J. Neurochem. (2011) 117, 1066–1074.
Neuregulin-1 (Nrg1) is genetically linked to schizophrenia, a disease caused by neurodevelopmental imbalance in dopaminergic function. The Nrg1 receptor ErbB4 is abundantly expressed on midbrain dopaminergic neurons. Nrg1 has been shown to penetrate blood-brain barrier, and peripherally administered Nrg1 activates ErbB4 and leads to a persistent hyperdopaminergic state in neonatal mice. These data prompted us to study the effect of peripheral administration of Nrg1 in the context of Parkinson’s disease, a neurodegenerative disorder affecting the dopaminergic system in the adult brain. We observed that systemic injections of the extracellular domain of Nrg1β1 (Nrg1β1-ECD) increased dopamine levels in the substantia nigra and striatum of adult mice. Nrg1β1-ECD injections also significantly protected the mouse nigrostriatal dopaminergic system morphologically and functionally against 6-hydroxydopamine-induced toxicity in vivo. Moreover, Nrg1β1-ECD also protected human dopaminergic neurons in vitro against 6-hydroxydopamine. In conclusion, we have identified Nrg1β1-ECD as a neurotrophic factor for adult mouse and human midbrain dopaminergic neurons with peripheral administratability, warranting further investigation as therapeutic option for Parkinson’s disease patients.
epidermal growth factor receptor
glial cell line-derived neurotrophic factor
Lund Human Mesencephalic
substantia nigra pars compacta
The neuregulin (Nrg) family of growth and differentiation factors plays a crucial role in development and plasticity of the nervous system (Buonanno and Fischbach 2001; Mei and Xiong 2008). It is transcribed from four genes (NRG-1 to NRG-4) and translated to diverse transmembrane and soluble isoforms. NRG-1 encodes at least 31 known Nrg1 isoforms, subclassified into the Types I–VI, with distinct time- and tissue-specific expression patterns. Its extracellular domain (ECD) is cleaved by transmembrane-proteases including β-amyloid converting enzyme-1 and released into the intercellular space, where it acts mainly as a paracrine trophic factor (Mei and Xiong 2008; Willem et al. 2009). The ECD contains an epidermal growth factor-like motif, which activates ErbB3 and ErbB4 receptors through dimerization and tyrosine phosphorylation. Both ErbB3 and ErbB4 transduce signals as tyrosine kinases (Buonanno and Fischbach 2001; Schrattenholz and Soskic, 2006; Mei and Xiong 2008; Birchmeier 2009).
Several lines of evidence suggest that Nrg1 affects dopamine (DA)-signaling. Nrg1 has been genetically linked to schizophrenia, a neurodevelopmental psychiatric disorder, in which altered DAergic neurotransmission has been implicated (Jaaro-Peled et al. 2009). Human and rodent midbrain DAergic neurons highly express ErbB4 throughout development into adulthood (Steiner et al. 1999; Thuret et al. 2004; Abe et al. 2009; Zheng et al. 2009). Interestingly, an N-terminally truncated ECD of human Nrg1β1 (nucleotides 46-634, 25.4 kDa) has been reported to pass the blood-brain barrier (BBB) in neonatal mice, to activate midbrain ErbB4 receptors, to increase the enzymatic activity of tyrosine hydroxylase (TH; the rate-limiting enzyme of DA biosynthesis) and to induce a persistent hyper-DAergic state (Kato et al. 2011). In this study (Kato et al. 2011), Nrg1β1-treatment coincided with the postnatal phase of ontogenic cell death (Oo and Burke 1997) and axonal differentiation (Graybiel 1984) of the mesencephalic DAergic system, suggesting that Nrg1β1 acts as neurotrophic factor during development. In line with this concept, glial growth factor-2, another member of the Nrg1-family of proteins, has been shown to be neuroprotecive for cultured fetal neurons (Zhang et al. 2004). In adult rodents, direct intracerebral infusion of the entire ECD (Ser2-Lys246, 26.9 kDa) of human Nrg1β1 into the hippocampus (Yurek et al. 2004) or of the epidermal growth factor-like domain only (Thr176-Lys246, 8 kDa) into the striatum (Kwon et al. 2008) increased local DA release, indicating that the DAergic system remains responsive to Nrg1β1 in adulthood.
