Molecular profiling of aged neural progenitors identifies Dbx2 as a candidate regulator of age‐associated neurogenic decline

Summary Adult neurogenesis declines with aging due to the depletion and functional impairment of neural stem/progenitor cells (NSPCs). An improved understanding of the underlying mechanisms that drive age‐associated neurogenic deficiency could lead to the development of strategies to alleviate cognitive impairment and facilitate neuroregeneration. An essential step towards this aim is to investigate the molecular changes that occur in NSPC aging on a genomewide scale. In this study, we compare the transcriptional, histone methylation and DNA methylation signatures of NSPCs derived from the subventricular zone (SVZ) of young adult (3 months old) and aged (18 months old) mice. Surprisingly, the transcriptional and epigenomic profiles of SVZ‐derived NSPCs are largely unchanged in aged cells. Despite the global similarities, we detect robust age‐dependent changes at several hundred genes and regulatory elements, thereby identifying putative regulators of neurogenic decline. Within this list, the homeobox gene Dbx2 is upregulated in vitro and in vivo, and its promoter region has altered histone and DNA methylation levels, in aged NSPCs. Using functional in vitro assays, we show that elevated Dbx2 expression in young adult NSPCs promotes age‐related phenotypes, including the reduced proliferation of NSPC cultures and the altered transcript levels of age‐associated regulators of NSPC proliferation and differentiation. Depleting Dbx2 in aged NSPCs caused the reverse gene expression changes. Taken together, these results provide new insights into the molecular programmes that are affected during mouse NSPC aging, and uncover a new functional role for Dbx2 in promoting age‐related neurogenic decline.

Dbx2-NSPCs (E-H) shows that ~100% of cells in both cultures are positive for the neural progenitor marker Nestin (red). Images taken from adherent NSPCs (A,B,E,F) and neurospheres that were attached to glass coverlsips (C,D,G,H), following culture in proliferating conditions. GFP signal (green) and Hoechst nuclear staining (blue) are shown. Scale bar, 20 µm. (I-P) Immunofluorescence microscopy with an anti-Dbx2 antibody (red) on adherent NSPCs (I,J,M,N) and attached neurospheres (K,L,O,P), following culture in proliferating conditions. Dbx2 protein levels are higher in Dbx2-NSPCs in comparison with GFP-NSPCs. Hoechst nuclear staining (blue) is shown. Scale bar, 20 µm. (Q) Quantification of the percentage of non-viable cells by trypan blue staining in adherent transgenic NSPCs at the indicated times after cell seeding reveals a slight increase in cell death in Dbx2-NSPCs. Data show mean ± s.e.m; n=3 biological replicates for each time point. (R) Quantification of the percentage of Ki67-positive cells in attached neurospheres using immunofluorescent microscopy reveals a slight decrease in the proportion of cycling cells in Dbx2-NSPCs. Data show mean ± s.e.m.; n=3 biological replicates. Over 2500 cells were scored for each replicate sample.  Table S1. List of differentially expressed genes. Table S2. List of differentially methylated regions. Table S3. List of regions with differential promoter H3K4me3 levels. Table S4. List of regions with differential promoter H3K27me3 levels.

