Neurite Outgrowth and Gene Expression Profile Correlate with Efficacy of Human Induced Pluripotent Stem Cell-Derived Dopamine Neuron Grafts

Abstract

Transplantation of human induced pluripotent stem cell-derived dopaminergic (iPSC-DA) neurons is a promising therapeutic strategy for Parkinson's disease (PD). To assess optimal cell characteristics and reproducibility, we evaluated the efficacy of iPSC-DA neuron precursors from two individuals with sporadic PD by transplantation into a hemiparkinsonian rat model after differentiation for either 18 (d18) or 25 days (d25). We found similar graft size and dopamine (DA) neuron content in both groups, but only the d18 cells resulted in recovery of motor impairments. In contrast, we report that d25 grafts survived equally as well and produced grafts rich in tyrosine hydroxylase-positive neurons, but were incapable of alleviating any motor deficits. We identified the mechanism of action as the extent of neurite outgrowth into the host brain, with d18 grafts supporting significantly more neurite outgrowth than nonfunctional d25 grafts. RNAseq analysis of the cell preparation suggests that graft efficacy may be enhanced by repression of differentiation-associated genes by REST, defining the optimal predifferentiation state for transplantation. This study demonstrates for the first time that DA neuron grafts can survive well in vivo while completely lacking the capacity to induce recovery from motor dysfunction. In contrast to other recent studies, we demonstrate that neurite outgrowth is the key factor determining graft efficacy and our gene expression profiling revealed characteristics of the cells that may predict their efficacy. These data have implication for the generation of DA neuron grafts for clinical application.

Introduction

Therapeutic interventions alleviating motor symptoms in Parkinson's disease (PD) typically target replacement of lost dopamine (DA) transmission within the basal ganglia to overcome the impact of nigrostriatal degeneration [1]. Cell replacement therapies seek to restore the DA neurons that have lost connections of medium spiny neurons in the striatum (STR). Clinical trials using fetal tissue containing DA neuron precursor cells as a reparative strategy for PD have demonstrated long-term recovery of function in some patients [2] and graft survival for over 20 years [3].

Human pluripotent stem cells [hPSCs: human embryonic stem cells or human induced pluripotent stem cells (iPSCs)] can be differentiated into ventral midbrain DA cells, and these cell therapy products have been widely reported to survive, integrate, release DA, and alleviate functional impairments when transplanted into rodent models of PD [4–7]. Cell therapies for PD are moving into an exciting era, with several first-in-human clinical trials using hPSC-derived DA neuron precursors recently begun, or due to commence [8–11].

Most current strategies for cell replacement have relied on a single immunologically unmatched hPSC line to generate DA neuron precursors for transplantation. Since these allogeneic approaches require immunosuppression and its associated risks, we are investigating the feasibility of autologous DA neuron replacement. In this study, we used iPSCs from two individuals with idiopathic PD. The primary aims were to assess the reproducibility of the generation of transplant-ready DA neuron precursors, and to identify the point during differentiation for optimal integration and efficacy of the induced pluripotent stem cell-derived dopaminergic (iPSC-DA) grafts. We chose 18 days (d18) and 25 days (d25) of differentiation in vitro, and phenotyped cultures at the two stages with transcriptome profiling. DA precursors from both donors were transplanted into a hemiparkinsonian rat model. Our data show that regardless of cell line or differentiation stage, all transplanted preparations survived and matured into grafts rich in DA neurons. However, we found that there was a striking difference in efficacy between d18 and d25-differentiated cells from both iPSC lines.

The d18 precursors from both lines consistently alleviated motor impairments in the rat model as measured 6 months after transplantation, while the d25 cells consistently failed to reverse motor symptoms in this model. Immunohistochemical analysis of the iPSC-DA grafts showed that classic measures of graft volume or DA neuron content did not distinguish efficacious from nonefficacious transplants. Instead, significant differences in the extent of neurite outgrowth from the grafts correlated with the ability to elicit behavioral recovery.

