Knockdown of Pnpla6 protein results in motor neuron defects in zebrafish

Motor neuron diseases (MNDs) are an etiologically heterogeneous group of disorders that are characterized by muscle weakness and/or spastic paralysis, which results from the degeneration of motor neurons. Hereditary spastic paraplegia (HSP) is the second-most common group of MNDs, whose main feature is progressive spasticity in the lower limbs. The identification of numerous spastic paraplegia (SPG) loci that are associated with different forms of HSP underscored a varied etiology and pathology in HSP. Currently, more than 40 SPG loci and 20 causative genes have been identified, which provides important insights into pathogenesis leading to HSP (Dion et al., 2009). In 2008, a relatively rare type of HSP with marked distal wasting in all four limbs was reported and the gene responsible for it was named SPG39 (Rainier et al., 2008; Rainier et al., 2011). SPG39 protein, i.e. patatin-like phospholipase domain containing 6 (PNPLA6), is a protein that had been studied for many years in neurotoxicology of organophosphate poisoning. But, until now, little is known about its exact role in HSP pathogenesis.

The protein product of PNPLA6 is neuropathy target esterase (NTE), a specific serine esterase that was originally identified as the target of organophosphorus esters that cause a delayed neuropathy in human and other vertebrates (Pope et al., 1993; Wilson et al., 1990). This organophosphate-induced delayed neuropathy (OPIDN) syndrome is characterized by paralysis of the lower limbs and degeneration of long axons in the spinal cord and peripheral nerves. Although much research has shown that functional loss of NTE is an initiating event in OPIDN (Glynn, 2005; Glynn, 2006), the exact role of NTE is not well established. The PNPLA6 mutations in HSP occur in the key catalytic domain of NTE and abolish NTE activity. The fact that OPIDN and the PNPLA6-related HSP exhibit similar symptoms also suggests a possible link between NTE function and motor neuron development.

PNPLA6, a conservative protein found in many species ranging from yeast to mammals (Chang et al., 2008; Kaur et al., 2006; Kretzschmar et al., 1997; Zaccheo et al., 2004), is abundantly expressed in the central nervous system and peripheral nerves (Johnson, 1982; Lush et al., 1998). PNPLA6 is anchored to the cytoplasmic side of endoplasmic reticulum (ER) and displays potent phospholipase B activity that is responsible for converting phosphatidylcholine to glycerophosphocholine in mammalian cells (Zaccheo et al., 2004). Although its tissue distribution and catalytic function are well established, how it functions in the nervous system has remained elusive. Drosophila was the first in vivo model established for PNPLA6 study. When swiss cheese (sws), the ortholog of PNPLA6 in Drosophila, was mutated, the activity of NTE was totally abolished. The sws mutant Drosophila exhibited neural degeneration including abnormal, multilayered wrappings around glia and neuron death. However, this phenotype can be rescued by mouse Nte, suggesting that PNPLA6 contributes to neural homeostasis (Kretzschmar et al., 1997; Mühlig-Versen et al., 2005). Complete inactivation of Nte gene results in embryonic lethality in mouse, whereas Nte^+/− mice do not exhibit apparent defects in the nervous system (Winrow et al., 2003). In a neuron-specific Nte knockout mouse strain, the absence of Nte results in disruption of ER, abnormal reticular aggregation and vacuolization of nerve cell bodies (Akassoglou et al., 2004). Furthermore, Read et al. observed distal degeneration of the longest spinal axons in nestin-cre:NTEfl/fl mice that closely resembled the axonal lesion in HSP (Read et al., 2009). These data indicate that PNPLA6 plays a key role in normal neural development and axon maintenance.

Zebrafish are becoming more popular as a model organism because the large and transparent embryos can be observed and manipulated with relative ease. Its genomic organization and embryonic development have been well characterized (Dodd et al., 2000). The relatively simple nervous system, gene knockdown technology with morpholinos (MOs), and the availability of many cell-type-specific antibodies provide additional advantages for study of genes related to neural function in zebrafish (Kabashi et al., 2011). We studied the function of Pnpla6 protein in zebrafish neural development and gained some insight into the role of PNPLA6 in HSP pathogenesis.

