Cell and Tissue Research

, Volume 318, Issue 1, pp 261–273

Stem cell therapy for Parkinson’s disease: where do we stand?


    • Section for Neuronal Survival, Wallenberg Neuroscience CenterLund University
  • Nicolaj S. Christophersen
    • Section for Neuronal Survival, Wallenberg Neuroscience CenterLund University
  • Patrik Brundin
    • Section for Neuronal Survival, Wallenberg Neuroscience CenterLund University
  • Jia-Yi Li
    • Section for Neuronal Survival, Wallenberg Neuroscience CenterLund University

DOI: 10.1007/s00441-004-0946-y

Cite this article as:
Roybon, L., Christophersen, N.S., Brundin, P. et al. Cell Tissue Res (2004) 318: 261. doi:10.1007/s00441-004-0946-y


A major neuropathological feature of Parkinson’s disease (PD) is the loss of nigrostriatal dopaminergic neuron. Patients exhibit motor symptoms, including bradykinesia, rigidity, and tremor. Neural grafting has been reported to restore striatial dopaminergic neurotransmission and induce symptomatic relief. The major limitation of cell replacement therapy for PD is the shortage of suitable donor tissue. The present review describes the possible sources of cells, including embryonic stem cells and somatic adult stem cells, both of which potentially could be used in cell therapy for PD, and discusses the advantages and disadvantages of each cell type.


Parkinson’s diseaseNeural graftingEmbryonic stem cellsSomatic adult stem cells


Parkinson’s disease (PD) is the second most-common neurodegenerative disease and affects around 2% of the population over 65 years of age. The neuropathology involves several brain regions, but the most crucial element is believed to be the loss of dopaminergic (DAergic) neurons in the substantia nigra (SN). This cell death is believed to play a central role in the development of the classic triad of symptoms, i.e. tremor, rigidity, and hypokinesia. The loss of DAergic neurons is commonly associated with the formation of intracytoplasmic protein aggregates known as Lewy bodies (Giasson and Lee 2003). Despite major advances over the past decade, the etiology of PD remains unclear. Occurring most commonly as the sporadic idiopathic form, PD has been hypothesized to be attributable to genetic predispositions to either endogenous toxins or environmental factors such as pesticides, herbicides, and industrial chemicals. The neurodegenerative process in PD has been associated with increased protein misfolding, oxidative stress, mitochondrial dysfunction, and excitotoxicity (Olanow and Tatton 1999; Ciechanover and Brundin 2003). The recent identification of genetic mutations in rare familial cases of PD has provided important insights into the molecular pathogenesis of this disease. At present, four identified genes have been clearly linked to PD: alpha-synuclein, parkin, ubiquitin C-terminal hydrolase L1 (UCH-L1), and DJ-1 (Dawson and Dawson 2003; Hardy et al. 2003). The list of other gene candidates that may also be involved in PD is steadily increasing.

The current primary treatment strategies revolve predominantly around pharmacological dopamine replacement. L-DOPA (L-3,4-dioxyphenylalanine), the precursor of dopamine, is the most commonly used and, in combination with a DOPA-decarboxylase inhibitor, can reverse many of the motor symptoms of PD in mild to moderately affected patients. However, after 5–10 years of treatment, there are typically daily and partly unpredictable fluctuations in the efficacy of L-DOPA treatment, and the onset of abnormal movements, termed dyskinesia, is common. Surgical therapies such as pallidotomy and deep brain stimulation or high frequency stimulation of the subthalamic nuclei are also used to treat some patients who have reached the complication phase in pharmacological therapy. High frequency stimulation has been applied clinically since 1993 and improves tremor, bradykinesia, and rigidity by up to 60% and reduces L-DOPA-induced dyskinesia (Limousin et al. 1995; Gross et al. 1999). The procedure may induce an inhibitory effect on the hyperactivity of the subthalamic nucleus of PD patients, producing a depolarization state in which the subthalamic neurons are unable to generate action potentials. However, the action mechanism is still under debate and unclear (Ashkan et al. 2004).

Another promising approach to the treatment of PD has been the replacement of the lost DA neurons by neural transplantation with embryonic nigral tissue. Preclinical trials have demonstrated that embryonic ventral mesencephalic tissue can survive, give rise to physiologically active neurons that are capable of releasing dopamine, and form connections with host neurons. Following transplantation of human embryonic neurons into the putamen of PD patients, successful cases display an increase in 18F-DOPA uptake and some amelioration of their motor symptoms (Piccini et al. 1999). However, the response of patients to transplantation is highly variable, raising questions about the standardization of the procedure. Many factors need to be considered, including the age of the patient, the use of multiple embryos from different sources, the storage time of embryonic tissue, and the immunosuppression treatment that follows transplantation (Björklund et al. 2003). Another issue that needs to be considered is graft-induced dyskinesia. Although cell transplantation is efficient in some patients, dyskinesia after embryonic nigral transplantation could be a serious problem. In open-labeled and double-blind clinical studies, 15%–50% of grafted patients have exhibited graft-induced dyskinesia of different severities (Freed et al. 2001; Hagell et al. 2002; Olanow et al. 2003). Moreover, the unpredictable variability in the outcome of the surgery is a cause for concern. Even when a relatively similar transplantation procedure is performed in two patients who are affected to a similar extent by PD, the outcome can differ dramatically. Without a better understanding of this variability in outcome, neural transplantation is unlikely to develop into a widely used therapy. Nevertheless, the greatest limitation to extensive testing and application of neural transplantation is probably the limited availability of human embryonic DAergic neurons (Brundin et al. 2000). Failure to deal with this issue alone may preclude the widespread use of human embryonic neural tissue for transplantation in PD. For example, without a reliable source of donor tissue, not only will it be impossible to graft large numbers of patients, but it will also be difficult to deepen our understanding of graft-induced dyskinesias and to identify which patients are most likely to benefit from the operation.

