Abstract

   Partial reprogramming (pulsed expression of reprogramming transcription
   factors) improves the function of several tissues in old mice. However,
   it remains largely unknown how partial reprogramming impacts the old
   brain. Here we use single-cell transcriptomics to systematically
   examine how partial reprogramming influences the subventricular zone
   neurogenic niche in aged mouse brains. Whole-body partial reprogramming
   mainly improves neuroblasts (cells committed to give rise to new
   neurons) in the old neurogenic niche, restoring neuroblast proportion
   to more youthful levels. Interestingly, targeting partial reprogramming
   specifically to the neurogenic niche also boosts the proportion of
   neuroblasts and their precursors (neural stem cells) in old mice and
   improves several molecular signatures of aging, suggesting that the
   beneficial effects of reprogramming are niche intrinsic. In old neural
   stem cell cultures, partial reprogramming cell autonomously restores
   the proportion of neuroblasts during differentiation and blunts some
   age-related transcriptomic changes. Importantly, partial reprogramming
   improves the production of new neurons in vitro and in old brains. Our
   work suggests that partial reprogramming could be used to rejuvenate
   the neurogenic niche and counter brain decline in old individuals.
     __________________________________________________________________

   Long thought to be irreversible, aging is in fact a malleable process.
   Several interventions can slow or even reverse aspects of
   aging^[37]1–[38]8. Among these, partial reprogramming has emerged as a
   potentially powerful way to restore function in old
   tissues^[39]9,[40]10. Expression of reprogramming transcription factors
   (OCT4 (POU5F1 or OCT3/4), SOX2, KLF4, c-MYC or ‘OSKM’), which converts
   somatic cells into embryonic-like stem cells, erases many hallmarks of
   aging in vitro^[41]11–[42]15. This is likely because reprogramming
   mimics aspects of the resetting process between generations that
   naturally occurs during gametogenesis and fertilization. However, for
   in vivo applications, full reprogramming to embryonic-like stem cells
   would be detrimental because of loss of cell identity and increased
   tumorigenic potential^[43]16–[44]18. Interestingly, the ‘rejuvenating’
   effects of reprogramming can be uncoupled from loss of cell identity by
   controlling expression of reprogramming factors through pulsed or cell
   type-targeted expression (‘partial reprogramming’)^[45]19–[46]31.
   Partial reprogramming improves function in a variety of tissues in
   mice^[47]20,[48]29–[49]34, including pancreas and muscle, to different
   extents depending on the cell type^[50]22,[51]27,[52]28. In the central
   nervous system, partial reprogramming has shown promising effects in
   the spinal cord of old mice^[53]20 and in the optic nerve, hippocampus
   and striatum of young to middle-aged mice^[54]26,[55]30,[56]33,[57]34.
   However, the impact of partial reprogramming on the brains of older
   mice remains largely uncharacterized. The brain has a remarkably large
   number of different cell types, and unbiased characterization at
   single-cell resolution is critical to understand how different cells
   are impacted by in vivo partial reprogramming.

   The adult brain contains neurogenic niches in the hippocampus and the
   subventricular zone (SVZ) that are unique regions to examine the
   effects of brain aging and rejuvenation^[58]35–[59]40. In particular,
   the SVZ neurogenic niche encompasses many cell types (neural stem cells
   (NSCs), neuroblasts, microglia, endothelial cells, etc.) and has
   regenerative potential, as NSCs from this niche can give rise to
   newborn neurons (neurogenesis) and glia throughout adulthood in
   mammals^[60]41–[61]46. Normally quiescent, NSCs can also activate and
   proliferate and commit to neuronal precursors
   (neuroblasts)^[62]41,[63]42,[64]47. In mice, SVZ neuroblasts migrate a
   long distance to the olfactory bulb and differentiate into newborn
   neurons, which are important for olfactory discrimination and
   learning^[65]48–[66]51. NSCs from the SVZ are also critical for repair
   after brain injury or stroke^[67]52–[68]54. However, the process of SVZ
   neuro genesis declines strikingly with age^[69]55–[70]60. Rejuvenation
   strategies such as parabiosis (the sharing of blood between young and
   old mice), exercise and rapamycin have been shown to improve old
   neurogenic niches with various degrees of efficacy^[71]61–[72]67,
   suggesting that aspects of these age-related changes are malleable.

   Here, we show that partial reprogramming in old mice, either in the
   whole body or targeted to the SVZ niche itself, restores the neuroblast
   proportion to a younger level and reverses several molecular signatures
   associated with aging. Partial reprogramming also improves the
   proportion of neuroblasts during differentiation in old primary NSCs,
   suggesting a cell-autonomous effect. Finally, partial reprogramming
   improves the production of new neurons both in vitro and in old brains,
   a characteristic of youthfulness. Our work raises the possibility that
   partial reprogramming in already old individuals could rejuvenate the
   neurogenic niche and counter brain decline with aging or
   neurodegeneration.

Results

Partial reprogramming boosts neuroblast proportion in old mice

   We leveraged single-cell transcriptomics to unbiasedly examine how
   partial reprogramming affects the SVZ neurogenic niche in old mice. To
   this end, we generated aging cohorts of a genetically engineered mouse
   line^[73]68 that allows doxycycline-inducible expression of the four
   reprogramming factors OCT4, SOX2, KLF4 and c-MYC in the whole body
   ([74]Fig. 1a) (hereafter iOSKM mice). In these iOSKM mice, a reverse
   tetracycline transactivator (rtTA) is constitutively and ubiquitously
   expressed (from the ROSA26 locus), and doxycycline treatment induces
   expression of OCT4, SOX2, KLF4 and c-MYC (from the Col1a1
   locus)^[75]68. We first verified induction of the reprogramming factors
   both in vivo and in culture. Treatment of iOSKM mice with doxycycline
   for 2 d led to increased expression of OCT4 in the whole SVZ niche and
   the olfactory bulb by western blot analysis ([76]Fig. 1b) as well as in
   more than ten other tissues at varying levels ([77]Extended Data Fig.
   1a). We also confirmed SOX2 expression by western blot ([78]Extended
   Data Fig. 1b). To assess OSKM expression in SVZ NSCs in these iOSKM
   mice (and not solely in blood cells present in the tissue, for
   example), we isolated NSCs from the brain and added doxycycline to
   these primary NSC cultures. Western blot analysis showed that rtTA was
   constitutively expressed in NSCs from iOSKM mice and that OCT4 was
   induced by doxycycline in a dose- and time-dependent manner
   ([79]Extended Data Fig. 1c,[80]d). Reverse transcription followed by
   quantitative PCR (RT–qPCR) confirmed expression of Oct4 (Pou5f1) and
   the polycistronic OSKM messenger RNA (mRNA) (primers specific for
   Oct4-P2A, Sox2-T2A and E2A-c-Myc) upon doxycycline treatment in iOSKM
   NSCs ([81]Fig. 1c). Expression of the reprogramming factors declined
   following doxycycline withdrawal ([82]Extended Data Fig. 1e). Thus, the
   iOSKM mouse model allows pulsed expression of reprogramming factors in
   culture and in vivo in multiple tissues, including the SVZ neurogenic
   niche.

Fig. 1 |. Effect of whole-body in vivo partial reprogramming on the old SVZ
neurogenic niche by single-cell transcriptomics.

   Fig. 1 |
   [83]Open in a new tab

   a, Transgenic whole-body ‘iOSKM’ mouse for doxycycline-inducible
   expression of OSKM reprogramming factors. P2A, T2A, E2A, self-cleaving
   peptides; tetO, tetracycline-responsive operator; UTR, untranslated
   region. b, Western blot from microdissected brain regions after in vivo
   doxycycline (Dox) treatment. OB, olfactory bulb; arrowheads, OCT4
   bands; KPNB1, karyopherin subunit beta 1 (loading control). Normalized
   densitometry quantification of OCT4 is indicated below each lane. c,
   RT–qPCR on primary NSC cultures treated with doxycycline. n = 4 young
   (3 months, two males and two females) and n = 4 old (24–26 months, two
   males and two females) cultures, each from one mouse. Oct4-P2A,
   Sox2-T2A, E2A-c-Myc, primers specific to the polycistronic OSKM mRNA.
   Each dot represents one culture. Mean ± s.e.m.; P values, two-sided
   Wilcoxon rank-sum test. d–f, Single-cell RNA-seq. n = 2 young (3–4
   months, two males), n = 3 old (24–26 months, two males and one female)
   and n = 2 old + OSKM (24–26 months, one male and one female) samples.
   See [84]Extended Data Fig. 2b–[85]f for another cohort. d, Experimental
   design for single-cell RNA-seq of the SVZ after whole-body partial
   reprogramming using iOSKM mice. e,f, UMAP on single-cell transcriptomes
   of 3,857 high-quality cells. Each dot represents one cell. Colors
   indicate cell type (e) or condition (f). g–i, Combined results from two
   independent experiments ([86]Supplementary Table 1 and [87]Source
   data). n = 2 young (3–4 months, two males), n = 5 old (18–20 months or
   24–26 months, four males and one female), 4 old + OSKM (18–20 months or
   24–26 months, three males and one female) samples. g, Proportion of
   each cell type from single-cell RNA-seq of the SVZ as a fraction of
   total cells recovered per mouse, normalized as log[2] (fold change)
   over the mean of old control mice per experiment. Each dot, triangle or
   square represents one mouse. Bar, mean; P values, two-sided Wilcoxon
   rank-sum test. h, Proportion of aNSCs–NPCs and neuroblasts from
   single-cell RNA-seq, as a fraction of total cells within the NSC
   lineage per mouse. Each dot, triangle or square represents one mouse.
   Median (center line), upper and lower quartiles (box limits) and 1.5×
   interquartile range (whiskers); shape, experimental cohort; P values,
   two-sided Wilcoxon rank-sum test. m, months. i, Predicted age per mouse
   from a linear regression model of cell type proportions in the SVZ
   neurogenic niche, trained on 28 mice at 3–29 months. Each dot, triangle
   or square represents one mouse. Line, median; shape, experimental
   cohort; P values, two-sided Wilcoxon rank-sum test.

   To unbiasedly examine how partial reprogramming affects the SVZ
   neurogenic niches of old mice in vivo, we aged cohorts of iOSKM mice
   for ~2 years and performed single-cell transcriptomics upon pulsed
   induction of the reprogramming factors. Old iOSKM mice (two independent
   cohorts, 18–20 months for cohort 1 and 24–26 months for cohort 2) were
   treated with a partial reprogramming regimen of 2 d of doxycycline in
   drinking water, followed by 5 d of withdrawal, repeated three times
   ([88]Fig. 1d and [89]Extended Data Fig. 2a,[90]b). A pulsed doxycycline
   regimen was chosen because it was shown to improve pancreas repair in
   middle-aged mice and lifespan in an accelerated aging disease
   model^[91]19. We verified that this regimen did not negatively impact
   weight or survival of old iOSKM mice during treatment ([92]Extended
   Data Fig. 1f,[93]g). We collected SVZ neurogenic niches from partially
   reprogrammed old mice immediately after the last 2-d pulse of
   doxycycline (‘old + OSKM’) and from untreated old mice as well as from
   untreated young mice for cohort 2 and then performed single-cell RNA
   sequencing (RNA-seq) ([94]Fig. 1d and [95]Extended Data Fig. 2b).
   Dimensionality reduction and clustering (uniform manifold approximation
   and projection (UMAP) and Louvain clustering) based on the
   transcriptomes of single cells identified the major cell types
   previously observed^[96]54,[97]69–[98]73, including the NSC lineage
   (astrocytes and quiescent NSCs (qNSCs), activated NSCs and neural
   progenitor cells (aNSCs–NPCs) and neuroblasts) and other cell types
   (for example, microglia, oligodendrocytes, endothelial cells) ([99]Fig.
   1e and [100]Extended Data Fig. 2j, cohort 2; [101]Extended Data Fig.
   2d, cohort 1).

   We examined how partial reprogramming affects cell composition in old
   SVZ neurogenic niches ([102]Fig. 1f,[103]h). As previously
   reported^[104]69,[105]73–[106]76, old age led to a decline in
   neuroblasts and their precursors: aNSCs–NPCs ([107]Fig. 1g and
   [108]Extended Data Fig. 2h). This age-dependent decline was also
   observed, especially for neuroblasts, when calculated as a proportion
   of cells recovered from the NSC lineage (astrocytes–qNSCs, aNSCs–NPCs
   and neuroblasts), which is less sensitive to variation in
   microdissection of the SVZ ([109]Fig. 1h). Interestingly, partial
   reprogramming in old mice (‘old + OSKM’) led to an increase in the
   proportion of neuroblasts that was consistent across both aging
   cohorts, either when calculated as a fraction of the entire niche (P =
   0.11; [110]Fig. 1g and [111]Extended Data Fig. 2h) or of the NSC
   lineage ([112]Fig. 1h). We also validated these cell type proportion
   changes using immunostaining approaches ([113]Fig. 3 and [114]Extended
   Data Fig. 4). By contrast, reprogramming did not strongly affect the
   proportion of neuroblast precursors (aNSCs–NPCs) in old mice ([115]Fig.
   1g,[116]h and [117]Extended Data Fig. 2h). Partial reprogramming also
   led to an increase in the proportion of mural cells and a drop in the
   fraction of oligodendrocyte precursor cells (OPCs) in the neurogenic
   niche of old mice, although the total number of these cells captured
   was very low ([118]Extended Data Fig. 2h,[119]i). Importantly, partial
   reprogramming did not result in the production of new cell types (for
   example, ‘undifferentiated’) ([120]Fig. 1e,[121]f and [122]Extended
   Data Fig. 2d–[123]g). Thus, partial reprogramming mainly affects
   neuroblasts in the old SVZ neurogenic niche, increasing their
   proportion within the NSC lineage to more youthful levels.

Fig. 3 |. Validation of neuroblast proportion changes after whole-body
partial reprogramming.

   Fig. 3 |
   [124]Open in a new tab

   a, Experimental design to assess SVZ neuroblast proportion in situ
   after whole-body partial reprogramming. The red rectangle shows the
   region in b. b, Representative images for immunofluorescence staining
   of the SVZ in coronal brain sections after in vivo partial
   reprogramming. Dorsal, up; lateral, left. Magenta, DCX (neuroblasts);
   green, Ki-67 (proliferative cells); blue, 4′,6-diamidino-2-phenylindole
   (DAPI) (nuclei). The yellow rectangle is the region shown in the insets
   in c. Scale bars, 100 μm. Note that only tiles containing the SVZ were
   imaged; black boxes are areas where no imaging data were acquired
   ([125]Methods). c, Inset region from b. One optical section of a
   confocal image acquired using a ×20 objective. Scale bar, 25 μm. d,e,
   Quantification of immunofluorescence. Multiple optical sections a total
   of 8 μm thick were acquired for each SVZ, and both hemispheres in five
   coronal sections at 120-μm intervals were quantified per mouse
   ([126]Methods). n = 11 young (3–4 months, five males and six females),
   n = 8 old (20–26 months, three males and five females) and n = 10 old +
   OSKM (20–26 months, four males and six females) mice over three
   independent experiments (for individual experiments, see [127]Source
   data). Each dot represents one mouse. Mean ± s.e.m.; P values,
   two-sided Wilcoxon rank-sum test. d, Neuroblast proportion (DCX^+ over
   DAPI^+) in the SVZ per mouse, normalized to the mean of old control
   mice. e, Neuroblast density (number of DCX^+ cells per mm^3) in the SVZ
   per mouse. f, Representative immunofluorescence images. Dorsal, up;
   lateral, right. Magenta, PSA-NCAM (neuroblasts); green, EGFR
   (aNSCs–NPCs); blue, DAPI (nuclei). Yellow and orange rectangles are the
   regions shown in the insets in g. Scale bar, 100 μm. g, High
   magnification of inset regions from f. One optical section of a
   confocal image acquired using a ×63 objective. Scale bars, 25 μm. h,i,
   Quantification of immunofluorescence, as above. n = 9 young (3–4
   months, four males and five females), n = 8 old (20–26 months, three
   males and five females) and n = 10 old + OSKM (20–26 months, four males
   and six females) mice over three independent experiments (Source data).
   Each dot represents one mouse. Mean ± s.e.m.; P values, two-sided
   Wilcoxon rank-sum test. h, Neuroblast proportion (PSA-NCAM^+ over
   DAPI^+) in the SVZ per mouse, normalized to the mean of old control
   mice. i, Neuroblast density (number of PSA-NCAM^+ cells per mm^3) in
   the SVZ per mouse.

   To quantify the impact of partial reprogramming on all cell type
   proportions in the old niche, we trained a machine learning model to
   predict chronological age by linear regression using as features the
   proportions of ten annotated cell types from single-cell RNA-seq data
   (astrocytes–qNSCs, aNSCs–NPCs, neuroblasts, oligodendrocytes,
   microglia, etc.) ([128]Methods). For the training data, we leveraged an
   independent dataset of single-cell RNA-seq from the SVZ neurogenic
   niches of 28 mice spanning an age of 3–29 months^[129]77. This cell
   proportion linear regression model had good accuracy in predicting the
   chronological age of the mice (R = 0.63, P = 0.00035; [130]Extended
   Data Fig. 2k). Applying this cell proportion model to our single-cell
   RNA-seq dataset revealed that partial reprogramming in old mice
   decreased median predicted age (‘rejuvenation’) by 2.7 months
   ([131]Fig. 1i), although this change was not significant at the mouse
   level ([132]Fig. 1i and [133]Extended Data Fig. 2l).

