MINIREVIEW
Neuronal growth-inhibitory factor (metallothionein-3):
evaluation of the biological function of growth-inhibitory
factor in the injured and neurodegenerative brain
Claire Howells, Adrian K. West and Roger S. Chung
Menzies Research Institute, University of Tasmania, Hobart, Australia
Introduction
Metallothioneins (MTs) are a family of unusual cyste-
ine-rich (30%), 6–7 kDa proteins synthesized predomi-
nantly by astrocytes within the brain. The MT3 isoform
was first isolated and identified as a neuronal growth-
inhibitory factor (GIF) in 1991, a brain-specific protein
whose synthesis was notably deficient in the Alzheimer’s
disease (AD) brain. It was found to possess a strong
ability to impair neurite outgrowth and neuronal
survival of cultured neurons, leading to its designation
as GIF. It was later discovered that GIF shares approx-
imately 70% amino-acid sequence similarity with the
MT family of proteins, leading to its renaming as MT3.
Most striking is the conservation within GIF of the
unique cysteine motifs found in mammalian MTs.
Given that GIF shares a biochemical structure similar
to those of the other MT isoforms, it is not surprising
that GIF has the characteristic metal-binding and
reactive oxygen species (ROS) scavenging capabilities
present in all MT isoforms. However, GIF has also
been found to exhibit several unique biological proper-
ties, suggesting that this MT isoform has different and
distinct functions within the brain. Furthermore, the
discovery and continued investigation of this brain-spe-
cific MT isoform has led to intense interest in the roles

during neurodegenerative disease progression.
Abbreviations
AD, Alzheimer’s disease; Ab, b-amyloid; CNS, central nervous system; GIF, neuronal growth-inhibitory factor; KO, knockout;
MT, metallothionein; NO, nitric oxide; ROS, reactive oxygen species.
FEBS Journal 277 (2010) 2931–2939 ª 2010 The Authors Journal compilation ª 2010 FEBS 2931
extracellular senile amyloid plaques. These physical
alterations are often accompanied by aberrant neurite
sprouting and significant neuronal loss, particularly
within the cerebral cortex and hippocampus. One
explanation for this impaired neuronal survival is that
there may be decreased synthesis of a neurotrophic
factor that promotes the survival and growth of neu-
rons within the hippocampus and cerebral cortex in
the AD brain. However, several studies observed that
in fact the AD brain has elevated neurotrophic activity
compared with the normal aged brain [1,2]. These
studies suggested that a soluble extract from AD brain
significantly enhanced neuronal survival and neurite
outgrowth of cultured cortical neurons in comparison
with aged brain extract.
It was proposed that this increased neurotrophic
activity could be a cause of the abnormal neurofibril-
lary changes that occur within the AD brain. Accord-
ingly, if neurite sprouting was left unchecked, then this
process could lead to neuronal exhaustion and eventu-
ally cell death, partly explaining the vast neuronal loss
in the AD brain. Further investigation of the increased
neurotrophic activity of the AD brain revealed that it
correlated with the loss of a specific neuroinhibitory
factor, rather than the presence of a neurotrophic fac-

GIF protein is primarily synthesized in astrocytes,
where it is found predominantly in the soma and fine
processes of these cells. Astrocytic synthesis of GIF
is mainly found in the cortex, brainstem and spinal
cord [6].
Additionally however, there are also consistent
reports of neuronal synthesis of GIF, albeit at seem-
ingly much lower levels than within astrocytes. Neuro-
nal synthesis of GIF is highly localized to specific
subsets of cortical and hippocampal neurons, and is
found within the axons and dendrites. Neuronal pro-
duction of GIF is particularly associated with granule
cells of the dentate gyrus in the hippocampus, in those
neurons that store and release zinc at synapses [7].
While there is substantial evidence for expression of
GIF mRNA in neurons from in situ hybridization
studies, there is a discrepancy between the co-localiza-
tion of GIF mRNA and protein, as very few studies
are able to demonstrate neuronal localization of GIF
at the protein level through immunohistochemistry or
similar techniques. Such confirmation at the protein
level has been hindered by the availability of appropri-
ate GIF antibodies, and further studies are needed to
resolve whether neurons synthesize basal levels of GIF
protein.
Neurodegenerative conditions under which GIF
levels are altered
The basal level of production of GIF in the CNS is
considered to be quite low, particularly in comparison
with the MT1 and MT2 isoforms that are more highly

