The Btg1 KO and WT mouse strains in the C57BL/6 background were generated as previously described (Farioli-Vecchioli et al., 2012). Genotyping was routinely performed by PCR analysis, using genomic DNA from tail tips as described (Farioli-Vecchioli et al., 2012).

Mice were maintained under standard specific-pathogen-free conditions and were housed in standard cages until P60. Then, mice were randomly assigned to running wheel or standard cages for 12 days. Wheel rotations were recorded daily with an automatic counter. After 12 days, mice were euthanized to dissect the dentate gyrus. The average running wheel distance over the whole experiment (12 days) was 7.55 km/day ± 0.54 (SEM) for WT mice and 7.17 km/day ± 1 (SEM) for KO mice, without significant differences (p = 0.75, Student’s t-test); the total distances run were on average 90.56 km ± 6.53 (SEM) for WT and 86.06 km ± 12.01 (SEM) for KO mice (p = 0.75, n WT mice = 10, n KO mice = 10, Student’s t-test). 15-month-old mice were subjected to the same protocol. All animal procedures were performed on male mice and completed in accordance with the current European (directive 2010/63/EU) Ethical Committee guidelines and the protocol of the Italian Ministry of Health (authorization 442-2016-PR). Btg1 KO mice are available upon request to J.P. Rouault.

A 2-month-old Btg1WT and KO mice were sacrificed by rapid decapitation. The isolation of bilateral dentate gyrus was performed under a stereomicroscope, either in sedentary mice or at the end of the 12-day running protocol, following a described procedure (Hagihara et al., 2009).

Dissected tissues were immediately homogenized in TRIzol Reagent (Invitrogen, San Diego, CA, United States) and total RNA extraction was performed as described previously (Farioli-Vecchioli et al., 2012). Extracted RNA was quantified and assessed for purity using a NanoDrop ND-1000 Spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, United States) and an Agilent 2100 bioanalyzer (Agilent Technologies, Santa Clara, CA, United States). RNAs were subsequently employed for Transcriptome sequencing and/or for real-time PCR experiments. The identity of the dissected dentate gyri was checked by comparing the relative expression of specific markers, i.e., Tdo2, Dsp, Meis2, Tyro3, as indicated in Hagihara et al. (2009) (data not shown). The same procedure was employed to analyze the dentate gyrus of 15-month-old Btg1 WT mice, submitted to exercise or sedentary.

For RNA-sequencing we used total RNA isolated from the dentate gyrus of Btg1 WT or KO mice, either sedentary or submitted to running. Five samples were used in total for each of the four experimental groups, and each sample was obtained by pooling together dentate gyri of two mice.

Purified RNA was delivered to IGA Technology Services¹ for RNA library preparation (by Illumina TruSeq Stranded mRNA Sample Prep kit, following the manufacturer’s instructions), RNA sequencing (Illumina HiSeq2500; 50 bp single-end reads, 6-plex run, 30 M reads) and standard bioinformatic analysis. Briefly, the CASAVA 1.8.2 version of the Illumina pipeline was used to process raw data for both format conversion and de-multiplexing. Reads were aligned on the mm10 genome reference assembly using TopHat/Bowtie tool, transcripts counts were performed via Cufflinks, while the pairwise differential expression analysis of gene transcripts were determined via Cuffdiff in the form of FPKM values (Trapnell et al., 2013; Ghosh and Chan, 2016). Furthermore, the RNA sequencing datasets are deposited at the Gene Expression Omnibus (GEO) repository with Accession Numbers GSE179081².

Gene Ontology enrichment analysis was performed in order to identify GO terms significantly over-represented in genes deregulated in specific comparisons and, as a result, to suggest possible functional characteristics of these genes. Enriched GO terms in the set of genes that are significantly over-expressed or under-expressed in a specific condition may suggest possible mechanisms of regulation or functional pathways that are, respectively, activated or repressed in that condition. As a first step, we built gene and GO term associations considering the known GO annotations for both Homo sapiens and Mus musculus organisms. In detail, GO Annotations for the gene products of both organisms were downloaded from the GO Consortium web site³. The GO terms common to both organisms were associated to the union of the lists of the organism-specific annotated genes and the terms exclusive for one of the organisms were considered together with their specific annotated genes. The genes assayed by RNA-seq in our study were annotated by using their associations with both the GO terms common and exclusive for the two organisms. The p-Values for enrichment were calculated by Fisher’s exact test by using MATLAB analysis code.

