Forskolin, 3-isobutyl-1-methylxanthine (IBMX), and dibutyryl-cAMP (db-cAMP) were from Serva, Germany. Cell culture media and Pen Strep are from Life Technologies Europe BV. KG-501 was from Società Italiana Chimici, SIC, Italy. b-FGF was from Merck Darmstadt Germany. Sphingosine-1-phosphate (S1P) and arginine-vasopressin (AVP) were from Sigma Aldrich.

C-MSCs were enzymatically isolated from human term placenta and characterized as previously described [15]. In the experiments shown, cells were passaged 6–10 times. For neuron-like cell induction, C-MSCs were plated in 96 well plates (10${}^4$/well) or on Ø12 mm coverslips coated with fibronectin for immunocytochemistry assay and grown at 37 °C, 5% CO${}_2$ and 20% O${}_2$ in DMEM 10% FCS and 1% Pen Strep. The medium was replaced after 24 hours with fresh growth medium reducing FCS to 1% and containing 50 $\mu$M forskolin (dissolved in DMSO), 0.5 mM 3-isobutyl-1-methylxanthine (IBMX, dissolved in ethanol), or 1 mM dibutyryl-cAMP (db-cAMP, dissolved in water) or vehicle as reported in the text. The culture media remained the same for up to 6 days.

After three days of treatment with the induction medium, C-MSCs were fixed using 4% paraformaldehyde (PFA) + 2% sucrose for 15 minutes and washed three times with phosphate buffer saline (PBS). Permeabilization and blocking was performed using PBS containing 10% FCS and 0.1% TritonX-100 for 30 minutes at room temperature. Coverslips were incubated overnight with primary antibodies raised against doublecortin (1:200 cat. #4645, Cell Signaling Technology, Inc. Danvers, USA), pro-neuregulin 1 (1:100 cat. CSB-PA09589A0Rb, Cusabio, Technology, Houston, USA), neuron-specific class III $\beta$-tubulin (1:500 Biolegend, 801202 San Diego, USA) for 3 hours at room temperature. After washing three times with PBS-0.05% Triton-X100, coverslips were incubated with secondary antibody (1:500 DyLightTM488-labeled IgG, KPL, Gaithersburg, USA) for 1 hour. Nuclei were stained with Hoechst 33342 (1:10000 SIC, Italy). The cytoskeleton was stained by phalloidin-alexa488 (1:500, SIC, Italy). Images were acquired by confocal microscopy (LSM710, Carl Zeiss) using a 20$\times$ Plan-Neofluor objective.

Neuron-like cells were clearly different from fibroblasts because they manifested a number of protrusions and smaller nuclei, as compared to the larger, flat fibroblast phenotype. The number of neuron-like cells was quantified by manual counting using the Plugin Cell Counter of the software ImageJ Fiji (version 2.1.0/1.53c, NIH, Bethesda, Maryland, USA).

The fluorescent signal was quantified in confocal images and analyzed using ImageJ Fiji software. The background was measured in non-populated areas and subtracted. The specific signal was calculated by measuring the mean fluorescence intensity of the whole cell surface for over 30 cells. The intensity was normalized to the average signal of a comparable number of untreated cells stained and analyzed in parallel.

C-MSCs were seeded in 96-well plates, after replacement of the growth medium with an induction medium, cells were placed in an EVOS® Onstage Incubator (Life Technologies) at 37 °C, 5% CO${}_2$ and 20% O${}_2$ under an EVOS-FL auto microscope (Life Technologies) for time lapse acquisition to monitor cell morphological changes. Images were taken with the high-sensitivity interline CCD monochrome (grayscale) camera in phase-contrast with a 10x Objective, fluorite, LWD, phase-contrast (Olympus AMEP 4681) every hour up to 6 days. The time lag of the neuron-like phase was determined for each cell in a number of representative fields (6 random fields at 10x, for each of triplicate well for each experiment).

Real-time PCR was performed on cells treated with forskolin (50 $\mu$M) and b-FGF (40 ng/mL) as indicated. Each condition was performed in triplicate wells. After treatment, cells were washed with PBS and total mRNA was extracted using the TRIzol extraction kit (Invitrogen) and reversed transcribed to cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystem). cDNA from samples was mixed with the primers (the sequence is detailed in the Supplementary File S primers) and PowerUp SYBR Green Master mix (Applied Biosystems), and run on a CFX96Touch Real-Time PCR Detection System. Amplification was performed utilizing the following cycling parameters: 50 °C for 2 minutes, 95 °C for 2 minutes, 95 °C for 15 seconds and 60 °C for 1 minute; the last two steps were carried out for 40 cycles. Glyceraldhyde-3-phosphate dehydrogenase (GAPDH) was used as housekeeping gene. Experiments were conducted in triplicates. Data were transformed using the $\mathrm{\Delta }$$\mathrm{\Delta }$ct method.

