Therapeutic Potential of PTBP1 Inhibition, If Any, Is Not Attributed to Glia-to-Neuron Conversion

Abstract

A holy grail of regenerative medicine is to replenish the cells that are lost due to disease. The adult mammalian central nervous system (CNS) has, however, largely lost such a regenerative ability. An emerging strategy for the generation of new neurons is through glia-to-neuron (GtN) conversion in vivo, mainly accomplished by the regulation of fate-determining factors. When inhibited, PTBP1, a factor involved in RNA biology, was reported to induce rapid and efficient GtN conversion in multiple regions of the adult CNS. Remarkably, PTBP1 inhibition was also claimed to greatly improve behaviors of mice with neurological diseases or aging. These phenomenal claims, if confirmed, would constitute a significant advancement in regenerative medicine. Unfortunately, neither GtN conversion nor therapeutic potential via PTBP1 inhibition was validated by the results of multiple subsequent replication studies with stringent methods. Here we review these controversial studies and conclude with recommendations for examining GtN conversion in vivo and future investigations of PTBP1.

1. INTRODUCTION

Since neuron loss is frequently associated with neural injuries or neurodegenerative diseases, in vivo reprogramming of somatic glial cells to neurons, also referred to as glia-to-neuron (GtN) conversion, is emerging as a potential regeneration-based therapeutic strategy (Barker et al. 2018, Chen et al. 2015, Heinrich et al. 2015, Tai et al. 2020, Wang & Zhang 2018). Such a strategy was initially demonstrated by the results of several studies (Grande et al. 2013, Guo et al. 2014, Heinrich et al. 2014, Niu et al. 2013, Torper et al. 2013); however, a few recent publications in top scientific journals have drawn more attention. It was reported that inhibiting a single protein [polypyrimidine tract-binding protein 1 (PTBP1)] rapidly induces fate switches of resident glial cells, which then become precisely connected mature neurons, and alleviates symptoms of neurological diseases and aging-associated cognitive impairments (Maimon et al. 2021, Qian et al. 2020, Zhou et al. 2020). If true, this would be revolutionary in regenerative medicine for treating numerous human neurodegenerative or neurological symptoms. Such phenomenal claims immediately called for confirmatory studies using more stringent methods; however, their results are rather disappointing as they were unable to observe GtN conversion or functional improvements via PTBP1 inhibition. Below, we review these controversial studies.

2. A CRITICAL ROLE OF PTBP1 IN NEURAL DEVELOPMENT

PTBP1 (also known as PTB or hnRNP I) is a member of the heterogeneous nuclear ribonucleoproteins expressed in various tissues and cells, including embryonic stem cells (Lillevali et al. 2001). It regulates many aspects of the messenger RNA (mRNA) life cycle such as splicing, 3′-end processing, localization, stability, and translation (reviewed by Black 2003, Busch & Hertel 2012, Corrionero & Valcarcel 2009, Hu et al. 2018). During neural development, PTBP1 is highly enriched in the nuclei of neural stem cells (NSCs) (Shibasaki et al. 2013) and its downregulation promotes neuron-specific mRNA splicing (Ashiya & Grabowski 1997, Lillevali et al. 2001, Markovtsov et al. 2000). Such downregulation is mediated by the abundantly expressed neuron-specific microRNA miR124 (Makeyev et al. 2007). While its constitutive knockout leads to embryonic lethality (Shibayama et al. 2009), conditional ablation of PTBP1 in embryonic NSCs causes hydrocephalus due to precocious differentiation of NSCs and subsequent dysfunctional ependymal cells (Shibasaki et al. 2013). These results clearly show a critical role of PTBP1 in maintaining NSCs and their normal differentiation during neural development.

3. PTBP1 INHIBITION: A REVOLUTION IN REGENERATIVE MEDICINE?

Despite it being discovered 30 years ago (Patton et al. 1991), an unexpected function of PTBP1 inhibition, showing efficient conversion of somatic glial cells to neurons in multiple regions of the adult mouse central nervous system (CNS), was only recently reported (Maimon et al. 2021, Qian et al. 2020, Yang et al. 2023, Zhou et al. 2020). While other studies showed gradual generation of neurons through GtN conversion in vivo (Gascón et al. 2016; Grande et al. 2013; Heinrich et al. 2014; Lentini et al. 2021, Niu et al. 2013, 2015; Su et al. 2014; Tai et al. 2020; Wang et al. 2016; Zhang et al. 2022), PTBP1 inhibition was reported to rapidly and efficiently convert glial cells to diverse mature neurons in multiple regions of the adult CNS. These neurons include dopaminergic (DA) neurons, retinal ganglion cells (RGCs), striatal neurons, cortical neurons, hippocampal neurons, or motor neurons (Maimon et al. 2021, Qian et al. 2020, Yang et al. 2023, Zhou et al. 2020). It is very perplexing, though, that PTBP1 inhibition through short hairpin RNA (shRNA)-, antisense oligonucleotide (ASO)-, or CRISPR-CasRx-mediated methods gives rise to different subtypes of neurons in the adult striatum: striatal neurons (Qian et al. 2020) or DA neurons (Zhou et al. 2020). Notwithstanding such discrepancies, all of the above-mentioned PTBP1 studies reported significant functional relevance: reversing the symptom of a Parkinson’s disease (PD) model, rescuing vision loss after retinal injury, improving cognition in aged mice, or promoting motor function after spinal cord injury (Maimon et al. 2021, Qian et al. 2020, Yang et al. 2023, Zhou et al. 2020). Below, we summarize these phenomenal discoveries associated with PTBP1 inhibition and raise concerns regarding the methods for detecting GtN conversion. Results of these studies are also summarized in Table 1.

