Dyskinesia in Parkinson's disease: mechanisms and current non-pharmacological interventions


  • Rolf Heumann,

    1. Molecular Neurobiochemistry, Ruhr-University Bochum, Bochum, Germany
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Rosario Moratalla,

    1. Cajal Institute, CSIC and CIBERNED, ISCIII, Madrid, Spain
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Maria Trinidad Herrero,

    1. Clinical & Experimental Neuroscience (NiCE-CIBERNED), School of Health Sciences, University Jaume I, Castelló, and School of Medicine, University of Murcia, Murcia, Spain
    Search for more papers by this author
  • Koushik Chakrabarty,

    1. Molecular Neurobiochemistry, Ruhr-University Bochum, Bochum, Germany
    Search for more papers by this author
  • René Drucker-Colín,

    1. Instituto de Fisiología Celular, Universidad Nacional Autónoma de México, Mexico City, México
    Search for more papers by this author
  • Jose Ruben Garcia-Montes,

    1. Cajal Institute, CSIC and CIBERNED, ISCIII, Madrid, Spain
    Search for more papers by this author
  • Nicola Simola,

    1. Department of Biomedical Sciences, Section of Neuropsychopharmacology, University of Cagliari, Cagliari, Italy
    Search for more papers by this author
  • Micaela Morelli

    Corresponding author
    1. Department of Biomedical Sciences, Section of Neuropsychopharmacology, University of Cagliari, Cagliari, Italy
    2. National Institute of Neuroscience (INN), University of Cagliari, Cagliari, Italy
    3. National Research Council (CNR), Neuroscience Institute, Cagliari, Italy
    4. Center of Excellence on Neurobiology of Dependence, University of Cagliari, Cagliari, Italy
    • Address correspondence and reprint requests to Micaela Morelli, Department of Biomedical Sciences, Section of Neuropsychopharmacology, University of Cagliari, Via Ospedale 72, 09124 Cagliari, Italy. E-mail: morelli@unica.it

    Search for more papers by this author


Dopamine replacement therapy in Parkinson's disease is associated with several unwanted effects, of which dyskinesia is the most disabling. The development of new therapeutic interventions to reduce the impact of dyskinesia in Parkinson's disease is therefore a priority need. This review summarizes the key molecular mechanisms that underlie dyskinesia. The role of dopamine receptors and their associated signaling mechanisms including dopamine-cAMP-regulated neuronal phosphoprotein, extracellular signal-regulated kinase, mammalian target of rapamycin, mitogen and stress-activated kinase-1 and Histone H3 are summarized, along with an evaluation of the role of cannabinoid and nicotinic acetylcholine receptors. The role of synaptic plasticity and animal behavioral results on dyskinesia are also evaluated. The most recent therapeutic advances to treat Parkinson's disease are discussed, with emphasis on the possibilities and limitations of non-pharmacological interventions such as physical activity, deep brain stimulation, transcranial magnetic field stimulation and cell replacement therapy. The review suggests new prospects for the management of Parkinson's disease-associated motor symptoms, especially the development of dyskinesia.


This review aims at summarizing the key molecular mechanisms underlying dyskinesia and the most recent therapeutic advances to treat Parkinson's disease with emphasis on non-pharmacological interventions such as physical activity, deep brain stimulation (DBS), transcranial magnetic field stimulation (TMS) and cell replacement therapy. These new interventions are discussed from both the experimental and clinical point of view, describing their current strength and limitations.

Abbreviations used



acetylated H3




abnormal involuntary movements


brain-derived neurotrophic factor


cAMP response element-binding protein


diacyl glicerol


dopamine-cAMP-regulated neuronal phosphoprotein


deep brain stimulation


epidermal growth factor


extremely low frequency magnetic fields


extracellular signal-regulated kinase 1 and 2


embryonic stem cells


fibroblast growth factor


glial cell line-derived neurotrophic factor


graft-induced dyskinesia


globus pallidus


histone H3


human embryonic stem cells


inositol 1,4,5-trisphosphate


induced pluripotent stem cells


Lewy bodies




L-DOPA-induced dyskinesia


long-term depression


long-term potentiation


primary motor cortex


mesodiencephalic dopamine


mitogen-activated protein kinase


motor evoked potentials




mitogen and stress-activated kinase-1


medium sized spiny neurons


mammalian target of rapamycin


nicotinic acethylcholine receptor


neural cell adhesion molecule


nerve growth factor


nuclear localization signal




platelet-derived neurotrophic factor


Parkinson's disease


phosphatidyl inositol


protein kinase A


phospholipase C


peroxisome proliferator-activated receptor




postsynaptic density 95 protein


protein transduction domain


Ras-guanine nucleotide-releasing factor




repetitive TMS


single/paired-pulse TMS


src family kinase signaling


supplementary motor area


substantia nigra


stage-specific embryonic antigen-1


subthalamic nucleus


tonically active neurons


theta-burst stimulation


transcranial magnetic field stimulation


thalamic ventrointermediate nucleus


ventral mesencephalic

Parkinson's disease (PD) is the second most common neurodegenerative disorder after Alzheimer′s disease (Thomas and Beal 2007). It is characterized by a slow and progressive degeneration of substantia nigra (SN) dopaminergic neurons projecting to the striatum. Although other neurons are affected, the dopaminergic neuronal loss is the primary cause for the characteristic motor symptoms of PD (Thomas and Beal 2007). Dopamine neuron loss in the SN correlates with the presence of neuronal cytosolic filamentous inclusions from protein aggregates called Lewy bodies (LB) that are positive for α-synuclein (Braak et al. 2004). LB pathology can appear around 10 years before motor symptoms (Spillantini et al. 1997), correlates with non-motor symptoms of PD and progresses in several neuropathological stages in accordance with the order of LB emergence (Braak et al. 2004).

Motor dysfunctions in PD are the main target of pharmacological treatment, with L-3,4-dihydroxyphenylalanine (L-DOPA) being the most effective drug (Jenner 2008). The primary complication of L-DOPA therapy is the development of dyskinesia and motor fluctuations; thus, the target of research is to prolong the antiparkinsonian effects of L-DOPA to improve the patients' quality of life. There exists a broad list of pharmacological agents which include: inhibitors of L-aromatic amino acid decarboxylase enzyme, monoamine oxidase inhibitors, catechol-O-methyl transferase inhibitors, dopaminergic agonists and anticholinergic drugs. Several dopaminergic agonists are available, but the current ones binding the dopamine receptors (R) -D2R, D3R, D4R- are the most effective. Monotherapy with agonists of dopamine R in the early stages of PD has shown to retard the side effects of L-DOPA. Other non-pharmacological strategies that are taken into account include cell transplant therapy and deep brain stimulation (DBS). With regard to non-invasive procedures, physical training and transcranial magnetic field stimulation (TMS) are also taken into consideration. To have a comprehensive view of all these strategies, this article will review the molecular substrates of dyskinesia, and discuss some of the recent therapeutic non-pharmacological strategies to manage this complication of L-DOPA therapy.

Long-term modifications induced by chronic dopamine replacement therapy

Effects induced by dopamine replacement therapy are widespread in the brain, however the striatum, the area most rich of dopamine R, is where the major neurochemical and functional changes that underlie dyskinesia take place. Moreover, striatal changes and imbalance between direct and indirect striatal efferent pathways containing D1 and D2R, respectively, are not only the most well-characterized molecular effects of long-term therapy with L-DOPA, but do also play a crucial role in the instatement and persistence of dyskinesia (Carta et al. 2008a). Elucidating the role of striatal dopamine R and their transduction mechanisms in these changes is therefore crucial, since altered function of the striatum has direct consequences on both other basal ganglia nuclei and motor cortex through the cortico-striatal loops, resulting in dyskinesia.

