Introduction

Direct neuronal reprogramming is the conversion of a somatic cell directly into a neuron, without passing through pluripotency1. The forced expression of neurogenic transcription factors in astrocytes, including NeuroD1, Sox2, Ascl1, and Neurog2, has been reported to convert transduced cells into new neurons in vitro and in vivo, primary using viral vectors for gene delivery2,3,4,5,6,7. The short human GFAP promoter, gfaABC1D (hGFAP)8 is most commonly used to target neurogenic factor expression specifically to astrocytes. The validity of in vivo astrocyte to neuron reprogramming has been challenged based on findings from lineage tracing models which demonstrated off-target neuronal expression from AAV based delivery of transcription factors using the hGFAP promoter9,10,11. This observation suggested that off-target neuronal expression was mistakenly interpreted as astrocyte to neuron reprogramming10,11. The mechanism(s) underlying this off-target expression remain largely unexplored.

AAVs are commonly employed for gene delivery due to their high gene transfer efficiency and relatively strong safety profile12,13,14,15. Often AAV capsids are selected because they have a desirable tropism for specific cell types. For instance, AAV5 has been reported to transduce astrocytes with greater efficacy than other AAV serotypes16 and accordingly, is commonly used for astrocyte to neuron reprogramming studies. However, numerous reports demonstrate that AAV5 transduces neurons in several brain regions including the striatum, hippocampus, auditory cortex and thalamus17,18,19,20. Indeed, GFAP mRNA transcripts have been previously detected in subpopulations of neurons21 which could drive vector expression in the transduced cells. Further, the potential for high AAV titres (1 × 1013 GC/mL) to alter viral tropism has also been proposed22. Hence, the delivery of AAV5 in different cortical regions, and at different doses, could alter cell-type specificity from the hGFAP promoter.

One factor that could contribute to the labeling of endogenous neurons following AAV delivery, is material transfer between astrocytes and neurons in the central nervous system. Studies in the retina have demonstrated the transfer of reporter proteins and mitochondria between transplanted and host mouse photoreceptors23,24,25. Hence, the transfer of proteins in reporter mice used in lineage tracking experiments, or fluorescent proteins driven from the viral construct could potentially contribute to the off-target labeling of neurons. Whether this phenomenon of cytoplasmic exchange occurs in the forebrain, remains unknown.

Toward designing better strategies for astrocyte targeting, various mechanisms were tested that could contribute to the non-specific labelling observed with AAV5-hGFAP promoter-based gene expression. Our findings reveal regionally distinct expression in neurons within the adult mouse cortex under the control of the hGFAP promoter. RNAscope was used to examine the presence of Gfap mRNA in neurons and found Gfap mRNA expression varied between different cortical regions. It was determined that material transfer occurs between astrocytes and neurons in vivo, but the rare frequency of the event is unable to account for the extent of neuronal labeling following AAV transduction of the hGFAP promoter-driven gene construct in the cortical parenchyma. Finally, hGFAP-promoter transcription factor binding site (TFBS) analysis was performed using the JASPAR Core 2022 database. This analysis identified numerous TFBSs within the hGFAP promoter that could potentially drive neuronal expression. Together, our study provides important insight into the cellular mechanisms that impact the use of hGFAP promoter to drive gene expression in neurons.

Materials and methods

Animal handling and ethics

Experiments were conducted in accordance with the animal use protocol (AUP200112508, 29-08-2019 start date) approved by the Animal Care Committee at the University of Toronto in adherence with guidelines published by the Canadian Council for Animal Care. The ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines were fully implemented during this study.

Mice

All mouse work was conducted in adherence to protocols approved by the University of Toronto Animal Care Committee and in accordance with the guidelines published by the Canadian Council for Animal Care. Mice were kept on a 12-h light/12-h dark cycle and were provided water and food ad libitum. Adult (8–12 weeks of age) male and female B6.129X1-Gt(ROSA)26Sortm1(EYFP)Cos/J (R26-EYFP; Jax strain 006148,26), B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J (Ai14, Jax strain 007914,27), nestin-CreERT2∷R26R-EYFP (Nestin-Cre, Jax Strain 016261), and B6.Cg-Tg(Gfap-EGFP)3739Sart/J (GFAP-GFP, Jax strain 010835,28) mice on the C57BL/6 J background were used.

