Tracing Synaptic Connectivity onto Embryonic Stem Cell-derived Neurons§


  • Isabella Garcia,

    1. Program in Developmental Biology, Baylor College of Medicine, Houston, Texas, USA
    2. Medical Scientist Training Program, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Longwen Huang,

    1. Department of Neuroscience, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Kevin Ung,

    1. Department of Molecular & Human Genetics, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Benjamin R. Arenkiel

    Corresponding author
    1. Program in Developmental Biology, Baylor College of Medicine, Houston, Texas, USA
    2. Department of Neuroscience, Baylor College of Medicine, Houston, Texas, USA
    3. Department of Molecular & Human Genetics, Baylor College of Medicine, Houston, Texas, USA
    4. Jan and Dan Duncan Neurological Research Institute, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    • Department of Molecular and Human Genetics, Jan and Dan Duncan Neurological Research Institute, Baylor College of Medicine, Houston, Texas 77030, USA

    Search for more papers by this author
    • Telephone: 713-798-1960; Fax: 832-825-1240

  • Author contributions: I.G.: experimental design, data collection, and manuscript writing; L.H.: data collection and data analysis; K.U.: data collection; B.A.: experimental conception and design, manuscript writing, and financial support.

  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    First published online in STEM CELLSEXPRESS July 31, 2012.


Transsynaptic circuit tracing using genetically modified rabies virus (RV) is an emerging technology for identifying synaptic connections between neurons. Complementing this methodology, it is now possible to assay the basic molecular and cellular properties of neuronal lineages derived from embryonic stem cells (ESCs) in vitro, and these properties are under intense investigation toward devising cell replacement therapies. Here, we report the generation of a novel mouse ESC (mESC) line that harbors the genetic elements to allow RV-mediated transsynaptic circuit tracing in ESC-derived neurons and their synaptic networks. To facilitate transsynaptic tracing, we have engineered a new reporter allele by introducing cDNA encoding tdTomato, the Rabies-G glycoprotein, and the avian TVA receptor into the ROSA26 locus by gene targeting. We demonstrate high-efficiency differentiation of these novel mESCs into functional neurons, show their capacity to synaptically connect with primary neuronal cultures as evidenced by immunohistochemistry and electrophysiological recordings, and show their ability to act as source cells for presynaptic tracing of neuronal networks in vitro and in vivo. Together, our data highlight the potential for using genetically engineered stem cells to investigate fundamental mechanisms of synapse and circuit formation with unambiguous identification of presynaptic inputs onto neuronal populations of interest. STEM Cells2012;30:2140–2151


The mammalian nervous system is built upon an elaborate collection of cells that form precise patterns of synaptic connectivity. Due to the complexity of neuronal tissue, cracking the code of brain wiring has remained a significant challenge. Limitations that have obscured our understanding of circuit formation stem from a lack of precise technology for marking, manipulating, and mapping patterns of connectivity between and among targeted neuronal subsets. To better understand how the mammalian brain forms and functions, we need to devise new approaches to elucidate patterns of synaptic connectivity. Genetically engineered embryonic stem cells (ESCs) harboring components that allow for direct visualization of connectivity between cells would aid this goal.

Historically, light microscopy has been the standard approach to dissect complex neuroanatomy. With advances in recombinant technologies, stem cell biology, in vitro culturing methods, fluorescent imaging, and electrophysiological analyses, it is now feasible to investigate the molecular mechanisms that underlie synaptogenesis, circuit formation, neuronal plasticity, and cell survival with impressive precision [1–3]. Additionally, manipulating the genome in ESCs now provides us with the ability to address fundamental aspects of gene or neuron function both in vitro and in vivo. Advances in stem cell technology raise the notion that transplantation of stem cell lineages into the intact brain could provide therapeutic avenues for treatment of some neurological disorders and/or neurodegenerative diseases [4–9]. Several groups have reported the successful transplantation of ESCs into the rodent nervous system [10–12]. However, implantation of undifferentiated stem cells has led to teratoma formation [11–15]. Consequently, significant effort has been placed on establishing in vitro differentiation methods of stem cells [14, 16–27]. It is now possible to obtain high numbers of cultured ESC-derived neurons in vitro following routine neuronal differentiation and isolation protocols. Following transplantation of neurons obtained from ESCs, teratomas form with significantly lower incidence compared to undifferentiated ESCs [11, 28, 29]. However, it remains unknown whether transplanted neurons form appropriate synaptic connections with established circuits in vivo. Not only is ectopic circuit formation a major concern but also the knowledge regarding the propensity for transplanted neurons to form proper connectivity in the intact brain is still lacking.