   Thus, whole-body partial reprogramming in vivo can boost neuroblast
   proportion in old mice, countering one aspect of age-associated cell
   proportion changes in the SVZ neurogenic niche.

Partial reprogramming improves some signatures of aging

   We examined transcriptomic changes for each cell type in the SVZ after
   whole-body partial reprogramming. Differential gene expression analysis
   revealed significant gene expression changes during aging (old versus
   young) and in response to reprogramming (old + OSKM versus old) across
   cell types ([134]Fig. 2a,[135]b and [136]Extended Data Fig. 3a). To
   examine molecular signatures affected by age and reprogramming, we
   performed gene set enrichment analysis (GSEA) and compared the
   direction of changes with age and reprogramming in each cell type using
   a scatterplot ([137]Fig. 2c and [138]Extended Data Fig. 3b). Several
   signatures were increased with age and changed in the opposing
   direction with reprogramming ([139]Fig. 2c). As previously
   reported^[140]69,[141]74,[142]75,[143]78, inflammation and adhesion
   signatures were upregulated in most cell types in old individuals
   (although inflammation was downregulated in neuroblasts and adhesion
   signatures were downregulated in astrocytes–qNSCs with aging)
   ([144]Fig. 2e,[145]f and [146]Extended Data Fig. 3d,[147]e). RNA
   processing signatures were also mostly upregulated with age ([148]Fig.
   2c and [149]Supplementary Tables 3 and [150]4), similar to previous
   reports^[151]73. Interestingly, reprogramming ‘reversed’ some of these
   age-dependent signatures ([152]Fig. 2c). For example, RNA processing
   and cell adhesion signatures were reversed by partial reprogramming in
   aNSCs–NPCs and oligodendrocytes ([153]Fig. 2d,[154]e and [155]Extended
   Data Fig. 3e). Generally, aNSCs–NPCs, neuroblasts, endothelial cells
   and mural cells had a high fraction of pathways that trended in
   opposing directions with aging and after partial reprogramming
   ([156]Fig. 2c and [157]Extended Data Fig. 3d–[158]g). By contrast,
   partial reprogramming exacerbated many age-related signatures in some
   cell types (microglia, astrocytes–qNSCs) ([159]Fig. 2c). Inflammation
   was further increased by reprogramming in many cell types including
   microglia ([160]Fig. 2f,[161]k). This is consistent with the
   observation that cell-specific aging clocks built to predict age from
   single-cell transcriptomes^[162]77 did not detect strong ‘rejuvenation’
   by partial reprogramming ([163]Extended Data Fig. 3c), as these clocks
   include many inflammation genes^[164]77. Differences between cell types
   could reflect differing levels of and responses to reprogramming factor
   expression. Thus, reprogramming has both positive and negative
   transcriptomic effects, with some cells (for example, aNSCs–NPCs)
   exhibiting more of the positive effects.

Fig. 2 |. Transcriptomic pathway changes in the old neurogenic niche after
whole-body in vivo partial reprogramming.

   Fig. 2 |
   [165]Open in a new tab

   a, Differential gene expression by MAST for each cell type from
   single-cell RNA-seq, comparing old versus young untreated iOSKM mice.
   Each dot is the Z score for one gene. Positive Z score indicates
   upregulation with age. Color indicates false discovery rate (FDR) <
   0.2. b, As in a, but comparing old + OSKM versus old samples. Each dot
   is the Z score for one gene. Positive Z score, upregulation with
   whole-body reprogramming; color, FDR < 0.2. c, Scatterplots showing
   normalized enrichment scores from GSEA with reprogramming (old + OSKM
   versus old) versus age (old versus young) for each cell type. The gene
   list from MAST was ranked by Z score, and enrichment analysis was
   performed with Gene Ontology (GO) pathways using fgsea ([166]Methods).
   Each point is the normalized enrichment score for one pathway. For c–h,
   adjusted P values (P[adj]) are Benjamini–Hochberg (BH)-adjusted
   two-sided fgsea P values. Pathways where P[adj] < 0.05 for at least one
   comparison are shown. Filled circles, significant enrichment (P[adj] <
   0.05) after reprogramming. Pathways in the top left quadrant and bottom
   right quadrant change in opposing directions with aging and after
   reprogramming. d–f, Dot plots of GSEA comparing the effects of aging
   and reprogramming on expression of selected pathways related to RNA
   processing (d), cell adhesion (e) and inflammation (f) across cell
   types. Names of GO pathways are listed on the left. Size of dot
   reflects P[adj] values. Color, normalized enrichment score (NES), where
   positive values (red) indicate increase with age (‘aging’) or increase
   with reprogramming (‘OSKM’). g,h, Dot plot showing normalized
   enrichment score from GSEA in aNSCs–NPCs (g) or neuroblasts (h) after
   reprogramming (old + OSKM versus old). GO pathway names are listed on
   the left. Significantly enriched pathways (P[adj] < 0.05) are shown.
   Color, P[adj]. Size of dot reflects the number of genes in the set.
   i–k, Violin plots of gene expression in aNSCs–NPCs and neuroblasts from
   single-cell RNA-seq. Each dot is the summed expression of genes within
   the RNA processing (i), cell adhesion molecule binding (j) or
   inflammatory response (k) GO pathway for a single cell. Line, median; P
   values, two-sided Wilcoxon rank-sum test.

   We next specifically examined transcriptomic changes in neuroblasts,
   the main cell type for which the proportion increased upon partial
   reprogramming ([167]Fig. 1g), and their immediate precursors,
   aNSCs–NPCs. Surprisingly, there were very few significant pathway
   changes in neuroblasts with partial reprogramming ([168]Fig. 2h and
   [169]Extended Data Fig. 3b). By contrast, aNSCs–NPCs exhibit reversal
   in several signatures, including RNA processing and cell adhesion
   ([170]Fig. 2d,[171]e,[172]g,[173]i,[174]j), which could contribute to
   the observed boost in neuroblast proportion.

Neuroblast proportion increases in situ after partial reprogramming

   While single-cell RNA-seq is a powerful tool to capture different cell
   types, it requires cell dissociation. As an orthogonal validation
   approach, we used immunostaining of the intact SVZ niche. We collected
   brain sections containing the SVZ niche from young (3–4 months) and old
   (20–26 months) iOSKM mice, with or without whole-body in vivo partial
   reprogramming (old + OSKM) ([175]Fig. 3a–[176]c and [177]Extended Data
   Fig. 4a). Immunostaining with antibodies against DCX, a neuroblast
   marker in the SVZ niche, confirmed that old SVZ niches exhibit a strong
   decrease in the proportion of neuroblasts compared to young
   counterparts ([178]Fig. 3d), as shown previously^[179]56,[180]58.
   Importantly, partial reprogramming in old mice boosted the proportion
   of DCX^+ neuroblasts in the SVZ niche ([181]Fig. 3d). This increase was
   also observed when we measured DCX^+ cell density ([182]Fig. 3e and
   [183]Extended Data Fig. 4b) or DCX intensity ([184]Extended Data Fig.
   4c). Additionally, the results were similar using PSA-NCAM, another
   marker of neuroblasts ([185]Fig. 3f–[186]i and [187]Extended Data Fig.
   4f–[188]i). These findings provide orthogonal validation of the
   neuroblast proportion increase observed in our single-cell RNA-seq
   data.

   By contrast, partial reprogramming did not affect the proportion (and
   density) of proliferating cells (Ki-67^+, which includes aNSCs and
   NPCs; [189]Fig. 3b,[190]c and [191]Extended Data Fig. 4d,[192]e) or the
   proportion (and density) of epidermal growth factor (EGF) receptor
   (EGFR)^+ cells (aNSCs and NPCs) in the old SVZ ([193]Fig. 3f,[194]g and
   [195]Extended Data Fig. 4j,[196]k). This observation is consistent with
   our single-cell RNA-seq data, which did not show an increased
   proportion of aNSCs–NPCs. Hence, the increase in neuroblasts upon
   reprogramming may be due to increased neuroblast differentiation (and
   possibly neuroblast survival) rather than enhanced proliferation.

   Thus, partial reprogramming of the whole body restores in part the
   proportion of neuroblasts in the SVZ of old mice, which could improve
   neurogenesis.

A mouse model for partial reprogramming targeted to the SVZ

   In the iOSKM mouse model, reprogramming factors are expressed not only
   in the SVZ neurogenic niche but also in other brain regions and organs,
   which could thereby influence neuroblast production indirectly. To test
   the direct effect of partial reprogramming in the SVZ niche itself, we
   established a mouse line to control OSKM expression in both a spatial
   and temporal manner. This mouse line has a lox-STOP-lox cassette before
   the sequences encoding rtTA (to allow Cre-dependent cell- and/or
   tissue-specific expression) and OSKM under the control of the
   tetracycline response element (to control timing of expression with
   doxycycline) ([197]Fig. 4a). In these mice, expression of Cre
   recombinase (which can be spatially restricted) is required to trigger
   excision of the lox-STOP-lox cassette and subsequent expression of rtTA
   and an enhanced green fluorescent protein (EGFP) reporter ([198]Fig.
   4a). Doxy cycline treatment then induces expression of the OSKM
   reprogramming factors. Thus, this mouse line, when coupled with Cre,
   allows for inducible expression of OSKM in specific tissues (hereafter
   Cre-dependent and inducible OSKM or c+iOSKM).

Fig. 4 |. Effect of partial reprogramming targeted to the SVZ neurogenic
niche by single-cell transcriptomics.

   Fig. 4 |
   [199]Open in a new tab

   a, Mouse model for conditional and inducible expression of OSKM
   (c+iOSKM). Stereotaxic injection of Cre targets expression of the
   reprogramming factors to the SVZ. b–d, Immunofluorescence of coronal
   sections after stereotaxic viral injection (AAV-mCherry-Cre, AAV
   serotype SCH9 carrying Ef1a-mCherry-IRES-Cre) into the lateral
   ventricle. b, Markers of viral infection (magenta, mCherry; green,
   Cre); blue, DAPI. Scale bar, 100 μm. c, Magenta, mCherry; green, Cre.
   Scale bar, 50 μm. d, Magenta, mCherry; yellow, GFAP (NSCs/astrocytes);
   green, DCX (neuroblasts); blue, DAPI. Scale bar, 50 μm. e, RT–qPCR in
   NSCs isolated from c+iOSKM mice and infected with AAV-mCherry (AAV-SCH9
   Ef1a-mCherry) or AAV-Cre (AAV-SCH9 Ef1a-Cre) in culture. n = 4 young
   (3–4 months, three males and one female) and n = 4 old (20–23 months,
   three males and one female) cultures for AAV-Cre, n = 3 young (3–4
   months, two males and one female) and n = 4 old (20–23 months, three
   males and one female) cultures for AAV-mCherry, over one independent
   experiment. Each dot represents one culture isolated from one mouse.
   Mean ± s.e.m.; P values, two-sided Wilcoxon rank-sum test. f,
   Experimental design for single-cell RNA-seq after SVZ-targeted
   reprogramming. Stereotaxic injection of AAV-Cre (AAV-SCH9 Ef1a-Cre) was
   performed in c+iOSKM mice. n = 5 young (two males and three females), n
   = 6 old (four males and two females) and n = 6 old + OSKM (SVZ) (three
   males and three females) samples over two independent experiments. g,h,
   Dimensionality reduction (UMAP) on single-cell transcriptomes of 16,291
   high-quality cells, colored by cell type (g) or condition (h). Each dot
   represents one cell. i, Proportion of each cell type from single-cell
   RNA-seq as a fraction of total cells recovered per mouse, normalized as
   log[2] (fold change) over the mean of old control mice. Each dot,
   triangle or square represents one mouse. Bar, mean; P values, two-sided
   Wilcoxon rank-sum test. j, Proportion of aNSCs–NPCs and neuroblasts
   from single-cell RNA-seq, as a fraction of total cells recovered from
   the NSC lineage. Each dot represents one mouse. Median (center line),
   upper and lower quartiles (box limits) and 1.5× interquartile range
   (whiskers); P values, two-sided Wilcoxon rank-sum test. k, Predicted
   age per mouse after SVZ-targeted reprogramming from a linear regression
   model of cell type proportions in the SVZ neurogenic niche. Each dot or
   triangle represents one mouse; shape denotes experiment. Line, median;
   P values, two-sided Wilcoxon rank-sum test.

   To target the SVZ neurogenic niche, we delivered Cre by stereotaxic
   viral injection into the lateral ventricle in close proximity to the
   SVZ of c+iOSKM mice. We used an adeno-associated virus (AAV) variant
   (SCH9) that is capable of infecting adult SVZ NSCs (and can also infect
   other cells)^[200]79 to deliver Cre and an mCherry reporter (AAV-SCH9
   Ef1a-mCherry-IRES-Cre or ‘AAV-mCherry-Cre’) ([201]Fig. 4b). Using
   immunostaining on brain sections, we verified that stereotaxic AAV
   injection delivered Cre and mCherry to the neurogenic niche and that
   there was a good overlap between Cre and mCherry ([202]Fig. 4b,[203]c).
   Co-staining with cell type markers showed that AAV infection (assessed
   by mCherry) occurred in NSCs (GFAP^+ (also expressed in astrocytes) or
   SOX2^+) and also other cell types (for example, SOX2^− cells and a
   small number of DCX^+ neuroblasts) in the neurogenic niche ([204]Fig.
   4d and [205]Extended Data Fig. 5a–[206]c). There was no AAV infection
   (assessed by mCherry) detected in other areas of the brain, including
   the olfactory bulb and the cerebellum ([207]Extended Data Fig. 5d). In
   NSCs isolated from c+iOSKM mice, we confirmed proper excision of the
   lox-STOP-lox cassette after in vitro infection with AAV expressing Cre
   (‘AAV-Cre’) ([208]Extended Data Fig. 5e,[209]f). Flow cytometry and
   immunofluorescence staining showed that both AAV infection and Cre
   recombination in NSCs were efficient and dose dependent and that
   AAV-Cre-infected NSCs expressed EGFP, indicative of Cre recombination
   at this locus ([210]Extended Data Fig. 5g–[211]k). Importantly,
   doxycycline treatment of cultured NSCs isolated from c+iOSKM mice after
   Cre recombination triggered Oct4 expression from the inducible OSKM
   transgene as detected by RT–qPCR ([212]Fig. 4e and [213]Extended Data
   Fig. 5l,[214]m) and western blotting ([215]Extended Data Fig. 5n).
   Thus, this c+iOSKM mouse system combined with stereotaxic delivery of
   Cre allows for targeted reprogramming of cells, including NSCs, within
   the SVZ niche.

SVZ-targeted partial reprogramming boosts neuroblast proportion

   To systematically probe the impact of reprogramming targeted to the SVZ
   niche, we aged cohorts of c+iOSKM mice for over 2 years and performed
   single-cell transcriptomics. We stereotaxically injected AAV-Cre into
   two independent cohorts of young (3–4 months) and old (26–28 months)
   c+iOSKM mice and then 7 d later treated them with doxycycline for
   another 7 d for targeted expression of OSKM in the SVZ (+OSKM (SVZ))
   ([216]Fig. 4f). Targeted OSKM expression allows for longer continuous
   doxycycline treatment (7 d compared to 2 d) without the mortality
   previously observed for long-term whole-body
   expression^[217]17,[218]19,[219]28,[220]30,[221]80. After this
   SVZ-targeted partial reprogramming, we performed single-cell RNA-seq of
   the SVZ niche ([222]Fig. 4f). UMAP and Louvain clustering analysis
   indicated that all major cell types were recovered ([223]Fig.
   4g,[224]h). We verified that AAV infection and Cre-mediated
   recombination indeed led to expression of rtTA-IRES-EGFP, which was
   detected mostly in the aNSC–NPC cluster but also in other cell types
   ([225]Extended Data Fig. 6c).