AAV injections

AAV5 was administered to R26R-EYFP and Ai14 mice by stereotaxic injection. AAV was injected bilaterally into mPFC at the following coordinates: medial prefrontal cortex (mPFC) injections: AP = + 1.9 mm from Bregma, ML = + /− 0.4 mm from midline, and DV =−1.5 mm from the skull surface. Unilateral AAV injections into mPFC were administered as above, but to ML = + 0.4 mm only. AAV was unilaterally injected into the motor cortex at the following coordinates: AP = + 0.6 mm from Bregma, ML =−2.2 mm from midline, and DV =−1 mm from the skull surface. AAV injections were performed at a rate of 0.1 μL/min, with 1 μL total volume administered per site. The AAV5 vectors used contained iCre-WPRE driven by either the hGFAP or CAG promoter, at titres ranging from 1 × 109 −1.1 × 1013 GC/mL. Viral particles were prepared and packaged by Vector BioLabs (Malvern, PA, USA).

Astrocyte transplantation

Cortices were collected from postnatal day four Ai14 mice and placed in ten milliliters of astrocyte media (79% DMEM (Cat# 10569010, ThermoFisher Scientific, Waltham, MA, USA), 20% FBS (Cat# A3840001, ThermoFisher Scientific), and 1% penicillin/streptomycin (Cat# 15070063, Thermo Fisher Scientific). Tissue was dissociated, pipetted into a Poly-D-Lysine coated T75 flask (Cat# 150687, ThermoFisher Scientific), and incubated at 37 °C overnight. The next day the cells were lifted following shaking of the flasks, the media was collected, and discarded. Three milliliters of TryplE (Cat# 12604013, ThermoFisher Scientific) was added to the flask and it was incubated for five minutes at 37 °C. TryplE was inactivated with nine milliliters of astrocyte media. Media was collected and centrifuged at 300xg for five minutes. The supernatant was then aspirated, the cells were resuspended in astrocyte media and plated at 70,000 cells per well (Poly-D-Lysine coated 24 well-plates (Cat# 142475, ThermoFisher Scientific) and incubated at 37 °C overnight. The next day the cells were counted, and the supernatant was aspirated. AAV5-GFAP-Cre (1 × 1013 GC/mL) was diluted to an MOI of 90,900 in DMEM with 1% pen/strep, and 200 μL of the resulting mixture was added per well. Four hours after applying the virus, 200 μL of astrocyte media was added per well, to a total volume of 400 μL of media per well. Cells were incubated at 37 °C overnight. The following day, the media was changed to fresh astrocyte media. Two days following the transduction the cells were lifted with TryplE, centrifuged for five minutes at 300xg and the supernatant was collected and placed on ice. The cells were then resuspended in aCSF (72.4% dH2O, 6.2% NaCl (Cat# S3014, MilliporeSigma), 0.5% 1 M KCl (Cat# P5405, MilliporeSigma), 0.1% MgCl2 (Cat# M8266, MilliporeSigma, 16.9% 155 mM NaHCO3 (Cat# S5761, MilliporeSigma), 1% 1 M Glucose (Cat# G5146, MilliporeSigma), 1.9% 108 mM CaCl2 (Cat# C7902, MilliporeSigma), 1% penicillin/streptomycin) (68,000 cells/μL) and placed on ice until the time of injection into the MC or mPFC of R26R-EYFP mice as described above.

Tamoxifen administration

Tamoxifen chow (250 mg/kg, Envigo, Haslett, MI, USA) was fed to nestinCreERT2::R26R-EYFP mice ad libitum for two weeks, followed by a two-week chase period where mice were provided regular chow prior to AAV injections.

Tissue preparation

Mice received an overdose of Avertin (250 mg/kg, i.p.) (Cat# 48402, MilliporeSigma, Burlington, MA, USA) and transcardially perfused with 0.01 M phosphate buffered solution (PBS) followed by ice cold 4% paraformaldehyde (PFA) (Cat# P6148, MilliporeSigma). Brains were dissected and postfixed in PFA for four hours then cryoprotected in 30% sucrose (Cat# S0389, MilliporeSigma) and stored at 4 °C (−80 °C for RNAscope tissue) until the time of sectioning. Coronal brain Sections (20 µm thickness for immunohistochemistry, or 10 µm for RNAscope) were cryosectioned on a HM525 NX cryostat (ThermoFisher Scientific) and were placed on Fisherbrand™ Superfrost™ Plus Microscope Slides (Cat# 1255015, Thermo Fisher Scientific). All perfusion reagents were prepared with diethyl pyrocarbonate (Cat# D5758, MilliporeSigma) to inactivate RNase enzymes for RNAscope processing.

