Mesenchymal stem cells (MSCs) exhibit immune-suppressive properties, follow a pattern of multilineage differentiation, and exhibit transdifferentiation potential. Ease in expansion from adult bone marrow, as well as its separation from ethical issues, makes MSCs appealing for clinical application. MSCs treated with retinoic acid resulted in synaptic transmission, based on immunostaining of synaptophysin and electrophysiological studies. In situ hybridization indicated that the neurotransmitter gene preprotachykinin-I was expressed in these cells. However, translation of this gene only occurred after stimulation with interleukin (IL)-1α. This effect was blunted by costimulation with IL-1 receptor antagonist. This study reports on the ability of MSCs to be transdifferentiated into neurons with functional synapses with the potential to become polarized towards producing specific neurotransmitters.
Research studies on adult and embryonic stem cells (ESCs) have undergone enormous advances during the past few years. The experimental evidence shows that both stem cells could have future benefits in the area of regenerative medicine. Adult stem cells have been shown to have potential benefits for such diseases or conditions as diabetes mellitus, liver disease, cardiac dysfunction, Alzheimer's disease, Parkinson's disease, spinal cord injuries, bone defects, and genetic abnormalities [1–7]. Furthermore, adult stem cells have been extensively studied for repair of orthopedic conditions, such as bone, cartilage, and tendon defects [8, 9].
ESCs are considered the prototype stem cells because of their innate ability to differentiate into all possible cells and are therefore invaluable sources for tissue repair . Despite the remarkable potential of ESCs in medicine, the obvious limit is the need for a safe match with respect to the major histocompatibility complex class II . ESCs could become functionally unstable when placed in an in vivo microenvironment and develop into tumors . Another important consideration for ESCs is the stage of development. These cells might be in transit to committed cells and would therefore, express various developmental genes. This molecular change might not be evident, because the ESCs might not show phenotypic changes. Thus, it is conceivable that cells generated from ESCs could be dysfunctional if the originating stem cells are already committed to form cells of another tissue. Given these arguments, clinical application of ESCs would require robust examination of gene expressions before the generation of different cell types.
Clinical application of hematopoietic stem cells (HSCs) is controversial. Some reports show evidence of transdifferentiation by HSCs . Others report that transdifferentiation of HSCs could be mistaken by cell fusion between HSCs and cells of other tissues . An issue that is mostly overlooked with respect to HSCs is the low efficiency that one could achieve in their expansion by current in vitro methods. Thus, to acquire sufficient HSCs for clinical use would require invasive procedures upon the donors. Another issue with HSCs is that a population of HSCs that is deemed stem cells by phenotypic analyses is generally heterogeneous and could include cells that are committed towards a particular lineage . Future application of HSCs in repair medicine requires further research with the appropriate cell subset. Given an ideal situation where the candidate HSC is identified, there are ethical issues on the amount of bone marrow (BM) aspirates that should be taken from a donor. Presently, the literature has opened potential avenues for all types of stem cells. The most efficient stem cell or cells with least ethical issues will be determined by future research studies.
Accordingly, alternative use of adult stem cells has been examined . Mesenchymal stem cells (MSCs) show promise among the adult stem cells. MSCs are major adult BM stem cells with multilineage potential [16, 17]. Theoretically, MSCs can be used across allogeneic barrier because of their unique immune property . This property of MSCs is demonstrated by the cell's ability to facilitate BM transplantation. [19–23]. Furthermore, MSCs are easily obtained from adult BM and can be expanded by simple in vitro procedures . MSCs have been shown to transdifferentiate into cells of other germ layers [25, 26].
The generation of MSCs into neurons has been studied [27–29]. However, for the most part, these studies characterized the transdifferentiation of MSCs based on morphology, phenotypic changes, and action potential [27–30]. As far as we are aware, synaptic transmission has not been reported for neurons derived from MSCs. Cells similar to MSCs have been shown to survive in the brain . In this report, we isolated MSCs from human BM aspirate and generated functional neurons as indicated by phenotype and electrophysiology. Synaptic transmission, as evident by immunofluorescence for synaptophysin, was supported by the presence of synaptic currents.