   We performed cell proportion analysis of single-cell RNA-seq data,
   comparing aging and reprogramming. There was a strong age-dependent
   decline in the proportion of aNSCs–NPCs and neuroblasts in c+iOSKM mice
   ([226]Fig. 4i and [227]Extended Data Fig. 6a), consistent with results
   from other mouse lines^[228]69,[229]73–[230]76. The proportions of
   other cell types in the niche were either not affected by age
   (astrocytes–qNSCs) or increased with age (oligodendrocytes, endothelial
   cells). Interestingly, SVZ-targeted partial reprogramming led to an
   increase in the proportion of both aNSCs–NPCs and neuroblasts in old
   mice, when normalized either by total cells ([231]Fig. 4i) or by cells
   from the NSC lineage ([232]Fig. 4j; P = 0.065 for neuroblasts).
   SVZ-targeted partial reprogramming also led to a trending (P = 0.093)
   decrease in the proportion of astrocytes–qNSCs and endothelial cells
   ([233]Fig. 4i and [234]Extended Data Fig. 6a). There was variability
   between mice in the proportions of mural cells and oligodendrocytes
   ([235]Fig. 4i and [236]Extended Data Fig. 6a). Finally, we noted a high
   prevalence of immune cells such as T cells in young mice in this model,
   perhaps due to an immune response to surgery and/or viral infection in
   young mice, which makes interpretation of changes by reprogramming in
   immune cells difficult ([237]Fig. 4g,[238]h and [239]Extended Data Fig.
   6a,[240]b). Thus, SVZ-targeted partial reprogramming increases the
   proportion of neuroblasts (albeit less than whole-body partial
   reprogramming) and their precursors (aNSCs–NPCs) and may also impact
   other cells in the niche.

   To quantify the impact of SVZ-targeted partial reprogramming on all
   cell type proportions in the old niche, we used our cell type
   proportion linear regression model for predicting age ([241]Extended
   Data Fig. 2k). This cell proportion model indicated that SVZ-targeted
   partial reprogramming decreased median predicted age (‘rejuvenation’)
   by 10.9 months ([242]Fig. 4k and [243]Extended Data Fig. 6d). The
   rejuvenating effect of SVZ-targeted reprogramming was larger than that
   of the whole-body reprogramming and could be due to the more pronounced
   effect on multiple cell types (including aNSC–NPC and astrocyte–qNSC
   proportions) in the niche.

   Thus, SVZ-targeted partial reprogramming in the neurogenic niche itself
   in old mice leads to an increased proportion of neuroblasts and their
   precursors, aNSCs–NPCs. This suggests that partial reprogramming
   targeted to the SVZ has a direct and intrinsic effect in this niche.

Signatures affected by SVZ-targeted partial reprogramming

   We asked which molecular signatures are impacted upon in vivo partial
   reprogramming targeted to the SVZ. Analysis of differential gene
   expression in our single-cell RNA-seq dataset revealed significant
   transcriptomic changes across cell types with aging and after
   SVZ-targeted partial reprogramming ([244]Fig. 5a,[245]b). We compared
   the direction of molecular signature changes with age and reprogramming
   across cell types using scatterplots to identify pathways most
   ‘rejuvenated’ by this intervention ([246]Fig. 5c and [247]Extended Data
   Fig. 6c). These pathways include chromatin and transcription factors in
   neuroblasts, oligodendrocytes, endothelial cells and microglia
   ([248]Fig. 5d,[249]i, [250]Extended Data Fig. 6g and [251]Supplementary
   Tables 5 and [252]6), and pathways related to RNA processing and
   ribosomes in aNSCs–NPCs and endothelial cells ([253]Fig. 5e,
   [254]Extended Data Fig. 6f and [255]Supplementary Tables 5 and [256]6).
   Changes in cell adhesion pathway expression with age were also reversed
   by reprogramming in aNSCs–NPCs and oligodendrocytes ([257]Fig.
   5f,[258]j). Immune and inflammation pathways increased in many cell
   types upon SVZ-targeted partial reprogramming, although they appeared
   to decrease with age, probably related to immune cell infiltration in
   young mice ([259]Extended Data Fig. 6b,[260]d and [261]Supplementary
   Tables 5 and [262]6). As with whole-body reprogramming, cell-specific
   transcriptomic aging clocks did not show strong improvements in
   predicted age across cell types ([263]Extended Data Fig. 6g),
   potentially due to these opposing effects on inflammation.

Fig. 5 |. Transcriptomic pathway changes after partial reprogramming targeted
to the SVZ.

   Fig. 5 |
   [264]Open in a new tab

   a,b, Differential gene expression by MAST for each cell type, comparing
   old versus young (a) or old + OSKM (SVZ) versus old (b). Each dot
   represents the Z score for one gene. Color, FDR < 0.2. c, Scatterplots
   showing normalized enrichment scores from GSEA with SVZ-targeted
   reprogramming (old + OSKM (SVZ) versus old) versus age (old versus
   young) for each cell type. The gene list from MAST was ranked by Z
   score, and enrichment analysis was performed with GO pathways using
   fgsea ([265]Methods). Each point is the normalized enrichment score for
   one pathway. For c–h, P[adj], BH-adjusted two-sided fgsea P value.
   Pathways where P[adj] < 0.05 for at least one comparison are shown.
   Filled circles indicate P[adj] < 0.05 after reprogramming. Pathways in
   the top left and bottom right quadrants change in opposing directions
   with aging and SVZ-targeted reprogramming. d–f, Dot plots of GSEA
   comparing the effects of aging and SVZ-targeted reprogramming on
   expression of pathways related to chromatin and/or transcription (d),
   RNA processing (e) or cell adhesion (f) across cell types. Names of GO
   pathways are listed on the left. Size of dot reflects P[adj] values.
   Color indicates NES, where positive values (red) indicate increase with
   age (‘aging’) or increase with SVZ-targeted reprogramming (‘OSKM’).
   g,h, Dot plot showing NES from GSEA in aNSCs–NPCs (g) or neuroblasts
   (h) after SVZ-targeted reprogramming (old + OSKM (SVZ) versus old
   control). GO pathway names are listed on the left. Color reflects
   P[adj]values. Size of dot reflects the number of genes in the set.
   Pathways are filtered by P[adj] < 0.03 for g and P[adj] < 0.01 for h,
   and the top and bottom pathways ranked by NES are shown. cGMP, cyclic
   GMP; ncRNA, noncoding RNA; rRNA, ribosomal RNA. i–k, Violin plots of
   gene expression from single-cell RNA-seq. Each dot is the summed
   expression of genes within the chromatin binding (i), cell-cell
   adhesion (j) or neuron migration (k) GO pathway for a single cell. The
   line represents median. P values, two-sided Wilcoxon rank-sum test.
   l,m, Heatmaps showing NES from GSEA in aNSCs–NPCs (l) and neuroblasts
   (m), comparing whole-body (old + OSKM versus old) and SVZ-targeted (old
   + OSKM (SVZ) versus old) in vivo partial reprogramming. GO pathway
   names are listed on the right. Color reflects NES. Selected pathways
   are shown.

Fig. 6 |. Impact of aging and reprogramming on primary NSC differentiation.

   Fig. 6 |
   [266]Open in a new tab

   a, Experimental schematic for differentiation of primary NSC cultures
   from iOSKM mice with partial reprogramming (c,g,h). PDL, poly-d-lysine.
   n = 8 young (3–4 months, four males and four females) and n = 8 old
   (24–26 months, four males and four females) samples over two
   independent experiments. Each culture was isolated from one mouse. b,
   Representative image of immunofluorescence staining for neuroblasts
   (DCX, magenta) after in vitro differentiation for 4 d. Scale bar, 50
   μm. c, Quantification of immunofluorescence staining for neuroblasts
   (DCX^+ over DAPI^+). Each dot represents the neuroblast proportion for
   one culture isolated from one iOSKM mouse. Median (center line), upper
   and lower quartiles (box limits) and 1.5× interquartile range
   (whiskers); P values, two-sided Wilcoxon rank-sum test. d, Experimental
   schematic for differentiation of primary NSC cultures from c+iOSKM mice
   for 4 d with partial reprogramming. Primary NSCs were infected with AAV
   carrying genes for mCherry (control) or Cre and then passaged to allow
   for recovery before in vitro differentiation with or without
   reprogramming. n = 4 young (3–4 months, three males and one female) and
   n = 4 old (20–23 months, three males and one female) samples over one
   independent experiment. Each culture was isolated from one mouse. e,
   Quantification of immunofluorescence staining for neuroblasts (DCX^+
   over DAPI^+). Each dot represents the neuroblast proportion for one
   culture; each culture was isolated from one c+iOSKM mouse. Median
   (center line), upper and lower quartiles (box limits) and 1.5×
   interquartile range (whiskers); P values, two-sided Wilcoxon rank-sum
   test. f, Representative images of immunofluorescence staining for
   astrocytes (GFAP, green) and oligodendrocytes (olig.) and OPCs
   (transcription factor SOX10, magenta) after in vitro differentiation
   for 4 d. Scale bar, 50 μm. g,h, Quantification of immunofluorescence
   staining for astrocytes (GFAP^+ over DAPI^+) (g) and oligodendrocytes
   and OPCs (SOX10^+ over DAPI^+) (h). Each dot represents the astrocyte
   proportion for one culture, and each culture was isolated from one
   iOSKM mouse. Median (center line), upper and lower quartiles (box
   limits) and 1.5× interquartile range (whiskers); P values, two-sided
   Wilcoxon rank-sum test.

   We next focused on aNSCs–NPCs and neuroblasts ([267]Fig. 5g,[268]h and
   [269]Extended Data Fig. 6f,[270]g). SVZ-targeted partial reprogramming
   reversed the age-associated increase in single-cell expression of cell
   adhesion-related gene signatures in these cells ([271]Fig. 5j) (similar
   to whole-body reprogramming for aNSCs–NPCs). In this case, neuroblasts
   also showed ‘rejuvenation’ changes in many molecular signatures upon
   partial SVZ-targeted reprogramming, including pathways related to
   neuron development and morphology, cell migration and projections
   ([272]Fig. 5h,[273]k, [274]Extended Data Fig. 6g and [275]Supplementary
   Tables 5 and [276]6).

   Finally, we compared the molecular signatures regulated by whole-body
   and SVZ-targeted partial reprogramming. We examined pathways with
   enrichment scores that were either positive or negative in both
   paradigms ([277]Extended Data Fig. 7a,[278]b). Immune response and
   immune signaling pathways in neuroblasts and endothelial cells were
   upregulated by both whole-body and SVZ-targeted reprogramming
   strategies ([279]Fig. 5m and [280]Extended Data Fig. 7c,[281]e). Cell
   morphogenesis, adhesion and migration pathways in aNSCs–NPCs and
   neuroblasts were downregulated by both reprogramming strategies
   ([282]Fig. 5l,[283]m and [284]Extended Data Fig. 7d,[285]e).
   Interestingly, chromatin-related pathways in neuroblasts were
   downregulated only in SVZ-targeted reprogramming ([286]Fig. 5m and
   [287]Extended Data Fig. 7e). Differences between whole-body and
   SVZ-targeted partial reprogramming paradigms could reflect the
   influence of cell non-autonomous and systemic factors or differences in
   the protocols (for example, length of doxycycline induction, use of
   viruses).

   Thus, SVZ-targeted partial reprogramming reverses some aspects of
   age-dependent changes within the niche, including cell type proportion
   changes and transcriptional signatures in the NSC lineage, indicating
   that the effects of reprogramming are at least in part intrinsic to the
   neurogenic niche.

Partial reprogramming in old NSCs improves neuronal lineage differentiation

   While stereotaxic injection in c+iOSKM mice allows spatial restriction
   of reprogramming to the SVZ niche, a key question remains: whether
   targeting a single cell type in the NSC lineage could recapitulate the
   effects of partial reprogramming. Cell type-specific in vivo
   reprogramming is limited by the tools needed to target only one cell
   state within the NSC lineage. We thus leveraged a primary NSC culture
   system^[288]63,[289]81 to assess how cell-autonomous reprogramming
   affects NSCs’ ability to differentiate into neuroblasts, given the
   increase in neuroblast proportion uncovered by single-cell RNA-seq.

   Proliferative NSCs in primary cultures can be induced to differentiate
   into a mixed population of both neuronal and glial cells by withdrawal
   of EGF and basic fibroblast growth factor (bFGF) from the cell culture
   medium^[290]81. Using immunofluorescence staining, we first verified
   that this NSC differentiation protocol led to the appearance of
   neuroblasts (DCX^+), astrocytes (GFAP^+) and oligodendrocytes and OPCs
   (SOX10^+) after 4 d ([291]Fig. 6a,[292]b,[293]f and [294]Extended Data
   Fig. 8a) and neurons (TUJ1^+(TUBB3), NeuN^+), astrocytes (GFAP^+) and
   oligodendrocytes (O4^+) after 8 d ([295]Extended Data Fig. 8b).

   We then analyzed how age and reprogramming influenced primary NSC
   differentiation. We isolated primary NSCs from young (3–4 months) and
   old (24–26 months) iOSKM mice and verified that doxy cycline also
   induced OSKM expression in these cells ([296]Fig. 1c and [297]Extended
   Data Fig. 1b–[298]d). To induce reprogramming and differentiation of
   NSCs, we treated NSC primary cultures with doxycycline overnight before
   inducing differentiation and throughout the 4-d differentiation
   protocol ([299]Fig. 6a). As expected, old NSC cultures produced fewer
   neuroblasts than their young counterparts after 4 d of differentiation
   ([300]Fig. 6b,[301]c and [302]Extended Data Fig. 8k,[303]l). In line
   with our in vivo data, partial reprogramming of old NSCs restored this
   age-associated deficit in neuroblast formation ([304]Fig. 6c and
   [305]Extended Data Fig. 8l). Partial reprogramming of young NSCs did
   not boost the neuroblast proportion ([306]Fig. 6c and [307]Extended
   Data Fig. 8l). Importantly, before differentiation, reprogramming in
   old NSCs did not impact cell type marker expression (nestin, EGFR) or
   proliferation status (Ki-67, 5-ethynyl-2′-deoxyuridine (EdU)) and did
   not induce a spurious increase in neuroblast differentiation (DCX)
   ([308]Extended Data Fig. 8c–[309]j). Partial reprogramming did not
   increase the proportion of proliferative cells (Ki-67^+) after
   differentiation ([310]Extended Data Fig. 8m), suggesting that this
   intervention ameliorates differentiation (or neuroblast survival)
   rather than promoting NSC expansion in differentiation conditions.
   Partial reprogramming also increased the neuroblast proportion in NSC
   cultures from old c+iOSKM mice infected with Cre virus and treated with
   doxycycline throughout differentiation ([311]Fig. 6d,[312]e and
   [313]Extended Data Fig. 8o,[314]p). Thus, partial reprogramming in
   culture boosts the ability of old NSCs to generate neuroblasts,
   suggesting a direct, cell-autonomous effect of reprogramming on old
   NSCs.

   Aging is also accompanied by a ‘glial skew’ of NSCs during
   differentiation, with a decrease in the proportion of new neurons and
   an increase in the proportion of glial cells (mostly
   astrocytes)^[315]82. Indeed, after 4 d of differentiation, cultures
   differentiated from old NSCs exhibited a large increase in the
   proportion of astrocytes (GFAP^+) compared to young counterparts
   ([316]Fig. 6f,[317]g and [318]Extended Data Fig. 8n). Partial
   reprogramming blunted this age-dependent increase in astrocytes
   ([319]Fig. 6g) (and did not affect oligodendrocytes, [320]Fig. 6h).
   Hence, partial reprogramming reverses the glial skew of old NSCs,
   favoring the production of neuroblasts, a feature of youthful NSCs.

   To identify transcriptional changes that could mediate the effects of
   reprogramming on old NSCs and their differentiation, we performed bulk
   RNA-seq on primary NSCs throughout a time course of differentiation and
   partial reprogramming ([321]Fig. 7a,[322]b and [323]Extended Data Fig.
   9c). As expected, principal-component analysis (PCA) separated samples
   by differentiation time points ([324]Fig. 7c), with a decline in cell
   cycle genes (for example, Pclaf, Mki67) and increases in neuronal (for
   example, Slc1a1, Draxin) and glial (for example, Gfap, Mbp) genes
   across differentiation ([325]Extended Data Fig. 9a,[326]b). Global GSEA
   in NSCs revealed a host of changes with age, some of which were
   reversed by reprogramming ([327]Fig. 7d and [328]Extended Data Fig.
   9d,[329]f). Similar to in vivo findings, primary NSCs exhibited
   increased expression of cell adhesion, RNA processing and
   chromatin-related gene pathways with age, and partial reprogramming in
   old NSCs reversed expression of these signatures ([330]Fig. 7e and
   [331]Extended Data Fig. 9d,[332]j). Interestingly, we identified a
   cluster of genes for which expression increased with age and that were
   restored to a youthful expression level by partial reprogramming
   ([333]Fig. 7f and [334]Extended Data Fig. 9i). These include Lrfn4,
   involved in neurite outgrowth and cell motility, Ddx19a, encoding a
   putative RNA helicase and the transcriptional regulator encoded by
   Brd2. Such genes and pathways may represent interesting candidates that
   mediate the age-specific effect of partial reprogramming on old NSC
   function.

Fig. 7 |. Gene expression in primary NSCs with aging, reprogramming and
differentiation.

