MANF family of neurotrophic factors in Parkinson’s disease

FEBS Letters 589 (2015) 3739–3748

journal homepage: www.FEBSLetters.org

Review

Therapeutic potential of the endoplasmic reticulum located and secreted CDNF/MANF family of neurotrophic factors in Parkinson’s disease Merja H. Voutilainen a, Urmas Arumäe a,b, Mikko Airavaara a, Mart Saarma a,⇑ a b

Institute of Biotechnology, University of Helsinki, Finland Department of Gene Technology, Tallinn University of Technology, Tallinn, Estonia

a r t i c l e

i n f o

Article history: Received 7 August 2015 Revised 23 September 2015 Accepted 30 September 2015 Available online 9 October 2015 Edited by Wilhelm Just Keywords: Neurotrophic factors Cerebral Dopamine neurotrophic factor Mesencephalic astrocyte-derived neurotrophic factor Parkinson’s disease Regeneration Dopamine

a b s t r a c t Parkinson’s disease (PD) is a progressive neurodegenerative disorder where dopamine (DA) neurons in the substantia nigra degenerate and die. Since no cure for PD exists, there is a need for diseasemodifying drugs. Glial cell line-derived neurotrophic factor (GDNF) and related neurturin (NRTN) can protect and repair DA neurons in neurotoxin animal models of PD. However, GDNF was unable to rescue DA neurons in an a-synuclein model of PD, and both factors have shown modest effects in phase two clinical trials. Neurotrophic factors (NTFs), cerebral DA NTF (CDNF) and mesencephalic astrocyte-derived NTF (MANF) form a novel family of evolutionarily conserved, endoplasmic reticulum (ER) located and secreted NTFs. CDNF and MANF have a unique structure and an unparalleled dual mode of action that differs from other known NTFs. Both protect cells from ER stress, and regulate the unfolded protein response via interacting with chaperons, and CDNF dissolves intracellular a-synuclein aggregates. By binding to putative plasma membrane receptors, they promote the survival of DA neurons similarly to conventional NTFs. In animal models of PD, CDNF protects and repairs DA neurons, regulates ER stress, and improves motor function more efficiently than other NTFs. Ó 2015 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

1. Introduction Parkinson’s disease (PD) is an age-related, progressive neurodegenerative disorder characterized by the cardinal motor symptoms of tremor, muscle rigidity, bradykinesia, and postural instability. These motor symptoms of PD result from the degeneration and death of neurons in the substantia nigra (SN) in midbrain, which leads to a loss of dopamine (DA) in the striatum and disrupts the neural circuitry that controls movement [1]. The incidence of neurodegenerative diseases, such as PD, increases with age, causing increasing costs for society. There is no treatment for PD which could prevent the cell loss or halt the progression of the disease. The ideal candidate therapy would be the one which prevents neurodegeneration and restores DAergic circuitry in the brain, thereby halting the progression of debilitating disease symptoms. During development neurotrophic factors (NTFs) regulate neuronal survival, differentiation and maturation, as well as neurite growth and branching [2]. In adult animals they control metabolic maintenance of the neurons but also protect and repair injured neurons [2]. Because of the therapeutic potential of NTFs for ⇑ Corresponding author. Fax: +358 2941 59366. E-mail address: [email protected] (M. Saarma).

neurological disorders, research in this area has been growing rapidly and a large number of NTFs and their receptors have been identified [3–5]. NTFs have been explored as a novel treatment for PD, where all current treatments are symptomatic and no disease-modifying therapy exists [2]. It should be noted that many NTFs can halt the PD symptoms in in vivo models of PD, but only a few restore DAergic circuitry. The concept to restore the nigrostriatal circuitry is largely based on the findings that degeneration occurs in a dying-back manner, starting from nerve endings in the caudate putamen and is followed by axon and cell body degeneration in the SN [6]. Indeed, at the onset of first motor symptoms the deficits are larger in the putamen than in the midbrain [7,8]. Moreover, it was indicated that at the onset of first motor PD symptoms only 30% of SN pars compacta (SNpc) DA neurons are lost [9]. Thus, at the onset of symptoms more than 50% of DA neurons are viable and neurorestorative properties (i.e. can protect and regenerate DA neurons when applied after the lesion) are an essential opportunity for potential PD therapy. This review is focused on the two families of NTFs that have shown to be the most promising in their neurorestorative effects in animal models of PD. These factors are glial cell line-derived neurotrophic factor (GDNF) family ligands GDNF and neurturin (NRTN) as well as cerebral dopamine neurotrophic factor (CDNF)

http://dx.doi.org/10.1016/j.febslet.2015.09.031 0014-5793/Ó 2015 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

3740

M.H. Voutilainen et al. / FEBS Letters 589 (2015) 3739–3748

and mesencephalic astrocyte-derived neurotrophic factor (MANF) from CDNF/MANF family. Brain-derived neurotrophic factor (BDNF) was the first protein identified that directly support the survival of DA neurons in vivo. However, only GDNF, NRTN, MANF and CDNF have well-established neurorestorative properties in the nigrostriatal DAergic system in animal models of PD. Since BDNF and most of the tested growth factors do not protect and repair nigrostriatal neurons after the lesion in animal models of PD, they do not have therapeutic potential in the disease. Vascular endothelial growth factor A (VEGF-A) and VEGF-B are also promising NTFs and are neuroprotective and restorative in 6-hydroxydopamine (6-OHDA) model of PD. However, these factors have not been tested in non-human primate models of PD or in clinical trials. We have studied the effects of VEGF-C in rat 6-OHDA neuroprotective model, but its effect was rather modest [10]. GDNF has been the most promising NTF studied so far, as GDNF has been able to protect and repair DA neurons in the 6-OHDA and 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) models of PD in rodents and non-human primates (NHP) [11–13]. However, GDNF showed no clinical benefit in two phase II clinical trials, and related factor NRTN gene therapy had a very modest clinical benefit. Moreover, GDNF did not protect DA neurons in two a-synuclein models of PD [14,15], which may be problematic since a-synucleinopathies are one of the hallmarks in the pathophysiological brain samples of PD patients. We have recently discovered CDNF, [16] that together with the related factor MANF [17,18], forms a novel conserved NTF family (Fig. 1A, [19,20]). MANF promotes survival of DA neurons in vitro [18], and the invertebrate orthologue DmMANF has been shown to be crucial for maturation and maintenance of the fruit fly (Drosophila melanogaster) nervous system [21]. We have shown the therapeutic potential of CDNF in the rat 6-OHDA and mouse MPTP models of PD and found that CDNF is a potent NTF to protect DA neurons and more importantly restore their function [16,22,23]. Furthermore, in the severe rat 6-OHDA model of PD CDNF was more effective than GDNF [23]. CDNF and the related MANF are stable proteins, which diffuse better than any other tested trophic factor in brain tissue [23,24]. Our present understanding of the molecular mechanism of action of CDNF and MANF suggest that they are involved in the regulation of endoplasmic reticulum (ER) stress and unfolded protein response (UPR) [19,25]. In a recent study we discovered that lack of MANF in vivo in mouse leads to chronic activation of UPR [25]. In addition, extracellular CDNF rescues only the neurons that degenerate via ER stress (Krieglstein, Saarma, submitted). Thus, CDNF and MANF have significant potential as a treatment of PD. Our hypothesis is that CDNF can rescue DA neurons by blocking neuronal apoptosis at least partially by regulating the ER stress response. The exact survival-promoting mechanism of CDNF and MANF is not yet fully solved, but recent data from us [19,25] and others [26] indicate that CDNF and MANF are critically involved in the regulation of ER stress. Thus, their mechanism of action is drastically different from other known NTFs. They can act like classical NTFs by extracellularly promoting cell survival via activating PI3K-Akt pathway (Krieglstein, Saarma, submitted) and possibly other pathways. However, differently from other NTFs, CDNF and MANF are mainly located in the ER, regulate ER stress and UPR intracellularly, but can be secreted from the cells to extracellular space mostly upon ER stress. ER stress is an important pathway that regulates cell homeostasis, but may also lead to cell death in PD. Recent studies of autopsied tissue samples from PD patients, and in vivo experiments with tissue from a PD model in rodents indicate that ER stress is involved in the pathogenesis of PD [27]. Therefore, in addition to an extracellular survival promotion mechanism that is similar to conventional NTFs, CDNF/MANF efficacy in

