-gal at pH 6.0 (Dimri et al., 1995). The external SA-β-gal staining analysis revealed that SA-β-gal activity was higher in the skin of rag1 −/− fish than in that of wt zebrafish (Figure 5d in the ALCAR-treated fish, suggesting that oxidative stress is a factor that may induce cellular senescence in zebrafish skin. The expression of different senescence-associated secretory phenotype (SASP) factors was also analyzed in wt and rag1 −/− at different ages by qPCR. The molecules secreted by senescent cells vary depending on cell type, but some of the most commonly secreted are interleukin 1b (il1b), tumor necrosis factor a (tnfa), interleukin 6 (il6), and colony-stimulating factor 1b (csf1b). Although in the microarray analysis only csf1b was significantly differentially expressed between wt and rag1 −/− , qPCR revealed a higher expression of il1b, tnfa, and csf1b in rag1 mutants compared to wt at the higher ages ( Figure 6a). Surprisingly, the cytokine il6 was not differentially expressed between both lines at any of the tested ages, and the lower expression values were observed for the oldest age groups ( Figure 6a). Immunofluorescence detection of Il1b in different liver sections of one-year-old zebrafish revealed that, whereas this cytokine was almost undetectable in wt, rag1 −/− zebrafish showed numerous Il1b-positive hepatocytes ( Figure 6b).
Next, we wanted to test the effect of ABT-263 and ALCAR in the expression of these cytokines classically attributed to the SASP and with a higher transcription level in rag1 −/− , but also in the expression of some of the immune-related genes checked above (caspa, mxa, mxe, and shisa5). The results showed that the higher expression of immune genes observed in rag1 −/− zebrafish is not a consequence of the accumulation of senescent cells in these animals, and ABT-263 and ALCAR only reduced the mRNA levels of il1b (Figure 6c).  et al., 2009). This ability may be related to the fact that immunity is an additive system throughout evolution, and in species with less advanced adaptive immunity, such as fish, innate immunity becomes more important (Tort, Balasch, & MacKenzie, 2003). In this work, we report an additional crucial in vivo role for Rag1 in senescence, demonstrating that the rag1 gene mutation induces premature aging in zebrafish, which is clearly reflected in the histological retina and liver sections from one-year-old zebrafish and leads to a shorter lifespan.

