Author contributions: I.J. and N.L.: conception and design, collection and assembly of data, and data analysis and interpretation; J.Y.L.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; I.H.P.: conception and design and provision of study material; K.S.P., S.H.S., C.C., D.J.C., J.K., S.H.O., D.A.S., J.T.D., and D.R.L.: collection and assembly of data; J.M.: statistical analysis; H.S.K., M.K., and K.S.K: data analysis and interpretation; G.Q.D.: conception and design and provision of study material; P.B.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; J.S.: conception and design, provision of study material, collection and assembly of data, data analysis and interpretation, financial support, manuscript writing, and final approval of manuscript. I.J., N.L., and J.-Y.L. contributed equally to this article.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS May 24, 2012.
Induced pluripotent stem cells (iPSCs) generated from somatic cells of patients can be used to model different human diseases. They may also serve as sources of transplantable cells that can be used in novel cell therapies. Here, we analyzed neuronal properties of an iPSC line derived from a patient with a juvenile form of Huntington's disease (HD) carrying 72 CAG repeats (HD-iPSC). Although its initial neural inducing activity was lower than that of human embryonic stem cells, we found that HD-iPSC can give rise to GABAergic striatal neurons, the neuronal cell type that is most susceptible to degeneration in HD. We then transplanted HD-iPSC-derived neural precursors into a rat model of HD with a unilateral excitotoxic striatal lesion and observed a significant behavioral recovery in the grafted rats. Interestingly, during our in vitro culture and when the grafts were examined at 12 weeks after transplantation, no aggregate formation was detected. However, when the culture was treated with a proteasome inhibitor (MG132) or when the cells engrafted into neonatal brains were analyzed at 33 weeks, there were clear signs of HD pathology. Taken together, these results indicate that, although HD-iPSC carrying 72 CAG repeats can form GABAergic neurons and give rise to functional effects in vivo, without showing an overt HD phenotype, it is highly susceptible to proteasome inhibition and develops HD pathology at later stages of transplantation. These unique features of HD-iPSC will serve as useful tools to study HD pathology and develop novel therapeutics. Stem Cells2012;30:2054–2062
Huntington's disease (HD) is a devastating, autosomal-dominant neurodegenerative disorder, caused by abnormal expansion of CAG repeats in the huntingtin gene. People carrying the HD mutation gradually develop personality changes, involuntary movements, weight loss, and eventually dementia. Aggregated huntingtin inclusions appear in the cytoplasm and nucleus. One of the affected neuronal populations is the striatal medium spiny projection neuron (MSN), which has largely degenerated in end-stage HD . Clinical trials have been performed using grafts of cells derived from human fetal striatal primordium. Outcomes have been variable with some reports describing no benefit while others have indicated transient clinical improvement with reduced motor dysfunction or slowed disease progression [2–4]. Because human fetal tissue transplants raise ethical issues and inevitably are genetically dissimilar to the recipient with the associated risk of immune rejection, other suitable nonfetal cell source of syngeneic donor tissue would be advantageous. The stem cell strategy has been proposed to restore GABAergic striatal projection neurons into the putamen and caudate and re-establish the degenerating striatopallidal circuit [5, 6]. In this study, we have demonstrated that it is possible to differentiate induced pluripotent stem cells (iPSCs) derived from a HD patient carrying 72 CAG repeats  into functional GABAergic projection neurons in vitro. After intrastriatal implantation of HD-iPSC-NPC, the grafted animals exhibited a clear functional recovery with no overt signs of pathology in the transplanted cells. However, we found that HD-iPSCs develop HD pathology when they are treated with a proteasome inhibitor or when they are grafted into animal models and examined at later stages of transplantation.
