Roles of Transcription Factors in the Development and Reprogramming of the Dopaminergic Neurons

The meso-diencephalic dopaminergic (mdDA) neurons regulate various critical processes in the mammalian nervous system, including voluntary movement and a wide range of behaviors such as mood, reward, addiction, and stress. mdDA neuronal loss is linked with one of the most prominent human movement neurological disorders, Parkinson’s disease (PD). How these cells die and regenerate are two of the most hotly debated PD research topics. As for the latter, it has been long known that a series of transcription factors (TFs) involves the development of mdDA neurons, specifying cell types and controlling developmental patterns. In vitro and in vivo, TFs regulate the expression of tyrosine hydroxylase, a dopamine transporter, vesicular monoamine transporter 2, and L-aromatic amino acid decarboxylase, all of which are critical for dopamine synthesis and transport in dopaminergic neurons (DA neurons). In this review, we encapsulate the molecular mechanism of TFs underlying embryonic growth and maturation of mdDA neurons and update achievements on dopaminergic cell therapy dependent on knowledge of TFs in mdDA neuronal development. We believe that a deeper understanding of the extrinsic and intrinsic factors that influence DA neurons’ fate and development in the midbrain could lead to a better strategy for PD cell therapy.


Introduction
The meso-diencephalic dopaminergic (mdDA) neurons are the primary dopamine (DA) sources in the mammalian central nervous system [1]. DA, as a neurotransmitter, plays a critical role in responding to ever changing environmental conditions, such as movement, reward, punishment, salience, learning, cognition, love, pleasure, and drug addiction [2][3][4][5][6]. Importantly, mdDA neuronal loss is linked with one of the most prominent human movement neurological disorders, Parkinson's disease (PD) [7]. PD is the second most common neurodegenerative disease, characterized by motor symptoms including static tremor, rigidity, bradykinesia, postural disorder, and non-motor symptoms including sensory and cognitive impairment [1,8]. The main pathological feature of PD is the irreversible mdDA neuronal degeneration in the substantia nigra pars compacta (SNpc), which primarily impairs striatal dopaminergic innervation and causes cardinal motor symptoms [1,9].
Currently, the main treatments for PD still focus on increasing DA levels or regulating DA transmission by pharmacotherapy. Deep brain stimulation is also applied for patients who experience a prominent tremor or uncontrolled motor fluctuations [10]. However, these treatments can only alleviate the physical symptoms rather than prevent or delay