Table 1.

Studies concluding GtN conversion via PTBP1 inhibition

Research group (reference)	Neural region	Targeting method	Detection method	Tracing glial origin	Disease model	Behavior benefit?
X.D. Fu (Qian et al. 2020)	Cortex	AAV2/2-LSL-shRNA	Cre-dependent viral reporter	mGfap-Cre (without a reporter)a	ND	ND
	Striatum
	SNc				PD (6-OHDA)	Yes
		ASO	Lineage reporter	hGFAP-CreERT2; R26R-tdTomato
H. Yang (Zhou et al. 2020)	Striatum	AAV2/PHP.eB-CRISPR-CasRx	Viral reporter	ND	PD (6-OHDA)
	Retina		Viral Cre-induced reporter		NMDA injury
D.W. Cleveland (Maimon et al. 2021)	Cortex	ASO	Lineage reporter	hGFAP-CreERT2; R26R-tdTomato	ND	ND
	Hippocampus				1.5-year-old mice	Yes
G. Chen (Yang et al. 2023)	Spinal cord	AAV-shRNA	Viral reporter	ND	Compression injury
		ASO	Neuronal density

^aWhen bred to the R26R-YFP reporter line, the genetic lineage reporter was detected in endogenous neurons of multiple brain regions such as the cortex and the SNc (Wang et al. 2021).

Abbreviations: 6-OHDA, 6-hydroxydopamine; AAV, adeno-associated virus; ASO, antisense oligonucleotide; GtN, glia-to-neuron; LSL, lox-STOP-lox; ND, not determined; NMDA, N-methyl-d-aspartate; PD, Parkinson’s disease; shRNA, short hairpin RNA; SNc, substantia nigra pars compacta.

3.1. Dopaminergic Neurons from Nigral Astrocytes and Rescue of a Parkinson’s Disease Model by shRNA?

By employing an adeno-associated virus (AAV) vector (serotype 2) and the Cre-dependent expression of a miR30-based shRNA (AAV-LSL-RFP-shPtbp1) to downregulate PTBP1, Qian et al. (2020) reported astrocyte-converted DA neurons in the adult substantia nigra pars compacta (SNc). These neurons were identified by the coexpressed fluorescent reporter RFP in transgenic mGfap-Cre mice. While only 1% of RFP⁺ cells were neurons at the beginning, this number increased to 20% and 80% when examined at 5 weeks and 10 weeks after virus injection, respectively. Such observations led Qian et al. to conclude that PTBP1 inhibition time-dependently converts midbrain astrocytes to mature neurons.

When examined at 12 weeks, 30–35% of the RFP⁺ cells in the SNc, but not the adjacent area, expressed markers of DA neurons (TH, DDC, DAT, VMAT2, EN1, LMX1A, and PITX3), indicating region-specific astrocyte-to-DA neuron conversion within the dopamine domain (Qian et al. 2020). These RFP⁺ DA neurons exhibit characteristic morphology and electrophysiology identical to endogenous DA neurons (Qian et al. 2020). Strikingly, these neurons provide axons to reconstruct the nigrostriatal circuit, with restoration of dopamine levels and rescue of motor deficits in a 6-hydroxydopamine (6-OHDA)-induced PD model (Qian et al. 2020). Notwithstanding these phenomenal findings, cell type specificity of the viral RFP reporter in mGfap-Cre mice was not examined by Qian et al. A recent report showed leaky neuronal expression of a reporter in multiple regions of the brain, including the SNc, in this transgenic mouse line (Wang et al. 2021), casting doubt on the astrocyte origin of those RFP⁺ DA neurons reported by Qian et al.

3.2. Dopaminergic Neurons from Striatal Astrocytes and Rescue of a Parkinson’s Disease Model by CRISPR-CasRx?

Zhou et al. (2020) employed the CRISPR-CasRx system to knockdown endogenous PTBP1 by using the PHP.eB-serotyped AAV-GFAP-CasRx-Ptbp1 virus. Virus-transduced cells were identified by the coinjected AAV-GFAP-mCherry virus such that cell type specificity was conferred by the human glial fibrillary acidic protein (GFAP) promoter. In contrast to striatal neurons reported by Qian et al. (2020) in the striatum, surprisingly, Zhou et al. (2020) concluded that PTBP1 inhibition converts striatal astrocytes into DA neurons. These neurons were claimed to exhibit molecular markers and electrophysiological features stereotypical to DA neurons. Importantly Zhou et al. also reported that PTBP1 inhibition by CRISPR-CasRx results in marked alleviation of motor dysfunction of the 6-OHDA-induced PD model.

