Midkine and Alk signaling in sympathetic neuron proliferation and neuroblastoma predisposition

Neurogenesis in sympathetic ganglia (SG) proceeds mainly by proliferation of immature but already differentiated neurons (DiCicco-Bloom et al., 1990; Rohrer and Thoenen, 1987; Rothman et al., 1978; Tsarovina et al., 2008). In contrast to other lineages in the peripheral and central nervous system (PNS and CNS), neuron differentiation is not linked to cell cycle withdrawal. Cell cycle exit and generation of postmitotic sympathetic neurons proceeds over a broad developmental period but the cellular and molecular mechanisms controlling proliferation in this lineage are not well understood.

NB, the major tumor of the developing PNS is restricted to the sympathoadrenal lineage, suggesting that predisposition and development of NB may be linked to the proliferation characteristics of sympathetic neuroblasts (Brodeur, 2003; Maris et al., 2007). Proliferation of precursors and immature sympathetic neurons are affected by the transcription factors Ascl1, Phox2b, Hand2, Insm1, Sox11 and Gata2/3 that were previously considered to control only initial specification and differentiation (Coppola et al., 2010; Hendershot et al., 2008; Morikawa et al., 2009; Pattyn et al., 1999; Potzner et al., 2010; Tsarovina et al., 2010; Wildner et al., 2008). Mutations in the PHOX2B gene were identified in familial cases of NB and result in tumor predisposition (McConville et al., 2006; Mosse et al., 2004; Trochet et al., 2004; van Limpt et al., 2004). NB PHOX2B mutations affect proliferation and differentiation of embryonic sympathetic neurons and the proliferation of NB cell lines (Raabe et al., 2008; Reiff et al., 2010). More recently, predisposing ALK mutations were identified in a large fraction of familial cases of NB (Janoueix-Lerosey et al., 2008; Mosse et al., 2008). In addition, in sporadic cases of NB, a relatively high frequency of ALK mutations (6-10%) and ALK amplifications (1.7%) was observed (De Brouwer et al., 2010; Janoueix-Lerosey et al., 2010; Mosse et al., 2008; Schulte et al., 2011). Mutational hotspots in ALK are located in the tyrosine kinase domain affecting the residues F1174 and R1275, leading to constitutively active mutant proteins independent of ligand binding (Chen et al., 2008; George et al., 2008; Janoueix-Lerosey et al., 2008; Mosse et al., 2008). Constitutively active ALK receptors stimulate cell proliferation, as shown by ectopic expression of F1174 and R1275 ALK receptor in 3T3 fibroblasts (Chen et al., 2008) and murine lymphoid Ba/F3 cells (George et al., 2008). Importantly, in NB cell lines that express mutated ALK, shRNA-mediated ALK knockdown and pharmacological kinase inhibition lead to strongly reduced proliferation and increased cell death (Chen et al., 2008; George et al., 2008; Janoueix-Lerosey et al., 2008; Mosse et al., 2008). ALK-dependent proliferation of NB cell lines with wild-type ALK implicates the presence of an activating ligand(s) (Janoueix-Lerosey et al., 2008; Mosse et al., 2008). Two potential Alk ligands, the growth factors Midkine (Mk) and Pleiotrophin (Ptn) have been identified with widespread functions in proliferation, differentiation and cell survival (Kadomatsu and Muramatsu, 2004; Muramatsu, 2010). Although Alk expression has been documented in developing SG (Hurley et al., 2006; Morris et al., 1997; Vernersson et al., 2006), the physiological role of Alk signaling in the sympathetic neuron lineage, as well as in other parts of the developing nervous system, is not well understood (Palmer et al., 2009).

In this study, an essential role for Alk signaling in sympathetic neuron proliferation was observed in culture and confirmed by Alk knockdown in vivo. The Alk ligand Mk was detected in SG and shown to be sufficient and required for neuron proliferation, which implicates Mk as physiological ligand in this lineage. Activation of Alk signaling, in particular overexpression of ALK^wt and NB ALK mutants resulted in the upregulation of NMyc and trkB, which may contribute to continued proliferation and NB predisposition.

PcDNA3.1 expression plasmids for human ALK^wt, ALK^F1174L and ALK^R1275Q, and pCAGGS expression plasmids for Hand2, Phox2b^wt and Phox2b^K155X have been described previously (Janoueix-Lerosey et al., 2008; Reiff et al., 2010).

