SOX2 reprograms resident astrocytes into neural progenitors in the adult brain

Abstract

Glial cells can be in vivo reprogrammed into functional neurons in the adult CNS; however, the process by which this reprogramming occurs is unclear. Here, we show that a distinct cellular sequence is involved in SOX2-driven in situ conversion of adult astrocytes to neurons. This includes ASCL1(+) neural progenitors and DCX(+) adult neuroblasts (iANBs) as intermediates. Importantly, ASCL1 is required, but not sufficient, for the robust generation of iANBs in the adult striatum. These progenitor-derived iANBs predominantly give rise to calretinin(+) interneurons when supplied with neurotrophic factors or the small-molecule valproic acid. Patch-clamp recordings from the induced neurons reveal subtype heterogeneity, though all are functionally mature, fire repetitive action potentials, and receive synaptic inputs. Together, these results show that SOX2-mediated in vivo reprogramming of astrocytes to neurons passes through proliferative intermediate progenitors, which may be exploited for regenerative medicine.

Graphical abstract

Figure 1

Expression and Reprogramming Ability of SOXB1 Factors (A) Immunohistochemistry (IHC) showing expression of SOXB1 factors in striatal regions with iANBs at 5 weeks post-injection (wpi) of SOX2 virus. SOX1 and SOX2, but not SOX3, can be identified in some of the induced DCX⁺ cells (right panels at higher magnifications). Hst, Hoechst 33342. The scale bars represent 20 μm. (B) Ectopic SOX2, but not the other SOXB1 factors, induce striatal DCX⁺ cells. Lentivirus expressing individual SOXB1 factors was injected into the adult mouse striatum and examined 4 or 5 weeks (wk) later. The scale bars represent 20 μm. See also Figure S1.

Figure 2

SOX2-Driven Reprogramming Does Not Pass through a NSC Stage (A) A genetic approach to trace NSCs and their progeny. Adult Nes-CreER^TM;Rosa-YFP mice were injected with tamoxifen (Tam) for 7 days and examined 3 wk later. (B) Endogenous NSCs and DCX⁺ cells in the lateral ventricle (LV) and dentate gyrus (DG) are genetically traced. The scale bars represent 20 μm. (C and D) iANBs do not pass through a nestin⁺ NSC stage. Adult mutant mice were treated with Tam 2 (C) or 3 (D) wpi of SOX2-expressing virus and examined at the indicated time points. DCX⁺ cells in the LV were used as endogenous controls. The scale bars represent 20 μm. See also Figure S2.

Figure 3

SOX2 Induces ASCL1⁺ Neural Progenitors (A) The expression of ASCL1⁺ cells in striatal regions with DCX⁺ iANBs at 5 wpi. The scale bar represents 20 μm. (B) A time course analysis of ASCL1⁺ cells in the reprogramming area (mean ± SD; n = 3 mice at each time point). (C) ASCL1 is detected in astrocytes transduced with SOX2 lentivirus. The co-expressed GFP marker is under the control of the human GFAP promoter. The scale bar represents 20 μm. (D) A genetic approach to trace derivatives of ASCL1⁺ progenitors. SOX2-driven reprogramming was induced in adult Ascl1-CreER^T2;Rosa-tdTomato (tdT) mice. (E) SOX2-induced DCX⁺ cells pass through an ASCL1⁺ progenitor stage. Confocal images show genetic labeling of iANBs. An orthogonal view is shown in the right panel. The scale bar represents 20 μm.

Figure 4

ASCL1 Is Required, but Not Sufficient, for SOX2-Driven Reprogramming (A and B) Schematic diagrams show experimental designs. Ascl1 is deleted in the adult astrocytes after Tam treatment of Ascl1^f/f;Cst3-CreER^T2 mice. Four weeks later, in vivo reprogramming was initiated by injection of SOX2-expressing lentivirus and examined at 5 wpi. (C) Deletion of Ascl1 dramatically reduces the induction of DCX⁺ cells (mean ± SD; n = 6 for vehicle-treated and n = 4 for Tam-treated mice; ^∗∗p = 0.0013 by Student’s t test). (D) The experimental design for examining ASCL1 reprogramming ability. (E) DCX⁺ cells were not detected in striatal regions with ectopic ASCL1 when examined at 1, 2, 3, or 5 weeks post virus injection. SOX2-induced DCX⁺ cells were used as positive controls. The scale bar represents 20 μm.

Figure 5

Reprogramming of Resident Astrocytes to Calretinin⁺ Neurons (A) Experimental schemes. Newly generated cells were labeled by BrdU incorporation. Neuronal survival and maturation were promoted by valproic acid (VPA) or neurotrophic factors. (B) Quantification of newly generated calretinin (CR)⁺ neurons under the indicated conditions (mean ± SD; n = 3 mice at each condition). BDNF, brain-derived neurotrophic factor; Nog, noggin. (C) Confocal images showing BrdU-labeled CR⁺ neurons that were induced by SOX2. An orthogonal view of the boxed region is also shown. The scale bar represents 20 μm. (D and E) SOX2-induced CR⁺ neurons originate from resident astrocytes. Adult resident astrocytes were genetically traced in Cst3-CreER^T2;Rosa-YFP mice after Tam treatment. These cells were then in vivo reprogrammed by ectopic SOX2 and analyzed 10 weeks later. Soma and processes of the astrocyte-converted CR⁺ neurons are indicated by arrows and arrowheads, respectively. The scale bar represents 20 μm. (F and G) SOX2-induced CR⁺ neurons pass through an ASCL1 progenitor stage. The descendants of ASCL1⁺ progenitors were genetically traced in Ascl1-CreER^T2;Rosa-tdT mice. A majority of the reprogrammed CR⁺ neurons were traced by the marker tdT. The scale bar represents 20 μm. See also Figure S3.

Figure 6

Functional Properties of SOX2-Induced Neurons (A) Representative confocal images of the four types of induced neurons traced in the striatum of adult Ascl1-CreER^T2;Rosa-tdT mice and infused with biocytin. The scale bars represent 10 μm. (B–D) Characterization of the four types of induced neurons based on input resistance (R_in) and resting membrane potentials (V₀). (E) A majority of the recorded neurons have a capacitance around 30 pF. (F) Representative action potential traces from each type of reprogrammed neurons. (G) All the induced neurons have a similar action potential (AP) threshold. (H and I) A majority of the induced neurons have higher AP amplitudes and frequencies. (J) Type IV induced neurons have wider AP half-width. (K–M) Properties of spontaneous postsynaptic currents (sPSCs). Whereas there is a broader range of sPSC frequencies, their amplitude is rather similar (n.a., not available). For all panels, n = 20 cells from 11 mice for type I neurons, n = 2 cells from two mice for type II neurons, n = 2 cells from one mouse for type III neurons, and n = 3 cells from three mice for type IV neurons. See also Figure S4.

Figure 7

Stepwise Reprogramming of Resident Astrocytes to Functional Neurons in the Adult Brain