As the adult DAergic neurons in the substantia nigra pars compacta (SNc) progressively degenerate in Parkinson’s disease (PD) (Dauer and Przedborski 2003), we studied here the Nrg1β1-ErbB4 system in the 6-hydroxydopamine (6-OHDA) model of PD in vivo and in vitro. Our results show that Nrg1β1-ECD can act as a neurotrophic factor for adult mouse midbrain DAergic neurons in vivo upon systemic administration. Moreover, we demonstrate that Nrg1β1-ECD also protects human dopaminergic neurons in vitro.
Materials and methods
Male wildtype C57/Bl6 mice (9–11 weeks old; Charles River, Sulzfeld, Germany), were kept at 23 ± 1°C under 12 h-dark/light conditions with ad libitum access to food and water and handled according to the EU Council Directive 86/609/EEC. All experiments were planned and executed to minimize the pain and discomfort for the animals used. All experiments were institutionally approved (Regierungspräsidium Giessen; MR20/15-Nr. 84/2007, 29/2008, 68/2009, 73/2009).
Human Nrg1β1-ECD (E. coli-derived, Ser2-Lys246, 26.9 kDa, GenBank accession no. AAA58639.1; R&D Systems, Minneapolis, MN, USA, no. 377-HB/CF) was dissolved at 10 ng/mL in 0.9% NaCl and injected i.p. into mice (50 ng/kg body weight). Control mice were injected with equal volumes of 0.9% NaCl. We used human Nrg1β1-ECD in both mouse and human experimental systems, because of the high amino acid sequence homology of both ligand and receptor between both species (Nrg1β1: 95.5%; ErbB4: 97%).
DA-measurement in vivo
Mice received five daily i.p.-injections of 50 ng/kg Nrg1β1-ECD and were perfused 1 h or 7 days later with 50 mL ice-cold 0.9% NaCl (n = 5 per group). Ventral midbrain, ventral striatum and dorsal striatum were micro-dissected and homogenized in 500 μL 0.4 M perchloric acid for 1 min. After centrifugation at 13 000 g at 4°C, the supernatants were passed through a 0.2 μm filter. DA levels were measured by reversed phase ion-pair HPLC, as described (Alvarez-Fischer et al. 2008).
DA-measurement in vitro
LUHMES (Lund Human Mesencephalic) neurons, derived of female human embryonic ventral mesencephalic cells by conditional immortalization (Tet-off v-myc over-expression) were differentiated for 6 days into post-mitotic neurons with a robust DAergic phenotype (Lotharius et al. 2005) and treated thereafter for 48 h with 0.01, 0.1 or 1 μg/mL Nrg1β1-ECD. Then, 300 μL 60% perchloric acid were added to the medium to denature potentially DA-metabolizing enzymes. Medium was removed and cells were treated with 100 μL Trypsin for 5 min. Then, 100 μL of 60% perchloric acid were added and cells scraped off. DA levels were measured on a Gynkotek HPLC with electrochemical detection (CC-4; Bioanalytical Systems, IN, USA) with a detection potential of 0.75 V. Samples were separated on a reversed phase column (Nucleosil 100-3 C18; Knauer, Berlin, Germany) using a mobile phase consisting of an aqueous solution with 0.65 mM 1-octanesulfonic acid sodium salt, 0.27 EDTA acid disodium salt (Na2-EDTA), 43 mM triethylamine and 0.67 M acetonitrile. The mobile phase was adjusted to pH = 3 with concentrated phosphoric acid. Each HPLC separation was carried out with a flow rate of 0.5 mL/min over 25 min.
Mice were injected i.p. with 50 mg/kg 5-bromo-2′-deoxyuridine (BrdU; Sigma, St Louis, MO, USA; 5 mg/mL in 0.9% NaCl) and 1 h later with either 50 ng/kg Nrg1β1-ECD or equal volumes of 0.9% NaCl (controls). These injections were repeated five times in 24-h intervals. Mice were killed 7 or 21 days later (n = 5 per group).