How could different subtypes of neurons be generated from striatal astrocytes by knocking down the same PTBP1 protein? This question awaits answers from Zhou et al. and Qian et al. Nonetheless, the image quality is too poor to discern the true identity of those claimed striatal DA neurons in the publication by Zhou et al. Recent follow-up studies by other groups showed that the AAV-GFAP-CasRx-Ptbp1 virus is inefficient in knocking down endogenous PTBP1 (Wang et al. 2021, Xie et al. 2022), casting serious doubts on the key conclusions by Zhou et al. (2020).

3.3. Cortical Neurons from Astrocytes by PTBP1 Inhibition?

Upon shRNA-mediated PTBP1 inhibition, Qian et al. (2020) reported the presence of astrocyte-converted RFP⁺ neurons in the cortex with a similar efficiency as in the midbrain or the striatum. Subsequently, Maimon et al. (2021) reported the generation of new neurons from cortical astrocytes after intracerebroventricular injection of an antisense oligonucleotide against Ptbp1 (PTB-ASO). They detected transient reduction of Ptpb1 mRNA and observed that 14% of the genetically labeled cells are neurons in the cortex of tamoxifen-induced hGFAP-CreER^T2;R26R-tdTomato mice (Maimon et al. 2021). Nonetheless, cell type specificity of genetically labeled cells was not rigorously examined for this mouse line, which is concerning since different lineage-tracing lines may exhibit variable and region-dependent neuronal leakage (Wang & Zhang 2022a, Wang et al. 2021).

3.4. Glia-Converted Hippocampal Neurons and Cognitive Improvement by Antisense Oligonucleotides?

Through ASO-mediated PTBP1 inhibition, Maimon et al. (2021) reported generation of new hippocampal neurons from astrocytes. The conclusion was based on lineage tracing in tamoxifen-inducible hGFAP-CreER^T2;R26R-tdTomato mice. Electrophysiology indicated integration of newly generated neurons into neural circuits receiving both glutamate and GABA inputs and firing action potentials that were indistinguishable from neighboring endogenous neurons. PTB-ASO also induced new neurons in 14-month-old mice. Excitingly, PTBP1 inhibition improved cognitive function of 1.5-year-old mice in an object-recognition test. Of note, despite the reported robust GtN conversion, the total number of GFAP-expressing astrocytes was not altered. It is also intriguing that the reported new neurons were all precisely positioned in the hippocampus despite a lack of regional and cell type specificity of the intracerebrally injected PTB-ASO.

3.5. Retinal Ganglion Cells from Muller Glia and Vision Restoration After Injury via PTBP1 Inhibition?

Through the CRISPR-CasRx system, Zhou et al. (2020) reported rapid and precise conversion of retinal Müller glia to RGCs. In this experiment, cells were traced with tdTomato after coinjection of AAV-hGFAP-CasRx-Ptbp1 and AAV-hGFAP-GFP-Cre into Ai9 (Rosa-CAG-LSL-tdTomato-WPRE) mice. AAVs were serotyped with PHP.eB. One month later, tdTomato-positive neurons were observed in the ganglion cell layer (GCL) and represented three subtypes of RGCs (Foxp2, Brn3c, or Parvalbumin) as well as amacrine cells. Strikingly, axons of reporter-positive RGCs precisely projected to the dorsal lateral geniculate nucleus and superior colliculus in the brain. This process was rapid, appearing at 1.5 weeks and being greatly enhanced by 4 weeks after virus injection. Even more strikingly, vision-dependent behavior, measured by visually evoked potentials and light/dark preference test, was also restored for mice with N-methyl-d-aspartate (NMDA)-induced retinal injury one month after virus injection. Despite these phenomenal discoveries, no convincing evidence was provided to show downregulation of PTBP1 in vivo, let alone Müller glia conversion to RGCs. Migrating intermediates and growth cones of axons were not reported. Furthermore, the PHP.eB serotype preferentially targets neurons rather than glial cells (Chan et al. 2017, Mathiesen et al. 2020), and AAV-hGFAP-Cre is highly leaky in endogenous neurons (Wang et al. 2021).

3.6. Motor neurons from Reactive Astrocytes and Motor Function Improvement After Spinal Cord Injury?

By using shRNA-mediated downregulation of PTBP1, Yang et al. (2023) reported that astrocytes were directly converted into motor neuron-like cells around the injured spinal area after compression injury. The GFP reporter in the pAAV-GFAP-EGFP-MIR155 vector was used to trace the virus-infected cells. About 30% of GFP⁺ cells coexpressed the neuronal marker NeuN and MAP2 in the injured spinal cord 11 weeks after virus infection. Interestingly, approximately 19% of the GFP⁺ cells were also ChAT⁺ motor neurons. These cells were in the ventral grey matter in which the endogenous spinal motor neurons normally reside. Yang et al. (2023) also reported significant behavioral recovery after shRNA- or ASO-mediated PTBP1 downregulation. Notwithstanding, direct evidence was not provided to unambiguously demonstrate that those viral GFP⁺ neurons were indeed converted from resident astrocytes. Stringent analyses of GtN conversion are needed for future independent replication studies.

