Rest represses maturation within migrating facial branchiomotor neurons
Introduction
Development of the vertebrate brain involves the generation and coordinated migration of neurons from their birthplaces to their final destinations. Defective neurogenesis and neuronal migration can lead to serious human neurological disorders including autism, epilepsy, and varying degrees of mental retardation (Couillard-Despres et al., 2001; Gleeson and Walsh, 2000; Lv et al., 2013; Ross and Walsh, 2001; Valiente and Marin, 2010; Wegiel et al., 2010). Neurogenesis is the generation of new functional neurons, and includes cell fate specification, nascent neuron migration, maturation, and integration with the surrounding circuitry (Jobe et al., 2012). During development neural progenitors at the ventricular surface of the neural tube divide to produce immature postmitotic neurons (Schmidt et al., 2013). Since a newly generated neuron must still undergo maturation to become functional, there is a window of time during which its maturation can be stimulated or delayed, depending on the neuronal subtype. Postmitotic neurons typically migrate radially along radial glia towards the pial surface of the neuroepithelium, however, specific classes of postmitotic neurons including telencephalic subpallial interneurons, granule cells of the cerebellum, and facial branchiomotor neurons (FBMNs) migrate tangentially, perpendicular to and independent of radial glia, often over significant distances (Chedotal and Rijli, 2009; Marin and Rubenstein, 2001).
This study focuses on the terminal maturation of FBMNs, a distinct and identifiable population of motor neurons that undergoes a tangential migration through the segmented hindbrain that is conserved from mammals to fish. Zebrafish FBMNs start to be born in hindbrain rhombomere (r) 4 at 16 hours post fertilization (hpf), and by 18 hpf have begun tangential migration posteriorly through the hindbrain. The FBMNs migrate in a chain-like fashion, with additional neurons joining the bilateral migration streams over a period of several hours, until the neurons reach r6 or r7. Upon reaching their rhombomeric destination, the neurons cease tangential migration and turn to perform a short radial migration to their final location. By 48 hpf the FBMNs have coalesced into major bilateral nuclei in r6 and minor nuclei in r7 (Chandrasekhar, 2004; Moens and Prince, 2002; Song, 2007; Wanner et al., 2013). Postmitotic but immature neurons, including FBMNs, demonstrate an elongated bipolar morphology as they migrate, but upon reaching their appropriate destinations they stop migration, become rounded in shape, undergo dendritic branching, and begin to express terminal maturation genes important for neuron function (Garel et al., 2000; Komuro and Rakic, 1998; Mapp et al., 2010; O’Rourke et al., 1995). As FBMNs must migrate 100–150 μm to reach their destinations (Grant and Moens, 2010), it is critical that they maintain their immature migratory state over this distance before fully maturing.
Transcription factors are thought to largely control the migratory potential of neurons (Chedotal and Rijli, 2009; Marin et al., 2006; Nobrega-Pereira and Marin, 2009; Valiente and Marin, 2010); however, the underlying mechanism of this regulation is poorly understood in vivo. The transcriptional repressor RE1-silencing transcription factor (Rest) is expressed in developing embryos in all non-neural tissues, as well as neuronal stem cells and progenitors within the nervous system (Chong et al., 1995; Schoenherr and Anderson, 1995). Rest has been linked to the terminal maturation of neurons in vitro, and is therefore considered a master regulator of neuronal differentiation (Ballas et al., 2005; Ballas and Mandel, 2005). Rest binds to a conserved RE1 consensus sequence upstream of a large subset of neuron-specific genes, and interacts with multiple co-repressors to form a stable repressive complex that silences its gene targets (Ballas et al., 2005; Lunyak and Rosenfeld, 2005).
We previously demonstrated that zebrafish Rest protein is localized to both the nucleus and the cytoplasm of migrating FBMNs as they move through r4 and r5; however, Rest becomes depleted from the nuclei of neurons that have completed migration to r6 (Mapp et al., 2011). This suggests that Rest function is necessary in the nuclei of FBMNs as they migrate, and consistent with this hypothesis we previously reported that FBMN migration is partially abrogated when Rest function is globally depleted, whether by using a dominant negative strategy, by morpholino knockdown (Mapp et al., 2011) or through genetic mutation (Kok et al., 2012). Surprisingly, ectopic neurogenesis in non-neural tissues does not occur in rest mutants although expression of Rest target genes increases, suggesting that additional epigenetic factors are required for neural fate acquisition (Lessard et al., 2007; Ooi and Wood, 2008). Indeed, neurons differentiate in a progression of epigenetic states, and Rest depletion is required for cells to acquire the more general aspects of the neuronal phenotype to become fully functional. As this process is best characterized in cell culture studies, relatively little is known about the mechanism of maturation in vivo.