D1R signaling cascades in striatonigral direct pathway

D1R are coupled to at least three distinct signaling cascades. One of these, the Gαs/olf/adenylyl-cyclase/cAMP/protein kinase A (PKA) signaling pathway is activated preferentially following stimulation of striatal D1R which in turn activates Gαs/olf adenylyl-cyclase and generates cAMP. The cAMP regulates the PKA which may lead to immediate-early-gene expression. The co-stimulation of dopamine D1/D2R triggers the dissociation of D1R from Gαq which finds its primary effector in phospholipase C. The phospholipase C converts the phosphatidyl inositol (PIP2) to the second messengers diacyl glicerol and inositol 1,4,5-trisphosphate (IP3), which produces a Ca2+ influx in the endoplasmic reticulum (Hasbi et al. 2009). Recent studies have further clarified the D1R signaling by demonstrating that D1R potentiate the NMDAR signalling via phosphorylation of the NR2B subunit by src family kinase signaling. This activation promotes the Ca2+ influx into the cell, activating the extracellular signal-regulated kinase 1 and 2 (ERK1,2) and mitogen-activated protein kinase/ERK signaling. Describing how these signaling cascades are initiated and what leads to the malfunction of the dopaminergic system in the striatum is crucial to understand L-DOPA-induced dyskinesia (LID).

Development of dyskinesia requires intact dopamine D1R

Experiments with dopamine R KO mice demonstrated that the D1R are required for the development of dyskinesia. Inactivation of D1R completely blocked LID and L-DOPA-induced phosphorylation of ERK, FosB, and dynorphin expression in the 6-hydroxydopamine (6-OHDA)-lesioned mouse striatum. Inactivation of D2R does not block dyskinesia or its molecular determinants but it seems to potentiate them (Darmopil et al. 2009) thereby indicating that D2R are not necessary for dyskinesia but modulate it. These results are consistent with earlier pharmacological studies which revealed that D1R antagonists almost completely inhibited LID and blocked L-DOPA-induced ERK1/2, and mitogen and stress-activated kinase-1 (MSK-1) phosphorylation (Darmopil et al. 2009).

Moreover, it is possible that the D3R, which are abnormally expressed in the dorsal striatum after repeated L-DOPA treatment, contribute to dyskinesia (Bezard et al. 2003). However, D3R antagonists do not prevent the induction of dyskinetic movements by D1R agonists (Kumar et al. 2009; Mela et al. 2010). Stimulation of D1R induces abnormal recruitment of D1R in the plasma membrane and intracellular trafficking of D1R is disrupted in dyskinetic animals where D1R expression is increased in all intracellular compartments (Berthet et al. 2009). This is associated with an increase in the postsynaptic density 95 protein (PSD95), a synapse-associated protein. PSD95 is associated with glutamatergic input and can alter NMDA and AMPA trafficking (Silverdale et al. 2010). Interestingly, PSD95 down-regulation decreases LID in monkeys (Porras et al. 2012). This study also found that PSD95 and D1R co-immunoprecipitate, indicating the formation of D1R/PSD95 complexes in vivo.

Dopamine depletion hypersensitizes the D1R PKA pathway

In the normal striatum, D1R activate cAMP/PKA leading to phosphorylation of PKA substrates such as dopamine-cAMP-regulated neuronal phosphoprotein (DARPP-32) and cAMP response element-binding protein which in turn induces the expression of immediate-early-genes (Greengard et al. 1999). This signaling cascade is robustly potentiated in the dopamine-depleted striatum, a phenomenon known as hypersensitization to differentiate it from the sensitization that occurs after repetitive dopamine stimulation in a normal animal. Evidence suggests that hypersensitization underlies LID, and that the intensity of the dyskinetic symptoms correlates with FosB expression in mice and rats (Pavon et al. 2006). Administration of selective D1R agonists also induces dyskinesia and increases FosB and dynorphin expression in the lesioned striatum, indicating that dyskinesia may be triggered by D1R stimulation. Further support comes from the finding that chronic treatment with L-DOPA strongly induced FosB expression primarily in dynorphin-containing neurons which co-express dopamine D1R (Darmopil et al. 2009; Suarez et al. 2013). In addition, Carta et al. (2007) suggested that 5-HT terminals in the lesioned striatum are involved in LID. As shown by Tanaka et al. (1999) and Navailles et al. (2011), abnormal release of dopamine synthesized from L-DOPA in 5-HT neurons and D1R-containing hypersensitized neurons underlies L-DOPA-induced dyskinesia. However, other studies showed that dopamine depletion is not sufficient to sensitize D1R (Aubert et al. 2005). Therefore, both the profound dopamine striatal denervation and repetitive intermittent, but not continuous, L-DOPA administration appear necessary to develop dyskinesia (Guigoni et al. 2007; Berthet et al. 2009).

D1R and DARPP intracellular signaling after L-DOPA treatment

Considering that LID is associated with modifications in the D1R-containing neurons, studies indicate that L-DOPA increases PKA activation, and DARPP-32 phosphorylation only in the medium-sized spiny neurons (MSN) of the direct pathway (Santini et al. 2007). Under basal conditions the phosphorylation of DARPP-32 is low at the Thr34 and high at the Thr75, Ser97, and Ser130 residues (Santini et al. 2007) (Fig. 1). In dopamine-denervated animals, the hypersensitization of D1R after L-DOPA is reflected in robust phosphorylation of DARPP-32 at Thr34 and dephosphorylation at Thr75 and Ser97, leading to changes in downstream signaling cascades and transcriptional activation of many genes like Arc, c-Fos, ΔFosB, zif-268, and brain-derived neurotrophic factor in direct pathway neurons (Santini et al. 2007) (Fig. 1). In agreement, dyskinesia is attenuated in DARPP-32 global KO mice (Santini et al. 2007) and in mice lacking DARPP-32 selectively in the MSN of the direct pathway, but not in those of the indirect pathway (Bateup et al. 2010). These studies indicate the importance of enhanced cAMP/PKA/DARPP-32 signaling in LID, and point to the MSN of the direct pathway as a key neuronal substrate for LID.

Figure 1.

Scheme of pre and postsynaptic changes induced by L-3,4-dihydroxyphenylalanine (L-DOPA). Shadows in the postsynatic component emphasize the R involved in the activation of molecular cascade. Presynaptic components of dyskinesia include the conversion of L-DOPA to dopamine by serotonergic terminals, as well as the aberrant release of dopamine and glutamate. The nicotinic R (nAChR) and cannabinoid R (CB1) exemplify the fine tune of presynapses. Chronic L-DOPA increases D1R-activated signal transduction in medium-sized spiny neurons (MSN) of direct pathway. D1R are implicated in development of dyskinesia by hyperactivation of Gs/olf protein which increases protein kinase A (PKA) levels that phosphorylates dopamine-cAMP-regulated neuronal phosphoprotein (DARPP-32) at Thr-34, increasing the inhibition of Protein Phosphatase-1 that activates mitogen-activated protein kinase (MEK). Hyperactivation of D1R activates Gβγ increasing the src family kinase signaling (SFK) that phosphorylates the NR2B subunit of NMDAR. Calcium entry via NMDAR promotes Ras-guanine nucleotide-releasing factor (Ras-GRF) which increases MEK-extracellular signal-regulated kinase (ERK) signal activating mitogen and stress-activated kinase-1 (MSK1) in the nucleus. ERK and MSK1 induce cAMP response element-binding protein (CREB) and H3 phosphorylation that stimulates gene expression. Cannabinoid R, located in direct pathway neurons, decrease D1R-signal possibly through TRPV1R activation. Modified from Nishi et al. (2011).