To validate RNA sequencing results, total RNA extracted from isolated dentate gyri was reverse-transcribed as previously described (Farioli-Vecchioli et al., 2012). In each of the four groups, two of the samples analyzed were the same already employed for RNA seq and two were from new extractions. Each sample consisted of dentate gyri from two mice.

Real-time PCR was carried out with a 7900HT System (Applied Biosystems) using SYBR Green I dye chemistry in duplicate samples. Relative quantification was performed by the comparative cycle threshold method (Livak and Schmittgen, 2001). The mRNA expression values were normalized to those of the TATA-binding protein gene used as endogenous control. One Btg1 WT sedentary mice was randomly chosen as control calibrator. Average ± SEM values of fold-changes relative to the control sample are shown. Specific RT-PCR primers used were deduced from published murine cDNA sequences and are listed in Supplementary Table 1.

A 2254 nucleotide sequence of DNA immediately upstream of the translation initiation codon of the mouse Snca gene was cloned into the 5′-MluI and 3′-BglII sites of the pGL3 basic vector, containing the sequence coding for the luciferase reporter gene, thus generating pGL3-prSnca-LUC. This promoter sequence corresponds to the active promoter region of Snca (Duplan et al., 2016) and was synthesized by Eurofins MWG (Ebersberg, Germany). All the constructs were verified by full sequencing. Transient co-transfections in HEK293T cells of pGL3-prSnca-LUC with a vector expressing Btg1 (pSCT-Btg1 or pSCT-empty; Ceccarelli et al., 2015) were carried out using lipofectamine (Invitrogen, San Diego, CA, United States) according to the manufacturer’s instructions. We included in all transfections the pRL-TK control reporter (Renilla luciferase driven by the thymidine kinase promoter). Luciferase assays were performed by the Dual-Luciferase reporter assay system (Promega, Madison, WI, United States) 48 h after transfection, according to the manufacturer’s instructions, as described previously (Micheli et al., 2011). We normalized the luciferase activity of each sample (L) for differences in transfection, by assaying the expression of Renilla luciferase (R) in each transfected cell extract. The normalized activity of the reporter gene was thus obtained as L/R. The fold activity was then obtained as ratio between each average value of the normalized reporter activity and the average normalized reporter activity of the corresponding control culture. Student’s t-test on normalized reporter activity values was used for statistical analysis.

The lentiviral vector pCCL-sin-PPT.hPGK.IRES.eGFP.Wpre, kindly provided by L. Naldini (see Dull et al., 1998) was used to express the cDNA sequence of mouse Snca in dentate gyrus cells. The construct pCCL-sin-PPT.hPGK.IRES.eGFP.Wpre-Snca was generated by cloning the optimized cDNA sequence of Snca in the BglII/XbaI sites of the lentiviral vector. The Snca sequence was synthesized by the Eurofins MWG (Ebersberg, Germany) and verified by sequencing.

The G glycoprotein vesicular stomatitis virus-pseudotyped lentiviral particles were generated by CaPh transfection of HEK293T cells with a mixture of either pCCL-sin-PPT.hPGK.IRES.eGFP.Wpre (pCCL-empty, control vector) or pCCL-sin-PPT.hPGK.IRES.eGFP.Wpre-Snca (pCCL-Snca) lentiviral vector and the three plasmids pMDL, pRSV-REV, and pVSV-G (kindly provided by L. Naldini; Dull et al., 1998) required to produce lentiviral particles. Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum, 100 U/mL penicillin G, and 100 μg/mL streptomycin at 37°C in 5% CO₂. 4 × 10⁶ cells were plated in a 10-cm dish 24 h before transfection. Virus-containing medium was harvested 48 and 60 h after transfection, and concentrated by two ultracentrifugation steps. The titers of the viral vectors were in the range of 108 TU/ml. The expression of transduced Snca was verified by real-time PCR and Western blot of RNA and protein from HEK293T infected cells (data not shown). The concentrated virus solution was infused (1.5 μl at 0.2 μl/min) by stereotaxic surgery in the right and left dentate gyrus of P60 Btg1 WT and KO mice (anteroposterior = −2 mm from bregma; lateral = 1.5 mm; ventral = −2.0 mm). Mice were euthanized after 5 days.

Lentiviruses generated were replicant-deficient. Their manipulation and stereotactic injection in mice were approved by the Italian Ministry of Health (authorization RM/IC/Op2/20/005) and performed using BSL-2 and ABSL-2 containment.

