Retinoic Acid-Treated Pluripotent Stem Cells Undergoing Neurogenesis Present Increased Aneuploidy and Micronuclei Formation

The existence of loss and gain of chromosomes, known as aneuploidy, has been previously described within the central nervous system. During development, at least one-third of neural progenitor cells (NPCs) are aneuploid. Notably, aneuploid NPCs may survive and functionally integrate into the mature neural circuitry. Given the unanswered significance of this phenomenon, we tested the hypothesis that neural differentiation induced by all-trans retinoic acid (RA) in pluripotent stem cells is accompanied by increased levels of aneuploidy, as previously described for cortical NPCs in vivo. In this work we used embryonal carcinoma (EC) cells, embryonic stem (ES) cells and induced pluripotent stem (iPS) cells undergoing differentiation into NPCs. Ploidy analysis revealed a 2-fold increase in the rate of aneuploidy, with the prevalence of chromosome loss in RA primed stem cells when compared to naïve cells. In an attempt to understand the basis of neurogenic aneuploidy, micronuclei formation and survivin expression was assessed in pluripotent stem cells exposed to RA. RA increased micronuclei occurrence by almost 2-fold while decreased survivin expression by 50%, indicating possible mechanisms by which stem cells lose their chromosomes during neural differentiation. DNA fragmentation analysis demonstrated no increase in apoptosis on embryoid bodies treated with RA, indicating that cell death is not the mandatory fate of aneuploid NPCs derived from pluripotent cells. In order to exclude that the increase in aneuploidy was a spurious consequence of RA treatment, not related to neurogenesis, mouse embryonic fibroblasts were treated with RA under the same conditions and no alterations in chromosome gain or loss were observed. These findings indicate a correlation amongst neural differentiation, aneuploidy, micronuclei formation and survivin downregulation in pluripotent stem cells exposed to RA, providing evidence that somatically generated chromosomal variation accompanies neurogenesis in vitro.


Introduction
Pluripotent stem cells, as capable of forming any cell type in the body, have been extensively used to study developmental biology, including neurogenesis. In the beginning, embryonal carcinoma (EC) cells were the unique choice for analyzing differentiation in vitro [1]. With the isolation of embryonic stem (ES) cells [2] and more recently, induced pluripotent stem (iPS) cells derived from reprogrammed somatic cells [3,4,5], new tools for studying the stemness behavior and the ability to differentiate into neurons became available.
The all-trans retinoic acid (RA) is a metabolic compound derived from vitamin A, widely associated with neurogenesis [6]. During embryogenesis, RA contributes to patterning of the neural plate and neural tube [7,8,9]. In adulthood, it plays a role in neuronal differentiation within the dentate gyrus [10], in the maintenance of motor neurons [11] and in mammalian nerve regeneration [12]. Indeed, RA is the most used morphogen to produce neural progenitor cells (NPCs) and neurons from stem cells in vitro [13,14,15,16].
The existence of variations in chromosome number (aneuploidy) in NPCs and neurons from mice, fish and humans has been previously described [17,18,19]. Aneuploid NPCs are present at a substantial rate of 33% in the developing brain [18,20,21] and 20% in the cerebellum [22]. Also, aneuploid neurons can be found functionally integrated into the neural circuitry of adults [21,23]. Even though the significance of aneuploidy within the nervous system is still unknown, the brain is a mosaic composed by euploid and aneuploid cells, which may contribute for brain complexity [18].
Despite chromosomal mosaicism has been extensively described within the brain in vivo, possible causes of aneuploidy during development of the central nervous system (CNS) have never been explored. Here, we asked whether neurogenesis triggered by RA in vitro is accompanied by aneuploidy in pluripotent stem cell types.