Human nigral grafts are only made up of 5%–10% of neurons that are destined to become DAergic neurons, the rest of the cells being other neuronal and glial cell types. Furthermore, only a small fraction, estimated at around 5%–10% of the cells destined to become DA neurons, actually survive the grafting procedure (Brundin et al. 2000). Despite the reported clinical benefit of neural grafting in PD patients, only about 400 patients have been transplanted worldwide, in part because of the shortage of embryonic donor tissue. Therefore, there is a need for an alternative cell source with the ability to expand indefinitely, constituting an unlimited standardized pool of cells. These cells should then be capable of differentiating into neurons that extend axons, form synapses, and produce and release dopamine in a regulated fashion.

Stem cells might provide a source of cells for replacement therapy. They are defined as undifferentiated cells that have high proliferative potential and can generate a wide variety of differentiated progeny (Gage 2000). They can be derived from either embryonic or adult tissue sources. They have several advantages. For example, they can be made available in large numbers, are relatively easy to maintain in culture and, by definition, exhibit a high differentiation potential and could therefore provide the missing DA neurons for grafting in PD. In this review, we give an overview of the various stem cells that might be used as a source of donor tissue for cell replacement therapy in PD.

Embryonic stem cells


Embryonic stem (ES) cell is a general term commonly applied to stem cells that are derived from the early developmental stages of an embryo. They can be divided into three different types: ES cells, embryonic germ (EG) cells, and embryonic carcinoma (EC) cells.

ES cells are derived from the inner cell mass of a blastocyst. The first ES cell lines were isolated in 1981 from mouse blastocysts, and then 17 years later, human ES cell lines were generated (Evans and Kaufman 1981; Martin 1981; Thomson et al. 1998; Reubinoff et al. 2000). They possess all the characteristics of true stem cells, being able to proliferate with no limitation of self-renewal and giving rise to cells that are derived from all three germ layers (Burdon et al. 2002). Their high telomerase activity means they are not so sensitive to senescence and are suitable for long-term culture (Thomson et al. 1998). EG cells are derived from the primordial germ cells of the embryo or fetus (5–10 weeks old for human EG cells). During development of the blastocyst, a subset of epiblast-derived cells forms the primordial germ cells, which undergo a complex migration to the genital ridges. It is from these EG cells that the gametes develop. After isolation by microsurgery, both ES cells and EG cells can be maintained in an undifferentiated stage in culture and generated as cell lines. EC cells are cells that can give rise to tumors, termed teratocarcinomas. Among the cells that compose teratocarcinomas, some undifferentiated cells possess stem cell properties. These cells are termed EC cells. Similar to ES and EG cells, EC cells can be expanded continuously and differentiated under certain conditions to give rise to the three types of germ layer cells.


As mentioned earlier, stem cells are defined as undifferentiated cells that can exhibit self-maintenance, generate a large number of progeny (principally of the phenotype of the tissue in which they reside, but also of other cell types), and retain their multilineage potential over time. ES, EG, and EC cells all possess such characteristics. Upon transplantation, ES cells may develop teratomas that contain tissues from all germ layers. When human ES cells are xenografted beneath the testis capsule of severe combined immunodeficient mice, they develop into tumors within 5 weeks. No metastasis has been found, and all the teratomas contain tissues representative of all three germ layers (Reubinoff et al. 2000). In addition, environmental factors can influence the differentiation fate of the ES cells into a specific type of cells. Mouse ES cells have been transplanted into the knee joint to assess their ability to form cartilage; 8 weeks after transplantation, the knee was destroyed by teratoma formation (Wakitani et al. 2003). These results indicate that ES cells are pluripotent. ES, EG, and EC cells have all been subjected to various manipulations to assess their differentiation potential. Engineered ES cells have been reintroduced into blastocystes after adult somatic nuclear transfer. They become a part of inner cell mass and contribute to the development of a variety of tissues in the resulting chimeric animals (Wakayama et al. 2001), also demonstrating their pluripotency. As mentioned above, they have high telomerase activity, which prevents the shortening of the telomeres at the ends of the chromosomes in conjunction with each mitosis. Maintenance of telomere length allows them to undergo unlimited division. Thus, human ES cells can be kept undifferentiated in culture and passaged for more than 1 year, undergoing at least 80 population doublings (Thomson et al. 1998; Rosler et al. 2004).

Several markers have been identified for undifferentiated ES, EG, and EC cells. They include Oct-3/4, Nanog, Sox2, LeftyA, Thy-1 cell surface antigen, FGF4, Rex1, SSEA-3, SSEA-4, TRA-1-60, TRA-1-81, and TRA-2-54 (Thomson et al. 1995, 1998; Chambers et al. 2003; Rosler et al. 2004). Certain markers are only expressed by one of the specific subtypes of ES cells, i.e., ES, EC, or EG cells. For example, human EG cells express the lactoseries glycolipid SSEA-1, which is not present in human EC and ES cells (Shamblott et al. 1998). Unlike human EC cell lines, both human ES and EG cell lines can maintain a normal karyotype during prolonged periods of culture (Thomson et al. 1998). However, human ES cells have recently been shown to have chromosomic changes like EC cells, resulting in the appearance of trisomy for chromosome 17q and/or chromosome 12 (Draper et al. 2004). The vast majority of cell culture protocols used for ES cells involve co-culture with another cell type that apparently supplies the stem cells with a range of essential, but largely uncharacterized, growth factors. It is a major challenge to grow ES, EG, and EC cells in the absence of such feeder cells under so-called feeder-free conditions (Xu et al. 2001).