6-OHDA in vivo
Surgery was performed under anesthesia with 10 mL/kg of 1% ketamine (Bela-Pharm, Vechta, Germany) and 0.2% xylazine (Bayer HealthCare, Leverkusen, Germany). 6-OHDA (Sigma; 5 μg in 2 μL 0.9% NaCl with 0.2 μg/μL ascorbic acid) was injected slowly (0.5 μL/min) into the right striatum (0.9 mm anterior and 1.8 mm lateral from bregma, 3.0 mm ventral from the dura). After the injection, the syringe was kept for additional 3 min in the brain, before it was slowly retracted. Controls were vehicle-injected (sham-OP). Mice received eight times 50 ng/kg Nrg1β1-ECD in 24-h intervals, starting instantly (6 h) or delayed (48 h) after 6-OHDA-injection (Sham NaCl, n = 10; Sham Nrg1β1, n = 12; 6-OHDA NaCl, n = 10; 6-OHDA Nrg1β1instantly, n = 10; and 6-OHDA Nrg1β1delay, n = 11). The 8-day paradigm was used in the in vivo neuroprotection experiment (as compared to the 5 days in the DA-measurement experiment) in order to optimize the NRG treatment to overlap with the most substantial DA neurodegneration that occurs within the first 7 days after 6-OHDA injection (Alvarez-Fischer et al. 2008).
At 24 days after the 6-OHDA lesion, mice received an i.p. injection of 5 mg/kg d-amphetamine (Sigma). Rotational behavior was monitored for 30 min (Viewer II with rotational plug-in; Biobserve, Bonn, Germany). Data are expressed as net 360° turns per minute ipsilateral to the lesioned side.
Tissue preparation for immunohistochemistry
Mice were killed 28 days after the surgery with 100 mg/kg pentobarbital (Sigma) and perfused transcardially with 50 mL 0.9% NaCl followed by 100 mL 4% paraformaldehyde. Brains were removed, post-fixed for additional 48 h in paraformaldehyde, cryoprotected for 24 h in 30% sucrose/phosphate-buffered saline and frozen on dry ice. 35-μm thick cryotome sections were collected in 10 regularly spaced series and stored in a cryoprotection solution (1.57 g NaH2PO4, 5.18 g Na2HPO4, 400 mL H2O, 300 mL ethylene glycol, 300 mL glycerin) at −20°C.
6-OHDA in vitro
LUHMES neurons, derived of female human embryonic ventral mesencephalic cells by conditional immortalization (Tet-off v-myc over-expression) were differentiated into post-mitotic neurons with a robust DAergic phenotype (Lotharius et al. 2005). For protection analysis, they were treated with or without Nrg1β1-ECD (0, 0.01, 0.1 or 1 μg/mL medium) and intoxicated 1 h later for 48 h with 32 μM 6-OHDA.
Free-floating tissue sections or cultures were stained with the following primary antibodies: rabbit polyclonal anti-TH (P40101-0; Pel-Freez Biologicals, Rogers, AR, USA; 1/1000), rat monoclonal anti-dopamine transporter (MAB369; Chemicon International, Temecula, CA, USA; 1/1000), rat polyclonal anti-BrdU (OBT0030CX; Immunologicals Direct, Oxfordshire, UK; 1/500), and rabbit polyclonal anti-ErbB4 (sc-283; Santa Cruz Biotechnology, Santa Cruz, CA, USA; 1/200). Visualization of bound primary antibodies was done directly with fluorochrome-conjugated secondary antibodies or indirectly with biotinylated secondary antibodies using 3,3′-diaminobenzidine as substrate.
Stereological counts were done on every 5th serial section in the mouse SNc (−2.4 to −4.1 mm from bregma) using the optical fractionator method (Stereoinvestigator; MicroBrightField, Magdeburg, Germany). Optical density of TH+ fibers was quantified on eight equally spaced sections in the striatum (1.7 to −0.5 mm from bregma) under bright-field illumination (ImageJ v1.42 software; NIH, Bethesda, MD, USA). Double-immunofluorescence was analyzed by confocal microscopy (TCS SP5; Leitz, Wetzlar, Germany). Pyknotic 4′,6-diamidin-2-phenylindol (DAPI)+ nuclei in cultures, identified as round chromatin clumps of irregular size, were counted in 10 randomly distributed visual fields per culture well (DM-IRB; Leica, Wetzlar, Germany).
LDH release assay
Lactatdehydrogenase (LDH) levels in the culture medium were measured (n ≥ 12 per condition) using the CytoTox-ONE™ Assay (Promega, Mannheim, Germany).
Data are shown as mean ± SEM. Normal, parametric data were compared with the two-sided, unpaired t-test or anova followed by post-hoc Fisher least significant difference (LSD) test. p < 0.05 was considered significant.