Reprogramming into pluripotent stem cells
Induced pluripotent stem (iPS) cells were developed in house [24] as previously described by Takahashi and Yamanaka [3]. The procedures were according to regulations set by international guidelines [25] and the local Biosafety Committee. Mouse embryonic fibroblasts (MEF) isolated from C57Black6 mice were transduced with retroviral vectors containing the human cDNAs of Oct4, Sox2, Klf4 and c-Myc inserted into moloney-based retroviral vectors (pMXs). 293T cells [26] (kindly provided by Dr. Bryan Strauss from Instituto do Coração, INCOR, Universidade de São Paulo) were used as packaging cells to the virus production after calcium phosphate based transfection with the following plasmids: pMXs (20 mg), pMDM (encoding mlv GAG/ POL, 10 mg) and pMD.G (encoding VSV G, 6 mg). The viral supernatants were collected 48 and 72 h post-transfection, filtered (0.22 mm) and concentrated at 32,000 g for 60 min. MEF were exposed to two rounds of transduction with the 293T concentrated supernatants. To improve transduction, polybrene (Sigma-Aldrich Corp., St. Louis, USA) was used at 8 mg/ml and the plate was centrifuged at 540 g for 45 min. Two days post-transduction, the cells were cultured in iPS cells-medium (detailed below) and treated with valproic acid ( Since the use of enzymes (to obtain single-cell suspensions) may affect the karyotype status of embryonic stem cells [30,31], we have adopted manual dissection (which produces cell clumps, instead of single-cells) to handle our cells and routinely monitored them for karyotypic changes. While ES cells exhibited the same number of chromosomes (40) after dozens of passages, iPS cells presented a mixed population formed by cells with 40 or 41 chromosomes even after 3 passages, as similarly described for human iPS cells [32].
Neural differentiation EC cells. EC cells were differentiated into NPCs as previously described [15,33,34]. Briefly, 5610 5 cells/ml were transferred to non-adherent dishes to form embryoid bodies (EBs). EBs developed in defined medium (DMEM supplemented with 2 mM L-glutamine, 2 mM sodium pyruvate, 2.4 mg/ml sodium bicarbonate, 5 mg/ml insulin, 30 mg/ml human apo-transferrin, 20 mM ethanolamine, 30 nM sodium selenite, 100 U/ml penicillin, 100 mg/ml streptomycin and 10 mM Hepes, pH 7. ES and iPS cells. ES and iPS cells differentiation into NPCs was carried out by RA treatment as previously described [35,36]. EBs were formed after transferring 2.5610 5 cells/ml to nonadherent dishes. EBs were cultured in medium with FBS without LIF. The percentage of FBS was 20% for ES-EBs and 15% for iPS-EBs. After 4 days of formation, EBs were exposed to 5 mM RA for additional 4 days. RA stock solution was prepared in DMSO, which was also added to the control groups at the final concentration of 0.01%. For immunofluorescence assays, EBs after 8 days in vitro were dissociated in 0.05% trypsin/EDTA and cells were plated onto precoated polyornithine (100 mg/ml, Sigma-Aldrich Corp., St. Louis, USA) and laminin (5 mg/ml, Invitrogen, Life Technologies, Carlsbad, USA) coverslips. After 2 h of plating, immunofluorescence was performed as below.
For analysis of DNA content and neural immunophenotyping, EBs exposed to RA or vehicle were dissociated as mentioned above. Then, cells were plated at a density of 12610 5 cells/cm 2 onto polyornithine and laminin-coated plates and maintained in neuronal medium, composed by high-glucose DMEM/F12 supplemented with 2% N2, 20 ng/ml fibroblast growth factor-2 (FGF-2) and 55 mM 2-mercaptoethanol (all from Invitrogen, Life Technologies, Carlsbad, USA) for 5 days, then DNA analysis by flow cytometry was carried out. For further analyses, EBs were also plated into polyornithine and laminin substrates and maintained in neuronal medium for 5 days.

Chromosome counts
Metaphase spreads were prepared as previously described [37]. Cells were incubated with 0.1 mg/ml KaryoMAX colcemid (Gibco Invitrogen, Life Technologies, Carlsbad, USA) for 6 h, then dissociated using trypsin/EDTA 0.05% and incubated for 15 min in a 75 mM KCl hypotonic solution at 37uC. The swollen cells were fixed in a methanol/glacial acetic acid 3:1 (v:v) solution (Merck KGaA, Darmstadt, Germany) and stored at 4uC overnight. On the next day, cell suspension was washed with the fixative solution and then spread onto glass slides. The slides were quickly exposed to water steam to burst the cells. Chromosomes were stained with DAPI, and the slides coverslipped with n-propyl galate. The mounted slides were examined on a fluorescence microscope (Nikon Eclipse 80i, Nikon Instruments Inc., NY, USA). At least 60 metaphases were counted on each condition per assay.
Once iPS cells present a mixed population formed by cells with 40 or 41 chromosomes, we adopted this interval as the modal chromosome number for this cell line. Chromosome numbers below and above range were considered as an increase in the aneuploidy rate for iPS cells.