Differentiation and application

Their high differentiation capacity makes ES cells interesting candidates for cell replacement therapy. ES cells have the potential to differentiate into a neural lineage, including DAergic neurons. Therefore, this type of stem cells could provide a source of donor tissue for transplantation to PD patients.

So far, the majority of studies have focused on mouse ES cells. To evaluate whether these cells can adopt a DAergic phenotype, most studies have examined the expression of tyrosine hydroxylase (TH), the rate-limiting enzyme in dopamine synthesis, either by using immunohistochemistry or Western blotting. Alternatively, the reverse transcription/polymerase chain reaction technique (RT-PCR) has been used to monitor the presence of message for the crucial enzyme. Although TH is an important marker for DAergic neurons, expression of TH alone is a not sufficient criterion to establish that DAergic neurons are present. For example, ventral DAergic neurons should co-express markers such as vesicular monoamine transporter (VMAT), dopamine transporter (DAT,) and TH, but at the same time, the cells should lack expression of dopamine beta hydroxylase (DBH), which is the enzyme for noradrenaline synthesis. Furthermore, many studies have employed other criteria such as the expression of DAergic neuron-related transcription factors, e.g., Nurr1, Pitx-3, and Lmx1b that appear to be important for DAergic differentiation (Smidt et al. 1997; Saucedo-Cardenas et al. 1998; Smidt et al. 2000) and assays to detect the production and release of DA itself. In addition, electrophysiological analysis of cells derived from differentiated ES cells has been used to determine whether they have neuronal membrane properties. The most stringent approach to testing whether ES cells have differentiated into true DAergic neurons is to graft them to the adult brain of animals with DA-depleting lesions. In this frequently employed paradigm, it is possible to study whether they can undertake behavioral functions normally associated with DAergic neurons. When grown in vitro under conditions that promote proliferation, ES cells do not spontaneously differentiate into DAergic neurons. A significant proportion of mouse ES cells dissociated into a single cell suspension has been reported to differentiate into DAergic neurons when grafted to the striatum of immunosuppressed rats (Björklund et al. 2002). The mechanisms that cause the mouse ES cells to adopt a neuronal phenotype when separated from their neighbors are not understood, and neuronal differentiation has been suggested to represent a “default” pathway of differentiation for mouse ES cells (Tropepe et al. 2001). However, although single cell suspensions of mouse ES cells can generate DAergic neurons when transplanted, they also frequently lead to the formation of teratomas. For safety reasons, this precludes the use of a similar method to generate donor tissue for transplantation in PD patients.

Various strategies have been tested to promote DAergic differentiation, such as growth factor exposure, co-culture with different types of feeder cells, genetic manipulation of the ES cells, or combinations of these strategies. With respect to soluble factors that promote DAergic differentiation, a combination of sonic hedgehog (SHH) and fibroblast growth factor 8 (FGF8) has been found to increase the differentiation of mouse ES cells into DAergic neurons. One of the first protocols that allowed ES cells to differentiate into DAergic neurons was based on a five-step culture procedure. In the final step, ascorbic acid was added to the culture medium and caused a dramatic increase in TH-immunopositive neurons (Lee et al. 2000). When transplanted into animals with unilateral 6-hydroxydopamine (6-OHDA) lesions of the nigrostriatal pathway, green fluorescent protein (GFP)-ES cells survive and exhibit immunoreactivity for TH after 8 weeks. Furthermore, they normalize motor asymmetry in a drug-induced rotation test (Nishimura et al. 2003). In addition, survival-promoting factors, such as interleukin-1 beta, transforming growth factor beta, and glial-cell-line derived neurotrophic factors (GDNF), applied at the latest stages of differentiation have been shown to promote and enhance DAergic differentiation of ES cells.

Other growth factors, such as neurotrophic factors, in combination with survival-promoting factors have been shown to promote the differentiation of ES cells into neurons (Rolletschek et al. 2001).

In an important series of studies, enhanced DAergic differentiation has been demonstrated in a co-culture system with ES cells grown on a feeder cell line (PA6) derived from bone marrow stroma of mouse skull (Kodama et al. 1982, 1984). This effect of the feeder cells on DAergic differentiation in ES cells has been called stromal-cell-derived inducing activity (SDIA; Kawasaki et al. 2000). When mouse ES cells or primate non-human ES cells were co-cultured with PA6 cells, up to 80% of the ES colonies contained TH-positive cells after 2 weeks induction. At the cellular level, among the 25% of the cells that expressed the neuronal marker beta-III tubulin, 35% were also TH-positive. Those neurons were able to release dopamine upon depolarization by high potassium stimulation. When differentiated cells derived from mouse (and later also primate) ES cell cultures were implanted into the striatum of mice with 6-OHDA lesions, surviving DAergic neurons were observed 2 weeks after grafting (Kawasaki et al. 2000, 2002; Morizane et al. 2002). It is still not clear which PA6-cell-derived factor(s) contribute to DAergic differentiation. When ES cells are grown in the presence of PA6 cells, SDIA is maintained even after the PA6 cells are fixed with 4% paraformaldehyde, suggesting that a cell surface molecule maintains its biological activity, despite formalin fixation. Other stromal cells, such as MS5 and S17 (both murine bone marrow stromal cell line) or primary stromal feeder cells obtained from the aorta gonad-mesonephros (AGM) region, have also been reported to promote the differentiation of ES cells into DAergic neurons (Barberi et al. 2003). Thus, when mouse ES cells are co-cultured with such stromal feeder cells, they differentiate into DAergic neurons if exposed to soluble growth factors added to the culture medium. DAergic differentiation can be induced in mouse ES cells growing on AGM stromal feeder cells when the culture medium is initially supplemented with SHH and FGF8, and later ascorbic acid and brain-derived neurotrophic factor. After grafting, TH-positive neurons survive and significant improvements in amphetamine-induced and apomorphine-induced rotation have been observed after only 3 weeks.