The shRNA against Gallus gallus Alk and Mk were designed using BLOCK-iT RNAi Designer (Invitrogen, Karlsruhe, Germany) and target the following sequences: shAlk, 5′-AAU GGU UUC UCU CUA UGU CCA ACU C-3′; shMk, 5′-GAG CUG ACU GCA AGU ACA AGU UUG A-3′. Scrambled shRNAs (sc-shRNA, Invitrogen, Karlsruhe, Germany) served as controls.

Non-radioactive in situ hybridization on cryosections and preparation of digoxigenin labeled probes for chick Phox2b was carried out as described previously (Ernsberger et al., 1997; Stanke et al., 1999).

Equal amounts of RNA were used to synthesize cDNA with Oligo(dT)-primers and Superscript-III-reverse-transcriptase according to manufacturer’s instructions (Invitrogen, Karlsruhe, Germany). The PCR was carried out using Abgene’s Absolute Blue SYBR-Green qPCR Mix (Abgene, Epsom, UK) in a Stratagene Mx3000p Light Cycler (Stratagene, Waldbronn, Germany). All primers (MWG Biotech AG, Ebersberg, Germany) were designed to anneal optimally at 58°C with PerlPrimer (Marshall, 2004) (95°C for 30 seconds, 58°C for 30 seconds, 72°C for 30 seconds, repeated for 50 cycles and a subsequent dissociation curve). The primer pairs (see Table S1 in the supplementary material) were analyzed for efficiency (≥95%) and used for quantitative analysis. At least duplicates of every condition were performed in parallel. Data were normalized using Gapdh and Islet1 as reference genes and evaluated by delta-delta C_t-method using Microsoft Excel. Experiments were repeated independently six times and statistically analyzed using paired two-tailed Student’s t-test and ANOVA.

Paravertebral lumbosacral sympathetic ganglia from E7 chicken embryos were dissociated (Rohrer and Thoenen, 1987; Zackenfels et al., 1995) and the Amaxa Nucleofector II was used to electroporate 200,000 cells [Program: Small Cell Number (SCN) #2, transfection efficiency ≈50%]. pMAX-GFP (0.5 μg) (Lonza, Cologne, Germany) was mixed with 0.5 μg of empty vector/sc-shRNA for controls or 0.5 μg of the appropriate pcDNA3.1-Alk variants or 0.5 μg of shAlk and shMk. Transfected cells were plated on poly-DL-ornithin/laminin-coated four-well culture dishes and cultured for 2 days as described (Reiff et al., 2010). NVP-TAE684 (Axon Medchem, Groningen, The Netherlands) was used to inhibit Alk kinase. Human recombinant MK (10 nM) (Biomol, Hamburg, Germany) was added to cultures where indicated. EdU was added to the culture medium 24 hours before fixation (1:1000, Invitrogen, Karlsruhe, Germany). For RNA isolation, 400,000 cells were electroporated, cultured for 2 days in vitro and harvested. RNA was isolated using the mRNEasy Mini Kit (Qiagen, Hilden, Germany).

Cells from dissociated E7 SG were separated from cellular debris by centrifugation through a step gradient with 3% BSA in DMEM medium (20 minutes, 150 g). All centrifugation and incubation steps were carried out at 12°C. After incubation with mouse anti-Q211 antibodies (Zackenfels et al., 1995) (1:50, 15 minutes), unbound antibody was removed by dilution with 5 ml DMEM medium and the cells were pelleted by centrifugation through a 3% BSA step gradient. Resuspended cells were incubated with microbead-coupled anti-IgM (1:5, 15 minutes) (Miltenyi Biotec, Bergisch Gladbach, Germany), separated from unbound anti-IgM by BSA gradient centrifugation. Magnetically labeled cells were applied to MACS SM column separator (Miltenyi Biotec, Bergisch Gladbach, Germany). Unbound cells were eluted by three washing steps, magnetically labeled Q211⁺ cells were eluted after removal of the column from the magnetic MACS separator using 3×500 μl of 0.1% BSA in DMEM medium. Both flow-through fraction and the magnetically labeled Q211⁺ fraction underwent a second cycle of MACS column purification. The final cell suspensions were pelleted by centrifugation for RNA isolation and qPCR analysis. Before pelleting, aliquots of cell suspensions were plated onto poly-DL-ornithin/laminin-coated dishes and analyzed after short-term (3 hours) culturing by staining for Q211, TuJ1, TH, O4 and DAPI for the proportion of neurons, glial cells and fibroblast-like cells (DAPI⁺, Q211^–, TuJ1^–, TH^–, O4^–).