In an effort to elucidate patterns of synaptic connectivity between neurons, various transsynaptic viral tracing methods have been developed [30–34]. The recently described method of monosynaptic circuit tracing using genetically engineered rabies virus (RV) provides a means to genetically target discrete neuronal subsets for transsynaptic circuit labeling [1, 31, 33–35]. RV is a negative-strand RNA virus that shows neurotropism and is propagated transsynaptically in a retrograde manner [33–36]. RV preferentially spreads to neighboring cells via chemical synapses. Thus, RV infection occurs primarily between functionally connected neurons [32, 33]. Once infected with genetically engineered RV, neurons can survive for more than 10 days before any cytopathic changes are visualized in vivo or in vitro [33]. The RV glycoprotein G mediates binding of wild-type (wt) virus to neurons and is responsible for the transsynaptic spread. Deleting G from the RV genome and replacing it with an enhanced green fluorescent protein (EGFP) reporter (ΔG-EGFP) renders the mutant RV replication incompetent [35, 36]. Pseudotyping the ΔG-EGFP RV with the foreign coat protein EnvA makes it infectious only to cells that express the TVA receptor [34–37]. If cells targeted for pseudotyped ΔG-EGFP RV infection harbor a cDNA encoding wt G, the virus can use available G protein to package “live” particles capable of retrograde transsynaptic transfer. Once transferred to presynaptic cells that lack G, the ΔG-EGFP RV can no longer spread. To differentiate between the source cell and presynaptic inputs, a red fluorescent protein such as tdTomato can be included. Consequently, the source cell would be labeled both with tdTomato and green fluorescent protein (GFP), while monosynaptic inputs will remain GFP positive only. This technology allows for unambiguous identification of presynaptic inputs onto neuronal populations of interest.

Here, we report the generation of a novel mouse ESC (mESC) line expressing the genetic components necessary to perform monosynaptic tracing. We further show high-efficiency differentiation of these ESCs into functional neurons, demonstrate their capacity to form functional synaptic connections with primary neurons and brain slices in culture, and reveal their ability to act as source cells for presynaptic tracing of neuronal networks in vitro and in vivo. Finally, we provide evidence that genetically engineered stem cells can be used to investigate fundamental mechanisms of synapse and circuit formation both in vitro and in vivo, which may ultimately lead to valuable insight toward designing new methods for cell replacement therapy.


Expression Plasmid Construction

To generate the monosynaptic tracing allele, cDNA encoding Rabies-G was excised from pHCMV-Rabies-G [38] and cloned into a custom expression construct harboring an Ef1α promoter, tdTomato, a 3′ WPRE element, and an hGH polyA sequence. A P2A peptide cleavage sequence [39] was introduced between tdTomato and Rabies-G by annealed oligos. For positive selection of ESC clones, bicistronic expression of a Neomycin resistance gene (pIRES-Neo3, Clontech, Mountain View, CA) was inserted downstream of Rabies-G. For tricistronic expression of the TVA receptor, the TVA cDNA from pCMMP-TVA800 [40] was PCR amplified and cloned downstream of an IRES2 element (Clontech) for insertion 3′ to NeoR, generating pEf1α-tdTomato-Rabies-G-IRES-Neo-IRES-TVA (pTomRITVA).

Generation of ROSA26-tomRITVA mESCs

pTomRITVA was shuttled into the pROSA-acceptor targeting plasmid using PacI and AscI [41] to generate the ROSA26-tomRITVA targeting vector. This targeting construct was linearized and electroporated into 129/SvJ ESCs. Following G418 selection, clones were picked, propagated, DNA isolated, and PCR genotyping was performed across the targeting vector junction.

Genotyping was performed using Herculase II Fusion DNA polymerase (Agilent Technologies, Santa Clara, CA) in 50 μl reactions. Primer sequences were as follows: wt/tgt forward primer -GCCTAAAGAAGAGGCTGTGC (112 bp upstream of 5′ ROSA homologous arm), wt reverse primer -GGAAGTCTTGTCCCTCCAAT, tgt reverse primer -ACCTGTGTTCTGGCGG (318 bp downstream of the EF1α promoter start site). Wt transcript size: 1,302 bp; tgt size: 1,530 bp. Two positively screened clones and the parental cell line were karyotyped via chromosomal metaphase spread G-banding to identify potential aneuploidies.

mESC Culture

mESCs were grown on gelatinized 10 cm tissue culture plates with leukemia-inhibitory factor (LIF)-containing Glasgow Minimum Essential Medium (GMEM) medium (Lonza Group Ltd., Switzerland) supplemented with 10% fetal bovine serum (FBS), 1% 10 mM nonessential amino acids, 1% 100 mM sodium pyruvate, 1% 100× Penicillin/Streptomycin, 560 μl β-Mercaptoethanol (55 mM in phosphate-buffered saline (PBS), Sigma), 1% 200 mM L-glutamine) in the absence of feeder layers.

Neuronal Differentiation of Targeted mESCs

The ESC differentiation assay was modified from Bibel et al. [24]. Briefly, ESCs were grown on 10 cm gelatinized tissue culture plates in ESC medium containing LIF. Cells were trypsinized with 0.05% trypsin/EDTA for 5 minutes, and 5 ml ESC medium without LIF was added to inactivate the trypsin. Cells were pelleted and resuspended in 15 ml of ESC medium without LIF and transferred to 10 cm bacteriological Petri dishes (Greiner #633102) for neurosphere culture in suspension. After 2 days, neurospheres were transferred to 50 ml Falcon tubes, allowed to settle, and the supernatant was replaced with 15 ml ESC medium without LIF. Next, the cell suspension was added to a new bacteriological Petri dish. After 4 days, culture medium was changed as above with the addition of retinoic acid (Sigma; 1:1,000 from 5 mM stock solution in DMSO). Cell aggregates were allowed to grow for 6 days, and the medium was changed once more to ESC medium containing retinoic acid but lacking LIF. Two days later, cell aggregates were washed twice with PBS and trypsinized for 5 minutes in 0.05% trypsin/EDTA. Ten milliliters of ESC medium without LIF was added to inactivate the trypsin, and the cell suspension was pelleted and resuspended in NB/B27 medium (Neurobasal medium [GIBCO], 2% B27, 5% fetal bovine serum, 1% 200 mM L-glutamine, 1 μg/ml Gentamycin). Approximately 5 × 105 cells were plated on poly(D-lysine) (PDL, Sigma) coated glass coverslips in 12-well plates in the presence of 2 ml NB/B27 medium. Within 5–7 days, elaborate neuronal morphology was observed.