Zebrafish pnpla6 gene (Gene ID: 560986) consists of 35 exons that span over 50 kb. The transcript of 4144 bp (XM_001921173.2) is predicted to encode for a protein of 1348 amino acids (XP_001921208.1). Based on the predicted zebrafish pnpla6 mRNA sequence, we designed five pairs of primers for PCR amplification and obtained the complete pnpla6 coding sequence of zebrafish. The zebrafish pnpla6 coding sequence we obtained contains 4026 bps and encodes 1342 amino acids. Zebrafish Pnpla6 protein shares 73% identity with human PNPLA6 and is highly conserved compared with Pnpla6 protein in other species including Drosophila and mouse. The catalytic domain of zebrafish Pnpla6 protein is similar to that in other vertebrate species, consisting of the serine hydrolase signature motif GXSXG (Glynn, 2006).

An antisense morpholino oligonucleotide targeting the pnpla6 mRNA translation start site was used to decrease the overall expression of Pnpla6 protein. Before analyzing the phenotypes of pnpla6 knockdown zebrafish, we first tested the effects of different concentrations of pnpla6 MOs on zebrafish development. The morphants that stayed alive were scored at different time points for each MO concentration injected. The survival rate of embryos decreased with increasing dose of MO injected (data not shown). The most prominent phenotype in pnpla6 MO morphants is curly tail (Fig. 1A,B). The proportion of morphants with curly tail increased with increasing pnpla6 MO concentration (Fig. 1O). This dose-dependent phenotypic severity indicates the specificity of pnpla6 knockdown. To avoid the toxic effects caused by high doses of MOs, 0.15 pmol of pnpla6 MO was chosen for subsequent experiments.

Injection of pnpla6 MO caused dramatic changes in the gross morphological appearance of pnpla6 morphants from 6 to 72 hours post-fertilization (hpf). Because the tail malformation in pnpla6 morphants is probably caused by anterior-posterior (AP) pattern defect, early AP pattern formation was examined. Before somite furrowing, no significant difference was observed between pnpla6 morphants and controls (injected with control MO). At the four-somite stage, the AP axis was shorter in pnpla6 morphants, or with a longer gap between head and tail (Fig. 1C,D). At the six-somite stage, the tails in pnpla6 morphants still appeared shorter than in control animals (Fig. 1E,F). A summary of the straight-line distances between head and tail of embryos measured at the four-somite stage is shown in Fig. 1P. Aberrant eye development (Fig. 1G,H) and reduced otic vesicle size (Fig. 1I,J) were also apparent in pnpla6 morphants. Compared with control animals, pnpla6 morphants have smaller eyes and some morphants had unclosed eye fissures. Other abnormalities, such as midbrain-hindbrain boundary (MHB) defect (Fig. 1K,L) and swelling of the pericardium (Fig. 1M,N), reduced heart rate and reduced blood flow (data not shown) were also observed in some pnpla6 morphants.

To further test the specificity of pnpla6 MO knockdown, rescue experiments with human PNPLA6 mRNA were performed. When human wild-type PNPLA6 mRNA was co-injected with pnpla6 MO, the pnpla6 morphant phenotypes were rescued. Most embryos receiving co-injection demonstrated normal tail formation (Fig. 2B,C,H). Only 14.2% of embryos co-injected with wild-type mRNA showed curly tails, a significant decrease from that in the pnpla6 MO morphants (53.7%).

To clarify whether the phenotypes of pnpla6 knockdown were a result of lack of catalytic activity conferred by the serine-centered domain (Fig. 2A), we prepared three mutant human mRNAs. The three key amino acids at the catalytic domain were mutated to alanine. None of the three mutant RNAs (D960A mRNA, S966A mRNA and D1086A mRNA) was able to rescue the abnormal phenotypes of the pnpla6 morphants (Fig. 2E-H). The results obtained with the rescue experiments confirmed that the phenotypes in the pnpla6 knockdown morphants were due to reduced esterase activity.