progressive supranuclear palsy and amyotrophic lateral
sclerosis, and around senile plaques in Down syn-
drome [17]. Interestingly, the level of GIF appears to
correlate with neuronal loss, as GIF immunoreactivity
disappears in areas with vast neuronal loss, but
remains in less affected areas.
Given its potent neurite growth-inhibitory proper-
ties, it was predicted that the GIF levels might corre-
late with the failure of the injured brain to regenerate.
Indeed, a number of studies have confirmed that levels
of GIF protein are up-regulated in reactive astrocytes
surrounding cerebral infarcts, stab wounds or excito-
toxic brain injuries. For instance, GIF mRNA was
substantially up-regulated in reactive astrocytes sur-
rounding degenerated neurons following ventricular
injection of kainic acid [18]. Also, GIF was elevated in
reactive astrocytes surrounding a stab wound to the
brain at 3–4 days post-injury and remained elevated
for almost a month [19,20]. The increase in GIF
expression in response to these different types of brain
injury was often observed in glial cells accumulating at
the degenerating area, where the glial scar was begin-
ning to form. Intriguingly, exogenously applied GIF
has been shown to induce astrocyte proliferation and
migration in vitro [14]. It is possible, then, that the up-
regulation and secretion of GIF by astrocytes in the
immediate vicinity of a lesion could contribute to the
accumulation of reactive astrocytes at the injury site.
Interestingly, one study has demonstrated that in the
absence of GIF [Gif gene knockout (KO) mice] there

However, in 2002 it was reported that GIF is secreted
by cultured astrocytes [24]. The amount of GIF
secreted by the cultured astrocytes was approximately
174 ngÆmg
)1
of protein, measured using ELISA [24].
This study also determined that the amount of extra-
cellular GIF was around 30% higher than the levels of
intracellular GIF in these astrocyte cultures. Further-
more, the absence of cell death suggested that GIF is
actively secreted by cells, although the precise mecha-
nism by which GIF is secreted remains unknown.
Intriguingly, secretion of the MT1 and MT2 isoforms
by cultured astrocytes has also recently been reported.
The secretion of MT1 and MT2 by astrocytes appears
to be regulated because the basal levels of protein
secretion were low and could be stimulated through
cytokine-activation of cultured astrocytes [25]. Whether
GIF is secreted by a similar mechanism is currently
not known. Understanding the specific situations
which stimulate GIF secretion by astrocytes will prob-
ably reveal important insights into the functional roles
of this protein in the brain, and in particular its ability
to inhibit neurite outgrowth (as described above).
The functional role of GIF synthesis and secretion
in the brain
Taken together, the regulation of GIF synthesis by
reactive astrocytes, and its subsequent secretion under
certain conditions, could be intimately involved in
regulating how the brain responds to and recovers

minireviews in this series [26,27]. In this minireview we
will briefly describe the role of GIF in (a) regulation of
metal ion homeostasis, (b) free radical scavenging and
(c) neurite growth inhibition.
Metal-binding properties of GIF and role in
synaptic activity
MTs are generally considered to have an important
role in maintaining the homeostasis of key metal ions
within the body. Like other MT isoforms, GIF binds
both monovalent and divalent metal ions with high
affinity. They have the ability to inhibit heavy metal
toxicity [i.e. Cd(II), Hg(II)] and regulate the transport
and storage of essential heavy metals [Cu(I) and
Zn(II)]. However, GIF does not seem to be essential
for metal-handling within the body, because studies
have reported that GIF-deficient mice were not more
prone to Zn or Cd toxicity compared with control ani-
mals [26]. However, as GIF has been localized both
intracellularly and extracellularly, it may have an
important role in metal-related extracellular neuro-
chemistry. It is important to note that the GIF is able
to bind Cu(I) and Zn(II) concurrently. For example,
GIF isolated from the human brain has been found to
contain both Zn(II) and Cu(I), in a Cu
4
Zn
3
-GIF struc-
ture [2]. The ability of GIF to bind redox-active metal
ions, such as Cu(I), may suggest a crucial mechanism