Administration or implantation of cultured neurons in a damaged tissue in vivo would be influenced by the microenvironment. Thus, the question is whether a neuron could be influenced to express a particular neurotransmitter. This is a relevant question, because microenvironmental changes would vary depending on the type of deregulation or injury. This study focused on the preprotachykinin-I gene (PPT-I), which encodes the neurotransmitter substance P (SP). Regulation of the PPT-I gene has been studied in the context of inflammatory mediators that are presumed to be at the site of neural injury. Interleukin (IL)-1 was selected mostly because of its role in neural function , its ability to induce the production of SP , and its signature as a proinflammatory mediator .
Materials and Methods
Dulbecco's modified Eagle's medium (DMEM) with high glucose, DMEM/F12, and L-glutamine were purchased from Life Technologies (www.lifetech.com). Defined fetal calf serum (FCS) was purchased from Hyclone (Logan, UT). All-trans-retinoic acid (RA), phosphate-buffered saline (PBS) (pH 7.4), human serum albumin (HSA), and Ficoll Hypaque were purchased from Sigma (St. Louis). Stock solution of RA was prepared in dimethylsulfoxide to 20 mM. Vectashield was purchased from Vector Laboratories (Burlingame, CA).
Antibodies and Cytokines
Rabbit anti-MAP2, nestin monoclonal antibody (mAb), rabbit anti-neurofilament (68 kDa), and glial fibrillary acidic protein (GFAP) mAb were purchased from Chemicon (Temecula, CA). Anti-synaptic vesicle 2 mAb was obtained from NOVO CASTRA (Newcastle, UK). Fluorescein isothiocyanate (FITC) streptavidin, synaptophysin mAb, and synaptovesicle protein 2 mAb were obtained from Vector Laboratories. FITC-goat anti-rabbit immunoglobulin G (IgG) and nonimmune mouse IgG were purchased from Sigma. Phycoerythrin (PE)-goat anti-rat IgG was obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). The following antibodies were purchased from Becton and Dickinson (San Jose, CA): PE-rat anti-mouse κ, rat anti-SP mAb, PE- and FITC-isotype control, PE-conjugated CD14 mAb, FITC-conjugated CD45 mAb, and PE-CD44 mAb. FITC fibroblasts mAb were purchased from Miltenyi Biotec (Auburn, CA). SH2 mAb was prepared from the ascites of mice injected intraperitoneally with hybridoma as described . Recombinant human IL-1α (rhIL-1α) was kindly provided by Hoffman LaRoche (Nutley, NJ). IL-1 receptor antagonist was purchased from R&D Systems (Minneapolis) and was stored at 2.5 mg/ml. Rabbit anti-SP was purchased from Biogenesis (Brentwood, NH).
Culture of MSCs
MSCs were cultured from BM aspirates as described . The use of human BM aspirates followed a protocol approved by the Institutional Review Board of University of Medicine and Dentistry of New Jersey-Newark campus. Unfractionated BM aspirates (2 ml) were diluted in 12 ml of DMEM containing 10% FCS (D10 media) and then transferred to tissue culture Falcon 3003 petri dishes. Plates were incubated, and at day 3, mononuclear cells were isolated by Ficoll Hypaque density gradient and then replaced in the culture plates. Fifty percent of media was replaced with fresh D10 media at weekly intervals until the adherent cells were approximately 80% confluent. After four cell passages, the adherent cells were asymmetric, CD14−, CD29+, CD44+, CD34−, CD45−, SH2+, and fibroblast− .
MSCs, approximately 80% confluence, were trypsinized and then subcultured in 35-mm Falcon 3001 petri dishes, on superfrost plus slides (Fisher Scientific, Springfield, NJ), or on round coverslips. Slides were placed in 17 × 100-mm tissue culture dishes, and coverslips were placed in 35-mm suspension tissue culture dishes. Superfrost slides and cover slips were seeded with 500 cells in D10 media. RA was added to culture media at 30-μM final concentration. After 3 to 4 days, media were replaced with fresh media containing RA. At different times after RA treatment, cells were studied by immunofluorescence, electrophysiology, and in situ hybridization for β-PPT-I and by immunocytochemistry for SP (see below).
