Bone marrow-derived mesenchymal stem cells differentiate into nerve-like cells in vitro after transfection with brain-derived neurotrophic factor gene

Bone marrow-derived mesenchymal stem cells can differentiate into a variety of adult cells. Brain-derived neurotrophic factor (BDNF) is briefly active during differentiation and induces mesenchymal stem cells to differentiate into nerve cells. In this study, we cloned human BDNF to generate a recombinant pcDNA3.1(−)-BDNF vector and transfected the vector into bone marrow-derived mesenchymal stem cells. We selected these cells with Geneticin-418 to obtain BDNF-BMSCs, which were induced with retinoic acid to obtain induced BDNF-BMSCs. The transfected cells displayed the typical morphology and surface antigen profile of fibroblasts and were observed to express clusters of differentiation 29, 44, and 90 (observed in matrix and stromal cells), but not clusters of differentiation 31, 34, and 45 (observed in red blood cells and endothelial cells), via flow cytometry. Enzyme-linked immunosorbent assays showed that transfected bone marrow-derived mesenchymal stem cells secreted more BDNF than non-transfected bone marrow-derived mesenchymal stem cells. Immunocytochemistry and real-time reverse transcription polymerase chain reaction analysis showed that non-induced BDNF-BMSCs maintained a higher proliferative capacity and expressed higher amounts of brain-derived neurotrophic factor, nestin, neuron-specific enolase, and glial fibrillary acid protein than non-transfected bone marrow-derived mesenchymal stem cells. An additional increase was observed in the induced BDNF-BMSCs compared to the non-induced BDNF-BMSCs. This expression profile is characteristic of neurocytes. Our data demonstrate that bone marrow-derived mesenchymal stem cells transfected with the BDNF gene can differentiate into nerve-like cells in vitro, which may enable the generation of sufficient quantities of nerve-like cells for treatment of neuronal diseases.


Introduction
The treatment of neurological diseases caused by irreversible neuronal cell damage, necrosis, or loss of neuronal cells is a global problem. In recent years, research into cell transplantation and gene therapy has provided a novel paradigm for addressing this problem. Bone-marrow mesenchymal stem cells (BMSCs) are multipotent stem cells (Kronenwett and Haas 2006) possessing plasticity (Grove et al. 2004) and can be induced to differentiate into a variety of adult cell lineages, including neurocytes (Qi et al. 2010), osteoblasts (Li et al. 2011), chondrocytes (Alves da Silva et al. 2011), and hepatocytes (Lange et al. 2006). Furthermore, BMSCs are easily isolated, with low rates of transplantation rejection. Thus, BMSCs have been deemed a resource for cell transplantation and an effective vector for gene therapy (Li et al. 2008). The primary limitation in the use of BMSCs is their low prevalence in bone marrow (only 0.1%) and their propensity to differentiate primarily into fibroblasts in vivo, with differentiation at very low rates into neurocytes and gliocytes. Thus, the therapeutic potential of BMSCs for treating neuronal conditions is currently severely limited.
Neurotrophic factors, especially brain-derived neurotrophic factor (BDNF), play an important role in the development, differentiation, survival, and damage-repair of neurocytes (Benraiss et al. 2001). It has been reported that BDNF, which plays a primary role in the maturation, proliferation, and differentiation of neuronal cells, can induce BMSCs to differentiate into nerve cells in vivo (Brazelton et al. 2000;Lin et al. 2010) and in vitro (Zhao et al. 2004).
However, as a macromolecule, BDNF has minimal ability to cross the blood-brain barrier and is only briefly active because of its short half-life in vivo. Previous studies have reported stable growth factor expression with high expression levels when a vector carrying the target gene is transfected into BMSCs (Lee et al. 2001), as the BDNF gene is characterized by continuous and stable expression and can maintain BDNF activity for an extended period of time.
In this study, we cultured BMSCs derived from guinea pigs in vitro and transfected the BDNF gene into the BMSCs by electroporation. We examined the characteristics of the transfected BMSCs. The aim of our experiments was to determine whether the BDNF gene could promote the nerve-like cell differentiation of BMSCs in vitro.