Involvement of ERK in dyskinesia

Molecular changes related to LID have been detected in other signaling cascades associated with dopamine D1R activation, including the ERK1/2 cascade. D1R stimulation induces ERK1/2 phosphorylation in the dopamine-intact striatum and is strongly potentiated in the dopamine-depleted one (Gerfen et al. 2002). More recent work established that ERK1/2 phosphorylation correlates with increased ΔFosB and dyskinesia, and that the expression pattern overlaps that of FosB in completely denervated striatal areas (Pavon et al. 2006). This work was substantiated by data demonstrating that the intensity of the p-ERK1/2 signal correlated with the intensity of dyskinesia in mice (Pavon et al. 2006), and that lovastatin, an inhibitor of Ras/ERK, reduced ERK activity as well as the incidence and severity of LID (Schuster et al. 2008). Inhibitors of ERK also prevented phosphorylation of mammalian target of rapamycin, and the mammalian target of rapamycin inhibitor rapamycin reduced the development of LID (Santini et al. 2009). Interestingly, L-DOPA-induced ERK phosphorylation is more prominent after acute treatment and induces p-ERK preferentially in direct pathway MSN, whereas chronic administration leads to an increase of p-ERK in cholinergic interneurons (Fig. 1).

Chromatin modifications, acetylated histone 3 phosphorylation

Changes downstream of the ERK1/2 signaling cascade have been detected in the striatal neurons of dyskinetic animals. Thr581 of the MSK-1, a nuclear target of ERK1/2, was found to be hyperphosphorylated in MSN of the direct pathway in rodents with LID, as was histone H3 (H3), which is phosphorylated by p-[Thr581]MSK-1 (Darmopil et al. 2009) (Fig. 1). Induction of p-MSK-1, p-ERK, and P-AcH3 (activated acetylated H3) by L-DOPA was completely inhibited by D1R-like antagonists (Westin et al. 2007) and by genetic deletion of D1R (Darmopil et al. 2009). AcH3 is co-expressed with dynorphin and FosB/ΔFosB in the denervated striatal direct pathway, further supporting that AcH3 modification is part of the transcriptional changes induced by L-DOPA (Darmopil et al. 2009). While NMDAR also activate ERK1/2 cascades (Hayashi and Huganir 2004), NMDAR antagonists do not modify ERK phosphorylation induced by D1R agonists in the dopamine-depleted striatum (Gerfen et al. 2002; Rylander et al. 2010). In contrast, mGluR5 antagonists attenuate LID and decrease ERK phosphorylation (Mela et al. 2007; Rylander et al. 2010), suggesting that mGluR5 play a more important role than NMDAR in the control of ERK1/2 by chronic L-DOPA treatment (Fig. 1). Modifications described above are triggered by intermittent stimulation of dopamine R, as they are totally prevented by continuous L-DOPA administration (Lebel et al. 2010).

Cannabinoids and their role in dyskinesia

In addition to dopamine, other neurotransmitter systems modify dopamine receptor-signaling cascades involved in dykinesia. Growing evidence implicates the endocanabinoid system in dyskinesia, since CB1R are highly expressed in both D1R-and D2R-containing MSN, and negatively modulate D1R- and D2R-mediated behaviors (Martin et al. 2008). This is the main reason to postulate CB1R as a therapeutic target to control the imbalance of GABAergic or glutamatergic neurons in PD and dyskinesia. There is also evidence that endocannabinoids and cannabinoid agonists inhibit dopamine reuptake by blocking dopamine transporters (Chen et al. 2003; Price et al. 2007) and thus may have implications for the fine tuning of the striatal circuitries involved in dyskinesia (Gonzalez-Aparicio and Moratalla 2013).

Cannabinoid agonists have shown antidyskinetic effects in animal models of PD (Fox et al. 2002; Morgese et al. 2007) and antagonists may have antidyskinetic properties as well (Kelsey et al. 2009). However, only cannabinoid agonists decrease dyskinesia in clinical studies (Sieradzan et al. 2001; Carroll et al. 2004). These paradoxical effects may be explained in part by the modulation of CB1R and vanilloid TRPV1R by endogenous cannabinoids such as oleoylethanolamide (OEA), which binds the peroxisome proliferator-activated R (PPARα) and TRPV1R.

CB1R and TRPV1R are co-expressed in several brain areas and appear to play opposite roles in diverse actions, including LID (Morgese et al. 2007). Recent studies have shown that binding to CB1R and CB2R is not the only mechanism by which cannabinoids show anti-dyskinetic effects. TRPV1R have been strongly implicated in the abnormal movements induced by L-DOPA since CB1R stimulation, in combination with TRPV1R blockade, leads to anti-dyskinetic effects (Morgese et al. 2007). Interestingly, OEA also potentiates the effect of anandamide by acting on TRPV1R (Garcia Mdel et al. 2009); moreover it reduced LID at doses that do not modify motor behavior in the mouse model of dyskinesia. These results are in line with those showing that capsaicin, a specific TRPV1R agonist, abolished both the OEA anti-dyskinetic effects and the increase in molecular markers of dyskinesia (Gonzalez-Aparicio and Moratalla 2013). By contrast, pre-treatment with a PPARα antagonist did not attenuate either LID or the associated molecular changes (Gonzalez-Aparicio and Moratalla 2013). This indicates that OEA reduces LID by blocking TRPV1R activation. These findings suggest the emergence of the endocannabinoid system as a current therapeutic target to treat PD while avoiding dyskinesia (Fig. 1).


The imbalance between dopamine and acethylcholine plays an important role in the long-term striatal changes induced by the dopaminergic lesion. When the dopamine signal decreases, cholinergic neurons also known as tonically active neurons lose their regulation and increase acethylcholine release. Moreover, dopamine release is strongly modulated by nicotinic R (nAChR) located in the dopaminergic terminals in the striatum (Partridge et al. 2002) (Fig. 1). Nicotine also induces sensitization of D1R and increases c-Fos expression in parvalbumin-positive interneurons (fast spiking interneurons) and in somatostatin-positive interneurons (low threshold interneurons), as well as in substance P positive neurons (direct pathway MSN) (Garcia-Montes et al. 2012). Thus, targeting nAChR could represent an alternative co-therapy with L-DOPA to decrease dyskinesia.

Abnormal synaptic plasticity in PD and dyskinesia

As described above, dopamine plays a critical role in modulating the excitation of MSN. Dopamine R importantly influence the corticostriatal connections by long-term potentiation (LTP) and long-term depression (LTD), the two classic forms of long-term plasticity (Calabresi et al. 1996; Wickens et al. 2003). In line with this, inactivation of D2R blocks striatal LTD but not LTP in mice (Calabresi et al. 1997), whereas inhibition of D1R reduces locomotor activity and blocks LTP but not LTD, implicating the D1R subtype in LTP (Rivera et al. 2002; Centonze et al.2003).

In PD, the loss of dopaminergic innervation of the striatal neurons is associated with the loss of their dendritic spines, which leads to an increase in glutamatergic transmission (Day et al. 2006). These changes in spines morphology are further associated with impaired synaptic plasticity in corticostriatal LTP and LTD and with impaired motor learning and motor control, which are directly modulated by the dopamine R located on the dendritic spines of striatal MSN (Murer and Moratalla 2011). Although it is unclear whether spine loss in parkinsonian animals preferentially affects D2 MSN or equally affects D1 and D2 MSN (Villalba et al. 2009), recent evidence indicates that both types of neurons are equally affected (Suarez et al. 2013). This evidence comes from a set of experiments using two lines of BAC transgenic C57/BL6 mice, D1R-tomato-, and D2R-eGFP-BAC mice. Following unilateral lesion with 6-OHDA, Lucifer Yellow injections revealed that spine pruning occurred in both D1R-positive and negative neurons from D1R-tomato BAC mice. Similarly, spine pruning after the lesion equally affects D2R-positive and negative neurons from D2R-eGFP BAC mice (Suarez et al. 2013), indicating that spine pruning after 6-OHDA lesion may underlie the impaired LTP observed in PD.