   Fig. 7 |
   [335]Open in a new tab

   a, Experimental schematic for bulk RNA-seq time course of partial
   reprogramming and differentiation in primary NSC cultures from iOSKM
   mice. n = 5 young (3–4 months, two males and three females) and n = 4
   old (25 months, two males and two females) samples. Each culture was
   isolated from one mouse. In the top row (‘NSC’), primary NSCs are
   treated with doxycycline for 6 h, 18 h or 48 h. In the bottom row
   (‘Diff.’, differentiation), primary NSCs are treated with doxycycline
   for 18 h before inducing differentiation and doxycycline treatment is
   continued for an additional 6 h, 18 h or 48 h. b, Table of the number
   of biological replicates collected for each condition at each time
   point for bulk RNA-seq. Each culture was isolated from one mouse. c,
   Dimensionality reduction (PCA) on all samples from bulk RNA-seq. Each
   dot represents one sample. PC, principal component. d, Scatterplots
   showing normalized enrichment scores from GSEA with partial
   reprogramming (old + OSKM versus old) versus aging (old versus young)
   for each time point. Each point represents the normalized enrichment
   score for one GO pathway. For d,e, P[adj], BH-adjusted two-sided fgsea
   P value. Pathways where P[adj] < 0.05 for at least one comparison are
   shown. Filled circles indicate P[adj] < 0.05 in both comparisons.
   Pathways in the top left quadrant and the bottom right quadrant change
   in opposing directions with aging and after reprogramming. e, Dot plots
   of GSEA comparing the effects of aging and 6 h, 18 h and 48 h of
   partial reprogramming in primary NSCs on expression of pathways related
   to cell adhesion, RNA processing and chromatin and histone marks. Names
   of GO pathways are listed on the left. Size of dot reflects P[adj]
   values. Color, NES. f, Expression of selected genes in primary NSCs at
   the 18-h time point. Each dot represents the normalized count for one
   culture. Dashed line, mean expression level in young control samples;
   mean ± s.e.m.; P values, two-sided Wilcoxon rank-sum test.

   Thus, partial reprogramming of NSCs during differentiation in culture
   is sufficient to boost the resulting proportion of neuroblasts,
   indicating a direct effect of partial reprogramming on old NSCs. While
   in vitro assays cannot fully recapitulate in vivo conditions, these
   results also suggest that the boost in neuroblast proportion after in
   vivo partial reprogramming may be due at least in part to improved
   differentiation of NSCs.

Newborn neuron production improves with partial reprogramming

   Can partial reprogramming boost not only the production of neuroblasts
   but also the generation of differentiated neurons? We first tested the
   impact of partial reprogramming in NSCs on mature neuron formation in
   culture. Using the in vitro differentiation protocol for 8 d ([336]Fig.
   8a,[337]b), we found that old primary NSCs led to fewer neurons
   (TUJ1^+) than young primary NSCs ([338]Fig. 8c). Partial reprogramming
   in old NSCs, starting before and continuing through differentiation,
   increased the production of more mature neurons ([339]Fig. 8c and
   [340]Extended Data Fig. 8q).

Fig. 8 |. Newborn neuron production with partial reprogramming.

   Fig. 8 |
   [341]Open in a new tab

   a, Experimental schematic for differentiation of primary NSC cultures
   for 8 d with partial reprogramming. n = 8 young (3–4 months, four males
   and four females) and n = 8 old (24–26 months, four males and four
   females) samples over two independent experiments. Each culture was
   isolated from one mouse. b, Representative image of immunofluorescence
   staining for immature neurons after in vitro differentiation for 8 d.
   Green, TUJ1 (neurons); magenta, DCX (neuroblasts and immature neurons);
   DAPI (nuclei). Scale bar, 50 μm. c, Quantification of
   immunofluorescence staining for neurons (TUJ1^+ over DAPI^+). Each dot
   represents the neuron proportion for one culture, and each culture was
   isolated from one mouse. Box plot indicates median (center line), upper
   and lower quartiles (box limits) and 1.5× interquartile range
   (whiskers). P values, two-sided Wilcoxon rank-sum test. d, Experimental
   design to assess the number of newborn neurons in the olfactory bulb
   after in vivo whole-body partial reprogramming. Blue bars represent
   doxycycline treatment (2 d on, 5 d off). EdU was administered once
   daily as indicated with brown arrows. i.p., intraperitoneal. e,
   Representative image for immunofluorescence staining of the olfactory
   bulb. Dorsal is up. Cyan, EdU; magenta, NeuN (neuronal nuclei); blue,
   DAPI (nuclei). The yellow square is the region shown in insets in f.
   Scale bar, 500 μm. f, Higher-magnification images of the olfactory
   bulb. Cyan, EdU; magenta, NeuN (neuronal nuclei); red, DCX (migrating
   neurons). Scale bar, 100 μm. g, Number of newborn neurons (EdU^+NeuN^+
   cells per mm^2) in the olfactory bulb per mouse. Both hemispheres in
   five coronal sections at 120-μm intervals were quantified per mouse
   ([342]Methods). n = 6 young (3–4 months, three males and three
   females), n = 4 old (20–22 months, two males and two females) and n = 5
   old + OSKM (20–22 months, three males and two females) samples. Each
   dot represents one mouse. Data are mean ± s.e.m.; P values, two-sided
   Wilcoxon rank-sum test.

   In vivo, neuroblasts from the SVZ migrate to the olfactory bulb where
   they terminally differentiate into neurons^[343]42,[344]83. To test the
   effect of partial reprogramming on the formation of newborn neurons in
   the olfactory bulb in vivo, we injected old mice with EdU to label all
   proliferating cells (including aNSCs–NPCs and proliferating
   neuroblasts) during each cycle of doxycycline treatment ([345]Fig.
   8d–[346]f). As neurons are postmitotic, EdU-positive neurons (labeled
   by the neuronal nuclear marker NeuN) are newly born from these labeled
   aNSCs and neuroblasts. Immunostaining showed that the number of newborn
   neurons (EdU^+NeuN^+) in the olfactory bulb declines dramatically with
   age, as has been previously observed ([347]Fig. 8g). Interestingly,
   partial reprogramming in old mice increased the number of newborn
   neurons in the olfactory bulb (neurogenesis) ([348]Fig. 8g).

   Thus, partial reprogramming in old NSCs and old mice increases the
   number of newborn neurons, an important aspect of youthfulness.

Discussion

   Using single-cell transcriptomics and immunostaining validation, we
   find that whole-body partial reprogramming in old mice partly reverses
   the age-associated defect in neuroblast proportion in the SVZ
   neurogenic niche. This ‘rejuvenation’ effect can be recapitulated by
   targeting the SVZ itself for partial reprogramming, indicating a
   niche-intrinsic phenomenon. Furthermore, partial reprogramming in old
   NSCs in culture cell autonomously improves their differentiation into
   neuronal precursors. Our study uncovers the impact of partial
   reprogramming in old brains by systematically probing its effect on
   multiple different cell types. Previous studies in the nervous system
   had found rejuvenation of visual acuity and bulk DNA methylation and
   transcriptome upon OSK expression in the optic nerve of 11–12-month-old
   mice^[349]30, improvement in remyelination after injury with c-MYC (M)
   expression in oligodendrocyte progenitor cells of 18-month-old
   mice^[350]20, improvement in epigenetic marks and memory in the
   hippocampus of 6–10-month-old mice upon OSKM expression^[351]26,
   improved neurodegeneration phenotypes in an Alzheimer’s mouse model
   with OSKM expression in neurons^[352]33 and improved recovery after
   ischemic injury in 2–4-month-old mice with OSKM expression^[353]34. Our
   study uses older animals (18–28 months old) and a combination of
   systematic single-cell RNA-seq analysis and immunocytochemistry
   validation, which reveal that NSCs and neurob lasts are key cell types
   influenced by partial reprogramming in a neurogenic niche of old
   brains.

   We find that partial reprogramming of the SVZ itself in vivo
   recapitulates some of the effects of whole-body partial reprogramming,
   suggesting that at least some of the beneficial effects are intrinsic
   to the SVZ niche. In muscle stem cells (satellite cells), the intrinsic
   effects of partial reprogramming have been assessed with various
   results^[354]19,[355]22,[356]28,[357]29: a genetic strategy to target
   reprogramming to the muscle stem cells in vivo does not seem to improve
   their function (in young mice)^[358]28, while ex vivo reprogramming in
   old muscle stem cells and reimplantation in old muscles does seem
   sufficient to improve repair after injury^[359]29. The levels of
   reprogramming factor expression^[360]22, the exact combination of
   factors used, the age of onset and duration of treatment could all be
   important factors to explain differences.

   We do not observe any appearance of a ‘new’ cell type (for example,
   embryonic-like stem cell) by single-cell RNA-seq after in vivo partial
   reprogramming. However, the possibility that transdifferentiation of
   another cell type might contribute to the increase in neuroblast
   proportion cannot be excluded in vivo. Using a culture system, we find
   that partial reprogramming of NSCs is sufficient to rescue the
   age-associated decline in neuroblast proportion, suggesting that at
   least some effects of in vivo partial reprogramming are due to
   improvement of NSCs. An alternative possibility could be that partial
   reprogramming factors directly promote neuronal differentiation^[361]84
   independent of a ‘rejuvenation’ effect. Our observation that the effect
   of partial reprogramming is specific to old NSCs in vitro suggests that
   this is unlikely. Nevertheless, it will be important to distinguish the
   impact of partial reprogramming on rejuvenation versus cell fate.

   Interestingly, reprogramming factors, which normally induce stemness,
   promote more fate-committed progenitor cells and boost the production
   of new neurons in the context of old mice. This is consistent with the
   idea that reprogramming factors could rescue aspects of age-associated
   defects in the neural lineage, thereby increasing neurogenesis. Partial
   reprogramming in 6–10-month-old mice did not show increased
   neurogenesis in the hippocampus^[362]26, perhaps because mice were too
   young. However, the authors observed increased neuroblast and neuron
   migration^[363]26, reminiscent of the changes in cell adhesion
   signatures we observe here, and this could be a possible step in the
   neurogenesis process affected by reprogramming. Partial reprogramming
   could thus improve aging hallmarks in old NSCs, favoring their
   commitment and differentiation toward the neuronal lineage.

   Many avenues for future study remain. It remains unknown how long the
   effects of reprogramming last and whether pulses of expression are more
   beneficial than continuous expression. It would be interesting to
   examine how long transcriptomic changes persist following doxycycline
   withdrawal, in addition to determining whether cell type proportion
   changes are stable or dependent on recent OSKM expression. The
   long-term effect on survival and the impact of seemingly negative
   effects like increased inflammation will be key to determine.
   Importantly, the c+iOSKM mouse, combined with strains allowing cell
   type-specific expression of Cre, could enable cell type-specific
   partial reprogramming in vivo as well as genetic lineage tracing, which
   should help elucidate cell-intrinsic effects, cell fate and population
   heterogeneity. Finally, it will be important to test whether partial
   reprogramming in the old SVZ leads to beneficial effects on behavior
   and whether pathways affected by reprogramming could be directly
   targeted to improve neurogenesis.

   Systemic rejuvenation interventions (exercise, young blood, dietary
   restriction) have different effects on cell proportions within the SVZ
   neurogenic niche. Exercise also boosts neuroblast proportion in the
   aged SVZ, whereas heterochronic parabiosis does not^[364]77,[365]85.
   There might be some overlapping mechanisms between reprogramming and
   exercise, and each intervention could improve specific aspects of aging
   (for example, neuroblast formation) while also having detrimental or
   orthogonal effects on the aging process (for example, inflammation).
   Comparing the effect of interventions on different cell types could
   help elucidate common or synergistic mechanisms to counter aging.

Methods

Animal housing and protocols

   All animal procedures were performed according to protocols approved by
   the IACUC and AAPLAC of Stanford University (protocol 8661). Mice were
   housed with 12-h light–dark cycles, ad libitum access to food and water
   in the cage, a temperature of ~21 °C and ~50% humidity. Mice were
   maintained under the care of the Stanford Veterinary Service Center and
   housed at the Stanford Comparative Medicine vivarium and the
   ChEM-H/Neuro vivarium.

Mouse strains, sex and age

   Whole-body iOSKM mice (ROSA26(rtTA-M2);Col1a1(tetO-OSKM)) were
   generated by the Jaenisch laboratory^[366]68. Mice were obtained from
   Jackson Laboratory ( JAX 011004) and maintained in an in-house breeding
   colony to produce aging cohorts.

   c+iOSKM mice were generated in-house by crossing whole-body OSKM mice
   (ROSA26(rtTA-M2);Col1a1(tetO-OSKM)) ( JAX 011004) with a strain
   allowing Cre-dependent expression of rtTA (ROSA26(
   lox-STOP-lox-rtTA-IRES-EGFP)) ([367]Fig. 3a) ( JAX 005670, generated by
   the Nagy laboratory^[368]86). F[1] heterozygotes were intercrossed to
   generate double homozygous mutants. Double homozygous mice (ROSA26
   (lox-STOP-lox-rtTA-IRES-EGFP);Col1a1(tetO-OSKM)) were intercrossed to
   establish the colony. Mice were maintained in an in-house breeding
   colony to produce aging cohorts.

   We maintained in-house aging colonies of iOSKM mice and c+iOSKM mice
   ranging in ages up to 28 months of age. Both male and female mice were
   used for this study. Old animals were assigned to control or treatment
   conditions such that sexes and mean body weights at the beginning of
   the experiment were balanced between conditions. Mice were considered
   young adults at 3 months of age and old after 18 months of age.
   Specific ages are indicated in the respective figures. Sexes of mice
   used in single-cell RNA-seq experiments can be found in the mouse
   metadata table ([369]Supplementary Table 1). Mice that died or were
   euthanized before the experiment endpoint ([370]Extended Data Figs. 1g
   and [371]6h) were excluded from the study.

In vivo doxycycline administration

   Doxycycline (Fisher, ICN19895505) was dissolved in drinking water at a
   concentration of 1 mg ml^−1 and provided ad libitum in amber water
   bottles to protect the solution from light. Mouse weight and condition
   were monitored three times per week, and no signs of dehydration were
   observed. Old mice were assigned to control or doxycycline treatment
   conditions such that sexes and mean body weights at the beginning of
   the experiment were balanced between conditions.

SVZ dissection for single-cell RNA sequencing

   Live cells were isolated from the SVZ for single-cell RNA-seq as
   previously published^[372]69,[373]77. The experimenter performing
   dissections was blinded to the sample condition; mice were taken in an
   alternating fashion to minimize impacts of sample processing order.
   Briefly, mice were terminally anesthetized with Avertin (0.5 mg g^−1
   mouse, Sigma-Aldrich, [374]T48402–25G) in PBS (Corning, 21–040-CV) and
   then perfused transcardially with 15 ml PBS with heparin (50 U ml^−1,
   Sigma-Aldrich, H3149–50KU) to remove circulating blood cells. Brains
   were removed into PIPES buffer on ice and dissected in ice-cold PIPES
   buffer. The lateral walls of the SVZs from both hemispheres were
   carefully dissected under a dissecting microscope using forceps, a
   scalpel with a no. 22 blade and a 22.5° stab knife. The dissected
   tissue includes NSCs (which are within a few cell layers of the
   ventricle) as well as part of the striatum (evidenced by the
   oligodendrocyte population recovered). SVZs were minced and then
   digested in papain solution (14 U ml^−1, Worthington, [375]LS003119)
   for 10 min at 37 °C. After centrifugation at 300g for 5 min at 4 °C,
   liquid was removed, and ovomucoid solution (0.7 mg ml^−1 ovomucoid
   (Sigma-Aldrich, T9253–1G), 0.5 mg ml^−1 DNase I (Sigma-Aldrich,
   DN25–100MG) in DMEM/F12 (Gibco, 11330057)) was added. Tissue was
   triturated with a P1000 pipette to obtain a single-cell suspension.
   Cells were then layered onto 22% (vol/vol) Percoll (Sigma-Aldrich,
   GE17–0891-01) in PBS and centrifuged at 700g for 10 min at 4 °C without
   brakes to remove myelin debris. The supernatant, containing myelin and
   debris, was removed. The cell pellet was then resuspended in FACS
   buffer (1% BSA (Sigma, A7979) with 0.1% glucose (Sigma, G7021) in PBS)
   to wash and filtered through a 40-μm strainer. Cells were centrifuged
   at 300g for 5 min at 4 °C and then resuspended in 70 μl FACS buffer
   with DAPI (1 μg ml^−1, Thermo, 62248). Live cells were isolated by flow
   cytometry ([376]Extended Data Fig. 10a) on a BD Aria II sorter (BD
   FACSDiva software version 8.0.1) using a 100-μm nozzle at 20 psi and
   collected in 750 μl FACS buffer in a protein low-bind microcentrifuge
   tube (Eppendorf, 0030108442).

Sample multiplexing with lipid-modified oligonucleotides

   Cells were labeled with lipid-modified oligonucleotides (LMOs) at 4 °C
   using a protocol^[377]77 adapted from MULTI-seq^[378]87. Lipid anchor
   and co-anchor reagents were generously provided by the Gartner
   laboratory at the University of California, San Francisco. Custom
   oligonucleotides were purchased from Integrated DNA Technologies. The
   entirety of each sample was labeled and used. Cells from two to six
   mice were combined after labeling and loaded onto each lane of a 10x
   Genomics chip. Metadata for all mice used in single-cell RNA-seq
   experiments, including sequences of multiplexing oligonucleotides and
   distribution of mice in each 10x lane, can be found in
   [379]Supplementary Table 1.

