Inducible motor neuron differentiation of human induced pluripotent stem cells in vivo

Abstract Objectives Transplantation of neural progenitor cells (NPCs) derived from human‐induced pluripotent stem cells (hiPSCs) is one of the promising treatment strategies for motor neuron diseases (MNDs). However, the inefficiency in committed differentiation of NPCs in vivo limits its application. Here, we tried to establish a potential therapeutic strategy for MNDs by in vivo directional differentiation of hiPSCs engineered with motor neuron (MN) specific transcription factors and Tet‐On system. Materials and Methods We engineered hiPSCs with three MN‐specific transcription factors and Tet‐On system. The engineered cells were directly transplanted into immunodeficient mice through subcutaneous, intra‐spinal cord and intracerebroventricular injections. Following doxycycline (Dox) induction, teratoma formation, and motor MN differentiation were evaluated. Results We generated genetically engineered hiPSCs, in which the expression of Ngn2, Isl1, and Lhx3 was controlled by a drug‐inducible transgenic system. These cells showed normal pluripotency and proliferative capacity, and were able to directionally differentiate into mature motor neurons (MNs) and NPCs with high efficiency in spinal cords and cerebral lateral ventricles under the induction of Dox. The grafts showed long‐term survival in the recipient mice without formation of teratoma. Conclusions The induced mature MNs and NPCs were expected to replace the damaged endogenous MNs directly, and play a role of de novo stem cell stock for long‐term neuron damage repair, respectively. Therefore, in vivo directional differentiation of the hiPSCs engineered with MN‐specific transcription factors and Tet‐On system via Dox induction could be a potential therapeutic strategy for MNDs with high efficacy and safety.

long-term neuron damage repair, respectively. Therefore, in vivo directional differentiation of the hiPSCs engineered with MN-specific transcription factors and Tet-On system via Dox induction could be a potential therapeutic strategy for MNDs with high efficacy and safety. progenitor cells (NPCs) obtained from embryonic tissues or pluripotent cells. NPCs can self-renew and differentiate into astrocytes, oligodendrocytes, and neurons to build neural networks. 4 However, these cells are likely to differentiate into glial cells rather than into functional neurons, which is a disadvantage for neuron replacement therapy. 5 As a solution, neuronal restricted progenitors and motor neuron progenitors (MNPs) derived from human embryonic stem cells (hESCs)/human-induced pluripotent stem cells (hiPSCs) have been engineered and developed into neurons rather than glial cells or other cell types in vivo and in vitro. 6,7 However, the purity and differentiation potential of these neural cells are substantially diminished during cell passage. 8 Although hESC/hiPSC-derived MNs can replace the lost neuronal population directly without glial formation, the terminally differentiated neurons are source limited, fragile, and likely to die after transplantation. 9 The in situ direct conversion of endogenous astrocytes to functional neurons by the ectopic expression of defined factors 10 or the knockdown of RNA-binding protein PTB in vivo is another therapeutic strategy for neurodegenerative disorders. 11 However, the potential adverse effects caused by local astrocyte depletion and microenvironment alteration are unknown. Neurodegenerative diseases are often accompanied by gene mutations that can hardly be repaired by in situ conversion, resulting in the failure of disrupted circuit reconstruction.
If the injury or degeneration is severe, then in vivo cell conversion might not be sufficient to generate enough cells. Exogenous functional cells or artificial tissues can provide a rich source to repair tissue loss. 12 Neural-specific genes can convert engineered human fibroblasts and astrocytes into neurons in vivo 10 ; however, the conversion efficiency is extremely low (0.4%-5.9%), resulting in limited cell regeneration.
Three MNs that induce transcription factors, namely, Ngn2, Isl1, and Lhx3, can efficiently induce functional MNs with mature electrophysiological properties from hESCs or hiPSCs. 13

| Plasmid construction
The PB-Ngn2-Isl1-Lhx3-BSD (PB-NIL) plasmid was described in De Santis et al. 16 The PB-Bcl-xL-Luciferase-GFP (PB-BLG) plasmid was generated by inserting the sequences of Bcl-xL (gene ID 397536) and firefly luciferase-GFP (LG, pGL4.21) in the PB vector. pGL4.21 was purchased from Promega. For further differentiation, the culture medium was switched to the MN induction medium. (i) In vitro differentiation was evaluated by immunofluorescence staining using MN, NPC, and MNP markers;

