In vivo fate analysis reveals the multipotent and self-renewal capacities of Sox2+ neural stem cells in the adult hippocampus

Abstract

To characterize the properties of adult neural stem cells (NSCs), we generated and analyzed Sox2-GFP transgenic mice. Sox2-GFP cells in the subgranular zone (SGZ) express markers specific for progenitors, but they represent two morphologically distinct populations that differ in proliferation levels. Lentivirus- and retrovirus-mediated fate-tracing studies showed that Sox2+ cells in the SGZ have potential to give rise to neurons and astrocytes, revealing their multipotency at the population as well as at a single-cell level. A subpopulation of Sox2+ cells gives rise to cells that retain Sox2, highlighting Sox2+ cells as a primary source for adult NSCs. In response to mitotic signals, increased proliferation of Sox2+ cells is coupled with the generation of Sox2+ NSCs as well as neuronal precursors. An asymmetric contribution of Sox2+ NSCs may play an important role in maintaining the constant size of the NSC pool and producing newly born neurons during adult neurogenesis.

Figure 1. Sox2-GFP cells in the SGZ represent dividing and undifferentiated cell populations

In Sox2-GFP transgenic mice, GFP expression in the SGZ (A) faithfully mimicked endogenous Sox2 mRNA in the adult hippocampus (B). A higher magnification view of in situ signal is displayed in the inset (B). Radial (arrow) and non-radial (arrowhead) Sox2-GFP cells were found in the SGZ (C). Radial Sox2-GFP cells showed co-localization with radial glial cell markers, including NESTIN (D), BLBP (E) and GFAP (F) in their processes. MUSASHI-1 was detected in Sox2-GFP cells in soma (G). Some Sox2-GFP cells were active in proliferation, displaying BrdU labeling (H) and co-expression with Ki67 (I). The majority of Sox2-GFP cells did not express the differentiated neuronal markers, such as NEUN (J), TuJ1 (K), and DCX (L). Sox2-GFP cells in the SGZ did not represent differentiated glial cells. Their co-localization with S-100β (M), GST-π, and NG2 (N) was not evident in the SGZ. Some GFP⁺ cells were co-stained with astrocyte marker, S-100β, in the hilus (M, arrow) or in the deep granular layer. Abbreviations in (A) and (C): g, granular layer; sg, subgranular zone; h, hilus; m, molecular layer. Scale bars: 50 μm (A and B), 10 μm (C to N).

Figure 2. Sox2-GFP cells are the origin of in vitro NSCs

The proliferation capacity of Sox2-GFP cells was measured by FACS analysis. Sox2-GFP cells, which comprised only 6% of the total hippocampus cells, survived and expanded, forming the majority of colonies in vitro (A). Western analysis showed the progress of differentiation in time course (B). The expressions of neuronal (TuJ1) and glial markers (GFAP) were associated with the down-regulation of SOX2. Immunocytochemistry revealed the differentiation of in vitro cultured Sox2-GFP cells to neural lineages (C). Note that co-culture paradigm with primary neurons was used for differentiation of Sox2-GFP cells to oligodendrocytes (RIP⁺ cells). Legends in (B); 0, one day after plating cells but still culturing in FGF2, EGF containing growth medium; 1-7, 1 to 7 days after substituting growth medium with Forskolin-containing differentiation medium. Scale bar: 50 μm.

Figure 3. Sox2⁺ cells proliferate to produce differentiated neural lineages as well as undifferentiated cells

A fate mapping is schematized in (A). Lentivirus containing Sox2-GFP/CRE was injected into the dentate gyrus of ROSA26-loxP-Stop-loxP-GFP reporter mice. Transduction of CRE recombinase deleted “STOP” codon to activate GFP-reporter expression (colored in green) in Sox2⁺ cells. The recombination in the genomic level allowed tracing of both targeted cells and their progeny. Ten or 28 days after virus injection, BrdU (colored in red) was administered to label newly born cells from the targeted cells (A). One month after BrdU injection, the fate of progeny was examined with cell-type specific markers. Sox2⁺ cells underwent cell proliferation and gave rise to neuronal precursors positive for NEUROD (B) or DCX (C) as well as PROX1⁺ granular neurons (D). GFAP⁺ (E) or S-100β⁺ (F) glial cells were also generated from Sox2⁺ cells. Sox2⁺ cells also have the potential to give rise to undifferentiated cells positive for SOX2 (G) or BLBP (H). Abbreviations: s, subgranular zone; g, granular layer; m, molecular layer. Scale bar: 10 μm.