Another approach to induce the differentiation of DAergic neurons is the over-expression of specific DA-neuron-related transcription factors by genetic engineering (Saucedo-Cardenas et al. 1998; Wagner et al. 1999; Cazorla et al. 2000). This approach has been successful to some extent, although one can predict at the outset that it is difficult to recapitulate the temporally controlled concert of transcription factors that control DAergic differentiation during embryogenesis. Nurr1 is an orphan nuclear receptor that has been implicated in the differentiation of mesencephalic DA neurons. Thus, mice that are null-mutant for Nurr1 have been found to lack mesencephalic DA neurons (Zetterstrom et al. 1997). Stable Nurr1 over-expression in mouse ES cells results in 80% of the cells being beta-III tubulin-positive, when the cells are grown in the presence of the growth factors SHH and FGF8. The differentiated Nurr1 over-expressing ES cells exhibit many molecular, morphological, and functional features of midbrain DAergic neurons. For example, hemiparkinsonian rats grafted with cells derived from the Nurr1 ES cells show marked behavioral improvement in several tests of drug-induced and spontaneous motor function (Chung et al. 2002; Kim et al. 2002). Furthermore, not only does Nurr1 induce a DAergic phenotype in neural cell populations; when expressed in late stages of differentiating ES cells, it can induce the expression of DAergic markers in non-neuronal cell populations (Sonntag et al. 2004).

Another method aimed at obtaining large numbers of DAergic neurons from ES cells uses the natural expression of specific transcription factors as a selection marker for development along the DAergic lineage. ES cells have been generated from transgenic mice that express GFP inserted into the transcription factor Pitx3. The Pitx3 gene is expressed in mesencephalic DAergic neurons (Smidt et al. 2000). The Pitx3-deficient aphakia mouse mutant has a double genomic deletion in the Pitx3 locus and can be considered as a null-mutant for the gene. Use of this animal model to study the role of the gene has shown that Pitx3 is involved in the terminal differentiation of mesencephalic DAergic neurons (Smidt et al. 2004). Fluorescence activated cell-sorting of the subpopulation of ES cells expressing GFP, and thereby also expressing Pitx3, has been found to yield a population enriched in DAergic neurons (Zhao et al. 2004). Thus, the use of reporter expression systems can be a powerful tool for the purification of differentiating DAergic neurons for transplantation.

In order to optimize DAergic differentiation, anti-apoptotic genes such as Bcl-XL have been over-expressed in ES cells. The in vitro differentiation of these cells compared with wild-type ES cells has resulted in the higher expression of midbrain DAergic markers. Following transplantation, Bcl-XL over-expressing ES cells exhibit more extensive fiber outgrowth and support a more pronounced amelioration of behavioral symptoms than do transplanted wild-type ES cells (Shim et al. 2004).

Advantages and disadvantages

As evident from the data discussed above, many advances have been made regarding our knowledge of mouse ES cells and their capacity to differentiate into DAergic neurons. However, it is not clear to what extent one can apply this knowledge to human ES cells, since species differences presumbaly exist with respect to the molecules that control differentiation in ES cells. A recent report has demonstrated that efficient neuronal differentiation of human ES cells can occur when the cells are cultured under serum-free conditions in the presence of 50% conditioned medium from human hepatocarcinoma cell line HepG2 (Schulz et al. 2003). Expanded neural-progenitor-derived human ES cells can indeed give rise to neurons upon differentiation. However, when transplanted into the lateral cerebral ventricles of newborn mice, the neurons survive, but DAergic differentiation is not efficiently achieved: less that 1% of the cells are TH-immunopositive (Reubinoff et al. 2001).

Neuronal differentiation from EC and EG cells has also been demonstrated. However, the differentiation potential into a DAergic phenotype has so far been low in EG and EC cells. Nevertheless, their ability to differentiate into neurons is interesting. Human EC cells can form neurospheres when cultured with differentiation medium supplemented with retinoic acid. These neurospheres, generated either as free-floating or adherent cultures, give rise to neurons immunopositive for the neuronal markers beta-III tubulin, microtubule-associated protein 2 (MAP2), and neurofilament-68kD (Horrocks et al. 2003). Interestingly, neuronal differentiation can also be achieved by over-expression of the soluble factor Wnt-1 in the EC cell line P19, even without exposing the cells to retinoic acid (Tang et al. 2002). This same EC cell line has been evaluated in a transplant paradigm in a rat model of PD. In this experiment, the objective was not to replace lost neurons, but to protect the endogenous DAergic neurons from death. The cells were first genetically engineered to produce GDNF. After an exposure of 48 h to retinoic acid, GDNF-producing EC cells were transplanted into the midbrain of intact rats. To examine whether the GDNF cells could protect the DA neurons of the rats against a neurotoxic insult, axonal terminals were lesioned by 6-OHDA injection into the striatum (Sauer and Oertel 1994). When the rats were examined 4 weeks later, the grafts expressing GDNF were found to have prevented the death of nigral DAergic neurons (Nakao et al. 2000). Neuronal differentiation has also been achieved with EG cells (Shamblott et al. 2001), and their ability to restore paralysis in animals with injuries to the spinal cord has also been evaluated (Kerr et al. 2003).