Confocal Imaging and Immunohistochemistry

ESC-derived neurons grown on glass coverslips were rinsed with PBS, fixed in 4% paraformaldehyde (PFA)/PBS for 15 minutes at RT, rinsed in PBS, and then incubated in blocking solution (10% normal goat serum, 0.3% Triton X-100 in PBS, pH 7.35) at 4°C for 2 hours. The following primary antibodies were used: anti-NeuN (1:700; Millipore MAB377, Billerica, MA), anti-β-III-Tubulin (1:500; Millipore MAB1637, Billerica, MA), anti-TUJ1 (1:500; Covance MMS-435P, Princeton, NJ), anti-Synapsin (1:50; Hybridoma Bank, Iowa City, IA), anti-Gephyrin (1:2,000; Synaptic Systems 147011, Goettingen, Germany), and anti-Bassoon (1:2,000; Synaptic Systems 141004, Goettingen, Germany). Antibodies were diluted in blocking solution and applied overnight at 4°C. The next day, coverslips were washed 4 × 10 minutes in PBS with 0.1% Triton X-100. Secondary Alexa-488 anti-mouse IgG and Alexa-488 anti-rabbit (Invitrogen, Carlsbad, CA) were then added to a final dilution of 1:300 and incubated for 1 hour at room temperature. Coverslips were washed 4 × 15 minutes each and mounted with Vectashield mounting medium (Vector Laboratories, Burlingame, CA). Microscopy was performed using a Leica TCS SPE confocal microscope under a 10× or 20× objective. Neuronal differentiation efficiency was quantified per culture per 400 × 400 μm2 field of view and is reported as percentage ± SEM. Statistical significance of neuronal differentiation efficiency between parental and ROSA-tomRITVA ESCs was determined by a Student's t test and one-way analysis of variance (ANOVA) using SPSS analysis software.

Primary Neuronal and Slice Coculture

Primary neurons were obtained from E18 C57Bl/6 wt embryos. Cortical tissue was dissected into cold dissection medium (DM) (Hanks' balanced saline solution [HBSS] [Invitrogen], 10 mM HEPES, 0.3% bovine serum albumin (BSA), 12 mM MgSO4, pH 7.4), centrifuged 2 minutes at 1,100 rpm, supernatant was aspirated, and 8 ml prewarmed digestion solution was added (4.2 mM NaHCO3, 25 mM HEPES, 137 mM NaCl, 5 mM KCl, 7 mM Na2HPO4, pH 7.4, 4 ml 2.5% trypsin, and 2 mg DNase [Sigma]). The solution was incubated 8 minutes at 37°C, centrifuged as above, supernatant removed, and 3 ml trypsin inhibitor solution (70 mg in 3 ml PBS, Sigma) added and incubated for 2 minutes at 37°C. The tissue was again centrifuged, supernatant removed, the pellet was washed twice with ice-cold DM, and resuspended in fresh DM. Cells were counted and seeded at a density of 2 × 105 primary neurons onto PDL-coated coverslips containing 0.5% ROSA-tomRITVA expressing neurons in 2 ml NB/B27 medium. Cells were grown for 5–7 days before infection with low titer SADΔG-EGFP-RV. SADΔG-EGFP-RV was produced as previously described [35]. Infected cells were allowed to grow for 2, 4, or 6 days postinfection and imaged for presynaptic jumping of SADΔG-EGFP-RV using confocal microscopy.

Cortical slice cultures were obtained from P4 C57Bl/6 wt pups. Brain tissue was sliced using a Compresstome (Precisionary Instruments, NC) at 300-μm thickness and collected in modified Gey's Balanced Salt Solution (Sigma) (0.5% D-glucose and 1% 3 M KCl). Slices were placed onto 30 mm Millicell-CM culture membranes (Millipore), transferred to six-well plates filled with 1.2 ml culture medium (25% horse serum [Gibco, Invitrogen, Carlsbad, CA], 25% HBSS, and 0.5% D-glucose in Opti-Minimum Essential Medium (MEM) [Gibco]), and allowed to recover for 1 day. Medium was changed to 1.2 ml neurobasal medium (NB)/B27 before seeding with 1 × 104ROSA-tomRITVA ESC-derived neurons. Slices were infected 3–5 days later with SADΔG-EGFP-RV. Images were taken 2, 4, and 6 days postinfection.