Immunohistochemistry

Tissue was stored at −20 °C for immunohistochemistry. Tissue sections were rehydrated in 0.01 M PBS with three washes of five minutes each, followed by permeabilization with 0.3% Triton-X (Cat# T8787, MilliporeSigma) for 20 min. A blocking solution was then applied for 1 h, consisting of 10% normal goat serum (Cat# 005-000-121, Cedarlane Labs, Burlington, ON, CA) and 0.3% Triton-X in PBS. Primary antibodies were applied (Supplementary Table 1) and placed in a humidified chamber at 4 °C overnight. The following day, tissue was washed in 0.01 M PBS, and secondary antibodies (Supplementary Table 1) were applied for one hour in PBS solution containing DAPI (D1306, ThermoFisher Scientific) (1:3000 dilution in PBS). The slides were washed in 0.01 M PBS for five minutes prior to applying mounting medium (Mowiol, Cat# 81381, MilliporeSigma) and cover slipped.

RNAscope

Experiments were conducted according to the manufacturer’s instructions for the RNAscope® Multiplex Fluorescent Reagent Kit v2 Assay (Cat #323,100, Advanced Cell Diagnostics, Newark, CA, USA). Briefly, cells were fixed in 4% PFA at room temperature for 30 min, stained for NeuN, then dehydrated in ethanol, treated with RNAscope Hydrogen Peroxide, treated with target retrieval, digested with RNAscope Protease IV (Cat# 322336, Advanced Cell Diagnostics), and were stained with probes for GFAP RNA (Cat #313,211-C2, Advanced Cell Diagnostics). Sections were then stained with DAPI for five minutes, washed in PBS, and mounted.

Microscopy and quantification

Images were taken on the Zeiss LSM 880 Super Resolution Confocal Microscope in the microscopy imaging laboratory (MIL) facility at the University of Toronto, using Zen 2011 software. For quantification, three representative sections from each slide were chosen, and three regions of interest (ROI) per section were imaged at 20× magnification. Quantification of cells was performed using ImageJ software (Version 1.53). NeuN+tdTom+and GFAP+tdTom+ cells were counted and then divided by total tdTom+ cells per 0.182 mm2 and represented as percent of total tdTom + cells. To evaluate the number of tdTom+ cells over time, total tdTomato numbers were averaged per section in each animal. For RNAscope, NeuN+ cells that contained at least one Gfap RNA signal were considered co-labeled. For the puncta quantification, the number of Gfap mRNA puncta were counted from Gfap mRNA+NeuN+ cells. To visualize RNAscope NeuN+ cells in Fig. 3A,B we used ImageJ to separate the DAPI (cyan) and NeuN (green) signals. A background subtraction was performed on the green channel at a value of 0.7 (70% of the cyan channel intensity was subtracted from the green channel). For the transplant experiments, the total numbers of cells expressing NeuN, tdTomato, and YFP expression, as well as the co-labeling of these markers was assessed in all tissue sections that contained tdTom + transplanted cells and averaged per section.

GFAP-promoter transcription factor binding sites analysis

The 681 bp hGFAP promoter sequence8 was aligned to the human GFAP promoter (GRCg38/hg38) using BLAT29 in the UCSC genome browser30. The JASPAR Core 2022 transcription factor binding tract31 was used to catalogue the transcription factor binding sites (TFBS) within the human GFAP sequence corresponding to the hGFAP promoter. A list of the corresponding transcription factors was entered into the Allen Brain Map Transcriptomics Explorer (mouse—whole cortex & hippocampus—10 × genomics)32. The TFBSs of transcription factors expressed in greater than ten cortical neurons at modest levels (more than 5.0 trimmed mean expression), were mapped onto the hGFAP promoter using SnapGene (Version 6.2.1, GSL Biotech LLC, San Diego, CA, USA).

Statistical analysis

GraphPad Prism 8 (GraphPad Software, La Jolla, CA, USA) was used for all statistical analysis. Analysis between two groups was performed using an unpaired t-test, and multiple unpaired t-tests were used to test two groups over several time points. Column analysis for more than two groups was assessed using one-way ANOVA with Tukey’s test for multiple comparisons. Group analyses were performed using two-way ANOVA with Tukey’s test for multiple comparisons. All analyses between proportions were conducted using the arcsin transformation prior to statistical tests. Parametric testing for ANOVA and t-test was conducted if all groups passed the Shapiro–Wilk normality test. Statistical significance was identified as p < 0.05, and the values were reported as means + /− SEM (standard error of the mean).