neurodegenerative diseases is likely also related to their ER stressrelieving properties, that is different from other known NTFs. 2. A novel CDNF/MANF family of proteins with unique structure 2.1. Three known classes of neurotrophic factors NTFs are small secretory proteins that, by binding to their specific receptors, regulate the development, survival and maintenance of neurons. During early stages of development, NTFs are involved in the differentiation and migration of neuronal precursors and during target innervation they act as target-derived NTFs regulating the survival of neurons and through that determine the final number of neurons and the density of innervation. In the adult nervous system NTFs maintain healthy neurons, protect them from toxins and injury, and regulate neuronal plasticity. Three major classes of NTFs have been described. The first discovered and the best studied is neurotrophin family that consist of nerve growth factor (NGF), BDNF, neurotrophin-3 (NT-3) and neurotrophin-4 (NT-4), which regulate neuronal survival and plasticity by activating transmembrane receptor tyrosine kinase (RTK) of the Trk family. These factors and their pro-forms also bind to the p75 neurotrophic receptor and trigger neuronal death. Neurotrophins have a vast variety of effects on various neuronal populations in the peripheral nervous system (PNS) and central nervous system (CNS). BDNF (Fig. 2) can support the survival of embryonic DA neurons and protect them from death in neurotoxin models of PD. However, BDNF is not protecting and repairing DA neurons when added to the midbrain after the neurotoxin lesion. Moreover, its receptor knockout (KO) animals have intact midbrain DA system and BDNF has not been tested in clinical trials on PD. The family of neurokines, also known as neuropoietic cytokines, are small, structurally related secretory proteins that all signal via transmembrane gp130 receptor. This family includes ciliary neurotrophic factor (CNTF), cardiotrophin-1 (CT-1), leukemia inhibitory factor (LIF), neuropoietin (NPN), oncostatin M (OSM), cardiotrophin-like cytokine (CLC), interleukin 6 (IL-6), IL-11 and IL-27. Neurokines use either a two or three component receptor system and support mostly the survival of motoneurons (MNs) [28]. However, the biological effects of neurokines in the nervous system are much less studied than other NTFs. In addition to MNs, CNTF also has survival promoting effects on DA neurons and parasympathetic neurons, and LIF supports sensory neurons. For the DA neurons the most important factors belong to the GDNF family ligands (GFLs). GDNF and related growth factors NRTN, artemin (ARTN) and persephin (PSPN) form a distant family in the transforming growth factor-b (TGF-b) superfamily of growth factors [29–31]. GFLs specifically bind to GPI-anchored co-receptors of the GFRa family and the GFL–GFRa complex binds to and activates RET receptor tyrosine kinase. GDNF that binds to GFRa1 and NRTN that binds to GFRa2 support the survival of DA, but also have wide effects on different neuronal populations in the CNS and PNS [29] (Fig. 2). Most remarkably, GDNF is absolutely required for the development of the enteric nervous system and together with NRTN they are critical regulators of the development of parasympathetic neurons [32]. NRTN and its receptor GFRa2 KOs have defects in the parasympathetic nervous system, but their brain DA system is completely intact. GDNF, GFRa1 and RET knockouts die at birth because of the lack of kidney, but without defects in the DA system [29]. However, conditional deletion of GDNF in two laboratories has produced completely different results. Pascual et al. [33] removed GDNF from adult mice using tamoxifen induction and reported significant loss of DA neurons in the midbrain resulting is motor disorders. The study concluded that GDNF is critically important for the maintenance and function of DA neurons in adult animals. Our laboratory developed a mouse model

3741

M.H. Voutilainen et al. / FEBS Letters 589 (2015) 3739–3748

A 26 Pre

Human CDNF

21 Pre

Human MANF

25 Pre 22 Pre

161 ER 187 -KTEL 158 ER 179 -RTDL 157 ER 182 -TEEF 158 ER 180 -RTDL 151

D.melanogaster MANF

22 Pre

152

C.elegans MANF

16 Pre

ER 173 -RSEL ER 168 -KEEL

B C-terminal SAP-like domain

CGKC loop

N-terminal Saposin-like domain Signal peptide

ER retention signal RTDL Helix-loop-helix motif Fig. 1. (A) Cerebral dopamine neurotrophic factor (CDNF) together with mesencephalic astrocyte-derived neurotrophic factor (MANF) forms a novel conserved NTF family. Human and zebrafish have CDNF and MANF proteins and D. melanogaster and C. elegans only one MANF-like protein that all have eight conserved cysteines with similar spacing, 16–26 amino acids long signal sequence (Pre), 152–161 amino acids long mature form and KDEL receptors retention sequence at the C-terminus. (B) 3D solution NMR structure of human MANF reveals the N-terminal saposin-like domain and the C-termininal SAP-like domain connected with flexible linker, modified from [19].

that was equivalent to the model used in the Pascual et al. study. In addition, two other strategies to remove GDNF from the midbrain were used. Results from those three different approaches demonstrated that GDNF is not a necessary component of the maintenance of the midbrain DA system [34]. Our results are in line with results from Rüdiger Klein and Jeff Milbrandt laboratories, who analyzed DA system in RET conditional KO mice. Jain et al. [35] analyzed RET-deficient mice until 9 month of age and concluded that RET is dispensable for the midbrain DA system. Kramer et al. [36], however, found that RET-deficient mice have age-dependent mild loss of SN DA neurons. In a series of elegant experiments Robert E. Burke laboratory studied whether GDNF is

the target derived-neurotrophic factor for DA neurons in the SNpc. They found that programmed cell death (PCD) of DA neurons starts shortly after birth and lasts for 3 weeks. Their results demonstrate that GDNF promotes the survival of DA neurons during developmental PCD, but has no significant effects during adulthood [37,38]. GDNF and NRTN have also been tested in animal models of PD. In rodent and non-human primate neurotoxin models of PD, both factors can quite efficiently protect and even repair DA neurons. However, recent data show that GDNF is not able to rescue DA neurons in a-synuclein overexpression models of PD. These results, together with the modest effect of GDNF and NRTN in phase 2 clinical trials on PD patients, warrant the search for

3742

M.H. Voutilainen et al. / FEBS Letters 589 (2015) 3739–3748

Fig. 2. CDNF’s mode of action is drastically different from other known neurotrophic factors such as BDNF, GDNF or NRTN. We show here only CDNF, but CDNF and MANF are most likely acting in a similar fashion. Unlike other known NTFs, CDNF and MANF do not bind to the surface of naive cells. Our current hypothesis of CDNF and MANF signaling is that, in the ER-stressed cells, monomeric CDNF or MANF binds to a putative soluble co-receptor that activates still uncharacterized plasma membrane receptors leading to the activation of intracellular signaling pathways, such as PI3K/Akt pathway. Alternatively or in parallel, CDNF (and also MANF) binds to oxidized lipids at the cell surface and is internalized by an unknown mechanism leading to the regulation of ER stress and UPR pathways (PERK, IRE1, ATF6). Moreover, endogenous CDNF or MANF can be retained to the ER where they, differently from other NTFs, directly affect ER stress, protein misfolding and UPR. Neuronal survival promoting actions of BDNF, GDNF and NRTN are mediated by transmembrane tyrosine kinase receptors. Homodimeric BDNF activates TrkB tyrosine kinase receptors leading to the activation of PI3K/Akt, PLCc and MAP kinases. Homodimeric GDNF and NRTN bind first to GFRa1–2 co-receptor, then ligand co-receptor complex binds to and activates the RET tyrosine kinase leading to the activation of PI3K/Akt, Src and MAP kinase pathways thus promoting the survival of DA neurons.

new, more potent growth factors for the disease-modifying therapy of PD. 2.2. The discovery of CDNF and MANF MANF is the founding member of CDNF–MANF family and was identified by the group of John Commissiong from the culture medium of rat astrocyte cell line as a novel neurotrophic factor supporting the survival of embryonic DA neurons in culture [18].