L-DOPA treatment in experimental parkinsonism has been shown to restore LTP in both dyskinetic and non-dyskinetic animals. However, low frequency stimulation in dyskinetic animals shows impaired depotentiation indicating that LTD is lost (Picconi et al. 2003). The loss of bidirectionality in synaptic plasticity has been proposed to underlie dyskinesia through D1R regulation of the NMDAR trafficking and the potentiation of NMDA responses (Ghiglieri et al. 2012). Interestingly, L-DOPA restores spine density in dyskinetic animals, although this occurred selectively in D2R-containing striatal projection neurons (Suarez et al. 2013). The imbalance in the dendritic spine number between the two types of striatal projection neurons might constitute the anatomical substrate of dyskinesia and the loss of bidirectional striatal synaptic plasticity (Fig. 1).

Non-pharmacological alternatives for overcoming dyskinesia in PD

Neurochemical changes are long-lasting and difficult to contrast by pharmacological interventions, since all drugs that are effective on dyskinesia impair motor performance. Therefore, new strategies to contrast dyskinesia in PD tackle both pharmacological and non-pharmacological aspects. These strategies mainly aim at mitigating the abnormal functional changes caused by dopamine replacement in the striatum and other basal ganglia areas as well as in cortico-striatal loops, thereby restoring motor function.

The remainder of this review summarizes some of the best characterized non-pharmacological strategies for the management of dyskinesia as physical activity and DBS, as well as TMS and cell replacement therapy, and discusses their proposed mechanisms.

Link between motor performance, physical activity and dyskinesia

Recent studies have investigated whether or not the performance of movement affects the motor symptoms of PD, as well as the emergence and/or severity of therapy-induced motor complications. The current evidence indicates that physical training can attenuate some features of motor impairment and mitigate LID severity.

Studies in PD patients

Different types of physical activity (e.g. treadmill exercise, intensive resistance training) have been evaluated for their therapeutic potential in PD. Improvements in bradykinesia, gait, balance, muscular force, and even quality of life have been observed in patients who performed some form of physical activity, compared with sedentary patients (Hackney and Earhart 2009; Frazzitta et al. 2013; Rose et al. 2013). However, recent meta-analysis studies have suggested that the effects of physical activity in PD patients, although clinically significant, are often small and involve a limited number of motor symptoms (Lima et al. 2013). Nevertheless, physical activity seems to impact both motor impairment and motor complications, as an amelioration of LID has been reported in patients undergoing intensive physical training (Reuter et al. 1999, 2000). Importantly, physical activity seems not to affect the pharmacokinetics of L-DOPA (Goetz et al. 1993; Reuter et al. 2000). Therefore, it is feasible that exercise may influence the neuronal circuits involved in motor control, hence modulating the functional mechanisms underlying motor performance and drug-induced motor complications. Taken together, these data have led to the vision of incorporating movement performance as a new strategy for the management of PD. This may acquire particular interest since some aspects of motor disability associated with PD are poorly sensitive to dopamine replacement therapy (King and Horak 2009), and no satisfying pharmacotherapy for dyskinesia exists. Therefore, a thorough investigation of the effects of physical activity in PD patients is warranted, particularly in view of the exhaustive follow-up information, as the majority of studies performed have focused on the improvement of symptoms within a short time window.

Studies in experimental animals

Evidence from rodent models of PD suggests that physical activity may counteract parkinsonian-like motor deficits. In this context, interesting data have been obtained in unilaterally 6-OHDA-lesioned rats characterized by a hemiparkinsonism feature with an asymmetry in their forelimb use and a neglect of the impaired limb contralateral to the 6-OHDA infusion, indicative of akinesia (Simola et al. 2007). Thus, 6-OHDA-lesioned rats subjected to a 7-day period of casting of the intact forelimb, which resulted in the forced use of the impaired forelimb, displayed reduced asymmetry of forelimb use when evaluated up to 60 days from 6-OHDA lesion (Tillerson et al. 2001). Moreover, the same procedure abolished the manifestation of apomorphine-induced contralateral rotational behavior (Tillerson et al. 2001), a parameter indicative of motor asymmetry (Deumens et al. 2002). These initial findings clearly demonstrated that forced movement of an akinetic limb may improve its functionality and mitigate the motor deficits. In agreement with this, are findings reported by other investigations employing the same animal model and attaining movement performance by either forcing the rats to place their impaired forelimb in response to vibrissae stimulation, a sensory-motor task, or having them exercising on a treadmill (Anstrom et al. 2007; Dutra et al. 2012). Amelioration of parkinsonian-like symptoms may also be elicited by the performance of voluntary exercise, as indicated by studies showing that 6-OHDA-lesioned rats allowed access to a running wheel exhibited scarce contralateral rotational behavior in response to apomorphine, compared with sedentary hemiparkinsonian rats (Mabandla et al. 2004; O'Dell et al. 2007). Similar results were obtained in mice treated with 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine, another widely used rodent model of PD. Both forced and voluntary exercise have been reported to ameliorate some of the parkinsonian-like motor deficits displayed by these mice, such as altered gait pattern, impaired balance and reduced spontaneous locomotion (Pothakos et al. 2009; Archer and Fredriksson 2010).

Studies in parkinsonian rodents indicate that movement performance may not only impact motor deficits, but also influences the behavioral parameters indicative of drug-induced motor complications. In this regard, interesting data have been obtained in 6-OHDA-lesioned rats subjected to the priming paradigm of repetitive dopaminergic stimulation (Simola et al. 2009; Frau et al. 2013). Priming involves an induction phase with a D1/D2R agonist followed days apart by an expression phase with a highly dyskinetic D1R agonist (Morelli et al. 1989). Upon priming expression, rats display a sensitized contralateral rotational behavior, indicative of an abnormal motor response to repetitive administration of dopaminergic drugs (Pinna et al. 2006). Using this paradigm, it was observed that rats' immobilization during the induction phase completely abolished rotational behavior on the expression phase (Simola et al. 2009). This suggests that the performance of movement triggered by dopaminergic drugs may itself promote the later emergence of motor complications. Another recent study in parkinsonian rodents has shown that behavioral parameters indicative of L-DOPA-induced motor complications are sensitive to exercise performance (Aguiar et al. 2013). This study evaluated the influence of voluntary running on abnormal involuntary movements (AIMs) of limbs and trunk, a reliable rodent model of human dyskinesia (Cenci et al. 1998) induced by chronic L-DOPA in unilaterally 6-OHDA-lesioned mice. It was observed that L-DOPA elicited AIMs of a lower severity in those mice that performed voluntary exercise in running wheels, compared with sedentary hemiparkinsonian mice (Aguiar et al. 2013). These data added to those observed in 6-OHDA-lesioned rats subjected to dopaminergic priming, where the performance of movement appeared to facilitate behaviors indicative of motor complications. Together, these findings suggest that drug-stimulated movement and voluntary exercise given their different nature might promote distinct neuroplastic adaptations in the basal ganglia, which would lead to divergent outcomes as to their role in motor complications. Therefore, it can be hypothesized that either performance of correct voluntary movement or impediment of performing incorrect drug-stimulated movement would compete with the immobility or abnormal movements. This would attenuate the generation of procedural mnemonic traces in the striatal circuits, which might be the basis of motor impairments and dyskinesia.