All ES cells can be expanded in culture while retaining the functional attributes of pluripotent, early embryo cells. This makes them highly interesting as a potential source of donor tissue for transplantation in PD. Furthermore, to their advantage, they are able to display a stable karyotype, although not without exception (Draper et al. 2004), and can be easily cryopreserved. As previously described, genetic modifications can be made by transgenesis or vector targeting, in order to increase ES differentiation potential. However, certain drawbacks must also be considered. So far, ES cells have not been shown to produce a pure specific population of cells after differentiation. Moreover, some of the ES cells in cultures derived from embryoid bodies do not undergo differentiation. Such cells should be removed or eliminated prior to transplantation, as otherwise they could be the source of unwanted cell proliferation. In some studies, this potential problem has been addressed experimentally, the ES cells having been treated with anti-mitotic drugs prior to transplantation in order to avoid tumor formation (Kawasaki et al. 2002). Another problem with the use of cells derived from ES cells for grafting is that they are often grown with feeder cells that typically are of xenogenenic origin. Therefore, purification of the ES cells also has to be carried out when ES cells have been cultured with feeder cells, in order to avoid contamination with non-desired cells.

Somatic stem cells

Somatic stem cells are undifferentiated stem cells that are tissue-/organ-specific and persist throughout adulthood. They are committed stem cells that can self-renew and differentiate, under physiological conditions or after injury, to regenerate the tissues in which they normally reside. The most commonly studied somatic stem cells are neural stem cells (NSC) and bone marrow stem cells (BMC). Recent studies have demonstrated that both NS and BMC can be considered as a new pool of donor cells for transplantation in neurodegenerative disease.

Neural progenitors

NSC can be isolated from various parts of the brain, such as the midbrain and forebrain (Fig. 1A), and expanded and differentiated into neurons, astrocytes, and oligodendrocytes. Neural progenitors can be cultured as a primary culture (Fig. 1B) or be expanded and grown as neurospheres in appropriate media supplemented with mitogenes, such as EGF and bFGF. The survival of differentiated neurons can be increased by neurotrophic factors (Svendsen et al. 1997; Caldwell et al. 2001; Kawasaki et al. 2002). However, the neuronal (including DAergic) differentiation properties of neural progenitors seem to be dependent upon the brain regions from which they have been isolated. In other words, they display regional specification. For example, neurospheres derived from cortical progenitors give rise to more neurons than those generated from the ventral mesencephalon (Fig. 1C). Interestingly, cortical neurospheres give rise to more TH-immunopositive neurons than ventral mesencephalic neurospheres when differentiated (L. Roybon et al. unpublished observations; Fig. 1D). In addition, the proliferation potential of neural progenitors differ substantially between different species. Human neurospheres can be passaged more than 40 times, whereas rat neurospheres can only be expanded for ten population doublings (Ostenfeld et al. 2002). Interestingly, mouse neurospheres can be expanded for much longer than rat neurospheres (Ostenfeld et al. 2002). Before grafting into rodent models of PD, NSC have been expanded in the presence of bFGF and differentiated into neurons with mature features by using soluble growth factors, such as SHH (Matsuura et al. 2001). Thus, expanded ventral mesencephalic precursors from E12 rat embryo were differentiated as floating neurospheres and transplanted into hemi-parkinsonian rats; the cells survived for up to 80 days and produced a dramatic reduction in amphetamine-induced rotation (Studer et al. 1998). However, the survival rate of grafted DAergic neurons derived from the expanded precursor was low (3%–5%) and similar to commonly nigral grafted cells (Brundin et al. 2000), This suggests a need to improve cell survival following cell grafting (Brundin and Björklund 1998). In other studies, genetic modifications of NSC have also been performed. In some cases, the genetic modifications initially performed in NSC have been instructive concerning factors that increase DAergic differentiation, and in later studies, the same approach has been used in ES cells (see above). For example, in addition to the orphan nuclear transcription factor Nurr1, the transcription factors Pitx3 and lmx1b (Sakurada et al. 1999; Smidt et al. 2000) have also been extensively studied and used to engineer NSC (Kim et al. 2003a, 2003b; Wallen-Mackenzie et al. 2003). When the capacity of Nurr1 to induce DAergic differentiation was assessed for different region-specific neural precursors, the virally Nurr1-transduced ventral mesencephalic and cortical progenitors had DAergic characteristics: cells were immunoreactive for DAergic markers and released dopamine in culture conditions in vitro (Kim et al. 2003a, 2003b). However, when Nurr1-transduced cortical or midbrain E14 neural precursors were transplanted into the ipsilateral striatum of 6-OHDA lesioned rats, no reduction of apomorphine-induced rotation scores was recorded. In contrast, a significant decrease in rotation scores was observed in rats grafted with E12 midbrain precursors. The lack of recovery might have been attributable to the low survival of grafted cells. In addition, and probably more importantly, the decreased capacity of dopamine release of grafted cells in vivo and the poor synaptic interaction with the host brains could also have contributed to the failure of behavioral improvement (Kim et al. 2003a, 2003b).