Immunofluorescence for Neural-Specific Markers
RA-treated MSCs were washed with PBS (pH 7.4) and then incubated for 1 hour at room temperature with the following antibodies: rabbit anti-MAP2 and/or synaptophysin mAb at final concentrations of 1/500 and 1/200, respectively. In other labeling studies, cells were incubated with anti-GFAP, anti-neurofilament, or anti-nestin, each at 1/500 final concentration. Antibodies were diluted in 0.05% Tween/PBS (PBT) for membrane permeabilization. Primary antibodies were developed with secondary FITC-goat anti-rabbit IgG and PE-rat anti-mouse κ, both at final concentrations of 1/500. Secondary antibodies were incubated for 45 minutes at room temperature in the dark. After labeling, cells were fixed with 0.4% paraformaldehyde and then covered with Vectashield. Slides were immediately examined on a three-color immunofluorescence microscope (Nikon Instruments Inc., Melvelle, NY).
Immunohistochemistry for SP
MSCs on superfrost slides were treated with RA. At day 6, cells were washed three times with sera-free DMEM and 25 ng/ml rhIL-1α in a total volume of 500 μl. Control cells were incubated in parallel with vehicle (PBS with 1% HSA). After 24 hours, cells were washed with PBS and then incubated in PBT containing rabbit anti-SP (1/3,000 final dilution) and/or synaptic vesicle protein 2 mAb. The latter determined whether SP was stored in synaptic vesicles. SP and anti-synaptic vesicle protein 2 were detected with anti-rabbit IgG FITC and anti-mouse IgG1 κ PE, respectively. After this, cells were fixed with 0.4% paraformaldehyde, covered with Vectashield, and immediately examined with a three-color immunofluorescence microscope.
In Situ Hybridization for β-PPT-I
MSCs were cultured as above on superfrost slide and then washed with 0.05% PBT. Cells were hybridized by first incubating with 500 μl of hybridization buffer (Sigma). After 2 hours, slides were incubated with 200 μl of hybridization buffer containing a cocktail (200 ng/ml) of denatured biotin-conjugated β-PPT-I--specific oligo probes (Table 1). Control slides were incubated with a mixture of similar oligos in the sense orientations. After this, slides were incubated overnight at 37°C in a humid chamber. Slides were then washed with standard saline citrate solution, and hybrids were detected by incubating for 1 hour at room temperature with FITC-streptavidin at 1/500 final concentration. Slides were covered with PBS and then immediately examined with an immunofluorescence microscope.
Table Table 1.. Oligonucleotide probes specific for β-PPT-I
Oligonucleotide sequences used for in situ hybridization to detect β-PPT-I. Abbreviation: PPT-I, preprotachykinin-I.
β-PPT-I (accession no. U37529)
5′act gcc gga gcc ctt tga 3′
5′atg gcc aga tct ctc aca 3′
5′tta tga aag gag tgc aat 3′
Electrophysiological recordings were conducted as described previously . In brief, whole-cell patch-clamp configuration was used to record electrical activity with an Axopatch 200B amplifier (Axon Instruments, Forster City, CA), a Digidata 1320A A/D converter (Axon Instruments, Foster City, CA), and pCLAMP 9 software (Axon Instruments). Data were filtered at 2 KHz and sampled at 10 KHz. The junction potential between the patch pipette and the bath solutions was nulled just before forming the giga-seal. The liquid junction potential between the bath and the electrode was 3.3 mV, as calculated from the generalized Henderson equation with the Axoscope junction potential calculator . Taking into account an initial series resistance of 15 to 25 MΩ, after the standard 80% compensation, there remained a 3–5-mV error for 1 nA of current. The cells were differentiated with RA in a 35-mm culture dish (Falcon 3001). These dishes were filled with standard external solution containing (in mM) NaCl 130, KCl 5, CaCl2 2, MgCl2 1, HEPES 10, and glucose 10 (320 mosmol; pH set to 7.4 with NaOH). The patch electrodes had a resistance of 3–5 MΩ when filled with (in mM) KCl 121, MgCl2 4, EGTA 11, CaCl2 1, HEPES 10, Mg-ATP 2, and GTP 0.1; the pH was adjusted to 7.2 with 1N KOH, and the osmolarity was adjusted to 280 mosmol with sucrose. All recordings were made in these solutions at an ambient temperature of 20–23°C. Postsynaptic currents and action potentials were counted and analyzed using the MiniAnalysis program (Synaptosoft, NJ). Spontaneous events were checked visually.