regeneration during the initial responses to
the injury. (C) At later stages, interference in
neuroglial interactions may decrease GIF
production and secretion by surrounding
astrocytes. The decreased levels of extra-
cellular GIF would allow ROS-mediated
attack on astrocytes and neurons, and may
also promote excessive neurite sprouting.
Function of GIF in injured and degenerative brain C. Howells et al.
2934 FEBS Journal 277 (2010) 2931–2939 ª 2010 The Authors Journal compilation ª 2010 FEBS
to seizure-induced injury to CA3 hippocampal neu-
rons, resulting in the GIF KO mice experiencing more
convulsions and greater mortality than their wild-type
littermates [28]. GIF-overexpressing mice experience
much less kainic acid-induced brain damage compared
with controls [26], implying that GIF may play an
important protective role in response to kainic acid-
induced seizures.
One explanation of how GIF can protect against
epileptic seizures is through its regulation of Zn(II) lev-
els. Synaptically released zinc is thought to play a key
role in neuronal signalling, and in response to kainic
acid insult, neurons release Zn(II), as well as gluta-
mate, from the synaptic cleft. Interestingly, neurons
with a high GIF mRNA content are those known to
store zinc in their axon terminals [7]. In addition, the
synthesis of GIF within cultured cells led to a signifi-
cant increase in intracellular Zn(II) stores within those
cells [7]. GIF, through its zinc-binding properties, may
act to recover synaptically released Zn(II) and enable

efficiently bind reactive ROS [29].
MT can function as a potent scavenger of hydrogen
peroxide, hydroxyl, nitric oxide (NO) and superoxide
radicals. Studies have shown that extracellular and
intracellular GIF can protect cells from oxidative
stress. There have been a number of studies investi-
gating the specific free-radical scavenging ability of
GIF. One such study reported that GIF was able to
directly scavenge hydroxyl radicals generated in a
Fenton-type reaction or by photolysis of hydrogen
peroxide. In the same study, GIF was unable to scav-
enge superoxide (generated in the hypoxanthine oxi-
dase reaction) or NO (generated from NOC-7) [24].
The neuroprotection provided by extracellular GIF
against free-radical-mediated attack was subsequently
explored in vitro. To induce a condition indicative of
oxidative stress, neurons were grown in a hyperoxic
environment. Hyperoxia stimulates neuronal cells to
differentiate or undergo cell death. Exogenous GIF
prevented the induction of neuronal differentiation,
quantified by the level of neurite outgrowth in young
cortical neuronal cultures. The protein also prevented
neuronal cell death in older cultures [24]. GIF not
only protected neurons from oxidative stress, but was
also shown to directly quench ROS within the cul-
tured neurons. Using an indicator for intracellular
ROS, GIF was shown to reduce ROS production in
neurons.
Intracellular GIF has been shown to protect against
free-radical-mediated attack. The knocking down of

differences in the protein sequence of this isoform
compared with all other mammalian MTs. The human
GIF amino acid sequence is 70% similar to human
MT2, retaining all the 20 conserved cysteine residues
[2]. The only structural differences between GIF and
MT1 and MT2 include a Cys-Pro-Cys-Pro(6–9) motif
and Thr5 in the b-domain of GIF and a different
amino acid structure in the a-domain [32]. Structural
and cell biology studies have revealed that the
Cys-Pro-Cys-Pro(6–9) motif seems to be critical for the
biological activity of GIF. Mutations in this motif,
specifically changing the two Pro residues to Ala and
Ser (found in human MT2) abolished the neuroinhibi-
tory activity of GIF [33]. Evidence from studies using
the inactive mutant of Zn
7
GIF also liberated similar
results, showing that despite the contrasting biological
activities of GIF compared with MT1 and MT2, the
metal-binding affinities remain comparable. However,
the precise spatial structure and the mechanism behind
the unique biological activity of GIF remain to be
elucidated.
Subsequent studies have investigated whether GIF
acts in a neuroinhibitory manner when administered
directly into the brain following traumatic brain injury.
In one such study, the injection of GIF into the site of
a stab wound promoted tissue repair, and the area of
the stab wound treated with GIF was significantly
smaller than those of control rats [34]. Interestingly,