MATERIALS AND METHODS
Culture and Neuronal Differentiation of HD-iPSC
We cultured and maintained HD-iPSC according to the method described previously  and induced neuronal differentiation of HD-iPSC by coculturing the cells with PA6 stromal cells (obtained from Riken Cell Bank, Japan) as described previously . For controls, we used a human embryonic stem cell (ESC) (H9, obtained from WiCell, Madison, WI), a normal iPSC (F5), and an additional HD-iPSC line (HD-iPS2) that was derived simultaneously when HD-iPSC was established. These undifferentiated pluripotent stem cell colonies were mechanically dissected and transferred onto freshly prepared PA6 cells in differentiation medium (DM-PA6) that consists of glasgow minimum essential medium (GMEM) containing 10% KO-SR (knockout serum replacements, Invitrogen, Grand Island, NY), and 4 days later, KO-SR in DM-PA6 was replaced by N2 supplements (Invitrogen). In the following 11–13 days, definitive neural rosette-like structures containing neuroepithelial (NE) cells were formed, which were mechanically detached and transferred onto a nonsticky Petri dish for suspension culture for 6 days to form neurospheres.
Stable Generation of Neural Precursor Cells
To make single cells at the neural precursor (NP) stage, we dissociated neurospheres following treatment with Accutase (Chemicon, Billerica, MA) and plated onto polyL-ornithine (PLO; 15 μg/ml, Sigma, St. Louis, MO)/fibronectin (FN; 1 μg/ml, Sigma)-coated 60 mm2 tissue culture dishes. NP cells (NPCs) were maintained in GMEM supplemented with 1% penicillin, 1% streptomycin, 1% nonessential amino acids, 0.1% β-mercaptoethanol, N2 supplements, and 20 ng/ml basic fibroblast growth factor (bFGF) (Invitrogen).
Differentiation into Mature Neurons
To differentiate iPSCs into mature neurons (MNs), we plated neurospheres directly onto PLO/FN-coated dishes in DM medium supplemented with 20 ng/ml brain-derived neurotrophic factor (BDNF, R&D Systems, Minneapolis, MN) in the absence of bFGF.
To analyze the marker expression of NP-stage and differentiated neuronal cells, we carried out immunocytochemical analyses using the following primary antibodies: human-specific nuclei (1:200; Chemicon), human-specific mitochondria (1:200; Chemicon), human-specific Nestin (1:200; Chemicon), SOX2 (1:200; Chemicon), type III β-tubulin (Tuj1) (1:500; Chemicon), Musashi (1:500; Chemicon), OTX2 (1:500; Chemicon), BF-1 (1:100; Santa Cruz, CA), GSH-2 (1:200; Santa Cruz), DLX2 (1:250; Chemicon), MAP2 (1:200; Chemicon), NeuN (1:500; Chemicon), Gamma-aminobutyric acid (GABA) (1:5,000; Sigma), glutamic acid decarboxylase 65/67 (GAD65/67) (1:200; Chemicon), DARPP-32 (1:100; Cell Signaling, Danvers, MA), Calbindin (1:250; Chemicon), TH (tyrosine hydroxylase, 1:1,000; Pel-Freez, Rogers, AR), SVP38 (1:200; Sigma), and EM48 (1:50; Chemicon). To detect proliferating or undifferentiated cells in the graft, proliferating cell nuclear antigen (PCNA) (1:50; Santa Cruz) and OCT4 (1:250; Santa Cruz) were also used. Secondary antibodies used were goat anti-mouse IgG-conjugated Alexa-555 (1:200; Molecular Probes, Eugene, OR), goat anti-rabbit IgG-conjugated Alexa-488 (1:200; Molecular Probes), and goat anti-mouse IgM-conjugated Alexa-555 (1:200; Molecular Probes). Staining patterns were examined and photographed using a confocal laser-scanning microscope imaging system (LSM510; Carl Zeiss, Inc., Thornwood, NY).
Fluorescence-Activated Cell Sorting Analysis
To quantify the percentage of GABAergic neurons formed within the neuronal population of HD-iPSC, we performed fluorescence-activated cell sorting (FACS) using the FACS Calibur System (BD Bioscience, San Jose, CA). Data analysis was carried out according to manufacturer's instructions (BD Bioscience).