Cells were washed with PBS and fixed using 4% paraformaldehyde in 0.1 M sodium phosphate buffer for 15 minutes. After washing with PBS, cells were stained with the EdU labeling kit according to the manufacturer’s protocol (Invitrogen, Karlsruhe, Germany, Click-It EdU Imaging Kit 594). Afterwards, primary antibodies were diluted 1:500 in staining buffer (PBS, 0.5% Triton X100 and 5% FCS) and incubated overnight (mouse, Th). After washing with PBS, appropriate secondary antibodies were used (1:500, goat anti-mouse Alexa 488, Invitrogen, Karlsruhe, Germany). Nuclei were stained with DAPI (Sanofi Aventis, Frankfurt, Germany). Stainings were covered with Aqua Polymount (Polysciences, Eppelheim, Germany) and analyzed at room temperature with a Zeiss Axiophot 2 microscope (Plan Neofluar Objective 40×, 0.75 Ph2). Transfected cells were identified by GFP fluorescence, whereas TH immunoreactivity served as a marker for immature noradrenergic neurons in non-transfected cells. GFP- or TH-positive neurons were analyzed for EdU staining and double-positive cells were counted in at least 20 visual fields per culture dish at a magnification of 40×. A Visitron Systems Spot RT3 camera was used for image acquisition and MetaVue (7.1.3.0) for digital image processing (adjustment of brightness and contrast). All results are given as mean±s.e.m. of six independent experiments and statistically analyzed with paired two-tailed Student’s t-test.

TUNEL-positive neurons were detected on sympathetic neuron cultures with the in situ cell death detection kit (Roche Applied Sciences, Grenzach, Germany) according to the manufacturer’s protocol followed by a DAB-reaction (Thermo Fisher, Schwerte, Germany). TUNEL staining on cryosections of chicken embryos was performed as described previously (Tsarovina et al., 2010). Total neuron number and TUNEL-positive neurons were counted in 20 visual fields per culture dish at a magnification of 40×. All results are given as mean±s.e.m. from six independent experiments and statistically analyzed with paired two-tailed Student’s t-test.

Short hairpin RNAi viruses for specific knockdown of Mk and Alk were constructed as described previously (Das et al., 2006) with the following primer pairs (flanking primers W,Y) (see Das et al., 2006): shAlk forward, GAG AGG TGC TGC TGA GCG CCA CAA AGA AGA AAG TTG ACA GGT GAA GCC ACA GAT GTAAAG GGA GGG TGT GCG ATG; shAlk reverse, ATT CAC CAC CAC TAG GCA ACA CAA AGA AGA AAG TTG ACA GCA TCT GTG GCT TCA CT; shMk forward, GAG AGG TGC TGC TGA GCG CCA AAC TTG TAC TTG CAG TCA GCT CCG TGA AGC CAC AGA TGT A; shMk reverse, ATT CAC CAC CAC TAG GCA ACA AAC TTG TAC TTG CAG TCA GCT CCC ATC TGT GGC TTC ACT. Underlined sequence hybridizes with respective mRNA.

Gallus gallus Mk was amplified by PCR from E7 SG cDNA using the following primer pairs that add an additional Kozak sequence and ClaI (forward, TAATCGATACCACCATGCAGCCCCGGGGCCTCCTCCTC) and SpeI (reverse, TAACTAGTCTAGTCCTTCCCCTTGCCCTTCTTTGC) restriction sites for RCAS cloning. DF1 chicken fibroblasts were transiently transfected with empty RCAS(BP-B), RCAS(BP-B)-Mk, RCAS(BP-B)-shMk and -shAlk (RCAS-shMk/RCAS-shAlk), and virus concentrate was prepared as described previously (Rüdiger et al., 2009).

Fertilized virus-free eggs (Charles River, Sulzfeld, Germany) were incubated and infected with empty RCAS virus for control or RCAS-shMk, RCAS-shAlk or RCAS-Mk virus concentrates as described (Rüdiger et al., 2009). Embryos were kept until stage 24 (E4) or E7, pulsed with EdU (400 μl of 10 mM EdU in PBS) for 3 hours, fixed in 4% paraformaldehyde and embedded. Cryosections (16 μm) were stained for EdU and TH, and analyzed with a Zeiss Apotome fluorescence microscope. The TH⁺ area from infected SG was quantified with Axiovision 4.6 (Zeiss Apotome, 40× magnification). In situ hybridization for Phox2b was carried out as described previously (Rüdiger et al., 2009). Images of Phox2b⁺-area were acquired by light microscopy (Zeiss Axiophot 2, 20× magnification) and analyzed with MetaVue 7.1.3.0 as described previously (Lucas et al., 2006; Tsarovina et al., 2004).