When the authors sequenced the new gene they noticed that it was homologous to a predicted human arginine-rich protein (ARP) of 234 amino acids [18,39] (Fig. 1A). Human ARP (ARMET; arginine-rich, mutated in early stage tumors) gene contains an amino-terminal arginine-rich region and mutations within were associated with different cancer types [39–41], but ARMET protein has never been purified. Since the arginine-rich region of ARP was not translated in vivo [18] it is fully justified that the protein was renamed MANF. Human MANF is a secreted protein having 21

M.H. Voutilainen et al. / FEBS Letters 589 (2015) 3739–3748

amino acids long signal sequence followed by 158 amino acids long mature part of the protein (Fig. 1A). Using bioinformatics and biochemical approaches our laboratory identified the second member of the family and named it CDNF [16]. CDNF is a vertebrate specific paralogue of the MANF with 26 amino acids long signal sequence that, after co-translational proteolytic cleavage, renders the mature CDNF of 161 amino acids (Fig. 1A). Human MANF shows 59% amino acid identity with human CDNF and both proteins have eight cysteine residues at similar spacing, suggesting a novel protein fold. Sequence analysis from different organisms revealed that vertebrates, including mouse and zebrafish Danio rerio, have orthologous genes for both CDNF and MANF. Human and zebrafish MANF proteins share 69% amino acid identity, whereas CDNF proteins from these species show 53% amino acid identity. Mammalian CDNF and MANF are rather unique among NTFs because they have a single ancestral MANF/CDNF gene orthologous in invertebrates, such as the nematode Caenorhabditis elegans and fruit fly D. melanogaster, which encode highly homologous proteins of similar size and conserved cysteine pattern (Fig. 1A). Human CDNF shows 49% amino acid identity with D. melanogaster and 46% identity with C. elegans MANF, whereas human MANF shows 53% identity with fruit fly and 50% of identity with the worm protein sequence. Thus, vertebrate and invertebrate CDNF and MANF form an evolutionarily conserved family of NTFs that have eight conserved cysteines with similar spacing. Surprisingly, MANF and CDNF contain a functional KDEL ER retention signal at their C-terminus [26] (Fig. 1A). Experimental evidence suggests that MANF can bind to KDEL receptors [42] and ER chaperone GRP78 [26] and thus function as the ER resident protein. The removal of the RTDL sequence from MANF C-terminus increases its secretion and supports the assumption that CDNF–MANF proteins may have dual modes of action: intracellular in the ER and extracellular acting via yet to be characterized plasma membrane receptors when secreted to the extracellular space. 2.3. Structure of CDNF and MANF Sequence analysis did not allow identifying homologous protein families and therefore the analysis of the three-dimensional (3D) structure of these proteins turned out to be very informative (Fig. 1B). We first solved the crystal structure of the human MANF protein and also the 3D structure of the N-terminal domain of human CDNF [28]. Crystal structure analyses revealed that CDNF and MANF proteins consist of two structurally independent domains: the N-terminal domain from 1 to about 100 amino acids, and the C-terminal domain from amino acids about 110–160. These two domains are linked with the relatively flexible loop sequence. The N-terminal domains of CDNF and MANF are rather similar, being composed of the five a-helices and one 310 helix and showing structural similarity to saposins [28]. Saposins are functionally pleiotropic proteins, but as a common property they can interact with lipids and membranes [43]. Our very preliminary data indicate that the N-terminal domain of CDNF and MANF can interact with specific phospholipids [Zhao, Lindholm, Lappalainen, personal communication]. This preliminary finding is consistent with two patches of conserved lysines and arginines on the surface of the molecule. Crystal structure demonstrated that the C-terminus of MANF is natively unfolded and, like reductases and disulfide isomerases, contains a CKGC motif with disulfide bridge, consistent with a role in the ER stress response. Importance of the CKGC sequence is underlined by the finding that mutating cysteines in this loop renders MANF biologically inactive [44], Mätlik et al. unpublished. Since the crystal structure of MANF showed a poorly defined and highly disordered C-terminal domain we analyzed MANF solution structure by nuclear magnetic resonance (NMR)

3743

(Fig. 1B). The 3D solution structure of full-length human MANF, determined by multidimensional NMR spectroscopy confirmed the earlier suggested two-domain architecture [19]. However, the solution structure of MANF reveals a well-defined globular structural module in the C-terminal domain (Fig. 1B). Remarkably, the C-terminal domain is homologous to the SAP (SAF-A/B, Acinus, and PIAS) domain of Ku70, a well-known inhibitor of proapoptotic Bax [19]. These data predicted a very tempting hypothesis that MANF and CDNF can inhibit apoptosis via direct interaction with Bax. Intracellularly expressed MANF and CDNF can indeed counteract the intrinsic Bax-involving apoptotic pathways in the sympathetic neurons [19]. However, our repeated attempts to verify the direct MANF-Bax interactions have failed, suggesting that the apoptosis could be blocked upstream of Bax. Hoseki and co-workers (2010) have independently determined the NMR structure of MANF that is highly similar to the MANF structure determined by us [45]. Latge et al. (2015) have recently reported the crystal structure of full-length CDNF and confirmed that the overall 3D structure of CDNF and MANF are very similar [46]. In addition, they have identified two surface patches in the N-terminal domain that have high conformational dynamics and therefore are potential candidates for functionally important sites. Both crystal structure and solution NMR analyses revealed a unique 3D structure of human MANF and CDNF that differs drastically from other NTFs. Remarkably, the C-terminal domain is homologous to the SAP domain of Ku70, whereas the N-terminal domain is homologous to saposins that can bind lipids and membranes. Thus, CDNF–MANF family proteins differ from other growth factors and NTFs by sequence and also by the 3D structure. Quite remarkably they differ from other NTFs also by their intracellular location. Usually secreted proteins are synthesized as proproteins on the rough ER and then co-translationally cleaved to the mature form and transported from ER through Golgi to the extracellular space. MANF and CDNF are also synthesized by the ER bound ribosomes, but due to the interaction with KDEL receptors and GRP78, retained in the ER. Our preliminary data demonstrate that in the healthy cells the vast majority of the CDNF and MANF proteins (up to 90%) are located in the ER and only a small proportion of the protein is secreted. This is very different from other NTFs, which are mostly secreted and only pass ER and Golgi. Where exactly in the ER CDNF and MANF are located is still not entirely clear. Interaction with GRP78 chaperone and immunocytochemical analysis indicated that they may localize on the luminal side of the ER [16,17,26,42]. Identification of the intracellular localization of CDNF and MANF is very important for the understanding of their mode of action. Additional immunocytochemical and immunohistochemical analyses with highly specific antibodies and also the use of electron microscopy are needed. 2.4. CDNF and MANF expression CDNF and MANF expression has mostly been analyzed in rodents and very little information about the expression of these proteins in human tissues exists. In general, MANF expression is widespread and it is detected in many tissues and organs of rodents starting from early stages of development [17,25]. High levels of MANF protein can be detected in rodent and human sera (Galli, Pulkkila, Saarma, unpublished). Relatively high MANF levels are detected in the cerebral cortex, hippocampus, cerebellar Purkinje cells, as well in the thalamus and hypothalamus. MANF was also expressed in the striatum and in the SN, but its expression levels are much lower compared to the Purkinje cells [17]. Wang et al. studied MANF expression using antibodies and also found widespread expression in the brain that is mostly localized in neurons [47]. CDNF expression has been studied less compared to MANF and its expression levels are generally lower [16]. In agreement with