Transdifferentiation of MSCs to Neuron-Like Cells
MSCs treated with RA were first examined microscopically at × 400 magnification. The purpose was to observe morphological changes as the cells were exposed to a known differentiating agent. Compared with the symmetrical morphology of untreated MSCs (Fig. 1, upper left panel), MSCs treated for 1 day with RA showed a transition to cells with asymmetrical morphology (Fig. 1, upper right panel). This change progressed to day 4, when the RA-treated MSCs showed neuron-like structures (Fig. 1, lower left panel). By day 7, the treated cells demonstrated structures of cell bodies with long thin processes and growth cones (Fig. 1, lower right panel).
We next determined whether the RA-treated MSCs express neuron-specific proteins. Cells were labeled for MAP2, which mostly labels the cell bodies and dendrites, neurofilament to stain for cell bodies, and axons and nestin for changes in morphology to neuronal cells. Cells were also studied for synaptophysin, which is an indicator of synaptic regions. Labeling with anti-MAP2 (FITC) and anti-synaptophysin (PE) were done at different times after RA treatment. Figure 2 shows pictures of individually labeled and overlaid MAP2 and synaptophysin (left columns).
Untreated MSCs stained dim with MAP2 but negative for synaptophysin (Fig. 2, top panels). In contrast, RA-treated cells stained bright for MAP2 at each time point shown in Figure 2. Cells stained for 6-day RA treatment showed predominance of MAP2 and slight fluorescence for synaptophysin (Fig. 2). At day 9, RA-treated cells showed brighter fluorescence for synaptophsin. This pattern of synaptophysin increased at up to day 12 of treatment. Figure 2A represents 10 different experiments, each with MSCs from a different BM donor.
Figure 2B shows representative staining for neurofilament and nestin. All cells treated with RA were negative for GFAP (not shown). The percentage of neurons developed by RA treatment were 80 ± 10 (n = 100) by immunostaining for neurofilament, MAP2, and nestin. However, when the RA-treated cells were analyzed by electrophysiology, 100% showed action potential (n = 25). Of note is that transfer of MSCs from propagation media (DMEM) to DMEM/F12 does not increase the percentages of neuron formation but enhanced the formation of nestin-positive cells by 2–3 days. The results show that MSCs treated with RA coexpress MAP2 and synaptophysin beginning at day 6 of RA treatment.
Electrophysiology in RA-Treated MSCs
To determine whether the RA-treated MSCs show functions consistent with native neurons, we studied the cells' electrophysiological properties using whole-cell patch clamp technique. First, record was made of action potentials under current clamp condition. As illustrated in Figure 3A, the cell fired spontaneously and regularly with a frequency of 20 s−1. The amplitude, measured from the resting membrane potential of −50 mV, was 63.4 ± 17.4 mV. The half-width and rise time of the action potentials were 19.25 ± 0.73 and 3.39 ± 0.11 ms (n = 768 events), respectively (Fig. 3D). To determine whether the RA-treated cells can communicate with each other, we next measured the postsynaptic currents under voltage clamp condition. Figure 3B illustrates a segment of such trace recorded from a cell treated with RA for 6 days. The frequency, amplitude, half rise time, and decay time of the postsynaptic currents of this cell were 2.49 s−1, 29.9 ± 1.3 pA, 12.96 ± 0.89 ms, and 36.5 ± 1.0 ms (n = 171 events), respectively. Figure 3C illustrates representative postsynaptic currents recorded from a cell in culture for 15 days. The frequency, amplitude, half rise time, and decay time of the postsynaptic currents of this cell were 1.24 s−1, 6.1 ± 0.2 pA, 1.84 ± 0.13 ms, and 31.3 ± 1.3 ms (n = 99 events), respectively (Fig. 3D). Kinetic analysis indicated that although the decay of the postsynaptic currents of the cell in culture for 6 days could be fitted by a single exponential function, the decay of the postsynaptic currents of the cell in culture for 15 days required two exponentials.