Induction of Midbrain FP
When the ectoderm thickens to form the neural plate, the trajectory of neural development begins. This plate begins to fold inward and merge, resulting in neural tube formation ( Figure 1A). Recent research has identified that the sex-determining region Y-box 2-to-Brachyury (Sox2-to-Bra) ratio is involved in neural tube specification [22]. The neural tube consists of a cluster of neural progenitor cells (NPCs) and post-mitotic neurons. Foxa2 and its downstream effector RNF152 regulates cell proliferation via the mTOR pathway to influence FP cell number [23]. The FP is a neural tube's ventral midline signaling center, where mdDA neurons come from, while the roof plate (RP) is placed in the dorsal midline [24]. The ectoderm and the RP produce a bone morphogenic protein (BMP) [25]. Shh is secreted from the notochord underlying the ventral neural tube and the FP [26]. Foxa2 regulates the expression of sonic hedgehog (Shh), which in turn activates Glioma-associated 3 of 23 oncogene homolog 1 (Gli1) and Foxa2 (Figure 2A(I)). The dorsoventral (DV) axis is defined by BMP and Shh morphogen gradients. The precise spatial and temporal interplay of signaling along the DV axis was used to specify NPCs into different neuronal types in each designated area [27,28].
pathway to influence FP cell number [23]. The FP is a neural tube's ventral midline signaling center, where mdDA neurons come from, while the roof plate (RP) is placed in the dorsal midline [24]. The ectoderm and the RP produce a bone morphogenic protein (BMP) [25]. Shh is secreted from the notochord underlying the ventral neural tube and the FP [26]. Foxa2 regulates the expression of sonic hedgehog (Shh), which in turn activates Glioma-associated oncogene homolog 1 (Gli1) and Foxa2 (Figure 2A(I)). The dorsoventral (DV) axis is defined by BMP and Shh morphogen gradients. The precise spatial and temporal interplay of signaling along the DV axis was used to specify NPCs into different neuronal types in each designated area [27,28]. Otx and Gbx2 function in opposition to one another to establish the position of the IsO, which defines the midbrain-hindbrain boundary. IsO regulates Fgf8, which, along with Shh, specifies the location of midbrain mdDA neuron growth. Shh stimulates Foxa2 expression, and Wnt1 is expressed in this region and required for midbrain development. Lmx1a is defined in the ventral midbrain; Pitx3 plays a role in the mdDA neuron differentiation. Abbreviations: TFs, transcription factors; IsO, isthmic organizer; Otx, Orthodenticle homeobox 2; Gbx2, Gastrulation brain homeobox 2; Fgf8, Fibroblast growth factor 8; Foxa2, Forkhead box A2; Lmx1a, homeobox transcription factor 1 A; Pitx3, Paired-like homeodomain3; and Shh, Sonic Hedgehog. Otx and Gbx2 function in opposition to one another to establish the position of the IsO, which defines the midbrain-hindbrain boundary. IsO regulates Fgf8, which, along with Shh, specifies the location of midbrain mdDA neuron growth. Shh stimulates Foxa2 expression, and Wnt1 is expressed in this region and required for midbrain development. Lmx1a is defined in the ventral midbrain; Pitx3 plays a role in the mdDA neuron differentiation. Abbreviations: TFs, transcription factors; IsO, isthmic organizer; Otx, Orthodenticle homeobox 2; Gbx2, Gastrulation brain homeobox 2; Fgf8, Fibroblast growth factor 8; Foxa2, Forkhead box A2; Lmx1a, homeobox transcription factor 1 A; Pitx3, Paired-like homeodomain3; and Shh, Sonic Hedgehog.
As the neural tube develops, it generates four major morphogenetic domains along the anterior-posterior (A-P) axis: the prosencephalon, the mesencephalon, the rhombencephalon, and the end of the neural tube, which will eventually form the cerebrum, the midbrain, the rest of the brainstem, the cerebellum, and the spinal cord, respectively ( Figure 1B). The isthmic organizer (IsO), which defines the midbrain-hindbrain boundary (MHB), appears during neural tube development on embryonic days 7-8 (E7-8). The IsO function is similar to the medial FP in that it is an important signaling center for mdDA neuron formation, and it determines the separate embryonic development of the midbrain and hindbrain [29,30]. The position of IsO is determined by the expression boundaries of TFs Orthodenticle homeobox 2 and Gastrulation brain homeobox 2 (Otx2 and Gbx2) in the central nervous system: the former is expressed in the forebrain and midbrain. At the same time, the latter is defined in the anterior hindbrain. Disruption of the Otx2/Gbx2 border causes an IsO positional shift with an expanded or diminished mid or hindbrain. Fibroblast growth factor 8 (Fgf8) is solely expressed on the hindbrain side of IsO, whereas Winglesstype MMTV integration site family member 1 (Wnt1) is expressed on the midbrain side ( Figure 1B) [31,32]. The primary role of Wnt1 is to regulate mdDA neuron development and maintain the expression of Engrailed-1 (En1), which may act through increasing En1 transcriptional activity to induce midbrain specification [33,34]. Furthermore, TFs including En1/2, Foxa1/2, Gbx2, Otx2, and secreted factors including Wnt1, Fgf8, and Shh are fundamental for induction and specification mdDA progenitor cells ( Figure 1B) [24,[35][36][37]. However, the interaction between morphogens and TFs is still not fully understood, which is crucial in exploring these mechanisms further.