The statistical measurements values were presented as mean $\pm$ variance as indicated in the figure legend. n indicates the number of independent experiments. Normal distribution was analyzed applying Kolmogorov-Smirnov test. Normal distributed data were analyzed utilizing t-test. Not normal distributed data were analyzed utilizing nonparametric tests (Kruskal-Wallis and Dunn’s multiple comparison as appropriate). In Fig. 3C nonparametric Kruskal-Wallis test was applied to compare the neuron-like cells number from day 2 to 4. Since at day 1 and 5 no neuron-like cell was present, the statistic was not applied because of the clear variation in respect to the other time point. To test at which day the effect was maximal we compared day 2 vs. day 3 and 4.

The p value was represented as follow: * p 0.05, ** p 0.01, *** p 0.001 and **** p 0.0001. The details of the analysis for each figure are provided in the Supplementary Data (S statistics).

C-MSCs were isolated from human placentas at term and propagated in vitro. We characterized the mesenchymal properties of these preparations in terms of immunophenotype and potency in a previous paper [15].

Similarly to what was repeatedly documented for other MSCs typically derived from bone marrow [4, 5, 6, 7, 8], a subset of C-MSCs acquired a neuron-like morphology in response to a treatment aimed to raise the concentration of intracellular cAMP. The effect was unambiguous because in response to forskolin the large flat fibroblasts raised around the nucleus to form a central dense body surrounded by a branched cytoplasm (Fig. 1, Supplementary Videos 1,2).

Fig. 1.

Forskolin induces neuron-like morphology in C-MSC. (A) Phase-contrast image of C-MSCs after 3 days of treatment with 50 $\mu$M forskolin. Part of cells assumed a neuron-like morphology while others conserved a fibroblast-like shape. The upper panel shows a very large cluster of neuron-like cells. White arrows point to fibroblast-like cells, black arrows point to neuron-like cells. In the panel below a magnification of a neuron-like cell from the same cluster. Arrowheads indicate contacts with other cells. (B) Crystal violet staining of cells treated with forskolin. In the upper panel a cluster of neuron-like cells (cluster formation can also be observed in the Supplementary Videos 1–5).

The robust stress fibers, typical of mesenchymal fibroblast-like cells, remained well organized in the majority of the cells that did not respond to the treatment. On the contrary, actin filaments were completely disorganized in the cells that assumed the neuron-like morphology (Fig. 2A).

Fig. 2.

Forskolin-induced rearrangement of the actin cytoskeleton and the expression of neuronal markers. (A) C-MSCs after 3 days of treatment with forskolin, phalloidin staining reveals the actin cytoskeleton in green, Hoechst-stained nuclei are in blue. The white arrow indicates a neuron-like cell. (B) Representative images of C-MSCs after 3 days of treatment with forskolin or vehicle. The panels on the left show phase-contrast images. All the other images present in panel (B) show cells stained for neuronal markers as indicated (green) or the nuclei (blue). (C) Fluorescence intensity was quantified in cells stained as in panel (B) and plotted as mean $\pm$ SD (unpaired t-test for the forskolin treated cells compares fibroblast-like vs. neuron-like, n = 3). (D) The expression of neuronal markers was measured by real-time PCR in C-MSC treated as indicated. For each condition, data are normalized to the gene expression of untreated cells.

The effect on cell morphology was associated with an increase in the expression of three neuronal markers, doublecortin (DCX), neuregulin 1 (NRG1), and neuron-specific class III $\beta$-tubulin (TUJ1) (Fig. 2B–D). This increase mainly concerned neuron-like cells. In fact, despite the treatment with forskolin, the cells that retained a fibroblast-like morphology displayed a level of fluorescence comparable to untreated cells.

A component of FCS was suggested to inhibit the acquisition of a neuron-like morphology by BM-MSC [4]. Likewise, 10% FCS concentration in the growth medium of C-MSCs reduced the number of neuron-like cells as compared to 1% FCS (Fig. 3A).

Fig. 3.

Neuron-like morphology is transient and mediated by transcription. (A) Neuron-like cells were counted 3 days after stimulation with 50 $\mu$M forskolin comparing two concentrations of FCS (n = 3 performed in triplicate, mean $\pm$ SEM). (B) Representative phase-contrast images of clusters of C-MSCs after 3 days after stimulation (50 $\mu$M forskolin, 500 $\mu$M IBMX, 1 mM db-cAMP) as indicated in each panel. (C) Neuron-like cells were counted over 5 days starting the first day after addition of the indicated stimuli applied as in panels (A) and (B) (n = 3 performed in triplicate, mean $\pm$ SEM. Statistical tests and representative movie files are available in the Supplementary Materials). (D) KG-501 was applied concomitantly to 50 $\mu$M forskolin, neuron-like cells were quantified on the third day of differentiation (n = 3, performed in triplicate, mean $\pm$ SEM, one way ANOVA).

The same dramatic change of morphology observed in response to forskolin was reproduced by two additional strategies, IBMX and the cAMP analog db-cAMP, that increase the intracellular concentration of cAMP acting on phosphodiesterases [9]. The three approaches replicated experimental settings applied by other authors to mesenchymal cells derived from other tissues [2, 4, 5, 6, 7, 8, 12, 16] and produced an identical effect on cell morphology (Figs. 1,3B). Although the treatments demonstrated different efficacy, in all cases the peak was after three days (Fig. 3C). Eventually, neuron-like cells returned to their original morphology indistinguishable from all other fibroblast-like cells (Supplementary Videos 1–4), which excludes a toxic effect.

