Origin of the Induced Pluripotent Stem Cells Affects Their Differentiation into Dopaminergic Neurons

Neuronal differentiation of human induced pluripotent stem (iPS) cells, both in 2D models and 3D systems in vitro, allows for the study of disease pathomechanisms and the development of novel therapies. To verify if the origin of donor cells used for reprogramming to iPS cells can influence the differentiation abilities of iPS cells, peripheral blood mononuclear cells (PBMC) and keratinocytes were reprogrammed to iPS cells using the Sendai viral vector and were subsequently checked for pluripotency markers and the ability to form teratomas in vivo. Then, iPS cells were differentiated into dopaminergic neurons in 2D and 3D cultures. Both PBMC and keratinocyte-derived iPS cells were similarly reprogrammed to iPS cells, but they displayed differences in gene expression profiles and in teratoma compositions in vivo. During 3D organoid formation, the origin of iPS cells affected the levels of FOXA2 and LMX1A only in the first stages of neural differentiation, whereas in the 2D model, differences were detected at the levels of both early and late neural markers FOXA2, LMX1A, NURR1, TUBB and TH. To conclude, the origin of iPS cells may significantly affect iPS differentiation abilities in teratomas, as well as exerting effects on 2D differentiation into dopaminergic neurons and the early stages of 3D midbrain organoid formation.


Introduction
One of the major goals in neurobiology is to understand the development of the nervous system and to model cellular interactions within the human brain.
The introduction of specific genes encoding transcription factors, such as MYC, OCT3/4, SOX2 and KLF4, could reprogram human somatic cells into induced pluripotent stem (iPS) cells, and subsequently, iPS cells can be differentiated into any type of human cell [1]. Generated iPS cell lines can be characterized by the expression of endogenous pluripotency markers, activity of alkaline phosphatase and the capabilities of the cells to form embryoid bodies in vitro and teratomas in vivo [1][2][3]. The technology of generating iPS cells from somatic cells has opened new perspectives, such as cellular replacement, regenerative therapy and disease modeling [3].
Our main goal was to compare iPS cell lines generated from different cellular origins. Both keratinocytes and PBMCs seem to be very promising cell types for reprograming, as they can be easily isolated and efficiently reprogrammed [21]. The highest safety can be achieved using the Sendai viral vector, because it does not integrate into the genome [22,23].
To isolate keratinocytes, up to 20 hairs from a healthy volunteer were collected. After cell culturing of the hairs' outer root sheath, a sufficient number of cells was acquired, within two to three weeks. Subsequently, the cells were transduced with the Sendai viral vector. Four days after transduction ( Figure 1AI), the cells changed their morphology, a process which characterizes the early stage of reprogramming ( Figure 1AII). The first colonies appeared on day 14, but numbers of colonies were not fully reprogrammed (pre-iPS) ( Figure 1AIII). The fully reprogrammed keratinocyte-derived iPS (iPS-K) colonies, which were round and tightly packed, were detected around day 18. After 28 days, the colonies were ready to transfer ( Figure 1AIV).  After isolation of PBMCs from whole blood, the cells were cultured in the expansion medium ( Figure 1BI) and then, on day 0, they were transduced. On day 2, the morphology of the infected cells did not change ( Figure 1BII). Between days 5 and 10 small colonies were noticeable (see Figure 1BIII). When iPS colonies were ready to transfer, their edges were regular and small cells were tightly packed ( Figure 1BIV). Acquired PBMC-derived iPS (iPS-P) clones were expanded up to the 10th passage before further analysis.
Firstly, the presence of the Sendai virus genome was investigated. At the 10th passage, all the selected iPS lines were Sendai-virus-free, which suggested that viral transgenes (SeV) were silenced ( Figure 1D). The iPS cell line, after reprogramming, at an early passage, was used as a positive control, whereas cancer cells (HCT116) were used as a negative control. The generated iPS cell lines expressed endogenous pluripotency markers at the mRNA level ( Figure 1C), such as homeobox protein NANOG, octamer binding transcription factor 3 (OCT3) and telomerase (TERT). A commercially available protein-induced iPS (piPS) cell line was used a positive control, whereas cancer cells (HTC116) served as a negative control. Furthermore, all analyzed iPS clones displayed activity of alkaline phosphatase, as shown in Figure 1E.
After characterization of iPS cells, the transcription and methylation profiles were analyzed in the selected clones of iPS-K and iPS-P cells ( Figure 2). The piPS cell line was our commercial control, generated from fibroblasts by protein introduction of transcription factors. The transcription profile, based on microarray analysis, showed differences for the 100 differentially expressed genes ( Figure 2A). The violin plot ( Figure 2B) showed that all three samples have similar methylation profiles. However, the methylation heatmap showed that iPS-P was clustered separately, compared to iPS-K and piPS samples ( Figure 2C). All of the plots were created using average methylation beta values (0-1).
the selected clones of iPS-K and iPS-P cells ( Figure 2). The piPS cell line was our commercial control, generated from fibroblasts by protein introduction of transcription factors. The transcription profile, based on microarray analysis, showed differences for the 100 differentially expressed genes ( Figure  2A). The violin plot ( Figure 2B) showed that all three samples have similar methylation profiles. However, the methylation heatmap showed that iPS-P was clustered separately, compared to iPS-K and piPS samples ( Figure 2C). All of the plots were created using average methylation beta values (0-1).  Heatmap for mRNA expression microarray data after filtering and NEQC (normexp background correction and quantile normalization) normalization for 3 samples of iPS cells (piPS, iPS-K, iPS-P). Yellow color indicates genes displaying lower expression and red refers to genes with higher expression. (B) Violin plot showing the distribution of average methylation β-value for 3 samples (piPS, iPS-K, iPS-P). (C) Manhattan and correlation models were used to generate sample clustered heatmap of iPS cell lines (piPS, iPS-K, iPS-P) after methylation analysis. For cluster analysis, the most variable CpG probes were selected based on the s.d. of the β-values. The heatmap scale is from 0 to 1, where the green color indicates unmethylated sites and red refers to hypermethylated sites (β-value = 1).