Possible mechanisms mediating the effects of movement performance on parkinsonian motor symptoms and drug-induced motor complications

Neuroprotection and neuroplasticity are envisioned as the major players in mediating the effects of movement performance on parkinsonian motor symptoms and drug-induced motor complications. Studies in parkinsonian rodents have provided evidence suggesting that physical activity can attenuate the degeneration of the dopaminergic nigrostriatal system. Thus, forced use of the impaired forelimb in 6-OHDA-lesioned rats has been found to be associated with sparing of striatal dopamine and metabolites, tyrosine hydroxylase immunoreactivity, and increase in the expression of glial cell line-derived neurotrophic factor (GDNF) (Tillerson et al. 2001; Cohen et al. 2003). Similarly, reduced dopaminergic degeneration has been reported in both 6-OHDA-lesioned rats and 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine-treated mice subjected to treadmill exercise (Zigmond et al. 2009; Smith et al. 2011; Sung et al. 2012). In contrast, studies in parkinsonian rodents also exist which suggest that movement performance is beneficial for motor symptoms but not neurodegeneration (O'Dell et al. 2007; Pothakos et al. 2009). These discrepancies may be explained on one hand by methodological issues and time window between toxin administration and exercise, with early exercise being most effective in providing neuroprotection (Tillerson et al. 2001). On the other hand, neuroprotection could be only one of the mechanisms underlying the effects of physical activity on motor impairment, and neuroplasticity might also be involved. Here, it is conceivable that neuroplastic adaptations could promote persistent changes in the functionality of basal ganglia circuits and eventually influence the proper execution of movement at a later time. This view appears particularly relevant to the effects of movement performance on therapy-associated motor complications. In fact, these effects usually emerge at the late stages of PD, when neurodegeneration is extensive; therefore, their modulation appears more susceptible to neuroplasticity rather than neuroprotection.

It has been hypothesized that repetitive intermittent administration of dopaminergic drugs may promote pathologic motor learning eventually leading to abnormal motor responses such as dyskinesia (Packard and Knowlton 2002; Picconi et al. 2003). In this regard, drug-stimulated abnormal movement seems detrimental as it appears to be a key player in the later emergence of motor complications (Simola et al. 2009; Frau et al. 2013), probably by favoring pathologic motor learning in the dopamine-denervated basal ganglia (Picconi et al. 2003). In addition, it can be hypothesized that movement performance according to specific therapeutic programs could compete with abnormal purposeless movements caused by dopamine replacement therapy, and thereby attenuate the generation of pathological procedural mnemonic traces in the striatum (Frau et al. 2013).

Recent experiments performed by Frau et al. (2013) have studied markers of long-term modifications in the priming model of dopaminergic stimulation, which can be useful in investigating the first molecular events that lead to dyskinetic movements in PD. In this study, the immediate-early-gene zif-268, which is rapidly and transiently induced by pharmacological manipulations, was used as an index of neuronal responsiveness (Harlan and Garcia 1998). The results obtained showed that abnormal drug-stimulated movement performance (rotational behavior) induces a selective activation of zif-268 mRNA in the dynorphinergic striatonigral efferent neurons, but not in the striatopallidal enkephalinergic pathway. This finding demonstrated that abnormal movement performance in response to dopaminergic drug administration under conditions of dopamine denervation promotes neurochemical changes in selected striatal efferent neurons. It is noteworthy that the different dyskinetic potential of dopamine agonists is associated with distinct long-term changes in striatonigral/striatopallidal zif-268 expression (Carta et al. 2010), thereby supporting the hypothesis that highly dyskinetic drugs increase zif-268 mRNA expression in the striatonigral pathway, whereas striatopallidal neurons remain unaffected. In contrast, drugs with a low dyskinetic potential, such as ropinirole, do not alter or decrease zif-268 mRNA either in striatonigral neurons or in the striatopallidal pathway (Carta et al. 2008b). These results further validated that long-term striatal changes in the striatonigral pathway might represent a molecular correlate of the pathological movements (AIMs) induced by highly dyskinetic drugs that are utilized in PD.

Deep brain stimulation

DBS is an effective surgical treatment for movement disorders refractory to medication (Williams and Okun 2013). DBS procedure consists of placing a microlectrode in a brain structure stereotactically and implanting a stimulator in the patient's chest wall where the stimulation parameters are adjusted by telemetry. The electrodes can be placed in the subthalamic nucleus (STN) in the globus pallidus or where the neurologist has programmed (Terzic and Abosch 2012). DBS is a non-destructive, adaptable, reversible, and safe procedure in bilateral surgery of the basal ganglia. However, DBS is expensive, laborious, and relies on strict patient selection criteria.

DBS was originally utilized to treat chronic pain or behavioral disorders by a DBS system implanted in the thalamus (Delgado et al. 1952). The first evidence of using depth stimulation to treat movement disorders was described by Bekhtereva et al. (1963) who performed a chronic implantation of electrodes to treat hyperkinetic disorders. However, the work received little attention as the communication was in Russian. With the introduction of L-DOPA in the 1960s, the surgery for PD dramatically declined. The renaissance of functional stereotactic neurosurgery occurred in the mid 1980s when Laitinen introduced the Leksell's posteroventral pallidotomy for PD with an excellent prospect to limit dyskinesia and other symptoms (Hariz and Hariz 2013). Then the idea of DBS re-emerged in France with the work of Benabid et al. 1987 on the thalamic ventrointermediate nucleus (Vim). This group demonstrated that DBS was safer than thalamotomy and pallidotomy. The clinical effect of DBS was incidentally found when these surgeons tried to localize the target with a stimulator. They observed that stimulation higher than 100 Hz produced a reversible effect on tremor (Benabid et al. 1991). At that time, it was shown in non-human primates that some basal ganglia nuclei were hyperactive, and DeLong and co-workers showed that lesioning or ‘stimulating’ the STN could relieve parkinsonian symptoms (Bergman et al. 1990). After the initial experiments in monkeys, the effects of the same procedure showed excellent results in PD patients affected by the ON–OFF motor fluctuations and LID where STN or globus pallidus (internal segment) were bilaterally implanted.

Mechanisms mediating DBS efficacy

DBS is a successful therapeutic approach in PD; however, the therapeutic mechanism of DBS remains under debate (Miocinovic et al. 2013). The main goal of DBS in PD is to provide therapeutic and clinical benefits by changing the electrical activity of the targeted brain structure in a controlled manner, in fact, the alleviation of PD symptoms by DBS originates in the functional decoupling of neurons in the stimulated area (Modolo and Beuter 2009).

Apparently DBS suppresses excessive pallidal outflow and desynchronizes parallel neuronal channels across frequency ranges. However, the most accepted hypothesis is that DBS suppresses pathological oscillatory synchronization at low frequencies (13–35 Hz) leading to symptomatic benefit. As abnormally synchronized activity patterns in the β band are the hallmark of untreated PD patients, if high frequency STN DBS suppresses this β synchrony, this could explain the motor improvement (Eusebio et al. 2012). However, other unknown consequences of STN-DBS could also contribute to the symptomatic motor and non-motor improvements. In fact, high-frequency DBS is considered more effective than low-frequency DBS due to its ability to replace pathological low-frequency network oscillations with a regularized pattern of neuronal firing (McConnell et al. 2012). Besides the mechanistic studies, the technological advancement is providing newer ways to control the brain neurosignals and to use local field potentials (Priori et al. 2013).

Issues in the utilization of DBS in PD

DBS is often used when patients have already developed several complications of the medical treatment. So far, surgical intervention for advanced PD has been an option to control symptoms when medical therapy fails. However, neurologists and neurosurgeons are now considering to proceed with DBS earlier, because DBS has a long-term efficacy in reducing both motor and non-motor symptoms of PD, increases the patients' quality of life and decreases LID by 60% (Benabid et al. 2009). PD patients who received DBS of the STN earlier than usual had better results than patients who received optimum medical treatment (Schupbach et al. 2007). Evaluation of the time spent in ON state without dyskinesia, showed that DBS of the STN plus best medical therapy improved quality of life significantly more than optimum therapy alone (Williams et al. 2010). The predictors of positive outcomes after DBS are the young age, the responsiveness to L-DOPA, and the persistence of motor benefits after L-DOPA treatment. Moreover, the patients should show a perfect cognitive function. However, STN-DBS is cognitively safe in patients with normal age-related cognitive testing.