Nrg1β1-ECD strengthens the DAergic system in healthy adult mice
We first studied the effects of peripheral administration of low doses of Nrg1β1-ECD (five times 50 ng/kg i.p. in 24-h intervals) on the nigrostriatal DAergic system in healthy adult mice. We found that the DA concentrations in the ventral midbrain and dorsal (‘motor’) striatum (Voorn et al. 2004) were unchanged immediately (day 0), but significantly elevated 7 days after Nrg1β1-ECD-treatment (+194.7% and +136.1%, respectively; p < 0.001 vs. NaCl-injected controls). The increase in the ventral (‘limbic’) striatum (nucleus accumbens and olfactory tubercle) (Voorn et al. 2004) was also significant (+63.8%; p < 0.05), albeit less pronounced (Fig. 1a).
Furthermore, we found that the total number of DAergic neurons in the SNc, identified by TH-immunostaining, was significantly increased at 21 days after five consecutive once daily i.p. injections of Nrg1β1-ECD (+16.7%; p < 0.001; Fig. 1b). Moreover, the number of large polygonal cresyl violet-stained neurons with the typical morphology of DAergic neurons in the SNc was increased after Nrg1β1-ECD-treatment (+21.5%; p < 0.001; Fig. 1c).
To analyze if the increase in the number of DAergic neurons in the SNc resulted from neurogenesis, we injected mice once daily with the thymidine analog BrdU to label mitotic cells, concomitantly with the 5-day Nrg1β1-ECD-treatment paradigm. We did not find BrdU and TH co-localization in any neuron in the SNc at 7 or 21 days after the last BrdU/Nrg1β1-ECD (Fig. 1d), arguing against Nrg1β1-ECD-induced neurogenesis. Also the absolute number of BrdU+ nuclei in the SNc did not increase at 7 or 21 days after the last BrdU/Nrg1β1-ECD-treatment as compared to 0.9% NaCl-treated controls (Fig. 1e), indicating that Nrg1β1-ECD is not mitogenic in the normal adult mouse SNc.
Nrg1β1-ECD protects mouse DAergic neurons against 6-OHDA in vivo
Next, we asked if Nrg1β1-ECD might be protective for DAergic neurons in a mouse model of PD. For this purpose, mice received a unilateral intrastriatal 6-OHDA injection or a sham-operation. They were then treated i.p. with 50 ng/kg Nrg1β1-ECD or vehicle (0.9% NaCl) for eight consecutive days in 24-h intervals. The Nrg1β1-ECD/vehicle treatment started either instantly (6 h) after 6-OHDA, when first oxidative stress is being generated, or with a delay (48 h), when severe axonal and mild neuronal loss had already occurred (Alvarez-Fischer et al. 2008).
Upon histological analysis 28 days after the 6-OHDA injection, the numbers of TH+ neurons (Fig. 2a and b) and cresyl violet (CV)+ large polygonal neurons (Fig. 2c) in the SNc were significantly reduced on the side of the 6-OHDA-lesion (6-OHDA NaCl group). This loss was attenuated in both Nrg1β1-ECD-treated groups (6-OHDA Nrg1β1 instantly and delay groups; Fig. 2a–c). Notably, the increases in the numbers of DAergic neurons in the SNc observed in the Nrg1β1-ECD-treated groups were not a mere consequence of the up-regulated cell number observed in healthy controls (Fig. 1b and c), because we expressed the cell numbers in the 6-OHDA-lesioned SNc (Fig. 2b and c) as percentage of the numbers of the contralateral intact SNc of each individual animal.
There was also a significant reduction in TH+ fiber density in the lesioned striatum as compared to the intact side (6-OHDA NaCl group; Fig. 2d and e). This loss of DAergic fibers was significantly mitigated in mice receiving Nrg1β1-ECD-treatment initiated instantly (6-OHDA Nrg1β1 instantly group). When Nrg1β1-ECD-treatment was initiated with delay at a time point, when severe axonal loss had already occurred (Alvarez-Fischer et al. 2008), it did expectedly not restore the lost DAergic fibers (6-OHDA Nrg1β1 delay group).
In addition, we examined amphetamine-induced turning behavior at 24 days after 6-OHDA injection, as a behavioral index of unilateral striatal DA deficiency (Iancu et al. 2005). Consistent with the histological findings of the striatal DAergic fibers, we found that the amphetamine-induced asymmetry was prevented when Nrg1β1-ECD-treatment was initiated after 6 h, but not when initiated with a 48-h delay (Fig. 2f).