iPS Cells of Different Origin Form Embryoid Bodies and Three Germ Layers with Similar Effectiveness
To validate the effects of the different origins of cells on the capabilities of iPS cells to differentiate, embryoid bodies were formed from iPS cell lines. This is a common method for induction of differentiation towards various kinds of somatic cells and is a standard method used to validate iPS pluripotency [1,24,25]. All selected clones from both origins formed three-dimensional aggregates in suspension without any noticeable differences ( Figure 3A).
Subsequently, markers of three germ layers and extraembryonic tissues (such as GBX2, HAND1, SOX17 and Brachyury) were investigated at the mRNA level ( Figure 3B,C). Brachyury is a transcription factor in early mesodermal cells [26]. HAND1 is a transcription factor critical for specification of extraembryonic tissues (trophoblasts) [27,28]. SOX17 is a transcription factor that plays an important role in early endoderm development [29]. GBX2 is the early ectodermal lineages marker [30,31]. We observed large differences in the investigated genes between individual clones, which resulted in large variations within the groups. Nevertheless, no statistically significant differences between iPS-K and iPS-P were detected in the expression of selected markers on day 4 and 6 of differentiation ( Figure 3B,C).
Subsequently, markers of three germ layers (such as CD140b, CD144-mesoderm; SOX2, PAX6-ectoderm; SOX17, CD184-endoderm) were also investigated at the protein level after differentiation of iPS-K and iPS-P cells in vitro ( Figure 4A). Flow cytometric analysis showed similar expression levels of the markers, characteristic of the first stage of differentiation into three germ layers for all three clones of iPS-K and three clones of iPS-P ( Figure 4B). The analysis confirmed the RT-qPCR analysis performed on embryoid bodies. No significant differences were detected at the early stage of differentiation into three germ layers at the protein level. specification of extraembryonic tissues (trophoblasts) [27,28]. SOX17 is a transcription factor that plays an important role in early endoderm development [29]. GBX2 is the early ectodermal lineages marker [30,31]. We observed large differences in the investigated genes between individual clones, which resulted in large variations within the groups. Nevertheless, no statistically significant differences between iPS-K and iPS-P were detected in the expression of selected markers on day 4 and 6 of differentiation ( Figure 3B,C). HAND1-marker of early trophoblast differentiation; SOX17-endodermal marker; BRACHYURYmesodermal marker. GAPDH was housekeeping gene control and the data were calculated using the ΔCt method. Significant differences between embryoid bodies (EBs) of different origin were not observed. The graph data show the results from 4 clones in duplicates, collected on day 4 (n = 8). The data represent the mean ± SEM. (C) Analysis of mRNA expression levels of markers of three germ layers in embryoid bodies on day 6. Significant differences between EBs of different origin were not GAPDH was housekeeping gene control and the data were calculated using the ∆Ct method. Significant differences between embryoid bodies (EBs) of different origin were not observed. The graph data show the results from 4 clones in duplicates, collected on day 4 (n = 8). The data represent the mean ± SEM. (C) Analysis of mRNA expression levels of markers of three germ layers in embryoid bodies on day 6. Significant differences between EBs of different origin were not observed on day 6. The graph data show the results from 3 clones, collected on day 6 (n = 3). The data represent the mean ± SEM. PAX6-ectoderm; SOX17, CD184-endoderm) were also investigated at the protein level after differentiation of iPS-K and iPS-P cells in vitro ( Figure 4A). Flow cytometric analysis showed similar expression levels of the markers, characteristic of the first stage of differentiation into three germ layers for all three clones of iPS-K and three clones of iPS-P ( Figure 4B). The analysis confirmed the RT-qPCR analysis performed on embryoid bodies. No significant differences were detected at the early stage of differentiation into three germ layers at the protein level.