Materials and Methods
Human BDNF clone and eukaryotic expression vector construction. RNA was extracted from human fetal brain tissue using Trizol reagent (Life Technologies, Carlsbad, CA). cDNA was synthesized by reverse transcription (Promega, Madison, WI). The cycling conditions for PCR were 94°C for 2 min; followed by 30 cycles of 95°C for 30 s, 58°C for 30 s, and 72°C for 60 s; and then 72°C for 5 min. BDNF primers, designed with Premier 5.0 software (PREMIER Biosoft, Palo Alto, CA), were added into a cDNA mixture for PCR amplification (Promega). The primers used were designed as follows (forward and reverse, respectively): 5′-GCGAATTCATGACCATCCTTTTCCTTACT-3′ and 5′-GCGGATCCTTATCTTC-CCTTTTAATGGTCAAT-3′. After the correct product sequence was verified, the pcDNA3.1(−) (Life Technologies) plasmid was used to generate the eukaryotic expression vector pcDNA3.1(−)-BDNF. The recombinant plasmids were transfected into HEK293 cells (ATCC, Rockville, MD), with BDNF expression determined by western blot (Bio-Rad Laboratories, Hercules, CA).
BMSC isolation and culture. Five red-eyed white guinea pigs weighing 250-300 g were bought from the Animal Feeding Center of the Medical College of Jinan University, PRC. All experimental procedures were approved by the Laboratory Animal Management Committee of Guangdong Province. BMSC isolation and culturing were performed as previously reported (Wang et al. 2009). The medullary cavities of guinea pig femurs and tibias were flushed repeatedly with Dulbecco's modified Eagle medium (DMEM; Gibco, Carlsbad, CA). The cells were rinsed with lymphocyte-separating medium (main ingredients: dextran and acid meglumine diatrizoate) and centrifuged at 1,000 rpm for 20 min. The pelleted cells were resuspended in DMEM and re-centrifuged for an additional 10 min. The pellet (with the leukocytic cream removed) was centrifuged at 1,000 rpm, washed twice for 10 min, and cultured in DMEM containing 10% (v/v) fetal bovine serum (FBS; Sigma, St. Louis, MO) at 37°C in a humidified atmosphere containing 5% CO 2 . The medium was changed after 48 h, followed by removal of non-adherent cells. The cells were passaged at a ratio of 1:2 after achieving 80% confluence, with the third passage being used in all subsequent experiments.
BMSC identification by flow cytometry. Passage 3 BMSCs were incubated with fluorescent-conjugated antibodies at 37°C for 1 h in the dark following two washes with phosphate-buffered saline (PBS, 1×). Cells were analyzed in a FACSCalibur flow cytometer using Paint-A-Gate software (Becton Dickinson, Franklin Lakes, NJ). The following antibodies were used: phycoerythrin (PE)-labeled rat anti-rat cluster of differentiation (CD) 29, CD31, CD34, CD44, CD45, and CD90 (Santa Cruz Biotechnology, Santa Cruz, CA). BMSC transfection. Passage 3 BMSCs were digested and centrifuged after adjustment of the density of the suspended cells to 2×10 6 /mL. Next, 0.5 mL of the cell suspension and 20 μg of pcDNA3.1(−)-BDNF were added to a 4-mm electrotransformation pool (Bio-Rad Laboratories). Electroporation was performed at settings of 1050 Uf for capacitance and 260 V for voltage. The cells were transferred into six-well plates and cultured for 48 h after electroporation, followed by selection with 50-200 μg/mL Geneticin (G418) sulfate (Sigma) for 10-14 d, with a medium change every 2-3 d.
BDNF-BMSC induction and differentiation. After removal of G418, the cells were cultured for an additional 5-7 d in DMEM containing 15% FBS. BMSCs that were transfected with the BDNF gene (BDNF-BMSCs) were induced by 0.5 μmol/L retinoic acid (RA; Sigma) for an additional 5 d.
BMSC and BDNF-BMSC proliferation-MTT assay. The viability of the cells was measured using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT; Sigma) assay. At each time point after seeding, the cells were incubated with 5 g/L MTT for 4 h at 37°C. The formazan granules generated by the live cells were dissolved in 150 mL of dimethyl sulfoxide (DMSO), and the absorbance at 490 nm was monitored using a model 550 microplate reader (Bio-Rad Laboratories).