It has been reported that PNPLA6 mutation contributes to motor neuron disease in humans (Rainier et al., 2008). Another study showed that PNPLA6 might play a crucial role in maintenance of axons in sensory and motor spinal tracts on adult axons (Read et al., 2009). To determine whether reduced pnpla6 expression could also affect the growth of spinal motor neurons in zebrafish, we examined the motor neurons and their axon outgrowth. The spinal cord motor neurons were immunostained with antibody 39.4D5, which recognizes the islet-1 transcription factor in the nuclei of motor neurons and Rohon-Beard sensory neurons. Immunostaining of islet-1 revealed that pnpla6 morphants displayed fewer motor neurons than control embryos (Fig. 3A,B). Quantification of motor neurons at 30 hpf showed that the number of islet-1-positive spinal motor neurons was significantly lower than in controls (Fig. 3I).

To observe the effect of pnpla6 knockdown on axon outgrowth, pnpla6 MO-injected embryos were fixed and whole-mount antibody staining was performed using the Znp-1 antibody (Melancon et al., 1997), which stains caudal primary motor axons and middle primary motor axons. The pnpla6 MO morphants showed aberrant axon branching (Fig. 3D). The axons in pnpla6 morphants also appeared to be truncated, failing to join the ventral muscles (Fig. 3D). At 36 hpf, the incidence of axon branching and truncation (10 axons at segments 11-20 per side or 20 axons per embryo) was also increased over that in controls. Although 69.2% of axons observed exhibited the phenotype of truncation in pnpla6 morphants, only 31.3% exhibited the phenotype in control embryos (Fig. 3J). When pnpla6 MO morphants were co-injected with wild-type human PNPLA6 mRNA, a significant drop in the number of aberrant motor axons was observed (Fig. 3E,J). However, these axon defects in pnpla6 morphants could not be rescued by co-injection of mutant mRNA (Fig. 3F,J).

Increase in apoptosis could contribute to the impairment in primary motor neuron development in pnpla6 morphants. We therefore evaluated the level of apoptosis in pnpla6 morphants. Using Acridine Orange staining, cell death and apoptosis were analyzed in pnpla6 morphant embryos. We found that there were more apoptotic cells in pnpla6 morphants than in control animals, especially in the central nervous system (data not shown). Because morpholino itself can lead to cell death through the p53 pathway, we used p53 knockdown co-injection for rescue. Knockdown of p53 can rescue cell death in control but not in pnpla6 MO-injected animals. Consistent with our observations using Acridine Orange for apoptosis analysis, TUNEL assay showed that more apoptotic cells were detected in morphants (Fig. 3G,H).

Because most of the pnpla6 morphants demonstrated curved tails, we determined whether the axon defect was due to the deformed body. First of all, we assessed the correlation between curly tail and axon defect in pnpla6 morphants. Although the morphants with severely impaired axons had severely curly tails, 15% of morphants with curly tails did not show axon defects. On the other hand, 65% of morphants without curly tails exhibited axon defects. Next, we scored the number of the motor axon defects located in the curved region (segments 13-20) and non-curved region (segments 5-12) in pnpla6 morphant embryos at 36 hpf. There was no significant difference in the percentage of sides with at least one of the axon defects between the curved and non-curved regions (truncation: 56.5±4.4% and 50.7±2.2%, respectively; branching: 46.6±2.4% and 49.0±4.2%, respectively; for all experiments, n=98). To further test whether the axon defects were secondary to the curved body axis of pnpla6 morphants, we performed spinal neuron cultures. Cells were prepared from pnpla6 morphant and control embryos. After the primary cells were cultured for 8 hours in vitro, the lengths and branches of motor neurons were determined using Znp-1 immunofluorescence staining. As shown in Fig. 4, motor neurons of pnpla6 morphants exhibited axons that were much shorter and more branched than in controls. These results are consistent with defects seen in the entire embryos and argue against deformed body being the cause of the impaired motor neuron development.

To examine whether pnpla6 functions in a cell-autonomous fashion, we established chimeric zebrafish embryos. Donor embryos were injected with rhodamine-lysine-dextran (RLDx), along with control MO or pnpla6 MO, at the one-cell stage. At 4-6 hpf, a small group of cells at the neuroectoderm were transplanted into the tail-bud region of wild-type recipients. At 36 hpf, the recipients were fixed and immunostained with Znp-1 antibody. As shown in Fig. 5, most motor neurons developed from pnpla6 MO cells in the chimeric embryos exhibited axon defects, resembling the motor neuron phenotype of pnpla6 morphants.