Differentiation of iPS Cells in Teratomas Is Dependent on Origin of iPS Cells
The iPS-K and iPS-P cell lines were subjected to teratoma formation assays in immunodeficient NOD-SCID mice. Histopathological analysis of tumor slices enabled us to observe structures characteristic of all three germ layers within the tumors ( Figure 5A). Subsequently, we analyzed the amount of tissue-specific structures in the generated teratomas ( Figure 5B). We observed that in teratomas from iPS-K the most numerous structure was neuroectoderm, whereas in teratomas from iPS-P the most numerous structure was the secretory epithelium. The average amounts of the indicated structures in teratomas from four different clones between iPS-K and iPS-P are compared in Figure 5C. We also noticed that iPS-P-derived teratomas tend to form more structures of pigmented cells and cartilage. In iPS-K-derived teratomas, we observed a higher number of neuroectoderm-like

Differentiation of iPS Cells in Teratomas Is Dependent on Origin of iPS Cells
The iPS-K and iPS-P cell lines were subjected to teratoma formation assays in immunodeficient NOD-SCID mice. Histopathological analysis of tumor slices enabled us to observe structures characteristic of all three germ layers within the tumors ( Figure 5A). Subsequently, we analyzed the amount of tissue-specific structures in the generated teratomas ( Figure 5B). We observed that in teratomas from iPS-K the most numerous structure was neuroectoderm, whereas in teratomas from iPS-P the most numerous structure was the secretory epithelium. The average amounts of the indicated structures in teratomas from four different clones between iPS-K and iPS-P are compared in Figure 5C. We also noticed that iPS-P-derived teratomas tend to form more structures of pigmented cells and cartilage. In iPS-K-derived teratomas, we observed a higher number of neuroectoderm-like structures and collagen fibers. Interestingly, structures characteristic of the mesoderm, such as bones and muscles, were detected only in teratomas generated from iPS-P.
These results confirm the efficient reprogramming of both keratinocytes and PBMC, and indicate the differences in the amount of three-germ-layer structures that could depend on the origin of iPS cells. structures and collagen fibers. Interestingly, structures characteristic of the mesoderm, such as bones and muscles, were detected only in teratomas generated from iPS-P. Graph presenting average amounts of structures in teratomas generated from iPS-K and iPS-P. Statistically significant higher amount of secretory epithelium was observed in teratomas from iPS-P (* p < 0.05), whereas a higher amount of neuroectoderm was noticed in teratomas generated from iPS-K. The photographs show representative images. The graph data represent the mean ± SEM.

Origin of iPS Cells Affects Early Stages of Differentiation into 3D Midbrain Organoids
Midbrain organoids were generated according to the protocols of Jo. et al., 2016, with our previous modifications [1,6]. The morphology of both kinds of organoids was similar ( Figure 6A), but the hematoxylin/eosin staining showed differences in their structures ( Figure 6B). More neuroectodermal rosettes were observed in organoids generated from iPS-K in each timepoint ( Figure 6D).
To further characterize the organoids, on day 27 we performed immunofluorescent staining for beta-III-tubulin (TUBB), which is a neuronal marker [32,33], and tyrosine hydroxylase (TH), which is a marker of dopaminergic neurons [34,35] ( Figure 6C). We detected both TUBB and TH in both kinds of organoids. Furthermore, some of the TUBB-positive neurons were also positive for TH, which suggests that we were able to generate human midbrain organoids from both kinds of cells. Nevertheless, TH protein levels were similar in organoids generated from iPS-P and iPS-K cells ( Figure 6C).
On days 4, 17, 27 and 39 we collected three organoids for evaluation of five markers at mRNA expression levels, such as LIM homeobox transcription factor 1 alpha (LMX1A), forkhead family of winged-helix transcription factor 2 (FOXA2), orphan nuclear receptor (NURR1), tyrosine hydroxylase (TH) and beta-III-tubulin (TUBB) ( Figure 6D). Expression of the FOXA2 gene was detected only on day 4 and was significantly higher in organoids from iPS-P cells (6.4-fold). On day 17, 27 and 39 expressions of LMX1A tended to be higher in organoids from iPS-K cells, but statistically a significant difference was observed only on day 27 (4.96-fold). From day 4 to day 39, the level of NURR1 expression in both kinds of organoids was rising, and differences between origin of iPS were not detected. On day 17, 27 and 39 we detected expression of TH gene. On day 17 the level of TH expression was lower than on day 27 and 39. On day 27 and 39 TH gene was expressed at the same level in both groups of organoids. On day 4, level of TUBB was at the low level. Then on day 17, 27 and 39 levels of TUBB were similarly high and stayed approximately at the same level for both groups in all timepoints.
These results suggest that the origin of iPS cells may have an impact only on early stages of differentiation in 3D organoids, but not the whole differentiation process.  and markers of neurons (NURR1, TH, TUBB) in organoids from iPS-K and iPS-P. Representative data show results from 3 organoids (n = 3) in each timepoint. Statistically significant differences were observed in expression levels of FOXA2 on day 4 and LMX1A on day 27. On day 4, higher expression of FOXA2 was noted in organoids from iPS-P, * p < 0.05. On day 27, higher expression of LMX1A was observed in organoids from iPS-K, *** p < 0.001. Relative expression of genes was calculated by RT-qPCR with the ∆Ct method, using GAPDH as a constitutive control. Statistically significant differences were analyzed using Student t-test. The experiments were performed using two clones of each type of iPS cells. Replicates for each clone were performed twice. In each experiment three organoids were collected for analysis. The data represent the mean ± SEM. TUBB-beta-III-tubulin, TH-tyrosine hydroxylase, NURR1-nuclear receptor related 1 protein, LMX1A-LIM homeobox transcription factor 1 alpha, FOXA2-forkhead box A2.

