Anterograde monosynaptic transneuronal tracers derived from herpes simplex virus 1 strain H129

To examine the labeling efficiency of H129, we constructed a replication competent recombinant virus H129-G4 that contains 4 GFP genes in the genome (Fig. 1a). H129-G4 is capable to spread transneuronally through the primary motor cortex (M1) pathway and efficiently label the downstream brain regions of M1 (Fig. 1b-d). The fluorescence intensity of H129-G4 is sufficient to clearly label and visualize the neuron morphological details, including the dendrites, spines and axonal fibers (Fig. 1e). The bright labeling makes H129-G4 compatible with the fluorescence Micro-Optical Sectioning Tomography (fMOST) (Fig. 1f-k) [15], while other versions containing 1 to 3 copies of GFP genes created by us, as well as the H129 tracers from Lo and McGovern [11, 12], failed to combine with fMOST due to insufficient labeling intensity (data not shown). The combination of H129-G4 tracing and fMOST imaging enables the high-throughput imaging and automated reconstruction of neuronal circuit in the entire brain with high resolution at submicron levels.

In addition, H129-G4 is also applicable to trace the neuronal circuit of the tree shrew (Tupaia belangeri chinensis) (Additional file 1: Figure S1), a small mammal more closely related to the primates than rodents at behavioral, anatomical, genomic, and evolutionary levels [16, 17]. These results indicate a high tracing efficiency of H129-G4.

Next, we employed microfluidic plates to characterize the invasion and transmission properties of H129 in vitro [18]. The plate consists of two isolated culture chambers connected by multiple microchannels (Additional file 2: Figure S2a), which only allows the grow-through of axons but impedes that of dendrites (Additional file 2: Figure S2b). Presynaptic marker synaptophysin and postsynaptic density protein (PSD-95) were observed in the afferent chamber, suggesting the potential synapse formation near the microchannels (Additional file 2: Figure S2c). Notably, the hydrostatic pressure generated from the medium height difference guarantees the flow direction in the microchannel and completely prevents the diffusion or leakage of the virus to the opposite chamber (Additional file 2: Figure S2d) [18]. The replication competent tracer H129-G4, which serve as a reference for the mono-synaptic tracer H129-ΔTK-tdT described below, was applied to the microfluidic plate to investigate its in vitro invasion and transmission property.

To determine the virus invasion feature, mouse hippocampal and cortical neurons cultured in one chamber, and H129-G4 was added to either the soma or axon terminal side (Fig. 2a). H129-G4 in the soma side well labeled the neurons with GFP by 24 h post infection (hpi) (Fig. 2b, left panel). When H129-G4 was added to the axonal terminal side to a final concentration of 1 × 10⁷ pfu/ml, no GFP labeled neuron was observed by 24 hpi (Fig. 2b, right panel) and even up to 96 hpi (data not shown). On the contrary, VSV labelled a vast number of somas by terminal invasion at the same concentration. (Additional file 3: Figure S3c). Terminal invasion of H129 has been reported previously [2], which was also observed in our microfluidic assay with higher H129-G4 concentrations and extended infection time (Additional file 3: Figure S3a-b). The positive correlation between the terminal invasion incidence and virus concentration suggests the necessity of optimizing the injection virus dose for in vivo tracing. Notably, H129-G4 displayed greater retrograde labeling incidence in ganglion neurons than in hippocampal and cortical neurons, suggesting that terminal invasion varies in different types of neuronal cells (Additional file 3: Figure S3b).

To test the transneuronal transmission direction in vitro, neurons were plated in both chambers with a 4-day interval between culture initiation in each chamber (Fig. 2c). H129-G4 was added to either the efferent or afferent chamber to a final concentration of 1 × 10⁷ pfu/ml, a dose causing no observed terminal invasion (Fig. 2c). During infection in efferent neurons, H129-G4 spread through the axons in the microchannels, and labeled the afferent neurons in the opposite chamber (Fig. 2d, left panel). But H129-G4 in the afferent chamber only efficiently infected the afferent neurons, and no GFP labeled efferent neuron was observed at 24 hpi (Fig. 2d, right panel) and even up to 72 hpi (data not shown). The fluorescence intensity in the chambers were quantified, and no retrograde transmission were observed in all experiments (Fig. 2e). These data indicate that H129-G4 transmits between the cultured neurons in the microfluidic plate in an anterograde manner.