Cell growth. As a direct cell growth assay, 5,000 cells were seeded per well in 24-well culture plates and counted daily by Trypan Blue exclusion dye staining (Sigma) for a 1-w period, and cell growth curves were plotted.
Cell analysis. Cell cycle analysis was performed using propidium iodide staining. The cell number at different phases of the cell cycle was analyzed by a FACSCanto II flow cytometer (Becton Dickinson). The proliferative index (PI) was calculated according to a previously published formula (Wang et al. 2007) Detection of BDNF in the culture supernatant by enzymelinked immunosorbent assay (ELISA). The BDNF concentration in the culture supernatant of all groups was determined by ELISA (R&D Systems, Minneapolis, MN). The culture supernatant was collected after the cells were cultured for 1 wk, 2 wk, 1 mo, and 2 mo. The concentration of FBS in the culture medium was decreased gradually from 15% to 10%, and then to 5%. Finally, Mesen Cult-XF medium (Gibco) was used to culture the cells before collection of the culture supernatant. All reagents and working standards were prepared following the manufacturer's instructions. In brief, 100 μL of the supernatant for testing was added to 96-well plates, gently mixed, and incubated at 4°C overnight. Each well was aspirated and washed according to the manufacturer's instruction, and 100 μL of the diluted biotinylated anti-guinea pig BDNF monoclonal antibody was added to each well. Subsequently, the plate was incubated for 2 h at room temperature. Aspiration and washing were repeated, and 100 μL of diluted streptavidin-HRP was added to each well and incubated for 2 h at room temperature. After 100 μL of TMB/E was added at room temperature for 15 min, the reaction was stopped by adding 50 μL of stop solution. The optical density (OD) in each well was determined within 30 min by using a microplate reader (Model 550, Bio-Rad Laboratories) at 450 nm.
Detection of BDNF, nestin, NSE, and GFAP by immunocytochemistry. Immunocytochemistry was performed using the SABC method (Shi et al. 1988): the culture medium was removed from adherent cells, and the cells were then fixed in 4% paraformaldehyde, incubated in a mixture of 30% hydrogen peroxide and pure methanol, incubated in goat serum, and then incubated with the anti-BDNF, anti-nestin, anti-NSE, and anti-GFAP (1:300 dilution for all; Wuhan Boster, Wuhan, China) primary antibodies overnight in a 4°C refrigerator. The following day, incubation with the HRP-labeled rabbit anti-mouse secondary antibody (Wuhan Boster, Wuhan, China) was performed after rewarming of the samples, followed by staining with streptavidin-biotin complex (SABC; Wuhan Boster) and 3,3′-diaminobenzidine (DAB; Wuhan Boster) and counterstaining with hematoxylin. The samples were then dehydrated, coverslipped, and sealed.
We selected ten non-overlapping visual fields in each well by inverted phase-contrast microscopy (Olympus, Tokyo, Japan); positively dyed and total cells were then counted.
Statistical analysis. The data are expressed as the mean±standard deviation (x AE SD). Statistical analysis was performed using SPSS 15.0 statistical software (Chicago, IL). ANOVA was employed for comparison among groups. P<0.05 was considered statistically significant.

Results
Construction of BDNF gene-containing eukaryotic expression vector. The coding sequence of BDNF was generated by RT-PCR (Fig. 1A). The eukaryotic expression vector pCDNA3.1(−)-BDNF was then successfully constructed. BDNF was confirmed to be present in HEK293 cells transfected with the pcDNA3.1(−)-BDNF plasmid. The size of the protein band conformed to that expected (Fig. 1B).
BMSC morphology and post-transfection change. After 72 h of culture, the adherent cells were primarily characterized by their long spindle shape and large, single nucleus, similar to fibroblasts. However, spherical hematopoietic cells existed in suspension. Roughly 2 wk later, the adherent cells were overgrown in the flask bottom, with a monolayer that was mutually fused among cell colonies, and arranged with certain directionality (Fig. 3A). The cellular morphology remained the same after passaging, although the rate of cellular proliferation gradually decreased with greater passage number. BMSC morphology and density did not change markedly after induction by RA (Fig. 3B).