Although the efficient generation in vitro of DAergic neurons from neural precursors is of great interest because of the clinical need for cell transplantation in PD cases, challenges remain to be overcome. Thus, limitations in the long-term propagation and difficulties in DAergic differentiation are still unresolved. The efficiency of DAergic differentiation decreases in midbrain neural precursors after proliferation or passaging for extended periods in vitro, although the reduction can be partly restored by ascorbic acid treatment (Yan et al. 2001). Furthermore, the maturation of expanded neural precursors in vivo is also an issue. Although expanded precursors can differentiate into dopamine-producing neurons in vitro, these cells appear, after transplantation, to be less mature and to integrate less well with the host brain compared with non-expanded precursors (Ostenfeld et al. 2002; Kim et al. 2003a, 2003b). These findings have important implications for our understanding of the nature of proliferating neural precursors isolated from the developing brain and their potential use in PD cell replacement therapy.
Fig. 1A–F

Possible donor cells for transplantation in PD. A E11 rat embryo. Neurospheres can be generated from neural stem cells isolated from the ventral midbrain (red arrowhead) and the forebrain (black arrowhead). B Ventral mesencephalic primary culture from a rat at embryonic day 14.5 (E14.5). Neurons were stained in red for the marker MAP2, and astrocytes in purple for glial fibrillary acidic protein (GFAP). C Three-day differentiation of E14.5 rat ventral mesencephalic neurosphere. Neurons were stained for β-III-tubulin (red) and astrocytes for GFAP (green). D Dopaminergic neurons derived from a E14.5 cortical neurosphere are immuno-positive for TH. E MesII cells expressing TH-immunostaining (red) after differentiation. F Differentiated adult mouse GFP hematopoietic stem cells (green) stained for the neural precursor marker nestin (red). B, C, E were counterstained using the nuclear marker Hoechst (blue)

Immortalized neural progenitors

The use of NS for cell banking and transplantation may not be optimal, since their mitotic competence is limited (Ostenfeld et al. 2000). This means that neural stem/progenitor cells can be stimulated to grow by using mitogens only until they approach their natural senescence limit in culture (epigenetic expansion). An alternative approach has been to immortalize neural stem and progenitor cells (Frederiksen et al. 1988), by arresting cells at specific stages of development and preventing their terminal differentiation (Cepko 1989).

The most extensively used approach is to immortalize cells from developing brain with a retroviral vector encoding the propagating-enhancing v-myc protein (Ryder et al. 1990; Snyder et al. 1992). V-myc propagated neural progenitor cell lines include C17.2 (derived from developing mouse cerebellum; Ryder et al. 1990; Snyder et al. 1992), H6 (derived from human 15-week-old telencephalon; Flax et al. 1998), HNSC.100 (derived from human 10-week-old forebrain cultures grown as neurospheres; Villa et al. 2000), and MesII (derived from human 8-week-old ventral mesencephalon; Lotharius et al. 2002). By constitutively expressing oncogenes such as v-myc, cell lines proliferate continuously in culture, although they are still dependent on the presence of mitogens, such as bFGF, EGF, or serum, to divide. In the absence of mitogens, the cells exit from the cell cycle and differentiate (Kitchens et al. 1994; Sah et al. 1997; Flax et al. 1998; Villa et al. 2000). As an alternative to developing cell lines by constitutively expressing oncogenes, some groups have taken advantage of a tetracyclin-controlled gene expression system (Hoshimaru et al. 1996; Sah et al. 1997). Use of this tetracyclin-regulated v-myc vector design has generated a human mesencephalic progenitor cell line, MesII (Lotharius et al. 2002). The constitutive expression of v-myc, in the absence of tetracyclin, allows the cloned MESII(1)C2.10 [MesII] cells to proliferate continuously in culture. Upon seeding on poly-d-lysine/laminin-treated plates in N2 medium containing tetracyclin with dibutyryl cAMP and GDNF, MesII cells develop into a neuronal phenotype. Under these conditions, they are found to display neurite extension, generate action potentials, express TH, and produce dopamine (Fig. 1E). Thus, human mesencephalic DAergic neurons can be generated in a controlled and predictable manner in vitro. Although a tetracyclin-regulated v-myc vector design may not be considered safe for clinical application, MesII cells provide a proof-of-principle that human DAergic cell lines can be generated and illustrate that immortalized cell lines may be of relevance for stem cell therapy for PD.

Another approach to the generation of neural progenitor cell lines of relevance to stem cell therapy for PD is the redirection of the differentiation of existing neural stem cell lines to a DAergic phenotype. For example, over-expression of Nurr1 in the C17.2 stem cell line promotes differentiation into TH-positive neurons, when co-cultured with astrocytes in vitro (Wagner et al. 1999). As mentioned earlier, maintenance mechanisms for telomere length play an important role for continued cell proliferation (Harley et al. 1990; Bryan et al. 1995; Sharma et al. 1995; Lundblad and Wright 1996; Bodnar et al. 1998). Interestingly, v-myc expression may be involved in telomerase activation, since telomerase is a direct target gene regulated by myc (Wu et al. 1999). Telomerase is active in germ cells, repressed early in the embryonic development of somatic cells, and reactivated in somatic cells immortalized during tumor genesis (Ostenfeld et al. 2000). Recently, neuronal progenitors from human embryonic spinal cord were genetically modified to over-express the telomerase reverse transcriptase that permitted the generation of lineage-restricted human neural progenitor cell lines; these cell lines remained phenotypically and karyotypically stable for at least 120 divisions and were able to differentiate into mature cells of relatively restricted and uniform phenotypes in vitro (Roy et al. 2004).