We transferred the cells that were attached to a glass coverslip into a bath mounted on the stage of an inverted microscope (IX-70, Olympus, Osaka, Japan). The bath (∼ 0.15 ml) was superfused at 5 ml/minute, and voltage-clamp experiments were performed at room temperature (22°C–25°C). Patch pipettes with a free-tip resistance of approximately 2.5–3 MΩ were connected to the head stage of a patch-clamp amplifier (Axopatch-1D, Axon Instruments, Foster City, CA). pCLAMP software v.9.2 and Digidata-1322A (Axon Instruments) were used to acquire data and apply command pulses. Whole-cell currents were recorded at 10 kHz and were low-pass filtered at 1 kHz, respectively. Current traces were stored and analyzed using Clampfit v.9.2 and Origin v.7.0 (Microcal, Northampton, MA). For comparison of whole-cell currents between cells, we normalized the current amplitudes to the membrane area measured by electrical capacitance. Receptor agonist, GABA, and antagonist, picrotoxin, were tested in whole-cell currents.
Quinolinic Acid-Induced Experimental HD Animal Model and Cell Transplantation
We carried out animal experiments in accordance with the CHA University IACUC (Institutional Animal Care and Use Committee; IACUC090012). Adult male Sprague-Dawley rats (Orient, Seoul, Korea) weighing 250–280 g were used. All rats received a stereotaxic unilateral lesion of the striatum via double injections of quinolinic acid (QUIN, Fluka, Milwaukee, WI; 0.3 mol/l × 1.5 μl) at the following coordinates: AP +0.7 mm, ML +2.5 mm, DV –4.5 mm, and AP +0.7 mm, ML +2.5 mm, DV –3.5 mm from the Bregma. At 7 days post-QUIN lesion, we injected a total of 12 rats with 2 μl of HD iPS-derived NP cells (100,000 cells per microliter) at the following coordinates: AP +0.2 mm, ML +2.2 mm, and DV –4.0 mm from the Bregma. For comparison, we transplanted QUIN-lesioned rats with 2 μl of suspension medium (GMEM) containing the equal numbers of H9-NPC (n = 8) or F5-NPC (n = 11). In the control group (n = 9), 2 μl of GMEM was injected in parallel. Transplanted animals received cyclosporine A intraperitoneally (10 mg/kg, Sigma) 24 hours before transplantation and daily up to 12 weeks.
To determine that the QUIN-induced striatal lesions were extensive and to evaluate whether rats recovered functionally following cell transplantation, we performed stepping test, staircase test, and apomorphine-induced rotation tests 1 day prior to transplantation and every 2 weeks following transplantation as previously described [10–12].
We performed statistical analyses on the behavior data using the Statistical Analysis System (Enterprise 4.1; SAS Korea, Seoul, Korea) on a CHA University mainframe computer. Performance measures were analyzed using the PROC MIXED program, a linear mixed models procedure. The dependent variables were mean scores from each behavioral test and independent variables were “Treatment” (H9-, HD-, F5-, and sham groups) and “Week” (0w pretransplantation, 2w, 4w, 6w, 8w, 10w, and 12w). The results are presented as the mean ± SEM. A p value of <.05 was considered significant.
For immunohistochemical analysis, we sacrificed the rats at 12 weeks post-transplantation, perfused, and fixed their brains with 4% paraformaldehyde. Forty micrometer frozen coronal sections were prepared using a cryostat (Microm, Walldorf, Germany). Antibodies used for immunohistochemistry were the same as described in immunocytochemical analyses (see above). To confirm the lesions induced by QUIN injections, we also carried out cresyl violet staining, Fluoro-Jade staining and conventional 3,3′-diaminobenzidine (DAB) immunohistochemistry using DARPP-32 antibody simultaneously.
Neuronal Differentiation of HD-iPSC
We first examined the neuronal properties of a HD-iPSC line derived from the skin fibroblast of a juvenile HD patient carrying 72 CAG repeats , which was generated by retroviral infection of four pluripotency factors (Oct4, Sox2, Klf4, and c-Myc).