Some BMSCs died after electroporation. After culture for 48 h, the viability of the transfected BMSCs was obviously higher than that of nontransfected BMSCs. Only cells transfected with BDNF survived G418 treatment. The transfection efficiency was approximately 28%. At 3-4 d after transfection, the cell density was still lower than that in nontransfected cells. Some surviving cells presented polygonal, fusiform, or ovoid shapes, along with cytoplasm retraction and concentration toward the nucleus together with a protruded soma (with the longer protruding end swelling into a cone shape) and a neuron-like form (Fig. 3C). Subsequently, the number of amoeboid cells increased at a slower rate, with cells returning to normal growth rate under appropriate conditions at 5-7 d after G418 removal. The number of amoeboid BDNF-BMSCs increased further after RA addition, with apparent cellular protuberances, and in some cases, cellular networks (Fig. 3D).
The effect of the BDNF gene on BMSC proliferation in BMSCs and BDNF-BMSCs was determined using the MTT assay. At 48, 96, and 144 h post-seeding, there was no notable variation in viability between non-induced and induced BMSCs, whereas non-induced BDNF-BMSCs displayed a significant increase in viability compared with non-induced and induced BMSCs. An additional increase in viability was observed in induced BDNF-BMSCs compared to uninduced BDNF-BMSCs (P<0.05; Fig. 4A). Similar results were obtained for cell counts from days 2 to 7 (P<0.05; Fig. 4B). Moreover, cell cycle analysis showed that BDNF-BMSCs were present in apparently lower proportions in the G 0 /G 1 phase and higher proportions in the S and G 2 /M phases compared to BMSCs (Fig. 4C, Table 1). The proliferative index (PI) of BDNF-BMSCs increased significantly compared with that of BMSCs (P<0.05; Table 1).
BDNF secretion. BMSCs (induced or non-induced) secreted BDNF at 1 wk. BDNF secretion gradually decreased with time and was almost undetectable at 2 mo. There was an obvious increase in BDNF secretion from noninduced BDNF-BMSCs compared with BMSCs at each time point, with differences of 661-fold (1 wk), 1,317-  fold (2 wk), 1,728-fold (1 mo), and 4,094-fold (2 mo) compared with the non-induced BMSCs. There was also an increase, although not significant, in the BDNF concentration in induced BDNF-BMSCs compared to noninduced BDNF-BMSCs. The maximum BDNF secretion of BDNF-BMSCs (induced or non-induced) was observed at 2 wk, although BDNF expression was still observed even at 2 mo (Table 2).

Discussion
BMSCs are derived from mesoblasts, adhere to the walls of tissue culture flasks, and possess a spindle morphology with fibroblast-forming colonies (Jiang et al. 2002). Our findings indicate that BMSCs do not express antigens observed on the surface of red blood cells and endothelial cells (e.g., CD31, CD34, and CD45), but express specific surface markers of matrix and stromal cells (e.g., CD29, CD44, and CD90) (Baddoo et al. 2003;Herzog et al. 2003). These findings agree with the accreditation standards of animal-origin BMSCs published by the International Society for Cell Therapy (Dominici et al. 2006). The particular mechanism of BMSC cross-germlayer differentiation into neurons, which are derived from ectoblasts, remains unclear. Undifferentiated BMSCs express not only mesodermal but also endodermal and ectodermal genes. Therefore, BMSCs are capable of producing mesodermal derivatives and perhaps also ectodermal derivatives through transformation, which may be the basis for crossgerm-layer differentiation (Woodbury et al. 2002).
It has been confirmed that the amino acid sequences of mammalian BDNF (Zhang et al. 2006), TrkB, and the p75 receptor (Dobrzańska et al. 2002) are highly homologous and conserved with regard to tissue distribution. Human and guinea pigs are both mammals. Therefore, we transfected the BMSCs of guinea pigs with the human BDNF gene in the study, expecting that the human BDNF would have an effect on guinea pigs.