Transcriptome analysis of cell culture phenotype revealed differences between efficacious and nonefficacious DA precursors in expression of genes related to neurite outgrowth, development, and plasticity. Interestingly, the efficacious pretransplantation stage was characterized in both cell lines by the transcriptional repression of a group of neural differentiation-associated genes regulated by RE1 Silencing Transcription Factor (REST); these genes were expressed at the later, nonefficacious stage, accompanied by a decrease in REST expression. This suggests that the optimal cell population would be at a plastic stage that enables response to the host environment by extending neurites and integrating into the brain.

Methods

The results of this study show, in two different patient-derived iPSC lines, that specific characteristics of the cultured cells determine their ability to reverse motor deficits in a rat model of PD, which is dependent on their ability to extend neurites into the host brain. Two iPSC lines (410, 411) were generated from fibroblasts harvested from two people with idiopathic PD (Fig. 1). The iPSC lines were differentiated in vitro to DA neuron precursors for 18 or 25 days to investigate the optimal stage for transplantation. To confirm that the precursors were capable of differentiation into mature DA neurons, both cell lines were differentiated further in vitro into mature DA neurons. The precursors harvested after 18 days (410-d18; 411-d18) or 25 days (410-d25; 411-d25) of differentiation were used for transplantation into a rodent model of PD. At the same stages, replicate cultures were analyzed by mRNA sequencing to examine their phenotypic characteristics.

FIG. 1.

Experimental Overview. (A) Study design showing experimental progression from fibroblasts to transplant-stage neuronal precursors, and the main outcome measures from both in vitro and in vivo investigation of their suitability as a DA neuron replacement source. (B) Schematics showing unilateral 6-OHDA MFB lesion model with accompanying unilateral TH depletion in immunolabeled brain sections in the STR and SNc (left), and cell replacement strategy, which consists of surgical implantation of cells into the STR (right). 6-OHDA, 6-hydroxydopamine; DA, dopamine; MFB, medial forebrain bundle; STR, striatum; SNc, substantia nigra pars compacta.

Both cell lines were transplanted into the 6-hydroxydopamine (6-OHDA) lesioned hemiparkinsonian model of PD, at both stages of the differentiation protocol. Other groups of rats remained as naive controls or lesioned controls. Rats underwent behavioral testing pre- and posttransplantation and brains were taken for histological analysis at 24 weeks postgraft. Histological analysis of the grafts included markers of human cells (HuNu) and mature DA neurons (TH, GIRK2, AADC), as well as two methods of measuring the extent of neurite outgrowth from the TH+ neurons. All experiments were conducted in compliance with the UK Animals (Scientific Procedures) Act 1986 under Home Office License No. 30/3036 and with the approval of the local Cardiff University Ethics Review Committee. See Supplementary Methods for further details.

Results

Both iPSC lines differentiated into mature DA neurons in vitro

To assess the abilities of the two iPSC lines to differentiate into mature DA neurons, we cultured them under differentiation conditions for 12 weeks in vitro. Both donor-derived iPSC lines differentiated into mature DA neurons, showing mature DA neuron markers by immunocytochemistry, bursting activity and mature membrane potential by patch clamp electrophysiology, and high performance liquid chromatography (HPLC) to detect evoked DA release (Fig. 2A).

FIG. 2.

Characterization of dopaminergic neuron differentiation and maturation from two PD donor iPSC lines. (A) iPSCs from 410 to 411 cell lines that were differentiated to day 80 in vitro express DA neuron marker TH (green). Single-cell patch clamp analysis of day-75 cells showed that most of the neurons had pace making activity and mature membrane resistance (Ra) and capacitance (Cm). Action potential amplitudes were maintained close to 60 mV. HPLC analysis of day 88–89 cultures stimulated with KCl shows release of DA and small amounts of NE and 5-HT. (B) Representative examples of immunohistochemical analysis of TH+ grafts in Control, Lesion, d18 and d25 graft groups. 5-HT, serotonin; HPLC, high performance liquid chromatography; iPSCs, induced pluripotent stem cells; NE, norepinephrine; PD, Parkinson's disease; TH, tyrosine hydroxylase.