The developmental potential of immortalized neural cells has been assessed in transplantation studies. Differentiation of immortalized progenitor cells into neurons requires the sufficient down-regulation of the oncogene in vivo to allow the differentiation programs to proceed unimpeded and to avoid the formation of tumors after grafting. V-myc has been reported to be down-regulated following intracerebral transplantation (Snyder et al. 1997; Flax et al. 1998), although detection of v-myc mRNA is still possible by RT-PCR (Rubio et al. 2000). Thus, in stem cell clones in which propagation is assisted by a cell cycle regulatory gene, that gene product is constitutively and spontaneously down-regulated as soon as the cells enter the brain. Whether this degree of down-regulation is sufficient to meet the level of safety required in clinical trials is unclear. Nevertheless, in experimental animals, cell lines generated by the constitutive expression of immortalizing oncogenes have been shown to be non-tumorigenic (Renfranz et al. 1991). Following transplantation into neurogenic regions, HNSC.100 cells stop dividing after 2 days and spontaneously generated neurons, astrocytes, and oligodendrocytes (Rubio et al. 2000; Villa et al. 2002). However, when transplanted into the striatum and substantia nigra of the adult intact rat brain, an experimental condition of relevance to neural replacement in PD, the HNSC.100 cell line displayed less neuronal differentiation in vivo than in vitro, and these cells failed to generate DAergic neurons after transplantation (Rubio et al. 2000).

An immortal source of cells would provide a supply of unlimited numbers of homogenous and stable cells for research, would allow controlled genetic modifications to be carried out, and would be highly suited for the extensive characterization and validation data needed for their clinical use. However, our awareness of the potential risks of creating transformed cells characterized by the loss of growth-control mechanisms, e.g., the loss of contact inhibition, the ability to grow in soft agar, the ability to give rise to tumors in the nude mouse, and the inability to respond to normal signals to withdraw from the cell cycle, make the clinical use of such cells unlikely. To provide an extra level of assurance, it would be prudent to engineer cells with a CRE-loxP recombinase system to remove the immortalizing genes just prior to or just following implantation (Westerman and Leboulch 1996). Insights from studies of genetically propagated NSC should be compared with those from investigations of mitogen-mediated expansion alone to provide a more complete picture of NSC biology and its applications. In this regard, it should be established whether v-myc-propagated neural progenitor cell lines display phenotypic drift from passage to passage over prolonged periods of time. The failure to generate DAergic neurons after transplantation in vivo may indicate that the generation of neurons from immortalized cells is not a cell-autonomous property in vivo, but rather that it requires some external cues or a permissive environment. In the case of transplantation, immune rejection can occur when the transplanted cells are derived from other individuals and animals (allogeneic and xenogeneic transplantation).

Bone marrow stem cells

In order to prevent immune graft rejection, a new concept of cell donor for transplantation for neurodegenerative diseases has recently emerged. Somatic stem cells have been identified in tissues outside the brain, and stem cells from bone marrow have been suggested as an alternative source of donor tissue, serving as an autologous system for transplantation. There are at least two types of stem cell residing in bone marrow: mesenchymal (stromal) stem cells (MSC) and hematopoietic stem cells (HSC).

MSC are some of the cells that comprise the bone marrow niche supporting hematopoiesis. They have plastic properties, allowing them to differentiate into tendon, cartilage, bone, and fat. Recent evidence has suggested that these cells are capable of changing their commitment to a neuronal phenotype. Freshly isolated MSCs from adult bone marrow can express neuronal markers, including neuron-specific enolase (NSE), NeuN, neurofilament-M (NF-M), and tau protein under certain differentiation conditions (Woodbury et al. 2000). Even prior to differentiation, undifferentiated MSCs can express several markers characteristic of neural cells, such as NSE, class III beta-tubulin isotype, MAP1, and vimentin. The expression level of these markers is increased by the addition of isobutylmethylxanthine and dibutyryl cyclic AMP to the differentiation medium (Deng et al. 2001). How can the bone marrow stromal cells express neural markers prior differentiation? Recent data support the idea that some bone marrow cells could be directly derived from organs and reside in the bone marrow “niche” waiting for a signal indicating that they need to migrate to the organ from which they are derived (Ratajczak et al. 2004). Furthermore, ES, NS, and HSC cells can have overlapping protein expression profiles (Ramalho-Santos et al. 2002) suggesting that common pathways exist between different types of stem cells when they are undergoing differentiation. Additional evidence suggests that the brain microenvironment stimulate the differentiation of MSCs into neural cells. When infused into the striatum of rat brain, human MSCs have been reported to integrate and migrate. By 72 days after transplantation, the cells are found in the cerebral cortex, the temporal lobe, through the rostro-caudal axis, and even contralateral to the injected site (Azizi et al. 1998). When mouse MSCs labeled with bromo-deoxyribo-uridine were injected into the lateral ventricle of a 3-day-old postnatal mice, they migrated to both the forebrain and the cerebellum without disruption of the normal brain structure of the host (Kopen et al. 1999). This in vitro evidence supports the application of MSCs in transplantation protocols.

