Methotrexate and Valproic Acid Affect Early Neurogenesis of Human Amniotic Fluid Stem Cells from Myelomeningocele

Myelomeningocele (MMC) is a severe type of neural tube defect (NTD), in which the backbone and spinal canal do not close completely during early embryonic development. This condition results in serious morbidity and increased mortality after birth. Folic acid significantly reduces, and conversely, folate antagonist methotrexate (MTX) and valproic acid (VPA) increase the occurrence of NTDs, including MMC. How these pharmacological agents exactly influence the early neurulation process is still largely unclear. Here, we characterized human amniotic fluid-derived stem cells (AFSCs) from prenatally diagnosed MMC and observed an effect of MTX and VPA administration on the early neural differentiation process. We found that MMC-derived AFSCs highly expressed early neural and radial glial genes that were negatively affected by MTX and VPA exposure. In conclusion, we setup a human cell model of MMC to study early neurogenesis and for drug screening purposes. We also proposed the detection of early neural gene expression in AFSCs as an additional MMC diagnostic tool.


Introduction
Myelomeningocele (MMC) is the most common form of neural tube defects (NTDs) with an average worldwide incidence of 4 per 10,000 live births [1]. It is caused by defective fusion of neural folds during day 25-28 of gestation, leading to the protrusion of dysplastic meninges and spinal cord from the spinal canal in a cyst-like sac (reviewed in [1]). During pregnancy, fetuses with MMC develop progressive motor and sensory deficits and hindbrain herniation (Arnold-Chiari Malformation II) and in some variable degrees of ventriculomegaly. Depending on the level of the lesion, there will be bowel and bladder dysfunctions. Some children may have an intellectual deficit, in particular in case of complicated hydrocephaly [1,2].
The etiology of MMC is unclear, with both environmental factors and genetic variations predisposing to the condition [3,4]. The use of folic acid (FA) supplementation during pregnancy reduces the risk of NTDs up to 70% [5,6], while prenatal exposure to folate antagonists, including methotrexate (MTX) and valproic acid (VPA), increases the risk of NTDs [7,8]. Hence, daily FA intake of at least 400 micrograms is recommended to all women of reproductive age to prevent NTD from occurring [4,5].
Folate coenzymes play an important role in several crucial processes, including nucleotide biosynthesis, generation of methyl donors and cell proliferation [9][10][11]. Both MTX and VPA are known to influence enzymes crucial for the folate metabolic pathway [12,13]. Yet, the exact mechanisms through which MTX and VPA cause NTDs remain unclear [5].
The embryonic neural plate and neural tube are formed from neuroepithelial (NE) cells. These polarized cells actively proliferate and express the earliest marker for neural plate SOX2 and an intermediate filament protein NESTIN. Subsequently, immature neurons expressing neural βIII-tubulin and radial glial (RG) cells expressing BLBP appear for further neural tube organization [14][15][16][17][18]. FA stimulates the proliferation and differentiation of neural stem cells (NSCs), whereas MTX and VPA impair cell proliferation of embryonic NSCs and amniotic fluid-derived neural stem cells, respectively [19,20]. It is likely that undesirable effects of VPA and MTX on the developing neural tube are due to the influence of genes crucial for neural tube development, including Ki67, SOX2, and NESTIN [7,20]. Obviously, the assessment of undesirable effects of drugs in pregnancy is limited for ethical reasons, with limited options to enroll pregnant women in clinical trials [20]. Thus, retrospective epidemiological studies or experimental animal models remain the main sources for studying drug effects. Therefore, novel platforms, such as human MMC cell models, are being developed to study the disease characteristics and early neurulation process and to reveal the effects of novel therapeutic drugs.
The amniotic fluid has been shown to contain a heterogeneous population of cells [21,22] that may be used as an invaluable source of patient-derived cells. In recent years, there has been growing interest in amniotic fluid-derived stem cells (AFSCs) due to their possible applications as clinical therapeutic tools for disease modeling, drug screening, and regenerative medicine. AFSCs are a subset of multipotent fetal stem cells that retain the ability to differentiate various cell types [21][22][23][24]. Previous studies have demonstrated successful differentiation of AFSCs to neural-like cells [25][26][27][28], and pharmacological agents, including VPA, have been shown to affect their characteristics [20].
In this study, we isolated and characterized AFSCs from fetuses prenatally diagnosed with MMC. Following induction into early neural differentiation, we described the effects of MTX and VPA on MMC-derived AFSCs compared to healthy AFSCs. This study offers a novel cell-based model to investigate human MMC characteristics and for diagnostics and drug screening purposes.  [29] with few adjustments. Briefly, amniotic fluid was centrifuged for 10 minutes at 800 rpm; the cell pellet was resuspended in mesenchymal basal medium, containing DMEM (Invitrogen), 20% Fetal Bovine Serum (FBS, Invitrogen), 1% Pen/Strep (Thermo Fisher Scientific), and 2 mM L-glutamine (Thermo Fisher Scientific). Cells were incubated at 37°C with 5% humidified CO 2 . After 36 hours, cells were washed with Phosphate-Buffered Saline (PBS, Invitrogen) to remove nonadherent cells, and the culture medium was replaced by fibroblast growth medium, containing KO-DMEM (Invitrogen), 2 mM L-glutamine, 10% FBS and 1% Pen/Strep. The medium was refreshed every three days. Cells were daily monitored by light microscopy. When reaching 80-90% of confluence, cells were detached by enzymatic treatment, using TrypLE (Gibco) and expended.