Supporting Information Figure S1A outlines our five-stage differentiation protocol, which involves coculture of undifferentiated HD-iPSC (Stage 1) with PA6 stromal cells  (Stage 2) and isolation of neural rosettes (Stage 3), and followed by neurosphere formation in suspension culture (Stage 4). Afterward, we either dissociated the neurospheres into single cells, consisting of neural precursor (NP) cells, for transplantation experiments (Stage 5-NP), or further differentiated the cells into MNs (Stage 5-MN). Although it is well known that PA6 coculture method drives the fate of neuronal differentiation into more midbrain-type neurons , isolation of NE cell types at early stage and treatment with BDNF render the nature of neuronal populations to be maintained as forebrain-type or GABAergic neurons [13, 14].
Thus, we first differentiated the HD-iPSC into neural rosette-like structures by coculturing them with PA6 stromal cells (Supporting Information Fig. S1B, Stage 3) and then grew them as neurospheres in suspension culture (Supporting Information Fig. S1C, Stage 4). We were able to maintain the resulting neural precursor cells (NPCs) (Supporting Information Fig. S1D, Stage 5-NP) in dissociated cultures for 5–10 passages without change of the normal karyotype (Supporting Information Fig. S2A). As described above, the HD-iPSC line was generated from a juvenile HD patient carrying 72 CAG repeats. We did not detect any alterations in CAG repeat numbers either after repeated expansion (Supporting Information Fig. S2B) or after neuronal differentiation (Supporting Information Fig. S2C).
Since the patterns of neuronal differentiation in HD-iPSC line showed some differences, compared with H9, F5, or HD2 lines, we compared the extent of neuronal differentiation by counting the numbers of colonies forming neural rosette-like structures out of total numbers of colonies at Stage 3 (Supporting Information Fig. S3A). Interestingly, we found that while H9 shows the highest rosette-forming efficiency (86.60% ± 1.83%), F5, HD, and HD2 exhibit significantly reduced efficiency compared with H9 cells but they also showed differences among each other (18.28% ± 0.81%, 32.97% ± 1.90%, and 26.82% ± 1.57%, respectively) (Supporting Information Fig. S3A). Similarly, we also found the total areas of rosettes to be formed in each colony at Stage 3 were the highest in H9 (71.37% ± 1.67%) and were reduced in F5, HD, and HD2 (27.32% ± 9.29%, 42.09% ± 8.53%, and 31.61% ± 3.56%, respectively) (Supporting Information Fig. S3B), indicating the efficiency of initial neuronal differentiation in HD-iPSC at Stage 3 is significantly lower than in H9 cells and slightly higher than in HD2 and relatively higher than F5 cells. As for the reduced efficiency of neuronal differentiation in F5 cells, compared with HD and HD2 cells, we found that Klf4 expression in F5 cells is considerably high (Supporting Information Fig. S3C), and the recent study suggests that neuronal differentiation is negatively regulated by the expression of miR-371-3, which is controlled by Klf4 . In this regard, the real-time quantitative polymerase chain reaction (PCR) results showing the endogenous Klf4 expression clearly correlate with the extent of neuronal differentiation to a certain extent (Supporting Information Fig. S3C). Since HD cells are shown to be better than HD2 cells in neuronal differentiation, we decided to focus mainly on HD cells in this study.
Formation of Striatal MSN and GABAergic Neurons from HD-iPSC
At the NP stage, the cells mainly expressed Nestin, Sox2, and Musashi, phenotypic markers of neural precursors (Fig. 1A, 1B, 1D). By contrast, they did not yet express MAP2, a mature neuronal marker or glial fibrillary acidic protein (GFAP), a glial cell marker (Fig. 1B–1D). Since Sox2 was used to generate iPSC, we examined whether the level of exogenous Sox2 gene was maintained at the NP stage. We observed that its level was still strongly maintained in F5-NPC but not in either HD- or HD2-NPC (Supporting Information Fig. S3D), indicating that high levels of Sox2 expression in HD-iPSC-NPC were unrelated to the exogenous gene.