Electrophysiological Recordings

ESC-derived neurons and primary neuronal cocultures were placed in a RT chamber mounted on an Olympus upright microscope (BX50WI, Center Valley, PA) and perfused with buffered (5% CO2 and 95% O2) artificial cerebrospinal fluid (ACSF) (in mM: 122 NaCl, 3 KCl, 1.2 NaH2PO4, 26 NaHCO3, 20 glucose, 2 CaCl2, 1 MgCl2, 305–310 mOsm, pH 7.3). Cells and slices were visualized under differential interference contrast imaging. Recording pipettes were pulled with tip resistance between 4 and 7 MΩ. Data were obtained via a Multiclamp 700B amplifier, low-pass Bessel filtered at 4 kHz, and digitized on computer disk (Clampex, Axon Instruments, Inverurie, Scotland). Whole-cell patch-clamp recordings were obtained from visually targeted GFP+/tdTomato+ cells or GFP+/tdTomato− cells using pipettes filled with internal solution (in mM: 120 K-gluconate, 5 KCl, 2 MgCl2, 0.05 EGTA, 10 HEPES, 2 Mg-ATP, 0.4 Mg-GTP, 10 creatine phosphate, 290–300 mOsm, pH 7.3). For paired cell recordings, 40 pA current was injected into the GFP+/tdTomato− cell (presynaptic) for 1 second to initiate action potentials, and postsynaptic recordings (GFP+/tdTomato+ cell) were performed in voltage clamp mode at −80 mV. Blockage of fast synaptic transmission was performed via bath application of 20 μM CNQX, 20 μM APV, and 50 μM Bicuculline. Presynaptic neurons were stimulated and recordings were made from the postsynaptic neuron before drug application, 5 minutes after drug application, and 10 minutes after wash out. The traces were averaged over 15 sweeps from the same neuron.

ESC and ESC-Derived Neuron Transplantation

For postnatal ESC transplants, P1 C57Bl/6 wt pups were injected into the lateral ventricle with 1 × 105 cells with 0.01% Fast green (Sigma) using a 30 g Hamilton syringe. Thirty days later, animals were anesthetized with a ketamine/dormatore mixture, placed in a stereotaxic injection frame, and 10 × 50 nl of SADΔG-EGFP RV (6 × 103 viral particles per microliter) was injected at a rate of 23 nl/second using a Drummond Nanoject II (Broomall, PA) at 20 second intervals into the core of the olfactory bulb, 800 μm from the dorsal surface. Tissues were harvested for transsynaptic analysis 7 days postinjection (dpi). For adult ESC-derived neuronal transplants, 3–6-week-old C57Bl/6 mice were anesthetized as above, placed in a stereotaxic frame, and 1 × 105 cells were injected into different parts of the brain using a Drummond Nanoject II (coordinates in mm from bregma and dorsal surface: cortex ML 1.12 mm, AP 2.46, DV 1.7, striatum ML 1.45, AP 1.1, DV 3.31, SVZ ML 0.65, AP 1.1, and DV 3.69). Thirty days later, SADΔG-EGFP RV was injected into the original injection site as described above.


A New ESC Line for Transsynaptic Viral Tracing In Vitro and In Vivo

Homologous recombination and ESC technology afford the routine ability to generate genetically engineered pluripotent cell lines that can be investigated at the cellular level in vitro, in the context of intact tissue, or in living mouse models. Understanding gene, cell, and tissue function in the mammalian nervous system where delineating precise patterns of synaptic connectivity is required to understand how the brain forms and functions benefits from multifaceted experimental manipulations. Toward this goal, we have generated a novel ESC line that harbors the genetic elements necessary to perform transsynaptic viral circuit tracing using the deletion-mutant RV strain SADΔG-EGFP RV [36]. Through gene targeting [42, 43], we generated ROSA-tomRITVA ESCs, differentiated these cells into neurons, and performed in vitro, ex vivo, and in vivo transsynaptic circuit tracing (Fig. 1A). The construct targeted to the ROSA26 locus included an EF1α promoter for high levels of tdTomato reporter expression, followed by a P2A element fused to the Rabies-G sequence. Neomycin resistance and TVA expression were linked using internal ribosomal entry sites (IRES). We included a Diphtheria toxin element outside vector homology to allow for negative selection (Fig. 1B). After linearizing the targeting vector, the construct was electroporated into wt 129SvJ ESCs by standard methods. Following positive and negative selection in vitro, we first visually identified potential clones using tdTomato expression and infectivity by SADΔG-EGFP RV. Out of 96 fluorescent clones, we picked 20 with the brightest tdTomato reporter and GFP expression following infection with pseudotyped virus (data not shown). We further screened the reporter-positive clones by PCR genotyping to validate correct targeting at the ROSA26 locus (Fig. 1B, 1C). As a control, we used wt 129SvJ ESC DNA and demonstrated the absence of the mutant allele. The correctly targeted ESC clones showed two bands, one at 1,302 bp (wt) and the other at 1,530 bp (tgt) (Fig. 1C). Following identification of positive ESC clones, visual inspection showed high-efficiency tdTomato reporter expression as well as the ESC marker Oct4 (Fig. 1D). To verify that the newly generated ROSA-tomRITVA ESCs did not show chromosomal abnormalities, G-banding was performed both on the parental cell line and two ROSA-tomRITVA ESC clones. Cytogenetic evaluation of the parental wild-type mESCs and one of the ROSA-tomRITVA ESC clones revealed a normal male diploid karyotype in all analyzed metaphases (Supporting Information Fig. S1). The other ROSA-tomRITVA ESC clone showed normal male diploid karyotype in 17/20 examined metaphases. We proceeded with the 100% normal ROSA-tomRITVA ESC clone for all experimentation. Thus, we have successfully generated and identified a new ESC line harboring the monosynaptic tracing components genetically targeted to the ROSA26 locus.

Figure 1.

Transsynaptic tracing in vitro and in vivo using ROSA-tomRITVA ESCs. (A): Scheme for generating ROSA-tomRITVA ESCs for transsynaptic tracing. (B): ROSA-tomRITVA targeting vector, highlighting primer pairs for genotyping. (C): PCR genotyping of ROSA-tomRITVA ESCs. The wild-type allele (wt) is 1,302 bp, whereas the ROSA-tomRITVA knockin allele (tgt) is 1,530 bp. (D): Undifferentiated ROSA-tomRITVA ESCs immunostained for OCT4. Far right panel represents inset shown in panel to the left. Scale bars left to right = 100 μm and 30 μm. Abbreviations: EGFP, enhanced green fluorescent protein; ESCs, embryonic stem cells.