Here, we report the precocious maturation of FBMNs in Rest-deficient zebrafish embryos, revealing the intimate link between suppression of maturation and the ability of these neurons to migrate. Our analysis of maternal-zygotic rest mutants reveals a critical function for Rest in normal FBMN development, and we show that Rest functions specifically within the FBMNs to suppress both cellular and molecular features of their maturation program. We additionally provide evidence that the Rest target gene serine/arginine repetitive matrix 4 (srrm4), which encodes an alternative splicing factor, plays a key role in promoting neuronal maturation. Finally, we show that localization of Rest to the nucleus is essential for regulating the timing of FBMN maturation. Overall, we provide in vivo evidence for a genetic and cellular mechanism of Rest function that regulates both neuronal maturation and the associated migration of FBMNs.
Materials and methods
Fish lines and husbandry
Zebrafish were maintained following standard procedures. Embryos were raised at 28.5°C and staged as described (Kimmel et al., 1995) under IACUC-approved protocols. The following transgenic and mutant lines were used: Tg(islet1:GFP) (Higashijima et al., 2000), Tg(zCREST1:membRFP) (Mapp et al., 2010), Tg(islet1:GFP);restsbu29 (Kok et al., 2012), and Tg(islet1:GFP); pk1bfh122 (Higashijima et al., 2000; Mapp et al., 2011). Embryos were transferred into 0.2 mM 1-phenyl 2-thiourea (PTU; Sigma) starting at 24 hpf to inhibit melanin synthesis.
Morpholino design and microinjection
A translation-blocking morpholino (MO; Gene Tools) was designed against the start site of rest (5′-CTGAGACATGCTGGACCACTGAAAC-3′). Other MOs, as previously described: splice-blocking Rest (5′-GGCCTTTCACCTGTAAAATACAGAA-3′) (Gates et al., 2010; Mapp et al., 2011), splice-blocking Srrm4 (5′-TCAATCACTACCTATGTCGCTTCCT-3′) (Calarco et al., 2009), and standard control MO (Gene Tools, Philomath, OR, USA; Dalgin et al., 2011). MOs were resuspended in water (Sigma W5402) to a stock concentration of 20 ng/nl, and diluted in water and phenol red for microinjections. Rest MOs were injected at 4 ng/nl each and Srrm4 MO at 6 ng/nl into embryos at one- to four-cell stages.
Microscopy and data analysis
Fixed embryos were deyolked and flat-mounted in glycerol for dorsal imaging on a Zeiss LSM710 confocal microscope. Fiji/ImageJ (NIH) was used to process and analyze data, and Prism (GraphPad) was used for statistical analysis. Cell nuclei were labeled with TO-PRO-3 (Invitrogen), and volume measurements were made using the Sync Measure 3D plugin in Fiji (Joachim Walter).
Genotyping rest mutants
Adult fins, embryo tails or whole embryos were used to genotype rest mutants as described (Kok et al., 2012).
Generation of Rest variant constructs
All constructs were tagged with mCherry at the amino terminus and cloned downstream of the zCREST1 enhancer (Uemura et al., 2005) and a minimal hsp70 promoter, into a plasmid containing Tol2 transposition sequences (Kawakami and Shima, 1999). The flexible linkers connecting the mCherry tag were (RSRITSLYKKAGFFQWSS) for the Rest4(+ECDLVG) and full-length Rest variants (Mapp et al., 2011), and (RSRITSLYKKAGFFQWGT) for the REST-VP16 construct (Immaneni et al., 2000). All constructs were sequence verified using primers to the Tol2 arms: for 5′-TTTGGCAAAGAATTCCTCGAC-3′ and rev 5′-CTTCGCAGATCTGATCTAGAG-3′. Nuclear localization and mutant variant signals were added to the carboxy-terminus of zebrafish Rest using the following reverse primers (IDT):
Rest-PKKKRKV (Rest-NLS): 5′-GTATGGATCCTCACACTTTGCGTTTCTTCTTGGGTTTGCCCCCCTGTGCCGC-3′
Rest-PAAARKV (Rest-mutNLS): 5′-GTATGGATCCTCACACTTTTCTAGCAGCGGCGGGTTTGCCCCCCTGTGCCGC-3′
Expression of Rest-mutNLS in FBMNs had no effect on Rest-deficient migration. These constructs were also cloned into pCS2+ for mRNA synthesis.
