There are few drugs to effectively rescue the patients with ischemic stroke in spite of the huge efforts to develop them for longer than 50 years [1]. As alternative approach, cell therapy has recently been expected as one of the promising strategies to enhance functional recovery after ischemic stroke. A variety of cells have been studied as the candidate donor cells for this purpose. These include embryonic stem (ES) cells, neural stem cells, induced pluripotent stem (iPS) cells, and bone marrow stromal cells (BMSCs). Of these, the BMSCs may have the most enormous therapeutic potential among them, because they can be obtained from the patients themselves and easily expanded without posing any ethical and immunological problems. The BMSCs are non-hematopoietic cells in the bone marrow and regulate the proliferation and differentiation of hematopoietic cells. The transplanted BMSCs significantly enhance functional recovery after the insults in animal models of ischemic stroke. On the other hands, recent studies have shown that the adult stem cells, including BMSCs, are observed in the peripheral blood and play an important role in repairing the injured tissues [2, 3].

Based on these observations, some of preliminary clinical testing has already been conducted to evaluate the safety and therapeutic effects of BMSC transplantation for the patients with both acute and chronic neurological disorders [4–10]. However, it should be reminded that a variety of questions or problems still remain to be solved in order to establish BMSC transplantation as scientifically proven entity in clinical situation [2]. This article reviews recent knowledge on therapeutic impacts of BMSC transplantation on ischemic stroke.

Recent studies have shed light on the mechanisms through which the transplanted BMSCs enhance functional recovery after cerebral infarct. They aggressively migrate toward the damaged lesions through some chemokines. Recently, CXCR4, a specific receptor for stromal cell-derived factor (SDF)-1α, is believed to play an important role in their migration in the CNS [11]. There are few studies whether the engrafted BMSCs retain their proliferative activity in the host brain or not. Yano et al. (2005) labeled the GFP-expressing BMSCs with a superparamagnetic iron oxide (SPIO) agent and transplanted into the ipsilateral striatum of the mice infarct brain. As the results, they found that the BMSCs actively proliferate, migrate toward the lesion, and partially express the neuronal phenotype in the host brain after transplantation [12].

Nowadays, the BMSCs are known to produce some neuroprotective or neurotrophic factors and support the survival of the host neural cells [13]. This hypothesis is readily reasonable because the BMSCs per se support the homing, proliferation, and differentiation of the hematopoietic cells in the bone marrow by producing a variety of cytokines [14]. The conditioned medium of BMSCs significantly promotes neurite outgrowth from the dorsal root ganglion [15]. They also release soluble neuroprotective factors, including nerve growth factor (NGF), hepatocyte growth factor (HGF), and brain-derived neurotrophic factor (BDNF), and significantly ameliorate glutamate-induced damage of neurons [16]. The BMSCs markedly promote the neurite extension from the neurons in the organotypic slice of the brain and spinal cord [17, 18]. The BMSCs also protect the neurovascular integrity between basement membrane and astrocyte end-feet and ameliorate brain damage in stroke-prone spontaneous hypertensive rats [19]. Recently, Shichinohe et al. (2014) demonstrated that the BMSCs serve the “nursing effect” to the damaged neurons and activate the neural stem cells in the host brain by producing BDNF [20]. Therefore, the transplanted BMSCs trigger endogenous signaling pathways of survival and repair in neurons by secreting soluble neurotrophic factors.

Both neutrophils and macrophages are well known to play an important role in the early inflammation after cerebral infarct [21]. Indeed, their inflammatory response may be an essential process to clear cellular debris and initiate the healing pathways. Simultaneously, however, these inflammatory reactions may also give rise to cytotoxic damage to the surviving neurons, astrocytes, and endothelial cells in the peri-infarct area [21]. The BMSCs have currently been investigated as donor cells for novel cell therapy to prevent and to treat clinical disease associated with aberrant immune response. In the host, the BMSCs may attenuate pro-inflammatory cytokine and chemokine induction and reduce pro-inflammatory cell migration into sites of injury and infection [22]. Therefore, the transplanted BMSCs may prevent excessive inflammatory response and prevent further tissue damage in the peri-infarct area.