dysfunction and death is the presence of increased neu-
rotrophic activity in the AD brain. It was on this
assumption that GIF was first discovered as a neuro-
nal growth-inhibitory factor, and a lack of GIF might
contribute to the generation of neurofibrillary changes
within the AD brain, including inappropriate neurite
sprouting. Abnormal neurite sprouting might then lead
to neuronal exhaustion and eventually cell death.
In vitro studies have clearly demonstrated the ability of
GIF to block neurite outgrowth, supporting the initial
proposal of Uchida et al. [2] that a lack of GIF might
be involved in AD pathogenesis. However, postmor-
tem human studies have provided mixed results on the
level of GIF present in the AD brain. The generation
of GIF KO or GIF overexpressing mice with tau-
based experimental mouse models of AD might pro-
vide considerable insight into the potential role of GIF
in neurofibrillary changes associated with AD.
Potential role of metal-binding/exchange
properties of GIF in AD
One of the primary pathological hallmarks of AD is
the formation of b-amyloid (Ab) plaques, composed
primarily of Ab(1–40) and Ab(1–42) peptides, which
associate to form abnormal extracellular deposits of
fibrils and amorphous aggregates [36]. Ab is a metallo-
protein that possesses high- and low-affinity Cu(I)-
and Zn(II)-binding sites. Furthermore, metal ion
homeostasis within the AD brain is dramatically
unbalanced and is proposed to be involved in the
pathology of Ab. For example, within the amyloid pla-

Cu(II)-induced Ab aggregation is related to the genera-
tion of ROS through the redox cycling of Cu. In this
regard, GIF has been demonstrated to not only prevent
Ab aggregation, but also block the generation of ROS
by Ab-Cu(II) [27,38]. Therefore, the ability of GIF to
redox-silence metal ions and directly scavenge ROS (dis-
cussed above) provide two distinct mechanisms through
which GIF can act to protect against Ab pathogenesis
and neurotoxicity in AD (Fig. 2).
Conclusion
The discovery of a brain-specific MT isoform, GIF, in
1991, sparked considerable interest in understanding
the role of all MTs in the brain, and particularly
within the injured or diseased brain. In the case of
GIF, the protein has many neuroprotective properties
that could provide vital protection during stress-induced
conditions. Studies have already focused on the ability
of GIF to protect against Ab pathology and neurotoxic-
ity in AD, with promising results. By contrast, GIF also
demonstrates a unique neuroinhibitory property, which
may promote cellular survival and repair following
injury, but could also be detrimental in neurodegenera-
tive diseases. The studies to date have only just started
to investigate and understand the physiological roles of
GIF within the brain. Further studies are required to
determine the function of GIF within the normal,
injured and neurodegenerative brain, and to establish
the therapeutic potential of the protein.
Acknowledgements
This work was supported by an Australian Research

β
-Cu(II) A
β
-Cu(II)A
β
-Cu(I)A
β
-Cu(II)
O
2
ROS O
2
ROS
+
ii
)
Zn-GIF
Cu-GIF
)
AB
Fig. 2. A diagrammatic model describing the possible protective properties of GIF against Ab pathogenesis and neurotoxicity in Alzheimer’s
disease. (A) In the presence of a reducing agent, the copper coordinated to the Ab can undergo redox cycling to produce harmful ROS that
damage the neurons. (B) MT is likely to protect against Ab-Cu(II) by a number of mechanisms: Zn
7
GIF is proposed to undergo a transmetalla-
tion event with the Ab-Cu(II), which removes the redox-active Cu(II) from the Ab, coordinating it to MT as redox-stable Cu(I), while simulta-
neously Ab coordinates redox-inert Zn. This event would prevent the generation of ROS. ZnGIF or CuGIF could also directly scavenge ROS.
C. Howells et al. Function of GIF in injured and degenerative brain
FEBS Journal 277 (2010) 2931–2939 ª 2010 The Authors Journal compilation ª 2010 FEBS 2937
3 Palmiter RD, Findley SD, Whitmore TE & Durnam