The function of Alk in normal sympathetic neuron development, which is implicated by the selective susceptibility of the sympathoadrenergic lineage to increased Alk signaling (Janoueix-Lerosey et al., 2010; Palmer et al., 2009) has remained elusive. Here, we study the role of Alk in cultures of proliferating immature neurons from embryonic (E7) chick SG, which contain proliferating cells virtually all of which harbor characteristic pan-neuronal and adrenergic markers and respond to proliferation stimulatory as well as antiproliferative signals (Reiff et al., 2010; Rohrer and Thoenen, 1987; Zackenfels et al., 1995). Glial cells and fibroblasts represent only 3±2% and 2±1% of E7 chick SG cells, respectively (Ernsberger et al., 1989). To investigate the effect of NB ALK mutations and mimic ALK locus amplification, cells were transfected with expression plasmids for two human NB ALK mutations (ALK^F1174L and ALK^R1275Q) and ALK^wt. Similar expression levels of ALK^wt and ALK variants were observed by quantitative PCR (qPCR) (data not shown). To pursue proliferation of transfected immature sympathetic neurons, an expression plasmid for GFP was co-transfected and the number of GFP⁺ neurons that are positive for the proliferation marker EdU was counted. Overexpression of all three ALK variants led to a strong increase in the proportion of EdU⁺ neurons compared with control (Fig. 1). To determine whether Alk is essential for basal proliferation in vitro, the effects of pharmacological Alk inhibition and shRNA-mediated knockdown of endogenous Alk were studied.

NVP-TAE684 is a potent and selective inhibitor of Alk signaling, which interacts specifically with the ATP-binding site (Galkin et al., 2007). In the presence of NVP-TAE684, sympathetic neuron proliferation was completely blocked, with an IC₅₀ of 67.6±1.8 nM (Fig. 2A). As NVP-TAE684 also blocks insulin-like growth factor receptor (IGFR) at high concentrations (IC₅₀ 1.2 μM), it was unclear whether the proliferation effects were entirely due to interference with Alk signaling (Galkin et al., 2007; Zackenfels et al., 1995). Thus, an Alk-specific shRNA-mediated knockdown was performed. A highly efficient shRNA (shAlk) was selected that resulted in a significant 71.3±3.8% (n=6) reduction in endogenous Alk mRNA levels 2 days after transfecting E7 sympathetic neurons. In shAlk-transfected sympathetic neuron cultures, a strong decrease in the number of proliferating EdU⁺ cells was observed (Fig. 2B). As the proportion of transfected cells was not significantly altered (81.4±5.9% compared with control), side effects of Alk knockdown on cell survival are excluded. The anti-proliferative effects of shAlk and NVP-TAE684 were completely rescued by simultaneous overexpression of human ALK^wt and the ALK variants (Fig. 2B,C), excluding off-target effects of shAlk. The finding that the overexpression of the ligand-dependent ALK^wt rescues Alk inhibition and knockdown (Fig. 2B,C) and increases the number of proliferating cells (Fig. 1) suggested that a receptor-activating ligand is present in SG during development.

Mk and Ptn have previously been identified as Alk ligands that are able to elicit phosphorylation of Alk and the downstream kinase PI3K (Stoica et al., 2001; Stoica et al., 2002), although the role of Ptn as Alk ligand is controversial (Mathivet et al., 2007). To determine whether these ligands are expressed in developing SG, we performed qPCR for Mk and Ptn mRNA on cDNA of E7 SG cultures. The analysis revealed a strong Mk expression, whereas Ptn is present at a much lower level (46-fold; n=6). Mk levels in SG decrease from E6 until E12, whereas Ptn expression decreases until E10 and is subsequently upregulated to reach similar levels as Mk at E12 (Fig. 3A). Alk expression levels in SG are not changed significantly (Fig. 3B), which is in agreement with a previous analysis by in situ hybridization (Hurley et al., 2006).

To address the issue of whether Mk is derived from the neuronal or non-neuronal SG population, E7 SG were dissociated and neurons were immunopurified using the neuron-specific Q211 antibody (Rohrer et al., 1985; Rohrer and Thoenen, 1987; Zackenfels et al., 1995). MACS sorting resulted in a virtually pure neuronal population (98.42±0.28% Q211⁺, 98.7±0.56% Tuj1⁺; n=3) and a neuron-depleted flow-through fraction with >35-fold higher proportion of non-neuronal cells (27±19% Q211⁺, 25±14% Tuj1⁺, 37±10% O4⁺; n=3). Analysis of Mk expression by qPCR revealed similar levels in the neuronal and neuron-depleted population [0.96-fold difference (± 0.08); s.e.m.; n=3]. Thus, Mk is expressed in both neuronal and non-neuronal cells. By contrast, Ptn is preferentially expressed [87.6-fold enriched (±30), s.e.m.; P<0.05; n=3] in the neuron-depleted population, i.e. in ganglion non-neuronal cells.