4. EVIDENCE CHALLENGING GLIA-TO-NEURON CONVERSION BY PTBP1 INHIBITION

A regenerative role of PTBP1 inhibition immediately attracted the attention of the research community. Unfortunately, the follow-up studies with more stringent analyses rather show that PTBP1 inhibition does not induce GtN conversion. Below, we review these replication studies and focus on their stringent methods and controversial conclusions (Chen et al. 2022; Guo et al. 2022a, 2022b; Hoang et al. 2021, 2022; Leib et al. 2022; Wang et al. 2021; Xie et al. 2022). Results of these replication studies are also summarized in Table 2.

Table 2.

Studies concluding a lack of GtN conversion via PTBP1 inhibition

Research group (reference)	Neural region	Targeting method	Detection method	Tracing glial origin	Disease model	Behavior benefit?
C.L. Zhang (Wang et al. 2021)	Striatum	AAV2/5-shRNA [same sequence as in Qian et al. (2020)]	Lineage reporter and viral reporter	Aldh1l1-CreERT2;R26R-YFP and mGfap-Cre;R26R-YFP	ND	ND
		AAV2/5-LSL-shRNA [same sequence as in Qian et al. (2020)]	Lineage reporter and Cre-dependent viral reporter
		AAV2/2-LSL-shRNA [vector from Qian et al. (2020)]		mGfap-Cre;R26R-YFP
		AAV2/5-LSL-shRNA [vector from Qian et al. (2020)]
		AAV2/2-LSL-shRNA [virus from Qian et al. (2020)]
		AAV2/PHP.eB-CRISPR-CasRx [vector from Zhou et al. (2020)]	Lineage reporter and viral reporter
S. Blackshaw (Hoang et al. 2021)	Striatum	Astrocyte-specific gene deletion	Lineage reporter	Aldh1l1-CreERT2;Sun1-GFPlox/lox;Ptbp1lox/lox
	Cortex
	SNc
	Retina	Müller glia–specific gene deletion		Glast-CreERT;Sun1-GFPlox/lox;Ptbp1lox/lox	NMDA injury	No
B.L. Davidson (Leib et al. 2022)	Striatum	AAV2/1-miRNA [same sequence as in Xue et al. (2013)]	Lineage reporter	Aldh1l1-CreERT2;R26R-tdTomato	ND	ND
	Cortex
	Hippo-campus
M. Li (Chen et al. 2022)	Striatum	AAV2/5-shRNA [same sequence as in Qian et al. (2020)]	Lineage reporter and viral reporter	Aldh1l1-CreERT2;Rpl22lsl-HA	PD (6-OHDA)	No
	SNc
		ASO [same sequence as in Maimon et al. (2021)]	Lineage reporter
B. Chen (Xie et al. 2022)	Retina	AAV2/ShH10-LSL-shRNA [vector from Qian et al. (2020)]	Lineage reporter and Cre-dependent viral reporter	Glast-CreERT;Sun1-GFPlox/lox	ND	ND
		AAV2/PHP.eB-CRISPR-CasRx [vector from Zhou et al. (2020)]	Lineage reporter and viral reporter
Y. Zhao (Guo et al. 2022a)	Striatum	AAV2/9-shRNA [same sequence as in Qian et al. (2020)]	Viral reporter	ND
	SNc
	Hippocampus				AD (5×FAD)	No
					AD (PS19)
Y. Zhao (Guo et al. 2022b)	Hippocampus	ASO [same sequence as in Qian et al. (2020)]	Neuronal density		ND	ND

Abbreviations: 5×FAD, transgenic mouse harboring five familiar Alzheimer’s disease–linked mutations; 6-OHDA, 6-hydroxydopamine; AAV, adeno-associated virus; AD, Alzheimer’s disease; ASO, antisense oligonucleotide; GtN, glia-to-neuron; miRNA, microRNA; LSL, lox-STOP-lox; ND, not determined; NMDA, N-methyl-d-aspartate; PD, Parkinson’s disease; PS19, transgenic mouse harboring P301S tau mutation; shRNA, short hairpin RNA; SNc, substantia nigra pars compacta.

4.1. PTBP1 Inhibition Fails to Convert Striatal Astrocytes to Neurons

Astrocytes are ubiquitously distributed and represent a large fraction of the neural cells in the adult mammalian brain. Unlike neurons, which frequently succumb to injuries or degenerative conditions, astrocytes become reactive and can replenish themselves. They are an ideal cell source for regenerative medicine if their fates can be reengineered in vivo (Smith et al. 2016, Tai et al. 2020, Wang & Zhang 2018). It is extremely intriguing that PTBP1 inhibition alone could rapidly and very efficiently convert striatal astrocytes into either mature striatal neurons or functional DA neurons (Qian et al. 2020, Zhou et al. 2020). However, such a phenomenal claim regarding the effect of PTBP1 inhibition on striatal astrocytes could not be independently validated (Guo et al. 2022a, Leib et al. 2022, Wang et al. 2021).