Behavioral analysis revealed motor recovery after transplantation of earlier (d18) but not later (d25) DA neuron precursors

Multiple behavioral analyses showed that the hemiparkinsonian rats transplanted with the earlier (d18) stage precursors showed reversal of deficits, while those transplanted with the later (d25) stage precursors did not recover. On the amphetamine-induced rotation test, motor recovery was evident after transplantation of earlier (d18) but not later (d25) DA neuron precursors (Fig. 3A, B; F_30,270 = 5.59, P < 0.001). For the apomorphine-induced rotation test, rats grafted with cells from the earlier stage of differentiation (410-d18 and 411-d18) did not differ from unlesioned control rats, indicating that improvement in the rotational bias was conferred by both cell lines differentiated to d18 (Supplementary Fig. S1A; F_5,50 = 6.34, P < 0.001). On the gait analysis test, rats grafted with cells from the earlier differentiation stage showed increased stride length relative to lesioned controls, suggesting improvement in gait performance (Supplementary Fig. S1B; F_5,50 = 3.27, P < 0.05).

FIG. 3.

Effect of iPSC-derived cells on motor function and immunohistochemical analysis of d18 and d25 iPSC-derived grafts. (A) Raw data showing performance of Control, Lesion, 410-d18, 411-d18, 410-d25, and 411-25 grafted rats on amphetamine-induced rotation test across the full test period, and (B) net rotation score at 24 weeks postgraft only. Based on histological analysis of the grafts from d18 and d25 preparations, we present (C) the total mean HuNu+ cells, (D) graft volume, (E) the density of TH+ cells per mm³, (F) total TH+ neurons, (G) percentage of TH+ cells out of total HuNu+ cells, (I) the total GIRK2+ cells, and (J) the percentage of GIRK2+ cells out of TH+ cells. The total AADC+ cell data are depicted in (L) and the percentage of AADC+ cells relative to HuNu+ cells is in (M). Representative immunohistochemistry is presented for (H) TH, (K) GIRK2 (blue) and HuNu (brown), and (N) AADC (brown) and HuNu (blue). P** ≤0.001, error bars = ±SEM. SEM, standard error of the mean.

Differences in graft size or composition did not explain the difference in efficacy between d18 and d25 precursor grafts

Immunohistological analysis was used to quantify the number of engrafted human cells (HuNu+), and cells positive for mature DA neuronal markers (TH+, AADC+, GIRK2+) that were detected in both d18 and d25 grafts (Fig. 3C–N). The analysis showed that none of these characteristics explained the difference in behavioral recovery between the d18 and d25 grafts (Fig. 3F).

The immunohistochemical analysis revealed variations dependent on cell line and/or stage (days in vitro); however, these characteristics could not explain the difference in functional efficacy between the d18 and d25 grafts. In general, d18 grafts were larger (Fig. 3C, D) and had more cells labeled for TH (Fig. 3F, G), GIRK2 (Fig. 3I), and AADC (Fig. 3L) (minimum F_1,11 = 6.02, P < 0.05). The density of TH+ neurons (Fig. 3E) and the percentage of GIRK2+ neurons (Fig. 3J) was also greater in d18 grafts (minimum F_1,11 = 6.32, P < 0.05). But comparison of similar-sized grafts makes clear that none of those factors was critical to the behavioral outcome. This is illustrated by comparing a group of grafts of similar size but with different behavioral outcomes. Day-25 (410-d25) and day-18 (411-d18) grafts had different behavioral outcomes, but the nonefficacious day-25 grafts were slightly larger (Fig. 3C, D), had comparable TH+ density (Fig. 3E), and contained more TH+ (Fig. 3F), GIRK2+ (Fig. 3I), and AADC+ neurons (Fig. 3L) than the day-18 grafts (minimum F_1,11 = 6.97, P < 0.05).

In summary, behavioral recovery did not correlate with the number of HuNu+ cells, graft volume, total number of TH+ cells, AADC+ cells, GIRK2+ cells, nor density of TH+ cells (Fig. 4F). Representative immunohistochemistry for TH+, GIRK2+, AADC+, and HuNu+ neurons is shown in Fig. 3H, K, and N. Figure 2B shows representative examples of TH immunolabeling for multiple d18 and d25 grafts at 24 weeks posttransplant.

FIG. 4.