Figure 2.
TFs and related molecules involved in developing mdDA neurons at various embryonic stages. (A) mdDA neuron induction, specification, differentiation, and maturation. The use of arrows denotes stimulatory effects, while perpendicular lines denote inhibitory effects. (A(I)) The orange area clusters the TFs and molecules involved in forming mdDA neurons from the regional specification and induction stage (see text for details). (A(II)) The green section groups together the TFs involved in the FP specification (see text for details). (A(III)) The TFs and related molecules implicated in mdDA differentiation are depicted in blue. (A(IV)) The pink area represents the TFs' involvement in the expression of mdDA neurons maturation markers. (B) The mdDA neuron migration paths. Black dotted arrows indicate radial migration regulated by chemokine (C-X-C motif) and ligand 12 (CXCL12). The red dotted arrows indicate tangential migration regulated by the L1 cell adhesion molecule (L1CAM), the L1CAM ligand-protein tyrosine phosphatase (L1CTP), and the reelin signaling pathway. Abbreviations: FP, floor plate; RP, roof plate; Gli1, Glioma-associated oncogene homolog1; En1, Engrailed-1; Msx1, Msh homeobox 1; Nurr1, Nuclear receptor-related fac-tor1; Th, tyrosine hydroxylase; Aadc, amino acid decarboxylase; Vmat2, vesicular monoamine transporter 2; Dat, dopamine transporter; Drd2, dopamine D2 receptor.
As the neural tube develops, it generates four major morphogenetic domains along the anterior-posterior (A-P) axis: the prosencephalon, the mesencephalon, the rhombencephalon, and the end of the neural tube, which will eventually form the cerebrum, the midbrain, the rest of the brainstem, the cerebellum, and the spinal cord, respectively (Figure 1B). The isthmic organizer (IsO), which defines the midbrain-hindbrain boundary (MHB), appears during neural tube development on embryonic days 7-8 (E7-8). The IsO function is similar to the medial FP in that it is an important signaling center for mdDA neuron formation, and it determines the separate embryonic development of the midbrain and hindbrain [29,30]. The position of IsO is determined by the expression boundaries of TFs Orthodenticle homeobox 2 and Gastrulation brain homeobox 2 (Otx2 and Gbx2) in the central nervous system: the former is expressed in the forebrain and midbrain. At the same time, the latter is defined in the anterior hindbrain. Disruption of the Otx2/Gbx2 border causes an IsO positional shift with an expanded or diminished mid or hindbrain. Fibroblast growth factor 8 (Fgf8) is solely expressed on the hindbrain side of IsO, whereas Wingless-type MMTV integration site family member 1 (Wnt1) is expressed on the midbrain side ( Figure 1B) [31,32]. The primary role of Wnt1 is to regulate mdDA neuron development and maintain the expression of Engrailed-1 (En1), which may act through increasing En1 transcriptional activity to induce midbrain specification [33,34]. Furthermore, TFs including En1/2, Foxa1/2, Gbx2, Otx2, and secreted factors including Wnt1,

Specification of mdDA Progenitors
The programming of FP cells results in the induction of the NPCs into mdDA progenitors, which is essential for the specification of mdDA progenitors. It has been demonstrated that a significant number of TFs are involved in this process. The first expression of Lmx1a appears around E8.5 in the midbrain FP, and the FP cells are differentiated into mdDA progenitors by expressing Foxa2 and Lmx1a [38]. Lmx1a/b cooperate to regulate neurogenesis via Msh homeobox 1 (Msx1)-mediated Ngn2 regulation (Figure 2A(II)) [24,36]. Furthermore, Lmx1a/b plays an essential role in regulating the proliferation of mdDA progenitors by Wnt1 and Ngn2 [39]. The number of mdDA progenitors and mdDA neurons reduces significantly in Lmx1a knock-out and dreher mice, partly attributable to a loss in neurogenesis and proliferation of mdDA progenitors [39,40]. However, the specification of mdDA progenitors failed to be affected in specific inactivation of Lmx1b in FP at E9. Therefore, Lmx1b can act as a compensatory function for Lmx1a by promoting the specification and proliferation of mdDA progenitors.
The study has shown that the absence of En1 does not affect the identity of mdDA progenitors and early post-mitotic neurons, but the down-regulation of some mdDA neuron markers (Pitx3; tyrosine hydroxylase, Th; dopamine transporter, Dat; vesicular monoamine transporter 2, Vmat2; and amino acid decarboxylase, Aadc) was found in the rostral-lateral DA domain of En1 −/− embryos [41]. Therefore, En1 may play a more critical role in the maturation of mdDA neurons than specification. Foxa1/2 have many overlapped bound sites, suggesting their cooperating role in developing mdDA neurons. Foxa1/2 activates Lmx1a/b, which regulates the specification and differentiation of mdDA neurons [39,42]. Moreover, Foxa1/2, an E-box transcription factor and Otx2 are directly involved in the transcriptional activity of a Neurog2 enhancer during the differentiation of mdDA neurons. Neurog2 starts to express at E10.75 in the rostral FP later than all three TFs, suggesting Neurog2 needs to be activated by other factors [43]. Otx2 plays an essential role in positioning MHB and regulates progenitor domains' specification in the ventral midbrain (VM). It has been identified that Otx2 conditional deletion in the VM results in abnormal expression of Shh, NK homeobox factor 6.1 and NK homeobox factor 2.2 (Nkx6.1 and Nkx2.2) and changed mdDA progenitors specification [44]. Furthermore, the Otx2 deletion in VM impairs the proliferation of Sox2 + mdDA progenitors, resulting in the inhibition of Lmx1a, Msx1, Ngn2, and Ascl1 expression in mdDA progenitors. Otx2 overexpression increases the number of mdDA neurons but not those in other progenitor domains, implying that Otx2 can selectively regulate the proliferation of mdDA progenitors [45].