Since the microchannel system can’t completely mimic the in vivo physiological condition, we further determined the invasion and transmission feature of H129 in vivo.

The unidirectional projection from retina to the lateral geniculate nucleus (LGN) provides an ideal pathway to investigate the terminal invasion and transmission direction of H129 in vivo [19]. In consistence with previous reports [2, 6, 11], H129-G4 efficiently transmitted anterogradely from the retina, labeled LGN and downstream brain regions (Additional file 4: Figure S4). LGN directly projects to layer IV of primary visual cortex (V1) [20], and GFP signal was indeed observed at this layer when H129-G4 was injected in to LGN. But retinas were also labeled when H129-G4 was injected to LGN, indicating terminal invasion to retinas and/or the viruses released from retinas secondarily infecting retina cells (Additional file 5: Figure S5a). The labeled retina cell number and their occurrence time are correlated with the dose of injected virus, again suggesting the importance of optimizing tracing condition and setting parallel retrograde tracer control in vivo. No retrograde transmission was observed under this applied experimental condition, as H129-G4 injected at V1 didn’t spread to retinas through LGN (Additional file 5: Figure S5b). Some non-ganglion cells in the retina were GFP positive when H129-G4 was injected into LGN, especially with higher dose of virus, which was probably due to the secondary infection of virus progeny released from the dead infected ganglion cells. It is of significance to test the retrograde transsynaptic transmission of H129 with the monosynaptic tracer H129-ΔTK-tdT, since the replication competent H129-G4 may cause secondary infection by the offspring virus released from infected neurons.

To investigate H129H transmission feature, H129H-G4 was also examined in the well documented neural circuit between ventral posteromedial thalamic nucleus (VPM) and the primary somatosensory cortex (S1) (Additional file 6: Figure S6a) [21‐23]. H129-G4 was injected into the VPM together with Alexa Fluor 594 conjugated-cholera toxin B subunit (CTB), which labels the soma of the upstream neuron via axon terminal absorption. GFP labeled neurons were sequentially observed at nRT (2 dpi, Additional file 6: Figure S6b) and layer 4 of S1 cortex (4 dpi, Additional file 6: Figure S6c). Notably, the GFP-positive neurons and CTB-labeled neurons formed two segregated cell populations located in layer 4 and 6 of S1 at 4dpi, respectively, and no overlap of the two populations was observed (Additional file 6: Figure S6c). These data indicate that H129-G4 anterogradely and transsynaptically transmits from VPM to layer 4 of S1, but does not label layer 6 neurons via axon terminals in the VPM.

Taken together, H129 mainly infects neurons through the soma and transmits to postsynaptic neurons anterogradely. However, terminal invasion of H129 was observed in cultured neurons and retina ganglia cells, whose efficiency is correlated with virus dose.

H129-ΔTK-tdT was generated by knocking out TK gene (Fig. 3a). Consistent with previous report [24], TK deficiency caused mild growth delay of H129 in Vero cells, but impaired viral replication in cultured fetal mouse neurons (Additional file 7: Figure S7). With helper virus AAV-TK-GFP or AAV-DIO-TK-GFP compensatorily expressing TK (Fig. 3a), H129-ΔTK-tdT transmits one step to the postsynaptic neuron.

We examined the in vivo transmission direction of this monosynaptic tracing system in the visual and olfactory pathways [19, 25, 26]. AAV-TK-GFP and H129-ΔTK-tdT were injected to LGN or the main olfactory bulb (MOB) as shown in the schematics (Fig. 3b and c). Neurons labeled with tdTomato were observed in the regions of the injection sites (LGN and MOB) (Fig. 3b1 and c1), and the downstream regions including cortical layer IV of V1 (transmitted from LGN) (Fig. 3b3–5), piriform cortex (Pir) (Fig. 3c2 and 4–7) and anterior cortical amygdaloid nucleus (ACo, transmitted from MOB) (Fig. 3c7–8), indicating the anterograde transmission of H129-ΔTK-tdT. No tdTomato-labeled neuron at the corresponding upstream regions (retina or the main olfactory epithelium, MOE) was observed (Fig. 3b2 and c3), suggesting no retrograde labeling of H129-ΔTK-tdT, including both terminal invasion and retrograde transmission, under the experimental condition. Different from H129-G4, H129-ΔTK-tdT didn’t label the retinas and MOE neurons via terminal invasion, probably because of the optimized injection dose and deficient replication (Materials and methods Tables 1 and 2). When H129-ΔTK-tdT was injected to these two regions alone, tdTomato-positive cells outside the injection sites were not observed (Additional file 8: Figure S8f and b).