The main advantages of DBS are its reversibility, programmability, ability to be safely performed bilaterally, and its adjustability at each moment. However, the procedure involves cerebral surgery and has some associated risks (transient adverse events at a rate of 15–25%). The shortcomings of DBS therapy are (i) complexity of the treatment; (ii) lack of evidence for neuroprotection; (iii) post-operative complications, and (iv) long-term hardware-related complications. DBS-associated complications are (i) surgery-related complications such as delirium, hypomania, hemiparesis, intracerebral hemorrhages, and infections; (ii) hardware-related problems such as electrode migration, lead fracture, and device failure; (iii) problems related to the stimulation itself, which could be weak and non-responding. Other DBS-associated risks are dysarthria, eye-lid apraxia, weight gain, and disorders such as mood, depression, (hypo)mania, anxiety, and thought disorder. Symptoms such as personality changes, hypersexuality, apathy, anxiety, and aggressiveness are less common. Moreover, studies have reported that DBS procedure could cause mood changes leading to transient suicidal behavior (Voon et al. 2008; Weintraub et al. 2013). Several studies have stated that even if there is a selective decrease in frontal cognitive functions, DBS of the STN does not reduce overall cognition or affectivity. DBS of the STN is safe with respect to neuropsychological and psychiatric effects in carefully selected patients during a 6-month follow-up period (Halpern et al. 2009). The potential of DBS is increasing, and the results from clinical studies on DBS show great potential, making it one of most promising techniques to address challenging neurological problems (Miocinovic et al. 2013).

Transcranial magnetic field stimulation

TMS is a safe and non-invasive method that affects the cerebral cortex but not the deep structures. TMS can be used to investigate causality in the brain-behavior relationship by utilizing neuronal depolarization or hyperpolarization (Barker et al. 1985).

Two primary protocols are established for TMS. In the first one, the single/paired-pulse TMS (s/p-pTMS) depolarizes and discharges the action potential under the region that is being stimulated. When applied to the motor cortex, s/p-pTMS can provoke motor evoked potentials, an output commonly used together with TMS (Kim et al. 2006). The second one, involves the repetitive TMS (rTMS) which has been shown to produce more durable effect after the initial stimulation (Stefan et al. 2000). rTMS can increase or decrease the excitability of the corticospinal tract depending on the intensity of stimulation, coil orientation and frequency, and properties of the stimulation coil. The inhibitory/excitatory effects of TMS on the neurons should be distinguished from the negative/positive outcome on behavior, which can occur in any combination. For instance, an excitatory effect of TMS in one area may (through inhibitory interconnections) induce inhibition of a different area that controls the execution of cognitive tasks, resulting in a negative behavioral outcome (Sack and Linden 2003).

TMS techniques can be used to diagnose movement disorders and evidence indicates their utility as a therapeutic tool in PD. TMS studies have documented several functional changes and agree that there is increased inhibition in patients with PD suggesting that cortical excitability is reduced (Lefaucheur 2005; Ueki et al. 2006). The level of cortical excitability in PD patients seems indirectly related to the nigrostriatal functioning and the level of disturbed intracortical excitability correlates with the total daily dose of L-DOPA. Several TMS techniques have been developed to investigate primary motor cortex (M1) inhibitory circuits. As demonstrated, there is a strong link between the plasticity of M1 and motor learning, as the latter is impaired in PD patients (Muellbacher et al. 2001). Other studies in PD patients, either at rest or during a voluntary muscular contraction, suggest that in PD hypoexcitability and hyperexcitability of M1 inhibitory circuits can paradoxically co-exist: (i) when patients are tested at rest, short-interval intra-cortical inhibition is reduced, suggesting increased excitability in cortical motor areas; (ii) when patients are tested during a voluntary muscle activation, the cortical silent period (which reflects inhibitory mechanisms) is significantly shorter than in healthy subjects and the long-interval intra-cortical inhibition (which inhibits motor evoked potentials) is significantly enhanced (Berardelli and Suppa 2011).

TMS has been used to study the effects of chronic L-DOPA and the development of LID. In PD, LID is in fact associated with aberrant plasticity of the M1 circuit and LTP-like plasticity is deficient in PD off medications, and is restored by L-DOPA in non-dyskinetic patients only. The continuous normalization of intra-cortical excitability in PD patients during a 12-month L-DOPA therapy, demonstrates the modulation of the neuronal inhibitory/facilitatory circuits of the motor cortex by dopaminergic stimulation (Strafella et al. 2000; Bagnato et al. 2006). These results would suggest that abnormal synaptic plasticity in the motor cortex might participate in the development of LID (Morgante et al. 2006). Some studies claim that chronic exposure to dopaminergic drugs does not substantially affect motor cortex excitability and plasticity and there is little interaction between plasticity and excitability features of motor cortex in PD (Gonzalez-Garcia et al. 2011). Other studies suggest that dyskinesia is associated with an abnormal effect of L-DOPA on cortical motor inhibitory circuits (Barbin et al. 2013). A recent study of two cohorts of PD patients, one optimally treated and the other never exposed to dopaminergic drugs, showed that the facilitatory responses to paired associative stimulation were similarly reduced in both cohorts compared with healthy subjects. These results show that in advanced PD cortical associative plasticity seems to be impaired, regardless of a previous chronic exposure to L-DOPA.

Previous studies have used rTMS to reduce motor impairment in PD. rTMS at frequencies of 5 Hz and higher can enhance motor cortex excitability (Pascual-Leone et al. 1994), whereas lower frequencies rTMS (1 Hz and lower) can transiently depress cortical excitability (Chen et al. 1997). Studies performed so far using rTMS (Elahi and Chen 2009; Gonzalez-Garcia et al. 2011) showed mixed results, which can be attributed to the variability of patients' profile, small sample size, large heterogeneity of cortical targets, and stimulation (Shirota et al. 2013). Theta-burst stimulation is a novel form of rTMS which uses a lower stimulation intensity and a shorter stimulation time compared with conventional rTMS protocols. Theta-burst stimulation applied over the M1 has been successfully used to induce changes in cortical excitability and this stimulation paradigm is able to produce strong and long-lasting effects (Morgante et al. 2006).

Mechanism mediating TMS efficacy

TMS uses electromagnetic induction by means of a rapidly changing magnetic field to induce weak electric currents that provoke activity in specific or diffuse parts of the brain, allowing the brain functioning and interconnections to be studied. rTMS could be a therapeutic tool in movement disorders, in particular by creating long-lasting changes in synaptic excitability within the motor system as a way to modulate symptoms. The motor improvement after rTMS could be attributed to its effects on the motor cortex or the supplementary motor area (Koch et al. 2005; Brusa et al. 2006), or to its ability to induce dopamine release from the basal ganglia (Strafella et al. 2003). The effects of rTMS and its functional outcome in PD patients are still unclear with no consensus on the symptoms most likely to respond. Therefore it can be perceived that the study of TMS as a treatment in PD and in LID needs a large multicenter trial taking into account the interindividual variability observed in PD patients, and considering the profile of cortical plasticity and its modulation by dopamine (Koch 2013).

Cell replacement therapy

No therapies are available that correct the neurodegenerative process underlying PD. To address these limitations, innovative approaches such as cell-based therapies had been conceptualized in the early 1970s, with a rather basic hypothesis of replacing the diseased dopamine neurons by transplanting cells that could release dopamine in the striatum and thereby restoring the characteristic nigrostriatal dopamine deficit, which is the hallmark of the PD pathophysiology (Madrazo et al. 1987).