Single-cell library preparation and sequencing

   The 10x Genomics Chromium Next GEM Single Cell 3′ kits were used to
   perform single-cell RNA-seq chip loading and library preparation
   according to the manufacturer’s instructions. Kit version 3 was used
   for one experiment ([380]Extended Data Fig. 2b–[381]f); version 3.1 was
   used for all other experiments. Gene expression libraries were prepared
   according to the manufacturer’s instructions. LMO libraries were
   prepared according to the MULTI-seq protocol^[382]87. All gene
   expression libraries and LMO libraries from an individual experiment
   were pooled and sequenced on the Illumina HiSeq (whole-body OSKM
   experiments 1 and 2) or the NextSeq (SVZ-specific OSKM experiment) with
   paired-end 150-bp reads.

Single-cell RNA-seq data processing and cell type identification

   Cell Ranger (version 3.0.2 for whole-body reprogramming experiments,
   version 6.0.2 for SVZ-targeted reprogramming experiments) default
   settings were used to distinguish cells from background. Reads were
   mapped to a custom genome (mm10), with the following transgene
   sequences added as additional chromosomes: for whole-body iOSKM mice,
   rtTA-M2; for c+iOSKM, NeoR and rtTA-IRES-EGFP. To determine and confirm
   these sequences, we performed PCR amplification from genomic DNA using
   multiple primer sets tiling the locus, and PCR products were Sanger
   sequenced (see ‘[383]Genomic DNA extraction and PCR’); primer sequences
   are in [384]Supplementary Table 3. P2A, E2A and T2A sequences were also
   added to the genome as separate chromosomes, but reads did not map to
   them, likely due to their very short length of 54–60 nucleotides.
   Seurat^[385]88 (version 3.1.2) was used for quality control. Cells were
   filtered out if they contained more than 10% mitochondrial reads or
   less than 500 genes.

   For LMO libraries, CITE-seq count (version 1.4.5) was used to map reads
   to sample barcodes. Samples were demultiplexed using HTODemux in Seurat
   or the MULTI-seq strategy^[386]87. Cells showing strong expression of
   multiple sample barcodes were considered doublets and excluded from
   subsequent analysis, as were cells that could not be assigned to any
   sample barcode.

   Seurat was used for SCT normalization, dimensionality reduction,
   clustering and marker gene identification. Cell types were manually
   annotated as previously described^[387]77. Major cell clusters were
   identified in Seurat using ‘FindClusters()’, and marker genes were
   identified using ‘FindAllMarkers()’. Expression of marker genes was
   used to label cell types based on literature and PanglaoDB^[388]89.
   This analysis identified 13 clusters of cells (depending on the
   dataset): astrocytes and qNSCs, aNSCs and NPCs, neuroblasts, neurons,
   oligodendrocytes, oligodendrocyte progenitor cells, endothelial cells,
   microglia, monocytes and macrophages, T cells, mural cells (pericytes
   and vascular smooth muscle cells) and ependymal cells. Very small cell
   clusters (fewer than 20 cells, labeled as clusters 27–30) from the
   whole-body reprogramming experiment were not annotated; they likely
   contain T cells and potentially doublets. The top five differentially
   expressed marker genes for each cluster are shown in [389]Extended Data
   Fig. 2j. Full lists of marker genes are in [390]Supplementary Table 2.

Single-cell RNA-seq cell type proportion analysis

   Cell type proportions for aNSCs–NPCs and neuroblasts were calculated as
   a fraction of the total cells within the NSC lineage (astrocytes–qNSCs,
   aNSCs–NPCs, neuroblasts) because small differences in SVZ
   microdissection could have an impact on the number of other cell types
   recovered (that is, oligodendrocytes). Cell type proportions as a
   fraction of total cells are also shown for completeness. It is possible
   that steps in the protocol, including tissue dissociation, live cell
   sorting and loading into the microfluidic chip might influence the
   recovery of specific cell types relative to others. The following cell
   types were excluded from cell type proportion analyses: (1) ependymal
   cells. Our single-cell RNA-seq datasets captured a small number of
   ependymal cells, as shown previously^[391]69,[392]77, even though
   ependymal cells are known to be more numerous in the SVZ niche. This
   could be because ependymal cells fall outside of the scatter gate used
   in flow cytometry or these cells were too big to be uploaded in
   droplets or were sheared in the 10x microfluidic device. Thus,
   ependymal cells were excluded from cell type proportion analyses. (2)
   Neurons. A small number of neurons were captured in the SVZ-targeted
   reprogramming experiments, all of which originated from old + OSKM
   (SVZ) mice but only in one independent experiment. In the other
   independent experiment (and in all other mice in both experiments),
   zero neurons were captured. We thus excluded neurons from cell type
   proportion analyses.

Single-cell RNA-seq differential expression and pathway enrichment analysis

   MAST (implemented in ‘FindMarkers()’ in Seurat) was used for
   differential gene expression analysis for each cell type. Cell types
   with very few cells per condition (OPCs and small clusters 27–30 in the
   iOSKM experiment) were excluded, as were ependymal cells and neurons as
   above. For iOSKM experiments, these analyses were performed with cohort
   2, which included young mice to compare the effects of aging and
   reprogramming. To identify pathways that change with aging and
   reprogramming, the gene list from MAST was ranked by decreasing Z
   score, and then pathway enrichment was performed using the R
   implementation of the Broad Institute’s preranked GSEA algorithm
   (fgsea) and the pathways provided in MSigDB version 7.1. To compare
   trends in whole-body and SVZ-targeted partial reprogramming paradigms,
   pathways were clustered by k-means clustering as implemented in
   pheatmap.

Machine learning models to predict age from cell proportion in the SVZ niche

   A cell proportion linear regression model was trained on previously
   acquired single-cell RNA-seq data from the SVZ neurogenic niche of 28
   male C57BL/6JN mice spanning 3–29 months of age^[393]77, using the
   fraction of each cell type (oligodendrocyte, endothelial, microglia,
   macrophage, astrocyte–qNSC, aNSC–NPC, neuroblast, mural (vascular
   smooth muscle cells and pericytes), ependymal, OPC) and age of the
   mouse of origin. For validation, models were built on 27 of the 28 mice
   and then tested on the remaining mouse (‘leave-one-mouse-out’
   validation). To quantify model performance, the data were presented as
   a correlation between the actual chronological age of the mouse and the
   predicted age for that mouse ([394]Extended Data Fig. 2k). Each mouse
   is represented as a dot. We fitted a linear model (blue line) through
   the points as well as the 95% confidence interval (light gray) using
   ‘geom_smooth’ (ggplot2). Pearson’s correlation (R) is indicated on the
   graph. Model coefficients are as follows: microglia, 176.32;
   oligodendrocyte, 106.15; neuroblast, 92.69; astrocyte–qNSC, 132.32;
   aNSC–NPC, 49.74; endothelial, 12.34; mural, 174.44; OPC, 65.86;
   macrophage, 33.24 (intercept, −94.98).

Protein extraction from tissues and cells for western blotting

   Mice were euthanized using CO[2] inhalation, and then tissues were
   immediately extracted and flash frozen in liquid nitrogen. Brains were
   immersed in ice-cold PBS while regions were dissected, and then
   dissected regions were flash frozen. For protein extraction, ice-cold
   RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% NP-40, 0.5% sodium
   deoxycholate, 0.1% SDS) (Alfa Aesar, J62524) with protease inhibitors
   (Roche, 11836153001) was added to pieces of frozen tissue. Tissues were
   homogenized on ice with a handheld electric homogenizer (Fisher,
   03–421-215) and a microtube pestle attachment (Fisher, 12–141-364).
   Lysates were centrifuged at 16,000g for 20 min at 4 °C to remove tissue
   debris and chromatin, and the pellet was discarded. Protein
   concentration was quantified using the BCA assay (Thermo Fisher
   Scientific, 23225).

   For protein extraction from primary NSCs, to minimize the impact of
   batch effects and confluency, all NSCs for each experiment were plated
   on the same day and collected on the same day. The same number of cells
   was plated per well within each experiment (dose–response,
   [395]Extended Data Fig. 1b, 300,000 cells per well in a 12-well plate;
   time course, [396]Extended Data Fig. 1c, 500,000 cells per well in a
   12-well plate). Cells were treated with doxycycline at the dose or time
   indicated in the respective figures, and then cells were rinsed with
   PBS and then lysed directly in the wells using 200 μl RIPA buffer with
   protease inhibitors and a cell scraper. Lysates were incubated with
   rotation for 30 min at 4 °C and then centrifuged at 16,000g for 20 min
   at 4 °C to remove chromatin and debris, and the pellet was discarded.
   Protein concentration was quantified using the BCA assay (Thermo Fisher
   Scientific, 23225).

   Laemmli buffer (Alfa Aesar, J61337) was added to lysates, and proteins
   were reduced and denatured at 70 °C for 10 min. Proteins were separated
   by electrophoresis in MOPS buffer (Invitrogen, NP0001) on precast 4–12%
   Bis-Tris polyacrylamide gels (Invitrogen, NP0323BOX). For blots with
   all tissues ([397]Extended Data Fig. 1a), equal protein amounts were
   loaded into each lane. For blots with only brain regions ([398]Fig.
   1b), an equal volume of cleared lysate was loaded in each lane to
   maximize signal due to the small size of dissected tissue. For blots
   from cell culture, an equal volume of lysate ([399]Extended Data Fig.
   1c) or the indicated protein amount ([400]Extended Data Fig. 1b) was
   loaded into each lane. Proteins were transferred onto nitrocellulose
   membranes. Ponceau staining was used to visualize protein bands to
   facilitate cutting the membrane to probe for loading controls.
   Membranes were incubated for 1 h at room temperature in blocking buffer
   (PBS with 5% (wt/vol) non-fat dry milk and 0.1% Tween-20). Primary
   antibodies diluted in blocking buffer were incubated overnight at 4 °C.
   Primary antibodies used were as follows: anti-OCT3/OCT4 (1:2,500,
   Abcam, ab181557, clone [401]EPR17929), anti-KPNB1 (1:2,500, Abcam,
   ab2811, clone 3E9), anti-SOX2 (1:1,000, R&D, AF2018, lot KOY0317071)
   and anti-TetR/rtTA (1:1,000, Takara Bio, 631132, clone 9G9). We tested
   other antibodies (R&D, AF3158; Abcam, ab106629; Santa Cruz, sc-40) that
   we were unable to confidently validate in our hands. After three washes
   with PBS with 0.1% Tween-20, secondary antibodies diluted in blocking
   buffer were incubated with the membranes for 1 h at room temperature.
   Secondary antibodies used were as follows: goat anti-rabbit peroxidase
   (1:10,000, Calbiochem, 401393), goat anti-mouse peroxidase (1:10,000,
   Calbiochem, 401215), rabbit anti-goat peroxidase (1:10,000, Calbiochem,
   401515). Membranes were washed with PBS with 0.1% Tween-20. Detection
   was performed using SuperSignal West Dura chemiluminescent substrate
   (Thermo Fisher Scientific, 34076) and film (Thermo Fisher Scientific,
   34090).

Primary neural stem cell isolation and culture

   Primary NSCs from adult mice (either iOSKM or c+iOSKM, see ‘[402]Mouse
   strains, sex and age’) were isolated and cultured based on previously
   published protocols^[403]63,[404]90,[405]91. Briefly, mice were
   euthanized using CO[2] inhalation, and then brains were removed and
   immersed in PBS (Corning, 21–040-CV) with 1%
   penicillin–streptomycin–glutamine (PSQ; Gibco, 10378016) and 0.1%
   gentamicin (Gibco, 15710064) on ice. Both SVZs were dissected as
   described previously^[406]92. Tissue was minced and then transferred to
   a conical tube on ice with fresh PBS with 1% PSQ and 0.1% gentamicin.
   After centrifugation at 500g for 5 min, supernatant was removed, and 5
   ml per brain of digestion buffer (HBSS (Corning, 21–021-CV), PSQ (1%),
   dispase II (1 U ml^−1, Stemcell Technologies, 07913), papain (2.5 U
   ml^−1, Worthington, [407]LS003126), DNase I (250 U ml^−1)) was added.
   Samples were incubated at 37 °C for 40 min on a rocker and then
   centrifuged at 300g for 5 min. Supernatant was removed, and samples
   were triturated 60× with a P1000 pipette in 1 ml Neurobasal-A (NBA;
   Gibco, 10888022) with 1% PSQ and 2% B-27 (minus vitamin A, Gibco,
   12587010), hereafter called ‘NBA/PSQ/B-27’, to obtain single cells.
   After trituration, an additional 9 ml NBA/PSQ/B-27 was added, and then
   cells were centrifuged at 300g for 5 min. The pellet was resuspended in
   5 ml of complete NSC growth medium (NBA/PSQ/B-27 with EGF (20 ng ml^−1,
   PeproTech, AF-100–15) and bFGF (20 ng ml^−1, PeproTech, 100–18B)) and
   transferred to a 6-cm dish. Cells were maintained in a humidified
   incubator at 37 °C with 5% CO[2]. Cells were fed every 2 d by replacing
   half of the conditioned medium with fresh NBA/PSQ/B-27, adding EGF and
   bFGF to a final concentration of 20 ng ml^−1 each. To passage,
   neurospheres were pelleted by centrifugation at 300g for 5 min and then
   incubated with 1 ml Accutase (Millipore, SCR005 or Stemcell
   Technologies, 07920) for 5 min at room temperature. Neurospheres were
   triturated 20× with a P1000 pipette to dissociate them into single
   cells, and then 5 ml PBS was added before centrifugation at 300g for 5
   min. Cells were resuspended in 10 ml NSC growth medium (up to 50% of
   which can be conditioned medium; EGF and bFGF were at a final
   concentration of 20 ng ml^−1 each) and then plated in a 10-cm dish.
   Cells were fed or passaged every 2 d. For protein extraction, RNA
   extraction or AAV infection (see respective sections below),
   neurospheres were dissociated into single cells as above, and then
   cells were plated in wells precoated with poly-d-lysine (0.05 mg ml^−1
   in PBS; Sigma-Aldrich, P6407) and maintained as adherent monolayer
   cultures.

Doxycycline treatment of primary NSCs

   Doxycycline was dissolved in PBS and added to the culture medium at 2
   μg ml^−1, unless otherwise indicated.

RNA extraction and reverse transcription followed by quantitative PCR

   For RNA extraction from primary NSCs, to minimize the impact of batch
   effects and confluency, all cells for each experiment were plated on
   the same day and collected on the same day. For the doxycycline time
   course ([408]Extended Data Fig. 1d), NSCs were plated at 200,000 cells
   per well in a six-well plate. For other experiments, NSCs were plated
   at 250,000 cells per well in a 12-well plate. Total RNA from cultured
   cells was isolated using the Qiagen RNeasy kit according to the
   manufacturer’s instructions. Cells were washed with PBS and then lysed
   in the well using Buffer RLT (350 μl for 12-well plates, 600 μl for
   six-well plates) with β-mercaptoethanol added and a cell scraper.
   On-column DNase digestion was performed to remove genomic DNA. RNA
   concentration was determined using a NanoDrop spectrophotometer or a
   Varioskan LUX reader with a μDrop plate. cDNA was synthesized using the
   High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems,
   4368814) using equal input of RNA for each sample within the
   experiment. cDNA was diluted 1:10 in water, and qPCR was performed
   using iTaq Universal SYBR Green Supermix (Bio-Rad) and run on the
   7900HT system or the QuantStudio 12K Flex instrument (Applied
   Biosystems). All reactions were performed with two to four technical
   replicates. Relative gene expression fold changes were calculated as
   2^−ddCt, normalizing to the geometric mean of housekeeping genes
   (Gapdh, Hprt1, Actb). For wells where product was undetected, the
   maximum cycle number (40) was used for calculations. Primer sequences
   are listed in [409]Supplementary Table 9.

Immunohistochemistry on brain sections

   Mice were terminally anesthetized with Avertin in PBS and then
   transcardially perfused with PBS and heparin followed by PBS with 4%
   paraformaldehyde. Brains were removed, postfixed in PBS with 4%
   paraformaldehyde overnight at 4 °C, washed with PBS and then
   cryoprotected in 30% sucrose at 4 °C until they sank. Tissue was
   embedded in optimal cutting temperature medium (Tissue-Tek, Electron
   Microscopy Services, 62550–12) in cryomolds and frozen in a dry
   ice–ethanol bath. Coronal sections (20 μm) were cut with a Leica
   cryostat and collected on Superfrost Plus slides (Fisher, 12–550-15).