To induce more MNs, we first attached the neurospheres (Stage 4 in the protocol) and further differentiated them for 1 week (Stage 5-MN). We found that this adhesion culture gave rise to a highly efficient neuronal differentiation, and majority of neurosphere-derived cells exhibited features of MNs, as shown below. Importantly, with our differentiation protocol, we predominantly generated forebrain-type neurons and relatively few midbrain-type neurons, as evident by the proportions of cells expressing OTX2 (76.0% ± 1.3%) and TH (3.1% ± 0.4%) (Fig. 1E). Formation of forebrain neurons was further confirmed by immunostaining for a forebrain-specific marker, such as BF-1 (also called as FOXG1; Fig. 1F; 38.4% ± 3.4%), in which type III-β tubulin (Tuj1) was used as a general neuronal marker. We also found that high proportions of HD-iPSC form lateral ganglionic eminence (LGE) progenitors, from which striatal MSNs subsequently develop. Formation of LGE progenitors was confirmed by expression of LGE-specific markers, such as GSH-2 (Fig. 1G; 28.8% ± 2.8%) and DLX2 (Fig. 1H; 34.1% ± 4.5%). Importantly, we found that HD-iPSCs also form DARPP-32 neurons (Fig. 1H; 27.0% ± 1.7%). It should be noted that while DARPP-32 is widely used as a marker in the striatum for MSNs, it is not specific to those cells. Widespread expression of DARPP-32 is observed in the CNS, including in non-neuronal cells (such as astrocytes). Therefore, co-localized expression of DARPP-32 with GSH-2 or DLX2 (as shown in Fig. 1H) definitively confirms that those double-positive neuronal cells differentiated from HD-iPSC are, in fact, striatal MSNs. We indeed detected that striatal neurons are formed, judged by Calbindin expression (Fig. 1I; 19.1% ± 2.1%). In addition, we also observed that several of the neurons were immunopositive for the synaptic protein synaptophysin (SVP38) (Fig. 1J; 20.0% ± 3.8%), and the proportion of MNs immunopositive for MAP was significantly high (Fig. 1J; 89.1% ± 5.5%). It is noteworthy that significant numbers of HD-iPSC-derived neurons can give rise to the neurotransmitter GABA (Fig. 1K; 75.1% ± 2.2%). FACS analysis further demonstrated that more than 88% of the NeuN-positive neurons was also immunoreactive for GABA (Fig. 1L).
Semi-quantitative reverse transcription polymerase chain reaction (RT-PCR) results further confirmed that the neuronal cells derived from HD-iPSC express high levels of forebrain (i.e., BF-1 and GSH-2), striatal (i.e., Calbindin), MSN (i.e., DARPP-32), and GABAergic (i.e., GAD67 and GABRAα2) markers (Fig. 1M). To verify the GABAergic neuronal property, we then performed patch-clamp analysis and found that the cells evoked whole-cell current when GABA (10 μM, 100 μM) was applied (Fig. 1N, left). These currents were fully blocked by the specific GABA receptor blocker, picrotoxin (Fig. 1N, right), implying the presence of GABA-sensitive currents from the HD-iPSC-derived neurons. High-performance liquid chromatography (HPLC) analysis further showed that the level of GABA release in the medium from HD-iPSC-derived neurons at Stage 5-MN was 42.07 ± 3.5 pmol/μl per 1 × 106 cells. Taken together, marker expression and electrophysiological and HPLC analyses strongly suggest that a large proportion of the HD-iPSC had differentiated into striatal MSN and functional GABAergic neurons. We finally compared the extent of neuronal or GABAergic differentiation in H9, F5, HD, and HD2 cells by real-time quantitative RT-PCR analysis (Supporting Information Table S3), and the results indicate that H9 cells again showed the highest efficiency compared with other three cell lines in forming MNs (Map2), and MSN (DARPP-32), or GABAergic (GAD65, GABRAα1 and GABRAα2) neurons. Interestingly, among three cell lines other than H9 cells, HD cells showed the highest efficiency. Recently, several groups have published efficient protocols to differentiate human ESCs into GABAergic neurons [16, 17], and it will be interesting to test whether they also work in HD-iPSCs efficiently, as we showed in this study.