Material and Methods
2.3. Cell Viability Assays. Viability assays were performed on AFSCs after each passage, exposed or not to VPA and MTX. Briefly, 10 μL of the cell pellet from each MMC and healthy line was resuspended in 10 μL of Trypan Blue (Life Technologies) and inserted in an automatic cell counter machine (Invitrogen). The number of live/dead cells and viability were evaluated.

Flow Cytometry Analysis.
Amniotic fluid cells were harvested from culture and dissociated by EDTA Accutase (Thermo Fisher Scientific) for 3 minutes at 37°C. Samples were collected in staining buffer (Hanks' Balanced Salt Solution, HBSS; Thermo Fisher Scientific) with CaCl2 and MgCl2 supplemented with 2% FBS, 10 mM HEPES, and 10 mM NaN3 (both from Sigma-Aldrich; pH 7.2). Cells were quantified and 100,000 cells from each line were stained for 30 minutes at room temperature (RT) with the following antibodies: CD44 (0.25 μg), CD117 (0.25 μg), CD105 (1 μg), CD73 (0.125 μg), and CD90 (1 μg/mL, all from Thermo Fisher Scientific, Bioscience). After incubation, cells were washed with PBS, centrifuged for 5 minutes at 800 rpm, and the cell pellet was resuspended in 200 μL staining buffer. Cells were analyzed and quantified by flow cytometry (BD FACSCanto I or II with HTS; BD Biosciences) and FlowJo Software (FlowJo LLC). and 1(b)). Increasing concentrations were tested for each pharmacological agent, and cell viability assay analyses were performed. 0.25 μM MTX and 1 mM VPA did not affect live/ dead viability assays and morphology of the cells. Thus, these concentrations were chosen to perform experiments. Pharmacological agents were added only during the first three days of neural induction.
2.9. Statistical Analysis. Data were analyzed using GraphPad Prism 6. All data were obtained from 4 independent experiments and reported as mean ± standard deviation (SD). In particular, 7 MMC-AFSC lines and 3 healthy AFSC lines have been used for RT-qPCR experiments, while for all other experiments 4 MMC-AFSC lines and 3 healthy AFSC lines have been used.

Morphology, Proliferation, and Characterization of MMC-AFSCs.
We first aimed at assessing the morphological characteristics of AFSCs to address whether differences between MMC-AFSCs and healthy AFSCs were present. No morphological differences could be observed between the two cell types (Figure 1(a)). We then sought whether MMC-AFSCs and healthy AFSCs retained multipotent characteristics. We checked for the expression of previously described mesenchymal stem cell markers [21,30] on MMC-AFSCs and healthy AFSCs by flow cytometry. Analysis revealed that MMC-AFSCs and healthy AFSCs similarly expressed a set of specific mesenchymal stem cell surface markers, including CD44, CD73, and CD90. Moreover, cells expressed low levels of CD117 (also known as C-Kit) and appeared to be almost negative for CD105 (Figures 1(b) and 1(c)). Therefore, we concluded that MMC-AFSCs and healthy AFSCs had similar morphology and retained multipotent stem cell characteristics.