The selective NB development in SG when compared with parasympathetic ganglia suggests that proliferation of parasympathetic neuron progenitors is not regulated by Alk signaling. Indeed, much lower Alk expression levels (8.3±0.1-fold, s.e.m.; P<0.008; n=3) were detected by qPCR analysis in ciliary ganglia (E5) when compared with sympathetic ganglia (E7). This is in agreement with the differential Alk expression observed in a SAGE screen comparing E7 sympathetic and E5 ciliary ganglia (J. Stubbusch and H.R., unpublished). Importantly, neither Mk nor Ptn was detected by qPCR during neurogenesis in E5 ciliary ganglia (n=3).

Alk is expressed in SG not only during neurogenesis (Hurley et al., 2006) (present data) but up to stages where neuron proliferation has finished (Rothman et al., 1978). The timing of Mk expression suggests that sympathetic neuron proliferation is controlled by Mk and that neurogenesis may be terminated by decreased Mk availability. To address this issue, we investigated whether Mk is responsible for Alk activation and basal proliferation of cultured sympathetic neurons.

Mk significantly increased the proportion of proliferating, EdU⁺/TH⁺ neurons when applied to SG cultures (Fig. 4A). The moderate effect of exogenous MK may be explained by the high expression level of endogenous Mk that already stimulates sympathetic neurons close to the maximal proliferation rate (Reiff et al., 2010). The proliferation increase of sympathetic neurons in response to MK is also maintained at later stages when neurogenesis and endogenous Mk levels decrease (Fig. 4A), and is mediated by Alk, as revealed by Alk knockdown (Fig. 4B).

To address the role of endogenous Mk, a shRNA knockdown strategy was chosen. Knockdown of endogenous Mk mRNA levels by shMk to 27.4±2.9% of control levels resulted in a strong decrease of proliferating immature neurons (Fig. 4B) without affecting cell survival (data not shown). The specificity of the shMk-mediated knockdown was demonstrated by the rescue through the addition of recombinant MK to the cell cultures (Fig. 4B). The Mk knockdown effect on proliferation was also rescued by overexpression of ALK^F1174L but not by ALK^wt. Thus, endogenous Mk is required for proliferation effects of transfected ALK^wt but not for increased proliferation by constitutively active ALK^F1174L (Fig. 4B,C). These findings show that Mk-dependent Alk signaling is essential for in vitro proliferation of immature sympathetic neurons and raise the question of to what extent sympathetic neurogenesis is controlled by Mk/Alk in the embryo.

To investigate whether Alk signaling is involved in sympathetic neurogenesis in the embryo, an in vivo approach to knock down endogenous Mk and Alk was chosen. Premigratory neural crest cells were infected with high titers of RCAS-virus concentrates that target either Mk or Alk by specific shRNAs. Infection efficiency was monitored by in situ hybridization against the viral reverse transcriptase mRNA and showed no difference between RCAS controls, RCAS-shAlk, RCAS-Mk and RCAS-shMk (data not shown). The analysis of RCAS-shAlk- and RCAS-shMk-infected embryos revealed a massive reduction in SG size when measured by the area occupied by TH⁺ and Phox2b⁺ cells (Fig. 5A,B; see Fig. S1 in the supplementary material). Furthermore, sympathetic neuron proliferation is severely affected, as shown by the reduced percentage of EdU⁺/DAPI⁺ cells in the TH⁺ ganglion area (Fig. 5A,C). By contrast, dorsal root ganglion (DRG) size (area of TuJ1⁺ cells) and the proportion of EdU⁺ cells in the DRG were not altered (see Fig. S2 in the supplementary material), although Alk is transiently expressed in developing DRG (Hurley et al., 2006). To analyze effects on cell survival, TUNEL-staining was performed on sections of shAlk- and shMk-infected embryos. As the number of TUNEL⁺ cells in shAlk- and shMk-infected embryos (0.5±0.2 and 0.6±0.2 TUNEL⁺ cells/1000 μm² TH⁺ area, respectively) was not significantly increased above RCAS-infected controls (0.3±0.1 TUNEL⁺ cells/1000 μm² TH⁺ area) cell death is excluded as cause of reduced SG size. Ectopic Mk expression significantly increased E4 SG size, as well as the proportion of EdU⁺ cells (Fig. 5B). Mk-overexpressing embryos could be maintained up to E7 but were growth retarded, most probably owing to impaired development of chorioallantoic blood vessels (not shown), which is in line with Mk functions in angiogenesis (van der Horst et al., 2008). The percentage of EdU⁺ cells in E7 Mk expressing embryos was significantly increased when compared with E7 control infected embryos (16.5±1.98% versus 9.7±0.5%; n=3, P>0.01). However, increased proliferation was not paralleled by increased ganglion size [9001±438 μm² in controls when compared with 6858±832 μm² in Mk infections (n=3; P>0.05 n.s.)]. The higher proportion of EdU⁺ cell in Mk-expressing growth-retarded embryos may thus be explained by increased proliferation in less mature ganglia, rather by a direct proliferation effect.