4.1.1. No glia-to-neuron conversion by CRISPR-CasRx.

The first replication study was initiated by Wang et al. (2021) because of their long-term interests in GtN conversion in vivo (Islam et al. 2015; Niu et al. 2013, 2015, 2018; Su et al. 2014; Tai et al. 2021; Wang et al. 2016; Zhang et al. 2022). They used the PHP.eB-serotyped AAV-CRISPR-CasRx-Ptbp1 virus that was constructed by Zhou et al. (2020) and traced the virus-transduced cells with the coinjected AAV2/5-hGFAP-mCherry, as described by Zhou et al. However, a distinction was that Wang et al. also genetically traced the fates of striatal astrocytes in mGfap-Cre;R26R-YFP mice. Because the Cre-dependent YFP reporter is controlled by the ubiquitously active ROSA locus, it will be continuously expressed even after GtN conversion. Despite the fact that approximately 22% of viral mCherry-positive cells were neurons in the CasRx-Ptbp1 group, none of them were traced with YFP, indicating a nonastrocyte origin (Wang et al. 2021). Immunohistochemistry also failed to confirm an effective inhibition of PTBP1 by CasRx-Ptbp1 virus in vivo. Although Zhou et al. provided an image on PTBP1 expression with cytoplasmic localization, it seemed to be nonspecific since PTBP1 is mainly a nuclear protein (Qian et al. 2020). The results of Wang et al. clearly indicate that CasRx-Ptbp1 virus neither efficiently downregulates PTBP1 nor converts striatal astrocytes in vivo (Wang et al. 2021).

4.1.2. No glia-to-neuron conversion by shRNA.

After failing to observe GtN conversion by CRISPR-CasRx, Wang et al. (2021) examined shRNA-mediated PTBP1 inhibition in four sets of experiments. First, they used serotype 5-packaged AAV-hGFAP-mCherry-shPtbp1, which had the identical Ptbp1 shRNA sequence as described by Qian et al. (2020). Additionally, fates of striatal astrocytes were genetically traced in tamoxifen-inducible Aldh1l1-CreER^T2;R26R-YFP mice, in which approximately 95% of striatal astrocytes could be permanently labeled by YFP (Wang et al. 2021). Despite a dramatic inhibition of PTBP1 expression in astrocytes, none of the genetic lineage-traced astrocytes became neurons. To rule out a potential effect of genetic background or tamoxifen treatments, they also traced striatal astrocytes in mGfap-Cre;R26R-YFP mice. Once again, none of the lineage-traced astrocytes were converted into neurons by PTBP1 inhibition. Second, Wang et al. employed the Cre-dependent AAV-CAG-LSL-mCherry-shPtbp1 virus. In addition to the virus-expressed mCherry reporter, they also lineage traced striatal astrocytes in Aldh1l1-CreER^T2;R26R-YFP or mGfap-Cre;R26R-YFP mice. Despite using these two different mouse lines, they failed to detect GtN conversion by PTBP1 inhibition (Wang et al. 2021). Third, Wang et al. obtained the AAV-CMV-LSL-RFP-shPtbp1 vector directly from Qian et al. and packaged it with serotype 5 or 2. Once again, they failed to detect any GtN conversion from lineage-traced striatal astrocytes. And finally, Wang et al. examined the AAV-CMV-LSL-RFP-shPtbp1 virus that was directly provided by Qian et al. Despite the identical virus being used, Wang et al. failed to observe the conversion of genetically traced astrocytes to neurons.

Supporting the negative results of Wang et al., recent studies by Leib et al. (2022) and Guo et al. (2022a) similarly failed to detect GtN conversion by shRNA-mediated PTBP1 inhibition in the striatum, hippocampus, or cortex. Together, these sets of experiments unambiguously argue against a role of PTBP1 inhibition in GtN conversion in vivo.

4.1.3. No glia-to-neuron conversion by conditional knockout of PTBP1.

To avoid virus-associated nonspecific labeling, Hoang et al. (2021) conditionally deleted Ptbp1 in astrocytes in Aldh1l1-CreER^T2;Sun1-GFP^lox/lox;Ptbp1^lox/lox mice. In these mice, astrocytes and their derivatives were also permanently traced with GFP in a tamoxifen-inducible manner. Despite PTBP1 deletion in astrocytes when examined at 2 or 8 weeks following tamoxifen treatments, GFP-labeled neurons were not detected in any of the brain regions, including the cortex, striatum, and substantia nigra. The results of these genetic studies support the findings by Wang et al. (2021) and clearly show that PTBP1 inhibition does not induce GtN conversion in vivo.

4.2. PTBP1 Inhibition Fails to Induce Dopaminergic Neurons in Vivo

Because of the enormous therapeutic potential of new DA neurons for PD, it would indeed be a revolutionary discovery to find that PTBP1 inhibition alone could rapidly and efficiently regenerate functional DA neurons from astrocytes in vivo (Qian et al. 2020, Zhou et al. 2020). However, subsequent studies failed to replicate the above findings (Chen et al. 2022, Guo et al. 2022a, Hoang et al. 2021).

