Repressing PTBP1 fails to convert reactive astrocytes to dopaminergic neurons in a 6-hydroxydopamine mouse model of Parkinson’s disease

Lineage reprogramming of resident glial cells to dopaminergic neurons (DAns) is an attractive prospect of the cell-replacement therapy for Parkinson’s disease (PD). However, it is unclear whether repressing polypyrimidine tract binding protein 1 (PTBP1) could efficiently convert astrocyte to DAns in the substantia nigra and striatum. Although reporter-positive DAns were observed in both groups after delivering the adeno-associated virus (AAV) expressing a reporter with shRNA or CRISPR-CasRx to repress astroglial PTBP1, the possibility of AAV leaking into endogenous DAns could not be excluded without using a reliable lineage-tracing method. By adopting stringent lineage-tracing strategy, two other studies show that either knockdown or genetic deletion of quiescent astroglial PTBP1 fails to obtain induced DAns under physiological condition. However, the role of reactive astrocytes might be underestimated because upon brain injury, reactive astrocyte can acquire certain stem cell hallmarks that may facilitate the lineage conversion process. Therefore, whether reactive astrocytes could be genuinely converted to DAns after PTBP1 repression in a PD model needs further validation. In this study, we used Aldh1l1-CreERT2-mediated specific astrocyte-lineage-tracing method to investigate whether reactive astrocytes could be converted to DAns in a 6-hydroxydopamine (6-OHDA) mouse model of PD. However, we found that no astrocyte-originated DAn was generated after effective and persistent knockdown of astroglial PTBP1 either in the substantia nigra or in striatum, while AAV ‘leakage’ to nearby neurons was easily observed. Our results confirm that repressing PTBP1 does not convert astrocytes to DAns, regardless of physiological or PD-related pathological conditions.

Nevertheless, without substantiating the exact origin of the nascent-induced DAns using a reliable lineage-tracing strategy, these two outstanding works soon arouse widespread debate and argument (Jiang et al., 2021;Arenas, 2020;Qian et al., 2021). Most recently, by adopting the stringent lineage-tracing method, two studies arguing against previous findings have been published. One group shows that adeno-associated virus (AAV)-shPtbp1-induced, presumed astrocyte-converted DAns are not truly converted from astrocytes but are merely AAV-infected endogenous neurons due to virus leakage . Another group reports that no astrocyte-derived neuron, including DAns are generated in multiple brain regions including the substantia nigra and striatum, in astrocyte-specific Ptbp1 deletion mice (Blackshaw et al., 2021). However, both studies only focus on quiescent astrocytes and whether reactive astrocytes could be converted to neurons more effectively after PTBP1 repression requires further verification. During brain injury or neurodegeneration, astrocytes become activated and acquire certain characteristics of neural stem cells (NSCs) such as proliferation, Nestin-or Vimentin-immunoreactivity, and multipotency (Buffo et al., 2008;Shimada et al., 2012;Robel et al., 2011;Sirko et al., 2013). Some researchers have claimed that reactive astrocytes with stem cell hallmarks can be reprogrammed to neurons more easily and efficiently than quiescent astrocytes (Grande et al., 2013;Guo et al., 2014;Brulet et al., 2017;Wan et al., 2014;Mattugini et al., 2019). Therefore, we adopted the 6-OHDA PD model with lineage-tracing method to investigate whether reactive astrocytes could truly be converted to neurons including DAns.
Next, to investigate whether repressing astroglial PTBP1 could gradually convert astrocytes to DAns in the substantia nigra and striatum, brain slices of different timepoints (1, 2, and 3 months) after AAV injection were collected for immunostaining analysis. PTBP1 expression was not affected by AAV-shscramble, but was downregulated to undetectable levels by AAV-shPtbp1 in GFP + cells from 1 to 3 months ( Figure 1B, C), indicating astroglial PTBP1 was consistently repressed.
However, these results are not sufficient to prove that astrocytes were truly converted to neurons including DAn, as AAV-mediated gene expression could be 'leaked' into neurons as indicated by a recent study . Thus, more solid evidence such as lineage-tracing is needed to verify the exact origin of these viral-reporter-labeled neurons and DAns.