Nrg1β1-ECD protects human DAergic neurons in vitro
Finally, we studied, whether Nrg1β1-ECD can also protect human post-mitotic mesencephalic DAergic LUHMES-neurons (Lotharius et al. 2005).
We observed that differentiated LUHMES-neurons indeed express the ErbB4 receptor (Fig. 3a). Treatment of differentiated LUHMES-neurons for 48 h with Nrg1β1-ECD led in a concentration-dependent manner to an increase in the DA content of the cells (0 μg Nrg1β1-ECD/mL: 100.0 ± 2.4%; 0.01 μg/mL: 110.6 ± 4.5%, not significant; 0.1 μg/mL: 124.2 ± 4.6%, p < 0.01; 1.0 μg/mL: 135.5 ± 1.6%, p < 0.001; anova, post hoc least significant difference test). Moreover, Nrg1β1-ECD significantly rescued cultured LUHMES-neurons against 6-OHDA-induced degeneration, as demonstrated microscopically by counting pyknotic nuclei (Fig. 3b). A concentration-dependent and almost complete protection was also confirmed biochemically using an LDH-release assay (Fig. 3c).
In the present study, we have shown that peripherally administered human Nrg1β1-ECD increases nigrostriatal DA levels and protects mesencephalic DAergic neurons in vivo against the neurotoxin 6-OHDA in mice. In addition, our data demonstrate that cultured human post-mitotic DAergic neurons are protected in a concentration-dependant manner by Nrg1β1-ECD against 6-OHDA toxicity. These observations indicate that the Nrg1β1-ECD may be an interesting candidate for both symptomatic and neuroprotective treatment of PD patients.
Current therapies of PD have clear limitations. They are primarily based on DA replacement, providing temporary improvement of motor deficits. Unfortunately, patients typically develop drug-induced motor complications (dyskinesia and on-off-fluctuations), and no presently available therapy halts the disease progression in a clinically relevant manner (Dauer and Przedborski 2003). Previous approaches to protect DAergic neurons in PD from dying with neurotrophic factors have been compromised. The glial cell line-derived neurotrophic factor (GDNF) for example, one of the best studied compounds of this group, potently protects midbrain DAergic neurons from a variety of toxic insults (Lin et al. 1993). However, GDNF does not cross the BBB, and therefore cannot be administered peripherally. Thus, several ways to circumvent the BBB have been studied (Kordower et al. 2000; Gill et al. 2003; Behrstock et al. 2006), but clinically relevant efficacy has not yet been achieved (Lang et al. 2006). Similar limitations also apply to other known neurotrophic factors (Thoenen and Sendtner 2002).
In contrast, we have demonstrated that peripheral administration of human Nrg1β1-ECD increases striatal DA concentrations in healthy adult mice. This effect was more pronounced in the dorsal (‘motor’) striatum, which receives DAergic afferents from the SNc, than in the ventral (‘limbic’) striatum, which is innervated by DAergic neurons located in the ventral tegmental area. This is consistent with previous reports on higher ErbB4 expression in the SNc as compared to the ventral tegmental area (Thuret et al. 2004; Abe et al. 2009). We observed increased DA levels after 7 days, but not immediately after Nrg1β1-ECD-treatment, suggesting that structural plasticity rather than acute changes underlies this phenomenon. Remarkably, Nrg1β1-ECD treatment also increased the number of DAergic neurons in the SNc. As Nrg1β1-ECD did not induce cell proliferation and neurogenesis in the adult SNc, the increase in number of DAergic neurons appears to result from an induction of a DAergic phenotype in pre-existing cells similar as previously described for striatal DAergic neurons (Tandéet al. 2006). In this regard, it is interesting to note that we observed a subpopulation of ErbB4+ cells that did not contain TH in the SNc in mice and humans. These cells might respond to Nrg1β1-ECD by induction of a DAergic phenotype.