To further examine the terminal invasion and infection tropism in the central nervous system (CNS), H129-ΔTK-tdT was injected into more brain regions (Additional file 8: Figure S8). Terminal invasion was not detected in 5 of 6 tested regions (Additional file 8: Figure S8b-g), except the only case in CA1. A few neurons in LEC, an upstream nucleus of CA1, were labeled by H129-ΔTK-tdT, indicating that terminal invasion may occur in certain brain regions (Additional file 7: Figure S7e). In all the regions tested, H129-ΔTK-tdT displayed neuron tropism, and no virus-infected glia (GFAP positive) was observed, since all tdTomato-positive cells were NeuN positive but GFAP negative (Additional file 8: Figure S8b-g).

These results show that the H129-ΔTK-tdT is a monosynaptic tracer which anterogradely transmits and labels the postsynaptic neurons. However, proper controls are strongly recommended to exclude the misleading labeling caused by the potential terminal invasion,

Next, the H129-ΔTK-tdT monosynaptic tracing system was tested in the well documented neuronal circuit between the ventral posteromedial thalamic nucleus (VPM) and the primary somatosensory cortex (S1) [21‐23, 27]. Neurons at VPM send their axons to S1 and form synaptic contacts in cortical layer IV, V and VI; projection neurons in layer VI in turn project backward to VPM and the nucleus of reticular thalamus (nRT). Reciprocal connections also occur locally in the thalamus between relay neurons and GABAergic neurons in nRT (Fig. 4b).

AAV-TK-GFP and H129-ΔTK-tdT were injected into the VPM of wild-type C57BL/6 mice in sequence (Fig. 4a). Similar to AAV-TK-GFP alone, injection of H129-ΔTK-tdT alone only weakly labeled neurons around the injection sites but not the connected regions, indicating no terminal invasion nor non-specific transmission occurred under this condition (Fig. 4c and d). When both viruses were injected to the same site at VPM sequentially (Day 1 and Day 22), neurons expressing both GFP and tdTomato were observed at the injection site at Day 25 (Fig. 4e). These double-labeled neurons were the starter cells, in which AAV-TK-GFP complementally expressed TK supporting H129-ΔTK-tdT replication and production of virus progeny. An average of 163.0 ± 21.83 starter cells were observed in three mice (Additional file 9: Figure S9a). The reproduced viral tracers anterogradely transmitted to the postsynaptic neurons and labeled them with tdTomato. However, due to the deficient viral replication by lack of TK, H129-ΔTK-tdT was restrained in the postsynaptic neurons. The defect of viral replication resulted in weak tdTomato expression, thus antibody staining was required to amplify the labeling signal for visualization.

A few nRT neurons expressed tdTomato were first observed at Day 25 (Fig. 4e2). At Day 32, tdTomato labeled neurons were then observed at various VPM innervating regions including cortical layer IV, V, VI of S1 (Fig. 4f-h). The quantitative analysis of the labeled neuron’s amount were performed and presented (Additional file 9: Figure S9a). The axonal fibers originated from the VPM could be visualized by AAV expressed GFP (Fig. 4f2 and f3). Notably, H129-ΔTK-tdT labeled neurons in nRT appeared earlier than those in S1, probably caused by, at least partially, the difference in projection distances from VPM to these two regions.

Precisely mapping the output neuronal circuits required not only the output information from given brain regions, but also the projectome paths from specific type of neurons. For this purpose, H129-ΔTK-tdT was applied with AAV-DIO-TK-GFP helper in appropriate Cre-transgenic mice. Under the control of the Double-floxed Inverted Orientation (DIO) Cre-On system, AAV-DIO-TK-GFP expresses TK and GFP only in the presence of Cre recombinase (Fig. 3a), which assists H129-ΔTK-tdT monosynaptic transmission specifically from the Cre expressing neurons [28].