PTBP1 repression fails to convert quiescent astrocytes to DAns
Genetic lineage-tracing (Kretzschmar and Watt, 2012) has been widely recognized as the most convincing strategy for cell source identification and is generally performed by combining cell-specific Cre recombinase-expressing mice with Cre-activated reporter mice. Aldh1l1-CreERT2 mice with the highest specificity to target astrocytes (Srinivasan et al., 2016) were chosen to cross-breed with a reporter mouse Rpl22 lsl-HA (Ribotag) (Sanz et al., 2009), in which the endogenous ribosomal protein Rpl22 was tagged with three copies of the hemagglutinin (HA) epitope after Cre-mediated recombination ( Figure 2A). After Tamoxifen (TAM)-mediated induction of CreERT2 activity, almost all of the AldoC-positive astrocytes were specifically labeled with the HA epitope ( Figure 2B) and barely no HA leaky expression was observed in the neurons of the substantia nigra and striatum of Aldh1l1-CreERT2;Rpl22 lsl-HA mice.
The online version of this article includes the following source data and figure supplement(s) for figure 1: Source data 1. Brain slices co-stained with PTBP1 (red) and GFP (green) at indicated timepoints after AAV-shPtbp1 or AAV-shscramble delivery in the substantia nigra for Figure 1B, C.
Source data 2. Brain slices co-stained GFP (green) with TH (purple) or NeuN (red) at indicated timepoints after AAV-shPtbp1 or AAV-shscramble delivery in the substantia nigra for Figure 1D, E.
Our results showed that 6-OHDA induced severe lesions in the nigrostriatal pathway, characterized by significantly reduced numbers of DAns in the substantia nigra and remarkably decreased densities  Tamoxifen (TAM) induction and representative images of the substantia nigra or striatum of Aldh1l1-CreERT2;Rpl22 lsl-HA mice co-stained hemagglutinin (HA) (red) with pan-astrocyte marker AldoC (green) and tyrosine hydroxylase (TH) (purple) 2 weeks after TAM administration. Scale bar, 100 μm. (C) Schematic of experimental design. Representative images of brain slices co-stained GFP (green), HA (red) with TH (purple) in the substantia nigra (D) or with NeuN (purple) in striatum (E) 3 months after AAV-shPtbp1 delivery. n = 3 biological repeats per group. Arrows indicate GFP/TH (D) or GFP/NeuN (E) double positive neurons that are HA negative. Scale bar, 75 μm.
The online version of this article includes the following source data and figure supplement(s) for figure 2: Source data 1. Indicated brain regions of Aldh1l1-CreERT2;Rpl22 lsl-HA mice co-stained hemagglutinin (red) with pan-astrocyte marker AldoC (green) and tyrosine hydroxylase (purple) 2 weeks after Tamoxifen administration for Figure 2B.
Source data 2. Brain slices co-stained GFP (green), HA (red) with TH (purple) in the substantia nigra 3 months after AAV-shPtbp1 delivery for Figure 2D.
Source data 3. Brain slices co-stained GFP (green), hemagglutinin (red) with NeuN (purple) in the striatum 3 months after adeno-associated virus-shPtbp1 delivery for Figure 2E.  of TH + fibers in the striatum ( Figure 3B). Meanwhile, astrocytes became remarkably activated, as indicated by classic cytoskeletal and morphological changes, including hypertrophy of the main processes and cell bodies and upregulation of intermediate filament protein GFAP ( Figure 3B). However, under such circumstance, no NeuN + neurons including TH + DAns positive for HA, could be detected, and no obvious morphological changes of astrocytes (indicated by HA staining) were observed after PTBP1 repression ( Figure 3C), suggesting that neither AtoN nor astrocyte-to-DAn (AtoDAn) conversion occurred. Moreover, the number of NeuN + neurons ( Figure 3D) and TH + DAns ( Figure 3E) did not increase after AAV-shPtbp1 delivery into the substantia nigra compared with AAV-shscramble delivery. The motor deficits induced by 6-OHDA lesion (reflected by apomorphine-induced rotation) were not improved by AAV-shPtbp1 injection, either ( Figure 3F).
Together, these data demonstrate that repressing PTBP1 also fails to generate DAns from reactive astrocytes in a mouse 6-OHDA model of PD.
ASO-mediated PTBP1 repression still fails to convert reactive astrocytes to DAns in a 6-OHDA mouse model of PD To rule out the possibility that AAV toxicity (Johnston et al., 2021) restrained the AtoN conversion process, we synthesized antisense oligonucleotide (ASO) against mouse Ptbp1 as an alternative strategy for PTBP1 repression ( Figure 4A). Immunofluorescence results showed that ASO was distributed broadly in the midbrain, as indicated by ASO-attached Cy3, and astroglial PTBP1 was significantly downregulated for 2 months after ASO-Ptbp1 delivery compared to ASO-Ctrl delivery (Figure 4-figure supplement 1A, B). Western blot analysis of the midbrain further confirmed the knockdown efficiency of ASO-Ptbp1 (Figure 4-figure supplement 2C, D). Using brain slices from the same mice, we did not find any astrocyte-originated neurons (YFP + NeuN + ) or DAns (YFP + TH + ) ( Figure 4B), suggesting that no neurons, including DAns, were converted from quiescent astrocytes after ASO-Ptbp1 delivery. Next, we injected ASO-Ptbp1 or ASO-Ctrl into the substantia nigra of 6-OHDA lesioned Aldh1l1-CreERT2;Rpl22 lsl-HA mice ( Figure 4C). Two months after ASO delivery, we still could not find any neurons including DAns positive for HA ( Figure 4D), indicating that no neurons, including DAn, were generated from reactive astrocytes. Furthermore, the motor deficits of the 6-OHDA lesioned mice were not alleviated by 2 months of ASO-Ptbp1 treatment ( Figure 4E), similar to those of ASO-Ctrl (Figure 4-figure supplement 2).
These data demonstrate that repressing astroglial PTBP1 via ASO also fails to generate DAns from either quiescent or reactive astrocytes in a 6-OHDA mouse model of PD.