Origin of iPS Cells Affects Their Capability to Differentiate into 2D Dopaminergic Neurons
Subsequently, we differentiated two selected iPS clones of each origin into dopaminergic neurons according to the 2D protocol described by Kriks et al., 2011 [5]. On day 11 neural progenitors were observed. Then, on day 20, cells with characteristic neuronal morphology started to appear. On day 30, mature dopaminergic neurons were detected ( Figure 7A).
The presence of dopaminergic neurons was confirmed by TH expression at the protein level ( Figure 7B). Therefore, staining for TH and TUBB was performed. Interestingly, the level of TH tended to be slightly increased in neurons generated from iPS-K cells ( Figure 7B). Furthermore, on days 11, 20 and 30, mRNA expression levels of FOXA2, LMX1A, NURR1, TH and TUBB were analyzed ( Figure 7C). During midbrain differentiation, co-expression of the roof plate marker LMX1A and the floor plate marker FOXA2 should be visible [5]. Cells from both iPS-K and iPS-P showed the expression of FOXA2 on day 11, and then the expression decreased. On day 30, the increased levels of FOXA2 in neurons generated from iPS-P were observed in contrast to neurons generated from iPS-K, as their level of FOXA2 was similar to the level from day 20. Furthermore, a significantly higher level of FOXA2 on day 20 (6.83-fold) and 30 (6.71-fold) was observed in cells from iPS-P cells. We found that, from day 11 to 30, the level of LMX1A in neural generation from iPS-K cells was rising. On day 30, we detected a significantly higher level of LMX1A (3.70-fold) in neurons from iPS-K, compared to neurons from iPS-P, whereas in cells generated from iPS-P, LMX1A expression on day 11 was higher than in cells generated from iPS-K, and then on day 20 it decreased, and on day 30 it increased again to the level from day 11. Cells from both groups showed increasing NURR1 levels over time. On day 20 expression of NURR1 was higher (2.63-fold) in cells generated from iPS-P, in contrast to day 30, when level of NURR1 was significantly higher in neurons generated from iPS-K (13.14 fold). The mRNA level of TH increased over time and was similar in neurons generated from iPS cells of both origins. Level of TUBB also rose over time for both groups. On day 11 and 20 the same level of TUBB for neurons generated from iPS-K and iPS-P was observed, whereas on day 30 expression of TUBB gene was significantly higher (2.04-fold) in neurons generated from iPS-K.
These results demonstrate that during neural generation in a 2D model, differences at the mRNA and protein levels of neuronal markers may appear, depending on origin of iPS cells.  (C) Gene expression of neuronal progenitors (FOXA2, LMX1A) and neurons markers (NURR1, TH, TUBB) in dopaminergic neurons generated from iPS-K and iPS-P in each timepoint (day 11, 20 and 30 of differentiation). Statistically significant higher expression of FOXA2 gene was detected on day 20 and 30 in cells generated from iPS-P cells compared to iPS-K cells, ** p < 0.01, * p < 0.05. Higher expression of LMX1A was observed on day 30 in neurons generated from iPS-K, * p < 0.05. On day 20, statistically significant higher expression of the NURR1 gene was observed in cells from iPS-P cells, but on day 30 higher expression of that gene was observed in neurons from iPS-K cells, * p < 0.05, ** p < 0.01. The expression of TUBB was definitely higher in neurons from iPS-K on the last day of differentiation and the differences were statistically significant, **** p < 0.0001. Relative expression of genes was calculated by RT-qPCR using the ∆Ct method with GAPDH as constitutive control. Statistically significant differences were analyzed using Student's t-test. The experiments were performed for two clones from every source of iPS cells. Replicates for each clone were performed twice (n = 4). The data represent the mean ± SEM. TUBB-beta-III-tubulin, TH-tyrosine hydroxylase, NURR1-nuclear receptor related 1 protein, LMX1A-LIM homeobox transcription factor 1 alpha, FOXA2-forkhead box A2.