Genetically Engineered ESCs Show Neuronal Differentiation In Vitro

Transient transfection, overexpression, and knockdown experiments in cultured cell lines have provided significant insight into neurodevelopmental pathways. Direct modification of the genome, however, provides stable and reproducible molecular expression and/or cellular phenotypes. Genetic manipulations of ESCs now bridge these two approaches; ESCs are readily manipulated genetically and can routinely be differentiated into desired cell and tissue types in vitro.

Following generation of ROSA-tomRITVA ESCs, we differentiated them in vitro into neurons (see Materials and Methods). ROSA-tomRITVA ESC-derived neurons showed neuronal morphology with extensive neurite outgrowth (Fig. 2). To verify neuronal differentiation, we assayed ESC lineages for neuronal marker expression via immunohistochemical analysis. Fourteen days post plating, ROSA-tomRITVA ESC-derived neurons expressed molecular markers of neuronal differentiation, including NeuN (Fig. 2B), β-III-Tubulin (Fig. 2C), and TUJ1 (Fig. 2D). In addition to general markers of neuronal differentiation, ESC-derived neurons also showed colocalization and punctate expression of synaptic proteins (Supporting Information Fig. S2). Notably, we did not observe statistically significant differences in the differentiation efficiency between the parental and genetically targeted ESC lines using Student's t test and ANOVA. Differentiated wt ESCs (81% ± 1.6%) expressed NeuN, whereas NeuN was detected in 82.3% ± 1% in ROSA-tomRITVA ESC-derived neurons (n = 4 coverslips each: p >.05 for all groups). Additionally, we also noted the presence of cells with glial morphologies (arrows, Fig. 2C, 2D). Cells that did not express NeuN were glial fibrillary acidic protein positive, suggesting differentiation of a minor fraction of the differentiated cells into an astrocyte lineage (21.5% ± 0.5%; n = 6 coverslips). Furthermore, no delay in maturation was observed as evidenced by quantification of β-III-Tubulin and TUJ1-positive cells (β-III-Tubulin: 77.1% ± 1.1% wt vs. 78.2% ± 1.2% ROSA-tomRITVA, n = 5 coverslips each; p >.05; TUJ1: 86.2% ± 0.6% wt vs. 78.7% ± 1.5% ROSA-tomRITVA, n = 5 coverslips each; p >.05). Worth noting, the longer we allowed the mESC-derived neurons to grow in culture, neuronal marker and synaptic protein expression became further enriched (data not shown), demonstrating that ESC-derived neurons mature with increased time in culture. Thus, ESCs harboring ROSA-tomRITVA show equal differentiation efficiency and maturation as the parental wt ESC line. Together, these data show that this new line of ROSA-tomRITVA ESCs can be efficiently differentiated into neurons in vitro by established protocols and with maturation show robust expression of molecular markers of neuronal differentiation and synapse formation.

Figure 2.

Neuronal differentiation of ROSA-tomRITVA ESCs in vitro. (A): ROSA-tomRITVA ESCs are grown as neurospheres with retinoic acid to induce neuronal differentiation in vitro (scale bars left to right = 60 and 100 μm). (B): Neuronally differentiated ESCs show expression of NeuN (scale bar = 60 μm), (C) β-III-Tubulin (scale bar = 60 μm), and (D) Tuj1 (scale bar = 60 μm). Arrows in (C) and (D) point to glial cells. Abbreviation: ESCs, embryonic stem cells.

Genetically Engineered ESC-Derived Neurons Show Synaptic Coupling and Transsynaptic Viral Spread In Vitro

Upon demonstrating that ROSA-tomRITVA ESCs can be efficiently differentiated into neurons in vitro, we next sought to determine whether the mESC-derived neurons expressed the necessary elements required for transsynaptic tracing, wire up in vitro as functional synaptic networks, and show properties of mature neurons. Toward this, we targeted ROSA-tomRITVA ESC-derived neurons for transsynaptic viral circuit tracing using an EnvA pseudotyped deletion-mutant strain of RV (SADΔG-EGFP RV), which does not infect wild-type neurons [35, 36]. We infected both wild-type and ROSA-tomRITVA ESC-derived neuronal cultures that underwent identical differentiation protocols both with a normal G-typed (nonpseudotyped) and pseudotyped SADΔG-EGFP RV 5–7 days postplating, which is sufficient time for extensive neurite formation and axonal outgrowth in vitro. Three days postRV infection, nearly all wild-type and ROSA-tomRITVA ESC-derived neurons became infected with the G-typed virus and robustly expressed GFP, demonstrating nonselective neuronal binding and uptake of the normal G-typed virus (Fig. 3A, 3B). Concurrently, we performed the same experiment using EnvA pseudotyped SADΔG-EGFP RV to determine the propensity for differential infectivity of ROSA-tomRITVA ESC-derived neurons. Using EnvA pseudotyped SADΔG-EGFP RV, we observed highly selective and targeted infection to ROSA-tomRITVA clones but not to wild-type ESC-derived neurons (Fig. 3C, 3D). We next aimed to determine whether the mESC-derived neurons showed membrane and ion channel properties of mature neurons, and thus were capable of firing action potentials. The mESC-derived neurons were infected with a low titer SADΔG-EGFP RV (100 μl of 2.4 × 104 infectious particles per milliliter) for subsaturating levels of infection, and whole-cell patch-clamp recordings were performed on tdTomato and EGFP double-positive neurons (Fig. 4A, 4B). In current clamp mode, ESC-derived neurons fired action potentials at high frequency to controlled current injections. Neurons reproducibly fired action potentials upon reaching threshold (−40 mV) following a series of stepwise current injections (Fig. 4C). These data show that neuronally differentiated ROSA-tomRITVA ESCs express the components necessary for transsynaptic tracing, are highly susceptible to EnvA SADΔG-EGFP RV infection, and that reporter expression does not interfere with in vitro neuronal maturation as shown by targeted whole-cell recordings.