Characterization of MMC-AFSCs by RT-qPCR.
We aimed to assess the expression levels of lineage-specific and pluripotent stem cell markers in MMC-AFSCs and healthy AFSCs. The RT-qPCR analysis showed that MMC-AFSCs and healthy AFSCs presented similar low expression profiles of pluripotency genes OCT4 and NANOG, the myogenic transcription factor PAX7 and the early neural transcription factor PAX6. Conversely, similar high expressions of the mesoendodermal genes PDGFR?, GATA4, osteogenic transcription factor RUNX2, and the chondrogenic genes SOX9 and COL2A1 was detected in both MMC-AFSCs and healthy AFSCs (Figure 2(a)). Additionally, all AFSC lines presented low expression of mesoendodermal genes MIXL1, GATA6, SOX17, and FOXA1 and negative expression of mesoendodermal and ectodermal genes, including BRACH, PDGFRa, ACAN, OCN, MYOD, MYH3, RELN, POU4F2, and FOXG1 (Supporting information Figure 1(c)). Intriguingly, overexpression of early neural genes SOX2 and NESTIN, the radial glial (RG) gene BLBP and the myogenic transcription factor PAX3 (also known as a neural crest stem cell migration marker) was observed in MMC-AFSCs (Figure 2(b)). We concluded that MMC-AFSCs presented significantly higher expression levels of neural and RG fate specification genes compared to healthy AFSCs.

Neural Differentiation Potential of MMC-AFSCs.
Because of the overexpression of early neural markers in MMC-AFSCs by RT-qPCR, we hypothesized that MMC-AFSCs could have better in vitro neural differentiation commitment compared to healthy AFSCs. Therefore, we induced differentiation of MMC-AFSCs and healthy AFSCs towards early neural lineage. Moreover, as the pharmacological agents MTX and VPA are clinically correlated with MMC in fetuses, we sought to explore their in vitro effect on early neuralderived MMC-AFSCs and healthy AFSCs. To this end, AFSCs were exposed to either MTX or VPA during the first 3 days of neural induction. At day 15, experiments were stopped and immunofluorescence (IF) analyses were MTX and VPA exposure, the amount of SOX2 + cells was dramatically reduced compared to untreated conditions in both MMC-AFSCs and healthy AFSCs. Interestingly, MMC-AFSCs appeared to be more sensitive to MTX and VPA exposure as evidenced by the significantly reduced amount of SOX2 + cells compared to healthy AFSCs (Figure 3(c)).  Conversely, MMC-AFSCs and healthy-AFSCs uniformly stained positive for NESTIN (Figures 3(a) and 3(b)). Following quantification, the exposure to either pharmacological agents resulted in a decreased amount of NESTIN + cells, with VPA exposure effect appearing to be more severe in MMC-AFSCs (Figure 3(d)).
We next evaluated the immature neural differentiation potential of MMC-AFSCs and checked the proliferation of neural-derived MMC-AFSCs. To this end, we assessed the subcellular localization of βIII-tubulin and Ki67. IF analysis demonstrated that double positive βIII-tubulin and Ki67 cells were much higher in MMC-AFSCs compared to healthy AFSCs (Figures 4(a) and 4(b)). Indeed, approximately 60% of MMC-AFSCs stained positive for βII-tubulin, while they were less than 5% in healthy AFSCs. The exposure to MTX and VPA dramatically decreased the number of βIII-tubulin + cells in MMC-AFSCs (Figure 4(c)). Additionally, analysis revealed that MMC-AFSCs and healthy AFSCs were highly proliferative at day 15 from neural induction (Figures 4(a) and 4(b)). Conversely, a dramatic decrease in the amount of Ki67 + cells was observed in all AFSC lines following exposure to VPA and MTX (Figure 4(d)).
Further, IF analysis showed that MMC-AFSCs had more BLBP + cells compared to healthy AFSCs (Figures 5(a) and  5(b)). Following quantification of BLBP + cells, it was confirmed that MMC-AFSCs presented significantly higher amount of BLBP + cells compared to healthy AFSCs. Both MTX and VPA severely reduced the amount of BLBP + cells at similar rates ( Figure 5(c)). Finally, IF staining for PAX3 also showed a higher percentage of PAX3 + cells present in MMC-AFSCs compared to healthy AFSCs. MTX exposure did not have any effect on MMC-AFSCs and healthy AFSCs as both presented similar amounts of PAX3 + cells compared to untreated conditions. Conversely, VPA exposure negatively affected the amount of PAX3 + cells only in MMC-AFSCs, with no significant changes in PAX3 + cells amount observed in healthy AFSCs (Figures 5(a), 5(b), and 5(d)).
Therefore, MMC-AFSCs exhibited a higher percentage of early neural and RG positive cells at day 15 from neural induction compared to healthy AFSCs. In addition, the exposure to MTX and VPA significantly decreased the number of early neural and RG positive cells in MMC-AFSCs. Intriguingly, PAX3 + cells appeared to be resistant to MTX treatment for all cell lines, and VPA affected only PAX3 + cell number in MMC-AFSC derivatives.