Discussion
Generating neurons in vitro provides a great opportunity to study neurodevelopment and pathogenesis of neurological diseases, and can be used in therapeutic approaches. Nowadays, many protocols for generating dopaminergic neurons in 2D models have been described. Efforts are being made to achieve the highest efficiency, standardization and reproducibility [4,5]. A new and also promising technique is generating 3D organoids, which consist of different kinds of cells apart from neurons, creating a more physiological environment for developing neurons. Three-dimensional neural tissue can mimic neurogenesis and the formation of particular neural structures [5,6]. Considering all applications of both models, their reproducibility is a very important feature. Our study examined whether the origin of iPS cells has an influence on their differentiation into neurons in 2D and 3D models, and, thus, on the reproducibility of the protocols.
The differences in transcriptional and epigenetic patterns in iPS cells of different origins, using a murine model, were firstly proposed in 2010 by Polo et al. [18]. Since that time, many publications have shown that iPS cells have epigenetic memory of their tissue of origin, which may affect their capability to differentiate [14,36,37]. On the other hand, there are also manuscripts suggesting that epigenetic memory has no impact on the generation of iPS cell lines from different cell origins [38], or indicating no impact of iPS origin on 2D differentiation to neuronal progenitor cells [39], so further research on that subject is still required. Our transcription and methylation analysis of iPS cells indeed showed differences between iPS cell lines of different origins. However, our other results demonstrated that at the level of basic characteristic of pluripotency features, there were no differences between iPS cells of different sources. Alkaline phosphatase staining, expression of endogenous pluripotency markers, embryoid body formation and differentiation into three-germ layers gave similar results for all investigated clones of iPS cell lines. Similarly in the literature, no significant changes have been observed in the basic characteristics of the iPS lines, both between clones of cells from one person and clones of the same source from different people [40,41]. Nevertheless, we noticed differences in the amount of structures characteristic for particular germ layers in the formed teratomas. More ectodermal structures were observed in teratomas formed from the keratinocyte-derived-iPS cell line, whereas some mesodermal structures, such as bones and muscles, were only detected in the PBMC-derived-iPS cell lines. These results suggest that the ectodermal origin of iPS cells may enhance their differentiation capabilities towards ectodermal structures. Importantly, the keratinocytes' origin promoted the formation of higher numbers of neuroectodermal structures. Furthermore, in comparison to the literature, our method of analyzing teratoma structures is a novel and unique approach [42,43].
After the first signal that differences can appear in teratomas, further analysis included evaluation of iPS cells' capabilities to differentiate into neurons. We selected two clones from every source of iPS cells, which displayed the highest variation between the selected structures in teratomas. Then, we started neural differentiation. Earlier research papers have also demonstrated that tissue of origin of iPS cells and their epigenetic signatures have an impact on differentiation preferences [13,14]. However, one paper suggested that it is not the cell type of origin, but rather, clones of iPS cells which may affect differentiation processes [44]. In our studies, we did not observe clonal differences in neuronal differentiation (data not shown), but we detected differences dependent on the origin of iPS cells. Both in 2D and 3D differentiation, we observed differences in gene expression levels during the process of neuron generation. These results are additionally supported by the higher levels of neuroectodermal structures formed in teratomas from keratinocytes-derived iPS cells.
An important aspect in the process of differentiation to dopaminergic neurons is the expression of neuronal progenitors' markers. LMX1A and FOXA2 are early markers of midbrain floor plate progenitors [5]. FOXA2 is required for in vivo development of dopaminergic neurons, whereas LMX1A is crucial factor in the development of dopaminergic neurons, which also regulates the survival of adult dopaminergic neurons [45][46][47]. NURR1 is responsible for differentiation of early progenitors to mature dopaminergic neurons [48]. Moreover, the interaction of FOXA2 and NURR1 protects neurons against toxins [49].
In both neuronal differentiation types, we observed statistically significant differences in FOXA2 and LMX1A levels, which are markers of midbrain floor plate progenitors [5]. Therefore, we can conclude that the source of iPS cells may have a significant impact on the initial phase of neural differentiation in both 2D and 3D models.
In later stages of midbrain organoid formation, we did not observe any differences in gene expression. The 3D midbrain organoids displayed similar levels of TUBB-and TH-positive neurons. In contrast, on the last day of 2D differentiation to dopaminergic neurons, the expression of NURR1 and TUBB was definitely higher in neurons generated from keratinocyte-derived iPS cells. An important conclusion to draw from our data is that in 2D differentiation to dopaminergic neurons, the origin of iPS cells may also have an impact on the final results. The significant effects of the origin of iPS cells on their neuronal differentiation capabilities in the 2D model are in agreement with previous research showing distinct origin-dependent neural cell identities after 2D differentiation of iPS cells, despite the lack of differences in TUBB and TH protein levels [12].
In contrast to the 2D protocol, in 3D midbrain organoids, we did not observe any differences between organoids at the last stage of differentiation. The differences were visible only in the first steps of differentiation, which suggests that the cells of differing origins may have a distinct timeframe of differentiation, although the final result is the same.
In previous research, both 2D and 3D differentiation protocols for different structures were compared [50,51]. Nevertheless, in our work, we compared these two types of differentiation models in the context of the origin of iPS cells for the first time. In the 3D model we observed statistically significant differences only at the early stages of differentiation. It is possible that differentiation in structures resembling the natural environment in developing tissue may slightly attenuate the effect of epigenetic differences dependent on the origin of iPS cells, which was also suggested by other groups [12,52].