TUNEL assay
Apoptosis was determined in NPCs derived from EC cells using an apoptosis detection kit (In Situ Cell Death Detection Kit, Fluorescein; Roche Applied Science, Indianapolis, USA). After harvesting, cells were fixed in 2% paraformaldehyde in PBS for 30 min at 4uC, rinsed with PBS and resuspended in 0.2% Triton-X-100 and 0.2% Tween in PBS for 10 min at 4uC. Then, cells were rinsed again with PBS and incubated with TUNEL reaction mixture as suggested by the manufacturer. For negative controls, instead of the TUNEL reaction mixture, cells were incubated with label solution (without terminal transferase). For positive controls, cells were incubated with DNase I (3 U/ml, Ambion, Austin, USA) for 10 min at room temperature. Cells were analyzed by flow cytometry (Beckman Coulter, Fc500) and data were analyzed using the WinMDI 2.8 software available at http://facs.scripps. edu/software.html. Thirty thousand events were acquired per sample with fluorescence measured in logarithmic scales.
For apoptosis evaluation in NPCs derived from ES cells, ApopTag Plus Fluorescein In Situ Apoptosis Detection Kit (Chemicon, Temecula, USA) was used according to the manufacturer's instructions. Briefly, EBs were dissociated and filtered through a 40 mm nylon cell strainer (BD Biosciences, San Jose, USA), cells were fixed in 1% paraformaldehyde in PBS for 15 min at 4uC, rinsed in PBS and resuspended in 70% ice-cold ethanol for at least 2 h. Then, cells were incubated with TdT enzyme and anti-digoxigenin fluorescein conjugate. As negative control, cells were not incubated with enzyme but with anti-digoxigenin fluorescein. Cells were analyzed by flow cytometry (BD FACS-Canto II, BD Biosciences, San Jose, USA) and data were analyzed using the Summit software (Dako, Glostrup, Denmark). Twenty thousand events were acquired per sample with fluorescence measured in logarithmic scales.

Flow cytometric analysis of DNA content
The DNA content of NPCs derived from EC cells was determined by flow cytometry. After harvesting, cells were fixed in 70% ethanol at 220uC for at least 2 h. Then, cells were rinsed twice with PBS, stained with propidium iodide (50 mg/ml, Sigma-Aldrich Corp., St. Louis, USA) in the presence of RNase (100 mg/ ml, Ambion, Austin, USA) for 1 h at 37uC and taken to the cytometer (Beckman Coulter, Fc500, Fullerton) for DNA content analysis as previous described by Capparelli et al. [38]. Chick erythrocyte nuclei (CEN) were used at one-tenth of sample concentration as internal control. Thirty thousand events were acquired per sample with fluorescence measured in logarithmic scales. For determining changes in DNA content, relative DNA content was expressed as the ratio of G0/G1 peak to the mean of the CEN peak.
For analysis of DNA content within neuronal cells, RA treated-EBs derived from ES cells, dissociated and plated for 5 days, were fixed in ice-cold ethanol, rinsed with PBS and permeabilized with 0.5% Tween (Sigma-Aldrich Corp., St. Louis, USA) for 5 min. To prevent nonspecific immunolabeling, 5% BSA was used for 1 h. Then, samples were incubated on ice with primary antibody for bIII-tubulin (Sigma-Aldrich Corp., St. Louis, USA) for 1 h. After rinsing twice in PBS, anti-mouse Alexa 488-conjugated secondary antibody (Life Technologies, Carlsbad, USA) was added for 30 min on ice. Following PBS washing, propidium iodide (PI) mix consisting of 20 mg/ml PI (Sigma-Aldrich Corp., St. Louis, USA), 100 mg/ml RNase (Ambion, Austin, USA) and 0.1% Triton X-100 (Sigma-Aldrich Corp., St. Louis, USA) diluted in PBS was added and the cells were taken to the cytometer (BD FACSCanto II, BD Biosciences, San Jose, USA). Mouse lymphocytes stained under the same conditions were collected to calibrate the cytometer regarding the position of G0/G1 peak. Fifty thousand events were acquired per sample with fluorescence measured in logarithmic scales. Relative DNA content between positive and