The beginning 1970–1985: it was the seminal work of Olson and Bjorklund (Bjorklund et al. 1975; Olson et al. 1972) which inspired and instilled hope in treating PD with cell replacement therapy. Due to ethical reasons and controversies over the use of fetal mesencephalic tissue, adrenal chromaffin cells were used in early animal experiments which were shown to grow neuron-like neurites after transplantation into the anterior eye chamber of rats. Subsequently, it was demonstrated the efficacy of adrenal medullary tissue as intraventricular grafts in unilaterally 6-OHDA lesioned rats. Although an up to 50% reduction in apomorphine-induced rotations after a couple of months after transplantation was achieved, the study indicated the inability of the adrenal chromaffin cells to re-innervate the striatum. Based on these observations the adrenal chromaffin cells were seen a viable alternative to fetal ventral mesencephalic (VM) (Nunes et al. 2003) tissue since at that period, use of fetal derived VM tissue was considered ethically challenging. Moreover, as pointed out by a recent review (Bjorklund and Kordower 2013), the adrenal chromaffin cells circumvented the ethical issues as these cells could be obtained directly from the patients providing the additional advantage of avoiding any immune reactions upon the use of autologous grafts. Although the limited clinical trials involving adrenal medullary transplants in humans failed to deliver the expected results, the idea of obtaining patient-derived dopamine neurons persisted till the present time as a viable alternative to face PD (Drucker-Colin and Verdugo-Diaz 2004).

The middle period (1985–2000): with subsequent mitigation of ethical issues on the use of fetal VM as transplants during the mid-1980s, a large number of clinical trials were performed where patients received intrastriatal implants of fetal VM tissue rich in post-mitotic dopamine neurons (Redmond 2002). The primary objectives in most these studies were to determine the survival of the grafted fetal tissue upon implantation in the patients' striatum, and if the dopamine neurons from the grafts could integrate and innervate other areas. Most of the early transplantation effort using fetal VM grafts was constituted of a number of small open-labeled clinical trials, and in few of these the patients could be withdrawn from L-DOPA treatment as they exhibited major recovery for several years. Although some of these trials were successful and visible clinical improvement was seen in a few of the patients, a high variability in the degree of symptomatic relief and overall efficacy of the transplanted fetal VM grafts was the general finding. The overall data forthcoming from these studies indicated the lack of consensus within and between the trials. Much reason has been attributed to variables intrinsic by nature of the transplant material, such as age and quality of the fetal VM tissue, and quantity of the tissue used per transplantation. As the VM tissue is sourced from aborted fetuses, quality control is also one of the contending issues. Other determining factors are the number and selection of patients and controls in the study, the nature of severity of the disease along with the age of the patients receiving transplants, and the study of endpoints or objectives are some of the critical parameters. In addition, apart from practical ethical limitations in using fetal dopamine neurons, potential immunological complications would require the use of immunosuppressive therapies, which could add to further complications. Taken together, the studies as such provide enough proof that cell replacement therapy in PD is a possibility.

In the clinical context, it has been observed that transplantation of the fetal VM grafts gives best results in patients with lesser severity of the disease (Olanow 1996). The other crucial point is the area of implant. Striatum and putamen are the two main areas where fetal VM grafts are implanted, and considerable variability in the clinical outcomes has been observed. The severity of the disease in patients is to be taken under consideration to address the issue. In fact, in patients with advanced PD and severe de-innervation, the intraputaminal graft implants are not able to innervate the striatum, and result in reduced efficacy (Lindvall and Bjorklund 2011). Several new developments in the field of stem cell biology have reignited the hopes for a cell-based procedure which addresses and redeems some of the key previously outlined limitations of fetal VM grafts.

In the past 15 years: cells with features or characteristics of midbrain dopamine neurons have been produced from stem cells of diverse sources and origins (Allan et al. 2010). So far most of the promising results in animal transplantation studies have been obtained using embryonic stem cells (ESc) from murine or primate origin (Sandberg et al. 2013). In principle, the midbrain dopamine neurons to a large extent are homogeneous, in that they express several of the mesodiencephalic dopamine (mdDA) neuronal markers. Similarly, human ESc (hESc) being homogeneous in character can be utilized as an ideal platform to obtain sufficient quantities of mdDA neurons (Cooper et al. 2012). One of the potential drawbacks of mdDA neurons derived from ESc, apart from the ethical concerns with hESc research, is the risk of teratoma formation (Cunningham et al. 2012). The upcoming field of cellular reprogramming has opened possibilities to address some of the key issues pertaining to the clinical application of ESc-derived mdDA neurons. Initial insights into cellular reprogramming was obtained with pioneering work of Takahashi and Yamanaka (2006), where it was shown that somatic cells (fibroblasts) could be reprogrammed using defined transcription factors (Oct3/4, Sox2, Klf4 and c-Myc using retroviral vectors) to pluripotent stem cells that very much resemble ESc in their characteristic features to possess the plasticity to be de-differentiated to cells belonging to diverse tissue types. By following a similar approach Wernig et al. (2008) obtained neural precursor cells from induced pluripotent stem cells (iPSC) which were in turn generated from mouse tail fibroblasts. This initial study demonstrated the ability of the neural precursors to differentiate to appropriate cell types upon transplantation. Electrophysiological recordings and morphological analysis demonstrated that the grafted neurons had mature neuronal activity and were functionally integrated in the host brain. Furthermore, the authors also reported that iPSC-derived neuronal precursors brought forth robust improvement in behavior in the 6-OHDA rat model. Interestingly, the study also reported teratoma formation after transplantation which was later rectified by depleting the cell suspension of SSEA-1 positive cell fraction prior to transplantation (stage-specific embryonic antigen-1 is a marker for murine pluripotent stem cells). To address the issue of immunosuppressive therapy often followed in conjunction with cell therapy, a recent study (Hargus et al. 2010), obtained transplantable mdDA neurons de-differentiated from iPSC from PD patient fibroblasts. This study provided evidence that PD patients' iPSC-derived mdDA neurons survive, integrate and mediate robust functional effects in an animal model of PD. Focusing on the immunosuppressive therapy issue, a recent report (Morizane et al. 2013), shows that monkey iPSC-derived dopamine neurons do not elicit an immune-response when autologously transplanted. These studies highlight the strength of the cellular reprogramming technology to generate mdDA neurons. It was then demonstrated that ectopic expression of defined transcription factors in mouse tail tip fibroblasts can induce early mdDA neuron progenitors, identified by the selective expression of the transcription factor Pitx3 which is a well-known marker for mdDA progenitors (Smidt et al. 2012). In a similar pursuit, it was demonstrated the direct generation of functional dopaminergic neurons from mouse and human fibroblast surpassing the intermediary stem cell stages (Caiazzo et al. 2011).

Robust and fast-paced advances are being made to devise and standardize protocols to maximize the generation of appropriate nigral dopaminergic neurons (Theka et al. 2013). A direct reprogramming from one somatic cell population to another would in essence reduce the need for the additional and multiple steps of cell purification procedures and also reducing the possibilities of developing teratomas from mdDA neurons generated from such direct reprogramming. The reprogramming technology offers great advantage to avoid the immunosuppressive therapy since the mdDA neurons utilized for the transplant are generated from the patient's own cells (Cunningham et al. 2012). However, one of the major drawbacks of generating iPSC is the use of viruses encoding the reprogramming factors which represents a major limitation of the current technology, since even low vector expression may alter the differentiation potential of the iPSC or induce malignant transformation (Griscelli et al. 2012; Steinemann et al. 2013). Aiming to circumvent insertional mutagenesis, Soldner et al. (2009) showed that fibroblasts from patients with idiopathic PD can be efficiently reprogrammed first to form iPSC using Cre-recombinase excisable viruses and subsequently differentiated into dopaminergic neurons. The authors reported that the factor free human iPSC maintain a pluripotent state and more closely relate to hESc than to hiPSC carrying the transgenes, thereby indicating that residual transgene expression in virus-carrying hiPSC can affect their molecular characteristics, and that factor-free hiPSC represent a more suitable source of cells therapy. Current research has involved the use of non-integration adenovirus and viral free vectors towards reprogramming procedures (Hu and Slukvin 2013; Sidhu 2011).