| D ISCUSS I ON
A simple explanation for the accelerated aging observed in these mutants may be that rag1 −/− fish pay a high maintenance cost, relying on only the innate immune response. Therefore, the increased energy resources required to fight against potential pathogens leads to faster aging. A few studies have tried to assess the energy cost of innate versus adaptive immunity, concluding that the metabolic costs of adaptive immunity are relatively low compared with those of innate immunity (McDade et al., 2016;Råberg et al., 2002). Another factor that may be linked to premature aging is inflammation: The higher innate immune activity of rag1 −/− fish may lead to a higher inflammatory status, thereby inducing oxidative stress and reducing the antioxidant capacity (Khansari, Shakiba, & Mahmoudi, 2009).
Nevertheless, recent observations revealed an unexpected role for RAG proteins that is linked to cell fitness and DNA repair in NK cells (Karo et al., 2014). The authors propose that RAG endonuclease can generate DNA breaks outside of V(D)J recombination with the purpose of efficiently developing DNA repair mechanisms and enhancing the maintenance of the genome stability, which is required for creating long-lived NK cells (Karo et al., 2014). If this observation is extrapolated to a whole organism, rag1 −/− zebrafish may be accumulating DNA damage due to the absence of the RAG endonuclease activity.
On the other hand, the absence of mature CD4 + T lymphocytes due to the lack of adaptive immunity, and a reduced lifespan of the F I G U R E 4 Comparative metabolomics between rag1 −/− and wt zebrafish and analysis of their oxidative status. (a) Plot in a twodimensional Cartesian coordinate system, with the axes (principal components, PC) representing the greatest variations in the data from mutant (blue circles) and wild-type (wt; green diamonds) metabolite extracts. Two injections of the rag1 −/− quality control (QC; red squares), wt QC (pink triangles), and QC of all the samples (black circles) are represented in the plot, and 95% confidence ellipses are also included. QC samples for each group (rag1 −/− QC, wt QC) are placed next to the middle of each cluster, assuring good data quality; in the same line, the QC ( (c) Effect of the one-month-long treatment with ABT-263 and ALCAR in the transcription of different genes differentially expressed between one-year-old rag1 −/− and wt zebrafish. The expression level of each gene was normalized to 18s gene expression and is expressed as the fold change with respect to the levels detected in the untreated wt fish. The graphs represent the means ± SEM of five independent biological replicates. Significant differences between wt and rag1 −/− fish at each sampling point are displayed as *(0.01 < p < 0.05) NK cells described by Karo et al. (2014), could be facilitating the accumulation of senescent cells in the organism. Indeed, it has been reported that these cell types, together with macrophages, play a pivotal role in the clearance of senescent cells (Hoenicke & Zender, 2012).
Therefore, three hypotheses may explain the premature senes- Because RAG is able to interact with a broad array of places on the genome outside of antigen receptor and immunoglobulin loci, the RAG-mediated cleavage at these sites might also generate epigenetic or gene expression modifications (Lescale & Deriano, 2017).
We cannot rule out the possibility that the absence of these alternative actions of RAG in the mutant fish could be affecting their phenotype.
Interestingly, IFN signaling activation has been previously linked with the stress response to DNA damage, and in this context, IFN activity could be related to the ability of antiviral IFNs to inhibit proliferation in cancerous cells and to promote apoptosis (Brzostek-Racine, Gordon, Scoy, & Reich, 2011;Yu et al., 2015). Furthermore, senescent cells acquire a senescence-associated secretory phenotype (SASP), which includes the secretion of pro-inflammatory cytokines, including IL-6 that promote chronic inflammation and consequently accelerate aging-associated dysfunctions (Childs, Durik, Baker, & Deursen, 2015;Freund, Orjalo, Desprez, & Campisi, 2010). Accordingly, inflammatory and IL-6 signaling pathways were The transcriptome analysis also revealed that the transcription of genes related to DNA repair and cell proliferation are