Micronuclei detection
For micronuclei detection, colonies of ES and iPS cells were

Protein Extraction, SDS-PAGE and Western Blotting
Cell lysates were obtained with lysis buffer (1% NP-40, 0.25% sodium deoxycholate, 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 2 mM EDTA, 1 mM Na 3 VO 4 ) supplemented with 16 phosphatase and protease inhibitors (Sigma-Aldrich Corp., St. Louis, USA). Cells were maintained in lysis buffer on ice and homogenized by shaking at 5 min intervals until its complete dissociation. Then, samples were centrifuged, the supernatants recovered and protein quantity was measured by Bradford method with albumin as standard. Forty mg of protein in sample buffer were boiled for 3 min and separated by SDS-PAGE in a 15% polyacrylamide gel at a constant voltage of 100 V. After separation, proteins were transferred into a nitrocellulose membrane (Hybond, Amersham, Piscataway, USA) in a semi-dry system (Bio-Rad, Hercules, USA) for 120 min at constant amperage of 0.1 mA.
For nonspecific blocking, 5% nonfat milk in tris-buffered saline and 0.05% Tween-20 (TBS-T) was added for 2 h under agitation at room temperature. After incubation with primary antibodies for survivin (Sigma-Aldrich Corp., St. Louis, USA) or a-tubulin (Santa Cruz Biotechnology, Santa Cruz, USA) overnight at 4uC, membranes were washed and probed with anti-rabbit or antimouse peroxidase-conjugated secondary antibodies (Invitrogen, Life Technologies, Carlsbad, USA) for 2 h under agitation at room temperature. Antibodies were diluted in 5% nonfat milk in TBS-T. Then, membranes were washed in TBS-T. Blots were revealed with SuperSignal West Pico (Pierce, Rockford, USA). The resulting bands were measured with ImageJ software. The values were expressed in arbitrary units while the survivin level was plotted in relation to a-tubulin in the same lane. The normalization was made considering vehicle as 1.

Statistical analysis
Results were expressed as mean 6 S.E.M. Analysis of statistical significance was performed using 4.0 GraphPad Prism software. Data were analyzed by unpaired t-test.

Chromosome instability is increased in NPCs derived from EC cells treated with RA
In order to investigate whether neural differentiation is accompanied by aneuploidy in EC cells, EBs were formed and concomitantly exposed to RA to trigger their differentiation into NPCs (RA-NPCs). On day 4, immunofluorescence revealed nestin and bIII-tubulin in, respectively, 96.462.3% and 94.363.6% of  To identify changes in chromosome number, metaphases were obtained from RA-NPCs and untreated cells. As previously reported, EC cells possess massive genomic instability even in the undifferentiated state [39]. Since we found that 89.060.6% of undifferentiated EC cells presented between 59 and 65 chromosomes (data not shown), we considered this interval as the basal level of chromosomal instability in this lineage. Chromosome numbers below and above this range were considered as an increase in aneuploidy for EC cells.
EC-derived RA-NPCs displayed a 2-fold increase (44.560.6%) in numerical chromosomal instability when compared to EC cells incubated with vehicle (21.664.9%) ( figure 2A). Moreover, analysis of DNA content by flow cytometry revealed that RA-NPCs displayed less DNA content than EC cells incubated with vehicle (figure 2B), indicating loss of chromosomes after RA treatment.