Behavioral Recovery Following Transplantation of HD-iPSC-Derived NP Cells into QUIN-Lesioned HD Animal Models
To assess whether the GABAergic neurons derived from HD-iPSC could exert functional effects after intracerebral implantation, we grafted them into a rat model of HD. First, we subjected rats to unilateral intrastriatal injections of QUIN. One week later, we grafted 200,000 HD-iPSC-derived NPC (Stage 5), in a volume of 2 μl, into the damaged striatum of each rat (n = 12). For comparison, we transplanted the equal numbers of H9-NPC (n = 8) or F5-NPC (n = 11). In sham control rats (n = 9), we only injected vehicle (GMEM). To evaluate functional effects of the grafts, we used three behavioral tests (i.e., stepping, staircase, and apomorphine-induced rotation tests) up to 12 weeks following transplantation.
To our surprise, all three groups transplanted with H9-, F5-, or HD-derived NPC gave rise to a significant behavioral recovery, compared with the sham group. In stepping test, there was a significant interaction between Treatment and Week (F21,280 = 2.71, p = .0001, Fig. 2A), with the main effects for Treatment (F3,40 = 6.19, p = .002, Fig. 2A) and Week (F7,280 = 253,19, p < .0001, Fig. 2A). Up until 2 weeks, four treatment groups did not show significant between-group differences. However, from 4 weeks, the F5 group started to reveal a significant improvement compared to the sham group and maintained the better performance until 12 weeks (ps < .002, Fig. 2A). The H9- and HD groups displayed a similar trend as seen in the F5 treatment group from 6 weeks to 12 weeks (ps < .04 for H9 vs. Sham and ps < .005 for the HD group vs. the sham group, Fig. 2A). There were no group differences between three cell-treated groups at any given time, except at 0 and 8 weeks. The HD group experienced a slight drop in stepping test and, as result, there was a significant difference between HD- and F5 groups (p = .03, Fig. 2A) at 8 weeks.
In staircase test, there was a significant interaction effect of Treatment and Week on the number of pellets (F18,240 = 2.97, p < .0001, Fig. 2B) and the both main effects, Treatment (F3,40 = 3.31, p = .03) and Week (F6,240 = 36.34, p < .0001) were significant. The H9- and F5 groups exhibited a significantly better performance than the sham group at any given time except for the 0 week (ps <.04 for the H9 group compared with the sham group and ps < .04 for F5 compared with the sham group, Fig. 2B). The HD treatment gave rise to a behavioral recovery from 4 weeks until 12 weeks compared to the nontreated group, the sham group (ps < .05, Fig. 2B), indicating the slower onset of recovery effect compared to the other cell-treated groups. There were no significant differences between three treatment groups from 4 weeks until 12 weeks.
Results of apomorphine-induced rotation test demonstrated the presence of a significant interaction (F18,240 = 2.97, p < .0001, Fig. 2C) between Treatment and Week. There was a significant main effect in Treatment (F3,40 = 3.31, p = .03) and Week (F6,240 = 36.34, p < .0001) on behavioral responses. The overall trend, as shown in Figure 2C, was that the cell-injected groups other than the F5 group tended to show a gradual recovery over time, whereas the H9- and HD groups induced a significant behavioral recovery from 6 weeks to 12 weeks compared to the sham group (ps = .04, Fig. 2C). The F5 group showed a significant improvement in comparison with the sham group at 2 weeks (p = .02, Fig. 2C), 6 weeks (p = .0002, Fig. 2C), and 8 weeks (p = .002, Fig. 2C). At 2 weeks after transplantation, a sharp decrease in the number of ipsi-lateral turns of the F5 group contributed to a significant difference between the HD- and F5 groups (p = .02, Fig. 2C). There was a significant difference between the H9- and F5 groups at 10 weeks (p = .03, Fig. 2C) and 12 weeks (p = .05, Fig. 2C) when the state of the F5 group became close to that of the sham group after recovery.