Further increasing the concentration did not show saturation of the effect, yet the morphological differences became less obvious, possibly as a consequence of toxic effects that over 100 $\mu$M reduced cell viability (Supplementary Fig. 1) while complete serum removal did not prevent morphological changes (not shown).

cAMP-responsive elements binding protein (CREB) phosphorylation by PKA controls neurogenesis by supporting newborn neurons and its role is conserved throughout evolution [17]. In order to verify whether the appearance of neuron-like cells involved CREB, the specific inhibitor KG-501 was simultaneously administered in culture with forskolin. In the presence of 1 $\mu$M KG-501, the number of neuron-like cells was already reduced by more than 50% and at 10 $\mu$M the morphology change was completely prevented (Fig. 3D) consistent with what previously reported for other cells [18].

b-FGF is central to a highly orchestrated and complex sequence of cues required for the acquisition of a neuronal phenotype [19] and has often been combined with stimuli raising cAMP to promote MSCs neural differentiation [9, 20]. Combining 40 ng/mL b-FGF with 50$\mu$M forskolin promoted the formation of a larger number of neuron-like cells (to 143 $\pm$ 32%, p 0.05) and the effect lasted longer (from 26 hours to 33 hours with an average difference of 6.7 $\pm$ 0.9 h, Fig. 4).

Fig. 4.

b-FGF promotes neuron-like morphology. The appearance of cells characterized by a neuron-like morphology was monitored in time lapse documenting the time extent before the cell would return to a fibroblast-like aspect. For each experiment (n = 3) data are normalized to the maximum number of neuron-like cells achieved after stimulation with forskolin alone (mean $\pm$ SD). Each experiment was performed in triplicate wells for each condition. At least 6 fields/well were analyzed. In the inset panels, representative pictures in phase-contrast showing frames of cell cluster at the corresponding time (the full movie file is available in the Supplementary Materials).

The expression levels of Sox2 and Oct4 are tightly balanced to support self-renewal and pluripotency. Forskolin treatment produced a dramatic decrease in both transcription factors by the third day (Fig. 5). Expression remained low when all cells recovered the fibroblast-like appearance and the decrease was not prevented by b-FGF.

Fig. 5.

Forskolin induced reduction in the expression of pluripotent markers. The expression of pluripotent markers was measured by real-time PCR in C-MSC treated as indicated. (A) OCT-4 mRNA expression. (B) SOX-2 mRNA expression. For each condition, data are normalized to the gene expression of untreated cells (n = 3, performed in duplicate, mean $\pm$ SEM, one-way ANOVA).

S1P is a bioactive lysophospholipid with versatile biological effects, mainly exerted through the binding of their specific receptors [21]. S1P is a powerful stimulator of neurogenesis by binding to various G protein coupled receptors (GPCRs) to activate or inhibit signalling pathways and by cross-talking with other cytokine and growth factor receptors [22, 23]. Our lab previously demonstrated that S1P activates ERK1/2 in C-MSCs in a dose-dependent manner. ERK1/2 activation was transient in time and it was completely inhibited by pertussis toxin, indicating that the pathway was fully Gi-dependent [15]. Combining 10 $\mu$M S1P with 50 $\mu$M forskolin was found to prevent the effect of forskolin (Fig. 6).

Fig. 6.

S1P and AVP prevent neuron-like morphology. The appearance of cells characterized by a neuron-like morphology was monitored in time lapse. For each experiment (n = 6) data are normalized to the maximum number of of neuron-like cells achieved after stimulation with forskolin alone (mean $\pm$ SD). Each experiment was performed in triplicate wells for each condition. At least 6 fields/well were analyzed.

AVP, a hormone synthesized from the AVP gene in hypothalamus neurons, regulates various aspects of stem cells life. All three subtypes of AVP receptors are found in mice adipose-derived MSCs (A-MSCs) whereas human A-MSCs expressed only AVPR1 subtype [24]. Activation of AVPR1 activates Gq-proteins and on turn PLC-$\beta$ and calcium release [25], with a direct effect on CREB through the MAP-Kinase pathway. Similarly to S1P, 500 nM AVP prevented the effect of forskolin on the appearance of neuron-like cells (Fig. 6).

The authors recognize some limitations to this study. Time-lapse recording allows only to observe morphology. This approach proved that neuron-like cells can return to the original fibroblast-like aspect and indistinguishable from all other cells. However, it did not detail the nature of neuron-like phenotype. Transcriptomic and proteomic analysis of MSCs derived from several origins and treated with IBMX suggested that a similar morphology alteration coincide with neurite substructure development [12]. A similar analysis could be instrumental to define more precisely the significance of this process also in MSC derived from placenta. Microdissection strategies or better markers will allow to selectively tag the small cell population and explore the specific nature and physiological significance of neuron-like cells in future RNAseq studies.

A full commitment to neural phenotype could be tested with electrophysiological analysis by patch-clamp of single cells to proof the excitability, if any, associated with this neuron-like phenotype [11].