Induction of SP in RA-Treated MSCs
This section describes studies to determine whether the neurons formed from MSCs could be induced to express a particular neurotransmitter. We selected IL-1α and the neurotransmitter gene PPT-I, because the regulation of PPT-I by inflammatory mediators such as IL-1 has been studied . Furthermore, the receptor for IL-1 has been reported in neurons . In addition, IL-1 can induce the production of SP, which is the major peptide derived from the PPT-I gene .
MSCs, treated with RA for 6 days, were stimulated with 10 ng/ml of IL-1α for 16 hours, and the cells were studied for SP production by immunofluorescence using anti-SP. The treated cells were also labeled for synaptovesicle 2, because it would be of interest to determine whether SP is stored in synaptic vesicle. The studies used MSCs exposed to RA for 6 days, because this time frame also showed functional synaptic activity (Fig. 3). RA-treated MSCs were labeled with anti-SP (FITC; Fig. 4, left column) and anti-synaptovesicle (PE; Fig. 4, middle panel). The two labels were overlaid to get insights on the location of SP. Because the anti-SP was diluted in 1% HSA and the immunofluorescence used indirect immunofluorescence, secondary antibody specificity was studied in parallel labeling with 1% HSA in lieu of anti-SP.
Representations of six experiments, each performed with a different donor, are shown in Figure 4. The results show bright fluorescence intensity for SP in cells stimulated with IL-1α (Fig. 4, middle row). Fluorescence intensities were significantly reduced when the cells were costimulated with IL-1 RA (Fig. 4, bottom row). This indicates that the effects of IL-1α were specific. Cells labeled with 1% HSA showed minimal fluorescence, which was subsequently used as background fluorescence to compare other labeling studies (Fig. 4, top row). The results show that IL-1α can induce the production of SP in neurons generated from RA-treated MSCs.
Expression of β-PPT-I in RA-Treated MSCs
Because dim fluorescence was observed for SP in studies with unstimulated MSCs treated with RA (not shown), we surmised that the gene from which SP is produced, PPT-I, might be expressed with low level of translation. We therefore asked whether the PPT-I gene was expressed in RA-treated MSCs by in situ hybridization with a cocktail of three oligomers (Table 1). Studies with neurons from three different BM donors are shown in Figure 5. Bright fluorescence was observed in the cell bodies of neurons (Fig. 5, right panels). Neurons labeled with sense oligonucleotides showed no fluorescence (not shown). The results indicate that the PPT-I gene is expressed in untreated MSCs differentiated with RA.
RA is a general differentiation and transdifferentiation agent for the generation of neurons . RA has been shown to induce the differentiation of cells during embryonic maturation into distinct organs, including the generation of neurons [38, 39]. This report similarly used RA to generate neurons from MSCs, asking two major questions.
First, we determined whether MSCs cultured in the laboratory could generate neurons. The techniques used superseded other methods, because we showed by electrophysiology that the MSC-derived neurons are functional with respect to synaptic transmission (Fig. 3). Immunostaining evidence for synaptic vesicles (Fig. 2) was further supported by synaptic currents recorded via patch clamp technique (Fig. 3). The method used in this study showed higher efficiencies in the generation of neurons compared with other reports . Although we do not have an explanation for these differences, we speculate that our starting population and differentiation methods could explain the differences in efficiencies.
Second, we determined whether the neurons generated from MSCs could be induced to express a specific neurotransmitter and, hence, arbitrary selection of SP. Indeed, the studies showed that SP was induced by IL-1 (Fig. 4). Its induction was specific, as evident by blunting when the cells were stimulated in the presence of IL-1 receptor type I antagonist (Fig. 4). The apparent descending of SP in the IL-1-stimulated cultures (Fig. 4) suggests that this cytokine, which is an inflammatory mediator, could cause bidirectional movement of the neurotransmitter, SP .