reduced in rag1 −/− zebrafish. The increased accumulation of DNA damage in rag1 −/− fish is reflected in the increased percentage of γ-H2AX-positive cells. In response to severe DNA damage, cells undergo senescence, which implicates cell cycle arrest (Blagosklonny, 2012). The tumor suppressor tp53 is more highly expressed in one-year-old rag1 −/− zebrafish than in wt zebrafish. This gene encodes an important cell cycle checkpoint protein that regulates cell cycle progression from G1 to S phase (Smith & Pereira-Smith, 1996;Stein & Dulic, 1995  Telomeres consist of short guanine-rich repeats, and telomere shortening is a marker of senescence and aging (Bernadotte, Mikhelson, & Spivak, 2016;Kawanishi & Oikawa, 2004). When telomere length was analyzed in rag1 −/− and wt zebrafish, significant telomere shortening was observed in one-year-old rag1 −/− fish.
Interestingly, guanine was one of the decreased metabolites in the rag1 −/− zebrafish. With each replication, telomeres are shortened due to the inability of DNA-polymerase to work on single-stranded 3′ ends, limiting the total number of cell divisions in somatic cells (Bernadotte et al., 2016). Oxidative stress is known to accelerate telomere shortening, especially because guanine residues are more sensitive to reactive oxygen species (ROS)-induced DNA damage (Kawanishi & Oikawa, 2004). Indeed, a higher level of 8-OHdG was observed in rag1 −/− mutants. This higher oxidative status of rag1 −/− fish was also confirmed by the measurement of lipid peroxidation.
Since general DNA damage and telomere shortening are highly induced by oxidative stress, the deficiency of the antioxidant metabolite ALCAR and the reduced activity of the antioxidant enzymes in rag1 −/− fish may be promoting these findings. The antioxidant activity of organisms declines in an age-dependent manner (Shigenaga, Hagen, & Ames, 1994;Tian, Cai, & Wei, 1998). Therefore, a parallel increase in ROS production and decrease in enzymatic and nonenzymatic antioxidants occurs during aging. Specifically, the content of carnitine and its derivatives, such as ALCAR, has been shown to decrease with the aging process (Costell, O'Connor, & Grisolía, 1989;Maccari, Arseni, Chiodi, Ramacci, & Angelucci, 1990), and this effect has been associated with some age-related diseases. ALCAR supplementation has been shown to reduce oxidative stress and to ameliorate related diseases and disorders (Hagen et al., 2002;Pettegrew, Levine, & McClure, 2000).
In this work, we observed increased SA-β-gal activity in the skin of rag1 −/− fish. Because aged dermis accumulates a high quantity of senescent cells (Dimri et al., 1995;Kishi et al., 2003), in zebrafish, SA-β-gal staining measurements have typically been conducted in the skin (Kishi et al., 2003(Kishi et al., , 2008Tsai et al., 2007). The strong infiltration of macrophages observed in the rag1 −/− zebrafish skin may be due to the increased accumulation of senescent cells in this tissue, since senescent cells release inflammatory chemokines and cytokines that attract leukocytes (Xue et al., 2007). We evaluated the effect of ALCAR supplementation on senescence, and the senolytic agent ABT-263 was used as a positive control. ABT-263 is a potent inhibitor of the Bcl-2 family of anti-apoptotic factors and is a promising chemotherapeutic drug for cancer cell treatment (Oltersdorf et al., 2005;Tse et al., 2008), but currently, ABT-263 is considered a highly selective senolytic agent (Chang et al., 2016;Pan et al., 2017, 62).
In addition, in vitro L-carnitine or ALCAR administration has been proven to prevent senescent cell formation during aging due to the antioxidant activity of these compounds (Mobarak, Fathi, Farahzadi, Zarghami, & Javanmardi, 2017;Wu et al., 2014). ABT-263 reduced tp53, mdm2, and p21 expression in wt, but this effect was extended As a summary of this work, we can conclude the following: (a) the rag1 −/− mutation accelerates aging, which is clearly supported by the numerous aging and senescence markers evaluated; (b) zebrafish are a very useful model for studying senescence in vivo; and (c) our study reveals that ALCAR could be considered a new biomarker for aging and its administration could reduce some agingrelated parameters, paving the way for future studies that aim to confirm the possible therapeutic use of ALCAR in premature aging syndromes.