mRNA generation and microinjection
Capped mRNA was generated using the MEGAscript SP6 Kit (Ambion). Tol2 transposase mRNA (60 ng/μl) and Tol2 plasmid DNA constructs (40 ng/μl) were kept on ice and carefully injected into one-cell stage embryos, avoiding the yolk, using a new cold needle every 15 min.
Immunohistochemistry
Embryos were fixed in 4% paraformaldehyde (PFA; Sigma), then washed in PBS containing 0.5% Triton-X (PBST). Embryos were permeabilized in water and cold acetone, incubated in blocking buffer (1% DMSO, 10% goat serum, 0.5% Triton-X in 1X PBS), and rinsed in 1X PBST + 1% DMSO. Primary and secondary antibodies were diluted in 1X PBST + 1% DMSO. The following primary antibodies were used: GFP at 1:500 (Clontech 632381), EphA4a at 1:2000 (a generous gift from David Wilkinson (Irving et al., 1996), RFP at 1:200 (Millipore AB3216), and DsRed at 1:200 (Clontech 632496). The following secondary antibodies were used: Alexa Fluor 488 (Molecular Probes A11001), Alexa Fluor 546 (Molecular Probes A11010), and anti-GFP Alexa Fluor 488 conjugate (Molecular Probes A21311). Embryos were then cleared in 80% glycerol, deyolked and flat-mounted.
Cell transplantation
As homozygous restsbu29/sbu29 males are not reliable breeders, crosses between heterozygous mutant restsbu29/+ males and homozygous mutant restsbu29/sbu29 females carrying Tg(zcrest1:mRFP) were used to generate donor embryos. Donor embryos were injected at the one-cell stage with 10 KDa lysinated Alexa Fluor 647 dextran (Life Technologies D-22914; 0.5% in 0.2 M KCl). Blastula stage donor cells were then transplanted as previously described (Rohrschneider et al., 2007) into the hindbrain primordia of Tg(islet1:GFP) gastrula stage hosts that were wild type for rest. Each individual donor was subsequently genotyped by PCR to identify homozygous mutants. Mosaic specimens were raised until 48 hpf then immunolabeled for GFP, RFP and EphA4 (to label r3 and r5), with donor-derived cells visualized using Alexa Fluor 647 lineage tracer. Donors from rest+/+ siblings carrying Tg(zcrest1:mRFP) were used as controls. Embryos were cleared in 80% glycerol, deyolked and flat-mounted for imaging.
Probe synthesis and whole mount in situ hybridization
Fixed embryos were dehydrated into methanol and rehydrated in PBS + 1% Tween 20 prior to in situ hybridization, as previously described (Prince et al., 1998) using digoxigenin-labeled RNA probes (Roche 11277073910). Following the in situ protocol, embryos were post-fixed before and after immunohistochemistry. Partial or full-length cDNA clones were obtained by PCR from 30 or 48 hpf zebrafish cDNA, cloned into pGEM-T vector (Promega) and verified by sequencing. The following accession numbers and primers were used:
adam23a ({"type":"entrez-nucleotide","attrs":{"text":"XM_686527.5","term_id":"528481937","term_text":"XM_686527.5"}}XM_686527.5): for 5′-CTCTCGTAATAATGCACTGCGAC-3′
rev 5′-AGCGCTAGAGAAGCAGTTCTATG-3′
cmet ({"type":"entrez-nucleotide","attrs":{"text":"NM_001007124","term_id":"55742199","term_text":"NM_001007124"}}NM_001007124) (Elsen et al., 2009)
nrp1a ({"type":"entrez-nucleotide","attrs":{"text":"NM_001040326","term_id":"94536927","term_text":"NM_001040326"}}NM_001040326) (Yu et al., 2004)
nrp2b ({"type":"entrez-nucleotide","attrs":{"text":"NM_212966","term_id":"47085724","term_text":"NM_212966"}}NM_212966) (Yu et al., 2004)
snap25b ({"type":"entrez-nucleotide","attrs":{"text":"NM_131434","term_id":"18859394","term_text":"NM_131434"}}NM_131434) (Thisse and Thisse, 2004)