Cell Culture
Induced pluripotent stem (iPS) cells were generated from keratinocytes and PBMCs of the same donor, or were bought (piPS, generated from fibroblasts, SBI System Biosciences, Palo Alto, CA, USA). iPS cells were cultured in serum-free iPS medium containing DMEM/F12, 20% KSR, 2 mM Glutamax, 100 U/mL Penicillin/Streptomycin, 100 µM non-essential amino acids, 10 ng/mL bFGF (all from Thermo Fisher Scientific, Waltham, MA, USA) and 100 µM β-mercaptoethanol (Sigma-Aldrich, Saint Louis, MO, USA). iPS cells were cultured feeder free on Matrigel (Corning, New York, NY, USA) or on feeder layers of inactivated mouse embryonic fibroblasts (iMEFs) on dishes coated with gelatin (Sigma-Aldrich). Medium was changed every day. Routine passages were done using Accutase (Lonza, Basel, Switzerland). After the passage, iPS cells were seeded at a density of 1:4-1:10 in medium for iPS cells supplemented with 10 µM ROCK inhibitor Y-27632 (Sigma-Aldrich). Before freezing, iPS cells were incubated with 10 µM ROCK inhibitor for one hour on a culture dish. Then, iPS cells were suspended in freezing medium containing 90% fetal bovine serum (FBS; Eurx, Gdansk, Poland), 10% DMSO (Sigma-Aldrich) and 10 µM ROCK inhibitor (Sigma-Aldrich), frozen and cryopreserved in liquid nitrogen.
Mouse embryonic fibroblasts (MEFs) (AMSBIO, Abingdon, United Kingdom) were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 4.5 g/L glucose and supplemented with 10% FBS, 2 mM L-glutamine and 100 U/mL Penicillin/Streptomycin (all from Thermo Fisher Scientific). To generate the feeder layer for iPS culture, MEF cells were inactivated with mitomycin C (Sigma-Aldrich). To generate MEF-conditioned iPS medium, MEF medium was incubated for 24h on MEF cells. Then, medium was centrifuged at 2000 rpm for 10 min and supplemented with 10 ng/mL bFGF (Thermo Fisher Scientific).
HTC116 is a human colon colorectal cancer that was used as a negative control for pluripotency features. This cell line was cultured in Dulbecco's modified Eagle's medium (DMEM) containing 4.5 g/L glucose and supplemented with 10% FBS, 2 mM L-glutamine and 100 U/mL Penicillin/Streptomycin (all from Thermo Fisher Scientific). All cells were cultured at 37 • C in a humidified atmosphere and 5% CO 2 level.