PV-Cre transgenic mice specifically express Cre recombinase in parvalbumin (PV) interneurons. The nRT contains abundant PV-neurons, which receive inputs from cerebral cortex and dorsal thalamus and project widely back to thalamic nucleuses to regulate the flow of information from thalamus to cortex (Fig. 5b) [29, 30]. So nRT was chosen as the virus injection site. Consistent with previous results, AAV-DIO-TK-GFP and H129-ΔTK-tdT restrictedly labeled the neurons around the injection site by itself (Fig. 5c-d). When both viruses were injected sequentially, GFP and tdTomato co-expressing neurons were observed at nRT at Day 25 (Fig. 5e). Since AAV-DIO-TK-GFP expresses GFP and TK specifically in Cre neurons, the yellow neurons were superinfected by both AAV-DIO-TK-GFP and H129-ΔTK-tdT, and represented the starters which initiated viral transmission. At Day 32, neurons expressing tdTomato were observed in various brain regions, including ventral posterior nucleus (VP), ventral medial nucleus (VM), posterior thalamic nuclear group (Po), parafascicular thalamic nucleus (PF), periaqueductal gray (PAG), red nucleus parvicellular part (RPC) and red nucleus magnocellular part (RMC) (Fig. 5f-j). The starter neurons and transsynaptically labeled neurons were counted, and the data were shown in Additional file 9: Figure S9b. Importantly, no neurons in the S1 was labeled, indicating absence of axon terminal invasion and retrograde transmission (Additional file 10: Figure S10).

To rule out the possibility that the nucleuses adjacent to nRT were labeled due to the virus diffusion instead of the transneuronal transmission, this system was further validated by tracing long distance projection.

DAT-Cre transgenic mice express the Cre recombinase in dopaminergic (DA) neurons under the control of the dopamine transporter (DAT) promoter. DA neurons in ventral tegmental area (VTA) widely implicate in the brain reward pathway and directly project to hippocampus, prefrontal cortex (PFC), nucleus accumbens (NAc) and amygdala (Fig. 6a) [31‐33]. Neurons labeled by individually injected viruses were restrained to the area around the injection site at VTA (Fig. 6b and c), which is consistent with the results described above. Starter neurons were observed at VTA at Day 25 (Fig. 6d). At Day 32, tdTomato labeled neurons in the Hipp-CA3, amygdala, NAc and as far as PFC, indicated that H129-ΔTK-tdT successfully transsynaptically spread from the starter DA neurons to these nuclei (Fig. 6e-g). The quantitative analysis was performed as presented (Additional file 9: Figure S9c).

Taken together, H129-ΔTK-tdT with the help of AAV-DIO-TK-GFP can monosynaptically trace the direct anterograde connections from a specific type of neurons expressing Cre.

The standards of performance (SOP) and animal studies have been approved by the Institutional Review Board and Institutional Animal Welfare Committee (WIVA10201502), including neuron isolation and intracerebral inoculation of mice and tree shrews with viral tracers. All the experiments with viruses were performed in bio-safety level 2 (BSL-2) laboratory and animal facilities.

Vero-E6 cell (Vero, ATCC#CRL-1586) was purchased from ATCC, maintained our own laboratory and tested to be Mycoplasma free. The culture medium is Dulbecco’s modified Eagle medium (DMEM, Cat. #12100046, Gibco/Life technologies) containing 10% fetal bovine serum (FBS, Cat. #12483020, Gibco/Life technologies) and penicillin-streptomycin (100 U/ml of penicillin and 100 μg/ml of streptomycin, Cat. #15140122, Gibco/Life technologies).

According to the protocol described previously [39‐42], hippocampal and cortical neurons were isolated from the forebrain of C57BL/6 mouse pups at embryonic day 18.5 (E18.5), and ganglion neurons were isolated from dorsal root ganglion (DRG) or trigeminal ganglion (TRG) of 4-week male C57BL/6 mice. Briefly, the cerebral cortex together with hippocampus were dissociated with trypsin (Cat. #15400054, Gibco/Life technologies)/DNase I (Cat. #D5025, Sigma) for 15 min at 37 °C. Isolated neurons were washed with Hank’s Balance Salt Solution (HBSS) without calcium and magnesium (Cat. #14170112, Gibco/Life technologies), resuspended and cultured in Neurobasal medium (Cat. #21103049, Gibco/Life technologies) supplemented with 2% B27 (Cat. #17504044, Gibco/Life technologies), 25 μM GlutaMAX (Cat. #35050079, Gibco/Life technologies) and penicillin-streptomycin. Medium was changed every other day. DRG and TRG dissection were treated with papain (Cat. #P4762, Sigma) solution for 15 min, then with collagenase (Cat. #17101015, Gibco/Life technologies)/dispase (D4693, Sigma) solution for addition 15 min, both at 37 °C. Percoll (Cat. #17–0891-01, GE Healthcare) gradient centrifugation was applied to separate myelin and nerve debris from sensory neurons. Acquired neurons were cultured with Neurobasal medium supplemented with 50 ng/ml mouse beta-nerve growth factor (β-NGF, Cat. #cyt-581, ProSpec), 2% B27, GlutaMAX and penicillin-streptomycin.