The BMSCs are believed to differentiate into neural cells in the host’s brain. This theory is based on the findings that BMSCs simulate neuronal morphology and express the proteins specific for neurons in vitro [23, 24] or in vivo [25, 26]. It may sound strange that the BMSCs have the ability to differentiate into the neural cells. However, the BMSCs per se express the genes related to neuronal and glial cells [27]. Recent studies also show that the BMSCs can alter their gene expression profile in response to exogenous stimuli and increase the genes related to the neural cells [27–29]. Sanchez-Ramos et al. (2000) showed that a small fraction of BMSCs cultured in epidermal growth factor (EGF) or retinoic acid/BDNF expressed nestin, NeuN, or GFAP and that the proportion of NeuN-expressing cells increased when BMSCs were cocultured with fetal mouse midbrain neurons [23]. Wislet-Gendebien et al. (2005) cocultured the BMSCs with cerebellar granule cells and assessed their fates. They found that the nestin-expressing BMSCs express other neuronal markers and that BMSC-derived neuron-like cells fire single-action potentials in response to neurotransmitters such as glutamate [30]. Hokari et al. (2008) also demonstrated that a certain subpopulation of the BMSCs morphologically simulated the neuron and expressed the neuron-specific proteins without any evidence of cell fusion, when cocultured with the neurons [16]. These findings strongly suggest that at least a certain subpopulation of the BMSCs have the potential to alter their gene expression profile and to differentiate into the neural cells in response to the surrounding environment. More importantly, the findings indicate that only the subgroup of BMSCs with potential of neural differentiation can survive in the host brain for a long time (>4 weeks). In fact, the engrafted BMSCs express γ-aminobutyric acid (GABA) receptor and improve the binding potential for ¹²⁵I-iomazenil in the peri-infarct area [31]. They also improve glucose metabolism in response to sensory stimuli when transplanted into the rat cold injury model [32]. Furthermore, ¹⁸F-fluorodeoxyglucose (FDG) PET study has very recently shown that the BMSCs markedly improve the recovery of glucose metabolism in the peri-infarct neocortex, when stereotactically transplanted into the infarct brain at 7 days postischemia [33]. According to recent work by Liu et al., the BMSC may enhance the axonal sprouting from the survived cortical neurons in the peri-infarct area [34]. Furthermore, Chiba et al. have recently found that the BMSCs are integrated into the neural circuits of the host spinal cord and promote functional recovery [35]. These biological properties of BMSC may play a key role to enhance functional recovery after ischemic stroke.

Based on these observations, the exogenous transplantation of BMSCs is now believed to enhance functional recovery through multiple mechanisms, including nursing effect, anti-inflammatory action, and neural cell differentiation, in patients with ischemic stroke.

As described above, the observations in basic experiments are quite encouraging, and some clinical trials of BMSC transplantation have already been started for patients with ischemic stroke. Bang et al. intravenously injected the autologous BMSC into five patients with severe neurological deficits due to ischemic stroke at 5–9 weeks after the onset and concluded that autologous BMSC infusion is a feasible and safe therapy that may improve functional recovery [4]. Honmou et al. intravenously transplanted the BMSC into 12 patients with ischemic stroke 36–133 days post-stroke [36]. Very recently, Lee et al. performed an open-label, observer-blinded clinical trial of 52 patients with ischemic stroke and followed up them for up to 5 years. As the results, they concluded that intravenous transplantation of autologous BMSC could be safe and effective strategy for ischemic stroke [9]. These studies indicate that BMSC transplantation may be at least safe and feasible for patients with ischemic stroke.