Generation of iPS Cells from Keratinocytes and PBMC
The Jagiellonian University Bioethical Committee in Kraków approved the study using human samples-decision number KBET/173/B/2012 on 31 May 2012, with two prolongations from 17 December 2015 and 23 June 2016. Blood and hair were isolated from a healthy volunteer who gave informed consent to participate in the study.
PBMCs were isolated by centrifugation of blood on gradient of Pancoll (PAN Biotech, Aidenbach, Germany), as described previously [1]. The acquired layer of mononuclear cells isolated from blood was washed three times with Phosphate Buffered Saline (PBS, Eurx). PBMCs were cultured in the expansion medium, described previously [1]. After 7 days, PBMCs were transduced with the Sendai viral vector from CytoTune ® 2.0 Sendai Reprogramming Kit (Thermo Fisher Scientific). Forty-eight hours after transduction, the cells were plated on inactivated mouse embryonal fibroblasts (iMEF). The first colonies (pre-iPS) appeared between 5 and 10 days. Mature colonies, ready to transfer, appeared around day 28. After 28 days, the colonies were picked and transferred to separate wells. After the 10th passage, iPS lines were cryopreserved and tested.
Keratinocytes were isolated from the outer root sheath (ORS) of the plucked hair, according the the protocols of Aasen et al., 2010, with modifications described previously by Sulkowski et al., 2018 [2,40]. The isolated hairs were rinsed with antimycotic-antibiotic solution (Thermo Fisher Scientific). Then, the hairs were cut and placed in a dish with coated with Matrigel (Corning) in MEF-conditioned iPS medium. The medium was changed until keratinocytes were surrounding the hair. Then, medium was changed to EpiLife (Thermo Fisher Scientific) and cells were cultured on dishes coated with Matrigel. For the dissociation of the cells, TrypLE solution (Thermo Fisher Scientific) was used. Keratinocytes displaying 40-60% confluence were reprogrammed using the Sendai viral vector-CytoTune ® 2.0 Sendai Reprogramming Kit (Thermo Fisher Scientific), as described previously [2]. Twenty-four hours after transduction, the medium was changed. Seven days after transduction, cells were plated on iMEF in iPS medium with 1 mM sodium butyrate (Sigma-Aldrich). Sodium butyrate was added to the medium until day 8. On day 11, the first iPS (pre-iPS) cell colonies were detected. Between days 18 and 35, iPS colonies were picked and transferred to separate wells in iPS medium with 10 µM ROCK inhibitor Y-27632 (Sigma-Aldrich). After the 10th passage, iPS cells were ready for further experiments.

RNA Isolation and Reverse Transcription
RNA was isolated using the Universal RNA/miRNA Purification Kit (Eurx), according to vendor's protocol. RNA was reverse transcribed to cDNA using the Moloney Murine Leukemia Virus reverse transcription kit (M-MLV Reverse Transcriptase, Promega, Madison, WI, USA), according to vendor's protocol.

Alkaline Phosphatase (AP) Activity
iPS cells were fixed in 4% paraformaldehyde for 10 min. After fixation, the cells were washed three times with PBS and incubated in AP staining solution NBT-BCIP (Roche, Basel, Switzerland), 5 M NaCl, 1 M Tris-HCl (pH 9.5), 1 M MgCl 2 for 10 min at room temperature in the dark. At the end, the cells were washed two times. The stained cells were analyzed using an IX70 microscope (Olympus Corporation, Tokyo, Japan) and images were collected using CellSensDimension software.

Gene Expression and Methylation Quantification
After isolation, RNA quality and concentration from all samples (3 samples: piPS, iPS-K and iPS-P) were checked on an Agilent TapeStation System (Agilent Technologies, Santa Clara, CA, USA) and a Promega QuantiFluor Dye System on a Quantus Fluorometer (Promega). To generate biotinylated cRNA from 300 ng of isolated RNA, the TargetAmp-Nano Labeling Kit for Illumina Expression BeadChip (Epicentre-an Illumina Company, Madison, WI, USA) was used. Prepared probes were fragmented and hybridized to an Illumina Whole Genome Expression Chip, HumanHT-12 v3.0. Then, BeadChips were washed and subsequently scanned. Gene expression profiles were made from 750 ng hybridized cRNA to Illumina HumanHT-12 v3.0 BeadChips. The hybridization assay was performed according to the Illumina whole-genome gene expression hybridization assay guide (Illumina, San Diego, CA, USA).
The heatmap was made after NEQC normalization and poor-quality samples were filtered out. Results were obtained for the 100 most variable genes.
The DNA methylation analysis was isolated by means of a QIAGEN QIAamp DNA Mini-Kit (QIAGEN, Hilden, Germany). The methylation assay was made according to the Infinium HD Methylation Assay Protocol Guide manual workflow for the Infinium Methylation BeadChips (Illumina) The data for 3 samples (piPS, iPS-K and iPS-P) were analyzed and visualized using GenomeStudio software (Methylation Module v1.8). Detected CpGs (0.05) for all samples were >95%. The methylation status of each CpG site is reported as the average B-value (0-1) which is the proportion of the methylated signal to the total signal, with a small constant added to the denominator for stabilization. For methylated CpG, signal_B is high and signal_A is low, then the calculated β is near 1 (the ratio of methylated to total signal). Hypomethylated probes (β values < 0.2) and hypermethylated probes (β values > 0.7) can be interrogated and compared across samples for large-scale studies. Raw data for average B-values and intensities for each CpG per sample were calculated. The Illumina internal controls and background subtraction were applied to the samples. All three samples passed the quality control steps.