| Wt and rag1 −/− zebrafish sampling
To analyze the expression of selected genes during the different developmental stages of wt and rag1 −/− zebrafish, samples were taken at 4 dpf (five biological replicates (BRs) of 10 larvae each) and at 1 (4 BRs of 4 whole fish), 4, 8, and 12 months (4 BR of 4 pooled kidney samples). Samples were stored at −80°C for RNA isolation.
Total RNA was extracted as explained below, and RNA quality was assessed with the Agilent 2,100 Bioanalyzer.

| Transcriptome analyses
Samples from one-year-old wt and rag1 −/− zebrafish were used for microarray hybridization. The 4x44K Zebrafish Gene Expression Microarray (V2, AMADID 019161) from Agilent Technologies was used to analyze differential gene expression. The data were complied with MIAME standards (Brazma et al., 2001). The signal was recorded, processed, and segmented using an Agilent G2565B scanner (Agilent Technologies) with the Agilent Feature Extraction Software (v9.5) protocol GE1-v5_95 and an extended dynamic range, and preprocessing was completed by the Agilent Feature Extraction software v9.5.5.1.
The results for the fluorescence intensity data and quality annotations were imported into GeneSpring GX version 13.0 (Agilent Technologies), and data were analyzed as previously described (Forn-Cuní et al., 2017). Genes differentially expressed between rag1 −/− and wt zebrafish were identified by unpaired t test, and data were considered significant at p ≤ 0.01. The fold change cutoff was set at 2. Raw and normalized data were deposited in the NCBI's Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) and are available under the accession number GSE91397.
Microarray expression values were validated with qPCR analysis of the expression of 8 different genes in the RNA samples used to hybridize the array. qPCR and microarray fold-changes are represented in Fig. S4, and their correlation was analyzed using Pearson's correlation coefficient.
The normalized log intensity ratios obtained were analyzed using GSEA (Subramanian et al., 2005). To characterize the differential processes enriched in the transcriptomes, the curated hallmark collection (Liberzon et al., 2015) of the Molecular Signature Database was used.

SA-β-gal staining
Three-month-old wt and rag1 −/− zebrafish were maintained for one month with three different water treatments: ABT-263 (2 µM + 0.1% DMSO), ALCAR (0.5 mM + 0.1% DMSO), or vehicle alone (0.1% DMSO). Tank water was refreshed every three days. After this period, twenty individuals from each group were sacrificed, and the total visceral mass was sampled, constituting 5 BR (four individuals/BR). The expression of the senescence-related genes p21, tp53, and mdm2 was analyzed by qPCR in these samples. The control samples were also used for analyzing the expression of nrf2, sod3b, and cat.
For SA-β-gal staining, the same treatments were conducted in one- year-old wt and rag1 −/− zebrafish. After one month, fish were fixed in

| RNA isolation, cDNA transcription, and qPCR
Total RNA isolation was performed using the Maxwell 16 LEV Simply RNA Tissue Kit (Promega) according to the manufacturer's instructions. cDNA synthesis was conducted with SuperScript II Reverse Transcriptase (Invitrogen). Specific qPCR primers were designed, and amplification efficiency was calculated using seven, serial twofold dilutions of cDNA from unstimulated zebrafish with the threshold cycle (CT) slope method (Pfaffl, 2001). The primer sequences are listed in Table S1. For the analysis of the bacterial load in wt and rag1 −/− zebrafish samples, a pair of universal primers for the 16S ribosomal RNA (rRNA) gene was used (F: 5'-AGGATTAGATACCCTGGTAGTCCA-3'; R: 5'-ACTTAACCCAACATCTCACGAGAC-3') (Campbell et al., 1995). qPCR amplifications were performed as previously described (Pereiro et al., 2015). The relative gene expression level was normalized using 18S ribosomal RNA as a reference gene following the Pfaffl method (Pfaffl, 2001).

H2AX-positive cells and TUNEL assays in kidney cells
Three adult wt fish and 6 rag1 −/− fish (1 year and 9 months old) were used in this experiment. Kidneys were sampled and individually disaggregated by being passed through a 40-µm mesh. Cells were fixed with 2% paraformaldehyde for 15 min at 4°C and washed in PBS.
Four microscope slides for each individual kidney were prepared by sample centrifugation at 800 rpm for 5 min in a Shandon Cytospin 4 cytocentrifuge (Thermo Scientific). Samples were used for immunocytochemical quantification of the histone γ-H2AX and apoptosis quantification by TUNEL assays.