| Cell experiments
(ii) the functional characteristics of hiPSC-derived MNs were determined by patch-clamp recordings; and (iii) the pro-survival capacity of Bcl-xL was determined by replating-induced stress 17 and glutamate toxicity assay. 18

| Statistical analysis
All statistics, including statistical tests, sample sizes, and types of replicates, were described in the figure legends. A p value of <0.05 was considered statistically significant.
Detailed experimental procedures are described in the Supporting Information Material S1.

| Validation of the directional differentiation of multi-gene-modified hiPSCs in vitro
Two different in vitro models were used to determine the prosurvival capacity of Bcl-xL: replating-induced stress 17 and glutamate excitotoxicity. 18 In the first model, 64.93% ± 8.47% of MNs from NILB-hiPSCs survived, which was higher compared with the 9.94% ± 2.68% of MNs from NIL-hiPSCs after passage in vitro      Figure S4 and S5 (w.p.t.) ( Figure 3A, lower panel). Engraftments were retrieved at different time points (1, 2, 4, and 6 w.p.t.). The NILB-hiPSCs in the untreated control group formed a teratoma-like structure with increasing size over time, and those in the Dox-treated group showed a white and loose engraftment at 2 w.p.t., that became small and compacted at 4 w.p.t. and further shrunk at 6 w.p.t. (Figure S3A). Histological test by H&E showed that the tissues from the untreated group exhibited the typical structures of tri-germ layers at 6 w.p.t. (Figure S3B, left panel). For the treated group, the structures of trigerm layers were not discerned, but a pyramid neuron-like structure was observed at 2 w.p.t., and massive ventricle-like cavities were formed at 4 w.p.t. (Figure S3B, right panel). The tissues from the treated group were further evaluated by immunofluorescence staining to determine the differentiation status of the transplanted cells. At 1 w.p.t., most of the cells were human nuclear antigen positive (hNuclei + ), that is, they were transplanted human cells. Many of the human cells expressed TUJ1 (68.93% ± 9.61%; Figures 3B and S3C).

| In vivo induction of NILB-hiPSCs after intraspinal cord and intracerebroventricular injections
Age-related decline of neurogenic niche in the brain, 19 as well as decreased migration capacity of grafts along with age of the recipient animals have been reported previously. 20,21 Therefore, neonatal mice (postnatal day 4) were chosen to test effects of in vivo induction after the engineered hiPSCs were transplanted into the spinal cords and cerebral ventricles. The luciferase signal was markedly reduced at 2 w.p.t., but bounced back at 4 w.p.t., and remained stable from 4 to 20 w.p.t. (Figures 4A,B and S4A,B). Immunofluorescence staining revealed that in the brain of mice at 2 w.p.t., the injected NILB-hiPSCderived cells were dispersed throughout the cerebral ventricles and formed extensive dendritic arborizations ( Figure S4C). At 2 w.p.t., the vast majority of hNuclei + cells in mice (68.81% ± 3.53% in spinal cords; 75.70% ± 2.88% in cerebral ventricles) were positive for the neuronal marker TUJ1 (Figures 4C and S4D,E), indicating the predominant differentiation of transplanted NILB-hiPSCs to neurons in the central nervous system (CNS). Further immunostaining showed that the grafts expressed mature MN markers, including MAP2 (57.56% ± 4.58% in spinal cords; 54.67% ± 11.21% in cerebral ventricles), HB9 (45.43% ± 9.75% in spinal cords; 40.26% ± 2.16% in cerebral ventricles), and ChAT (59.92% ± 8.72% in spinal cords; 62.81% ± 14.50% in cerebral ventricles; Figures 4C and S4D,E). Moreover, the electrophysiological characteristics of transplanted cells were examined at 4 w.p.t. Patch-clamp recording in GFP + neurons showed the robust trains of action potentials and large voltage-gated sodium/potassium currents ( Figures 4D and S4F), suggesting that the transplanted NILB-hiPSCs can be directionally converted to electrophysiologically mature MNs in the CNS after Dox induction.
Rosette-like structures and ventricle-like cavities were also observed in the brain and spinal cord cryosections of mice at 4 w.p.t.
( Figures 4E and S5A). Proliferating marker Ki67 + was also intermixed with the cells expressing SOX2 and PAX6 ( Figures 4F and S5A), and this finding was consistent with the bioluminescent imaging data above ( Figures 4A,B and S4A,B). Unexpectedly, a small portion of transplanted cells expressed OLIG2, suggesting the presence of MNPs ( Figures 4G and S5A). None of the grafts remained pluripotent (negative for NANOG) or formed teratomas ( Figure S5B). All these data indicated that the transplanted cells can differentiate into MNs, NPCs, and MNPs in the CNS after Dox treatment and the differentiated cells could survive for a long term.