Formation of Embryoid Bodies
For dissociation, feeder-free iPS cells were incubated with 1 U/mL dispase (Thermo Fisher Scientific) for 5-10 min. The cells were then washed with PBS with Mg and Ca (Eurx). Cell clumps were gently collected to new tubes and centrifuged at 300 rpm for 3 min. The aggregates of cells were suspended in iPS medium without bFGF (Thermo Fisher Scientific). The medium was changed every day. After 4 days, embryoid bodies were ready for further experiments or analysis. The representative pictures on day 4 were shown. Analysis was also performed on day 6. The graph data show the results from 4 clones in duplicate, collected on day 4 (n = 8) and 3 clones (n = 3) on day 6.

Differentiation to Three Germ Layers
Functional characterization of iPS cell lines was performed using a StemMACS Trilineage Differentiation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany). Each cell line was seeded on 3 wells of 24-well plate. The media were changed according to the Differentiation Kit Protocol. After 7 days, cells were passaged and counted for further analysis.

Staining and Flow Cytometry
Expression of specific markers of the mesoderm, ectoderm and endoderm were analyzed according to the StemMACS Trilineage Differentiation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany). Five hundred thousand cells were stained with antibodies. For staining of surface markers, the cells were alive, whereas for intracellular staining, the cells were fixed by BD Cytoperm/Cytofix Plus (Becton Dickinson, Franklin Lakes, NJ, USA). The differentiated cells were incubated with fluorochrome-conjugated antibodies (fluorescein allophycocyanin-APC and phycoerythrin-PE): anti-CD144-PE, anti-CD140b-APC (mesodermal markers); anti-PAX6-APC, anti-SOX2-PE (ectodermal markers); and anti-CD184-PE and anti-SOX17-APC (endodermal markers). According to the protocol, the negative controls were the unstained cells. The staining was analyzed using Attune NxT Software v2.2 on the Attune Nxt Flow cytometer (Thermo Fisher Scientific).

Formation of Teratomas
All experimental protocols on NOD-SCID mice were approved by the 2nd Local Institutional Animal Care and Use Committee (IACUC) in Cracow, decision numbers: 162/2015 and 11/2018. iPS cells were passaged by Accutase (Lonza), centrifuged and suspended in Matrigel (Corning) and PBS (Eurx) in proportions of 1:1. Two hundred microliters (2 × 10 6 cells) of the suspended cells were injected into the left dorsal flank of an adult female non-obese diabetic/severe combined immunodeficiency (NOD-SCID) mouse. Three mice were injected with each iPS clone; four iPS clones of each origin of iPS cells were tested. Tumor growth and mice health were controlled every day. Two months after the injection, teratomas were excised. Hematoxylin-eosin (H/E) staining was performed, as described previously [1]. After the staining, the amount of different structures in the tumor were analyzed. After characterization of particular structures, their amount was counted for all analyzed teratomas from clones from iPS-K and iPS-P. The structures were counted from 4 independent preparations for each clone. The counted structures were assessed by an experienced pathomorphologist. Graphs present the average amounts of structures in teratomas generated from iPS-K (generated from keratinocytes) and iPS-P (generated from PBMC); n = 4. Pie charts represent the average amount of structures in all analyzed teratomas from particular types of iPS origin. Bar graphs compare the average amount of particular structures between both types of teratomas from different iPS origins.

Statistical Analysis
GraphPad Prism version 8.4.2 was used to analyze the acquired data. To analyze statistically significant differences between two compared groups, Student's t-test was performed. Data on graphs were presented as means ± SEM. Statistically significant differences (p < 0.05) are represented in graphs by the * symbol.

Conclusions
To conclude, our findings suggest that the origin of iPS cells may significantly affect their differentiation capabilities in different models, such as differentiation in teratomas in vivo, as well as exerting effects in models of both 2D and 3D differentiation into dopaminergic neurons in vitro. Keratinocyte-derived iPS cells may display a bias towards the formation of a higher number of neuroectodermal structures in teratomas in vivo and increased expression levels of different neural factors in 2D and 3D differentiation into dopaminergic neurons in vitro. These limitations should be taken into consideration in the design of good models of different neuronal diseases, such as Parkinson's disease and Alzheimer's disease [1,53].