| FLOW-FISH
Zebrafish were euthanized, briefly rinsed in 0.5% chilled bleach, crushed, and incubated in PBS supplemented with antibiotics for 30 min. Then, the samples were centrifuged (600 g, 5 min), incubated in trypsin (0.5 mg/ml)/EDTA (0.1 mg/ml) in PBS for 1 min, centrifuged (600 g, 5 min), and then incubated in collagenase (0.5 mg/ml) in RPMI medium supplemented with CaCl 2 2H 2 O (0.7 mg/ml) for 30 min. The cell suspensions were obtained by pipetting, smashing, and finally filtering the digested tissues through a 100-µm mesh and then were washed and resuspended in PBS. One million cells from each sample were then washed in 2 ml of PBS supplemented with 0.1% BSA. Each sample was divided into two replicate tubes: one pellet was resuspended in 500 µl of hybridization buffer and another in hybridization buffer without FITC-labeled telomeric peptide nucleic acid probe as the negative control. Samples were then denatured for 10 min at 80°C under continuous shaking and hybridized for 2 hr in the dark at RT.
Then, cells were washed twice in a washing solution (70% deionized formamide, 10 mM Tris, 0.1% BSA and 0.1% Tween-20 in dH2O (pH 7.2)). The cells were then centrifuged, resuspended in 500 µl of propidium iodide solution, incubated for 2 hr at RT, stored at 4°C, and analyzed by flow cytometry within the following 48 hr.
Afterward, the fish were dehydrated, embedded in Paraplast Plus and sectioned at a thickness of 5 µm. Sections were stained with H&E or periodic acid-Schiff (PAS) reagent, followed by a slight counterstain with hematoxylin. For immunohistochemistry, sections were immunostained with a 1:300 dilution of anti-Lcp1 antibody (GeneTex) and a 1:100 dilution of anti-Mpx antibody (GeneTex) and were slightly counterstained with hematoxylin. All sections were examined under a Leica microscope equipped with a digital camera Leica DFC 280, and photographs were processed with Leica QWin Pro software.
Confocal images were captured using a TSC SPE confocal microscope (Leica) using the LAS AF software (Leica).

| Comparative metabolomics
Whole, frozen 10-month-old rag1 −/− and wt zebrafish (n = 5/group) were ground in liquid nitrogen using a mortar and pestle, and the fish powder was weighed. A mixture of 80:20 methanol:water was used as the extraction solvent at −80°C. Samples were mixed and allowed to rest for 15 min at 4°C, and then, the supernatant was clarified by centrifugation at 13,000 g for 10 min, removed, and stored at −80°C for LC-MS/MS analysis. Metabolomic analyses were performed using an UHPLC system 1,290 (Agilent Technologies) connected to an Agilent 6,540 quadrupole-time-offlight mass spectrometer (Q/TOF MS) equipped with an orthogonal ESI interface (Agilent Jet Stream) and operating in positive ion mode. The instrument was controlled by a PC running Mass Hunter Workstation software 4.0 from Agilent. The sample was injected into an Agilent ZORBAX C8 Rapid Resolution HD system and maintained at 40°C. Each sample was analyzed in triplicate.
Reverse-phase chromatographic separation was performed using water with 0.1% formic acid as phase A and acetonitrile with 0.1% formic acid as phase B. Accurate TOF MS mass spectra were recorded across the range of 50-1000 m/z at 1.5 spectra/s. After all samples were analyzed in duplicate, raw LC-MS data were processed using R language. Signal detection, deconvolution, and alignment were performed using XCMS and mzMatch packages. Then, a recursive analysis was carried out using the PeakML.

| Measurement of the catalase activity, lipid peroxidation, and oxidative DNA damage
The three experiments were conducted in 1-year-old zebrafish and by following the manufacturer's recommendations. For catalase activity evaluation blood, liver and kidney were sampled (n = 7) and the Catalase Colorimetric Activity Kit (Invitrogen) was used.
The enzymatic activity was referred to the total protein content of each sample. The experiment was replicated three times. Lipid oxidation was analyzed in liver and kidney (n = 7) with the Lipid Peroxidation (MDA) Assay Kit (Sigma). This experiment was replicated twice. Finally, the detection of 8-OHdG was conducted in DNA extracted from liver and kidney (n = 4) with the DNA Damage (8-OHdG) ELISA Kit (StressMarq Biosciences). In this case, the concentration of 8-OHdG was referred to the DNA contain of each sample.