The studies did not perform labeling experiments to identify additional neurotransmitters. Thus, this report cannot make claim that stimulation by IL-1α caused polarization with respect to a single neurotransmitter. However, because we observed high levels of SP, it could be argued that our model has the potential to study mechanisms to achieve neurotransmitter polarization of neurons. It would be interesting to examine SP-positive neurons for the neurotransmitter calcitonin gene-related peptide, because this peptide generally colocalizes with SP .
An interesting observation is the presence of bright immunofluorescence for β-PPT-I by in situ hybridization (Fig. 5). Because the RA-treated cells were not stimulated with IL-1α, the results showing β-PPT-I in the cell body indicate that this gene is available for translation in the event that the appropriate signal is received. Theoretically, PPT-I should be expressed so that the production of neurotransmitters could be rapid, given the appropriate signal. Ongoing studies will identify whether miRNA might be involved in halting translation or other mechanisms. Regardless, the data pertaining to PPT-I are exciting observations that form the focus for current and future studies. Because MSCs show potential for application in repair medicine, future studies are required, and planned, to determine the influence of microenvironmental factors on the behavior of neurons. Timeline studies are required to identify the point at which IL-1 receptor is expressed during transdifferentiation of MSCs to neurons. This question is important to determine how inflammatory mediators, present within a microenvironment, would influence the function or development of MSCs to neurons.
Although IL-1α is a single cytokine, several growth and inhibitory factors are expected in a physiological situation during tissue damage or cell dysfunction. Recent report showed that cAMP is important for repair of spinal cord injury . IL-1α is linked to an increase in the levels of cAMP . Thus, the findings in this report could be relevant to the mechanisms by which cAMP might be a mediator of injury to the spinal cord . It would be important to show how a microenvironment could be manipulated to allow for the maturation and development of neurons from MSCs while allowing for polarization with respect to the type of neurotransmitter.
The spontaneous discharges of RA-treated cells indicate ongoing electrical activity (Fig. 3). The spontaneous postsynaptic currents reveal the existence of the functional synapses in these cells. Because communication is a unique characteristic of neuronal cells, these functional synapses indicate that our MSCs have this important ability. Interestingly, although the decay time is almost the same for these two cells, the rise time is much shorter for the cell in culture for 15 days, with a half-rise time of 12.96 ms (n = 171 events) for 6-day—treated cell versus 1.84 ms (n = 99 events) for 15-day—treated cell (p < 0.01). The rise time of the postsynaptic current mainly reflects the time requests for the transmitters release from presynaptic vesicle and travel to the postsynaptic membrane, with immature synapses having longer rise times. This may indicate the maturation of these synapses. Note that some of the spontaneous postsynaptic currents recorded from cells in culture for 6 days, such as the one shown in Figure 3, have a half rise time around 2 ms, which is close to the rise time for synapses in native neurons (neurons in mid-brain slice, 1.23 ± 0.02 ms, n = 1,729 events, data not shown).
Future work will address the transmitters released by unmanipulated cells. Furthermore, insight into the electrophysiology and microenvironmental manipulation of the cells has clinical implications. A missing point in this study is to determine whether the neurons can exhibit retraction. These studies are in progress. The results of these studies, including retrograde uptake of molecules, are the subject of another manuscript. Once the mechanism of microenvironmental manipulations is understood by in vitro methods, the studies would be extended to in vivo models. The studies are necessary before MSCs are applied to patients with neural disorders. As the identities of relevant molecules and an understanding in their roles become apparent, there will be understanding of networks that link the microenvironment and MSCs. Ultimately, these studies should provide avenues for appropriate cell therapies.
MSCs can be induced, via RA treatment, to become functional neurons that express neuron-specific phenotypic markers and exhibit synaptic properties similar to those of native neurons. The MSC-derived neurons express the tachykinin gene PPT-1 and can be induced to produce SP by IL-1 stimulation.
This work was supported by grants CA-89868, AA11989, and AT001182 from the National Institute of Health and a grant from the F.M. Kirby Foundation.