Differentiation and Maturation
After completing the cell cycle and entering the post-mitotic stage, mdDA progenitors differentiate into mature mdDA neurons. Nurr1 continues to be expressed in the midbrain from E10.5 to adulthood [46]. Nurr1 can influence mature mdDA neuron markers such as Th, Dat, Aadc, and Vmat2 (Figure 2A(III)), and Nurr1-deficient mice cannot develop mdDA neurons and die shortly after birth [46,47]. In either Foxa1 or Foxa2 conditional deletion at E11.5, the number of Nurr1 + post-mitotic mdDA neurons is reduced, and the differentiation from Nurr1 + immature to Nurr1 + Th + mdDA neurons is inhibited. Foxa2 has been found to influence neuronal differentiation by binding directly to a Neurog2 enhancer and activating Smarca1, promoting the induction of mature mdDA neurons in the developing mouse midbrain [43]. Furthermore, Foxa2 binding to Nurr1 promotes the expression of DA phenotypes throughout the development of mdDA neurons [48,49].
The onset of Th and Pitx3 expression is approximately around E11.5 [50]. The ventrolateral mdDA neurons express Pitx3 before Th, whereas the dorsomedial ones express Th before Pitx3. In the absence of Pitx3, SNpc neurons lose at the beginning of their terminal differentiation. Consistent with this, a recent study demonstrates that the lack of Pitx3 in Pitx3 fl/fl/ Dat CreERT2 mice causes a rapid reduction of DA content and a severe loss of mdDA neurons but not ventral tegmental area (VTA) [51]. Therefore, Pitx3 is essential in the terminal differentiation of the SNpc [52]. Other TFs, including En1/2 and Otx2, also play crucial roles in the differentiation and maturation of mdDA neurons. Previous studies have indicated that En1-null mice have a significant loss in the entire SNpc and most VTA [41,53]. En1/2 is also required to survive the mature mdDA neurons (Figure 2A(III,IV)) [54,55]. Overexpressed Otx2 in the midbrain results in the increase of Nurr1 + immature mdDA neurons and Pitx3 + mature neurons through enhancing the proliferation of mdDA progenitors, while the deletion of Otx2 leads to the failure of mdDA progenitors differentiation into Nurr1 + post-mitotic mdDA neurons because the expressions of Lmx1a, Msx1, Ngn2 and Mash1 in mdDA progenitors are lacking. Therefore, Otx2 might regulate the expression of Lmx1a to involve mdDA neurons' differentiation.

Migration
The differentiation and maturation of mdDA neurons are accompanied by their migration. mdDA neurons migrate away from the FP of the ventral mesencephalon, called the ventricular zone (VZ), into the mantle layer and form three different mdDA neuron clusters: SNpc (A9) on the lateral side, VTA (A10) on the medial side, and the posterior retro-rubral field (RRF; A8) [37]. Previous research shows that mdDA neurons migrate first radially between E11.5-13.5 after exiting the cell cycle, and then some mdDA neurons begin to migrate tangentially into the laterally-situated SNpc medially-located VTA at E12.5 ( Figure 2B) [56]. Furthermore, not all SNpc mdDA neurons migrate tangentially after radial migration; some mdDA neurons migrate radially into the dorsal SNpc [57]. It has been proposed that the chemokine (C-X-C motif) ligand 12 (CXCL12) and its receptor CXCR4 [58] are involved in regulating the radial migration, and L1 cell adhesion molecule (L1CAM) and potential L1CAM ligand-protein tyrosine phosphatase, receptor type Z, polypeptide 1 (PTPRZ1) [37] modulate the tangential migration of mdDA neurons as well as Reelin signaling [56,59]. Recent research found that Netrin-1 mediates dorsal mdDA neuron migration into SNpc along radial glia fibers, and axon-derived Netrin-1 attracts GABAergic neurons into substantia nigra pars reticula (SNr) [57]. Moreover, early B-cell factor 1 (Ebf1), an essential TF in B-lymphocyte differentiation, has been implicated in mdDA neuron migration. Cellular Ngn2 disruption causes both an arrest of cell migration and a failure of cell differentiation [60].
During the differentiation of mdDA progenitors at E11.5, Sox2 + Lmx1a + progenitors with low Sox6 expression are located in the medial VZ domain, while Nolz1 expresses laterally with high Sox6 expression. After exiting the cell cycle, mdDA neurons begin to migrate: Sox6 + mdDA neurons, located in the medial area, migrate ventrally and then tangentially to end up in the lateral SNc at E18.5; Otx2 + or Otx2 + Nolz1 + cells, positioned in the lateral domain, extend ventrally and are confined to the VTA [61]. Furthermore, many studies have found that Otx2 is limited to VTA neurons at the post-mitotic stage [62,63]. Even with what has already been done, it is still difficult to ascertain how Otx2 and Sox6 are involved in mdDA neuron migration and the precise interaction mechanism between these TFs and mdDA neuron subtypes.