3744

M.H. Voutilainen et al. / FEBS Letters 589 (2015) 3739–3748

the rodent data, a wide expression of CDNF transcripts was detected in the human brain and non-neuronal tissues. cdnf mRNA was found in the hippocampus and thalamus, but also in the striatum and SN. Using antibodies for CDNF, protein was detected in the adult cerebral cortex, hippocampus and striatum. In the SN, CDNF was detected in solitary cells that did not express tyrosine hydroxylase (TH), a marker for DA neurons. High levels of CDNF protein were detected in the Purkinje cells of the cerebellum and in regions of the brain stem, including the locus coeruleus.

and MANF differ in several aspects from classical NTFs and growth factors: they have unique amino acid sequence and 3D structure, they function inside the cells regulating ER stress, but also via still unknown plasma membrane receptors by promoting neuronal survival. CDNF and MANF seem not to affect un-lesioned, healthy neurons in vitro and in vivo, but efficiently protect and repair lesioned DA neurons.

3. CDNF and MANF work differently than other growth factors and neurotrophic factors

Although MANF and CDNF are secreted proteins, they have been shown to mostly localize in the ER [26,53,54]. MANF secretion to the culture medium from cardiac myocytes and HeLa cells has been induced by thapsigargin treatment, which depletes calcium from the stores of the ER. The level of both proteins is also induced in ER stress and MANF interacts with chaperone GRP78 (also known as Bip, a member of heat shock protein 70kDa family, Hsp70) in a calcium-dependent manner [26]. This suggests that MANF and CDNF secretion is enhanced in pathological situations involving reduced ER calcium concentrations such as brain and heart ischemia as well as other pathological situations where ER calcium is misbalanced due to accumulation of misfolded proteins. These results provide a plausible mechanism where extracellular MANF and CDNF could protect cells from death in response to ER calcium depletion. Although CDNF and MANF are known to function in the ER, their exact intracellular mode of action is not fully understood. MANF contains a C-terminal sequence (RTDL) and CDNF the sequence (KTEL) closely resembling the classical ER retention signal (KDEL) (Fig. 1A). MANF and CDNF trafficking and secretion could be regulated by their RTDL/KTEL interaction with the ER retention receptor (KDELR1, 2 and 3b) [42]. In a recent study, indirect evidence indicated binding of MANF to the KDELR, which in ER-stressed cells was also expressed on the cell surface [42]. The C-terminal SAP-like domain contains a CXXC cysteine bridge motif, suggesting that MANF and CDNF may be involved in protein folding in the ER [28]. The ER is an intracellular organelle that contributes to the proper folding and processing of nascent translated proteins destined for secretion. Under normal physiological conditions, unfolded, misfolded or aggregated proteins are degraded. Accumulating unfolded proteins in the ER lumen can trigger ER stress and activate the UPR, a cellular defence mechanism to combat ER stress. Continuous protein aggregation and chronic ER stress are toxic to the cells leading to ER stressinduced apoptosis. UPR regulates the ER stress of the cell by three different pathways: mRNA degradation and suppression of further protein translation, facilitation of protein refolding by the induction of ER chaperons, and activation of ER-associated degradation of unfolded proteins. The UPR is mediated through three ER transmembrane receptors: inositol-requiring enzyme 1 (IRE1), pancreatic ER kinase-like ER kinase (PERK), and activating transcription factor 6 (ATF6) [55]. These three transmembrane receptors are activated in ER stress by dissociating from an ER chaperone, GRP78 that is associated with several proteins. Phosphorylated PERK blocks general mRNA translation by phosphorylating eukaryotic initiation factor 2 (eIF2) a subunit. However, transcription factor ATF4 is produced despite translational block to restore ER homeostasis, or later CHOP to trigger the apoptotic response. ATF6 and IRE1 pathways regulate the expression of ER chaperone genes and IRE1 degrades mRNA and, through spliced XBP1, induces the expression of genes for protein degradation. Several lines of evidence show that MANF and CDNF are involved in the regulation of UPR and unresolved ER stress. Firstly, MANF KO mice develop severe diabetes mellitus due to progressive postnatal reduction of b-cell mass caused by decreased b-cell proliferation and increased b-cell apoptosis [25]. Importantly, MANF-deficient mice show chronically activated UPR pathways

3.1. The mode of action of CDNF and MANF When we first characterized and purified recombinant human CDNF protein we were very surprised to find out that CDNF had no effects on cultured rodent neurons when added to the culture medium. We tested sensory, sympathetic, hippocampal, motoneurons and also embryonic DA neurons in culture that were dying due to deprivation of NTFs or treatment with genotoxic toxin etoposide. Having seen no survival promoting or neurogenic response, that is typical to all known NTFs that we used as positive controls, we first concluded that for some reason our recombinant protein is biologically inactive. After almost one year of testing recombinant CDNF expressed in different host cells, including bacterial, baculovirus and mammalian cells, we finally concluded that the lack of response to CDNF in cultured neurons as well as in all tested cell lines is its biological property [16]. Thus the in vitro mode of action of CDNF and MANF is very different from other NTFs, such as GDNF and BDNF (Fig. 2). These factors bind to their specific receptors, activate intracellular signaling pathways and promote the survival of DA neurons both in vitro and in vivo (Fig. 2). We then tested the same recombinant CDNF in rat 6-OHDA model of PD and to our great satisfaction CDNF most efficiently protected and even repaired SN DA neurons and improved motor behavior of lesioned animals [16]. CDNF also efficiently protects lesioned hippocampus in animal models of Alzheimer’s disease [48], but fails to influence rat and mouse DA neurons in naïve, healthy rodents [23,49]. Thus, CDNF differs from other known growth factors and NTFs in amino acid sequence, 3D structure and in their in vivo mode of action. CDNF has no effects on naive, healthy cells and neurons, but seem to require a lesion for its action. The mechanism of action of MANF may be somewhat different from CDNF. MANF was first characterized as the NTF based on its survival promoting effects on embryonic DA neurons when added to the culture medium [18]. Moreover, we have recently reported that recombinant human MANF significantly and dose-dependently increased pancreatic b cell proliferation [25]. Furthermore, MANF seems to be involved in protein kinase C signaling [50] and protects cardiac myocytes [51] and cultured astrocytes [52] from apoptosis. In these experiments, MANF prevented death of the neurons and cells and also stimulated growth of the DAergic fibers, thus acting as a typical NTF. Similarly to CDNF, most types of neurons do not respond to MANF in culture. The situation changes very dramatically when MANF or CDNF are expressed inside the cells. Differently from all known NTFs, MANF is a resident protein of the ER and protects cells intracellularly against ER stress [26,42,51,53,54]. We have also shown that microinjected intracellular MANF protects cultured superior cervical ganglion (SCG) neurons from toxin-induced or ER stress-induced cell death, where extracellularly added MANF had no effect [19]. Our recent unpublished data also show that intracellular CDNF can rescue SCG neurons and DA neurons from ER stress-induced cell death. It is important to note that classical NTFs usually have no survival promoting or neuritogenic effects on neurons when expressed inside the cells. Taken together, CDNF