Greater innervation from d18 grafts. (A) Differences were evident in neurite outgrowth from the d18 and 25 grafts into the surrounding host STR up to 1,400 μm from the graft border. (B–E) Representative images of TH+ neurites. (F) Functional recovery did not correlate with graft volume or cellular composition of the grafts. Functional recovery did correlate significantly with greater overall neurite outgrowth and greater fiber projections into both lateral and medial neostriatum (*P ≤ 0.01; **P ≤ 0.001). (G) Manual counting of fibers revealed 4-fold higher average density of projections from d18 grafts compared with d25 grafts. (H) Significantly more neurite outgrowth was also evident in d18 grafts when the tissue was analyzed using unbiased optical density estimates through computerized image analysis (ImageJ). Analysis of the y-intercept of projection trend lines confirmed significantly more neurite outgrowth from d18 grafts (Supplementary Fig. S1C). This is further supported by correlation analysis of both projection density measurements and optical density of projection fibers (Supplementary Fig. S1E–J).

Functional improvements correlated with significantly more extensive neurite outgrowth from grafts of earlier (d18) cultures than later cultures (d25)

Both manual and automated estimates of graft-derived TH+ fibers revealed significantly more neurite outgrowth and innervation from d18 grafts into the host brain than from d25 grafts (Fig. 4A, G, H; Supplementary Fig. S1C, D; minimum F_1,11 = 14.31, P < 0.01). Even when innervation data were normalized to graft size or TH+ neurons, more innervation was evident in d18 grafts, suggesting that differences in graft integration are due to underlying features of the transplanted cells and are irrespective of graft size or TH+ cell number (Supplementary Fig. S1D). Moreover, functional recovery on the amphetamine-induced rotation test correlated with all measures of neurite outgrowth (Fig. 4F, and Supplementary Fig. S1E–J). Representative images of neurite outgrowth are depicted in Fig. 4B–E.

Gene expression analysis identified stage-associated signatures of developing DA neuron precursors

An mRNA sequencing was used to compare phenotypic profiles of the cell cultures as they progressed through development toward DA neurons. Principal component analysis of the time course of differentiation (d0, d13, d18, d25) showed no overlap among the time points; PC1 separated the undifferentiated iPSC lines from the others and PC2 resolved the later time points in a consistent rank order (Fig. 5A). When only the three differentiation phases (d13, 18, 25), were compared, PC1 separated the three stages of differentiation for both cell lines, and PC2 distinguished the two cell lines that had similar but not identical profiles (Fig. 5B). The factor loading of the top- and bottom-ranked genes on these PC axes are listed in Supplementary Table S1A. Differential expression analysis comparing the two stages used in the animal studies (d18 and d25) showed 949 genes that were significantly [false discovery rate (FDR) <0.05] differentially expressed. Of these, 520 genes (FDR <0.05) were upregulated in d25 compared with d18 and 429 genes (FDR <0.05) were downregulated (Fig. 6A; Supplementary Table S1B: genes with negative log₂ fold change have higher expression at d25 relative to d18).

FIG. 5.

Gene expression analysis. RNA was sequenced at the two stages used for efficacy assessment (d18 and d25) and two earlier stages, day 0 (undifferentiated iPSCs) and d13 of differentiation. (A) Principal component one (PC1) explained 56.1% of the variance in transcriptome expression in the d0, d13, d18, and d25 dataset and largely separated the undifferentiated cells (triangles: day 0) from the differentiating dopaminergic precursors (circles, diamonds, and squares representing day 13, 18, and 25, respectively). The second axis (PC2) delineated the DA precursors in a rank order corresponding with their development duration. (B) Comparison of the differentiation stages (d13, d18, d25) showed a strong signal of development stage along PC1 (49.4% of transcriptome variance), and PC2 separated the two cell lines (410: blue and 411: orange) within a time point. (C) Transcription factor binding-site motifs in differentially expressed genes that were enriched in genes upregulated and downregulated between d18 and d25 of differentiation. Genes regulated by the repressors REST (RE1 Silencing Transcription Factor) and SUZ12 (SUZ12 Polycomb Repressive Complex 2 Subunit) were expressed at higher levels at d25. Day-25 cells had lower levels of genes regulated by cell cycle-associated factors FOXM1 and EFF4. (D, E) Normalized, zero-centered gene expression levels of REST and neuronal genes repressed by REST. As cells differentiated, REST decreased and REST targets increased in both iPSC line 410 (D) and iPSC line 411 (E). (F) List of neuron-associated genes with REST motifs upregulated at d25. EFF4, E2F transcription factor 4; FOXM1, Forkhead Box M1.