Because our results demonstrated a role for pnpla6 in motor neurons, we wondered whether pnpla6 function is also crucial to other types of neurons. Hindbrain Mauthner neurons are the most important interneurons in zebrafish. At 34 hpf, embryos were immunolabeled with 3A10 antibody, which specifically labels the commissural axon of the Mauthner neuron (McWhorter et al., 2003). Compared with controls, pnpla6 MO morphants did not exhibit a significant difference in the formation and arrangement of Mauthner neuron axons. Moreover, pnpla6 knockdown was not observed to affect the projection from interneurons (Fig. 6A,B).

To determine the effect of pnpla6 knockdown on sensory neurons, acetylated tubulin monoclonal antibody staining was employed to analyze sensory neuron development. As shown in Fig. 6C,D, the Rohon-Beard, lateral line sensory neurons and axons from them appeared normal in pnpla6 morphants. We next used neuronal antibodies HuC/D to analyze pnpla6 morphants and control embryos, and found that neither the number nor the arrangement of mature neurons was significantly affected by the knockdown of pnpla6 (Fig. 6E-H).

It has recently been reported that bone morphogenetic protein (BMP), one of the major secreted morphogens, plays a key role in HSP pathogenesis. Several HSP proteins, including NIPA1, spastin, spartin and atlastin, are all confirmed to be inhibitors of BMP signaling (Fassier et al., 2010; Tsang et al., 2009). Taking into consideration their similarity to Pnpla6 protein in cellular location and in phenotypes caused by reduced gene expression in zebrafish, we next investigated whether the BMP pathway is involved in the abnormal neural development in pnpla6 morphants. As shown in Fig. 7A, there was a striking increase in the level of phosphorylated Smad1/5/8 in pnpla6 morphants compared with control embryos at 24 hpf. Furthermore, the expression of phosphorylated Smad1/5/8 appeared to be reduced to the normal level in embryos rescued with wild-type human PNPLA6 mRNA. Consistent with the results of the rescue experiments mentioned above, mutant RNA failed to attenuate the expression of phosphorylated Smad1/5/8 in pnpla6 morphants.

To determine whether the developmental defects observed in pnpla6 morphants were caused by the upregulation of BMP signaling, dorsomorphin (DM), a small molecule that blocks the kinase activity of BMP receptors (Yu et al., 2008), was used to test possible involvement of BMP signaling in zebrafish pnpla6 morphants. The pnpla6 morphants and control embryos were each divided into two groups that were treated with DM or dimethyl sulfoxide (DMSO) at the ten-somite stage, and the typical phenotype (curly tail) was examined at 36 hpf. The phenotypes of pnpla6 morphants, including curly tail, could be rescued by treatment with DM, but not by DMSO (Fig. 7B,C).

Here, we describe a vertebrate model of pnpla6 insufficiency. Decreased pnpla6 activity in embryonic zebrafish resulted in developmental abnormalities including curly tail, short AP axis and small eyes. Knockdown of pnpla6 also results in motor neuron loss and axon growth defects including truncation and branching. The following lines of evidence suggest that loss of pnpla6 is directly responsible for these phenotypes: first, a mismatch control MO does not cause the same phenotypes; and second, co-injection of human wild-type mRNA together with the pnpla6 MO significantly rescued all the abnormalities in pnpla6 morphants.

Because Pnpla6 is essential for embryonic development and Pnpla6 knockout mice died around embryonic day 9 (Moser et al., 2004; Winrow et al., 2003), it is difficult to study its function in early development using mice. Employing zebrafish as a model system, our in vivo studies revealed several new functions for pnpla6 in motor neuron development, AP axis development and eye development. With regard to the role of pnpla6 in motor neuron development, pnpla6 morphants showed prominent motor neuron loss and defects of motor neuron axon outgrowth, including abnormal branching and truncation. Using a neuron-specific Nte knockout mouse model, Akassoglou et al. showed that loss of NTE results in neuron defects in hippocampus, thalamus and cerebellum (Akassoglou et al., 2004). Recently, the distal degeneration of the spinal axons was observed in 3-week-old NTE-deficient mice (Read et al., 2009). Furthermore, progressive degeneration involved both motor and sensory spinal tracts, although the incidence of lesions in sensory tracts was much lower than that in motor tracts. In contrast to these results, no obvious defects of interneurons and sensory neurons were found in pnpla6 morphant zebrafish. In our zebrafish model, morpholino antisense oligonucleotides were used to knockdown NTE, and the effect of morpholino can last only about 5-7 days. The effect caused by morpholino knockdown should be much less than that in knockout mice, in which sustained NTE deficiency over many months can result in much severer phenotypes. Furthermore, the axon lesion in Pnpla6 conditional knockout mice, including Wallerian-like morphology, closely resembled that in HSP. It was proposed that PNPLA6 deficiency affects membrane trafficking, which eventually results in distal axon lesion and consequent motor defects in HSP patients (Read et al., 2009). In our study, zebrafish pnpla6 MO morphants were found to exhibit more truncation and branching in motor neuron axons. Therefore, it is reasonable to assume that pnpla6 plays a key role in axon development and maintenance and, when deficient, leads to the axon pathology of HSP.