3.2. MANF and CDNF regulate ER stress

M.H. Voutilainen et al. / FEBS Letters 589 (2015) 3739–3748

[25]. Secondly, Drosophila-null mutants of DmMANF showed chronically activated ER stress and increased PERK phosphorylation [21]. Thirdly, MANF interacts with GRP78 in the ER in a calcium-dependent manner [26]. Fourthly, MANF and CDNF regulate ER stress and are up-regulated in vitro and in vivo in UPR (reviewed in [20,56]). Finally, Tadimalla and co-workers made an interesting observation that ATF6 activation decreased ischemic damage in an ex vivo model of myocardial ischemia/reperfusion and induced numerous genes, including MANF. Additionally, simulated ischemia induced MANF in an ATF6-dependent manner in myocyte cultures [51]. Recent work shows that CDNF can protect DA neurons from the injury caused by a-synuclein oligomers, as is evident in PD [46]. However, the exact mechanism of action of MANF and CDNF in the ER stress and UPR is still unclear. 3.3. Hypothetical mode of action What is the possible mode of action of CDNF and MANF? CDNF and MANF are growth factors with unique sequence, structure and with dual modes of action. They are located to the ER but can also be secreted. Since the majority of CDNF and MANF are retained in the ER, most probably their most important function inside the cells is to regulate ER stress and UPR. As MANF binds to GRP78 it can also interact (at least indirectly) with UPR receptors IRE1, PERK and ATF6 and thus participate in their regulation. The ability of MANF and CDNF to alleviate ER stress and in particular the chronic activation of all these three UPR pathways in MANF-deficient mice supports this hypothesis. It should be mentioned that how exactly the alleviation of ER stress leads to the blockage of apoptotic pathways is not yet studied. CDNF and MANF can reach ER by two possible mechanisms: firstly, they can, after synthesis, retain in the ER by the virtue of their ER retention signals (Fig. 2); secondly, when extracellular, they can possibly interact with membranes or membrane lipids, or proteins, being then internalized and carried to ER by an unknown mechanism. When ER stress is becoming chronic MANF and CDNF are secreted, as has been shown in several cell lines and also in cardiac myocytes. The most striking property of CDNF (and partially also for MANF), compared to all other known growth factors, is CDNF’s inability to stimulate any effects in any tested cells extracellularly in the culture medium [16,19,23]. In collaboration with Kerstin Krieglstein’s laboratory we have recently solved this puzzle by demonstrating that extracellular CDNF can have effects on the cells, but only after applying ER stress or cellular injury, but not physiological death stimuli (NTF deprivation) (Zhao, Saarma, Krieglstein, submitted). It is not entirely clear what signaling pathways are activated, but there is some evidence that PI3K-AKT (Zhao, Saarma Krieglstein, submitted), PKC [50,57] and NFvB [58] pathways may be involved. It is not yet clear whether and how these survival pathways are related to ERmediated survival mechanisms of MANF and CDNF. Hypothetical modes of action for CDNF are shown in Fig. 2 and we think that CDNF and MANF function quite similarly. We postulate that these factors can directly bind to an injury-induced plasma membrane receptor or that their action may be mediated by an injury-induced cofactor that can be a protein or a lipid. It is important to stress that at least CDNF has no effects on DA neurons in healthy untreated animals. CDNF and MANF effects on DA neurons in animal models of PD are described below. However, it is important to note that these factors have survival promoting and injury protecting effects also on other neurons. MANF protein and gene therapy can protect striatal and cortical neurons and reduce ischemic brain injury [59,60]. In the spinocerebellar ataxia 17 mouse model MANF transcription is reduced and overexpression of MANF prevents Purkinje cell degeneration via PKCdependent signaling [50]. For comparison, the survival-promoting mechanism of classical NTFs is also shown on Fig 2. GDNF activates

3745

RET and BDNF TrkB triggering intracellular signaling pathways that stimulate the survival of DA neurons (Fig. 2). 3.4. MANF ablation reveals an interesting phenotype in mice The data on the in vivo effects of exogenously added or transgenetically overexpressed CDNF and MANF on DA, cortical, hippocampal and cerebellar Purkinje neurons suggest that MANF and CDNF-deficient mice may have interesting neuronal phenotypes. Quite surprisingly, MANF-deficient mice develop severe diabetes mellitus (DM) characterized by progressive reduction of pancreatic insulin producing b cell mass. Loss of b cells is caused by decreased proliferation and increased apoptosis. As stressed earlier, MANF-deficiency in mouse renders chronic activation of PERK, ATF6 and IRE1 receptor pathways of UPR. Since a severe form of DM in mice is developed relatively early it may also have a strong influence in the nervous system. Therefore we decided not to analyze the defects in the nervous system in MANF-full KOs, but rather in MANF conditional KO mice that we have developed [25]. CDNF KO mice were developed by Dr. Lindahl in our laboratory and further analysis of the phenotype, particularly in DA neurons of CDNF-deficient mice is ongoing. Interestingly, a recent clinical exome sequencing study on 149 Qatar patients with mainly neurocognitive phenotypes identified a patient with MANF mutation, who presented type 2 diabetes, hypothyroidism, primary hypogonadism, short stature, mild intellectual disability, obesity, deafness, high myopia, microcephaly and partial alopecia [61]. 4. Effect of CDNF and MANF on dopamine neurons 4.1. Studies on D. melanogaster NTFs that regulate the development, maintenance and functioning of DA neurons in D. melanogaster have not been described. However, cell ablation studies in Drosophila suggest trophic interaction between neurons and glia i.e. putative NTFs that can potentially support the survival of fly DA neurons should be expressed by glial cells. Differently from mammals, D. melanogaster MANF is expressed in glial cells, whereas in the mammalian nervous system MANF and CDNF are mostly expressed by the neurons. This is quite remarkable and suggests that a fly CDNF–MANF homolog may be the long searched NTF for fly DA neurons. D. melanogaster has a single protein homologous to mammalian CDNF and MANF proteins. The fly protein is carrying 8 conserved cysteines and shows 51% homology with human MANF and 49% homology with human CDNF. Since the homology with MANF is more extended we named the protein DmMANF [21]. The protein sequence has similar elements to mammalian counterparts: a 22 amino acid long signal sequence followed by 151 amino acid long mature part of the protein and the RSEL as putative ER retention signal sequence (Fig. 1A). The conserved cysteine pattern predicts that DmMANF 3D structure resembles that of mammalian CDNF and MANF. To study the biology of DmMANF we deleted both maternal and zygotic DmMANF and found that this leads to the degeneration of axonal bundles of DA neurons in the embryonic CNS and subsequent non-apoptotic cell death. Furthermore, MANF-deficient flies had significantly lower DA levels than wildtype (WT) counterparts demonstrating that DmMANF is important for the development and maintenance of DA neurons and DA levels in the fly [21]. We then tested whether human MANF and CDNF can rescue fly lethality when overexpressed in MANF-deficient flies. The rescue experiments demonstrated that both mammalian MANF, but also CDNF can rescue mutant fly lethality confirming DmMANF as a functional ortholog of the human MANF–CDNF genes [21,44]. In subsequent studies using a transgenic rescue approach in the DmMANF mutant background we found that only