4.2.1. PTBP1 inhibition fails to convert quiescent astrocytes to dopaminergic neurons.

As described above and contradictory to the findings of Zhou et al. (2020), CRISPR-CasRx failed to initiate GtN conversion let alone the generation of DA neurons in the adult mouse striatum (Wang et al. 2021). To reexamine the role of PTBP1 inhibition in the SNc, Chen et al. (2022) employed AAV-shPtbp1 under the full-length (2.2 kb) hGFAP promoter. Efficient PTBP1 inhibition was confirmed in the viral reporter-positive cells in the SNc when examined at 1, 2, or 3 months after virus injection. In addition to the viral reporter, Chen et al. genetically traced astrocytes and their derivatives with tamoxifen-inducible HA-Rpl22 in Aldh1l1-CreER^T2;Rpl22^lsl-HA mice. Although viral reporter-positive DA neurons were observed in the SNc, none of them were labeled with HA, indicating a nonastrocyte origin (Chen et al. 2022). Using the shRNA-based knockdown approach, Guo et al. (2022a) reached the identical conclusion that PTBP1 inhibition is unable to convert astrocytes into neurons in the SNc. Results that further supported these conclusions came from genetic deletion of PTBP1 in the SNc (Hoang et al. 2021). Together, this series of studies rigorously shows that PTBP1 inhibition is insufficient to convert nigral astrocytes to DA neurons.

4.2.2. PTBP1 inhibition fails to convert reactive astrocytes to dopaminergic neurons.

Though unlikely, it is possible that only reactive astrocytes could be converted to DA neurons by PTBP1 inhibition. Such a possibility was examined through shRNA-mediated PTBP1 inhibition in the SNc of adult Aldh1l1-CreER^T2;Rpl22^lsl-HA mice after a 6-OHDA lesion (Chen et al. 2022). The 6-OHDA lesion was confirmed by the loss of tyrosine hydroxylase (TH)⁺ neurons and the presence of reactive astrocytes in the SNc. Despite efficient downregulation of PTBP1 expression by shRNA AAVs, neurons from lineage-traced astrocytes were not observed even under such a pathological condition.

To rule out a potential toxic effect or leaky neuronal expression of AAVs (Le et al. 2022, Wang & Zhang 2022b, Wang et al. 2021), Chen et al. (2022) further employed an ASO-based approach for PTBP1 inhibition in vivo. Efficient PTBP1 inhibition was confirmed by immunostaining 2 months after ASO-Ptbp1 delivery. However, none of the genetically traced astrocytes were converted into neurons, let alone DA neurons, in the SNc of adult Aldh1l1-CreER^T2;Rpl22^lsl-HA mice with or without 6-OHDA injury.

4.3. PTBP1 Inhibition Fails to Induce Glia-to-Neuron Conversion in the Hippocampus

The hippocampus is critically involved in learning and memory. Loss of hippocampal neurons is associated with the aging process and is more severely affected by neurodegenerative conditions such as Alzheimer’s disease (AD). If PTBP1 inhibition induced GtN conversion and improved hippocampal function, as reported (Maimon et al. 2021), it would be a milestone advancement for treating age-related cognitive decline or AD. Unfortunately, replication studies failed to see GtN conversion through shRNA- or ASO-mediated PTBP1 inhibition in the hippocampus of wild-type mice or two models of AD mice (Guo et al. 2022a, b; Leib et al. 2022).

4.3.1. PTBP1 inhibition via shRNA fails to induce glia-to-neuron conversion in wild-type or Alzheimer’s disease mice.

For PTBP1 inhibition, Guo et al. (2022a) used AAV-hGFAP-mCherry-shPtbp1, which employs a longer form of the hGFAP promoter to drive expression of the mCherry reporter and a Pthp1 shRNA with an identical sequence to what was reported by Qian et al. (2020). Despite a nearly complete depletion of PTBP1 (~90%) in hippocampal astrocytes, no difference was observed between the shPtbp1 group and the scramble control group in terms of the number of NeuN⁺mCherry⁺ mature neurons or DCX⁺mCherry⁺ immature neurons, indicating a failure of GtN conversion in the hippocampus of adult wild-type mice (Guo et al. 2022a). To examine whether a pathological condition could promote GtN conversion, Guo et al. used two well-established AD mouse models: 5xFAD mice with progressive amyloid-β pathology and PS19 mice with tauopathies. Once again, both NeuN⁺mCherry⁺ neurons and DCX⁺mCherry⁺ immature neurons in the shPtbp1 group did not differ from the scramble control, suggesting a lack of GtN conversion even in a microenvironment with abundant reactive astrocytes (Guo et al. 2022a). Consistent with these findings, another study also showed that shPtbp1 failed to convert lineage-traced astrocytes into neurons in the hippocampus of adult Aldh1l1-CreER^T2;R26R-tdTomato mice (Leib et al. 2022).

4.3.2. PTBP1 inhibition via antisense oligonucleotides fails to induce adult hippocampal neurogenesis.

By using the identical nucleotide sequence as reported by Qian et al. (2020), Guo et al. (2022b) also examined adult hippocampal neurogenesis after ASO-mediated PTBP1 inhibition. Cell type specificity and inhibition efficiency were determined by the fluorescein amidite (FAM)-labeled ASO. FAM signals were detected predominantly in NeuN⁺ neurons but also in GFAP⁺ glial cells. Compared to the control ASO, Ptbp1-ASO only resulted in incomplete, though significant, reduction (~40%) of PTBP1 in astrocytes. Notably, the number of DCX⁺ immature neurons or NeuN⁺ mature neurons was indistinguishable between Ptbp1-ASO and the control group when examined 2 weeks after hippocampal injections. Such a result suggests that PTBP1 inhibition by ASO fails to induce immature or mature neurons, consistent with the data generated by using a shRNA-based approach (Guo et al. 2022a).

