Human Pluripotent Stem-Cell-Derived Cortical Neurons Integrate Functionally into the Lesioned Adult Murine Visual Cortex in an Area-Specific Way

Human PSC-Derived Cortical Neurons Integrate into the Adult Lesioned Murine Cortex following Transplantation

To determine whether transplanted human PSC-derived cortical neurons can integrate in the lesioned adult mouse cortex, we used focal cortical lesioning mediated by injection of neurotoxic ibotenic acid, as used for the transplantation of mouse ESC-derived cortical neurons (Michelsen et al., 2015) (Figure 1A). Three days later, we transplanted cortical-like cells differentiated from GFP-expressing human ESCs (Espuny-Camacho et al., 2013) into the same lesioned area. This in vitro corticogenesis model generates human neurons from all different cortical layers in a temporal manner. We showed that such human cortical cells transplanted into immunodeficient newborn mice generate neurons with a visual-like areal identity at early stages that efficiently integrate into mouse neural networks in vivo (Espuny-Camacho et al., 2013). We, therefore, first focused on transplantation into the lesioned visual cortex, as performed previously for mouse ESCs (Michelsen et al., 2015). Following 3 weeks of in vitro differentiation, GFP-expressing human ESC-derived cortical cells were dissociated and transplanted in the lesioned visual cortex of immunodeficient adult mice. The presence of the human transplants was assessed by GFP immunostainings at 1–2 and 6 months post-transplantation (MPT) and found in 11 of 21 (52%) animals that underwent transplantation. As expected, transplants were located in the cortex or in the underlying white matter (Figures 1B, 2A, S1A, S1B, S2E, and S3A). We then compared the transplants at 2 and 6 MPT for maturation and integration.

At 2 MPT, transplants were still rather immature, with the presence of neural progenitors, displayed in rosette-like structures and expressing markers of proliferation (Figures S1A, S1C, and S1E), while the expression of mature neuronal markers such as VGlut1 and NeuN remained low (Figures S1G and S1I). In contrast, at 6 MPT, the transplants, found in 58% of animals, presented a more mature pattern, with far fewer or no rosette-like structures and with few proliferative cells or progenitors (Figures S1B, S1D, and S1F). GFP⁺-transplanted cells were positive for the human specific markers HuNuclei and human neural cell adhesion molecule (NCAM), and broadly expressed the telencephalic marker FoxG1, the neuronal marker beta 3 class III tubulin (Figures 1C–1E), and mature neuronal markers such as VGlut1, NeuN, and MAP2 (Figures S1H, S1J, and 1B). Moreover, these neurons expressed a diverse repertoire of layer-specific cortical pyramidal neuronal markers, including deep layer VI; V markers Tbr1, Foxp2, and Ctip2; and upper layer IV-II markers Satb2 and Brn2 (Figures 1F–1J and S2I–S2L), in accordance with results previously obtained with human cortical cells transplanted in neonatal cortex (Espuny-Camacho et al., 2013).

We then examined qualitatively the patterns of axonal projections of the transplants. At 2 MPT, consistent with immaturity, the transplants displayed little neurite outgrowth within the host, with GFP⁺ axons that were still positive for immature markers such as doublecortin (DCX) (Figures S2A–S2D). Axonal processes were detected mostly close to the site of the graft in the ipsilateral cortex, with few GFP⁺ axons extending distantly along the corpus callosum, along the internal capsule, and in the striatum (Figures S2E–S2H). In contrast, at 6 MPT, in accordance with the expression of generic and specific markers of mature pyramidal neurons from all cortical layers, the transplants displayed robust axonal outgrowth, with a high number of GFP⁺ projections located at short- and long-range distances from the graft location, including ipsilateral and contralateral cortical projections (target of layer II-III cortical neurons), and subcortical projections to striatum (target of layer V cortical neurons), thalamus (target of layer VI cortical neurons), and midbrain/hindbrain (target of layer V cortical neurons) (Figure 2).