Mouse NSPC culture and manipulation
For neurosphere culture, NSPCs were propagated on uncoated T25 flasks (Corning) in chemicallydefined media as described (Soldati et al. 2012;De Luca et al. 2013;Soldati et al. 2015), supplemented with 20 ng/ml human recombinant Epidermal Growth Factor (R&D systems), 10 ng/ml human recombinant Fibroblast Growth Factor-basic (Peprotech), 1:50 B27 supplement (Invitrogen) and 2 g/ml heparin (Sigma-Aldrich). NSPCs grown as neurospheres were passaged approximately once per week by pelleting the cell aggregates, dissociating with Accutase (Sigma-Aldrich) and seeding on new uncoated flasks at a density of 20-50000 cells/ml. For the establishment of adherent NSPC cultures, dissociated NSPC-derived neurospheres were initially plated in 6-well plates coated with 10 g/ml poly-ornithine (Sigma-Aldrich) and 5 g/ml laminin (Sigma-Aldrich) at a density of 10-15000 cells/cm 2 using the same media as described above without heparin. Subsequent passages were performed once or twice per week using Accutase to detach NSPCs and seeding the cells in coated T25 flasks at a density of 10-20000/cm 2 . For differentiation assays, adherent NSPCs seeded at a density of 20-25000 cells/cm 2 were induced to differentiate by incubating them in media devoid of EGF from the day after seeding.
To generate transgenic adult SVZ NSPCs, 2-4x10 6 cells expanded in adherent conditions were transfected with 2-4 µg of the following plasmids: pTP6-hrGFP control plasmid (Vallier et al. 2004) plasmids were generated for this study, starting from independent derivations of adult SVZ NSPCs.
Dbx2 expression from the epB-Puro-TT-Dbx2 plasmid was induced by treating transgenic NSPCs with 125 ng/ml doxycycline hyclate (Sigma-Aldrich) diluted from a 1 µg/ml stock in PBS. Cells were routinely expanded in the absence of doxycycline, which was added at the time of seeding for experimental assays and maintained in the media until the cells were harvested for endpoint analyses. In each experiment, half of the cultures were left untreated as an internal control.
For neurosphere assays with transgenic NSPCs, adherent transgenic cultures were seeded in uncoated T25 flasks at a density of either 1000 cells/ml or 25000 cells/ml in media for non-adherent culture supplemented with 600 ng/ml of puromycin. Neurospheres arising in low density cultures were plated after 5-7 days post-seeding in 24-well plates containing glass coverslips coated with 20 µg/ml poly-ornithine and 10 µg/ml laminin and left to attach for 4-6 hours. Attached colonies were subsequently fixed with 4% methanol-free formaldehyde (Pierce) in PBS, followed by Hoechst nuclear staining and quantification of the total number of neurospheres and the number of cells per sphere in each sample. Neurospheres generated in high density cultures were harvested after 4-8 days post-seeding and dissociated with Accutase, followed by the quantification of the total number of cells per flask and freezing of cell pellets for qRT-PCR. Differentiation of transgenic NSPCs was initiated by culturing adherent NSPCs in EGF-free media as described above, in the presence of 600 ng/ml of puromycin.

Isolation of Tlx-GFP NSPCs
Tlx-GFP reporter mice were obtained from the GENSAT project (Gong et al., 2003). Mice were housed according to international standard conditions and all animal experiments complied with local (Regierungspräsidium Karlsruhe) and international guidelines for the use of experimental animals. SVZ tissues from three adult (7 month old) and three aged (18 months) C57BL/6 mice were dissected separately in phosphate buffered saline (PBS) and enzymatically dissociated with 0.05% Trypsin/EDTA (Life Technologies) in HBSS containing 2mM glucose at 37°C for 30 min.
During incubation, the tissues were repeatedly triturated with a fire polished Pasteur pipette.

Immunocytochemistry and in situ hybridization
Immunocytochemistry was performed on adherent NSPCs cultured on glass coverslips coated with 10 µg/ml poly-ornithine and 5 µg/ml Laminin or neurospheres attached to glass coverslips as described above. Cells were fixed with 4% formaldehyde in PBS for 20min, followed by washes with PBS, permeabilization with 0.5% Triton X-100 in PBS, incubation with blocking solution (5% Sections were analysed and photographed with a Leica DM microscope equipped with a DFC500 digital camera.

Quantitative RT-PCR
Total RNA extraction from frozen cell pellets and qRT-PCR were performed as previously described (Carucci et al. 2017). Relative gene expression levels were determined using Eef1a1 and Rpl19 as reference genes. Total RNA from Tlx-GFP isolated cells was harvested with an RNeasy Mini Kit (QIAGEN) and reverse transcribed into cDNA using random primers (dN6, Roche).