Results

Regionally distinct tdTomato reporter expression following AAV-GFAP-Cre injections in vivo

AAVs have been widely used with the combination of the hGFAP promoter to deliver reprogramming factors to astrocytes in the brain4,6,22,33,34. The fact that AAV-hGFAP has been shown to display off-target expression in endogenous neurons4,6,10 has implications for AAV-hGFAP-mediated expression systems to study astrocyte-to-neuron reprogramming. It was first asked whether regional tropism occurred within the brain by examining two cortical areas in the adult mouse forebrain. AAV5-hGFAP-Cre was injected into the mPFC or MC of Ai14 reporter mice and the brains were harvested at 5, 14 and 21 days post-AAV injection (Fig. 1A). As predicted, at all times examined, the presence of GFAP+tdTom+ cells (transduced astrocytes) was identified in both the mPFC and MC (Fig. 1B,C). In addition, NeuN+tdTom+ cells (transduced neurons) were observed in both regions (Fig. 1D,E). The numbers of tdTom+ (transduced cells), transduced astrocytes, and transduced neurons at 5, 14 and 21 days post-AAV injection were quantified (Fig. 1F). Interestingly, a trend towards an increase in the total numbers of transduced cells over time in the mPFC (67.65 ± 24.10 vs. 137.13 ± 98.33 vs. 209.82 ± 76.6 cells/section at 5, 14 and 21 days, respectively) was observed, which was not apparent in the MC (37.29 ± 11.06 vs. 25.68 ± 7.67 vs. 58.46 ± 31.30 cells/section at 5, 14 and 21 days, respectively) (Fig. 1F). Regarding GFAP+ astrocyte transduction, as shown in Fig. 1G, there was no change in the percentage of transduced GFAP+ cells in the MC over time (day 5: 77.17 ± 0.15%, day 14: 74.86 ± 2.03%, and day 21: 73.51 ± 4.46% GFAP+tdTom+ cells). However, a significant reduction in the percentage of transduced GFAP + in the mPFC was observed from day 5 to day 21 (day 5: 69.29 + /− 6.61 vs. day 21: 24.95 + /− 8.29% GFAP+tdTom+ cells, p = 0.007 (Fig. 1G). The reduction in transduced astrocytes in the mPFC was concomitant with a significant increase in the numbers of transduced neurons in the mPFC over time (day 5: 17.60 ± 5.53% vs. day 21: 51.85 ± 6.09% of NeuN+ TdTom+ cells, p = 0.038) (Fig. 1H). In contrast, the percentage of transduced neurons in the MC remained unchanged over time (day 5: 2.41 ± 1.40% vs. day 21: 11.38 ± 4.215% of NeuN+tdTom+ cells, p = 0.986) (Fig. 1H). Moreover, a significant ~ fivefold increase in the percentage of transduced neurons at day 21 between the MC and mPFC (p = 0.010) was observed, with a concomitant ~ threefold reduction in transduced GFAP+ astrocytes in the mPFC at this same time (p = 0.003) (Fig. 1G,H). Taken together, these findings reveal regional heterogeneity of AAV-hGFAP-Cre expression in the cortices of the adult mouse brain.

Fig. 1
figure 1

Regional comparison of AAV5-hGFAP-Cre transduction between the MC and mPFC over time (A) Experimental paradigm. Mice received an injection of AAV5 with Cre under the control of the hGFAP promoter into the mPFC and MC (one per hemisphere). Mice were sacrificed on days 5, 14, and 21 post-injection. (B, C) Images of transduced, tdTom + (red), GFAP + (green), and DAPI + (blue) in the MC (B) and mPFC (C) at 21 days post-injection. Arrowheads indicate tdTom + GFAP + DAPI + cells. Boxes are enlarged images of cells within the composites. (D, E) Images of transduced, tdTom + (red), NeuN + (green), and DAPI + (blue) in the MC (D) and mPFC (E) at 21 days post-injection. Arrowheads indicate tdTom + NeuN + DAPI + cells. Boxes are enlarged images of cells within the composite. (F) The average numbers of tdTom + cells per section in the MC and mPFC over time (n = 3–7 mice per group, mean ± SEM). (G) The percentage of GFAP + tdTom + cells at 5-, 14-, and 21-days post-injection (n = 3–7 mice per group, mean ± SEM). (H) The percentage of NeuN + tdTom + cells at days 5, 14, and 21 post- injection (n = 3–7 mice per group, mean ± SEM). One-way ANOVA tests followed by Tukey’s multiple comparisons tests were used for both (G) and (H). Statistical significance is reported between bars (p values).