   Coronal sections were washed with PBS, permeabilized in ice-cold
   methanol with 0.1% Triton X-100 for 15 min and then blocked in 5%
   normal donkey serum (ImmunoReagents, SP-072-VX10) with 1% BSA in PBS
   for 30 min. Primary antibodies were diluted in blocking buffer and
   incubated overnight at 4 °C. Primary antibodies used were as follows:
   anti-DCX (1:1,000, Cell Signaling Technology, 4604, lot 7), anti-Ki-67
   (1:1,000, Invitrogen–eBioscience, clone SolA15), anti-mCherry (1:500,
   Invitrogen, [410]M11217, clone 16D7), anti-Cre (1:500, MilliporeSigma,
   MAB3120, clone 2D8), anti-GFAP (1:1,000, Abcam, ab53554, lot
   GR3402854–1), anti-SOX2 (1:500, R&D, AF2018, lot KOY0317071), anti-NeuN
   (1:250, MilliporeSigma, MAB377, clone A60), anti-PSA-NCAM (1:2,000,
   Invitrogen, 14–9118-82, clone 12E3), anti-EGFR (1:100, Millipore,
   06–847, lot 3900816). After three washes with 0.1% Tween-20 in PBS,
   secondary antibodies and DAPI diluted in blocking buffer were incubated
   for 2 h at room temperature. Secondary antibodies used were as follows:
   donkey anti-rabbit, donkey anti-mouse, donkey anti-goat and donkey
   anti-rat, conjugated to AF 488, AF 568, AF 594 or AF 647
   (1:500–1:1,000, Invitrogen), or donkey anti-mouse IgM AF 488 (1:500,
   Jackson ImmunoResearch, 715–545-020). DAPI was used at 1 μg ml^−1.
   Sections were washed three times with 0.1% Tween-20 in PBS and twice
   with PBS, and then a coverslip was added with ProLong Gold antifade
   mounting medium (Invitrogen, [411]P36930). Images were acquired on a
   Zeiss LSM 900 confocal microscope using a ×20 objective and ZEN Blue
   3.0 software.

Imaging and quantification of in situ neuroblast proportion in brain sections

   For quantification of neuroblasts in the SVZ by immunostaining, 20-μm
   coronal brain sections were cut with a Leica cryostat. Every seventh
   section was collected onto the same slide, with five sections per
   slide. One slide per mouse (five coronal sections) was stained as above
   (see ‘[412]Immunohistochemistry on brain sections’). To quantify the
   number of neuroblasts in the SVZ, the entire lateral wall of each
   lateral ventricle was imaged on a Zeiss LSM 900 confocal microscope
   using a ×20 objective and automatic tiling, taking five optical
   sections at 2-μm intervals. Only tiles containing SVZ were imaged; note
   that ImageJ fills the non-imaged tiles with 0 values to make an overall
   rectangular image, and thus these areas appear as black boxes in the
   representative images shown. Cell counting was performed in a blinded
   and semi-automated manner using QuPath 0.2.3. For this, a region of
   interest (ROI) was manually drawn (on blinded and randomized images)
   around the SVZ to exclude irrelevant areas (ventricle, corpus callosum,
   striatum) from total cell counts. Counting was then performed in an
   automated manner in QuPath. Nuclei were segmented based on DAPI
   staining, and then nuclear masks were expanded by 2 μm to define the
   cytoplasm. Cells were identified as DCX^+ based on a threshold for mean
   cytoplasmic DCX intensity and as Ki-67^+ based on a threshold for mean
   nuclear Ki-67 intensity. Cell proportion was calculated as the number
   of positive cells over the total number of cells. Cell density was
   calculated as the number of positive cells over volume (area of ROI × 8
   μm, the thickness of the z stack). For quantification of DCX intensity,
   ImageJ–Fiji version 2.9.0/1.53t was used to generate maximum intensity
   projections for each z-stack image. The integrated density (intensity
   and area) of DCX signal above background was measured in an automated
   manner for each image. Images were blinded and randomized for analysis.

In vivo EdU administration

   EdU (Invitrogen, A10044, [413]E10187) was dissolved in sterile PBS to
   make a 5 mg ml^−1 stock solution and administered by i.p. injection for
   a dose of 50 mg kg^−1 body weight. To count newborn neurons, EdU was
   injected once daily for 2 d (concurrent with doxycycline
   administration), followed by a 5-d chase, repeated for the duration of
   the partial reprogramming experiment. The endpoint for analysis was 16
   d after the first EdU injection and 1 d after the last EdU injection.

Imaging and quantification of newborn neurons in brain sections

   For quantification of newborn neurons in the olfactory bulb by
   immunostaining, 20-μm coronal brain sections were cut with a Leica
   cryostat. Every seventh section was collected onto the same slide, with
   five sections per slide. One slide per mouse (five coronal sections)
   was stained. EdU detection was performed on brain sections after
   permeabilization and before blocking. Following permeabilization,
   sections were washed twice with PBS, and then EdU was detected using
   the Click-iT Plus EdU imaging kit (Invitrogen, [414]C10637) according
   to the manufacturer’s instructions. Blocking and antibody incubation
   were then performed as usual (see ‘[415]Immunohistochemistry on brain
   sections’). The entirety of each olfactory bulb was imaged on a Zeiss
   LSM 900 confocal microscope using a ×10 objective and automatic tiling,
   taking one optical section. Cell counting was performed manually in
   QuPath 0.2.3. An ROI was drawn based on DAPI signal around the central
   layers of the olfactory bulb, up to and including the mitral layer.
   DAPI^+EdU^+NeuN^+ cells within the ROI were counted manually using a
   counting grid overlay in QuPath. Images were blinded and randomized for
   analysis.

Generation of adeno-associated virus

   Viruses were generated and purified by the Stanford Gene Vector and
   Virus Core using the following plasmids. The AAV serotype SCH9
   packaging plasmid^[416]79 was kindly provided by D. Schaffer at the
   University of California, Berkeley. pAAV-EF1a-mCherry-IRES-Cre was a
   gift from K. Deisseroth at Stanford University (Addgene plasmid 55632,
   [417]http://n2t.net/addgene:55632 RRID Addgene_55632). pAAV-EF1a-Cre
   was a gift from K. Deisseroth (Addgene plasmid 55636,
   [418]http://n2t.net/addgene:55636, RRID Addgene_55636).
   pAAV-Ef1a-mCherry was a gift from K. Deisseroth (Addgene plasmid
   114470, [419]http://n2t.net/addgene:114470, RRID Addgene_114470). Viral
   titers (viral genomes (vg) ml^−1) were as follows: AAV-SCH9 Ef1a-Cre,
   1.66 × 10^13 ([420]Figs. 4e and [421]6d,[422]e and [423]Extended Data
   Fig. 5f,[424]i,[425]l,[426]m) or 2.75 × 10^12 ([427]Fig. 4f and
   [428]Extended Data Fig. 5h); AAV-SCH9 Ef1a-mCherry, 2.12 × 10^13
   ([429]Figs. 4e and [430]6d,[431]e and [432]Extended Data Fig.
   5f,[433]i,[434]l,[435]m); AAV-SCH9 Ef1a-mCherry-IRES-Cre, 1.38 × 10^13
   ([436]Extended Data Fig. 5i) or 1.07 × 10^12 ([437]Fig. 4b–[438]d and
   [439]Extended Data Fig. 5h,[440]j,[441]k,[442]m).

Stereotaxic injection of viral vector

   To target reprogramming to the SVZ, stereotaxic surgery was performed
   on c+iOSKM mice (see ‘[443]Mouse strains, sex and age’) to inject virus
   into the right lateral ventricle according to published
   protocols^[444]93. Mice were anesthetized with isoflurane and
   positioned on heating pads in a Kopf (model 940) stereotaxic frame set
   up with a World Precision Instruments (UMP3T-1) UltraMicroPump3 and a
   Hamilton 1710RN 100-μl syringe with a 30G Small Hub RN needle with a
   point 2 beveled end. Injections were made into the right lateral
   ventricles at the following coordinates relative to the bregma:
   lateral, 1.0 mm; anterior, 0.3 mm; ventral depth, 3.0 mm from the skull
   surface. After drilling the skull and inserting the needle into
   position, we waited 5 min before injecting virus. We injected 3 μl of
   virus (AAV-Cre, 2.75 × 10^12 vg ml^−1; AAV-mCherry-IRES-Cre, 1.07 ×
   10^12 vg ml^−1) at a rate of 10 nl s^−1 and then waited 7 min before
   removing the needle and suturing the skin. A single dose of
   buprenorphine SR (0.5 mg kg^−1) was administered subcutaneously for
   postoperative pain management, and animals were monitored daily until
   full recovery.

AAV infection of primary neural stem cells in culture

   Primary NSCs were plated at 250,000 cells per well in a 12-well plate
   precoated with poly-d-lysine. After 24 h, virus was added to the cells
   (1 μl of virus, unless otherwise indicated). After overnight
   incubation, medium was replaced with fresh NSC growth medium. After 48
   h, cells were fixed for immunocytochemistry, lysed for genomic DNA
   extraction, processed for flow cytometry or dissociated for passaging.

   For immunocytochemistry, 8-mm German glass coverslips (Electron
   Microscopy Services, 72296–08) were sterilized in 70% ethanol, air
   dried and then placed into each well and precoated with poly-d-lysine.
   Primary NSCs were plated onto the coverslips and infected with AAV as
   above, and then cells were fixed in 4% PFA in PBS for 15 min at room
   temperature and washed twice with PBS before immunostaining.

   For AAV infection followed by passaging, primary NSCs were plated and
   infected with AAV as above, targeting 95% infection. Cells were then
   dissociated using Accutase, washed and replated in suspension in
   six-well plates. Neurospheres were then cultured and passaged as usual
   for 1–2 weeks to allow for recovery from virus and then plated for
   RT–qPCR (see ‘[445]RNA extraction and reverse transcription followed by
   quantitative PCR’).

Immunocytochemistry of primary neural stem cell cultures

   To detect AAV infection of primary NSCs or to test NSC identity and
   differentiation by immunostaining, cells were grown on coverslips and
   fixed (see ‘[446]AAV infection of primary neural stem cells in culture’
   and ‘[447]In vitro differentiation of primary neural stem cells’).
   Coverslips were permeabilized with 0.1% Triton X-100 in PBS for 10 min
   and then blocked in 2% BSA with 0.2% Tween-20 in PBS (PBST) for 30 min
   (for staining using anti-O4 or anti-PSA-NCAM antibody, cells were not
   permeabilized and detergent was omitted from all steps). Primary
   antibodies were diluted in blocking buffer and incubated for 1 h at
   room temperature or overnight at 4 °C. Primary antibodies used were as
   follows: anti-DCX (1:1,000, Cell Signaling Technology, 4604), anti-DCX
   (1:500, MilliporeSigma, AB2253, lot 3777998), anti-GFAP (1:1,000,
   Abcam, ab53554), anti-Ki-67 (1:1,000, Invitrogen–eBioscience, clone
   SolA15), anti-mCherry (1:1,000, Invitrogen, clone 16D7), anti-GFP
   (1:500, Abcam, ab13970, lot GR3361051–14), anti-Cre (1:500,
   MilliporeSigma, MAB3120, clone 2D8), anti-TUJ1 (1:500, BioLegend,
   802001), anti-O4 (1:500, R&D, MAB1326, clone O4), anti-EGFR (1:500,
   Millipore, 06–847, lot 3173794), anti-nestin (1:250, BD Pharmingen,
   556309, lot 6084618), anti-PSA-NCAM (1:2,000, Invitrogen, 14–9118-82,
   clone 12E3). After three rinses with PBST, secondary antibodies and
   DAPI diluted in PBS were incubated for 30 min at room temperature.
   Secondary antibodies used were as follows: donkey anti-rabbit, donkey
   anti-mouse, donkey anti-goat, donkey anti-rat, conjugated to Alexa
   Fluor (AF) 488, 568, 594 or 647 (1:1,000, Invitrogen, A-21206, A-31573,
   A10042, A-21202, A-31571, A-21447, A-21208, A-21209, A21247); donkey
   anti-mouse IgM AF 488 (1:1,000, Jackson ImmunoResearch, 715–545-020);
   goat anti-chicken IgY AF 488 (1:1,000, Invitrogen, A-11039); donkey
   anti-chicken IgY AF 488 (1:1,000, Jackson ImmunoResearch, 703–545-155);
   donkey anti-guinea pig AF 594 (1:250, Jackson ImmunoResearch,
   706–585-148). DAPI was used at 1 μg ml^−1. Coverslips were rinsed three
   times with PBST, twice with PBS and once with water before mounting
   onto glass slides with ProLong Gold antifade mounting medium.

In vitro EdU treatment and detection

   EdU was dissolved in sterile PBS to make a 10 mg ml^−1 stock solution.
   EdU was added to prewarmed complete NSC growth medium (see
   ‘[448]Primary neural stem cell isolation and culture’), and then one
   volume of this mixture was added to the existing medium in each well
   for a final EdU concentration of 10 μg ml^−1. Cells were incubated for
   1 h before fixation for immunostaining. Following permeabilization, EdU
   was detected using the Click-iT Plus EdU imaging kit (Invitrogen,
   [449]C10639) according to the manufacturer’s instructions. After EdU
   detection, primary and secondary antibody incubations were performed as
   usual (see ‘[450]Immunocytochemistry of primary neural stem cell
   cultures’).

Imaging and quantification of immunocytochemistry

   Samples were imaged using a ×20 objective on a Zeiss LSM 900 confocal
   microscope, taking three optical sections at 1-μm intervals. Eight
   fields of view were imaged for each coverslip. Maximum intensity
   projections were generated in ImageJ–Fiji version 2.9.0/1.53t, and then
   cell counting was performed in an automated manner in QuPath. Nuclei
   were segmented based on DAPI, and then nuclear masks were expanded by 2
   μm to define the cytoplasm. Dead cells were excluded based on nuclear
   size and DAPI intensity, and then cells were classified as positive or
   negative based on signal intensity thresholds for each antibody. Images
   were blinded and randomized for analysis.

Flow cytometry of primary neural stem cells in culture

   Primary NSCs were dissociated using Accutase into a single-cell
   suspension and washed with FACS buffer (PBS with 1% BSA and 0.1%
   glucose). Cells were resuspended in FACS buffer with DAPI (1 μg ml^−1)
   and filtered through a filter-cap 5-ml FACS tube before running on a BD
   LSR II flow cytometer. Debris and doublets were excluded based on
   scatter, and dead cells were excluded based on DAPI ([451]Extended Data
   Fig. 10b–[452]d). Analysis was performed using FlowJo version 10.

Genomic DNA extraction and PCR

   To confirm AAV-Cre-mediated excision of the lox-STOP-lox cassette in
   c+iOSKM cells, genomic DNA was extracted from primary NSCs in culture.
   NSCs were lysed in DirectPCR lysis reagent (Viagen, 101-T) with
   proteinase K added (1:100, Invitrogen, 25530049) and then incubated at
   65 °C for 25 min and 95 °C for 15 min. For PCR, 0.5 μl of crude genomic
   DNA extract was used per 25-μl reaction, with 2× Phusion Master Mix
   with GC Buffer (Thermo, F-532), 1 μM each primer and 2.5% DMSO. Primer
   sequences are listed in [453]Supplementary Table 9. Cycling was
   performed on a preheated block as follows: 98 °C for 3 min, 35 cycles
   (98 °C for 15 s, 65.5 °C for 30 s, 72 °C for 2 min 30 s), 72 °C for 10
   min. The PCR product was mixed with 6× purple loading dye (NEB) and run
   alongside the 1 kb Plus ladder (NEB) on a 1% agarose gel with SYBR
   Safe. Images were acquired on a UV gel imaging station (Analytik Jena
   UVP UVsolo Touch). To confirm the sequence of transgenes for
   single-cell RNA-seq read mapping, portions of the ROSA26 locus and
   transgene were amplified by PCR as above, followed by Sanger sequencing
   using primers tiling the locus. All primer sequences are listed in
   [454]Supplementary Table 9.

Induction of quiescence in primary neural stem cell culture

   Primary NSCs were plated at a density of 250,000 cells per well in a
   12-well plate precoated with poly-d-lysine. Two days later, the medium
   was replaced with quiescence medium (NBA/PSQ/B-27 with bFGF (20 ng
   ml^−1) and BMP4 (50 ng ml^−1, BioLegend)) to induce quiescence^[455]90.
   Cells were maintained in quiescence medium for at least 5 d before AAV
   infection, replacing the medium every 2 d.

In vitro differentiation of primary neural stem cells

   Primary NSCs were freshly isolated from the SVZ and cultured as above
   and used at low passage (passages 3–5), with the exception of the
   experiment in [456]Fig. 4e (used at passages 9–10 due to passaging of
   cells after AAV infection to allow recovery from virus). To assess
   differentiation by immunostaining and microscopy, 8-mm German glass
   coverslips (Electron Microscopy Services, 72296–08) were sterilized in
   70% ethanol, air dried and then placed into 12-well plates (four to
   five per well, arranged to not overlap). Coverslips were coated with
   laminin (5 μg ml^−1, Gibco, 23017015) and poly-d-lysine (50 μg ml^−1)
   in PBS, for 4 h to overnight at 37 °C. Wells were washed twice with PBS
   before plating cells. NSCs were plated into prepared wells at a density
   of 500,000 cells per well in NSC growth medium with or without
   doxycycline and incubated overnight. Next, wells were gently washed
   once with differentiation medium (NBA with 1% PSQ and 2% B-27, no
   growth factors), and then fresh differentiation medium with or without
   doxycycline was added to each well. Half of the medium was replaced
   with fresh differentiation medium with or without doxycycline every 2
   d, avoiding exposure of cells to air. To fix cells for immunostaining,
   one volume of 4% PFA in PBS was added to the existing culture medium in
   each well, and plates were incubated for 5 min at room temperature.
   Liquid was then removed and replaced with 4% PFA in PBS for 10 min at
   room temperature. Wells were washed with PBS, and plates were stored at
   4 °C before immunocytochemistry.