Discussion
In this study, through stringent and convincing lineage-tracing technology, we substantiated that neither AAV-shRNA-nor ASO-mediated astroglial PTBP1 repression could achieve AtoN or AtoDAn conversion either in the substantia nigra or in the striatum of a 6-OHDA mouse model of PD.
The online version of this article includes the following source data for figure 3: Source data 1. Brain slices of the substantia nigra or striatum after 6-OHDA lesion, co-stained with TH (green) and GFAP (red) for Figure 3B.
Source data 3. Original data and statistical analysis of Figure 3D and E for Figure 3D&E.
Source data 4. Original data and statistical analysis of Figure 3F. expressed in astrocytes 7 days post-infection. After 1-3 months of infection, AAV-shPtbp1 allowed for low levels of reporter protein (GFP) expression in neurons, whereas AAV-shscramble-mediated GFP expression was still restricted to astrocytes. Without stringent lineage-tracing, the result can be easily misinterpreted as PTBP1 repression-mediated AtoN conversion. The reason why AAV-shPtbp1 rather than AAV-shscramble leaked into neurons is currently unclear. According to a recent study, coding sequences of some proneural genes, such as Neurod1, could activate GFAP promoter elements in cis Source data 1. Brain slices of Aldh1l1-CreERT2;Rosa26 lsl-YFP mice co-stained with YFP (green) and NeuN (red) or TH (purple) after ASO-Ptbp1 or ASO-Ctrl delivery in the substantia nigra for Figure 4B.
To exclude the potential toxicity of AAV, we adopted ASO as an alternative method for PTBP1 repression and found ASO-Ptbp1 also failed to convert astrocytes into neurons or DAns. This result is inconsistent with that of a previous study claiming that AtoN conversion occurred after ASO-Ptbp1 delivery (Qian et al., 2020). However, we believe that their result is not convincing since the reporter mice-Rosa-Tdtomato (Ai14) used for lineage-tracing by Qian et al., 2020 has been questioned for occasional leakage to neurons . Other possible reasons for the discrepancy may be different ASO-mediated PTBP1 repression efficiency, different lineage-tracing mouse types and genetic backgrounds, and different experimental time lengths.
The major weakness of the present study is that we only rule out the possibility of astrocyte conversion to neuron, including DAn using Aldh1l1 promoter-based lineage-tracing mice. Whether other latent neurogenic cell types such as NSC (Maimon et al., 2021), oligodendrocytes (Weinberg et al., 2017), or NG2 glia, could be converted to neurons upon PTBP1 repression requires further investigation. Our results showed that ASO had no cell selectivity and could non-specifically enter different cell types to repress PTBP1 expression. These neurogenic cell types could be converted into neurons after PTBP1 repression Therefore, a reliable lineage-tracing method targeting these neurogenic cell types is necessary for future studies to identify the genuine cell identity that might contribute to neuron restoration.
One important question of the present study is whether and to what extent the reactive state of astrocytes in the 6-OHDA model could truly reflect the real state of astrocytes in PD patients. Acute lesions induced by neurotoxins, such as 6-OHDA, usually results in substantial neuron loss, creating an inflammatory microenvironment characterized by the presence of both A1 (pro-inflammatory) and A2 (anti-inflammatory) astrocyte subtypes (Ryu et al., 2020). In contrast, as one of the most important risk factors for PD, normal aging induces pro-inflammatory A1 like astrocyte reactivity (Clarke et al., 2018), which may accurately reflect the real state of PD patients. In particular, during ischemia stroke, a classic model characterized by the presence of A2 astrocytes, astrocytes spontaneously become neurogenic and the Notch pathway is repressed (Magnusson et al., 2014). We therefore assume that pro-inflammatory cytokines, such as tumor necrosis factor-α (TNF-α), could be detrimental to the AtoN conversion process and the survival, maturation, and subsequent neurite outgrowth of the newborn neurons (Karimi-Abdolrezaee and Billakanti, 2012;Belenguer et al., 2021;Park and Bowers, 2010), whereas the anti-inflammatory cytokines and neurotrophic factors, such as brainderived neurotrophic factor might be beneficial and even critical (Niu et al., 2013). Therefore, whether A2 astrocytes could be converted to neurons including DAns more efficiently, and more importantly, how to induce beneficial A2 astrocytes in the brains of PD patients for neural repair and regeneration needs further investigation.