Both of H129-G4 and H129-ΔTK-tdT were derived from the H129 bacterial artificial chromosome (H129-BAC). H129-BAC was constructed by inserting the F-plasmid vector pUS-F5 into the H129 genome between UL22 and UL23 by homologous recombination as described previously [43]. Two binary GFP elements were inserted into H129-BAC sequentially at the indicated sites to generate H129-G4 (Fig. 1a). H129-ΔTK-tdT and H129-tdT were obtained by inserting tdTomato genes (tdT) and the Zeocin resistant gene (Zeo^R) into H129-BAC with or without deleting thymidine kinase (TK) gene (UL23) respectively (Fig. 3a). The viral genome manipulations were performed in E. coli DY380 strain via homologous recombination, and validated by PCR and sequencing. The recombinant viruses were reconstituted and produced in Vero cells following the protocol described previously [43, 44]. The average viral titer after concentrating were about 2–5 × 10⁹ pfu/ml for H129-G4 and 5–10 × 10⁸ pfu/ml for H129-ΔTK-tdT.

Two helper viruses, AAV-TK-GFP and AAV-DIO-TK-GFP were constructed using the vector of adeno-associated virus serotype 9, and were packaged by the Obio Technology Corporation (Shanghai, China). AAV-TK-GFP constitutively expresses TK and GFP, supporting H129-ΔTK-tdT monosynaptically transmission. AAV-DIO-TK-GFP conditionally expresses TK and GFP only in the presence of the Cre recombinase (Fig. 2a).

Specimen for fMOST imaging was embedded with Technovit 9100 Methyl Methacrylate (MMA, Electron Microscopy Sciences) using an optimized protocol to preserve the fine structures labeled with GFP. Briefly, PFA fixed animal brain was rinsed in 0.01 M PBS for 12 h, and completely dehydrated in a series of alcohol (50%, 75%, 95%, 100% and 100% ethanol, 2 h for each) followed by immersion in xylene twice (2 h for each) for transparentization. After dehydration, the brain was ready for infiltration: it was successively soaked in a graded series of infiltration solutions (50%, 75%, 100% and 100% resin in 100% ethanol, 2 h each for the first three solutions and 48 h for the final solution), Then the specimen was transferred into gelatin capsule and immersed in polymerization solution. Finally, the capsule with the specimen was closed and kept in a dry chamber at -4 °C in dark for 72 h. After complete polymerization, the resin-embedded mouse brain was immersed into 0.05 M Na₂CO₃ buffer to enhanced the GFP signal. With real time reactivating fluorescence, the whole brain was imaged using fMOST system at 0.5 μm × 0.5 μm × 1 μm voxel size [45, 46]. Lastly, the image stack of the acquired data set was transformed into Large Data Access using the Amira software (Visage Software, San Diego, CA, USA) for 3D image reconstruction [47].

The microfluidic plates were fabricated following the protocol described previously [18, 40, 48]. The microchannel mask and chamber mask were designed accordingly and produced by Microclear Electronics Technology (Nanjing, China). Photoresist SU-8 GM 1050 and GM 1075 (Gersteltec Sarl), propylene glycol methyl ether acetate (PGMEA, Sigma), poly (dimethylsiloxane) (PDMS) (Sylgard 184, Dow Corning) were used to fabricate the microfluidic devices. The microchannels of the plate are 700 μm length, 10 μm width and 3 μm depth (Additional file 2: Figure S2a).

To culture neurons in the microfluidic plate, freshly isolated neurons (5 × 10⁵ of fetal mouse cortical and hippocampal neurons, and 1 × 10⁵ of ganglion neurons) were plated into one chamber and cultured in 400 μl medium (Day 1) (blue chamber in Fig. 2a and c). For bi-chamber culture, a new batch of neurons (1 × 10⁵ neurons) were added into the opposite chamber (red chamber in Fig. 2c) and cultured in 200 μl medium at Day 5 of initiation of neurons culture, when the axons of the first plated neurons had grew into the microchannels. Medium was refreshed every day, and the volume in the chamber was maintained at 400 μl and 200 μl to generate the hydrostatic pressure. H129-G4 was added into the chamber with negative hydrostatic pressure to avoid virus diffusion to the opposite chamber, and the medium volume was changed accordingly prior to infection.