4.4. PTBP1 Inhibition Fails to Convert Müller Glia to Retinal Neurons

Müller glia in the retina play a similar role as astrocytes in the brain (Bringmann et al. 2006). They can be reprogrammed by controlling fate-determining factors to produce certain types of retinal neurons (Hoang et al. 2020, Jorstad et al. 2017, Yao et al 2018). The claim that PTBP1 inhibition alone could rapidly convert Müller glia into new RGCs and restore visual function after retinal injury was remarkable (Zhou et al. 2020; however, it was not supported by results of recent replication studies (Hoang et al. 2021, 2022; Xie et al. 2022).

4.4.1. PTBP1 inhibition via CRISPR-CasRx fails to convert Müller glia into retinal ganglion cells.

To verify a role of PTBP1 inhibition in reprogramming Müller glia, Xie et al. (2022) first used the original CRISPR-CasRx system as employed by Zhou et al. (2020); however, they simultaneously traced the fates of Müller glia with GFP in Glast-CreER^T;Rosa-CAG-LSL-Sun1-GFP mice after tamoxifen treatments. Consistent with results in the mouse brain previously reported by Wang et al. (2021), Xie et al. also only detected a very mild reduction of PTBP1 expression upon subretinal injection of AAV-hGFAP-CasRx-Ptbp1. When examined at 4 weeks after virus injections, GFP⁺ Müller glia neither became RGCs nor migrated to the GCL in which endogenous RGCs are located, indicating an unsuccessful conversion of Müller glia into RGCs via the CRISPR-CasRx system.

4.4.2. PTBP1 inhibition via shRNA fails to convert Müller glia into retinal ganglion cells.

By using the identical viral vectors as described by Qian et al. (2020), Xie et al. (2022) next employed the shRNA-based approach for PTBP1 inhibition in Müller glia. Müller glia were transduced with AAV-CMV-LSL-RFP-shPtbp1 or the control AAV-CMV-LSL-RFP via intravitreal injections into adult Glast-CreER^T;Rosa-CAG-LSL-Sun1-GFP mice. Tamoxifen treatments not only induced Müller glia–specific expression of RFP-shPtbp1 or the control RFP but also genetically traced Müller glia with GFP. Despite robust PTBP1 inhibition, GFP⁺ RGCs were not detected at 4 weeks after virus injection (Xie et al. 2022). Soma migration of GFP⁺ Müller glia toward the GCL, an essential first step for Müller glia-to-RGC conversion, was also not observed (Xie et al. 2022).

4.4.3. PTBP1 inhibition via genetic deletion fails to convert Müller glia into retinal ganglion cells.

To rule out any concerns associated with virus-based methods, Hoang et al. (2021, 2022) used a genetic approach to conditionally delete PTBP1 in Müller glia. This was accomplished in Glast-CreER^T;Sun1-GFP^lox/lox;Ptbp1^lox/lox mice after tamoxifen treatments. Despite an approximate 90% reduction in the number of PTBP1-positive Müller glia, none of them expressed RBPMS and BRN3B (markers for RGCs), arrestin (cone-specific marker), or OTX2 (marker for photoreceptor and bipolar cells). Neither did the authors find any evidence of Müller glia converting to RGCs or other retinal neurons, even under an injury condition.

5. MOLECULAR EFFECTS OF PTBP1 INHIBITION ON GLIAL CELLS

As described above, all replication studies with stringent methods failed to show GtN conversion via PTBP1 inhibition in any CNS regions, including the retina. Nonetheless, PTBP1 inhibition could still induce significant molecular changes in glial cells that might be indicative of fate switches or be beneficial to neural regeneration. This possibility was examined by Hoang et al. (2021) through single-cell RNA sequencing (scRNA-seq) of cells from both the adult mouse brain and retina.

5.1. PTBP1 Deletion Leads to Subtle Changes in Astrocytes

Hoang et al. (2021) performed scRNA-seq of cells isolated from the cortex, striatum, or substantia nigra of wild-type, heterozygous, and homozygous Ptbp1 mice. Reduction of Ptbp1 expression was confirmed in astrocytes of heterozygous and homozygous mice. A correlation plot showed that differential gene expression was greater in astrocytes among different brain regions than between different Ptbp1 genotypes. Neuron-specific markers were not detected in Ptbp1-deleted astrocytes. These cells still retained key markers of astrocytes, such as Sox9, and showed only very subtle changes in gene expression with either upregulation (mt-Nd4, Son, Hes5, and Mt3) or downregulation (mt-Nd3, Lars2, and Ivd) (Hoang et al. 2021). Importantly, Ptpb1-deleted cells exhibited electrophysiological features stereotypical to astrocytes but not to neurons.