Full size image

Regional heterogeneity is not due to viral titre

It was then asked if viral titre influences the regional difference in tdTom reporter expression22. A dose–response experiment using AAV5-hGFAP-Cre was performed over a range of titres (1 × 109 to 1 × 1013 GC/mL), delivered to the MC or mPFC, and examined expression at 21 days post-injection. As predicted, higher titres resulted in an increase in transduced cells (Fig. 2A), with lower titres (1 × 109 and 1 × 1010 GC/mL) not producing detectable tdTom expression. Quantification was assessed in the brains from mice that received titres ≥ 1 × 1011 GC/mL. No difference was found in the relative percentage of NeuN + tdTom + cells within the MC or mPFC with increasing titres. However, the percentage of NeuN + tdTom + cells was significantly greater in the mPFC compared to the MC at all viral doses (2.2 ± 0.2 fold (p = 0.040); 2.7 ± 0.3 fold (p = 0.001), and 4.6 ± 0.5 fold (p = 0.0002) at 1 × 1011, 1 × 1012, and 1 × 1013 GC/mL, respectively) (Fig. 2B). Hence, viral titre does not account for the increased percentage of transduced neurons in the mPFC compared to the MC.

Fig. 2
figure 2

Viral titre and tropism do not account for regionally distinct neuronal transduction (A) Mice were injected with AAV-hGFAP-Cre on day 0 and sacrificed on day 21 with different viral titres (1 × 109–1 × 1013 GC/mL; designated 9–13 on graph). The average number of tdTom + cells per section in the MC and mPFC was quantified. (B) The percentage of NeuN + tdTom + /tdTom + cells in the MC (1 × 1011 GC/mL: n = 4 mice, 1 × 1012 GC/mL: n = 6 mice, 1 × 1013 GC/mL: n = 3 mice) and mPFC (1 × 1011 GC/mL: n = 4, mice, 1 × 1012 GC/mL: n = 4 mice, 1 × 1013 GC/mL: n = 7 mice) of mice that received 1 × 1011–1 × 1013 GC/mL doses of AAV-hGFAP-Cre. One-way ANOVA followed by Tukey’s multiple comparisons tests were used to assess differences between the groups. (C, D) Images of AAV5-CAG-Cre injected MC (C) and mPFC (D) at day 21. Arrowheads indicate transduced neurons (tdTom + (red), NeuN + (green), and DAPI + (blue)). Boxes are enlarged images of selected cells within the composite image. (E, F) At 21 days post-injection there is no significant difference between the average number of tdTom + cells per section in the MC and mPFC (E) or the percentage of NeuN + tdTom + cells between regions (F). Unpaired t tests were used to assess the differences between the groups in (E) and (F) (n = 5 mice per group). Data represents mean ± SEM.

Full size image

To determine whether AAV5 tropism differed between the MC and mPFC of Ai14 mice, AAV5 with Cre-recombinase expression driven by the ubiquitous CAG promoter was injected into the MC and mPFC of Ai14 mice to reduce the bias for expression between astrocytes and neurons that was inherent to GFAP promoter driven Cre-expression. Twenty-one days following AAV administration, the numbers of tdTom+ and NeuN+tdTom+ cells were counted (Fig. 2C,D). No differences in the number of transduced cells between the MC and mPFC (245.7 ± 18.59 and 343.1 ± 44.53 tdTom+ cells/section, respectively (p = 0.078)) was observed (Fig. 2E). Similarly, the percentage of transduced neurons was not different between the MC and mPFC (50.69 ± 4.42% and 47.35 ± 2.81% NeuN+/tdTom+ cells, respectively (p = 0.541)) (Fig. 2F). These findings reveal that AAV5 tropism is similar between the MC and mPFC and suggests that another mechanism is driving the regionally distinct AAV5-hGFAP-Cre expression.

GFAP mRNA is transcribed in cortical neurons

Our findings reveal that the proportion of neurons that express tdTomato following AAV5-hGFAP-Cre injection increases over time in the mPFC, but not in the MC (Fig. 1H). This finding is consistent with the interpretation that Gfap is transcribed at low levels in mPFC neurons, which suggests that mPFC neurons have the transcriptional machinery to drive Cre expression from the hGFAP promoter. To test if Gfap mRNA is indeed transcribed in MC and mPFC neurons, RNAscope was used to visualize Gfap mRNA transcripts and co-labeled with NeuN antibody staining in sections from the MC and mPFC of control (non-AAV injected) mice. Gfap mRNA in NeuN+ cells in both the MC and mPFC was observed (Fig. 3A,B). Quantification revealed significantly less Gfap mRNA colocalized with NeuN in the MC compared to the mPFC (41.49 ± 0.72% and 50.85 ± 3.03% NeuN+GFAPmRNA+ , MC and mPFC, respectively (p = 0.040)) (Fig. 3C). Moreover, Gfap mRNA expressing neurons of the mPFC have significantly more Gfap puncta than Gfap mRNA-expressing neurons of the MC. Specifically, 1.717 + /− 0.1092 and 2.450 + /− 0.1637 SEM puncta per Gfap-expressing neuron in the MC and mPFC, respectively (p = 0.0003) (Mann–Whitney, nonparametric t-test). Hence, the increased expression of GFAP mRNA in neurons in the mPFC is consistent with the increased numbers of tdTom+NeuN+cells at 21 days post-transduction in inducible Cre reporter mice that were injected with AAV5-hGFAP-Cre.