Aneuploidy is increased in NPCs derived from ES and iPS cells after treatment with RA
In order to evaluate whether RA also increases aneuploidy in NPCs derived from chromosomal stable stem cells, NPCs were derived from both ES and iPS cells.
Neural differentiation of ES and iPS cells was carried out as previously described [35,36]. On day 8, nestin and bIII-tubulin were present, respectively, in 62.760.1% and 29.061.6% in RA primed ES cells and in 54.063.9% and 3662.1% in RA primed iPS cells ( figure 3).
Analysis of the chromosomal complement of RA-treated ES cells (figure 4A) demonstrated that aneuploidy increased from 23.362.7% in cells incubated with vehicle only to 38.765.4% in cells undergoing differentiation by RA ( figure 4B). The same phenomenon was demonstrated for RA-treated iPS cells, which presented a 2-fold increase in the aneuploidy rate when compared to cells incubated with vehicle (31.563.8% compared to 14.061.4%) ( figure 4C). On the other hand, the parental MEF with 40 chromosomes, from which our iPS cells were generated, did not show any increase in aneuploidy rate when exposed to RA under the same conditions as ES and iPS cells (5 mM RA for 4 days) ( figure 4D).
Neural commitment with RA revealed that chromosome loss (hypoploidy) was more prevalent than chromosome gain (hyperploidy) in pluripotent stem cells. In ES-derived RA-NPCs, 33.164.6% of the cells were hypoploid compared to 12.762.0% in cells incubated with vehicle, whereas the proportion of hyperploidy was not increased by RA (10.562.4% in vehicle compared to 5.661.5% in RA) (figure 5A). In iPS-derived RA-NPCs, hypoploidy was again prevalent (29.864.2% in RA compared to 7.360.4% in vehicle) while hyperploidy was not increased by RA (6.761.0% in vehicle compared to 1.661.6% in RA) (figure 5B). Table 1 shows the most common aneuploidies by chromosome number in RA-treated and untreated ES and iPS cells.
In order to verify whether the RA-increased aneuploidy was more prevalent in neural cells than in non-neural cells, we evaluated DNA content by flow cytometry in cells expressing or not bIII-tubulin. RA-treated EBs, dissociated and plated for 5 days, presented 47.066.4% of young neurons with long processes expressing bIII-tubulin (figure 6A). To identify decreases in DNA content, we assessed the PI mean fluorescence intensity (MFI). Young neurons expressing bIII-tubulin presented less DNA content than non-neural cells (figure 6B).   Aneuploid NPCs are not prone to cell death Next, we asked whether aneuploid RA-NPCs would be prone to cell death. Both EC and ES cells undergoing neural differentiation by RA were analyzed for DNA fragmentation. TUNEL technique followed by flow cytometry demonstrated no difference in apoptosis levels between RA-NPCs and cells incubated with vehicle ( figure 7A and B), demonstrating that an increase in aneuploidy is not necessarily followed by cell death in RA-treated stem cells.

Micronuclei formation is increased in pluripotent stem cells after RA treatment
During mitosis, micronuclei can be formed when a chromosome does not appropriately attach to the mitotic spindle and become excluded from daughter cells nuclei. Micronucleus is an indicative of mitotic checkpoint failure and a possible mechanism by which aneuploidy emerges [40]. Our results revealed that RA-treated ES cells presented micronuclei scoring 1.760.2 times higher than cells incubated with the vehicle. In iPS cells, treatment with RA also resulted in micronuclei increase (1.560.1 times higher than vehicle) (figure 8).

Survivin expression is reduced in RA-treated embryonic stem cells undergoing neurogenesis
Since a downregulation of survivin correlates with RA exposure [41] and micronuclei formation in cancer cell lines [42], we decided to investigate whether survivin expression could also be reduced in embryonic stem cells undergoing neurogenesis after treatment with RA. Quantitative real-time PCR (figure 9A) and western blotting (figure 9B) analysis revealed that RA-treated cells presented a 50% reduction in survivin mRNA and protein levels when compared to untreated cells. These results indicate that a member of the chromosomal passenger complex may be involved in the mechanism by which RA increases aneuploidy and micronuclei formation in pluripotent stem cells.