Transplantation of HSCs has been used for many diseases, for example, to treat leukemia or congenital immunodeficiencies. HSCs are able to generate different blood cells continuously throughout life. After irradiation, HSCs can repopulate the whole blood system. It is still debatable whether HSCs have the potential to transdifferentiate into other cell types, crossing the border and leaving the hematopoietic lineages. At the end of 2000, two papers suggested that blood cells can differentiate into neurons; the authors used similar approaches in different model systems (Brazelton et al. 2000; Mezey et al. 2000). They observed that small fractions of neuronal cells (0.3%–2.3%) in the central nervous system contained donor cell markers (sex mismatched, Y chromosome marker, or GFP marker) after the infusion of bone marrow cells into neonatal or irradiated adult mice. Thus, the data strongly suggest that blood-derived stem cells can differentiate into neurons. If these findings are validated, the implications for the treatment of various degenerative or traumatic diseases, including PD, will be enormous. However, as discussed below, the results and their interpretation are controversial. When injected into the lateral ventricle of neonatal mice, HSCs differentiate into microgial cells or astrocytes, but not neurons (Vitry et al. 2003). The neuronal differentiation of bone marrow cells (both MSCs and HSCs) has been suggested to be a rare event promoted by experimental systems rather than being a general or natural phenomenon (e.g., stroke, ischemia; Fig. 1F). Until now, no strong evidence has been provided showing that pure hematopoietic stem cells are capable of expressing neural markers upon differentiation, merely that they differentiate into microglial cells after transplantation into newborn rats. Furthermore, recent studies have demonstrated that BMC can fuse with different types of host cells after transplantation. Fusion phenomena are most robust in the liver, in which cell fusion has been shown to be the principal source of bone-marrow-derived hepatocytes in a model of liver disease with fumarylacetoacetate hydrolase deficiency (Vassilopoulos et al. 2003; Wang et al. 2003). However, in the brain, Purkinje cells are so far the only cell type with which grafted bone marrow stem cells can fuse (Alvarez-Dolado et al. 2003; Weimann et al. 2003); the formation of fused cells increased in a linear manner over 1.5 years, and after fusion, fused Purkinje cells exhibited dispersed chromatin and an active Purkinje-cell-specific transgene (Weimann et al. 2003). The significance of the cell fusion between the grafted bone marrow cells and the host Purkinje cells is not clear. However, cell fusion can occur during normal development, e.g., between myoblasts (Mintz and Baker 1967) and liver cells. Mouse hepatocytes normally have ploidy values of 2N, 4N, 8N, or 16N (Bohm and Noltemeyer 1981); this phenomenon has not been seen in neurons. Therefore, clear differences probably exist between different cell types regarding their ability to undergo cell fusion. These intrinsic natural differences may also contribute to the different observations of the differentiation of adult stem cells between different tissues. Moreover, cell fusion probably does not explain all the earlier reports of transdifferentiation, especially with the high efficiency of neural differentiation in 7%–57% of the neuronal stem cell population in co-culture assays (Galli et al. 2000; Rietze et al. 2001), especially when cells are not in co-cultured conditions (Jiang et al. 2002).

Mutipotent adult progenitor cells

Another cell type for transplantation is the multipotent adult progenitor cells (MAPC). MAPCs are a sub-population of MSCs and have the potential to differentiate into mesenchymal, visceral mesodermal, endodermal, and neuroectodermal cells. Such cells can be isolated from bone marrow, muscle, and brain (Jiang et al. 2002). In an elaborate culture system, rodent MAPCs have differentiated into specific mature neurons, such as GABAergic and DAergic neurons. Furthermore, the differentiated cells express functional sodium channels (Jiang et al. 2003). However, the propagation of these cell is not trivial, as they demand special handling with regard to cell density and additives.

Umbilical cord blood cells

Cells derived from blood taken from the umbilical cord have also been suggested to have the potential to differentiate into both neuronal and glial cell types. Around 30% and 40% of the cells adopt a neuronal and astrocytic fate, respectively (Buzanska et al. 2002). Upon differentiation, the neural-type precursor cells from cord blood also give rise to a relatively high proportion of oligodendrocytes (11% of the total population of differentiating cells). The advantage of using cord blood cells is that they are easy to obtain at birth and can be preserved for several years by cyropreservation. When co-cultured with primary cortical cells, neuronal differentiation is enhanced, suggesting that brain cells have the potential to induce and enhance neural differentiation by providing optimal paracrine neurotrophic effects. Such cod blood cells could potentially serve as a routine starting material for the isolation and expansion of cells for allogenic and autologous transplantations.

Concluding remarks

Among the stem cell types discussed above, each has its advantages and disadvantages when considered for cell replacement therapy in PD. For example, ES cells may have the highest proliferation and differentiation rates, but teratoma formation may constitute a real safety risk. Embryonic neural progenitor cells may readily undergo neuronal differentiation and, less commonly, also DAergic differentiation. However, these cells share with ES cells many of the problems related to ethical issues and immune responses after transplantation. Adult stem cells from bone marrow and from other tissues may serve as a potential source of cell replacement therapy, especially with the advantage of autologous grafting. However, the proliferation and differentiation of adult stem cells are still poorly understood. In particular, it is still debatable whether transdifferentiation can occur across the border of the germ layers. Therefore, in order to apply either of these cells in clinical trials in PD patients, many basic scientific issues need first to be solved.


We thank Emma Lane for critical reading of the manuscript.

Copyright information

© Springer-Verlag 2004