Figure 3.

ROSA-tomRITVA neurons are selectively infected by SADΔG-EGFP RV. (A, B): Both wt neurons (marked by DAPI) and ROSA-tomRITVA neurons show infectivity by G-EGFP RV. (C): WT neurons are not infected by the pseudotyped SADΔG-EGFP RV. (D): ROSA-tomRITVA neurons infected with SADΔG-EGFP RV (scale bars = 70 μm). Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; EGFP, enhanced green fluorescent protein; ESCs, embryonic stem cells; RV, rabies virus.

Figure 4.

ROSA-tomRITVA ESC-derived neurons show neuronal properties in vitro. (A): Fluorescent merge of recorded ROSA-tomRITVA neuron infected with GFP-expressing RV (scale bar = 20 μm). (B): Differential interference contrast image of recorded cell, arrow (scale bar = 20 μm), and (C) representative responses to ramped current injections. Abbreviations: EGFP, enhanced green fluorescent protein; RV, rabies virus.

Genetically Engineered ESC-Derived Neurons Form Functional Synapses with Primary Neurons and Slice Explants In Vitro

A hallmark of neuronal differentiation is the formation of functional synapses. We next sought to determine whether the mESC-derived neurons functionally connect with primary neurons cocultured in vitro, and whether these artificially generated neuronal networks are capable of transsynaptic circuit tracing and synaptic transmission. To test this, we cocultured ROSA-tomRITVA neurons at low density (0.5% of total cells) with excess wild-type primary cortical neurons and waited 7 days before infection with SADΔG-EGFP RV. Three days postinfection, transsynaptic tracing was evident by the presence of few doubly labeled source cells and high numbers of GFP+ presynaptic cells (Fig. 5A). Next, to determine whether transsynaptically labeled neurons showed functional synaptic transmission, we performed paired whole-cell recordings of infected source cells (tdTomato+ and GFP+ ROSA-tomRITVA neurons) and their presynaptic partners (GFP+ only neurons) 3 days post-RV infection (Fig. 5B, 5C). This time point is ideal for our studies since transsynaptic labeling is the primary means for RV transfer, whereas nonspecific particle shedding into the neuronal culture and/or RV-mediated labeling from cell death or glial-mediated synaptic remodeling processes in vitro is minimized. Upon establishing whole-cell configuration in presynaptic and postsynaptic pairs, current was injected into the presynaptic (GFP+) neurons to initiate depolarizing events and drive action potentials (Fig. 5B). Postsynaptic responses were simultaneously measured in double-positive (tdTomato+/GFP+) source cells. In multiple pairs (n = 7), we found that cells were synaptically coupled (Fig. 5D), and that the GFP+ cell was indeed presynaptic to our source cell (Fig. 5B–5D). To rule out the possibility of other nonsynaptic electrical connections or field effects, we established paired whole-cell recordings and bath-applied both GABA and glutamate receptor blockers (CNQX, APV, and Bicuculline). Drug application blocked electrophysiological responses between RV-labeled pairs, consistent with transsynaptic transfer of RV particles and synaptic connectivity between recorded cell pairs. Synaptic responses were recovered following washout (Supporting Information Fig. S3).

Figure 5.

ROSA-tomRITVA neurons synaptically couple with primary neurons in vitro. (A): Primary neuronal coculture shows transsynaptic spread of SAD-ΔG-EGFP RV. Arrows, postsynaptic neurons; arrowheads, presynaptic neurons (scale bar = 20 μm). (B): Fluorescent images of synaptic pairs targeted for paired whole-cell recordings (scale bar = 15 μm). (C): Differential interference contrast image of (B) (scale bar = 15 μm). (D): Representative trace showing synaptic coupling. Top, changes in membrane potential with presynaptic current injection; bottom, postsynaptic current responses. Abbreviations: EGFP, enhanced green fluorescent protein; RV, rabies virus.

To determine whether ROSA-tomRITVA ESC-derived neurons could form functional connections with neurons in more intact nervous tissue, we next seeded ROSA-tomRITVA neurons onto cortical slice explants that were made from wild-type postnatal day 4 (P4) pups. Three to five days after seeding, the coculture explants were infected with SADΔG-EGFP RV. Three days postinfection, we observed extensive GFP reporter expression in presynaptic neurons as well as a minor fraction of glial cells (Fig. 6A, 6B). Together, these data demonstrate that neurons derived from ROSA-tomRITVA ESCs form functional synaptic connections with neuronal cell types in vitro. These connections can be revealed by transsynaptic reporter expression, and electrophysiological recordings validate synaptic connectivity between RV-labeled neuronal pairs.

Figure 6.