Sample preparation for bulk RNA sequencing

   Primary NSCs were freshly isolated from the SVZ and cultured as above
   and used at passage 3. Twelve-well plates were coated with laminin (5
   μg ml^−1, Gibco, 23017015) and poly-d-lysine (50 μg ml^−1) in PBS.
   Wells were washed twice with PBS before plating cells. Primary NSCs
   were plated into prepared wells at a density of 500,000 cells per well
   in NSC growth medium with or without doxycycline, except for the NSC
   48-h time point, for which cells were plated at a density of 250,000
   cells per well to prevent overgrowth during the experiment. For
   differentiation time points, cells were incubated overnight in NSC
   growth medium before switching to differentiation medium (see ‘[457]In
   vitro differentiation of primary neural stem cells’). One culture
   isolated from an old mouse did not expand sufficiently by passage 3 and
   was thus excluded from the experiment.

   At the appropriate time point, cells were lysed in the well using 350
   μl Buffer RLT with β-mercaptoethanol added and a cell scraper. Lysates
   were transferred to microcentrifuge tubes, flash frozen in liquid
   nitrogen and stored at −80 °C until all samples were collected. Total
   RNA was isolated using the Qiagen RNeasy kit according to the
   manufacturer’s instructions, with on-column DNase digestion. RNA
   concentration and purity were determined using a Varioskan LUX with a
   μDrop plate. For samples with A[260]/A[230] < 1.7, RNA cleanup was
   performed using the Qiagen RNeasy kit according to the manufacturer’s
   instructions. RNA integrity analysis was performed using the
   Bioanalyzer for 16 randomly chosen samples.

   Library construction was performed using 500 ng of each sample as input
   and the Illumina Stranded mRNA Prep (ligation) kit (20040534) and IDT
   for Illumina DNA/RNA UD Indexes, plate A index adaptors, according to
   the manufacturer’s instructions. Insert sizes were verified using the
   Bioanalyzer for a random selection of samples. Concentrations of all
   individual libraries were determined using a Qubit, and then samples
   were combined into a single pooled library. A final AMPure cleanup was
   performed on the pooled library before quality control and sequencing.
   The pooled library was sequenced across three lanes of an Illumina
   NovaSeq X Plus with paired-end 150-bp reads.

Bulk RNA sequencing data processing and analysis

   Clipping of 5′ T overhangs, quality trimming and adaptor trimming were
   performed using Trim Galore (version 0.6.6). Trimmed reads were mapped
   to a custom genome (mm10 with rtTA and each 2A sequence added as
   chromosomes) using STAR (2.7.10b), allowing for chimeric mapping
   (minimum length of 25). Multimapped reads were removed with SAMtools,
   and then features were counted with Subread.

   The count matrix was used to analyze gene expression using DESeq2
   (1.40.1) with a single design variable incorporating time point, age
   and treatment. Differentially expressed genes were identified for
   pairwise comparisons with Wald test P values as implemented in DESeq2.
   Genes were considered significantly differentially expressed if P[adj]
   < 0.1. Heatmaps, hierarchical clustering and PCA were performed on
   VST-transformed values (implemented in DESeq2). To identify pathways
   that change with reprogramming, the complete gene list was ranked by
   decreasing Z score (calculated using log[2] (fold change) and P values
   from DESeq2), and then pathway enrichment was performed using the R
   implementation of the Broad Institute’s preranked GSEA algorithm
   (fgsea) and the GO pathways provided in MSigDB version 7.1.

Statistics and reproducibility

   No statistical methods were used to predetermine sample sizes. Data
   distribution was not formally tested for normality. Two-sided
   Kruskal–Wallis rank-sum test was used to compare more than two groups,
   and two-sided Wilcoxon rank-sum test was used for comparison between
   specific groups, unless otherwise stated in figure legends. Statistical
   tests were performed in R (version 4.0.2) or in GraphPad Prism (version
   9.3.1). Plots were generated in R (ggplot2, Seurat or UpSetR^[458]94)
   or Prism. For sequencing experiments and representative images of
   immunostaining with quantification, the number of independent
   experiments is listed in the figure legend and all data points are
   shown in the figures. For other panels, the numbers of independent
   experiments are as follows: [459]Fig. 1b, 1; [460]Fig. 4b–[461]d, 2;
   [462]Fig. 8d–[463]g, 1; [464]Extended Data Fig. 1a,[465]b, 1;
   [466]Extended Data Fig. 4f,[467]g, 1; [468]Extended Data Fig. 5d, 1;
   [469]Extended Data Fig. 5h, 1 for each batch of virus; [470]Extended
   Data Fig. 5n, 1; [471]Extended Data Fig. 8a,[472]b, 2. For in vivo
   experiments, group allocation was not randomized; rather, animals were
   assigned to experimental versus control groups based on starting weight
   and sex to minimize the impact of these differences. For in vitro
   experiments, randomization was not performed, as each culture was split
   into both experimental and control groups. Randomization and blinding
   were performed to minimize bias from sample preparation for single-cell
   RNA-seq and image quantification. For single-cell RNA-seq, the
   experimenter performing dissections was blinded to the sample
   condition; mice were taken in an alternating fashion to minimize
   impacts of sample processing order. For image quantification, images
   were blinded and randomized for analysis.

Extended Data

Extended Data Fig. 1 |. Characterization of whole-body inducible iOSKM mice
and NSCs.

   Extended Data Fig. 1 |
   [473]Open in a new tab

   a-b, Western blot of tissues from whole-body reprogramming iOSKM mice
   after treatment with doxycycline. Two different exposures of the same
   OCT4 blot are shown due to varying levels of transgene expression
   across tissues. Size in kDa is indicated on the left. Arrowheads
   indicate OCT4 bands. Expression differences between tissues may be due
   to differences in accessibility of the Col1a1 genomic locus and
   doxycycline bioavailability. c, Western blot of primary NSCs from iOSKM
   mice (n = 2 cultures, each from one mouse) treated in culture with
   doxycycline at different doses. For NSCs, protein concentration was
   equalized within each culture; total amount of protein loaded is
   indicated in parentheses. iPSC: induced pluripotent stem cell, as a
   positive control for OCT4 expression. Size in kDa is indicated on the
   left. Normalized densitometry quantification of OCT4 is indicated below
   each lane. KPNB1 is a loading control for OCT4 and a sample processing
   control for SOX2 and rtTA. d, Western blot of primary NSCs from iOSKM
   mice (n = 2 cultures, each from one mouse) treated in culture with
   doxycycline for the indicated amount of time. Normalized densitometry
   quantification of OCT4 is indicated below each lane. KPNB1 is a loading
   control. e, RT-qPCR of OSKM transgene expression in primary NSCs from
   iOSKM mice (n = 8 cultures (Days 0, 2, 4) or 4 cultures (Days 1, 3),
   each culture from one mouse, over 2 independent experiments) during
   doxycycline treatment (green bars) and withdrawal (gray bars). Each dot
   represents expression level from one culture. Bar plot indicates mean
   +/− SEM. FDR-corrected P-values are as follows, using Welch’s ANOVA
   with multiple comparisons, for Day 1–4 vs. Day 0: P = 0.041, 0.019,
   0.017, 0.019. f, Body weight of old (24–28 months) mice during partial
   reprogramming. n = 10 old (5 males and 5 females) and 15 old+OSKM (6
   males and 9 females) mice. Note that not every mouse was weighed at
   every time point. Each dot represents weight for one mouse, normalized
   to starting weight. Data are mean +/− SEM. All P-values > 0.2 by
   two-sided Wilcoxon rank sum test. g, Survival of old (18–28 months)
   mice during partial reprogramming. n = 40 old (25 males and 15 females)
   and 55 old+OSKM (36 males and 19 females) mice. P-value: Mantel-Cox
   test.

Extended Data Fig. 2 |. Single-cell transcriptomics after whole-body partial
reprogramming.

   Extended Data Fig. 2 |
   [474]Open in a new tab

   a, Body weight of mice during doxycycline treatment. n = 3 old and 6
   old+OSKM, all old (18–20 months) males. b, Experimental design for
   first single-cell RNA-sequencing experiment with old (18–20 months)
   whole-body iOSKM mice using 10x Genomics v3 kit. n = 2 male mice per
   condition. c, Example of expression of the lipid-modified
   oligonucleotides used for sample multiplexing of cells from six mice
   into one lane of a 10x Genomics chip. d-e, Dimensionality reduction of
   1194 high-quality cells from single-cell RNA-seq of cohort 1, colored
   by celltype (d) or condition (e). f, Dimensionality reduction from
   single-cell RNA-seq of cohort 1, split by condition. Cell number is
   downsampled to 550 cells per condition. g, Dimensionality reduction
   from single-cell RNA-seq of cohort 2, split by condition. Cell number
   is downsampled to 732 cells per condition. h, Cell type proportions by
   single-cell RNA-seq, as a proportion of total cells recovered per
   mouse. Each dot represents cell proportion for one mouse. Bar plot
   indicates mean. Mice from both independent cohorts are combined: n = 2
   young (3–4 months; 2 males), 5 old (18–20 months or 24–26 months; 4
   males and 1 female), 4 old+OSKM (18–20 months or 24–26 months; 3 males
   and 1 female), over 2 independent experiments (see [475]Fig. 1d,
   [476]Extended Data Fig. 2b, [477]Supplementary Table 1). P-values:
   two-sided Wilcoxon rank sum test. i, Cell type proportions for clusters
   with very few cells, as a proportion of total cells per mouse, and
   normalized as log2 fold change over the mean of old untreated mice
   within each experiment. Bar is mean, each dot is one mouse. n = 2
   young, 5 old, 4 old+OSKM, over 2 independent experiments, as in (h).
   P-values: two-sided Wilcoxon rank sum test. j, Heatmap of the
   expression of the top 5 differentially expressed genes in each cell
   type cluster. Annotation of cell types is indicated on top. Full list
   of marker genes is in [478]Supplementary Table 2. k, Performance of
   linear regression model trained to predict age based on cell type
   proportions in the SVZ from single-cell RNA-seq of 28 mice from 3–29
   months old (see [479]Methods). Scatter plot shows correlation between
   predicted age from using leave-one-mouse-out testing vs. actual
   chronological age of the mouse. Each point represents one mouse. R is
   Pearson’s correlation coefficient. P-value: two-sided. Blue line is
   linear model, gray band is 95% confidence interval. l, Difference
   between predicted age from cell type proportion model and actual
   chronological age for each mouse. Line represents median. n = 2 young,
   5 old, 4 old+OSKM, over 2 independent experiments, as in (h). P-value:
   two-sided Wilcoxon rank sum test.

Extended Data Fig. 3 |. Pathways and transcriptomic clocks after whole-body
partial reprogramming.

   Extended Data Fig. 3 |
   [480]Open in a new tab

   a, Differential gene expression by MAST for each cell type from
   single-cell RNA-seq, comparing old+OSKM vs. young untreated iOSKM mice.
   Each dot represents the Z-score for one gene. Positive Z-score
   indicates upregulation in old+OSKM. Color indicates FDR < 0.2. b, UpSet
   plot showing overlap between significantly changing pathways
   (BH-adjusted two-sided FGSEA P-value, padj < 0.05) after whole-body
   partial reprogramming across cell types. Purple bars at lower left
   indicate total number of significant pathways for each cell type. Main
   bar plot shows number of pathways shared between the cell types
   indicated in the matrix below (note that direction of change may be
   different). c, Cell type-specific transcriptomic aging clocks
   previously trained to predict chronological age using single-cell
   RNA-seq of the SVZ of 28 mice from 3–29 months old were applied to each
   cell type after partial reprogramming. Density plots show the
   distribution of predicted ages of each cell type from each condition.
   The 2 independent experiments ([481]Fig. 1d, [482]Extended Data Fig.
   2b) are shown separately. d-g, Dot plot of gene set enrichment analysis
   comparing the effects of aging and reprogramming in neuroblasts (d),
   aNSCs-NPCs (e), mural cells (f), and endothelial cells (g). Names of
   gene ontology (GO) pathways are listed on the left. padj: BH-adjusted
   two-sided FGSEA P-value. Pathways that are significantly enriched (padj
   < 0.05) in at least one comparison are shown. Size of dot reflects
   adjusted P-value (padj). Color indicates normalized effect size (NES),
   where positive values (red) indicate increase with age (‘aging’) or
   increase with partial reprogramming (‘OSKM’).

Extended Data Fig. 4 |. Neuroblasts and proliferation in the intact SVZ after
whole-body partial reprogramming.

   Extended Data Fig. 4 |
   [483]Open in a new tab

   a, High-magnification images of [484]Fig. 3a. Immunofluorescence
   staining of the SVZ in coronal sections of brains after in vivo partial
   reprogramming. Dorsal is up, lateral is left. Magenta, DCX
   (neuroblasts); green, Ki-67 (proliferative cells). Dashed line
   indicates border of ventricle. One optical section (1.0 μm thick) of a
   confocal image acquired using 63x objective. Scale bars are 25 μm. b-e,
   Quantification of immunofluorescence of the SVZ in coronal brain
   sections, related to [485]Fig. 3. Multiple optical sections totaling 8
   μm thick were acquired for each SVZ, and both hemispheres in five
   coronal sections at 120 μm intervals were quantified per mouse (see
   [486]Methods). n = 11 young (3–4 months; 5 males and 6 females), 8 old
   (20–26 months; 3 males and 5 females), and 10 old+OSKM (20–26 months; 4
   males and 6 females) mice, over 3 independent experiments (for
   individual experiments, see [487]Source Data Table). Each dot
   represents value for one mouse. Data are mean +/− SEM. P-values:
   two-sided Wilcoxon rank sum test. b, Same as [488]Fig. 3e, but symbols
   represent sex. Neuroblast density (number of DCX^+ cells per mm^3) in
   the SVZ per mouse. c, Total DCX fluorescence intensity (intensity x
   area) per mouse, normalized to the mean of old control mice. d,
   Proportion of proliferative cells (Ki-67^+ over DAPI^+) in the SVZ per
   mouse, normalized to the mean of old control mice. e, Density of
   proliferative cells (number of Ki-67^+ cells per mm^3) in the SVZ per
   mouse. Right, symbols represent sex. f, Representative images for
   immunofluorescence staining of the SVZ. Magenta, DCX (neuroblasts);
   green, PSA-NCAM (neuroblasts); blue, DAPI (nuclei). Dorsal is up,
   lateral is right. Yellow rectangle is region shown in insets in (g).
   Scale bar is 100 μm. g, High-magnification images of inset region from
   (f). One optical section of a confocal image acquired using 63x
   objective. Scale bars are 25 μm. h, Quantification of
   immunofluorescence. Fraction of cells expressing DCX or PSA-NCAM that
   are positive for one or both markers. Each bar is one image of an SVZ
   (as shown in i), n = 20 images across 2 young mice. i-k, Quantification
   of immunofluorescence of the SVZ in coronal brain sections, related to
   [489]Fig. 3. Multiple optical sections totaling 8 μm thick were
   acquired for each SVZ, and both hemispheres in five coronal sections at
   120 μm intervals were quantified per mouse (see [490]Methods). n = 9
   young (3–4 months; 4 males and 5 females), 8 old (20–26 months; 3 males
   and 5 females), and 10 old+OSKM (20–26 months; 4 males and 6 females)
   mice, over 3 independent experiments (for individual experiments, see
   [491]Source Data Table). Each dot represents value for one mouse. Data
   are mean +/− SEM. P-values: two-sided Wilcoxon rank sum test. i, Same
   as [492]Fig. 3h, but symbols represent sex. Neuroblast density (number
   of PSA-NCAM^+ cells per mm^3) in the SVZ per mouse. j, Proportion of
   aNSCs/NPCs (EGFR^+PSA-NCAM^− over DAPI^+) in the SVZ per mouse,
   normalized to the mean of old control mice. k, Density of aNSCs/NPCs
   (number of EGFR^+PSA-NCAM^− per mm^3) in the SVZ per mouse. Right,
   symbols represent sex.

Extended Data Fig. 5 |. Characterization of Cre-dependent and inducible
c+iOSKM mice and AAV-SCH9.