3746

M.H. Voutilainen et al. / FEBS Letters 589 (2015) 3739–3748

full-length MANF containing both the N-terminal saposin-like and C-terminal SAP-domains can rescue the larval lethality of the DmMANF mutant. Separate N- or C-terminal domains of MANF, even when co-expressed together, failed to rescue the fly. Deleting the signal peptide or mutating the CXXC motif in the C-terminal domain destroyed the activity of full-length DmMANF [44]. Palgi et al. (2012) analyzed in parallel the ultrastructural data and transcriptome changes of maternal/zygotic and only zygotic MANF mutants and found significant alteration in the ER stress and membrane traffic-related proteins and structures [21]. They postulated that DmMANF is an important cellular survival factor needed to overcome the UPR, especially in tissues with high secretory function. Remarkably, they revealed the changes in the expression of several genes involved in PD in Drosophila MANF mutants. Taken together, studies on fly MANF provide a clue on the mechanisms by which this novel NTF protects DA neurons. 4.2. CDNF and MANF in Zebrafish Unlike D. melanogaster, in zebrafish CDNF and MANF are duplicated into separate genes (Fig. 1A). Zebrafish has become an increasingly used vertebrate animal model to study gene functions and also to model diseases, including neurodegenerative diseases such as PD, Huntington’s disease and Alzheimer’s disease. Both zebrafish CDNF and MANF have, like other family members, eight conserved cysteines. The zebrafish CDNF consists of 182 amino acids: the signal peptide of 25 amino acids and the mature CDNF of 161 amino acids with non-conserved TEEF KDELR retention signal at the C-terminus. Zebrafish MANF consists of 168 amino acids with 16 amino acids in the signal peptide, the mature form of 152 amino acids and it has the conserved RTDL sequence at the C-terminus. The RTDL sequence has been shown to be important for cellular localization and secretion of MANF [42]. Similarly as in mouse brain [47], it has been shown that MANF expression is widespread during embryonic development as well as in adult [62]. In zebrafish, the highest mRNA expression has been shown to be in the whole embryos at two hours post-fertilization and then the mRNA expression gradually decreases which indicates that manf mRNA is maternally contributed [62]. In adult tissues manf mRNA is approximately 30-fold higher in the liver than in the brain, kidney or eye. Similarly to expression in rodent brain, zebrafish brain MANF is mainly expressed in neurons and few MANF positive glial cells have been observed [62]. Few MANFpositive cells in the zebrafish have been shown to co-localize with markers of DA neurons. In a study where morpholino oligonucleotides were used to reduce levels of MANF, a clear DA phenotype was found [62]. In zebrafish larvae with reduced levels of MANF the DA concentration was decreased as compared with wild-type larvae. Moreover, the expression levels of the two tyrosine hydroxylase (th) genes: th1 and th2 were decreased as well as the number of th1 and th2 cells in the diencephalon. The effect was specific to DA neurons and no alterations were observed in GABA, 5-HT or epinephrine levels. Interestingly, it was found in the zebrafish that reduced levels of MANF were associated with increased levels of transcription factor pax2a as well as in nr4a2b, which play role in differentiation and survival of DAergic progenitor cells [62]. Thus, the results on zebrafish MANF are very similar to those obtained in fruit fly and suggest that MANF is involved in the regulation of the development of DAergic system in these organisms. The effects of CDNF deletion in zebrafish have not yet been reported. 4.3. Effect of CDNF and MANF in 6-OHDA model of PD We have shown that a single intrastriatal injection of CDNF and MANF protects nigrostriatal DAergic neurons when administered six hours before the neurotoxin 6-OHDA [16,24]. More

importantly, the function of the lesioned nigrostriatal DAergic system was partially restored even when the NTFs were administered four weeks after 6-OHDA [16,24] (Fig. 3). A 14-day continuous intrastriatal infusion of CDNF, but not of MANF, was able to restore the function of the midbrain neural circuits controlling movement when initiated two weeks after intrastriatal 6-OHDA. Continuous infusion of CDNF also protected TH-positive neurons in the SNpc and fibers in striatum from toxin-induced degeneration [23]. Although most effects of CDNF are thought to directly target stressed neurons, recent studies suggest that CDNF may also support neuroregeneration by suppressing neuroinflammation via effects on astrocytes and microglial cells [63–65]. 4.4. Effect of CDNF in MPTP model 1-Methyl-4-phenyl-1,2,3,6-tetrahydropyridine is a neurotoxic precursor for DA neuron toxic MPP+. MPP+ causes DA neuron degeneration in C57/Bl6 mice, primates and in humans. MPTP is metabolized into MPP+ by the enzyme MAO-B and MPP+ interferes with complex I in the mitochondrial electron transport chain. In a study where the effects of CDNF were studied in the MPTP model of PD in C57BL/6 mice it was found that CDNF has both neuroprotective and neurorestorative properties against MPTP [49]. When CDNF was given one day before MPTP it was found that CDNF protects TH-positive neurites in the striatum and SN as well as THpositive neurons in the SN. The proactive effects observed in immunohistochemistry were associated with behavioral improvement. Moreover, when administered one week after MPTP, bilateral CDNF microinjections into striatum improved locomotor behavior and increased TH immunoreactivity in the striatum and the number of TH+ cells in the SN. These results are important since MPTP can cause Parkinsonian symptoms in humans and CDNF can not only restore the DAergic circuitry but also protect cell bodies. 4.5. Protein versus gene therapy approach for PD The prospects and challenges of the gene therapy approach of NTFs on PD has been recently reviewed in [56]. The gene therapy approach of targeting DA neurons in the SN or DA fibers in the striatum needs careful consideration of promoter, viral vector type as well as its serotype. Intra-striatal injections of adeno-associated viral (AAV) vectors serotype 2 under CMV promoter and carrying cDNA of CDNF have been shown to protect and recover 6-OHDAinduced behavior deficits and result in a significant restoration of TH immunoreactive (TH-ir) neurons in the SNpc and TH-ir fiber density in the striatum [66,67]. Moreover, AAV7-MANF with CMV promoter has been shown to be protective in cortical neurons against stroke [60]. In contrast when CMV promoter, which is not DA neuron-specific, was used to drive intranigral CDNF and MANF expression with lentiviral vectors in rat it was found that only combination of both CDNF- and MANF-expressing viral vectors showed neuroprotection [68]. Additionally, see above discussion of studies with recombinant proteins in rats and mice. These results demonstrate the importance of choosing injection site in order to gain neuroprotection. Moreover, for SN injections and targeting DA neurons, the choice of promoter is important in order to observe neuroprotection. Although direct comparison of AAV and lentiviral gene therapy approaches are difficult, AAVs are thought to a be safe and efficient approach, particularly for neurodegenerative diseases [56]. The result that only combination of CDNF and MANF showed neuroprotection after lentiviral vector injections into SN also suggest towards the possibility that the lentivirus driven expression of a single factor is not efficacious enough to be protective. It should also be remembered that both CDNF and MANF are mainly located in the ER and secreted from cells upon

3747

M.H. Voutilainen et al. / FEBS Letters 589 (2015) 3739–3748

After CDNF

CDNF

Dopamine neurons lose synapses, degenerate and die resulting in impaired dopaminergic circuitry.

Regenerative dopamine neurons

Degenerative dopamine neurons

Before CDNF

CDNF

The remaining axons are stimulated to regenerate and branch.

Death of dopamine neurons is blocked and dopaminergic circuitry is restored.

Fig. 3. Schematic figure of degenerating DA neurons and CDNF-induced regeneration. On the left DA neurons degenerate due to 6-OHDA or MPTP. From the top: (1) DA neurons have their synaptic connections to the target striatal cells. (2) DA neurons have lost their synaptic connections and axons have degenerated. (3) DA neurons have lost their DA phenotype but remain their cytoarchitectual structures and are alive. (4) DA neurons have degenerated and died. On the right DArgic circuitry is regenerated after CDNF treatment. From the top: (1) DA neurons have their synaptic connections to the target cells and CDNF protects these neurons from degeneration and death. (2) DA neurons have regenerated the axons and re-established their synaptic connections. (3) DA neurons have restored their DA phenotype, and restored their synaptic connections to the target cells. (4) CDNF cannot affect DA neurons that have died.