Dopamine Cell Therapy
The current pharmacological approach for PD patients is still the primary management to alleviate or control motor symptoms. The treatment is generally aimed at increasing DA bioavailability, either by replenishing the DA precursors or inhibiting DA's breakdown. However, it cannot directly replace the lost pathway. Regenerative medicine-based solutions are being aggressively pursued to restore dopamine levels in the striatum via several emerging techniques designed to reconstruct the nigrostriatal pathway. Currently, several different types of stem cells have attempted to regenerate mdDA neurons. Due to their self-renewing and multipotent features, the most commonly used stem cells are embryonic stem cells (ESCs), mesenchymal stem cells (MSCs), and pluripotent stem cells [24]. Several trials with grafts integrated into the host brain have shown restored DA release, re-innervated striatum, and alleviated clinical symptoms of motor dysfunction. In certain hopeful situations, patients may be able to discontinue L-dopa medication following transplantation [64,65]. However, multiple concerns remain that most in vivo transplanted NSCs become glial cells rather than neurons, and only 5-10% of NSCs survive after transplantation due to the toxic effect of the inflammatory state [66][67][68]. As a result, it is critical to identify a method to protect DA neurons from neuroinflammation and enhance their survival and differentiation. Among the most effective and attractive methods, TFs-based therapeutics have gained considerable interest. The TFs involved in enhancing mdDA neuron cell development will be reviewed in depth below, and they can be employed alone or in combination.

Nurr1
Nurr1, also known as NR4A2/NOT/TINUR, is a member of the orphan nuclear receptor family 4 (NR4A), necessary for the mdDA neurons' development, maturation, and functional maintenance [46,69,70]. It is also involved in neuroprotection and neuroinflammation regulation by inhibiting pro-inflammatory factors in microglial and astrocytes cells [71][72][73]. Notably, in mdDA neurons, the decreased levels of Nurr1 were characterized during aging, which may be related to the increased morbidity of PD [74,75]. Furthermore, Nurr1 is required for neuronal plasticity remodeling. Previous research found two functional Nurr1 binding sites in the proximal Topoisomerase IIβ (Top IIβ) promoter [76], since it was known that Topo IIβ deficiency affects axon growth through Rho-GTPase dysregulation [77]. Additionally, genome-wide analysis in human NSCs identified many Nurr1 direct genes involved in synapse formation [78]. However, little is yet known about how Nurr1 regulates synaptogenesis. Moreover, Nurr1 appears to be an essential TF for maintaining mdDA neurons' distinct traits by positively regulating many nuclear-encoded mitochondrial genes [79], manifested by the evidence that mitochondrial impairment modulated by MPP (+) was improved by Nurr1 agonists' treatment [80].
Nurr1 is required for DA phenotype genes (Th, Dat, and Vmat2) expression. Overexpression of Nurr1 was thought to induce the development of mature DA neurons during cell differentiation. ESCs and embryo-derived cells were the most common cells used in cell reprogramming due to their pluripotency. As expected, Nurr1 transduction promotes in vitro differentiation of mouse ESCs into mature DA neurons (Table 1) [81,82]. Coexpressing Nurr1 and GPX-1 (Glutathione peroxidase 1, a neuroprotective enzyme against oxidative stress) in mouse ESCs results in the differentiation of DA-like cells with increased survival ability [83]. Additionally, exogenous Nurr1 expression in embryonic corticoidderived NPCs and embryonic midbrain-derived NSCs achieves comparable results to the above ESCs [84,85]. In the presence of Nurr1-Mash1, thyroid hormone derivatives have been shown to stimulate differentiation of embryonic corticoid-derived NPCs into DA neurons [84]. Besides this, these derivatives have been displayed in vitro to protect DA neurons from neurotoxic damage caused by 6-hydroxydopamine (6-OHDA) and hydrogen peroxide (H 2 O 2 ). Similarly, Urocortin (UCN), a corticotropin-releasing hormone family peptide, has been shown to enhance Nurr1 + NPCs differentiation into Th + DA neurons in vitro and in vivo as Nurr1, Foxa2, and Pitx3 expression [85].
Previous studies have found that most transplanted NSCs fail to differentiate into neurons but glial cells and survive after transplantation [66], which might significantly contribute to a poor host cellular environment. In light of the role of Nurr1 in glial cells [40,72], co-culture of embryonic mesenchymal NSCs and primary microglial cells overexpressing Nurr1 prolonged the survival of transplanted NSCs, decreased the number of microglial cells and showed long-term survival [86,87]. Recent innovative research has found that cografting NNSC and NMG (NSCs and microglia both with Nurr1 overexpression) improved the behavior of PD rats. Furthermore, with a better understanding of MSCs' accessibility, multilineage potential, and non-tumorigenic potential, MSCs are thought to have a better potential for clinical applications without ethical constraints [88]. A recent study reveals that the grafts of MSCs overexpressing Nurr1 in the striatum of PD rats not only survive and migrate in the brain but also alleviate PD symptoms, increasing the number of Th + cells in the SNpc, and, most importantly, inhibiting glial cell activation [89]. In vivo Transplate these iPSCs-derived NPCs into PD models