Discussion
AFSCs have recently emerged as a novel tool for regenerative medicine purposes. They can be obtained from fetuses with minimal risk and can be used for in vitro drug screening purposes [22,31]. Indeed, the effects of several pharmacological agents on fetal development are also observed in AFSCs used as in vitro cell model. In fact, VPA treatment alters proliferation and inhibits neural gene expressions in AFSCs [20]. The use of VPA and MTX during pregnancy interferes with normal neural tube formation by blocking dihydrofolate reductase, methylenetetrahydrofolate reductase, and several other key enzymes of the folate pathway [12]. Thus, the folate pathway is critical for the regulation of the neurulation process, which is also evidenced by the benefit of FA supplementation to avoid abnormal neural tube closure [13].
In this study, we used human AFSCs derived from 7 patients diagnosed with MMC to study the early neural differentiation process and to evaluate the effects of MTX and VPA exposure on neurulation event. Our results revealed that human MMC-AFSCs showed similar growth, proliferative potential, pluripotency gene expression, and mesenchymal stem cell characteristics as healthy AFSCs. Previous studies demonstrated that AFSCs derived from MMC Sprague-Dawley rat model showed significantly increased neural stem cell (NSC) markers, including SOX2 and NESTIN [32,33]. Moreover, in a recent study, MMC-AFSCs isolated from a rat model were characterized, showing high expression of neural progenitor markers SOX2 and NESTIN [34]. Similarly, in our study, human MMC-AFSCs significantly overexpressed early neural genes SOX2 and NESTIN and in addition PAX3 when compared to healthy AFSCs. The paired box transcription factor PAX3 is expressed in neural crest stem cells during vertebral development [35], having a crucial role in neural tube closure as seen in Splotch mutant mouse embryos exhibiting NTDs [36]. In our previous   study, we demonstrated the overexpression of PAX3 gene in a fetal lamb model of MMC [37]. Consistently, our results showed that in humans MMC-AFSCs PAX3 gene was overexpressed, and the number of PAX3 + cells was higher in neural-derived MMC-AFSCs compared to that of the healthy lines. It is possible that upon the failure of neural tube closure, PAX3 progenitors are getting activated in order to force neural crest stem cell migration. Interestingly, MTX exposure did not sort any effect on PAX3 expression in all samples, and VPA exposure slightly affected only the amount of MMCderived PAX3 + cells. RG cells guide neuronal migration processes and maintain the shape of the cerebral cortex during development. While the increase of the RG marker GFAP has already been described in association with spinal cord injury [38], we sought to analyze the levels of RG marker BLBP in MMC-AFSCs. Intriguingly, BLBP was overexpressed in MMC-AFSCs where, at day 15 from neural induction, the number of BLBP + cells was higher compared to healthy AFSCs. To our knowledge, this is the first report that describes the overexpression of the BLBP gene in human MMC-AFSCs. Intriguingly, our results showed that VPA and MTX exposure negatively affected the amount of SOX2 + , NESTIN + , βIII-tubulin + , and BLBP + cells in AFSC-derived neural progenitors. These data suggested that MTX and VPA severely affected neural and RG commitment. In addition, as expected [39,40], both drugs negatively impacted cell proliferation, as shown by a decrease on the amount of Ki67 + cells.
In summary, MMC-AFSCs retained an enormous potential to differentiate toward early neural progenitors and could be further considered for in vivo therapeutic and regenerative applications. Moreover, although samples from more patients would be required to confirm our findings, our results suggested that validation of SOX2, NESTIN, PAX3, and BLBP expression in amniotic fluid could be used as an additional tool to confirm MMC diagnosis.

Conflicts of Interest
All the authors declare no potential conflict of interests.