Fig. 3
figure 3

RNAscope reveals Gfap mRNA within neurons of the mPFC (A, B) Images of sections from the MC (A) and mPFC (B) following processing for RNAscope for Gfap mRNA (GFAPRNA, red) and immunohistochemistry for NeuN (green) and DAPI (blue) staining. Arrowheads indicate the presence of Gfap mRNA in NeuN + cells. (C) Quantification of the percentage of neurons expressing NeuN + cells reveal significantly more neurons in the mPFC are expressing Gfap mRNA. Unpaired t-test used for statistical analysis. n = 3 mice per group. Data represents mean ± SEM. (D) Quantification of the number of Gfap mRNA puncta per Gfap mRNA-expressing cell, revealing significantly more Gfap mRNA puncta in Gfap + Neurons of the mPFC. Mann–Whitney, nonparametric test was performed, n = 60 cells per group distributed over n = 3 mice per group. (E) Cortical and hippocampal single-cell RNAseq data was used to analyze the expression of 131 transcription factors with binding sites in the hGFAP promoter. Thirteen transcription factors are expressed in numerous cortical and/or hippocampal cell types in the mouse brain. Image was adapted from the Allan Brain Atlas Transcriptomics Explorer (https://celltypes.brain-map.org/rnaseq/mouse_ctx-hpf_10x). Examples of layer 2—layer 6 cortical neurons are highlighted with a dashed white line. (F) Transcription factor binding sites (green) of the thirteen cortically and/or hippocampally expressed transcription factors were mapped onto 15 TFBS within the hGFAP promoter (blue), revealing clustering of twelve of the fifteen TFBS at two loci in the promoter (160–216 bp, and 531–568 bp). CGE—Caudal ganglionic eminence neighborhood, MGE—medial ganglionic eminence neighborhood, Sst—Somatostatin subclass, NP—Cortex layer 5/6 near projecting (NP)/ Layer 6 corticothalamic (CT)/layer 6b (L6b)(NP/CT/L6b) neighborhood, CA3—Cornu Ammonis 3 subclass, V—Vascular and immune cells.

Full size image

The hGFAP promoter possesses binding sites for TFs expressed in cortical neurons

A transcription factor binding site analysis (TFBS) was performed in the hGFAP promoter to gain insight into whether TFs expressed in cortical neurons could explain the neuronal hGFAP expression in transduced cells. To employ the JASPAR 2022 tract on the UCSC genome browser, the hGFAP promoter was aligned to the human GFAP sequence. Within the aligned human promoter, 225 TFBS were catalogued, corresponding to 131 transcription factors (Supplementary Table 2). Of these transcription factors, 13 were at least modestly expressed in subpopulations of mouse cortical neurons, corresponding to 15 TFBS within the hGFAP promoter (Fig. 3D). Interestingly, 12 of the 15 TFBS are located in two loci within the promoter (160–216 bp, and 531–568 bp), representing regions that could potentially drive cortical neuron expression from the hGFAP promoter (Fig. 3E). Hence, the TFBS analysis supports the possibility that transcription factors expressed by cortical neurons are sufficient to drive the expression of Cre-recombinase from AAV-hGFAP-Cre transduced cells.

Material transfer occurs at low frequencies between donor astrocytes and host neurons