Pitx3
Pitx3 is a TF required for the development and survival of mdDA neurons [110]. After birth, Pitx3 is expressed constitutively in the midbrain's SNpc and VTA. ESCs are also used to study Pitx3's role in PD cell therapy. Research shows that exogenous Pitx3 in ESCs-derived progenitor cells promotes the generation of DA neurons in vitro through regulating the expression of Th, Ngn2, and β-tubulin III genes [91]. High Th expression was also significantly correlated with increased Pitx3 expression [111]. In vivo, ESCs have been shown to have the potential for PD cell therapy [13]. Correspondingly, transplanting ESCs-derived progenitor cells overexpressing Pitx3 restores the functional deficits in PD rats. However, grafts of Pitx3-eGFP + or Pitx3-eGFP − cells sorted from these cells show a different result: cell grafts from Pitx3-eGFP − cells improve motor behavior deficits, but not Pitx3-eGFP + cells [92]. Furthermore, the size of the Pitx3-eGFP + cell grafts decreases [92], implying that these grafts do not survive and integrate into the host brain. The timing of Pitx3 expression could cause this during DA neuron development. The onset of Pitx3 expression coincides with the cell cycle exit and the entry into the post-mitotic state. Thus, a poor survival rate may be achieved following transplantation. On the other hand, incorporating GDNF and GFR-1 into embryonic brain-derived NSCs could improve the behavior of PD models. In a rat PD model, GDNF, and its receptor GFR1 signaling, activate Nurr1 and Pitx3 to increase the survival of transplanted midbrain-derived NSCs [112,113]. During embryogenesis, Pitx3 is required to activate BDNF expression in a rostro-lateral SNpc mdDA, and GDNF transient expression in the murine VM induces Pitx3 transcription via NF-B signaling. Meanwhile, overexpression of Pitx3 protects mdDA neurons by increasing GDNF and BDNF expression [114,115]. Thus, more research may be needed in the future to determine whether co-transplantation of non-neuronal cells overexpressing Pitx3 with NSCs promotes graft survival and integration.

Lmx1a/b
Lmx homeodomain TFs (Lmx1a/b) are required to develop mdDA neurons. Lmx1a is expressed first at E8.5 in the RP [116]. During mdDA neurogenesis, Lmx1a is expressed by the mdDA progenitors and maintains its expression in the post-mitotic mdDA neurons. Lack of Lmx1a/b results in impaired respiratory chain activity, increased oxidative stress, mitochondrial DNA damage, and axonal pathology. These disturbed molecular pathways eventually lead to synuclein accumulation and autophagy defects, as well as the loss of mdDA neurons [117]. Moreover, it has been known that Lmx1a/b are essential for mdDA neuron excitatory synaptic inputs and dendritic development [118]. As previously stated, Lmx1a is more important in developing mdDA, and Lmx1b acts as a compensatory factor for Lmx1a. It was reported that Lmx1a, Foxa2 and Pitx3 could increase Th expression when delivered to NPCs during neural proliferation, but only Lmx1a increases expression after induction, which is more efficacious [15]. When embryoid body cells (EBCs) transplanted into the adult intact SNpc, they differentiate into NPCs without acquiring DA phenotypes. In contrast, EBCs that overexpress Lmx1a develop DA neuronal markers. Notably, when EBCs are transplanted into 6-OHDA-lesioned SNpc, they develop into Th+ cells without the need for exogenous Lmx1a expression. However, these Th+ cells did not survive long [94], implying that additional factors are required to ensure long-term survival and terminal differentiation. Furthermore, Lmx1a-eGFP + cell grafts are more predictable and enriched in DA neurons when isolated from ESC-derived progenitor cells overexpressing Lmx1a and transplanted into PD rodents. This study demonstrated that the appropriate integration might accelerate functional recovery. Additionally, mature Th + DA neurons from Lmx1a-eGFP + cell grafts are two-fold higher than unsorted grafts [92], indicating that pure grafts are more predictable and safer, which may be a future trend in PD cell therapy.