We also studied, whether Nrg1β1-ECD can protect DAergic neurons against 6-OHDA, a neurotoxin leading to impaired mitochondrial energy production, protein mishandling, oxidative stress and excitotoxicity (Blum et al. 2001; Dauer and Przedborski 2003). We used a partial and subacute lesion paradigm with intrastriatal 6-OHDA injection in mice, which allows oxidative stress and structural damage to be temporally separated (Alvarez-Fischer et al. 2008). Indeed, we observed that Nrg1β1-ECD markedly reduced the 6-OHDA-induced loss of neurons in the SNc and their terminals in the striatum, resulting in improved motor dysfunction in the amphetamine-induced rotation test, a test which shows robust correlation with the imbalance in DA transmission in the rodent 6-OHDA lesioned PD model. The effect was pronounced, when Nrg1β1-ECD was administered early after intoxication (i.e. when oxidative stress, but no structural damage was present (Alvarez-Fischer et al. 2008). When Nrg1β1-ECD was administered later (i.e. when pronounced axonal damage was already established (Alvarez-Fischer et al. 2008), it did expectedly not revert the already established loss of DAergic terminals in the striatum and the corresponding motor asymmetry in the experimental period. Nevertheless, delayed administration of Nrg1β1-ECD still protected neurons in the SNc from retrograde degeneration. Importantly, however, although Nrg1β1-ECD treatment showed positive effects on drug-induced rotation, it is crucial to evaluate its effect on behavior in more complex tests, such as the cylinder test, stepping test and corridor test (Grealish et al. 2010), to clearly understand its functional neuroprotective effect. In addition, as over-expression of NRG-1 has shown to induce some schizophrenia-like symptoms in transgenic mice models (Kato et al. 2010), it is also needed to evaluate possible side effects of the Nrg1β1-ECD treatment, such as hyperlocomotion, anxiety and disturbed social behavior. Both the beneficial as well as possible adverse effect should be addressed in follow-up studies. Thus, the neuroprotective potency of Nrg1β1-ECD appears to be comparable to GDNF (Lin et al. 1993), cerebral DA neurotrophic factor (Lindholm et al. 2007) and mesencephalic astrocyte-derived neurotrophic factor (Voutilainen et al. 2009), which are presently the only known neurotrophic factors with efficacy in the severe 6-OHDA in vivo model of PD. However, Nrg1β1-ECD has the important advantage that it does not depend on intracerebral infusion, but can be administered peripherally (Kastin et al. 2004; Kato et al. 2011).
Finally, we demonstrated that Nrg1β1-ECD also protects human mesencephalic DAergic neurons from 6-OHDA-induced degeneration in vitro, which suggests that Nrg1β1-ECD-treatment might also be effective in PD patients.
The precise molecular mechanism, how Nrg1β1-ECD provides neuroprotection downstream of the activation of its receptor is currently not clear. Unpublished proteomic data indicate that Nrg1β1-ECD treatment increases several neuronal proteins implicated in defense mechanisms against impaired mitochondrial energy production, protein mishandling, oxidative stress and excitotoxicity. Further evaluations of these proteins are ongoing and may provide insights into the molecular mechanism of action of Nrg1β1-ECD (Höglinger and Schrattenholz unpublished). Cautiously it has to be kept in mind, however, that the precise mode of action has not yet been elucidated for other neurotrophic factors such as GDNF although their biological efficacy has long been demonstrated.
In conclusion, Nrg1β1-ECD treatment may presumably have a benefit in PD. Foremost, it has been shown that the ErbB4 expression is relatively maintained in PD patients (Iwakura et al. 2005). Second, as our data shown the increase in endogenous DA production in the motor striatum may provide symptomatic relief and postpone the time until DAergic drugs are required, which are known to induce unwanted side effects in human PD patients. Third, the induction of a DAergic phenotype in pre-existing non-DAergic neurons in the SNc may suggest a regenerative therapeutic option in PD. Fourth, a protection of the endogenous DAergic neurons in the SNc from degeneration may slow or even halt the progression of the disease. Therefore, further basic research into safety and efficacy of Nrg1β1-ECD and subsequent clinical trials are highly warranted. In fact, systemic Nrg1 injection has recently been reported in a clinical trial, aimed towards treatment of chronic heart failure, which indicates that the peripheral administration of Nrg1 may be safe for applications in humans (Jabbour et al. 2011).
We thank Sabine Anfimov, Aynur Cabuk, Silke Caspari and Susanne Stei for technical assistance. We also thank Marcel Leist, Konstanz University for supplying LUHMES cells. This work was supported by the German Ministry of Research and Technology (BMBF-Biochance4 0315123 to AS), German National Genome Research Network (01GS08136-4 to GUH), [European Community’s] Seventh Framework Programme (FP7/2007-2013) under grant agreement nr. 220656 (to TC) and the University Clinics Giessen and Marburg (UKGM) (to CD). Conflict of interest: ProteoSys holds a patent on the use of Nrg1β1-ECD in neurodegenerative disorders. The other authors declare that they have no competing financial interests.