ER stress. Therefore, it may be that unlike many readily secreted proteins, the efficacy of neuroprotection may be different whether the protein is expressed in DA neurons versus in post-synaptic neurons, and whether the proteins are readily secreted or not. At the moment the optimal settings for the gene therapy approach for CDNF and MANF have not been thoroughly studied. Results from the CDNF protein infusion and gene therapy studies suggest a retrograde transport system from the striatum to the SN for CDNF as measured with ELISA or with 125I-labelled proteins. Indeed, we have shown that CDNF has a retrograde transportation profile similar to GDNF [23] and that the transportation profile of CDNF and MANF is different [23,24]. Delivery of CDNF did, however, not induce changes in the intact rat DAergic system [23] so CDNF differs in this property from many NTFs. 4.6. Summary and future directions CDNF and MANF differ from other known NTFs by sequence, 3D structure and mode of action. They share some key features of NTFs, as they both protect DA neurons, Purkinje neurons and cortical neurons in vivo when injected into the injured animal brain. In opposition to other NTFs, such as GDNF and BDNF, CDNF and MANF do not rescue neurons in vitro when added to the culture medium, from the physiological apoptosis due to absence of NTFs, but rescue the ER-stressed neurons. CDNF and MANF do not have effects on DA neurons when injected to the brain of naïve mice and rats [23,49]. This is also different from GDNF, BDNF and other NTFs. Thus, like other NTFs they protect injured neurons from death by activating still unknown plasma membrane receptors and classical anti-apoptotic signaling pathways. However, they also act inside the cells, where they regulate ER stress and UPR, additionally contributing to the survival and maintenance of DA neurons. In the first studies in animal models of PD, CDNF and

MANF have shown strong therapeutic effects in rodents, but in the coming years important new information is required to understand how they act on non-human primate models of PD and whether they can rescue DA neurons in an a-synuclein model of PD, where GDNF does not work. As our current data show that CDNF and MANF function only on lesioned neurons, it is interesting to see whether they have disease-modifying therapeutic effect in clinical trials on Parkinson’s disease. It is of utmost importance to identify CDNF/MANF plasma membrane receptors and clarify their mode of action. Acknowledgements We thank Katrina Albert for critical reading of the manuscript. MS and MHV were supported by the Grant from Jane and Aatos Erkko Foundation – Finland. UA was supported by the Estonian Research Council Grant PUT110. MHV and MA were supported by the Academy of Finland. References [1] Thomas, B. and Beal, M.F. (2007) Parkinson’s disease. Hum. Mol. Genet., R183– R194. 16 Spec No. 2. [2] Airavaara, M. et al. (2012) Neurorestoration. Parkinsonism Relat. Disord. 18 (Suppl. 1), S143–S146. [3] Hefti, F. (1997) Pharmacology of neurotrophic factors. Annu. Rev. Pharmacol. Toxicol. 37, 239–267. [4] Lanni, C. et al. (2010) The expanding universe of neurotrophic factors: therapeutic potential in aging and age-associated disorders. Curr. Pharm. Des. 16 (6), 698–717. [5] Skaper, S.D. and Walsh, F.S. (1998) Neurotrophic molecules: strategies for designing effective therapeutic molecules in neurodegeneration. Mol. Cell. Neurosci. 12 (4–5), 179–193. [6] Bjorklund, A. et al. (1997) Studies on neuroprotective and regenerative effects of GDNF in a partial lesion model of Parkinson’s disease. Neurobiol. Dis. 4 (3–4), 186–200.

3748

M.H. Voutilainen et al. / FEBS Letters 589 (2015) 3739–3748

[7] Agid, Y. (1991) Parkinson’s disease: pathophysiology. Lancet 337 (8753), 1321–1324. [8] Bernheimer, H. et al. (1973) Brain dopamine and the syndromes of Parkinson and Huntington. Clinical, morphological and neurochemical correlations. J. Neurol. Sci. 20 (4), 415–455. [9] Cheng, H.C., Ulane, C.M. and Burke, R.E. (2010) Clinical progression in Parkinson disease and the neurobiology of axons. Ann. Neurol. 67 (6), 715– 725. [10] Piltonen, M. et al. (2011) Vascular endothelial growth factor C acts as a neurotrophic factor for dopamine neurons in vitro and in vivo. Neuroscience 192, 550–563. [11] Hoffer, B.J. et al. (1994) Glial cell line-derived neurotrophic factor reverses toxin-induced injury to midbrain dopaminergic neurons in vivo. Neurosci. Lett. 182 (1), 107–111. [12] Rosenblad, C., Martinez-Serrano, A. and Bjorklund, A. (1998) Intrastriatal glial cell line-derived neurotrophic factor promotes sprouting of spared nigrostriatal dopaminergic afferents and induces recovery of function in a rat model of Parkinson’s disease. Neuroscience 82 (1), 129–137. [13] Tomac, A. et al. (1995) Protection and repair of the nigrostriatal dopaminergic system by GDNF in vivo. Nature 373 (6512), 335–339. [14] Decressac, M. et al. (2011) GDNF fails to exert neuroprotection in a rat alphasynuclein model of Parkinson’s disease. Brain 134 (Pt. 8), 2302–2311. [15] Lo Bianco, C. et al. (2004) Lentiviral nigral delivery of GDNF does not prevent neurodegeneration in a genetic rat model of Parkinson’s disease. Neurobiol. Dis. 17 (2), 283–289. [16] Lindholm, P. et al. (2007) Novel neurotrophic factor CDNF protects and rescues midbrain dopamine neurons in vivo. Nature 448 (7149), 73–77. [17] Lindholm, P. et al. (2008) MANF is widely expressed in mammalian tissues and differently regulated after ischemic and epileptic insults in rodent brain. Mol. Cell. Neurosci. 39 (3), 356–371. [18] Petrova, P. et al. (2003) MANF: a new mesencephalic, astrocyte-derived neurotrophic factor with selectivity for dopaminergic neurons. J. Mol. Neurosci. 20 (2), 173–188. [19] Hellman, M. et al. (2011) Mesencephalic astrocyte-derived neurotrophic factor (MANF) has a unique mechanism to rescue apoptotic neurons. J. Biol. Chem. 286 (4), 2675–2680. [20] Lindholm, P. and Saarma, M. (2010) Novel CDNF/MANF family of neurotrophic factors. Dev. Neurobiol. 70 (5), 360–371. [21] Palgi, M. et al. (2009) Evidence that DmMANF is an invertebrate neurotrophic factor supporting dopaminergic neurons. Proc. Natl. Acad. Sci. U.S.A. 106 (7), 2429–2434. [22] Airavaara, M. et al. (2011) CDNF protects the nigrostriatal dopamine system and promotes recovery after MPTP treatment in mice. Cell Transplant.. [23] Voutilainen, M.H. et al. (2011) Chronic infusion of CDNF prevents 6-OHDAinduced deficits in a rat model of Parkinson’s disease. Exp. Neurol. 228 (1), 99– 108. [24] Voutilainen, M.H. et al. (2009) Mesencephalic astrocyte-derived neurotrophic factor is neurorestorative in rat model of Parkinson’s disease. J. Neurosci. 29 (30), 9651–9659. [25] Lindahl, M. et al. (2014) MANF is indispensable for the proliferation and survival of pancreatic beta cells. Cell Rep. 7 (2), 366–375. [26] Glembotski, C.C. et al. (2012) Mesencephalic astrocyte-derived neurotrophic factor protects the heart from ischemic damage and is selectively secreted upon sarco/endoplasmic reticulum calcium depletion. J. Biol. Chem. 287 (31), 25893–25904. [27] Colla, E. et al. (2012) Accumulation of toxic alpha-synuclein oligomer within endoplasmic reticulum occurs in alpha-synucleinopathy in vivo. J. Neurosci. 32 (10), 3301–3305. [28] Parkash, V. et al. (2008) The structure of the glial cell line-derived neurotrophic factor-coreceptor complex: insights into RET signaling and heparin binding. J. Biol. Chem. 283 (50), 35164–35172. [29] Airaksinen, M.S. and Saarma, M. (2002) The GDNF family: signalling, biological functions and therapeutic value. Nat. Rev. Neurosci. 3 (5), 383–394. [30] Andressoo, J.O. and Saarma, M. (2008) Signalling mechanisms underlying development and maintenance of dopamine neurons. Curr. Opin. Neurobiol. 18 (3), 297–306. [31] Aron, L. and Klein, R. (2011) Repairing the Parkinsonian brain with neurotrophic factors. Trends Neurosci. 34 (2), 88–100. [32] Rossi, J. et al. (1999) Retarded growth and deficits in the enteric and parasympathetic nervous system in mice lacking GFR alpha2, a functional neurturin receptor. Neuron 22 (2), 243–252. [33] Pascual, A. et al. (2008) Absolute requirement of GDNF for adult catecholaminergic neuron survival. Nat. Neurosci. 11 (7), 755–761. [34] Kopra, J. et al. (2015) GDNF is not required for catecholaminergic neuron survival in vivo. Nat. Neurosci. 18 (3), 319–322. [35] Jain, S. et al. (2006) RET is dispensable for maintenance of midbrain dopaminergic neurons in adult mice. J. Neurosci. 26 (43), 11230–11238. [36] Kramer, E.R. et al. (2007) Absence of Ret signaling in mice causes progressive and late degeneration of the nigrostriatal system. PLoS Biol. 5 (3), e39. [37] Burke, R.E. (2006) GDNF as a candidate striatal target-derived neurotrophic factor for the development of substantia nigra dopamine neurons. J. Neural Transm. Suppl. 70, 41–45. [38] Oo, T.F., Kholodilov, N. and Burke, R.E. (2003) Regulation of natural cell death in dopaminergic neurons of the substantia nigra by striatal glial cell linederived neurotrophic factor in vivo. J. Neurosci. 23 (12), 5141–5148.