srrm4 ({"type":"entrez-nucleotide","attrs":{"text":"XM_001336887","term_id":"528478686","term_text":"XM_001336887"}}XM_001336887) (Calarco et al., 2009)
syt4 ({"type":"entrez-nucleotide","attrs":{"text":"NM_199948","term_id":"41152407","term_text":"NM_199948"}}NM_199948) for 5′-GTGACGACAGAAGAAGCACATTT-3′
rev 5′-ATTACTCGCGACATAAGTCCACA-3′
Fish lines and husbandry
Zebrafish were maintained following standard procedures. Embryos were raised at 28.5°C and staged as described (Kimmel et al., 1995) under IACUC-approved protocols. The following transgenic and mutant lines were used: Tg(islet1:GFP) (Higashijima et al., 2000), Tg(zCREST1:membRFP) (Mapp et al., 2010), Tg(islet1:GFP);restsbu29 (Kok et al., 2012), and Tg(islet1:GFP); pk1bfh122 (Higashijima et al., 2000; Mapp et al., 2011). Embryos were transferred into 0.2 mM 1-phenyl 2-thiourea (PTU; Sigma) starting at 24 hpf to inhibit melanin synthesis.
Morpholino design and microinjection
A translation-blocking morpholino (MO; Gene Tools) was designed against the start site of rest (5′-CTGAGACATGCTGGACCACTGAAAC-3′). Other MOs, as previously described: splice-blocking Rest (5′-GGCCTTTCACCTGTAAAATACAGAA-3′) (Gates et al., 2010; Mapp et al., 2011), splice-blocking Srrm4 (5′-TCAATCACTACCTATGTCGCTTCCT-3′) (Calarco et al., 2009), and standard control MO (Gene Tools, Philomath, OR, USA; Dalgin et al., 2011). MOs were resuspended in water (Sigma W5402) to a stock concentration of 20 ng/nl, and diluted in water and phenol red for microinjections. Rest MOs were injected at 4 ng/nl each and Srrm4 MO at 6 ng/nl into embryos at one- to four-cell stages.
Microscopy and data analysis
Fixed embryos were deyolked and flat-mounted in glycerol for dorsal imaging on a Zeiss LSM710 confocal microscope. Fiji/ImageJ (NIH) was used to process and analyze data, and Prism (GraphPad) was used for statistical analysis. Cell nuclei were labeled with TO-PRO-3 (Invitrogen), and volume measurements were made using the Sync Measure 3D plugin in Fiji (Joachim Walter).
Genotyping rest mutants
Adult fins, embryo tails or whole embryos were used to genotype rest mutants as described (Kok et al., 2012).
Generation of Rest variant constructs
All constructs were tagged with mCherry at the amino terminus and cloned downstream of the zCREST1 enhancer (Uemura et al., 2005) and a minimal hsp70 promoter, into a plasmid containing Tol2 transposition sequences (Kawakami and Shima, 1999). The flexible linkers connecting the mCherry tag were (RSRITSLYKKAGFFQWSS) for the Rest4(+ECDLVG) and full-length Rest variants (Mapp et al., 2011), and (RSRITSLYKKAGFFQWGT) for the REST-VP16 construct (Immaneni et al., 2000). All constructs were sequence verified using primers to the Tol2 arms: for 5′-TTTGGCAAAGAATTCCTCGAC-3′ and rev 5′-CTTCGCAGATCTGATCTAGAG-3′. Nuclear localization and mutant variant signals were added to the carboxy-terminus of zebrafish Rest using the following reverse primers (IDT):
Rest-PKKKRKV (Rest-NLS): 5′-GTATGGATCCTCACACTTTGCGTTTCTTCTTGGGTTTGCCCCCCTGTGCCGC-3′
Rest-PAAARKV (Rest-mutNLS): 5′-GTATGGATCCTCACACTTTTCTAGCAGCGGCGGGTTTGCCCCCCTGTGCCGC-3′
Expression of Rest-mutNLS in FBMNs had no effect on Rest-deficient migration. These constructs were also cloned into pCS2+ for mRNA synthesis.
mRNA generation and microinjection
Capped mRNA was generated using the MEGAscript SP6 Kit (Ambion). Tol2 transposase mRNA (60 ng/μl) and Tol2 plasmid DNA constructs (40 ng/μl) were kept on ice and carefully injected into one-cell stage embryos, avoiding the yolk, using a new cold needle every 15 min.