The knockdown of Alk and Mk seems to be compensated with time in vivo, as proliferation is no more reduced at E7 and ganglion size is partially recovered (see Fig. S3 in the supplementary material). Having confirmed the importance of Alk signaling for sympathetic precursor proliferation in vivo, the mechanisms that underlie these effects were studied.

To identify potential downstream target genes of Alk signaling, E7 sympathetic neurons were treated by MK (10 nM) or transfected with shMk and subsequently analyzed for expression levels of candidate target genes by qPCR. The expression of the differentiation markers DBH, TH, Phox2b, Gata2/3, trkA and trkC was not significantly changed by Mk (data not shown). Interestingly, Alk expression was affected both by increasing and decreasing Mk levels, which indicates a positive autoregulatory control (Fig. 6A). A similar Mk-dependent expression was observed for the transcription factor Hand2 (Fig. 6A). This finding, together with the previous demonstration that Hand2 controls proliferation of sympathoadrenergic precursors and immature neurons in vitro and in vivo (Hendershot et al., 2008; Reiff et al., 2010; Schmidt et al., 2009) and the presence of Hand2 expression in sympathetic ganglia (Fig. 6B), implicated Hand2 in the regulation of Alk expression. It also suggested that the effects of Hand2 on sympathetic neuron proliferation may be caused indirectly by controlling Alk expression levels. Indeed, we now demonstrate that forced Hand2 expression leads to significantly increased Alk expression (Fig. 6A) and that shAlk knockdown strongly reduces the proliferation stimulatory effects of Hand2 overexpression (see Fig. S4 in the supplementary material). By contrast, cell cycle genes Cdk6 and Ccnblip identified as Hand2 target genes in developing heart tissue (Holler et al., 2010) were not affected in sympathetic neurons (not shown). As Phox2b has been implicated, together with Hand2, in the control of sympathetic neuron proliferation (Bachetti et al., 2010; Coppola et al., 2010; Reiff et al., 2010) we investigated whether wild-type Phox2b (Phox2b^wt) or a NB-related Phox2b variant (Phox2b^K155X), ectopically expressed in cultured immature sympathetic neurons, would affect Alk expression (Reiff et al., 2010). However, we were unable to detect effects of Phox2b^wt and Phox2b^K155X on Alk expression levels by qPCR (data not shown).

Forced Hand2 expression also increased the expression of two additional proliferation regulators in sympathetic neurogenesis, trkB and NMyc. Activation of trkB through its ligand BDNF, as well as forced NMyc expression, lead to a strong increase in immature sympathetic neuron proliferation (Reiff et al., 2010; Straub et al., 2007). As the effects of Hand2 on sympathetic neuron proliferation are due to increased Alk expression, we investigated whether overexpression of ALK^wt and NB ALK variants may also lead to increased NMyc and trkB expression. Indeed, trkB and NMyc expression levels were elevated about twofold upon transfection of sympathetic neurons with expression vectors for ALK^wt, ALK^F1174L and ALK^R1275Q (Fig. 6C), taking the transfection efficiency (about 50%) into account.

The present work demonstrates that Alk signaling, stimulated by its ligand Mk, is essential for the proliferation of immature sympathetic neurons. Increased Alk receptor signaling is sufficient to enhance sympathetic neuron proliferation, which is accompanied by the upregulation of Alk, NMyc and trkB. Our findings uncover a novel mechanism that controls sympathetic neuron proliferation in vivo and in vitro, and provide insight into how activating ALK mutations may circumvent the termination of neurogenesis and predispose to NB.

Neurogenesis in SG differs from other lineages, i.e. immature but already differentiated neurons proliferate (Rohrer and Thoenen, 1987; Rothman et al., 1978; Tsarovina et al., 2008). Thus, the expression of differentiation markers such as TH, Isl1 and TuJ1 is not linked to cell cycle exit; the mechanisms that lead to withdrawal from the cell cycle and the generation of postmitotic neurons remained largely unclear.