FIG. 6.

RNAseq analysis of cell preparations. (A) A volcano plot showing that a total of 520 genes were upregulated at day 25 relative to day 18 (pink). Four hundred twenty-nine genes were downregulated between day 18 and 25 (green) (FDR <0.05). (B) RNAseq data for cell preparations harvested at d13, d18, and d25, with genes grouped according to association with cell type or stage. Expression values are displayed as variance stabilized data (vsd), calculated using the vst command in the DESeq2 package [12]. FDR, false discovery rate.

The key processes active in d13, d18, and d25 cultures are listed in Fig. 6B. Gene Ontology analysis indicated that the main gene set that was upregulated at d18 relative to d25 was largely associated with proliferation. On d25, the upregulated genes had an overwhelming signal of synapse-related ontologies (Supplementary Table S1C).

Transcription binding-site motifs for the transcription factors E2F4, Forkhead Box M1 (FOXM1), SIN3A, and NFYA (Fig. 5C; Supplementary Table S1D) were enriched in the genes expressed at higher levels in day 18 cultures compared with the later stage (d25), which is consistent with a predifferentiated state. Among the genes upregulated at the later stage (d25) relative to d18, there was a strong enrichment for transcription factor binding-site motifs for REST, SUZ12, EZH2, and SMAD4 (Supplementary Table S1D). Additionally, as REST levels decreased between d18 and d25, expression of neuron-associated genes with REST motifs increased (Fig. 5D–F).

Discussion

Success of regenerative therapies using pluripotent stem cell-derived cell types is dependent on the demonstration of their functional efficacy, and the maturation stage and plasticity of the cultures is key to achieving this. The efficacy [13] and size [14] of hPSC-derived DA neuron precursor grafts has been linked to the length of time that the cells were differentiated before transplant. We took this concept a step further, using two different iPSC lines derived from people with PD at two time points during in vitro differentiation of DA neurons, and found that for both cell lines, the critical factor determining efficacy of the transplants was not the size or composition of the grafts, but the extent of neurite outgrowth into the host STR. Both cell lines at both time points, 18 or 25 days of differentiation, engrafted well. The grafts were variable in volume, density, and quantity of iPSC-derived dopaminergic neurons, but these characteristics did not correlate with the functional efficacy. Only the cells from the earlier time point were efficacious, and differed from the later time point's ineffective grafts in only one characteristic: the ability to react to the host brain by producing extensive neurite growth.

These results suggest that the efficacious cells were in a poised state, activated to extend neurites when they encountered target cells in the STR, while the ineffective cells, while able to survive, had decreased ability to form functional connections with the host brain. To investigate this idea, we analyzed the cells in vitro by gene expression profiling. Gene expression profiling predicts the differentiation abilities of pluripotent stem cells [15,16] and it is likely to prove useful for prediction of other qualities, such as posttransplantation characteristics. Our results are consistent with the idea that the cultures that were functionally effective were poised to begin to mature, but had not quite initiated the processes leading to functional neurons. One gene that is of particular interest is REST (RE1 silencing transcription factor), which codes for a transcription factor that acts as a repressor of genes involved in neural maturation; its expression is thought to allow a pool of neural precursors to accumulate during processes of neural differentiation in embryogenesis [16].

In our cultures, REST decreased considerably at d25; in addition, at d25, genes with REST transcription factor-binding motifs were upregulated, which is consistent with removal of REST-mediated repression. REST expression and expression of the regulated genes could prove to be useful markers to assess an optimal stage of transplantation of future cell therapy products.

