Chronic, controlled GDNF infusion promotes
structural and functional recovery in advanced
parkinsonian monkeys
Richard Grondin,1,2 Zhiming Zhang,1,2 Ai Yi,1 Wayne A. Cass,1 Navin Maswood,1
Anders H. Andersen,1,2 Dennis D. Elsberry,3 Michael C. Klein,4 Greg A. Gerhardt1,2 and
Don M. Gash1,2
1Department of Anatomy and Neurobiology and 2Morris K.
Udall Parkinson's Disease Research Center of Excellence,
University of Kentucky Medical Center, Lexington, KY,
3Medtronic Inc., Medtronic Neurological Division,
Minneapolis, MN and 4AMGEN Inc., Thousand Oaks, CA,
USA

Correspondence to: Dr Richard Grondin, Department of
Anatomy and Neurobiology, University of Kentucky
Medical Center, Room 305, Davis Mills Building, 800
Rose Street, Lexington, KY 40536-0098, USA
E-mail: XXXXXXXXXX
Summary
The powerful trophic effects that glial cell line-derived
neurotrophic factor (GDNF) exerts on midbrain dopamine
neurones suggest its use in treating Parkinson's
disease. However, some important questions remain
about the possible therapeutic applications of GDNF.
Here we demonstrate that the chronic infusion of 5 or
15 mg/day GDNF into the lateral ventricle or the striatum,
using programmable pumps, promotes restoration
of the nigrostriatal dopaminergic system and signi®-
cantly improves motor functions in rhesus monkeys
with neural de®cits modelling the terminal stages of
Parkinson's disease. The functional improvements were
associated with pronounced upregulation and regeneration
of nigral dopamine neurones and their processes
innervating the striatum. When compared with vehicle
recipients, these functional improvements were associated
with (i) >30% bilateral increase in nigral dopamine
neurone cell size; (ii) >20% bilateral increase in
the number of nigral cells expressing the dopamine
marker tyrosine hydroxylase; (iii) >70 and >50% bilateral
increase in dopamine metabolite levels in the striatum
and the pallidum, respectively; (iv) 233 and 155%
increase in dopamine levels in the periventricular striatal
region and the globus pallidus, respectively, on the
lesioned side; and (v) a ®ve-fold increase in tyrosine
hydroxylase-positive ®bre density in the periventricular
striatal region on the lesioned side. In addition, chronic
GDNF treatment did not induce the side-effects generally
associated with chronic administration of levodopa,
the most widely used treatment for Parkinson's disease.
Thus, the results suggest that the prolonged and controlled
delivery of GDNF into the brain could be used
to intervene in long-term neurodegenerative disease
processes like Parkinson's disease. Additional studies
are required to determine the potential differences
between chronic, intraventricular and intraputamenal
(or intranigral) delivery of GDNF to maximize the
ef®cacy of infusion treatments.
Keywords: GDNF; Parkinson's disease; dopamine neurones; regeneration; non-human primates
Abbreviations: DOPAC = 3,4-dihydroxyphenylacetic acid; GDNF = glial cell line-derived neurotrophic factor; HVA =
homovanillic acid; MPTP = 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine; TH = tyrosine hydroxylase
Introduction
Converging evidence from a number of laboratories suggests
that glial cell line-derived neurotrophic factor (GDNF) is
capable of halting or reversing the progressive degeneration
of the nigrostriatal dopamine system in animal models of
Parkinson's disease (Tomac et al., 1995a; Gash et al., 1996;
Choi-Lundberg et al., 1997; Bjorklund et al., 2000; Kordower
et al., 2000). The most likely mechanism is through the
trophic actions of GDNF on midbrain dopamine neurones in
the substantia nigra. These neurones are the principal target of
the pathophysiological processes underlying Parkinson's
disease, and the consequent neuronal injury and subsequent
degeneration lead to the profound depletion of basal ganglia
ãGuarantors of Brain 2002
Brain (2002), 125, 2191±2201
dopamine levels that characterizes the disease
(Lang and
Lozano, 1998a, b).

While there is good general agreement in results between
laboratories using a variety of animal models of Parkinson's
disease, some important questions remain about the possible
therapeutic applications of GDNF. For instance, GDNF
appears to be both neuroprotective and neurorestorative for
dopamine neurones (Gash et al., 1998; Kordower et al., 2000;
Costa et al., 2001); the protective effects are manifested
within hours, whereas the regenerative actions are not evident
for days to weeks. In many published animal studies, it is
dif®cult to distinguish between the results that arise from
protective (injury prevention) and restorative (recovery after
an injury) effects because GDNF treatment is initiated in the
hours or days following a lesion while the injury sequelae are
still unfolding. However, the distinction is important in
assessing treatment strategies for Parkinson's disease using
GDNF. If the primary effects were protective, then GDNF
treatment would be most bene®cial in the early stages of
Parkinson's disease, before devastating losses of dopamine
neurones have occurred. On the other hand, if the primary
actions of GDNF are on restoration, then treatment at all
stages of Parkinson's disease could be bene®cial.
Another important issue is the titre of biologically available
GDNF necessary to produce bene®cial effects.
Information in this area is especially limited for methods
involving the extended release of GDNF into the nigrostriatal
pathway in animal Parkinson's disease models using viral
vectors or encapsulated cells genetically engineered to
produce GDNF (Lindner et al., 1995; Choi-Lundberg et al.,
1997; Bensadoun et al., 2000; Kordower et al., 2000). While
these procedures show promise, techniques for determining
and controlling dosing and the timing of delivery are in the
developmental stage (Bjorklund and Lindvall, 2000; Olson,
2000). Thus, while signi®cant bene®cial effects can be
quanti®ed on host dopamine neurones and neuronal processes
after viral vector GDNF transfection or implantation of
GDNF-producing cells, the levels of biologically available
GDNF producing these effects are unclear.
The present study had the following aims. (i) To assess
the restorative actions of GDNF under conditions where
neuroprotection would have only a minor role, the late
stages of human Parkinson's disease were modelled in
rhesus monkeys with stable, advanced hemiparkinsonian
features induced by the neurotoxin 1-methyl-4-phenyl-
1,2,3,6-tetrahydropyridine (MPTP) (Bankiewicz et al.,
1983; Smith et al., 1993; Emborg-Knott and Domino,
1998). In this model, MPTP infusion through the right
carotid artery results in ~75% loss of dopamine neurones
expressing the phenotypic marker tyrosine hydroxylase
(TH) in the right substantia nigra and >99% depletion of
dopamine in the right putamen (Gash et al., 1996). These
reductions are comparable with advanced human parkinsonism,
in which cell counts typically show 60±70% loss
of nigral dopamine neurones (Jellinger, 1986) and 99%
dopamine depletion in the putamen (Kish et al., 1988).
(ii) To determine the titre of biologically available GDNF