ROSA-tomRITVA embryonic stem cell-derived neurons form synapses with brain slice explants. (A): Slices seeded with ROSA-tomRITVA neurons and infected with SADΔG-EGFP RV. The square shows the region of the seeded brain slice that was imaged. (B): Transplanted neurons show transsynaptic viral spread in intact brain tissue. Arrows, postsynaptic ROSA-tomRITVA neurons; arrowheads, presynaptic input neurons; hollow arrowheads, labeled glial cells (scale bar = 150 μm). Abbreviations: EGFP, enhanced green fluorescent protein; RV, rabies virus.

Genetically Engineered ESCs Allow Transsynaptic Circuit Tracing In Vivo

Numerous studies have reported the introduction, differentiation, and survival of ESCs into living rodent nervous tissues [11–13, 28, 44]. Although evidence of circuit integration has been substantiated by the expression of fluorescent reporters [44–46], recovery of pathological conditions [11, 45, 47–50], detection by magnetic resonance imaging or positron emission tomography signals [45, 51, 52], and behavior [17, 48, 50, 53], direct visualization of synaptic connectivity of transplanted cells in these preparations has been lacking. Having shown that the ROSA-tomRITVA mESC-derived neurons form functional synaptic connections with primary neurons and slice explants in vitro (Figs. 5 and 6), we next wanted to determine whether these cells could be used for transsynaptic tracing in vivo. Exploiting the feature of continued neurogenesis in the murine olfactory system [54–60], we set out to determine whether this cell line would appropriately differentiate, migrate, and integrate into olfactory bulb circuits in vivo. For this, we transplanted ROSA-tomRITVA ESCs (2 × 104 cells) into the lateral ventricles of postnatal C57Bl/6 mice, followed by a 30-day period of maturation and subsequent introduction of SADΔG-GFP RV into the olfactory bulb. Points of infection and transsynaptic labeling were analyzed 7 days later (Fig. 7A). We observed numerous presynaptic partners as evidenced by extensive SADΔG-GFP RV spread throughout the olfactory bulb. Moreover, we determined that transplanted ESCs differentiated, became infected by SADΔG-GFP RV, and subsequently formed synaptic connections with local olfactory bulb cell types previously described to form connections with newborn neurons [31] (Fig. 7B). Interestingly, however, we also observed some GFP labeling in cells with glial morphology. Glial cell labeling via RV transfer has been documented [31], but it remains to be determined whether this occurs through glial cell-mediated synaptic remodeling or actual synaptic connectivity between neurons and glial subtypes. Together these data show that ROSA-tomRITVA ESCs are capable of neuronal differentiation and synapse formation following in vivo transplantation.

Figure 7.

ROSA-tomRITVA embryonic stem cells (ESCs) show transsynaptic circuit tracing in vivo. (A): ESC-derived neurons migrate to the olfactory bulb and show transsynaptic tracing of host circuits. Scale bar = 500 μm. Square is the magnified region shown in (B). (B): High magnification of highlighted region shown in (A) (scale bar = 100 μm). (C): ROSA-tomRITVA neurons transplanted into prefrontal cortex of adult mice, followed by infection with SADΔG-EGFP RV. Arrows in (B) and (C) point to postsynaptic source cell, arrowheads point to presynaptic host neurons, hollow arrowheads point to labeled glial cells (scale bar = 20 μm). Abbreviations: EGFP, enhanced green fluorescent protein; EPL, external plexiform layer; GCL, granule cell layer; GL, glomerular layer; RV, rabies virus.

In agreement with previous reports [11–14, 28], one major limitation we observed following transplantation of undifferentiated ESCs was teratoma formation. Injection of undifferentiated ESCs resulted in a 20% (four of 20 transplanted animals) incidence of teratomas. To decrease the incidence of teratoma formation, and to increase the chances of cell integration into other parts of the adult brain, we next performed transplantations of in vitro-differentiated ROSA-tomRITVA neurons into the brain parenchyma. After differentiation, we transplanted ROSA-tomRITVA neurons at low density (2 × 104 cells) into adult mouse somatosensory cortex. Following transplantation, we waited 30 days and injected SADΔG-EGFP RV into the transplantation sites. Seven days later, we observed successful integration of ROSA-tomRITVA expressing cells into the pre-existing host brain circuits. Seventeen of 20 transplanted animals contained ROSA-tomRITVA cells 1 month later. Transsynaptic tracing was evident in cortical regions where numerous cells were identified as being presynaptic to our source cells (Fig. 7C). Interestingly, when implanted into the subventricular zone (SVZ), differentiated ROSA-tomRITVA neurons incorporated into the walls of the lateral ventricle and SVZ but did not give rise to known adult-born SVZ-derived neuronal lineages. This may be due to the postmitotic state of ESC-derived neurons, and therefore integration into the olfactory bulb was rarely observed following SVZ transplantation. Taken together, we were able to show that ROSA-tomRITVA ESC-derived neurons functionally integrate into pre-existing host circuits in vivo and that these circuits are susceptible to transsynaptic viral circuit tracing using genetically modified RV.


Delineating precise patterns of synaptic connectivity in the mammalian nervous system is essential if we are to ultimately gain mechanistic insight into normal neural development, neurological disease, or potential avenues for brain repair. Numerous experimental approaches ranging from imaging and electrophysiology, to genetic engineering and stem cell biology, have been broadly implemented toward our working knowledge in these areas. Unfortunately, progress in this field has been somewhat tempered due to the innate complexity of brain tissue. Thus, new experimental approaches and methodologies for investigating patterns of synaptic connectivity are both useful and necessary.