The future: iPSC generated by these approaches are not completely free of the risk of the transgene insertion as there is always the possibility of a vector DNA remaining. One procedure that could derive iPSC free of insertional mutation is by means of delivering the necessary transcription factors directly in the form of functional proteins. Studies in this area have demonstrated the need for specific amino acid sequence derived from the HIV-1-TAT protein, referred to as the TAT protein transduction domain (PTD), which has the intrinsic ability to penetrate the cell membrane and thereby deliver the fused heterogeneous protein inside the cells. High cellular uptake with an efficiency of 90% has been reported with the PTD-based protein transduction system in a variety of cell types. The subcellular localization of the transduced proteins has been suggested to depend on their biophysical and biochemical characteristics, the cell type to be transduced, and the delivery approach. If the proteins contain a nuclear localization signal, they would exhibit nuclear localization without further modifications (Zhang et al. 2012). In 2009, Zhou et al. reported the successful reprogramming of mouse somatic cells to pluripotency using recombinant proteins of Sox2, Oct4, Klf4, and c-Myc. Soon after, in the same year it was reported the successful induction of pluripotency in human fibroblast cells using 293T cell extracts containing the same transcription factors with a C-terminal PTD (Kim et al. 2009; Zhou et al. 2009). However, in both studies, the efficiency of generating iPSC by protein transduction was extremely low (about 0.001%) and the process was very slow.

The future of cell replacement therapy to treat PD, mandates developing a robust multidimensional perspective involving cellular reprogramming and neuroprotective strategies to address issues such as the loss or death of the transplanted mdDA neurons. These issues could be immediate upon transplantation due to mdDA poor survival in the degenerative environment depleted of trophic support or later as the disease progresses and globally affects the brain. Moreover, in the long term the disease process could eventually reach the grafted neurons especially if they are transplanted ectopically into the striatum far from the degenerated SN site. Alternatively, fibroblast growth factor-4-secreting schwannoma cells or GDNF-secreting schwann cells were transplanted and GDNF was applied to enhance the survival and neurite outgrowth of the transplanted cells. Although the subsequent fiber growth that was achieved using these strategies is sparse, partial restoration of function was observed (Gaillard and Jaber 2011).

However, a major risk of ectopic transplantation is thought to be the development of graft-induced dyskinesia (Barker and Kuan 2010; Lane et al. 2010). In order to re-establish proper brain circuitry and afferent innervation of the dopaminergic neurons, cells may have to be transplanted into the SN. The critical factor of recipient age-related innervation was further substantiated (Bentlage et al. 1999) which showed that transplanted dopamine neurons in the SN have the ability to extend their axons and reconnect with the dopamine-depleted striatum when transplanted at postnatal day 3 and 10, but not at postnatal day 20. This study points towards the requirement of developing suitable strategies methods to stabilize the transplanted neurons and to promote their directed outgrowth into the striatum.

A possibility to circumvent the limitations in the appropriate application of extracellular growth factors to elicit specific cellular responses is to alternatively activate the downstream intracellular signaling cascades. Extracellular factors and mitogens, such as epidermal growth factor, brain-derived neurotrophic factor, platelet-derived neurotrophic factor, nerve growth factor, and fibroblast growth factor are often used to promote survival of the transplant and neurite outgrowth. Research so far performed has demonstrated various beneficial effects of neurotrophins on neurons (Hartmann et al. 2001; Heumann et al. 1987), the actions of which are partially mediated by the intracellular membrane anchored Ras-signaling protein (Borasio et al. 1989; Heumann et al. 1981). Furthermore, several studies have neatly shown that neuronal activation of Ras protects from toxic insults, increases brain synapse formation, promotes neuronal regeneration (Arendt et al. 2004; Heumann et al. 2000; Makwana et al. 2009), and leads to a significant increase in the expression of Nurr1, a key activator of dopaminergic genes (Saucedo-Cardenas et al. 1998). Specifically, neuronal activated Ras induces an increase in Nurr1 via an activating phosphorylation of cAMP response element-binding protein and increased phosphorylation of the pro-apoptotic Bad protein through the PI3K/Akt pathway (Fig. 2) culminating in the enhancement of the dopaminergic properties and neuroprotection of dopamine neurons (Chakrabarty et al. 2007).

Figure 2.

The model depicts Ras activation by the ligand bound R tyrosine kinase eliciting the activation of downstream cascades by phosphorylating the mitogen-activated protein kinases (MAPK) and PI3k pathways. In dopaminergic neurons, the MAPK pathway phosphorylates cAMP response element-binding (CREB) protein, which upon entry into the nucleus leads to activation of the Nurr1 gene culminating with the activation of the tyrosine hydroxylase gene. In parallel, the PI3k pathway regulates cell survival by controlling various mediators of the apoptotic machinery.

An important aspect of transplanting dopamine neurons in the SN region would require the axons to correctly navigate to their specific targets through a complex molecular environment. Axon guidance represents an important step in the formation of neuronal networks which are carried out by various inhibitory or activating guidance cues, such as nogo, semaphorins, slits, ephrins, and netrins (Pernet and Schwab 2012; Prestoz et al. 2012). Taken together, further understanding of the molecular mechanisms involved in the process of survival of the transplanted dopamine neurons and their ability to integrate, navigate, and form a functional neural circuitry would help us to determine the ideal transplantable solutions.


Dyskinesia is a very complex phenomenon which involves all basal ganglia nuclei and stems from an imbalance in the functionality of the two types of striatal efferent neurons, together with the loss of bidirectional synaptic plasticity. Pharmacological therapies for contrasting dyskinesia are very limited and not effective; therefore, non-pharmacological strategies are the only options for the treatment of this highly disturbing side effect. The clinical results from DBS show a great potential, making it one of the most promising techniques to address dyskinesia. At the same time, physical activity has attracted great interest for the positive results so far obtained. Another promising strategy is TMS, which is a safe and non-invasive method that affects the cerebral cortex, but not deep structures. Finally, an important aspect of future therapies includes cell replacement in the SN region which, however, requires further information on the complex molecular environment capable to guide axons to properly move to their specific targets. Further studies would be important to define the ideal combination of non-pharmacological therapies to efficiently contrast dyskinesia and improve the quality of life of PD patients.

Acknowledgements and conflict of interest disclosure

RM acknowledges grants from the Spanish Ministries de Economía y Competitividad and of Sanidad Política Social e Igualdad, ISCIII: BFU2010-20664, PNSD, CIBERNED ref. CB06/05/0055 and Comunidad de Madrid ref. S2011/BMD-2336, JRGM is supported by ICyTDF México; MTH acknowledges the support by CIBERNED CB05/05/505, SAF2007-062262 and FIS PI10-02827. RH and KC were supported by the German Bundesministerium für Bildung und Forschung, Grant 01GN1006B. NS gratefully acknowledges Sardinia Regional Government for financial support (P.O.R. Sardegna F.S.E. Operational Programme of the Autonomous Region of Sardinia, European Social Fund 2007–2013 - Axis IV Human Resources, Objective l.3, Line of Activity l.3.1 “Avviso di chiamata per il finanziamento di Assegni di Ricerca”). The authors have no conflicts of interest to declare.

All experiments were conducted in compliance with the ARRIVE guidelines.