Some genes that were specifically upregulated in d18 cells are associated with neurite outgrowth, including LIN28A, FLRT3, and ITGA5. LIN28A (log2FC = 3.8) codes for a posttranscriptional regulator of miRNAs associated with embryogenesis and has been linked to axonal regeneration [17–20]. Overexpression of LIN28A in DA neurons has been reported to increase dendrite length, graft volume, and TH+ content, and enhance functional recovery posttransplantation [21]. FLRT3, which was upregulated in d18 cells, is implicated in neurite outgrowth and has been identified as a positive regulator of FGF signaling and cell adhesion [22,23]. FLRT3 codes for a coreceptor for Robo1; the attractive response to the guidance cue Netrin1 has been shown to be controlled by Slit/Robo1 signaling and by FLRT3 [23,24]. Thus, the expression of FLRT3 may promote neurite outgrowth from the grafted d18 precursors. ITGA5 codes for subunit alpha 5 in the integrin alpha chain family (Integrin α5β1), which has been identified as having a role in specific dopaminergic neuron outgrowth onto striatal neurons [25].

Thus, given that the expression of REST-regulated genes may be key to identifying the optimal state for transplantation, and several key genes associated with neurite outgrowth are differentially expressed between d18 and d25, there is clear impetus to better understand the role these genes play as both biomarkers of optimal transplantation and as pathways for modulation to optimize graft efficacy.

It is important to note that, despite working with autologous cell therapy products, using cells harvested from people with PD, the results of the current study are generalizable to all hPSC-derived DA cell therapy products. For example, it is well established that cell maturity can impact upon on graft viability and functionality [eg, (13,26)]. Previous studies have demonstrated that individual cell therapy products are likely to each have an optimal day of harvest for transplantation, making it critical to determine this time point for each cell line or batch.

In contrast to our study, however, a previous study investigating the effect of differentiation state on graft outcome used one cell line and identified graft size and total TH+ neuron content as the key factor driving graft efficacy [13]. In this study, we present evidence, across two cell lines, that similarly sized grafts, with comparable TH neuron content, may not produce equivalent behavior. Instead, the key determinate is the extent to which grafts integrate into the host brain, not the size of the graft per se. Finally, we present corresponding gene expression data to further characterize these distinct cell products and we identify REST as a potentially exciting biomarker of optimal cell state.

Our study is unique in that we have distinguished two distinct cell preparations that are both capable of surviving long-term in vivo and producing DA neuron-rich grafts in a rat model of PD, but observed that only one of the cell profiles restored motor function. The effective cells behaved differently in the host brain, producing extensive neurite outgrowth. It is important to note that different differentiation protocols result in distinct maturation timescales, so the specific number of days in culture that produces optimal in vivo transplants will differ according to the protocol [13,26–28]. Our work does suggest, however, that it is likely that the optimal time of differentiation before transplantation may be identifiable for all protocols. Phenotyping cultures by gene expression analysis offers insights into the characteristics that distinguish cells that develop into effective grafts from those that fail, and is useful for discovering markers that can predict the success of cultures.

Gene expression profiles may be used to identify an optimized product to ensure extensive neurite outgrowth after grafting and consequent functional recovery. This is especially important for autologous cell replacement therapy for PD [29] for which effective cell types must be derived reproducibly from each graft recipient.

Author Disclosure Statement

There are no conflicts of interest. All of the data were generated at Scripps Research Institute or Cardiff University with the financial support indicated. J.F.L. and A.B.L. are stockholders and founders of Aspen Neuroscience, Inc., (Aspen). H.T., R.M.W., and J.A.M. are stockholders in Aspen.

Funding Information

M.J.L. was supported by a Parkinson's UK Senior Research Fellowship (F-1502); R.H. was supported by a grant from Summit for Stem Cell Foundation; and J.F.L., H.T., A.B.L., and R.M.W. were supported by grants from Summit for Stem Cell Foundation and the California Institute for Regenerative Medicine (CIRM; CL1-00502, RT3-07655, GC1R-06673-A). J.F.L., H.T., A.B.L., R.M.W., and D.G.S. were supported by CIRM (DISC2-09073); P.P.S. and I.V.S. were supported by NIH (DA046170, DA046204-04, DA043268).

Supplementary Material

Supplementary Methods

Supplementary Figure S1

Supplementary Table S1A

Supplementary Table S1B

Supplementary Table S1C

Supplementary Table S1D