Brains were collected after transcardiac perfusion with 4% PFA in PBS 1x and kept overnight in 4% PFA. Brains were then equilibrated in 30% sucrose and cryopreserved at −80°C. Immunohistochemistry was performed on serial free-floating coronal sections cut at 40 μm thickness in a cryostat at −25°C from brains embedded in Tissue-Tek OCT (Sakura Finetek, Torrance, CA, United States). Sections were previously permeabilized with 0.3% TritonX-100 in PBS, and then incubated with primary antibodies with 3% normal donkey serum in 0.3% TritonX-100 in PBS for 16-18 h at 4°C.

Proliferating stem and progenitor cells infected with lentiviruses were visualized by means of a rabbit monoclonal antibody against Ki67 (Invitrogen, San Diego, CA, United States; MA514520; 1:200), and a goat polyclonal antibody against Sox2 (Abcam, Cambridge, United Kingdom; Ab239218; 1:300) and a mouse monoclonal antibody against GFAP (Sigma–Aldrich, St Louis, MO, United States; G6171; 1:200). Apoptotic cells were identified with a rabbit polyclonal antibody against cleaved (activated) Caspase-3 (Cell Signaling Technology, Danvers, MA, United States; 9661; 1:100). GFP-positive cells generated by the lentiviral construct were directly visualized by fluorescence microscopy.

Secondary antibodies used to visualize the antigen were from Jackson ImmunoResearch (West Grove, PA, United States) as follows: a donkey anti-rabbit antiserum conjugated to tetramethylrhodamine isothiocyanate (TRITC) (Ki67 and Caspase-3) and a donkey anti-goat antiserum conjugated to Alexa-647 (Sox2), both incubated for 1 h; a donkey anti-mouse antiserum conjugated to DyLight 405 (GFAP) incubated for 5 h. Nuclei were counterstained by Hoechst 33258 (Sigma–Aldrich, St. Louis, MO, United States; 1 μg/ml in PBS).

For quadruple labeling (Ki67, Sox2, GFAP, and GFP), immunofluorescence for antibodies against Ki67 and Sox2 was performed as described above, and sections were post-fixed with 4% PFA for 10 min. Then, we proceeded with incubation with the anti-GFAP antibody with 3% normal donkey serum in 0.3% TritonX-100 in PBS for 16-18 h at 4°C, followed by incubation with the secondary antibody.

In negative-control sections, the primary antibodies were omitted, to exclude non-specific signal.

Confocal Z-stacks and single plane-images of the immunostained sections were obtained using a TCS SP5 confocal laser scanning microscope (Leica Microsystem, Wetzlar, Germany).

Cells expressing the indicated markers were counted throughout the whole rostrocaudal extent of the dentate gyrus, by analyzing with confocal microscopy one-in-six series of 40-μm free-floating coronal sections (240 μm apart). The total estimated number of positive cells within the dentate gyrus was obtained by multiplying the average number of positive cells per section by the total number of 40-μm sections including the entire dentate gyrus (about 50-60 sections), as described (Jessberger et al., 2005; Farioli-Vecchioli et al., 2008; see also about the cell counting theory: Noori and Fornal, 2011). Therefore, about 8-10 sections (16-20 dentate gyri) per mouse and four animals per group were analyzed. Cell number analyses were performed manually by trained experimenters, in blinded fashion, using the IAS software to register positive cells (Delta Sistemi, Rome, Italy).

Analysis of pairwise comparison of differential gene expression, i.e., the comparison of the mean fold-expression changes between samples of two groups from the different data sets (n = 5 samples per group), was performed with Cuffdiff p-Value (p), corrected for False Delivery Rate to obtain the q-Value (q) (Figures 3, 4).

The statistical significance of the overlap between differentially expressed (DE) genes belonging to different data sets (Table 1), or the enrichment probability of the DE genes in each GO functional class (Table 2 and Supplementary Figures 1, 3, 4), were calculated using the Fisher’s Exact test.