Discussion
It has been suggested that aneuploidy in the developing cerebral cortex is the consequence of chromosome missegregation and multipolar spindles [18,43]. Hypoploidy is the prevalent form of aneuploidy observed during neurogenesis in vivo [18,19,20]. Here we show that an increase in aneuploidy, especially chromosome loss, is also observed in pluripotent stem cells undergoing neurogenesis in vitro.
Recently, aneuploidy was described in neural progenitor cells derived from the human fetal brain (hNPCs) after long-term cell culture [44]. Sareen and colleagues reported that long-term cultured hNPCs are susceptible to chromosome gains. They observed that after 9 weeks and 25 weeks in vitro, five NPCs lines and one NPC line, out of 21 cell lines, acquired trisomy 7 and trisomy 19, respectively.
While Sareen et al. described chromosome gain upon selective pressure in vitro, which emphasizes the need for careful cytogenetic evaluation of NPCs before their use in clinical or basic science, our data point to an increase in chromosome loss during neurogenesis in vitro, similarly to the observed in vivo [18]. Moreover, chromosome loss observed in pluripotent stem cells after retinoic acid treatment occurs independently of the number of passages, differently from Sareen et al., in which NPCs must be cultured for at least 9 weeks to spontaneously generate chromosome gains.
A crucial question is whether aneuploidy here described was an authentic characteristic of neural commitment of pluripotent stem cells or a corollary effect of RA treatment. In this sense, MEF exposed to the same concentration of RA for the same period of time did not present numerical chromosome alterations. These  results suggest at least 2 interpretations: 1) aneuploidy promoted by RA is associated to neural differentiation; 2) somatic cells are not prone to aneuploidy triggered by RA. In fact, embryonic stem cells exhibit rapid cycles of cell division by bypassing certain checkpoints [45,46,47], which is postulated to be a property of stemness [48] but could also make these cells more prone to aneuploidy.
It was previously reported that RA induces apoptosis in EC and ES cells [49,50]. However, in both studies RA triggered apoptosis immediately upon its exposure, which was attenuated after cells start to differentiate [51]. In our case, apoptosis was measured in embryoid bodies at least 2 days after RA treatment, when cells were already committed to the neural fate. No difference in cell death was detected, indicating that an increase in aneuploidy is not directly associated with elimination of pluripotent stem cells undergoing neural differentiation in vitro.
Another issue addressed in this work was the description of a possible mechanism by which aneuploidy is triggered in pluripotent stem cells after RA treatment. Micronuclei reflect an inability to repair DNA lesions or deficits in checkpoints [40,52]. ES and iPS cells exposed to RA presented an increase in micronuclei formation when compared to naïve cells. Consistently  with our findings, Yang and coworkers reported micronucleation in NPCs in vivo [43]. It is worth mentioning that NPCs derived from the developing cerebral cortex show tolerance to chromosomal instability associated with an inefficient checkpoint mechanism [53], similarly to the observed in pluripotent stem cells [48]. Interestingly, the expression of survivin, which exerts multiple roles in chromosome segregation [54,55], was reduced in embryonic stem cells after RA treatment.
It is well known that the depletion of survivin, functioning as tension-sensor at kinetochores-microtubules interactions in mitotic checkpoint [54], leads to unaligned chromosomes, failure in the mitotic checkpoint arrest [56] and inaccurate chromosome segregation [57,58]. Indeed, in breast cancer cell lines, Pratt et al. demonstrated reduced expression of survivin in response to RA [41], while delayed expression of survivin results in cancer cells with micronuclei [42].
These data provide evidence that somatically generated chromosomal variation accompanies neurogenesis. We suggest that pluripotent stem cells undergoing neural differentiation are prone to mitotic checkpoint failure and defects in kinetochoresmicrotubules attachment, as indicated by the reduction in survivin expression, culminating with micronuclei formation and aneuploidy (figure 10). We cannot assure if aneuploidy is a cause or a consequence of neural differentiation but our results indicate that loss of entire chromosomes is a common feature of neurogenesis induced in vitro, which may lead to altered gene expression [20].
These results support the idea that aneuploidy is a common feature of neural progenitor cells both in vivo and in vitro and may play a crucial role in neurogenesis. Nevertheless, alternative protocols, without using RA, to induce neural differentiation should be applied to characterize a broader association between aneuploidy and neurogenesis. To our knowledge this is the first demonstration that aneuploidy and micronuclei are amplified during neurogenesis induced by RA in pluripotent stem cells. The physiological consequences of neurogenic aneuploidy should be carefully considered before exploring the full potential of neural cells derived in vitro towards therapy for neurodegenerative diseases.