The most distinct morphological phenotype in pnpla6 MO morphants is the curly tails, resembling the phenotype observed in zebrafish with morpholino knockdown of KIAA0196. KIAA0196, also named SPG8, is the eighth HSP locus whose mutation leads to an adult onset HSP with marked spasticity (Valdmanis et al., 2007). We previously identified a mutation in SLC33A1 (SPG42) as the cause of HSP in a Chinese family and showed that the slc33a1 MO zebrafish exhibit varying degree of curly tails (Lin et al., 2008). Morpholino knockdown of several genes related to neurodegenerative diseases were also shown to lead to curly tail or other forms of impaired tail development (Anichtchik et al., 2008; Lumsden et al., 2007). These data suggest that abnormality in tail development might be a common feature of zebrafish with deficiency in genes involved in the neural development.

Because it is well known that PNPLA6 functions as a serine esterase, the physiological role of its catalytic activity was also addressed in this study. Like most of the members of the patatin-like phospholipase family protein family (Wilson et al., 2006), PNPLA6 contains a GXSXG motif, which includes a nucleophilic serine essential to its catalytic domain. It has been confirmed that Ser966, Asp960 and Asp1086 are key loci for esterase activity (Atkins and Glynn, 2000). We constructed three mutant pnpla6 vectors (with D960A, S966A and D1086A point mutations) to test whether the esterase activity of PNPLA6 is required for its function in neural development. The zebrafish pnpla6 knockdown phenotype could be rescued by co-injecting the pnpla6 MO with wild-type human PNPLA6 mRNA. By contrast, co-injection of pnpla6MO with each of the three mutated PNPLA6 mRNAs failed to rescue the phenotypes of pnpla6 MO morphants, confirming the key role of the catalytic domain for its physiological function. In Drosophila, Ser985 of sws, which aligns precisely with Ser966 of vertebrate NTE, has been shown to be the active residue. The observation that sws with a point mutation in Ser985 failed to functionally replace wild-type sws further confirmed the importance of the active site (Mühlig-Versen et al., 2005). Each of two point mutations (M1012V, R890H) in PNPLA6 can lead to a motor neuron disease in humans (Rainier et al., 2008). A recent report revealed that introduction of these mutations can also change the NTE catalytic activity (Hein et al., 2010). Both mutations are located in the conserved region of the serine-centered catalytic domain. These findings are consistent with our data demonstrating a role for NTE catalytic activity in vertebrate development and motor neuron maintenance.