The Alk tyrosine kinase receptor is implicated in the control of sympathoadrenergic proliferation as gain-of-function mutations leading to constitutively active ALK were identified as a major cause for familial neuroblastoma (Janoueix-Lerosey et al., 2008; Mosse et al., 2008). A function in proliferation was also supported by overexpression experiments in various cell lines (Chen et al., 2008; De Brouwer et al., 2010; George et al., 2008) and knockdown in ALK-expressing NB cell lines (Chen et al., 2008; George et al., 2008; Janoueix-Lerosey et al., 2008; Mosse et al., 2008). The present results confirmed the proliferation stimulatory role of NB ALK mutations in embryonic sympathetic neuron cultures – the potential tumor founder cells. Interestingly, Alk signaling is not only sufficient but is also essential for normal sympathetic neuron proliferation in vitro and in vivo, as revealed by pharmacological inhibition and Alk knockdown. Our finding that forced expression of normal ALK^wt increased proliferation of sympathetic neurons, whereas ALK^wt had no effect in NIH3T3 and Ba/F3 cells (Chen et al., 2008; George et al., 2008) suggested the presence of (an) endogenous ligand(s) in sympathetic neuron cultures available for activation of the ectopically expressed receptor.

Two potential ligands for the Alk receptor are known, Mk and Ptn (Ardini et al., 2010; Palmer et al., 2009), with similar expression patterns in rodents (Mitsiadis et al., 1995). However, besides Alk (Stoica et al., 2002), a number of additional receptor types have been identified for Mk, including PTPζ (Maeda et al., 1999), integrins (Muramatsu et al., 2004), Notch2 (Huang et al., 2008), LDL receptor-related protein (Muramatsu et al., 2000) and neuroglycan C (Ichihara-Tanaka et al., 2006). Both MK and PTN are expressed at high levels in neuroblastoma, but only MK expression correlates with aggressive NB (Fiegel et al., 2008; Nakagawara et al., 1995). Mk-deficient mice have not been analyzed specifically for defects in SG development (Nakamura et al., 1998). A reduction in ganglion size may have remained inconspicuous with respect to mouse behavior under normal animal housing conditions. In our study, Mk was identified as the physiologically relevant Alk activator for sympathetic neurons because: (1) Mk is expressed at much higher levels in SG when compared with Ptn; (2) Mk stimulates sympathetic neuron proliferation in vitro and in vivo; (3) the effect of MK is mediated by Alk, as shown by Alk knockdown; and (4) Mk knockdown results in a strong decrease in sympathetic neuron proliferation in vitro and in vivo. Mk is expressed both in immature sympathetic neurons and in the ganglion non-neuronal population, and may act in both an autocrine and paracrine manner, as previously shown for IGF-II (Zackenfels et al., 1995). It remains to be shown whether the activation of the Alk receptor involves direct binding of Mk or the PTPζ signaling pathway (Maeda et al., 1999; Perez-Pinera et al., 2007).

In vivo knockdown of Mk and Alk initially leads (E4) to a massively reduced number of proliferating EdU⁺ immature sympathetic neurons and to smaller ganglion size. The proliferation of Sox10⁺/TH^– sympathetic neuron progenitors may also depend on Mk/Alk signaling and contribute to the effect on ganglion size. The finding that E7 sympathetic ganglia of Mk/Alk knockdown embryos show partially recovered ganglion size and a normal proportion of EdU⁺ cells suggests that interruption of Alk signaling may be compensated over time by other signals like IGF-I and IGF-II (Zackenfels et al., 1995). It should be noted, however, that normal neuron proliferation at E7 depends on Alk signaling, as shown by our acute gain- and loss-of-function experiments in vitro. The recent observation in Drosophila CNS that neuroblast lineages devoid of Alk signaling fail to produce the full complement of neurons and display a reduced proportion of BrdU⁺ cells demonstrate an evolutionary conserved role for Alk in the control of neurogenesis (Cheng et al., 2011).

The mechanisms that lead to the termination of neurogenesis in sympathetic ganglia are unsolved and may either involve an increased probability of cell cycle exit or a regulated process, e.g. by the accumulation of cell cycle inhibitors or the downregulation of positive regulators such as Mk. It is unclear to what extent the decrease in Mk expression between E6 and E12 contributes to the strong reduction in neuron proliferation during this period (Rothman et al., 1978) as the effects of Mk overexpression could not be analyzed beyond E7. A tight control of Mk expression levels and/or availability would be required in order for Mk to function in controlling the extent and timing of neurogenesis. Several explanations for the decrease in Mk expression can be provided: (1) downregulation by extrinsic signals such as TGFβ, which is known to interfere with proliferation (Rahhal et al., 2004); (2) reduction by an intrinsic timing mechanism that determines when precursors withdraw from the cell cycle, similar to the reduction of Id4 in oligodendrocyte precursors (Kondo and Raff, 2000); (3) as Mk expression is regulated by a HIF responsive element, downregulation of HIF-2α protein may also lead to decreased Mk levels (Favier et al., 1999; Reynolds et al., 2004).