Visual Cortical Transplants of Human PSC-Derived Cortical Neurons Display Visual-like Long-Range Axonal Projections in the Adult Brain

Next, we examined in more detail the pattern of projections from the transplants at 6 MPT, especially in terms of areal specificity. We found remarkable areal specificity of the axonal patterns, corresponding to visual and limbic cortex targets, within both cortical and subcortical structures (Figures 3A–3E). In the basal ganglia, transplant-derived projections preferentially targeted the dorsomedial striatum (DM) (a preferential target of visual cortex), rather than the dorsolateral striatum (DL) (preferential target of motor areas) (Figures 2A, 2D, 3A, 3B, and 3F). In the thalamus, transplant-derived axons were detected abundantly within visual/limbic thalamic nuclei (e.g., lateral geniculate nucleus [LG], lateral posterior thalamic nucleus [LP], laterodorsal thalamic nucleus [LD]), while only a minority of fibers were detected in other primary non-visual thalamic nuclei (e.g., ventrobasal complex [VB], posterior thalamic nucleus group [Po], ventrolateral thalamic nucleus [VL], medial geniculate nucleus [MG]) (Figures 2A, 2F, and 3C–3F). Within the midbrain/hindbrain, transplant-derived axonal fibers were detected mostly in the superior colliculus (SC) and periacqueductal gray matter (PAG), which receive mainly inputs from the visual cortex and the limbic cortex, while fewer or no projections were detected in non-visual targets such as the inferior colliculus (IC), pedunculopontine nucleus (PPTg), red nucleus (RedN), and pyramidal tract (Figures 2A, 2G, and 3F). In the contralateral cortex, we also observed projections at the level of the V1/V2 visual cortex and less so at the level of the motor cortex (Figure 2A). A comparison with data from the Allen Mouse Brain Connectivity Atlas following anterograde injection into the V1 visual cortex shows that projections of human transplanted neurons match the normal projections of the visual cortex (Oh et al., 2014).

These data indicate that human PSC-derived cortical neurons transplanted into the lesioned adult visual cortex display a mostly visual-like pattern of axonal projections, which is in line with the patterns observed following neonatal xenotransplantation (Espuny-Camacho et al., 2013) and with what was observed with transplanted mouse ESC-derived visual-like cortical neurons (Michelsen et al., 2015, Gaspard et al., 2008).

Long-Range Projections and Transplant Maturation Require a Visual Areal Identity Match

We found that the efficient integration of transplanted mouse ESC-derived cortical neurons or murine embryonic cortical tissue was achieved only if there was a precise match in the areal identity between lesioned and transplanted cells (Gaillard et al., 2007, Michelsen et al., 2015). Therefore, we compared our results obtained with transplants into the visual cortex with the outcome of the transplantations into a non-matching area, the motor cortex at 6 MPT (Figures S3B and S4A). We detected a transplant in 23 of 37 (62%) animals grafted in the motor cortex, including in 82% at 6 MPT, a rate that is slightly higher than that obtained with visual cortex transplantation, while graft volume was not significantly different between visual and motor locations (Figure 4A). However, examination of the pattern of axonal projections revealed that while a similar proportion of transplants projected fibers to the striatum (91% in visual transplants versus 71% in motor-located transplants), fewer motor-located transplants (<50%) sent long-range projections to the thalamic or midbrain/hindbrain levels compared to transplants in the visual cortex (91% in visual transplants versus 47% in motor-located transplants) (Figure 4B). Moreover, quantification of the total number of axonal fibers revealed that motor-located transplants sent far fewer axonal projections to specific subcortical targets, such as LG, LP, LD, SC, and PAG, with the exception of the striatum when compared to visual grafts (Figures 4D and S4B–S4E). It is interesting, however, that the distribution of fibers found in the striatum, thalamus, and midbrain/hindbrain from motor-located transplants also corresponded to a visual areal identity, albeit at a much lower efficiency (Figure 4E).