Formation of Neural Tissues from Transplanted HD-iPSC
After completion of the behavioral tests at 12 weeks, we sacrificed the rats and analyzed their brains histologically. To identify the transplanted human cells, we used antibodies against human-specific nuclear antigen and human mitochondria. Using confocal microscopy and double immunostaining, we found that some of the transplanted HD-iPSC-derived cells were still Nestin-positive neural precursors (Fig. 2A). Importantly, a large number of the surviving human neurons were MAP2-positive (Fig. 2B), and some neurons exhibited MSN and GABAergic features, that is, they were immunoreactive for DARPP-32 (Fig. 2C), GAD65/67 (Fig. 2D), and GABA (Fig. 2E). We also found that transplanted human cells expressed the synaptic vesicle protein synaptophysin (SVP38) (Fig. 2F), suggesting the possibility of synapse formation in the graft.
From our results, HD-iPSC-NPC can give rise to GABAergic neurons following transplantation, which could contribute to functional recovery to a certain extent. However, we cannot exclude the possibility that transplanted cells could provide the host with non-cell autonomous effects, which may include secretion of growth factors, reduction of inflammatory responses, and so forth.
Previous studies, using other sources of donor tissue, have indicated that the GABAergic subpopulation of neurons mediates graft-induced functional recovery [18, 19]. Human ESC-derived striatal progenitors can be differentiated into DARPP-32-positive neurons . Following intracerebral implantation, however, the resulting grafts were often found to have grown excessively, making them unsafe and unsuitable for cell therapy in HD , although more recent study suggests a significantly reduced risk for tumor growth . In our study, we never found any signs of overgrowth in all 40 grafted rats 12 weeks after transplantation, and no PCNA-positive cells (Fig. 3G) or OCT4-positive cells (data not shown) were detected from the transplanted cells. Taken together, these results indicate that surviving transplanted HD-iPSC-derived GABAergic neurons support behavioral recovery.
Increased Susceptibility to HD Aggregate Formation in HD-iPSC
Intranuclear inclusions of huntingtin are a neuropathological hallmark of HD, which is readily demonstrated using an antibody against aggregated huntingtin (EM48) [20, 21]. We examined whether iPSC-derived neurons develop huntingtin aggregates after differentiation or intracerebral transplantation. To our surprise, we found no EM48-positive aggregates were formed in cultured (data not shown) or grafted HD-iPSC-derived cells (Fig. 3H), while the neural stem cells that we isolated from the YAC (yeast artificial chromosome) 128 transgenic mice and their brain tissues exhibited a distinct EM48-positive staining (data not shown). To explain this, we speculated that human cells simply develop aggregates much later than murine counterparts or that the reprogramming of the cells in order to obtain iPSC might have affected processes that influence development of a cellular HD phenotype. To test this hypothesis, we treated undifferentiated HD-iPSC and F5-iPSC with a proteasome inhibitor, MG132, at varying concentrations (0, 2, 5, and 10 μM). Interestingly, we observed that HD-iPSC started forming EM48-positive aggregates at 5 μM and showed more pronounced effects at 10 μM, whereas F5-iPSC formed no EM48-positive aggregates even at the highest dose (Fig. 4A).
We also speculated that, if HD-iPSC shows slow protracted huntingtin aggregate formation, such aggregates might develop at later stages of transplantation. As shown above, our initial analysis was completed at 12 weeks following transplantation into QUIN-lesioned rat brain with no signs of HD pathology. To test long-term effects, we transplanted HD-iPSC-NPC into the lateral ventricle of the postnatal day 2 (P2) normal CF-1 mouse brains and examined EM48 expression at 33 weeks and 40 weeks. At these stages, we observed clear signs of HD pathology (Fig. 4B), although the precise time at which the aggregates developed has yet to be determined. By contrast, we observed that no EM48-positive aggregates developed in human ESC (H9)-NPC, which was transplanted and analyzed in the same way (Fig. 4B).