Hayakawa T (1997) Expression of growth inhibitory
factor mRNA after focal ischemia in rat brain. J Cereb
Blood Flow Metab 17, 745–752.
11 Zheng H, Berman NE & Klaassen CD (1995) Chemical
modulation of metallothionein I and III mRNA in
mouse brain. Neurochem Int 27, 43–58.
12 Tsuji S, Kobayashi H, Uchida Y, Ihara Y & Miyatake
T (1992) Molecular cloning of human growth inhibitory
factor cDNA and its down-regulation in Alzheimer’s
disease. EMBO J 11, 4843–4850.
13 Yu WH, Lukiw WJ, Bergeron C, Niznik HB & Fraser
PE (2001) Metallothionein III is reduced in Alzheimer’s
disease. Brain Res 894, 37–45.
14 Carrasco J, Giralt M, Molinero A, Penkowa M, Moos
T & Hidalgo J (1999) Metallothionein (MT)-III:
generation of polyclonal antibodies, comparison with
MT-I+II in the freeze lesioned rat brain and in a
bioassay with astrocytes, and analysis of Alzheimer’s
disease brains. J Neurotrauma 16, 1115–1129.
15 Erickson JC, Sewell AK, Jensen LT, Winge DR &
Palmiter RD (1994) Enhanced neurotrophic activity in
Alzheimers-disease cortex is not associated with
down-regulation of metallothionein-III (GIF). Brain
Res 649, 297–304.
16 Amoureux MC, Van Gool D, Herrero MT, Dom R,
Colpaert FC & Pauwels PJ (1997) Regulation of
metallothionein-III (GIF) mRNA in the brain of
patients with Alzheimer disease is not impaired. Mol
Chem Neuropathol 32, 101–121.
17 Sogawa CA, Asanuma M, Sogawa N, Miyazaki I,

by astrocytes in response to neuronal injury. J Neuro-
chem 88, 454–461.
24 Uchida Y, Gomi F, Masumizu T & Miura Y (2002)
Growth inhibitory factor prevents neurite extension and
the death of cortical neurons caused by high oxygen
exposure through hydroxyl radical scavenging. J Biol
Chem 277, 32353–32359.
25 Chung RS, Penkowa M, Dittmann J, King CE, Bartlett
C, Asmussen JW, Hidalgo J, Carrasco J, Leung YKJ,
Walker AK et al. (2008) Redefining the role of metallo-
thionein within the injured brain – extracellular metallo-
thioneins play an important role in the astrocyte-neuron
response to injury. J Biol Chem 283, 15349–15358.
26 Faller P (2010) Neuronal growth inhibitory factor
(metallothionein-3): reactivity and structure of metal-
thiolate clusters. FEBS J 277, 2921–2930.
27 Ding ZC, Ni FY & Huang ZX (2010) Neuronal growth
inhibitory factor (metallothionein-3): structure-function
relationships. FEBS J 277, 2912–2920.
28 Erickson JC, Hollopeter G, Thomas SA, Froelick GJ &
Palmiter RD (1997) Disruption of the metallothionein-
Function of GIF in injured and degenerative brain C. Howells et al.
2938 FEBS Journal 277 (2010) 2931–2939 ª 2010 The Authors Journal compilation ª 2010 FEBS
III gene in mice: analysis of brain zinc, behavior, and
neuron vulnerability to metals, aging, and seizures.
J Neurosci 17, 1271–1281.
29 Meloni G, Faller P & Vasak M (2007) Redox silencing
of copper in metal-linked neurodegenerative disorders
reaction of Zn(7)metallothionein-3 with Cu
2

36 Mattson MP (2004) Pathways towards and away from
Alzheimer’s disease. Nature 430, 631–639.
37 Lovell MA, Robertson JD, Teesdale WJ, Campbell JL
& Markesbery WR (1998) Copper, iron and zinc in Alz-
heimer’s disease senile plaques. J Neurol Sci 158, 47–52.
38 Meloni G, Sonois V, Delaine T, Guilloreau L, Gillet A,
Teissie J, Faller P & Vasak M (2008) Metal swap
between Zn-7-metallothionein-3 and amyloid-beta-Cu
protects against amyloid-beta toxicity. Nat Chem Biol 4,
366–372.
39 Holmes C, Boche D, Wilkinson D, Yadegarfur G,
Hopkins V, Bayer A, Jones RW, Bullock R, Love S,
Neal JW et al. (2008) Long-term effects of Ab 42
immunisation in Alzheimer’s disease: follow-up of a
randomised, placebo controlled phase I trial. Lancet
372, 216–223.
C. Howells et al. Function of GIF in injured and degenerative brain
FEBS Journal 277 (2010) 2931–2939 ª 2010 The Authors Journal compilation ª 2010 FEBS 2939