Despite the genetic heterogeneity of HSP, the similar clinical phenotypes shared by different SPGs suggests a common biochemical pathway. Spichthyin, the Drosophila ortholog of SPG6, is the first HSP protein reported to function as a BMP inhibitor (Wang et al., 2007). The elevated levels of phosphorylated smad and BMP receptors in spichthyin mutant animals suggests that spichthyin antagonizes BMP signaling by regulating its receptor traffic. It has also been demonstrated in human cells that NIPA1 inhibits BMP signaling by promoting endocytosis and lysosomal degradation of BMP receptors. SPG4 (spastin) and SPG20 (spartin) protein are also demonstrated to be BMP inhibitors (Tsang et al., 2009). Another study showed that SPG3A protein, also known as atlastin, can inhibit BMP signaling in zebrafish, and that the atlastin morphant phenotype can be rescued by BMP repression (Fassier et al., 2010). A recent study confirmed that two HSP-associated proteins, ATLASTIN-1 and NIPA1, are direct binding partners and proposed that these two proteins are members of a common biochemical pathway for maintenance of motor neuron axons (Botzolakis et al., 2011). Our present study also demonstrates the link between pnpla6 and the BMP pathway. In pnpla6 morphants, phosphorylated smad was expressed at abnormally high levels, hinting that pnpla6 is an inhibitor of BMP signaling. This notion was substantiated by the result that the BMP inhibitor DM could rescue the abnormal phenotype. It was also reported that deletion of PNPLA6 results in disruption of ER and aberrant membrane trafficking. Although the exact mechanism is still unknown, results obtained on BMP signaling strongly suggested that PNPLA6 might affect the trafficking of BMP-related protein which, when impaired, could lead to motor neuron defects. Furthermore, previous study has suggested that knockdown of atlastin leads to abnormal architecture of motor neurons with excessive branching. In the current study, we have shown that knockdown of pnpla6 also leads to abnormal architecture of motor neurons with excessive truncation, although branching is also observed. Interestingly, both atl1 and pnpla6 genes affect BMP signaling. We believe that both Atlastin and Pnpla6 might be involved in motor axon development. However, they might play different functional roles in this process. It is possible that BMP signaling is the common downstream pathway.

In summary, we have described a zebrafish model of neural dysfunction as a result of the knockdown of the pnpla6 gene, which is responsible for HSP in humans. This zebrafish model offers the opportunity to study the irregularities in BMP signaling or in other pathways during the pathogenesis of HSP.

Wild-type TU-strain zebrafish (Danio rerio) were maintained at 28.5°C under standard laboratory conditions (Westerfield, 1995). Embryos were obtained by natural cross and were staged by hours or days after fertilization at 28.5°C using morphological criteria (Kimmel et al., 1995).

Dorsomorphin (DM; EMD Biosciences/Calbiochem), a BMP type I receptor antagonist, was used to downregulate the BMP pathway. For the BMP inhibition experiment, pnpla6 morphant and control embryos at 10 hpf were dechorionated and each divided into two groups, which were treated with 6 μmol/l DM or DMSO and the phenotype of different groups scored at 36 hpf (Yu et al., 2008).

D16 vector was kindly provided by Yong Li (Leicester University, UK) and total cDNA of human PNPLA6was cloned into pcDNA3.1 vector. Point mutations were introduced to PNPLA6 cDNA at positions corresponding to amino acids 960, 966 and 1086, respectively.

Based on the predicted zebrafish pnpla6 sequence, five pairs of primers were designed for the amplification of zebrafish pnpla6. Total cDNA from 72 hpf zebrafish embryos was used as template and five overlapping fragments were amplified. Using appropriate endonucleases distributed in the five fragments, the complete pnpla6 coding region was obtained. Sequencing of the complete sequence further verified its pnpla6 identity.

Injections were performed on wild-type zebrafish embryos using a PV830 Pneumatic PicoPump (WPI) under a Nikon SMZ 645 stereomicroscope. Antisense MOs were purchased from Gene Tools (Philomath, OR). One translation blocking morpholino against the ATG-containing sequence (+1 to +25) was designed (5′-CTGTGTCCGATGTGCTCTGTCCCAT-3′) and named pnpla6 MO. The control MO oligo (5′-CTcTGTgCGATGTcCTCTcTCCgAT-3′) had five mismatched bases (shown in lowercase) compared with pnpla6MO. The MOs were resuspended in RNase-free water with Phenol Red and injected into the yolk of one- to two-cell stage embryos. Different doses (0.15 pmol or 0.3 pmol) of MO were injected. Human wild-type and mutant PNPLA6 mRNA were in vitro transcribed from linearized pCS2+ constructs using SP6 Message Machine Kit (Ambion). For the rescue experiments, injections were performed as mentioned above, with 0.15 pmol MO plus 60 pg human PNPLA6 mRNA.

Embryos were examined with an Olympus SZX16 dissecting microscope, and images were captured using a DP-71 camera and Olympus DP2-BSW controller program. Figures were generated using Adobe Illustrator CS4 or Adobe Photoshop CS2. For the analysis of distance between the head and tail, embryos were captured and measured using the DP2-BSW controller program.