Finally, it was interesting to assess whether material transfer, the process of exchanging intracellular components between cells, was contributing to the presence of tdTom+ neurons in the brains of Ai14 mice injected with AAV5-hGFAP-Cre. In our experimental mouse model, the transfer of Cre recombinase or tdTomato protein between astrocytes and neurons could potentially lead to the detection of NeuN+tdTom+ cells. To test our hypothesis, astrocytes were isolated from postnatal day 4 Ai14 mice and transduced with AAV5-hGFAP-Cre before transplantation into the MC or mPFC of lox-stop-lox YFP reporter mice and brains were analyzed at 21 days after astrocyte transplantation (Fig. 4A). If material transfer of tdTom occurs between astrocytes and neurons, then the presence of tdTom+ neurons in the transplanted brains would be an expected observation. If material transfer of Cre occurs between astrocytes and neurons, then the presence of YFP+ neurons would be an expected observation. Similar numbers of tdTom + cells in the MC and mPFC (184.00 ± 70.12 and 202.25 ± 30.58 tdTom+ cells per section, respectively; p = 0.819) were found (Fig. 4B,C), indicating that the survival of transplanted astrocytes was not dependent on the cortical region. Interestingly, the evidence of material transfer in both the MC and mPFC was observed. NeuN+tdTom+ cells were observed in both the MC and mPFC (1.00 ± 0.71 and 8.00 + /− 3.49 NeuN+ tdTom+ cells per section, MC and mPFC, respectively), indicating the transfer of tdTom protein from the donor astrocytes to the host neurons. (Fig. 4D,E). Rare YFP + cells were also found in the MC and mPFC (4.50 ± 2.40 and 4.25 ± 1.49 NeuN+ YFP+ cells per section, MC versus mPFC, respectively (p > 0.999)) suggesting the material transfer of Cre protein from donor astrocytes to host cells (Fig. 4E,F). These results indicate that material transfer can occur between astrocytes and neurons and may contribute, in small part, to the NeuN+tdTom+ cell population seen in AAV5-GFAP-Cre injected Ai14 mouse brains.

Fig. 4
figure 4

Material transfer occurs between transduced astrocytes and neurons (A) Experimental paradigm schematic. Astrocytes from early postnatal Cre-inducible mice were transduced with AAV5-hGFAP-Cre and injected into the MC and mPFC of Cre-inducible YFP mice. Mice were sacrificed at 21 days post-transplantation. Created in BioRender. Lab, L. (2024) https://BioRender.com/z30e902. (B) Images of sections from transplanted brains reveal the presence of tdTom + astrocytes (red). NeuN + (neurons, white), DAPI (nuclei, blue). Arrowheads are colocalized cells. Box inset is an enlarged example of a cell in the composite. (C) Quantification reveals similar numbers of tdTom + cells in the MC and mPFC at 21 days post-transplantation. Unpaired t-test was used to assess the differences between the groups, n = 4 mice per group. Data represents mean ± SEM. (D) Image from the mPFC of NeuN + tdTom + cells visualizing material transfer of tdTom protein from donor astrocytes to host neurons. Arrowheads are colocalized cells. Box inset is an enlarged example of a cell in the composite. (E) Quantification of material transfer events: tdTom + NeuN + are endogenous neurons with tdTomato protein (material transfer); YFP + cells had Cre protein transferred; YFP + NeuN + represent endogenous neurons with material transfer of Cre protein; tdTom + YFP + are cells with tdTomato and Cre protein transfer;and tdTom + YFP + NeuN + indicate endogenous neurons that had both tdTomato and Cre transferred. Two-way ANOVA was used to assess the differences between the groups, n = 4 mice per group. Data represents mean ± SEM. (F) Representative image of material transfer of Cre from transplanted, transduced astrocytes to host neurons (NeuN+ /YFP + cell).

Full size image

Discussion

The hGFAP promoter has been an important tool for driving astrocyte-specific gene expression in mouse models35,36. However, it has also been reported that the transgenes driven by the hGFAP promoter show nonspecific expression in neurons10,11,37. Here, regionally distinct transgene expression in the adult mouse forebrain was demonstrated. Specifically, in the MC, hGFAP promoter-driven transgenes show predominant expression in astrocytes and a minority of neurons, consistent with previous findings8,38,39,40. In the mPFC, hGFAP-driven transgene expression shows a lack of astrocyte specificity. A greater percentage of neurons in the mPFC express GFAP mRNA compared to the MC was demonstrated, which is consistent with the increased transgene expression from the hGFAP promoter. Further, rare instances of material transfer between astrocytes and neurons was observed, which was more prominent in the mPFC compared to the MC. Our study highlights regionally distinct variation in neurons that warrants consideration when designing genetic constructs for gene therapies that target astrocytes, including astrocyte to neuron reprogramming.

A number of studies have demonstrated that high AAV titres can be toxic41,42,43 and can decrease promoter specificity44. Here, whether AAV titre impacts the specificity of the hGFAP-driven transgene expression was tested. Consistent with previous reports45, no differences in the cell specificity of hGFAP expression following AAV5 transduction was found across a broad range of titres (1 × 109–1 × 1013 GC/mL) in the MC and mPFC. Similarly, it has been previously established that the tropism of AAV vectors can differ between different brain regions18,46,47,48. However, when the ubiquitous CAG promoter was used to explore AAV5 tropism in the MC and mPFC, a similar proportion of neurons were transduced in both brain regions, suggesting that AAV5 does not show a neuron-biased tropism. This finding suggests that the hGFAP promoter restricts Cre expression to astrocytes in the MC, but not in the mPFC. These regional differences are consistent with previous studies showing contradictory neuronal48,49 versus glial transduction bias50 following AAV transduction. The findings have implications for the development of therapeutics aimed at targeting specific cell populations cells, highlighting the importance of understanding the fundamental biology and heterogeneity of target cells.