5.2. PTBP1 Deletion Leads to Subtle Changes in Müller Glia

Hoang et al. (2021, 2022) also performed scRNA-seq of retinal cells of wild-type, heterozygous, and homozygous Ptbp1 mice. Reduction of Ptbp1 expression in Müller glia was confirmed in heterozygous and homozygous mice; however, they failed to detect any Ptbp1 deletion–induced changes on the relative fraction of retinal cell types. This analysis also failed to show significant changes of Müller glia–specific markers, such as Sox9 and Aqp4, among all genotypes. Genes were not induced for retinal progenitors or mature neurons in Ptbp1-deficient Müller glia. Again, Ptbp1 deletion only induced subtle changes in gene expression with either upregulation (Id3, Trf, Eno1, Mt3, and Lgals3) or downregulation (Msi2, Dio2, Rnf121, and Atp1a2). GFAP, a marker of reactive gliosis, was not induced by Ptbp1 deletion. Such unbiased molecular analyses clearly demonstrate that Ptbp1 deletion in Müller glia significantly alters neither gene expression nor their identity.

6. A THERAPEUTIC POTENTIAL OF PTBP1 INHIBITION FOR NEUROLOGICAL DISEASES?

Regardless of GtN conversion, is there a therapeutic potential for PTBP1 inhibition? The results of subsequent replication studies are summarized in Table 2.By using shRNA- or ASO-mediated PTBP1 inhibition in the SNc, Chen et al. (2022) examined but failed to observe any amelioration of motor deficits in the 6-OHDA model of PD mice. On the other hand, Guo et al. (2022a) explored the therapeutic potential of shRNA-mediated PTBP1 inhibition in the hippocampus of AD mice. Despite efficient PTBP1 inhibition, they were unable to observe improvements in either synaptic or cognitive function measured by electrophysiology or multiple behavioral paradigms. Neither did they detect a reduction of amyloid pathology in 5xFAD mice or tauopathy in PS19 mice (Guo et al. 2022a). Genetic deletion of Ptbp1 also failed to exert significant changes in retinal function examined by either dark- or light-adapted electroretinogram responses (Hoang et al.(2021). Together, this series of replication studies from multiple labs rather argues against a therapeutic potential of PTBP1 inhibition for treating neurological diseases, at least for the conditions examined in these replication studies.

7. CONCLUDING REMARKS

In reexamining PTBP1 inhibition through multiple methods, including CRISPR-CasRx, shRNA, ASO, and conditional cell type–specific knockouts, these replication studies failed to show GtN conversion in any brain regions or the retina. Neither was a therapeutic potential of PTBP1 inhibition observed when examined with electrophysiology or multiple behavioral paradigms. These results contradict those initial publications showing phenomenal biological effects and therapeutic potentials of PTBP1 inhibition. What accounts for such irreconcilable discrepancies? Besides additional investigations, stringent methods for detecting GtN conversion are critically important (Calzolari & Berninger 2021, Svendsen & Sofroniew 2022, Wang & Zhang 2022a).

As recently proposed in a perspective by Wang & Zhang (2022a), four complementary criteria should be considered for investigations of GtN conversion in vivo. (a) Cell proliferation–based labeling of new neurons has been a gold standard in the field of adult neurogenesis. This method employs BrdU or EdU, analogs of the nucleoside thymidine, to label proliferating cells and trace their progenies. Such a method is possible for studying GtN conversion since glial cells often undergo proliferation upon injury or under neurodegenerative conditions (Niu et al. 2013, 2015; Su et al. 2014; Tai et al. 2020; Wang et al. 2016; Zhang et al. 2022). (b) Genetic lineage tracing of glial cells and their progenies is another important technique (Heinrich et al. 2014, Hoang et al. 2020, Niu et al. 2013, Su et al. 2014, Tai et al. 2020, Todd et al. 2021, Wang et al. 2016, Zhang et al.(2022). The lineage-tracing technique has been employed for decades in the field of developmental biology and is used to tag the cells under investigation with a unique and permanent marker and then follow their fates in well-characterized mutant animal lines, (c) Time-lapse in vivo imaging (Pilz et al. 2018) speaks to the adage that seeing is believing. After specific labeling of the glial cells, their fates can be directly followed by in vivo time-lapse imaging through two-photon or multi-photon microscopy and postimaging confirmation of neuronal identity of those exact imaged cells. (d) scRNA-seq analysis can be used to study GtN conversion (Hoang et al. 2020, Todd et al. 2021, Zhang et al. 2022). The molecular identity of genetically traced glial cells can be transcriptionally analyzed through scRNA-seq. Fate conversion is then defined by a set of genes for immature or mature neurons. Pseudotime trajectory analysis may also reveal the sequential order of cell state progression from glial cells to immature and mature neurons.

Although a miraculous role of PTBP1 inhibition in regenerative medicine has not been confirmed by many replication studies, it is still formally possible that targeting this protein may have therapeutic values for certain disease conditions due to its critical role in controlling many aspects of RNA biology. It is puzzling that PTBP1 inhibition causes only subtle changes of gene expression in glial cells (Hoang et al. 2021), especially when considering that its inhibition induces neurons from multiple cell types in culture (Hu et al. 2018). Clearly the results from cell cultures cannot be directly extrapolated to the in vivo condition; such extrapolation is a frequently made mistake in the scientific field. Although the current replication studies do not support the idea that PTBP1 inhibition alone is sufficient to induce GtN conversion in vivo, will it still be critically involved in the conversion process by other factors? Or will it require other factors in combination to initiate the conversion? Future investigations are required.