The statistical significance of the expression data obtained by real-time PCR was evaluated by two-way ANOVA when the main effects of both genotype and running were tested (Figure 5), or by one-way ANOVA when running and age effects were analyzed together (Supplementary Figure 7); individual between-group comparisons to test simple effects were carried out by Fisher’s PLSD ANOVA post hoc test. Instead, Student’s t-test was used for analysis of real-time PCR expression data when only two groups were tested (Set C: sedentary Btg1 WT vs. running Btg1 WT) (Figure 2). The immunohistochemical proliferation data of Btg1 WT and Btg1 KO dentate gyri injected with Snca or empty lentivirus were analyzed with two-way ANOVA followed by Fisher’s PLSD ANOVA post hoc test (Figure 6 and Supplementary Figure 6); apoptosis data presented unequal variance, as indicated by Levene’s and Bartlett’s tests, therefore simple effects were analyzed with the non-parametric Mann-Whitney U-test (Figure 6). These analyses were carried out using the Stat View 5.1 software (SAS Institute, Cary, NC, United States) and XLSTAT (Addinsoft, Paris, France). Differences were considered statistically significant at p < 0.05. Real-time PCR and immunohistochemistry data were expressed as mean ± SEM.

Through RNA sequencing, we examined the transcriptomic profiles of the dentate gyrus isolated from 2-month-old mice, either Btg1 WT or Btg1 KO with aging-like neural phenotype such as reduced proliferative capability of stem and progenitor cells, submitted to physical exercise or sedentary. The WT and KO mice were submitted to voluntary running (called here WT RUN and KO RUN) for 12 days, according to a previously used protocol (Farioli-Vecchioli et al., 2014). We checked that no significant differences in running ability occurred between the two genotypes (see Materials and Methods section Mouse cell lines, genotyping and husbandry). At the end of the 12-day exercise schedule the WT RUN and KO RUN dentate gyrus was isolated following a described procedure (Hagihara et al., 2009) and processed for transcriptomic analysis, together with the dentate gyrus from the control sedentary Btg1 WT (WT CTL) and Btg1 KO mice (KO CTL).

The four pairwise comparisons of DE genes presented here were performed with significance threshold of the log2 fold change of gene expression set to p < 0.05. The two pairwise comparisons that we took chiefly in consideration are defined as Set B and Set A, which comprise the genes whose expression is significantly changed by Btg1 knockout, relative to Btg1 WT (Set B: Btg1 KO CTL vs. Btg1 WT CTL, 1107 genes; Figure 1A, Venn diagram), or changed by running in Btg1 KO mice, relative to Btg1 KO sedentary mice (Set A: Btg1 KO RUN vs. Btg1 KO CTL, 545 genes; Figure 1A).

The other two pairwise comparisons taken into consideration are Sets C and D that include the genes with a significant change of expression triggered by running, either in Btg1 WT compared with sedentary (Set C: Btg1 WT RUN vs. Btg1 WT CTL), or in Btg1 KO, compared with Btg1 WT sedentary (Set D: Btg1 KO RUN vs. Btg1 WT CTL). We found 737 DE genes belonging to Set C and 2081 genes to Set D (see Venn diagram, Figure 1A).

If we limit the significance threshold of the log2 fold expression change of DE genes to q < 0.05 (i.e., the p-Value adjusted in order to reduce the False Discovery Rate for false positive), it turns out that 103 genes belong to Set B, 68 genes to Set A, 65 DE genes to Set C and 436 genes to Set D (Venn diagram, Figure 1B).

This report focuses on the DE genes that, in sedentary mice, are significantly altered by Btg1 deletion, relative to Btg1 WT (Set B: Btg1 KO CTL vs. Btg1 WT CTL; Figure 1) and on the DE genes that are changed by running in Btg1 KO mice, relative to Btg1 KO sedentary mice (Set A: Btg1 KO RUN vs. Btg1 KO CTL; Figure 1).

In particular our interest is addressed to identify genes whose expression is modified by Btg1 knockout (Set B) and whose change is reversed by running (Set A). These DE genes should be associated to the phenotype of early neural aging and its recovery as well as to the control of stem cell quiescence. Setting the threshold of significance at p < 0.05, we identified, as mentioned above, 1107 genes in Set B whose expression was significantly changed by Btg1 knockout (either up- or down-regulated), and 545 genes in Set A changed by running. Of all these genes, 42 were down-regulated in Set B, i.e., by Btg1 knockout, and correspondingly up-regulated in Set A, by running (Figure 3 and Supplementary Figure 2). These genes comprise among them Arc, Egr1, Epha6, Fos, Fxyd7, NeuroD6, Npas4, Snca, which are involved in synaptic plasticity, memory and aging (Chauhan and Siegel, 1996 [Fxyd7]; Liu et al., 2004 [Snca]; Uittenbogaard et al., 2010 [NeuroD6]; Mo et al., 2015 [Egr1]; Minatohara et al., 2016 [Fos]; Sun and Lin, 2016 [Npas4]; Nikolaienko et al., 2018 [Arc]; Petschner et al., 2018 [Epha6]). Furthermore, another 42 genes resulted up-regulated in Set B, i.e., by Btg1 knockout, and correspondingly down-regulated in Set A, by running (Figure 4 and Supplementary Figure 2). These genes include Wnt7b, Grik3, Gpr88, involved in stem cells proliferation, dendritic growth, signaling and learning (Papachristou et al., 2014; Benes, 2015 [Grik3]; Meirsman et al., 2016 [Gpr88]; Ferrari et al., 2018 [Wnt7b]).