| DISCUSSION
In this study, we engineered hiPSCs with the three MN transcription factors Ngn2, Isl1, and Lhx3 and Tet-On system. The engineered cells could directionally and efficiently differentiate into MNs and NPCs in vitro and in vivo by Dox induction.
Ngn2, Isl1, and Lhx3 can directionally induce pluripotent cells and fibroblasts into mature MNs. 16,17 This method can bypass the neural progenitor stage, which has been observed in conventional multi-step differentiation techniques using small molecules. 22 form MAP + mature neurons in ALS rats. 24,25 The MNs induced by our strategy can rapidly replace the damaged endogenous MNs, which is particularly an advantage in curing acute spinal cord injury.
In this work, NILB-hiPSCs had differentiated not only into mature MNs but also to NPCs in vitro and in vivo; this phenomenon was not observed in previous studies. 16,17 Previous dynamics studies showed that Ngn2 plays a crucial role in the outcomes of neural differentiation: Ngn2 induces functional neurons when its expression is sustained but induces NPCs when its expression oscillates. 26 However, the regulating mechanism for oscillatory versus sustained Ngn2 expression remains to be determined. A recent study showed that Ngn2-modified messenger RNA (mmRNA) can simultaneously program hiPSCs into neurons and NPCs due to the intrinsic fluctuations in mmRNA and protein levels. 27 28 Only small portion of NPCs exist in the adult brain and are restricted in the hippocampus and striatum. They remain in a rest state and will be activated to proliferate, migrate, and differentiate to replace lost neurons after an injury. 29 However, these NPCs have an extremely limited capability for regeneration and thus cannot effectively replace injured neurons. The de novo NPCs derived from NILB-hiPSCs could survive long-term in vivo and may play the role of stem cell backup to fix injured neurons. This characteristic should be a potential advantage in curing chronical neurodegenerative diseases, such as ALS.
HiPSCs have a tendency to form a teratoma, which is the main concern of clinical practices. The generated NILB-hiPSCs also developed a teratoma after their subcutaneous transplantation without Dox treatment. However, after Dox treatment for only 5 days, the destination was narrowed to neural cell lineage, and no tumour formation was noted in the subcutaneous tissues. The induced MNs gradually dwindled and ultimately disappeared at 8 w.p.t., which differed from the situation in the CNS. We speculate that this phenomenon probably occurred because the neurons could not adapt to the non- increasing angiogenesis should be conducted prior to the transplantation to enhance their survival capacity. 30 Despite these beneficial potentials, the safety risks of engineered pluripotent cells are still the most important challenge for their clinical application. Overdose of Dox treatment may lead to side effects on gastrointestinal tract, 31 while the insufficient induction caused by the inaccessibility of Dox to some cells will make grafts out of control, resulting in teratoma formation. In addition, the insertion of a foreign gene in an inappropriate site in the stem cells might give rise to tumorigenesis. Therefore, before human clinical trials, it is necessary to conduct experiments on the large animal models, such as pigs, dogs and monkeys to determine the optimized dosage and administration route. Besides, inducible suicide systems could be introduced into hiPSCs to reduce the risk of uncontrollable pluripotent cells by killing the grafted tumorigenic cells 32 or the undifferentiated cells. 33 In summary, we describe a rapid, effective, and long-lasting