We then examined in more detail the cellular composition of the transplants in visual versus motor transplants. As expected, both visual and motor-located grafts showed broad expression of the pan-neuronal marker beta 3 class III tubulin (Figures 5A–5F). Notably, the percentage of Tbr1⁺ or Foxp2⁺ deep layer cortical neurons was not significantly different between visual versus motor-located grafts (Figure 5G). Alternatively, the expression of mature neuronal and mature glutamatergic markers NeuN and VGlut1 displayed lower levels of expression in motor transplants (Figures 5A–5D). Similarly, Couptf1, an early marker of the occipital cortex (Armentano et al., 2007) that is widely expressed in ESC-derived cortical cells in vitro and following their transplantation in the lesioned visual cortex (Espuny-Camacho et al., 2013, Gaspard et al., 2008), also was expressed at lower levels in the motor-located transplants (Figures 5E, 5F, and 5H). Collectively these data indicate that heterotypic transplants in the motor cortex, despite the fact that they contain similar numbers of cortical neurons, fail to re-establish long-range axonal projections and to mature properly at the molecular level.

Human Transplants in the Visual Cortex Are Synaptically Integrated into the Adult Mouse Brain

To determine whether the transplanted neurons actually integrate at the synapse level, we performed specific marker analyses and patch-clamp recordings. Marker analyses revealed robust expressions of maturation markers for dendrites, axons, and synapses (Figures 6A–6D).

To test functional maturation and synaptic integration, we then performed patch-clamp recordings of transplanted neurons from brain tissue slices prepared from late-stage transplanted animals (8–9 MPT) into the visual cortex. This revealed that the majority of the cells recorded (8 of 13) presented a regular accommodating firing pattern that is characteristic of pyramidal neurons (Figures 6E–6G, S5A, and S5C). Stuttering firing profile, characteristic of some types of cortical interneurons, also was detected in a minority of the cells (3 of 13) (Figure S5B), reflecting that a minority of putative GABAergic cells also was present in the transplants, as shown by Espuny-Camacho et al. (2013. To test whether the grafted human neurons received functional synaptic connections, we measured an electrophysiological correlate of incoming synaptic connectivity. We observed that the human neurons displayed at rest excitatory postsynaptic potentials (EPSPs), present in 3 of 13 cells (Figures 6H and 6I), confirming that the human grafted cells received functional synapses.

Finally, we tested whether the human cells displayed signatures of functional incoming synaptic activity from mouse neuronal networks. We measured their evoked response to a train of extracellular electrical stimuli applied onto mouse host cortical cells located at a distance from the graft. The human neurons (three cells of three tested) showed robust evoked synaptic responses that could be blocked by specific glutamatergic (cyanquixaline [CNQX] and 2-amino-5-phosphonopentanoic acid [AP-V]) and GABAergic (gabazine) receptor antagonists (Figure 6J).

These data indicate that human ESC-derived cortical neurons transplanted into the adult mouse cortex present a mature electrical neuronal profile and are effectively integrated into the glutamatergic and GABAergic adult mouse host circuitry at 8 MPT.

Human ESC Differentiation into Cortical Cells

For the differentiation of human ESCs toward cortical progenitors and neurons (Espuny-Camacho et al., 2013), cells were dissociated using StemPro Accutase on day 2 (Invitrogen, catalog no. A11105) and plated on Matrigel (BD, human ESC-qualified Matrigel)-coated coverslips/dishes at low confluency (5,000–10,000 cells/cm²) in MEF-conditioned human ESC/human iPSC medium supplemented with rho-associated protein kinase (ROCK) inhibitor (Y-27632; 10 μM Calbiochem, Sigma-Aldrich, catalog no. 688000). On day 0 of the differentiation, the medium was changed to default differentiation medium (DDM) (Gaspard et al., 2008) supplemented with B27 (10 mL B27 per 500 mL DDM, GIBCO) and Noggin (100 ng/mL, R&D Systems) (Espuny-Camacho et al., 2013). At 16 days in vitro (DIV), Noggin was withdrawn and the medium was changed to DDM and supplemented with B27 (GIBCO). At 24 DIV the human cortical progenitors were manually dissociated and a dilution of 100,000 cells/μL was prepared for transplantation experiments, as described by Espuny-Camacho et al. (2013).