Animals
All animal experiments were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee of University (Approval number: 2018-059). The protocol was reviewed and approved by the Ethics Committee on Laboratory Animal Care. The mice were housed in rooms with controlled 12 hr light/dark cycles, temperature, and humidity, and food and water were provided ad libitum. Eight-to 10-week-old C57BL/6 mice weighing 22-26 g were obtained from the Vital River Laboratory Animal Technological Company (Beijing, China). Aldh1l1-CreERT2 transgenic mice (Stock number #029655), Rpl22 lsl-HA (Ribotag) mice (Stock number #011029), and Rosa26 lsl-YFP mice (Stock number #006148) were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Aldh1l1-CreERT2 mice were used for breeding to the Rpl22 lsl-HA mice or Rosa26 lsl-YFP mice. Eight-to 10-week-old Aldh1l1-CreERT2;Rpl22 lsl-HA and Aldh1l1-CreERT2;Rosa26 lsl-YFP mice were used for lineage-tracing experiments.

Tamoxifen (TAM) administration
The protocol of TAM administration was determined according to previous work (Srinivasan et al., 2016) with little modifications. Briefly, TAM-free base (Sigma, Shanghai, China) was dissolved in corn oil (Aladdin, Shanghai, China) at a concentration of 10 mg/mL in a 60°C water bath for 30 mins. TAM was orally administered at a daily dose of 100 mg/kg body weight for 5 consecutive days. Experiments were performed 2 weeks after the last TAM administration.

Apomorphine-induced rotation
Apomorphine-induced rotation was performed 3 weeks after 6-OHDA lesion and 3 months after AAVs delivery or 2 months after ASOs delivery. Briefly, 10 mins after intraperitoneal injection of apomorphine (Sigma-Aldrich, 5 mg/kg dissolved in ice-cold saline solution), each mouse was placed in an opaque cylinder (30 cm diameter) for free moving with a camera recording above for 20 mins as reported .
Before injection into the mouse brain, the AAVs were adjusted to 1E+12 vg/mL using sterile Dulbecco's phosphate buffered saline (DPBS, Gibco, Thermo Fisher Scientific, Inc, Waltham, MA, USA). Wildtype or lineage-tracing mice were subjected to AAV injection into the substantia nigra (1 μL) or striatum (2 μL), respectively, at a speed of 100 nL/min. The coordinates indicating distance from bregma were A/P = -2.90 mm, M/L = 1.30 mm, and D/V = -4.35 mm for the substantia nigra, and A/P = 0.80 mm, M/L = 1.60 mm, and D/V = -2.80 mm for the striatum. After injection, the needle remained in place for at least 5 mins to prevent retrograde flow along the needle track, and the needle was slowly removed from the mouse brain. Cleaning and suturing of the wound were performed after the needle was removed.

Antisense oligonucleotides (ASOs) synthesis and delivery
ASOs were synthesized by Synbio Technology (Suzhou, China), the sequence and modification for ASO-Ptbp1 (5ʹ-GTGG AAAT ATTG CTAG GCAC -3ʹ) and control ASO (5ʹ-CCTA TAGG ACTA TCCA GGAA -3ʹ) were performed as reported (Maimon et al., 2021). Briefly, 10 core 2ʹ-deoxyribonucleotides in the central were flanked on both 5ʹ and 3ʹ sides by 5 2ʹ-methoxyethyl (MOE)-modified nucleotides. The backbones of all ASOs contain phosphorothioate modifications and all cytosine residues were modified as 5ʹ-methylcytosines. Cyanine dye Cy3 was attached to the 3ʹ end of those ASOs for fluorescence detection. After dissolved in sterile and Rnase-free DPBS at a concentration of 1 μg/μL, ASOs were subpacked and stored at -80°C to avoid repeated freezing and thawing. A 2 μL of ASO was injected into the substantia nigra (A/P = -2.90 mm, M/L = 1.30 mm, and D/V = -4.35 mm) of astrocyte-specific lineage-tracing mice with or without 6-OHDA lesion.

Statistics
GraphPad Prism (GraphPad software, version 9.0) was used for the statistical analysis. All data are presented as mean ± SEM (standard error of the mean). When comparing data from two groups, a two-tailed Student's t test was used. When there were two variables, ANOVA followed by Tukey's multiple comparisons test was used. For all analyses, statistical significance was considered when probability value of p<0.05. R&D Program of Guangdong Province (2018B030337001), the Guangdong Provincial Key Laboratory of Brain Function and Disease (2020B1212060024).