What is the link between Alk and other signals also known to be involved in the proliferation control of sympathoadrenergic precursor cells and immature neurons? Hand2 and Alk have been identified as genes affected by raised extracellular Mk levels and its knockdown. Elevated Alk expression in response to increased Alk signaling suggests a positive autoregulation of Alk. As Alk receptors may be limited in SG, this autoregulatory loop will ensure proliferation of all cells with access to Mk. Notably, higher levels of ALK mRNA were observed in NB samples harboring ALK mutations when compared with ALK^wt (Caren et al., 2008; Mosse et al., 2008; Schulte et al., 2011). The recent observation that NB with high-level ALK^wt expression display similar unfavorable phenotypes and clinical courses as do NB with activating ALK mutations demonstrates the important role of ALK^wt and its ligand in individuals with NB with elevated ALK expression (Schulte et al., 2011). The autoregulatory expression of Alk has also been shown for glioblastoma and may confer a proliferative advantage to the cells (Powers et al., 2002).

Hand2 is not only increased in parallel but is responsible for the upregulation of Alk and previously described proliferation effects of Hand2 (Reiff et al., 2010) are due to increased Alk expression and signaling. Supporting this notion, a function for Hand2 downstream of Alk has been demonstrated in Drosophila midgut visceral muscle development (Varshney and Palmer, 2006). It is presently unclear how Alk expression is initiated in primary SG and whether Hand2 is involved in this initial process.

High level expression of constitutively active ALK were shown to transform 3T3 fibroblasts and Ba/F3 lymphoid cells in vitro, which involves the PLCγ, PI3K/AKT, RAS/MAPK and JAK/STAT pathways (Chen et al., 2008; George et al., 2008). In addition, in vivo forced expression of ALK^F1147L in TH- or DBH-expressing embryonic mouse sympathetic neurons is able to induce NB in 4- to 6-month-old mice (J. Schulte, personal communication). The mouse model displays the incomplete penetrance and delayed tumor development of human ALK germline mutations, i.e. some carriers of human ALK mutations in familial NB remain asymptomatic and other mutation carriers in affected families do not develop detectable tumors during embryonic life but during postnatal months and years (Janoueix-Lerosey et al., 2008; Mosse et al., 2008). Delayed tumor development results from cells of solid tumors, such as NB, accumulating several ‘driver’ mutations in cancer genes and genomic aberrations in order to achieve a neoplastic status (Bozic et al., 2010; Parsons et al., 2011; Vogelstein and Kinzler, 2004). NB can also be elicited in mice by forced NMYC expression under the control of the TH promoter (Weiss et al., 1997). Interestingly, in this mouse model of NB, there is also a large delay between tumor-mediated initiation of low level NMYC expression and overt tumor generation, which suggests that additional mutations and chromosomal aberrations are accumulated during this period (Hansford et al., 2004). Our finding that sustained and increased Alk signaling leads to increased NMyc expression, together with the conclusion from the TH-NMYC mouse that elevated NMYC levels are sufficient to elicit NB development (Alam et al., 2009; Hansford et al., 2004; Weiss et al., 1997) implicates NMyc as an important component of the ALK-induced NB predisposing mechanism. Interestingly, NMyc amplification occurs postnatally in the TH-NMYC mouse and is also associated with the ALK F1174 mutation in human NB (De Brouwer et al., 2010; Hansford et al., 2004). Elevated trkB levels may contribute by increased proliferation of sympathetic progenitors (Reiff et al., 2010; Straub et al., 2007).

The differential Mk and Alk expression supports our notion that the restriction of NB development to sympathoadrenergic cells when compared with parasympathetic, enteric and sensory precursors is caused by different modes of neurogenesis and cell cycle control (Reiff et al., 2010; Tsarovina et al., 2008). The present data suggest that Alk-dependent normal neurogenesis defines the lineage and the developmental time period when ALK mutations are effective for tumor predisposition.

We thank J. Andrees for excellent technical assistance, T. Wunderle and J. Mordel for support, U. Ernsberger and M. Schmidt for comments on the manuscript, and J. H. Schulte for discussion and communicating results prior to publication. We thank S. Morris for full-length ALK plasmid and V. Raynal for assistance in the preparation of the ALK pcDNA3.1 vectors.