We have generated a novel ESC line with all the necessary elements for transsynaptic viral tracing in nervous tissue using genetically modified RV. Through immunohistochemistry, transsynaptic fluorescent reporter analysis, and electrophysiological recordings, we have demonstrated the efficient differentiation of these cells into functional neurons in vitro and have shown that that they can wire up effectively with host neuronal circuits in vitro and in vivo. Transsynaptic tracing technology using genetically engineered RV is a powerful tool given its general utility to genetically identify neurons that are connected via chemical synapses [32–37]. In contrast to most lipophilic dye-based synaptic tracers, lectin-based conjugates, and other viral technology, RV spread occurs exclusively in a retrograde manner, is not thought to cross gap junctions, and requires the presence of intact presynaptic structures for neuronal infection and transsynaptic transfer [31–36]. However, it is important to consider other mechanisms of viral labeling and spread that can occur in vitro. Unlike the specificity of synapse formation within intact nervous tissue, in vitro preparations facilitate numerous “unnatural” cellular interactions. For example, differentiating cells and neurons in culture can form transient synaptic connections through time. This poses the potential for nonspecific viral transfer and thus labeling of cell types that do not show long-lived functional synaptic connectivity. Moreover, in vitro culture environments do not foster the longevity of cells in vivo, and thus cultured neuronal cells can die off earlier than in vivo. In this scenario, infected neurons might shed RV into the culture medium upon death, and these viral particles have the capacity to be taken up by nearby neurons without synaptic connectivity. Importantly, these nonspecific labeling phenomena can be minimized when in vitro RV infection times are kept as short as possible. For this reason, we performed all tracing experiments within 3–7 days after infection. During this time, infected cells remain alive, healthy, and function as expected for in vitro growth [33].

Neuronal differentiation of modified ESCs allows the unique opportunity to investigate basic molecular and cell biological mechanisms that underlie synapse and circuit formation in vitro but also affords the ability to efficiently crossover to in vivo experimentation via cell transplantation or generation of mutant mice. A beneficial facet of our approach that complements pre-existing tracing methods is that circuit integration can be studied at any stage of neuronal differentiation and in any part of the nervous system. Having a genetically targeted ESC line that harbors the elements for transsynaptic viral tracing allows for differentiation of these cells into virtually any cell type and facilitates the identification of the various types of presynaptic inputs in diverse regions of the nervous system, both healthy and diseased. Experimentation becomes feasible in which ROSA-tomRITVA ESCs are differentiated into specific neuronal subtypes and subsequently combined with either neurons or tissues harboring known genetic lesions to address cell nonautonomous mechanisms of synapse formation, or alternatively, introduce compound mutations or shRNA expression into ROSA-tomRITVA cells to address cell autonomous gene function. These approaches could also be extended to in vivo applications, where transplanted ESCs could interact with intact brain tissue. Alternatively, the generation of novel mouse lines from these ESCs could be powerful in delineating patterns of connectivity in the context of normal development or disease. One application that could prove to be extremely useful would be to generate mosaic mice through morula aggregation, in which select subsets of neurons could be targeted for transsynaptic tracing in a semistochastic manner. This could also be combined with additional genetic alterations to tease out mechanisms of synapse wiring.

Of course, we must also consider the limitations. ESC technology is in its infancy. Although we have learned a great deal about driving neural differentiation in vitro, we still have much to learn about the generation and maintenance of certain neuronal subtypes. It will be imperative for future work to more precisely characterize and describe methodology to enrich and isolate neuronal subtypes of given neurotransmitter or receptor expression properties. Additionally, current applications for ESC-derived neuronal transplants remain prominent in rodents, and in particular inbred strains of mice. Teratoma formation and immune responses have stifled applications in other mammalian systems. Induced pluripotent stem (iPS) cells offer promise in this area. It is conceivable that similar genetic expression systems for transsynaptic analysis could be introduced into tailored iPS cell lines, and that these could subsequently be used to investigate synaptic mechanisms in other tissues or model systems.

Considering the widespread use of ESCs to generate mouse models of brain development and dysfunction, the broad array of available alleles expressed in the nervous system, and advancements in iPS cell technology, in vitro experimentation that directly transfers from the culture dish to in vivo application is thus becoming much more feasible. Permutations of the experimentation we performed here using the described ROSA-tomRITVA ESCs hold promise to provide future insight into synaptic wiring mechanisms at work in the mammalian brain.


With the ultimate goal of elucidating the mechanisms that underlie synaptogenesis, circuit formation, circuit integration, and neuronal survival, we have developed a novel ESC line harboring all the necessary elements for transsynaptic circuit tracing using genetically modified RV. We have shown that these cells efficiently undergo differentiation into functional neurons and can be used for transsynaptic tracing both in vitro and in vivo. With this novel technology, it will be feasible to identify precise presynaptic inputs in various healthy and diseased brain regions. This knowledge will allow us to better understand why certain neurons form correct synapses while others fail to properly integrate. In the long-term, such knowledge could potentially lead to novel therapeutic methods in neurodegenerative diseases.


The authors indicate no potential conflicts of interest.


This work was supported by the McNair Medical Institute and NINDS Grants R00NS064171 and R01NS078294 to B.R.A. We would like to thank Michael Ehlers and Irina Lebedeva for input and assistance with previous versions of this experimentation, and Mirjana Maletic-Savatic, Ben Deneen, and Fan Wang for critical review of this manuscript.