Figure 4. Lineage tracing of Sox2⁺ cells at a single cell level

Sox2-GFP/Cre retrovirus was injected into the dentate gyrus of ROSA26R mice to activate a reporter (β-gal or GFP reporter), which was used to trace the fate of progeny of Sox2⁺ cells. The majority of clones were single-cell clusters showing that targeted Sox2⁺ cells (GFP⁺ or β-gal⁺ cells) became a neuron (A, NeuN⁺), an astrocyte (B, GFAP⁺) or a NSC. Sox2⁺ cells were able to give rise to multiple NSCs (C, SOX2⁺, arrow and arrowhead) or neurons (D, NeuN⁺, arrow and arrowhead) in the hippocampus. Multi-cell clusters containing heterogeneous cell populations were also identified. One clone showed Sox2⁺ NSC could give rise to one neuron (NeuN⁺, arrow) and one astrocyte (GFAP⁺, arrowhead) (E). Some clones contained a Sox2⁺ NSC (SOX2⁺, arrowhead) and a neuron (NeuN⁺, arrow) that were physically associated each other (F). Sox2⁺ NSC was able to give rise to a neuron (NeuN⁺, right arrow) and a Sox2⁺ NSC (SOX2⁺, left arrow) that has GFAP expression (two arrowheads) in the radial process (G). Abbreviations: s, subgranular zone; g, granular layer; m, molecular layer. Scale bar: 10 μm.

Figure 5. Increased proliferation of Sox2⁺ NSCs is indicative of enhanced neurogenesis in the adult hippocampus

Transgenic mice were housed in running wheel cages for 7 days. During the running period, they received a daily BrdU injection to examine the proliferation effects of Sox2-GFP cells in the SGZ (A). Representative images with antibody staining used for quantitative analysis are displayed in (B). Total number of dividing cell population (BrdU and Ki67) significantly increased in running mice (C). While numbers of DCX⁺ precursors or immature neurons expanded, the number of Sox2-GFP cells remained unchanged (D) even after more newly born Sox2-GFP cells were generated (E). Increased numbers of Sox2⁺ NSCs and DCX⁺ precursors were in cell cycle, indicating that running affected those two cell types (F). When neurogenesis increased in running mice, radial Sox2-GFP cells proliferated, showing Ki67 expression and BrdU incorporation (G). However, the total number of radial Sox2-GFP did not increase significantly (H), although 6% of total radial GFP cells proliferated in running mice (I). Abbreviations in graphs: NR, non-runner; R, runner; rGFP, radial Sox2-GFP cells. Scale bars: 50 μm (B) or 10 μm (G).

Figure 6. Contribution of Sox2⁺ NSCs to adult neurogenesis

A. Non-radial Sox2⁺ NSCs have potentials to generate identical cells (SOX2⁺) and give rise to downstream neural cell types, suggesting that non-radial Sox2⁺ cells retain the self-renewing and multipotent NSCs. Sox2⁺ NSCs preferentially gave rise to neurons, presumably due to the influence of stronger neurogenic niche. Non-radial Sox2⁺ NSCs also have the potential to give rise to radial Sox2⁺ NSCs, suggesting that the equilibrium between radial and non-radial Sox2⁺ NSCs may have a significant role in the homeostatic control of adult neurogenesis. B. Sox2⁺ NSCs proliferated and generated increased numbers of newly born Sox2⁺ cells in response to mitotic signals (running induced). However, the increase in new Sox2⁺ cells did not contribute to expansion of the Sox2⁺ NSC pool; the total number of Sox2⁺ NSCs remained unchanged. Increased proliferation of Sox2⁺ NSCs was associated with the generation of new DCX⁺ precursors, which subsequently led to production of new neurons in the SGZ. Thus, population-wise, proliferation of Sox2⁺ NSCs was coupled with the generation of neuronal precursors as well as maintenance of NSCs. Note that asymmetric contribution is used to explain the generation of two different cell types but does not imply asymmetric division of individual cells.