Our finding that Gfap mRNA is expressed in cortical neurons in the adult mouse brain is consistent with previous reports of Gfap mRNA splice variants being endogenously expressed in neurons51,52. More pronounced Gfap mRNA expression in neurons in the mPFC compared to the MC was observed. Accordingly, it was hypothesized that the hGFAP promoter is more active in the neurons of the mPFC and, albeit weak, this would be sufficient to drive expression of the transgene in neurons in a regionally distinct manner. Additionally, it was recently hypothesized that there may be a time-dependent activation of the hGFAP promoter in neurons, possibly due to the expression of specific genes or the result of a neurodegenerative state10. Our data supports this hypothesis as an increase in the numbers of transgene-expressing neurons in mice injected with AAV5-hGFAP-Cre from day 5 to 21 was found in the mPFC, but not in the MC, further establishing the impact of regionally distinct cell heterogeneity. Notably, we also report significant differences in the amount of Gfap mRNA in individual cells within the mPFC compared to the MC. Together, these findings suggest that the frequency of neurons that express Gfap mRNA combined with increased expression of Gfap mRNA transcripts is regionally distinct and this may account for the gradual increase in tdTomato expression in mPFC neurons over time. While many reprogramming studies focus on the MC, where we did not observe a significant difference in the numbers of transduced neurons over 21 days post-transduction, Livingston et al.,6 did demonstrate that the numbers of transduced neurons at longer survival times (56 days post-transduction and 63 days post-stroke) was increased in the MC compared to 28 days post-stroke. Hence, our findings highlight the important considerations of regional heterogeneity and the impact of injury when using the hGFAP promoter as an astrocyte targeting vector.

Towards the aim of designing astrocyte specific tools, a better understanding of how astrocyte promoters are regulated is needed. Here, it is shown that there are transcription factor binding sites in the hGFAP promoter that correspond to transcription factors expressed in neurons. This could help explain why neuronal expression of the hGFAP promoter was observed and it may also contribute to further modifications to the hGFAP promoter that can confer greater astrocyte specificity. One possible solution to avoid the non-specific expression of reporter proteins such as tdTom in neurons in the Cre/lox system could be to include a target sequence for miRNA124, or other neuronal miRNAs, in the AAV5-hGFAP-Cre construct. This microRNA is expressed in neurons and could inhibit the expression of Cre upon neuronal transduction40 allowing for more refined expression of transgenes.

Material transfer between photoreceptors has been well documented following cell transplantation in the retina and involves the exchange of macromolecules through nanotubes24. Astrocytes have been shown to donate mitochondria to neurons following injuries such as stroke53. Here, it was demonstrate that material transfer can occur between cortical astrocytes and neurons. Different from photoreceptors, where material transfer frequency can be quite high (for example, 81.4% of all fluorescently labelled ‘transplanted cells’ were actually endogenous photoreceptors in an early study of material transfer24), the frequency of observing material transfer between astrocytes and neurons is very rare (less than five cortical neurons had evidence of material transfer). One possibility is that the frequency of material transfer between astrocytes and neurons would be greater between cells that were already integrated within the cortical cytoarchitecture, in which case we would be underestimating the impact of material transfer on the neuronal labeling following viral transduction in our model. Future studies to assess material transfer at later timepoints post transplantation may reveal differences in the occurrence of material transfer in cortical tissue. Nevertheless, while material transfer from transduced astrocytes may contribute to neuronal labeling, it does not can account for the high percentage of transgene expressing neurons in our study.

The use of gene therapy approaches to replace lost neurons following injury or disease through direct cellular reprogramming of astrocytes to neurons has great potential to treat a wide range of neural disorders. Our findings identify important factors to consider when designing AAVs for astrocyte-specific gene expression. A critical factor in these AAVs is the choice of astrocyte-specific promoter. Traditionally, the GFAP promoter has been the gold standard for targeting expression to astrocytes54. However, advances in genomics have driven the development of new astrocyte specific promoters, such as a promoter built from the human S100B gene55,56. Future studies may include the design of new astrocyte-specific promoters that are rigorously tested across brain regions in both health and disease.