Mice Lesioning and Grafting

All of the animal experiments were carried out in full compliance with the Université Libre de Bruxelles Committee for animal welfare and Belgian and European regulations. Data for this study are derived from a total of 58 adult mice of both sexes (n = 37 for motor-located transplants and n = 21 for visual-located transplants). Lesioning and grafting experiments were performed in 6- to 8-week old adult NOD/SCID mice, as described by Michelsen et al. (2015). Lesioning was performed under anesthesia by focal stereotactic injections of ibotenic acid either in the visual or motor cortex (20 mg of ibotenic acid in 1 μL distilled water; see Supplemental Experimental Procedures for coordinates). Three days after lesioning, approximately 100,000 human ESC-derived cortical progenitors (in 0.5–1 μL) were injected in the visual or the motor cortex.

Quantification of Axonal Projections

GFP⁺ fibers were quantified using a Zeiss Axioplan upright fluorescence microscope equipped with 20x and 40x dry objectives. The GFP was enhanced by immunofluorescence staining against GFP. For the striatum, the total number of fibers was counted in one-sixth of the coronal sections collected for each individual brain. For the thalamus and midbrain, fibers were counted in half of the coronal sections collected for each individual brain. Data are represented as means ± SEMs. For the percentage of fibers, 10 and 8 animals were used for quantification following transplantation into the visual and motor cortex, respectively. For the total number of fibers in visual and motor transplants, 11 and 17 animals were used for quantification following transplantation into the visual cortex and the motor cortex, respectively.

Quantification of the Tbr1, Foxp2, and Couptf1 Cells in Transplants

The percentage of Tbr1⁺, Foxp2⁺, and Couptf1⁺ cells detected among the total number of GFP⁺ cells (human grafted cells) was counted on confocal acquired images from immunofluorescence stained brain sections. Data are represented as mean percentages ± SEMs (n = 4−3 animals grafted into the visual cortex; n = 4−3 animals grafted into the motor cortex).

Quantification of Graft Volume

The graft volume was measured from confocal images from the graft in serial coronal sections corresponding to one-sixth of the mouse brain. Mouse brain sections were immunostained with GFP antibody and z stacks of 10 μm were acquired for a total of 6 steps (60 μm thickness). The volume of the graft was calculated using Fiji software. Data are represented as mean percentages ± SEMs (n = 8 animals grafted into the visual cortex; n = 8 animals grafted into the motor cortex).

Quantification of the Proportion of Transplants with Subcortical Projections

The proportion of visual and motor-located transplants projecting GFP⁺ fibers into the striatum, thalamus, or midbrain/hindbrain was quantified by manual counting from immunofluorescence stained brain sections using the GFP antibody. Data are represented as mean percentages ± SE of the proportions (SEPs) (visual grafts: n = 11 animals; motor grafts: n = 17 animals).

Brain Tissue Slice Electrophysiology

Electrophysiological recordings were performed intracellularly on grafted neurons in acute brain slices from mice at 8 and 9 MPT as previously described (Espuny-Camacho et al., 2013). Briefly, 300-μm thick coronal slices were cut in a sucrose-based solution and subsequently stored in artificial cerebrospinal fluid (aCSF) before recording. Transplanted cells were identified by their EGFP fluorescence and visualized using an upright microscope equipped with infrared differential interference contrast (DM LFS, Leica Microsystems). Whole-cell patch-clamp recordings and bipolar extracellular stimulation were performed and analyzed as described in more detail in the Supplemental Experimental Procedures.

Additional Methods

An extended version of the experimental procedures is described in the Supplemental Experimental Procedures.