When indicated, the neurons grown on microfluidic plate were stained. The primary antibodies included rabbit anti-MAP2 polyclonal antibody (Cat. #17490–1-AP, Proteintech), mouse monoclonal antibody of anti-Tau (IgG1, Cat. #ab80579, Abcam), −SYP (IgG1, Cat. #sc-17,750 SantaCruz), −PSD95 (IgG2a, Cat. #75–028, NeuroMab). The secondary antibodies included Alexa Fluor 488 conjugate Goat anti-Rabbit IgG (H + L) (Cat. #A-11008, Invitrogen), Alexa Fluor 594 conjugate Goat anti-Mouse IgG1 (Cat. #A-21125, Invitrogen), Alexa Fluor 488 conjugate Goat anti-Mouse IgG1 (Cat. #A-21121, Invitrogen), and Alexa Fluor 594 conjugate Goat anti-Mouse IgG2a (Cat. #A-21135, Invitrogen).

For quality control, a few microfluidic plates from each fabrication batch (50 plates) were randomly selected to check the inter-compartment leakage. 5 × 10⁴ Vero cells were cultured in one chamber for 5 days, and H129-G4 was added to the opposite chamber (2.5 × 10⁹ pfu/ml) with less medium volume and lower hydrostatic pressure (Additional file 2: Figure S2d). GFP signals were daily monitored up to 3dpi. The batch of microfluidic plates were applied for experiment only when no leakage was observed in all tested plates.

Intracerebral virus injection was performed using a stereotaxic system in a BSL-2 animal facility following the approved SOP on 8 week-old male mice or adult tree shrews without randomization or blinding. The mice included wild-type, PV-Cre and DAT-Cre transgenic C57BL/6 mice. DAT-Cre mice specifically express Cre recombinase in dopaminergic (DA) neurons under the control of the dopamine transporter (DAT) promoter, and PV-Cre mice express Cre recombinase in parvalbumin (PV) interneurons. The anesthetized animals received intracerebral virus injection with a motorized stereotaxic injector (Stoelting Co.). The exact parameters of the mouse nucleus location was determined according to the Mouse Brain Atlas by the mediolateral (ML), anteroposterior (AP) and dorsoventral (DV) distances to Bregma [49].

H129-G4 injection was performed as listed in Table 1. When indicated, Alexa Fluor 594-conjugate cholera toxin subunit B (CTB, Cat. #C22842, Molecular Probes) was injected together with the virus to mark the injection site. Animals were anesthetized and perfused with sterile normal saline and 4% paraformaldehyde (PFA) solution at the indicated time points. The whole brain was carefully removed, fixed with 4% PFA, dehydrated in 30% sucrose, and sectioned at -20 °C for imaging. Animals were monitored daily after the virus injection, and experiment would be terminated and animal would be excluded if severe sickness was observed.

For monosynaptic tracing, helper viruses and H129-ΔTK-tdT were injected at Day 1 and 22 respectively. Perfusion were performed on Day 25 to 32. The helper viruses and H129-ΔTK-tdT were also individually injected to the same site as controls. The detailed injection parameters are listed in Table 2.

The obtained brains were coronally cryo-sectioned to 40 μm thickness slices using a microtome (HM550, Thermo/Life technologies). Neurons were stained with rabbit anti-NeuN (Cat. #ab104225, Abcam) and Alexa Fluor 647-conjugated goat anti-rabbit antibodies (Cat. #A-21245, Thermo/Life technologies) when indicated. Otherwise, the cell nuclei were counterstained with Hoechst dye 33,342 (Cat. #H3570, Thermo/Life technologies). All images were obtained using a Nikon’s A1R MP+ confocal microscope equipped with a fast high resolution galvanometer scanner. The tdTomato signal was amplified by staining with rabbit anti-DsRed polyclonal antibody (Cat. #632496, Takara) and Alexa Fluor 594-conjugated goat anti-rabbit IgG (H + L) (Cat. #A-11037, Invitrogen). When stained with NeuN, mouse anti-tdTomato monoclonal antibody (IgG2b, Cat. #TA180009, Origene) and the secondary antibody of Alexa Fluor 594-conjugate goat anti-mouse IgG2b (Invitrogen, Cat. #A-21145) were used.