Moreover, if we select the genes differentially regulated with higher statistical significance (q < 0.05), among all the Btg1 KO-modified genes (set B) also significantly counter-regulated by running (Set A), only four genes emerged that were all down-regulated by Btg1 knockout and up-regulated by running (Figure 3). These DE genes were Arc (set B: q = 0.036, set A: q = 0.015), Fos (set B: q = 0.022, set A: q = 0.015), Npas4 (set B: q = 0.011, set A: q = 0.015), and Snca (set B: q = 0.011, set A: q = 0.015); (Figure 3, genes written in red).

The expression changes of the four genes Arc, Fos, Npas4, and Snca, differentially regulated with statistical significance q < 0.05, were validated by real-time PCR in the dentate gyrus isolated from 2-month-old mice belonging to the four groups analyzed. The real-time PCR expression values of all these genes resulted significantly decreased in Set B and increased in Set A (Figure 5; set B: Btg1 KO CTL vs. Btg1 WT CTL: Arc p = 0.0025, Fos p = 0.0036, Npas4 p = 0.0023, Snca p < 0.0001; set A: Btg1 KO RUN vs. Btg1 KO CTL: Arc p < 0.0001, Fos p < 0.0001, Npas4 p = 0.0002, Snca p = 0.001; two-way ANOVA, PLSD post hoc test). By real-time PCR we further confirmed the expression changes of Egr1 and Ptgs2, genes differentially regulated in both set B and set A with statistical significance p < 0.05, and known to be involved in synaptic plasticity and memory (Figure 5; set B: Btg1 KO CTL vs. Btg1 WT CTL: Egr1 p = 0.0004, Ptgs2 p = 0.0165; set A: Btg1 KO RUN vs. Btg1 KO CTL: Egr1 p = 0.0019, Ptgs2 p = 0.0046; two-way ANOVA, Fisher PLSD post hoc test).

By analyzing the GO databases, we then sought to identify the biological processes that are significantly enriched in the genes either down-regulated (42) or up-regulated (42) in set B in the Btg1 KO aging model, and correspondingly counter-regulated in set A by running (Supplementary Figures 3, 4). In particular the 42 genes down-regulated in Set B and up-regulated in Set A are significantly enriched in biological processes that appear representative of the Btg1 KO phenotype and of the counter-effect of running: a selection is shown in Table 2. These processes are involved mainly in the regulation of synaptic plasticity (Arc, Snca), GABAergic synaptic transmission (Npas4), synaptic vesicle transport (Snca), dendritic spine morphogenesis (Arc), memory (long and short-term memory, Npas4, Arc, Egr1), learning (Arc, Ptgs2, Npas4) and long-term synaptic potentiation (Arc, Snca), aging (Fos, Snca, Ptgs2), cell cycle regulation (Ptgs2/Cox2, Nr4A1/Nur77) and stem cell proliferation (Wnt7b); see Table 2. Therefore, these GO processes impact on the dentate gyrus function of neuronal synaptic activity and plasticity, learning and memory, and on aging.

We will analyze here the biological processes enriched in genes with opposite regulation in the two Sets A and B and relevant to the Btg1 KO phenotype, in order to interpret the transcriptomic signature of the effect of exercise on the Btg1 KO neural aging model.

We will mainly focus on the most statistically significant DE genes, i.e., with q < 0.05, which presented great differential change between the two sets, and which are, notably, all down-regulated in Set B and counter-regulated in Set A (i.e., Snca, Fos, Arc, Npas4; Figure 3), while no DE gene with high statistical significance is up-regulated in Set B and down-regulated in Set A (Figure 4 and Table 2). This suggests that the deletion of Btg1 is most effective in depressing the expression of specific genes while running rescues their down-regulation. We also tested whether, by restoring Snca expression in the dentate gyrus, is possible to rescue the defective processes.