In this study, we have extensively analyzed the neuronal properties of HD-iPSC derived from a HD patient carrying 72 CAG repeats. Although its initial neural differentiation efficiency was lower than that of human ESC, we demonstrate that HD-iPSC can produce functional GABAergic projection neurons resembling MSNs, the cell type most severely affected in HD patients. We show that such neurons can survive after transplantation into the striatum of a rat model of HD and can support significant behavioral recovery in the host.
Interestingly, we found no signs of huntingtin aggregation during in vitro culture or in the grafted cells. However, when the culture was treated with a proteasome inhibitor (MG132) or when the grafted cells were analyzed after longer survival times (e.g., 33 weeks), huntingtin aggregation was evident. These results indicate that HD-iPSC carrying 72 CAG repeats is highly susceptible to proteasome inhibition and can develop pathological symptoms at later stages, although it can form GABAergic neurons and elicit functional recovery in vivo at early stages. Previous studies using HD-iPSC have not reported HD pathology, except for a mild increase in caspase activity  or enhanced lysosomal activity . By contrast, our results demonstrate that HD pathology can be induced directly from HD-iPSC, providing a unique experimental platform to study HD pathology and develop drug screening. Since HD pathology is known to increase by CAG repeats, it will be interesting to investigate the effects of CAG repeat numbers on the development of HD pathology using iPSC technology. Although our results indicate that HD-iPSC can be stimulated to differentiate into GABAergic neurons efficiently and promote functional recovery without development of HD pathology for at least 3 months in rats, it turned out that HD pathology can develop following 33-week survival periods after grafting into neonatal brains. For this reason, it is not relevant to use HD-iPSC directly for cell therapy purpose. To do this, it will be essential to target the HD mutation and correct it to a normal gene in the HD-iPSC before grafting [24–26]. For clinical applications, it will be essential to use transgene-free reprogrammed iPSCs that are derived from patients under good manufacturing protocol (GMP) conditions, which may provide a suitable source for autologous transplantation. At the same time, since mesenchymal stem cells (MSCs) have been widely used for clinical trials, it will be worthwhile to consider applying MSC-based cell therapies to treat HD  until iPSCs can be readily used for clinical purpose.
In summary, here we show that HD-iPSCs carrying 72 CAG repeats can form GABAergic neurons efficiently, and are functional when they were transplanted into a rat model of HD. However, they are susceptible to proteasome inhibition and recapitulate the disease phenotype. They can also develop HD pathology at later stages of transplantation. These unique features of HD-iPSCs can serve as useful tools to study HD pathology and develop novel therapeutics. For autologous cell transplantation, it will be essential to correct the mutated HD gene prior to grafting.
This work was supported by grants from the Seoul R&D Program (#10548), Korea Food & Drug Administration (S-11-04-2-SJV-993-0-H), Korea Health Technology R&D Project, Ministry of Health & Welfare (A111016), and Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2012R1A1A2006827), Republic of Korea to J.S.; and the Swedish Research Council to P.B. P.B. and J.Y.L. are part of the Bagadilico Linneaus consortium. We are grateful to Britt Lindberg for assistance with histology and Thomas Deierborg for useful discussion on HD animal models. We are also grateful to Woo-Seok Im, Hyun Woo Choi, Hyo Jin Jang, and Yun-Hwa Jeong for the studies using YAC128 mice, stem cell maintenance, and assistance with statistical analysis. J.S. is particularly grateful to Seung Jae Lee for advice on proteasome inhibitor treatment, Hoon Ryu for useful discussion on HD pathology, and Soo Young Kim at Korea Basic Science Institute for HPLC analysis.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.