Under normal circumstances, BMSCs secrete low levels of BDNF, and few differentiate into neurocytes. This conclusion is supported by the observation that non-induced BMSCs were, en toto, only slightly positive by immunocytochemistry and had low levels of mRNA  expression for BDNF, nestin, NSE, and GFAP in our experiments. The differentiation pathways of BMSCs are affected by a variety of factors, including differentiation signals, growth factor types in a given microenvironment, and the extracellular matrix (Deng et al. 2001;Seki 2003). The mechanism by which clinicians may induce BMSCs to differentiate into target cells has been the subject of intensive research in cellular transplantation and gene therapy.
In the present study, BMSCs were transfected with the BDNF gene. With continuous BDNF expression, the transfected BMSCs displayed significantly higher proliferative capacity, more BDNF secretion, and a higher proportion of differentiation into nerve-like cells than primary BMSCs. Differentiated BDNF-BMSCs possessed the morphological features of neuronal cells and expressed the neuronal cellspecific markers nestin, NSE, and GFAP. Nestin is a neuronal stem cell-specific marker protein, whereas NSE and GFAP are neuron-and astrocyte-specific markers, respectively.
The immunoreactivity and mRNA expression levels of nestin, NSE, and GFAP were higher in BDNF-BMSCs (induced or non-induced) than in BMSCs (induced or noninduced). Moreover, the immunoreactivity and mRNA expression level of NSE were greater than those of GFAP in BDNF-BMSCs (induced or non-induced), indicating that BDNF-BMSCs possessed characteristics of neuronal cells and suggesting that BDNF overexpression could enhance the differentiation of BMSCs into nerve-like cells, principally into neuron-like cells. These observations are consistent with previously reported studies (Jouhilahti et al. 2008). Although the mechanism underlying the BDNF-mediated enhancement of proliferation and differentiation remains unclear, it has been widely reported that BDNF binds to HNGFR TrkB or LNGFR P75NGFR on the membranes of BMSCs, activating tyrosine kinase (TK) and subsequently activating intracellular signaling pathways such as those involving mitogen-activated protein kinase (MAPK), calmodulin kinase (CaMK), and phosphatidylinositol-3-kinase (PI3K). This process causes BMSCs to proliferate more rapidly and differentiate from undifferentiated cells into mature neurons (Lu 2003;Tapia-Arancibia et al. 2004;Chai et al. 2007).
RA can induce cells to proliferate and differentiate by activating signal transduction upon binding to retinoic acid receptors (RARs) on the nuclear membrane, thereby affecting the transcriptional regulation of downstream genes and exerting biological effects (Nezzar et al. 2007). RA also enhances the expression of RARβ and the transcription factor islet-1/2, which is important for neural development. It is therefore used in neuronal induction media to promote the proliferation and differentiation of BMSCs into nerve-like cells . In our experiment, the immunoreactivity and mRNA expression levels of nestin, NSE, and GFAP in induced BDNF-BMSCs were higher than those in non-induced BDNF-BMSCs, consistent with the results of Bi et al. (2010). RA induces TrkB expression and switches on the TrkB-BDNF signal transduction pathway, enhancing the inductive effect of BDNF on cell differentiation (Kaplan et al. 1993). However, RA is a chemical reagent and thus has concentration-dependent effects. In particular, it has little effect on cells at low concentrations (for example, 0.5 μmol/L), whereas it has cytotoxic effects such as teratogenesis and apoptosis at high concentrations (Lu et al. 2009). To prevent toxicity, we treated the cells with a low concentration of RA (0.5 μmol/L). Thus, in this study, the RA did not have a marked effect on BMSCs. However, it evidently promoted the proliferation and differentiation of BDNF-BMSCs synergistically with BDNF.

Conclusions
In conclusion, we have confirmed that BDNF enhances the differentiation of BMSCs into nerve-like cells in vitro. We examined the characteristics of the differentiated nerve-like cells with regard to morphological features, BDNF secretion, and expression of neural cell markers. Some investigators have evaluated whether the differentiated cells possess the structural and functional characteristics of neuronal cells based on the ultrastructural and electrophysiological aspects of the cells, obtaining promising, although preliminary, results (Akiyama et al. 2002). The cellular differentiation mechanisms will be the focus of future research. The present findings may be useful for the treatment of nervous system diseases.