En1
En1 expression starts in the mid and hindbrain at E8 [119]. En1 protein-coding genes are linked to NSCs and lineage-specific markers [24,120]. Paradoxically, in the absence of En1, the ectopic expression of mdDA neuron markers was diminished in the metencephalon, suggesting that En1 influences the relocation of MHB [121]. It is proposed that En1 is significant for the correct establishment of IsO by controlling the proper expression of Fgf8, Otx2, and Wnt1 [122]. Overall, En1 is essential for the appropriate maintenance and function of IsO [20,122]. In addition, En1 induces epigenetic modifications in the nucleus, involves the guidance of retinal ganglion cell axons and maintains synapse integrity by influencing mitochondrial function [123]. Similarly, En1 can also protect mdDA neurons from oxidative stress and preserve axonal integrity [124,125]. The combination of En1 and other TFs has been certified to reprogram cells into DA neurons successfully [17,108] ( Table 1).

Foxa1/2
Despite being required for Lmx1a/b, Nurr1, En1, Aadc, and Th expression, Foxa1/2 are critical TFs in the early development, specification, and maturation of mdDA neurons [42,126]. Foxa1/2 are also involved in neuron projection development and axon guidance [43]. Aside from that, Foxa2 is necessary for Foxa1 expression at E8.5, and Foxa1 has been shown to induce the differentiation of pluripotent P19 cells into neural stem-like cells [127]. Recent studies found that Foxa1/2 deletions resulted in down-regulation in Th and DA synthesis, as well as burst-firing activity in SNpc mdDA neurons [128,129]. As previously stated, Foxa1/2 plays overlapping roles in the specification and development of mdDA neurons. Furthermore, Foxa1/2 also regulates Shh expression. While Foxa2 deficiency causes a transient decrease of Shh, Foxa1 deficit does not. Foxa1 and Foxa2 both positively and negatively control Shh signaling to determine the identity of ventral midbrain progenitors [130]. Therefore, more research now focuses on the generation of DA neurons mediated by Foxa2 rather than Foxa1. As stated below and in Table 1, Foxa2, like other TFs, has been identified as a potentially excellent forward-looking DA inducing neuron when combined with other TFs.