However, there are many variables that may affect the efficacy of BMSC transplantation in clinical setting. Thus, they include donor cell factors (safety, autologous, or allogeneic, ex vivo cell expansion), patient factors (age, stroke type), treatment factors (interval since onset, delivery route, cell dose), and validation factors (neurological assessment, imaging) [37, 38].

First, allogeneic cells would permit “off-the-shelf” use even within 24 h after the onset, but force a long-term medication of immunosuppressant. Autologous BMSC from patients themselves would be ideal as the donor cells for restorative medicine, but require several weeks for ex vivo expansion. Therefore, it should be scientifically proven that the BMSC can enhance functional recovery after ischemic stroke even when transplanted several weeks after the onset. More importantly, it would be critical to establish the feasible protocol to “safely and rapidly” expand the BMSC. Thus, the BMSCs have been cultured in the medium including fetal calf serum (FCS) in the majority of animal experiments and even clinical trials [4]. However, the FCS carries the potential risk of prion, viral, or zoonoses contamination. Alternatively, autologous serum is employed to expand the BMSC, but may require a large amount of serum [36]. Very recently, human platelet lysate (PL) is proven useful to expand the BMSC as the alternative substitute. The human BMSCs expanded with the FCS-free, PL-containing medium retain their capacity of migration, survival, and differentiation and significantly promote functional recovery when stereotactically transplanted into the infarct brain. Therefore, PL may be a clinically valuable and safe substitute for FCS in expanding the hBMSC to regenerate the infarct brain [39–41].

Second, the BMSCs are transplanted within 24 h or 7 days after the insults in the majority of animal studies, whereas they are usually transplanted several weeks (or even several months) after stroke onset in previous clinical trials [4, 9, 36]. In order to resolve this problem, pharmacological modulation may be useful to promote in vitro proliferation of the cultured BMSC to shorten the interval between stroke onset and cell therapy. For example, granulocyte colony-stimulating factor (G-CSF) significantly enhances the proliferation and growth factor production of the cultured BMSC and accelerates functional recovery by BMSC transplantation into the infarct brain [42]. Such pharmacological modulation would be essential in considering autologous stem cell therapy for older patients with ischemic stroke, because adult stem cells, including BMSC, suffer the effect of aging and reduce their self-renewal and differentiation capacity [43]. Very recent study has also demonstrated that G-CSF significantly promotes the proliferative capacity of BMSC harvested from the aged rats [44]. These observations should be taken into considerations when establishing the treatment protocol in clinical situation.

Third, the BMSC can be transplanted directly, intravenously, intra-arterially, or intrathecally. Although direct, intracerebral, or stereotactic injection permits most efficient delivery of the donor cells to the damaged tissue, less invasive procedure would be optimal. Intravenous or intrathecal transplantation is attractive because of its noninvasive, safe technique for the host CNS, but has been reported to result in less pronounced cell migration and functional recovery than direct cell transplantation [45]. Alternatively, the intra-arterial injection of BMSC may be valuable to noninvasively deliver them to the damaged CNS [46, 47]. Therefore, optimal transplantation technique should be developed to serve maximally safe and efficient results. However, there are limited numbers of studies that directly compare the therapeutic effects of these delivery routes under the same conditions. It is urgent issue that tests the effects of each delivery route on functional recovery after cerebral infarct. Recently, Kawabori et al. transplanted the BMSC into the infarct brain directly or intravenously at 7 days after the insult, that is, clinically relevant timing. As the results, they concluded that intravenous administration of BMSC has limited effectiveness at clinically relevant timing and intracerebral administration should be chosen for patients with ischemic stroke [48]. Furthermore, they directly transplanted the BMSC (1 × 10⁵ or 1 × 10⁶) into the infarct brain at 1 or 4 weeks after the insult and found that earlier transplantation requires a smaller number of donor cells for beneficial effects [49]. These observations strongly suggest the importance of timing, delivery route, and cell dose to yield therapeutic effects of BMSC transplantation for ischemic stroke. Similar translational research should thoroughly be conducted to establish the scientifically proven protocol prior to the start of clinical testing.