[39] Shridhar, V. et al. (1996) A gene from human chromosomal band 3p21.1 encodes a highly conserved arginine-rich protein and is mutated in renal cell carcinomas. Oncogene 12 (9), 1931–1939. [40] Shridhar, R. et al. (1996) Mutations in the arginine-rich protein gene, in lung, breast, and prostate cancers, and in squamous cell carcinoma of the head and neck. Cancer Res. 56 (24), 5576–5578. [41] Shridhar, V. et al. (1997) Mutations in the arginine-rich protein gene (ARP) in pancreatic cancer. Oncogene 14 (18), 2213–2216. [42] Henderson, M.J. et al. (2013) Mesencephalic astrocyte-derived neurotrophic factor (MANF) secretion and cell surface binding are modulated by KDEL receptors. J. Biol. Chem. 288 (6), 4209–4225. [43] Bruhn, H. (2005) A short guided tour through functional and structural features of saposin-like proteins. Biochem. J. 389 (Pt. 2), 249–257. [44] Lindstrom, R. et al. (2013) Characterization of the structural and functional determinants of MANF/CDNF in Drosophila in vivo model. PLoS One 8 (9), e73928. [45] Hoseki, J. et al. (2010) Solution structure and dynamics of mouse ARMET. FEBS Lett. 584 (8), 1536–1542. [46] Latge, C. et al. (2015) The solution structure and dynamics of full-length human cerebral dopamine neurotrophic factor and its neuroprotective role against alpha-synuclein oligomers. J. Biol. Chem.. [47] Wang, H. et al. (2014) Spatiotemporal expression of MANF in the developing rat brain. PLoS One 9 (2), e90433. [48] Kemppainen, S. et al. (2015) Cerebral dopamine neurotrophic factor improves long-term memory in APP/PS1 transgenic mice modeling Alzheimer’s disease as well as in wild-type mice. Behav. Brain Res. 291, 1–11. [49] Airavaara, M. et al. (2012) CDNF protects the nigrostriatal dopamine system and promotes recovery after MPTP treatment in mice. Cell Transplant. 21 (6), 1213–1223. [50] Yang, S. et al. (2014) Age-dependent decrease in chaperone activity impairs MANF expression, leading to Purkinje cell degeneration in inducible SCA17 mice. Neuron 81 (2), 349–365. [51] Tadimalla, A. et al. (2008) Mesencephalic astrocyte-derived neurotrophic factor is an ischemia-inducible secreted endoplasmic reticulum stress response protein in the heart. Circ. Res. 103 (11), 1249–1258. [52] Zhao, H. et al. (2013) Mesencephalic astrocyte-derived neurotrophic factor inhibits oxygen–glucose deprivation-induced cell damage and inflammation by suppressing endoplasmic reticulum stress in rat primary astrocytes. J. Mol. Neurosci. 51 (3), 671–678. [53] Apostolou, A. et al. (2008) Armet, a UPR-upregulated protein, inhibits cell proliferation and ER stress-induced cell death. Exp. Cell Res. 314 (13), 2454– 2467. [54] Mizobuchi, N. et al. (2007) ARMET is a soluble ER protein induced by the unfolded protein response via ERSE-II element. Cell Struct. Funct. 32 (1), 41– 50. [55] Walter, P. and Ron, D. (2011) The unfolded protein response: from stress pathway to homeostatic regulation. Science 334 (6059), 1081–1086. [56] Domanskyi, A., Saarma, M. and Airavaara, M. (2015) Prospects of neurotrophic factors for Parkinson’s disease: comparison of protein and gene therapy. Hum. Gene Ther.. [57] Huang, R.R. et al. (2015) Chronic restraint stress promotes learning and memory impairment due to enhanced neuronal endoplasmic reticulum stress in the frontal cortex and hippocampus in male mice. Int. J. Mol. Med. 35 (2), 553–559. [58] Chen, L. et al. (2015) Mesencephalic astrocyte-derived neurotrophic factor is involved in inflammation by negatively regulating the NF-kappaB pathway. Sci. Rep. 5, 8133. [59] Airavaara, M. et al. (2010) Widespread cortical expression of MANF by AAV serotype 7: localization and protection against ischemic brain injury. Exp. Neurol. 225 (1), 104–113. [60] Airavaara, M. et al. (2009) Mesencephalic astrocyte-derived neurotrophic factor reduces ischemic brain injury and promotes behavioral recovery in rats. J. Comp. Neurol. 515 (1), 116–124. [61] Yavarna, T. et al. (2015) High diagnostic yield of clinical exome sequencing in Middle Eastern patients with Mendelian disorders. Hum. Genet.. [62] Chen, Y.C. et al. (2012) MANF regulates dopaminergic neuron development in larval zebrafish. Dev. Biol. 370 (2), 237–249. [63] Nadella, R. et al. (2014) Transient transfection of human CDNF gene reduces the 6-hydroxydopamine-induced neuroinflammation in the rat substantia nigra. J. Neuroinflammation 11, 209. [64] Zhao, H. et al. (2014) Transplantation of cerebral dopamine neurotrophic factor transducted BMSCs in contusion spinal cord injury of rats: promotion of nerve regeneration by alleviating neuroinflammation. Mol. Neurobiol.. [65] Zhao, H. et al. (2014) Mechanisms of anti-inflammatory property of conserved dopamine neurotrophic factor: inhibition of JNK signaling in lipopolysaccharide-induced microglia. J. Mol. Neurosci. 52 (2), 186–192. [66] Back, S. et al. (2013) Gene therapy with AAV2-CDNF provides functional benefits in a rat model of Parkinson’s disease. Brain Behav. 3 (2), 75–88. [67] Ren, X. et al. (2013) AAV2-mediated striatum delivery of human CDNF prevents the deterioration of midbrain dopamine neurons in a 6hydroxydopamine induced Parkinsonian rat model. Exp. Neurol. 248, 148– 156. [68] Cordero-Llana, O. et al. (2015) Enhanced efficacy of the CDNF/MANF family by combined intranigral overexpression in the 6-OHDA rat model of Parkinson’s disease. Mol. Ther. 23 (2), 244–254.