   Extended Data Fig. 5 |
   [493]Open in a new tab

   a-c, Representative images (a-b) and quantification (c) of
   immunofluorescence staining to detect mCherry (virus), SOX2 (NSC
   marker), and DCX (neuroblasts) in the SVZ, 7 days post injection of AAV
   into the lateral ventricle. Scale bars are 50 μm (a) and 25 μm (b).
   AAV-mCherry-Cre (AAV SCH9 Ef1a-mCherry-IRES-Cre) was injected into the
   lateral ventricle of the right hemisphere for experimental animals
   (pink, n = 4 mice), and ipsilateral and contralateral hemispheres
   relative to injection site were quantified separately. The right
   hemisphere was counted for uninjected control (n = 1 mouse). In (a),
   yellow arrows indicate mCherry+SOX2+ cells; mCherry+ population
   includes both NSCs (SOX2+) and non-NSC cells (SOX2-). Each dot
   represents cell counts for one image, shape denotes an individual
   mouse. Box plot indicates median (center line), upper and lower
   quartiles (box limits), and 1.5x interquartile range (whiskers).
   P-values: two-sided Wilcoxon rank sum test. d, Representative images of
   immunofluorescence of the olfactory bulb and cerebellum using an
   antibody against mCherry (virus), 4–6 days post injection of
   AAV-mCherry-Cre (AAV SCH9 Ef1a-mCherry-IRES-Cre) into the lateral
   ventricle. No mCherry signal was detected. Scale bars are 500 μm. e,
   Primer sites for PCR from genomic DNA to detect Cre-mediated
   recombination of lox-STOP-lox cassette in c+iOSKM cells. f, PCR from
   genomic DNA of NSCs after infection with AAV (0.125 μl to 1μl, 2-fold
   dilutions). AAV-mCherry: AAV SCH9 Ef1a-mCherry. AAV-Cre: AAV SCH9
   Ef1a-Cre. Colored arrows indicate expected size of products with and
   without recombination. Equal numbers of cells were plated before
   infection; note that we observed a correlation between cell death and
   high virus doses that is reflected in the amount of PCR product (for
   example, see 3.2 kb band in mCherry-infected cells). Representative of
   n = 2 cultures. g, Scheme for in vitro AAV infection of primary NSCs
   isolated from c+iOSKM mice. All lots of virus were tested for
   infectivity/recombination in primary NSC culture by flow cytometry
   and/or immunofluorescence. h, Immunofluorescence staining of NSCs after
   infection with the indicated AAVs, showing efficient infection (Cre,
   mCherry) and recombination (GFP). Composite includes DAPI (blue). Scale
   bars are 50 μm. i, Flow cytometry of primary NSCs after infection with
   AAV carrying mCherry only, Cre only, or mCherry-IRES-Cre. j-k, Flow
   cytometry of NSCs after infection with increasing doses of AAV
   mCherry-IRES-Cre ( j) and quantification of populations (k). l, Related
   to [494]Fig. 4e. RT-qPCR in primary c+iOSKM NSCs after infection with
   AAV-mCherry or AAV-Cre, showing expression of the respective viral
   genes. n = 4 young (3–4 months; 3 males and 1 female) and 4 old (20–23
   months; 3 males and 1 female) cultures for Cre, n = 3 young (3–4
   months; 2 males and 1 female) and 4 old (20–23 months; 3 males and 1
   female) cultures for mCherry, over 1 independent experiment. Each dot
   represents expression level in one culture; each culture was isolated
   from one mouse. Bar plot indicates mean +/− SEM. P-values: two-sided
   Wilcoxon rank sum test. m, RT-qPCR in quiescent NSCs, showing AAV-SCH9
   infection (mCherry), Cre-mediated recombination (Egfp), OSKM transgene
   induction (Oct4-P2A), and total Oct4 expression (Oct4). n = 2 cultures;
   one independent experiment. Each dot represents expression level in one
   culture; each culture was isolated from one mouse. Bar plot indicates
   mean +/− SEM. n, Western blot of primary NSCs from c+iOSKM mice,
   infected in culture with AAV-mCherry (control) or AAV-Cre, then treated
   with doxycycline. Size in kDa is indicated on the left.

Extended Data Fig. 6 |. Single-cell transcriptomics after SVZ-targeted
reprogramming.

   Extended Data Fig. 6 |
   [495]Open in a new tab

   a, Cell type proportions by single-cell RNA-seq, as a proportion of
   total cells recovered per mouse. Each dot represents cell proportion
   for one mouse. Bar plot indicates mean. n = 5 young, 6 old, 6 old+OSKM
   (SVZ), over 2 independent experiments. P-values: two-sided Wilcoxon
   rank sum test. b, Cell type proportions for small clusters and immune
   cells, as a proportion of total cells per mouse, and normalized as log2
   fold change over the mean of old untreated mice. Each dot represents
   cell proportion for one mouse. Bar plot indicates mean. n = 5 young, 6
   old, 6 old+OSKM (SVZ), over 2 independent experiments. P-values:
   two-sided Wilcoxon rank sum test. c, Dot plot of rtTA-IRES-Egfp
   expression by cell type from single-cell RNA-seq. Size of dot reflects
   percentage of cells within cluster with non-zero expression. Color
   indicates average expression level. d, Difference between predicted age
   from cell type proportion model and actual chronological age for each
   mouse. Line represents median. n = 5 young, 6 old, 6 old+OSKM (SVZ),
   over 2 independent experiments. P-value: two-sided Wilcoxon rank sum
   test. e, UpSet plot showing overlap between significantly changing
   pathways (BH-adjusted two-sided FGSEA P-value, padj < 0.05) after
   SVZ-targeted partial reprogramming across the six most abundant cell
   types. Purple bars at lower left indicate total number of significant
   pathways for each cell type. Main bar plot shows number of pathways
   shared between the cell types indicated in the matrix below (note that
   direction of change may be different). f, Dot plot of gene set
   enrichment analysis comparing the effects of aging and SVZ-targeted
   reprogramming on expression of inflammation pathways across cell types.
   Names of gene ontology (GO) pathways are listed on the left. padj:
   BH-adjusted two-sided FGSEA P-value. Size of dot reflects adjusted
   P-value (padj). Color indicates normalized effect size (NES), where
   positive values (red) indicate increase with age (‘aging’) or increase
   with SVZ-targeted reprogramming (‘OSKM’). g, Cell type-specific
   transcriptomic aging clocks previously trained to predict chronological
   age using single-cell RNA-seq of the SVZ of 28 mice from 3–29 months
   old were applied to each cell type after SVZ-targeted reprogramming.
   Density plots show the distribution of predicted ages of each cell type
   from each condition. h, Survival of old (19–28 months) mice during
   SVZ-targeted partial reprogramming. n = 14 old (8 males and 6 females)
   and 14 old+OSKM (SVZ) (7 males and 7 females) mice. P-value: Mantel-Cox
   test. i-j, Dot plot showing normalized enrichment score from gene set
   enrichment analysis in aNSCs-NPCs (f) or neuroblasts (g) after
   SVZ-targeted reprogramming (old+OSKM (SVZ) vs. old control). GO pathway
   names are listed on the left. padj: BH-adjusted two-sided FGSEA
   P-value. Color reflects adjusted P-value (padj). Size of dot reflects
   number of genes in set. Significantly enriched pathways (padj < 0.05)
   are shown.

Extended Data Fig. 7 |. Comparison of whole-body and SVZ-targeted
reprogramming.

   Extended Data Fig. 7 |
   [496]Open in a new tab

   a, Scatter plots showing normalized enrichment scores from gene set
   enrichment analysis, comparing SVZ-targeted reprogramming (old+OSKM
   (SVZ) vs. old) vs. whole-body partial reprogramming (old+OSKM vs. old)
   for common cell types. Each point represents the normalized enrichment
   scores for one pathway. For all panels, padj: BH-adjusted two-sided
   FGSEA P-value. Pathways where adjusted P-value (padj) < 0.05 for at
   least one comparison are shown. Color indicates padj for SVZ-targeted
   reprogramming. Pathways in the upper right quadrant and lower left
   quadrant change in the same direction in both partial reprogramming
   paradigms. b, Heatmap of pathway enrichment for each cell type,
   comparing normalized enrichment score (NES) in whole-body (old+OSKM vs.
   old) and SVZ-targeted (old+OSKM (SVZ) vs. old) reprogramming. Pathways
   where padj < 0.05 for at least one paradigm are included. Pathways are
   divided into 4 clusters by k-means clustering. Color reflects NES. Size
   is number of pathways falling into each cluster. c, UpSet plot showing
   overlap between cell types of pathways increasing (top) or decreasing
   (bottom) in both whole-body and SVZ-targeted partial reprogramming
   (padj < 0.05 for at least one paradigm). Bars at lower left indicate
   total number of pathways meeting these criteria for each cell type.
   Main bar plot shows number of pathways shared between the cell types
   indicated in the matrix below. d-e, Heatmap showing normalized
   enrichment score (NES) from gene set enrichment analysis in aNSCs-NPCs
   (d) and neuroblasts (e), comparing whole-body and SVZ-targeted partial
   reprogramming. GO pathway names are listed on the right. Pathways are
   divided into 4 clusters by k-means clustering. Color reflects NES.
   Pathways where padj < 0.05 for at least one comparison are included.

Extended Data Fig. 8 |. Differentiation of primary NSCs in culture.

   Extended Data Fig. 8 |
   [497]Open in a new tab

   a, Representative images of immunofluorescence staining of primary NSC
   cultures. Primary NSCs express markers of activated NSCs (Ki-67, GFAP,
   EGFR, and Nestin). Scale bar is 50 μm. b, Representative images of
   immunofluorescence staining after in vitro differentiation of primary
   NSCs using markers for neurons (TUJ1, NeuN), astrocytes (GFAP), and
   oligodendrocytes (O4). Scale bar is 25 μm. c, Experimental schematic
   for immunofluorescence of primary NSC cultures from iOSKM mice with
   partial reprogramming, used in panels d-j. PDL: poly-D-lysine. n = 9
   young (3–4 months; 4 males and 5 females) and 9 old (24–26 months; 4
   males and 5 females), over 2 independent experiments. Each culture was
   isolated from one mouse. d-e, Representative images of
   immunofluorescence staining for NSC markers (panel d: EGFR, green;
   Nestin, magenta), proliferation markers (panel e: Ki-67, magenta; EdU,
   green) and neuroblast markers (panel e: DCX, red) in primary NSCs from
   young mice before differentiation. Scale bar is 50 μm. f-j,
   Quantification of immunofluorescence in primary NSCs. n = 9 young (3–4
   months; 4 males and 5 females) and 9 old (24–26 months; 4 males and 5
   females), over 2 independent experiments. Each dot represents the
   proportion of cells positive for the indicated marker in one culture,
   where each culture was isolated from one iOSKM mouse. Box plot
   indicates median (center line), upper and lower quartiles (box limits),
   and 1.5x interquartile range (whiskers). P-values: two-sided Wilcoxon
   rank sum test. k, Representative image of immunofluorescence staining
   for neuroblasts (PSA-NCAM, green) after in vitro differentiation for 4
   days. Scale bar is 50 μm. l, Quantification of immunofluorescence
   staining for neuroblasts (PSA-NCAM^+ over DAPI^+). n = 8 young (3–4
   months; 4 males and 4 females) and 8 old (24–26 months; 4 males and 4
   females), over 2 independent experiments. Each dot represents
   neuroblast proportion for one culture, where each culture was isolated
   from one iOSKM mouse. Box plot indicates median (center line), upper
   and lower quartiles (box limits), and 1.5x interquartile range
   (whiskers). P-values: two-sided Wilcoxon rank sum test. m-n, Related to
   [498]Fig. 6a–[499]f. Quantification of immunofluorescence for
   proliferative cells (Ki-67^+, m) and glia (sum of GFAP^+ cells and
   SOX10^+ cells, n) after 4 days of differentiation with partial
   reprogramming. n = 8 young (3–4 months; 4 males and 4 females) and 8
   old (24–26 months; 4 males and 4 females), over 2 independent
   experiments. Each dot represents proliferative proportion for one
   culture, where each culture was isolated from one iOSKM mouse. Box plot
   indicates median (center line), upper and lower quartiles (box limits),
   and 1.5x interquartile range (whiskers). P-values: two-sided Wilcoxon
   rank sum test. o-p, Related to [500]Fig. 6d,[501]e. Quantification of
   immunofluorescence for neuroblasts (DCX^+) and proliferative cells
   (Ki-67^+) after AAV infection and 4 days of differentiation with
   partial reprogramming. n = 4 young (3–4 months; 3 males and 1 female)
   and 4 old (20–23 months; 3 males and 1 female), over 1 independent
   experiment. Each dot represents proliferative proportion for one
   culture, where each culture was isolated from one c+iOSKM mouse. Box
   plot indicates median (center line), upper and lower quartiles (box
   limits), and 1.5x interquartile range (whiskers). P-values: two-sided
   Wilcoxon rank sum test. q, Related to [502]Fig. 8a–[503]c.
   Quantification of immunofluorescence for total neuronal cells (sum of
   DCX^+ cells and TUJ1^+ cells, representing neuroblasts and neurons)
   after 8 days of differentiation with partial reprogramming. n = 8 young
   (3–4 months; 4 males and 4 females) and 8 old (24–26 months; 4 males
   and 4 females), over 2 independent experiments. Each dot represents
   total neuronal proportion for one culture, where each culture was
   isolated from one iOSKM mouse. Box plot indicates median (center line),
   upper and lower quartiles (box limits), and 1.5x interquartile range
   (whiskers). P-values: two-sided Wilcoxon rank sum test.

Extended Data Fig. 9 |. Bulk RNA-seq of primary NSCs with partial
reprogramming and differentiation.

   Extended Data Fig. 9 |
   [504]Open in a new tab

   a, Heatmap of expression of the top variable genes across all samples
   from bulk RNA-seq of NSCs with reprogramming and differentiation. Color
   indicates expression, scaled by row. Condition and time point are
   indicated at the top. Sex-related genes are removed for visualization.
   b, Expression of selected genes that change across differentiation.
   Each dot represents normalized count for one culture, where each
   culture was isolated from one iOSKM mouse. n = 5 young (3–4 months; 2
   male and 3 female) and 4 old (25 months; 2 male and 2 female) cultures.
   Box plot indicates median (center line), upper and lower quartiles (box
   limits), and 1.5x interquartile range (whiskers). P-values: two-sided
   Wilcoxon rank sum test. c, Expression of the reprogramming factors in
   old control and old+OSKM NSCs (6h time point). Each dot represents
   normalized count for one culture, where each culture was isolated from
   one iOSKM mouse. n = 4 cultures (25 months; 2 male and 2 female). Box
   plot indicates median (center line), upper and lower quartiles (box
   limits), and 1.5x interquartile range (whiskers). P-values: two-sided
   Wilcoxon rank sum test. d-f, Dot plot showing normalized enrichment
   score from gene set enrichment analysis after partial reprogramming in
   NSCs (old+OSKM vs. old control) at 6h (d), 18h (e), and 48h (f) time
   points. GO pathway names are listed on the left. padj: BH-adjusted
   two-sided FGSEA P-value. Color reflects padj. Size of dot reflects
   number of genes in set. Top and bottom 15 (by NES) significantly
   enriched pathways (padj < 0.05) are shown. g, Dotplots of gene set
   enrichment analysis comparing the effects of aging and reprogramming at
   each differentiation time point. Names of gene ontology (GO) pathways
   are listed on the left. padj: BH-adjusted two-sided FGSEA P-value. Size
   of dot reflects padj. Color indicates normalized effect size (NES). h,
   Differential gene expression by DESeq2 for each time point, comparing
   old vs. young (left) or old+OSKM vs. old (right). Each dot represents
   the Z-score for one gene. Positive Z-score indicates upregulation.
   Color indicates FDR < 0.1. i, Heatmap of expression of genes that are
   significantly differentially expressed in NSCs between old and old+OSKM
   cultures at the 18h time point. Color indicates expression, scaled by
   row. Condition and time point are indicated at the top. Genes are
   separated into 4 clusters by k-means clustering. j, UpSet plot showing
   overlap between differentially expressed genes (DEGs) after partial
   reprogramming (old+OSKM vs. old control) in in vitro time points
   (adjusted P-value < 0.2 by DESeq) and in vivo paradigms (adjusted
   P-value < 0.3 by MAST in aNSC-NPC, neuroblast, or astrocyte-qNSC
   clusters).

Extended Data Fig. 10 |. FACS gating and Western blots.

   Extended Data Fig. 10 |
   [505]Open in a new tab

   a, Representative gating for FACS sorting of live cells from the SVZ
   for single-cell RNA-seq. Population shown is listed above each plot. b,
   Representative gating strategy to identify live cells from primary NSCs
   in culture. Population shown is listed above each plot. c, Experimental
   schematic for flow cytometry analysis of AAV infection of primary NSCs
   in culture. d, Representative gating strategy for analysis of mCherry
   (virus) and EGFP (recombination marker). Live cells (see b) are shown.
   Each plot is a culture treated with the virus listed above the plot.
   Note that these plots are also shown in [506]Extended Data Fig. 5i.
   e-j, Full Western blots. Corresponding figure panels are indicated
   above the blots. Dotted rectangles indicate cropped area shown in
   corresponding figure.

Supplementary Material

   Supplementary Material
   [507]NIHMS2069809-supplement-Supplementary_Material.xlsx^ (79.5MB,
   xlsx)
   Source Data
   [508]NIHMS2069809-supplement-Source_Data.xlsx^ (67.7KB, xlsx)

Acknowledgements