RNA-sequencing and analysis
Total RNA was purified from frozen NSPC pellets using Trizol and indexed mRNA-seq libraries were constructed and sequenced as described (Collinson et al. 2016). Basecalls were converted to fastq using bcl2fastq v2.15.0.4 and read quality was checked using FastQC v0.10.1. Reads were aligned to the mouse genome (GRCm38) using STAR v2.3.0 with default parameters. Where necessary, subsampling was performed using samtools view with the -s option. Read counts per annotated feature were determined using htseq-count v0.6.0. Gene annotation is taken from Ensembl release 75 and small RNA annotation from miRbase release 20. Raw read counts were further analyses using the BioConductor package limma. Briefly, features with zero reads in all samples were removed and the remaining data were converted to logCPM using the voom function, taking into account mean associated variance and library size. Gene ontology (GO) functional annotation for the identified gene sets was done using the DAVID analysis tool

ChIP-sequencing analysis
Sequencing reads were trimmed using trim galore v0.4 using default parameters to remove the standard Illumina adapter sequence. Reads were mapped to the mouse GRCm38 genome assembly using bowtie2 v2.2.5 using default parameters. BAM files imported to SeqMonk and reads were extended by 200bp at their 5' end to approximate the true insert size. Sequence reads were quantitated over gene promoter regions (defined as -2kb to +0.5kb over transcriptional start sites) and the values were globally normalised per million reads. Gene promoters with log2 RPM values greater than 1, and that differed by more than 1.5-fold in log2 RPM values between adult and aged NSPCs, were defined as differentially methylated. For drawing the quantitation trend plots, sequence reads were quantitated over 100bp running windows, values were globally normalised to the data set with the highest coverage, and used to calculate the mean signal over all transcriptional start sites -/+ 5kb. For the genome browser screenshots, data were quantitated over 20bp running windows, separated by 10bp and the values were globally normalised to the data set with the highest coverage.

BS-sequencing and oxBS-sequencing
Genomic DNA was extracted using the AllPrep DNA/RNA Mini Kit (QIAGEN). DNA (250ng per sample) was fragmented by sonication (Covaris) and spiked with pooled sequencing control samples (3% w/w; CEGX). Samples were end-repaired and ligated with Illumina-supplied methylated adaptors using the NEBnext Ultra Kit (NEB). Indexed BS-seq and oxBS-seq libraries were constructed using the TrueMethyl Seq (CEGX) following the manufacturer's protocol. Final library amplification (11 cycles) was performed with Kapa Uracil Plus (Kapa Biosystems) and purified using 1x XP Ampure beads (Beckman Coulter). Samples were sequenced on an Illumina HiSeq 1000 as 100bp single-end libraries.
Sequence reads were trimmed to remove poor-quality reads and adapter contamination using Trim Galore (v0.3.7). The remaining sequences were mapped using Bismark (v0.12.5) (Krueger & Andrews 2011) with default parameters to the mouse reference genome GRCm38 in paired-end mode. Reads were deduplicated and CpG methylation calls were extracted from the deduplicated mapping output using the Bismark methylation extractor (v0.13.0) in paired-end mode.
Methylation calls were imported into SeqMonk and three replicates were merged into data groups for Adult and Aged, and bisulfite and oxidative bisulfite samples, respectively. Coverage outliers (10-fold above the median) were excluded. The average percentage methylation was calculated over windows generated over 50 consecutive CpGs where at least 10 had been measured in each sample. The percentage methylation of genomic features were calculated in SeqMonk and exported as annotated probe reports. Probes were created over CpG Islands (Illingworth et al. 2008) and probes with at least 10 observations in both data groups were quantified using the bisulfite methylation over features pipeline in SeqMonk. Probes were compared using a Chi-squared test with multiple testing correction, and an adjusted p-value of <0.05 was categorised as significant.