Combinations of TFs
As noted previously, transplanted NSCs primarily differentiated into glial cells rather than DA neurons in vivo, resulting in a poor outcome. As a result, it is necessary to develop a method to protect DA neurons from neuroinflammation and promote the survival and differentiation of transplanted NSCs into DA neurons [87]. Forced expression of ALN (Ascl1, Nurr1, and Lmx1a) in glial cells resulted in the generation of DA neurons with DA release and spontaneous pace-making activity [102,104]. In contrast, neurons derived from astrocytes or NG2 glial do not express Th in vivo [131,132]. Transplantation of human astrocytes reprogrammed by NeAL218 (Neurod1, Ascl1, Lmx1a, and microRNA miR218) can generate induced DA neurons (iDANs) directly in vivo, which are capable of expressing typical DA neuron markers, adopting mature neuronal morphology and functionality, and rescuing some of the motor behavior in PD animal models. Unlike the above two types of research, these iDANs do not derive from or acquire a striatal neuron phenotype [16]. Even so, reprogramming efficiency alone is insufficient, and the next step is to improve its efficiency and safety in humans. A new strategy for co-grafting NPCs and midbrain-derived astrocytes with Nurr1 and Foxa2 overexpression has recently been revealed, boosting graft maturation and survival and resulting in an improvement in the therapeutic impact on NPCs cell transplantation. [97]. Astrocytes are transformed from pro-inflammatory, immunogenic astrocytes to regeneration-friendly astrocytes when activated by regionspecific TFs [133]. Only these neurotrophic, anti-inflammatory astrocytes promote the maturation and survival of VM-NPC transplants.
Although ESCs or embryo-derived cells are being studied for PD cell therapy, many technical and ethical issues remain. Induced pluripotent stem cells (iPSCs), derived from adult somatic tissues and differentiated into specific cells similar to ESCs, remove ethical barriers and reduce the likelihood of immune rejection. A growing body of research has explored how to make self-renewing NPCs from fibroblasts instead [95,101,134]. In vitro, mouse embryonic and postnatal fibroblasts can be transdifferentiated into induced neural progenitor cells (iNPCs) using the Wernig factors Ascl1, Brn2, and Myt1 [95,101], but their regional identity is lost, and iNPCs preferentially differentiate into astrocytes. Likewise, mature DA neurons were generated from iNPCs forced expression of Nurr1 and Foxa2, but the neuronal maturity of engrafted neurons was different between in vivo and in vitro when the iNPCs were transplanted into the striatum of PD rats [95], which may result from poor graft quality, immune response or poor host microenvironment. Recent studies found that autologous iPSCs derived from human skin fibroblasts are obtained through a new reprogramming method combining TFs (Oct4, Sox2, Klf4, and c-Myc) and miRNAs (miR-302s, and miR-200c) [109,135]. The transplantation of differentiated and purified iPSCsderived NPCs into PD animal models can restore motor symptoms and form functional connections with other neurons without tumorigenicity or toxicity [109,136,137], implying that purified graft cells are safer and more predictable. More notably, cells have been transplanted into PD patients in a clinical trial [138] and a case report indicates that a patient with PD may benefit from these autologous NPCs [135]. Autologous transplantation can avoid host immune rejection. Nevertheless, graft quality, including mutations and contamination, is crucial because it is linked to the safety of transplantation trials [139]. Additionally, the source of the iPSCs is a concern: it is unknown if autologous iPSC-derived NPCs from PD patients can develop defective DA neurons. Even when iPSC-derived DA neurons from genetically related healthy donors are transplanted, the host brain's inadequate microenvironment may affect healthy grafts [13]. The results of ongoing clinical trials will provide us with a clear idea regarding keeping this strategy safe and efficient for a long time.
Moreover, some studies directly converse fibroblasts into DA neurons by different TF combinations [100,107,108]. The combination of six TFs (Ascl1/Nurr1/Lmx1a/Pitx3/En1/ Foxa2) could induce the expression of DA neuronal marker genes more efficiently than any other combination. Importantly, when combined with Shh and Fgf8, these six factors can induce mature DA neurons in vitro [108]. In addition, micro-34b/c has been shown to promote cell cycle exit by regulating Wnt1 and enhancing mesencephalic DA neuron differentiation when combined with Ascl1 and Nurr1 [100]. Therefore, small chemical molecules also influence neuron differentiation as well as TFs [36]. However, the grafts subsequently integrate and survive poorly in host tissues after transplantation and further research is needed to improve these issues. Multiple methods using TFs combinations to induce DA neurons are shown in Table 1.
Cell reprogramming provides an ideal strategy for generating DA neurons from nonneural cells. However, induced DA cells by simple overexpression in the developing midbrain are prone to immature neurons and short maintenance for their phenotype after transplantation [140,141]. It is important to note that tumors may form when grafting undifferentiated cells or immature neural progenitors into the therapy. Future research may focus on using both forced expression of compound TFs and co-grafting different cell types simultaneously to promote DA neuron generation at specific regions and times and to help the graft mature and survive. iPSCs still have a promising future in PD cell therapy. Moreover, many unanswered questions still need to be further addressed [95,142]: to what extent, if any, do the various TFs described above interact? what are their downstream targets and upstream activators? to obtain the best results, which sources of cells need to be utilized (xeno or allotransplantation; cell types; pure DA neurons or other cells)? and how to prevent cancer formation?

Conclusions
The evolution of the DA system has been a topic of extensive research in recent years. Many molecules have been demonstrated to facilitate mdDA neuron development. Studies of TFs that play critical roles in various stages have greatly improved our understanding of how these neurons are generated and paved the way for developing new strategies for transplantation therapy. In vitro and in vivo, using isolated or mixed TFs has successfully induced the generation of DA neurons, which manipulate cell fate for a specific cell type. The grafts from neural cells induced by TFs can survive and play a DA neuron's role in PD model animals, such as DA release and motor behavior improvement. However, many critical issues need to be further explored regarding DA neurons' differentiation, survival, and maturation after iNPCs transplantation. The clinical trials are the final manifestation to determine the efficacy of cell transplant therapy. We believe that in-depth knowledge of the critical regulatory proteins and hierarchical networks involved in mammalian mdDA neuron differentiation will benefit future clinical applications and regenerative medicine.