Immunohistochemistry was performed using standard procedures for zebrafish (Beattie, 2000). The following monoclonal antibodies (mAb) were used: mAb Znp-1 (1:2000; Developmental Studies Hybridoma Bank, DSHB, Iowa City, IA), mAb acetylated tubulin and HuC/D (1:200; Santa Cruz Biotechnology), 39.4D5 (1:2000; DSHB), 3A10 (1:200; DSHB). Primary antibody binding was visualized using a 3,3′-diaminobenzidine tetrahydrochloride substrate kit (Zhongshan Goldenbridge Biotechnology, Beijing, China).

To generate chimeric embryos, cell transplantations were performed as previously described with modifications (Ho and Kane, 1990; van Eeden et al., 1996). Donor embryos for the transplants were injected with rhodamine-lysine-dextran (RLDx, 25 mg/ml), along with control MO or pnpla6 MO, at the one-cell stage. These embryos were allowed to develop to 4-6 hpf. Several cells near the animal pole of donor embryos were transplanted into the midline of the neural plate adjacent to the tail bud of the recipients. After immunostaining with mAb Znp-1, images were acquired with a confocal laser scanning microscope (Zeiss LSM 780) or with a fluorescent microscope.

The 18-somite stage embryos were used for primary neuron culture. Briefly, the spinal cords were isolated by removing the yolk bag, head and tail. They were then treated with Custom ATV solution (0.6 mM EDTA, 5.5 mM glucose, 5.4 mM KCI, 136.8 mM NaCI, 5.5 mM Na₂CO₃ and 0.05% w/v trypsin). The fully dissociated spinal cords were plated onto a cover glass in a Petri dish. Eight hours after plating at room temperature, cells were fixed with 4% paraformaldehyde for immunofluorescence analysis. The mAb Znp-1 (1:100; DSHB) was used for analysis of motor neuron growth and the lengths of axons were measured.

Apoptotic and necrotic cells were detected by incubation of 30-hpf embryos in Acridine Orange solution (5 μg/ml) (Furutani-Seiki et al., 1996). Cells loaded with the dye were visualized using a tetramethylrhodamine isothiocyanate filter on the Olympus BX51 fluorescent microscope. For TUNEL assay, embryos were fixed with 4% paraformaldehyde overnight, rehydrated with decreasing concentrations of methanol, washed with PBST (PBS containing 0.1% Tween 20), treated with proteinase K (10 mg/ml) at 37°C for 30 minutes, refixed with 4% paraformaldehyde, washed in PBST, and lastly incubated with 0.1 M citrate solution (0.1% citrate in PBST). TUNEL staining was performed with the One Step TUNEL Apoptosis Assay Kit (Beyotime, Jiangsu, China).

Embryos collected at suitable developmental stages were dechorionated, deyolked and lysed in RIPA lysis buffer supplemented with 1 mM PMSF, 1 mM NaF and 1 mM sodium vanadate (Na₃VO₄). Equal amounts of protein (40 μg) of the embryos were electrophoresed into 10% SDS-PAGE gel and transferred to nitrocellulose membranes. Membranes were blocked with PBS buffer containing 5% non-fat dry milk and 0.01% Tween 20 for at least 1 hour at room temperature, followed by overnight incubation at 4°C with primary antibodies against phosphorylated Smad1/5/8 (1:400; Cell Signaling), Smad1 (1:500; Abcam), BMPRI (1:300; Santa Cruz Biotechnology) or actin (1:5000; Sigma). After washing, the membranes were incubated with horseradish peroxidase-labeled anti-rabbit (Sigma), anti-mouse (Sigma) secondary antibodies for 1 hour at 20-23°C. Chemiluminescence detection was performed by ECL kit (Thermo).

Statistical significance between experimental groups was determined by one-way analysis of variance (ANOVA), and data from two groups was compared by Student's t-test. A difference between means was considered significant at P<0.05.

The authors thank Yong Li for the gift of D16 vector.

FUNDING: This study was supported by the Ministry of Science and Technology of China [grant number 2011CB966200], National Natural Science Foundation [grant numbers 30500190, 81070905, 81021001] and Natural Science Foundation of Shandong Province [grant number ZR2012HQ016].