Immunohistochemistry
Embryos were fixed in 4% paraformaldehyde (PFA; Sigma), then washed in PBS containing 0.5% Triton-X (PBST). Embryos were permeabilized in water and cold acetone, incubated in blocking buffer (1% DMSO, 10% goat serum, 0.5% Triton-X in 1X PBS), and rinsed in 1X PBST + 1% DMSO. Primary and secondary antibodies were diluted in 1X PBST + 1% DMSO. The following primary antibodies were used: GFP at 1:500 (Clontech 632381), EphA4a at 1:2000 (a generous gift from David Wilkinson (Irving et al., 1996), RFP at 1:200 (Millipore AB3216), and DsRed at 1:200 (Clontech 632496). The following secondary antibodies were used: Alexa Fluor 488 (Molecular Probes A11001), Alexa Fluor 546 (Molecular Probes A11010), and anti-GFP Alexa Fluor 488 conjugate (Molecular Probes A21311). Embryos were then cleared in 80% glycerol, deyolked and flat-mounted.
Cell transplantation
As homozygous restsbu29/sbu29 males are not reliable breeders, crosses between heterozygous mutant restsbu29/+ males and homozygous mutant restsbu29/sbu29 females carrying Tg(zcrest1:mRFP) were used to generate donor embryos. Donor embryos were injected at the one-cell stage with 10 KDa lysinated Alexa Fluor 647 dextran (Life Technologies D-22914; 0.5% in 0.2 M KCl). Blastula stage donor cells were then transplanted as previously described (Rohrschneider et al., 2007) into the hindbrain primordia of Tg(islet1:GFP) gastrula stage hosts that were wild type for rest. Each individual donor was subsequently genotyped by PCR to identify homozygous mutants. Mosaic specimens were raised until 48 hpf then immunolabeled for GFP, RFP and EphA4 (to label r3 and r5), with donor-derived cells visualized using Alexa Fluor 647 lineage tracer. Donors from rest+/+ siblings carrying Tg(zcrest1:mRFP) were used as controls. Embryos were cleared in 80% glycerol, deyolked and flat-mounted for imaging.
Probe synthesis and whole mount in situ hybridization
Fixed embryos were dehydrated into methanol and rehydrated in PBS + 1% Tween 20 prior to in situ hybridization, as previously described (Prince et al., 1998) using digoxigenin-labeled RNA probes (Roche 11277073910). Following the in situ protocol, embryos were post-fixed before and after immunohistochemistry. Partial or full-length cDNA clones were obtained by PCR from 30 or 48 hpf zebrafish cDNA, cloned into pGEM-T vector (Promega) and verified by sequencing. The following accession numbers and primers were used:
adam23a ({"type":"entrez-nucleotide","attrs":{"text":"XM_686527.5","term_id":"528481937","term_text":"XM_686527.5"}}XM_686527.5): for 5′-CTCTCGTAATAATGCACTGCGAC-3′
rev 5′-AGCGCTAGAGAAGCAGTTCTATG-3′
cmet ({"type":"entrez-nucleotide","attrs":{"text":"NM_001007124","term_id":"55742199","term_text":"NM_001007124"}}NM_001007124) (Elsen et al., 2009)
nrp1a ({"type":"entrez-nucleotide","attrs":{"text":"NM_001040326","term_id":"94536927","term_text":"NM_001040326"}}NM_001040326) (Yu et al., 2004)
nrp2b ({"type":"entrez-nucleotide","attrs":{"text":"NM_212966","term_id":"47085724","term_text":"NM_212966"}}NM_212966) (Yu et al., 2004)
snap25b ({"type":"entrez-nucleotide","attrs":{"text":"NM_131434","term_id":"18859394","term_text":"NM_131434"}}NM_131434) (Thisse and Thisse, 2004)
srrm4 ({"type":"entrez-nucleotide","attrs":{"text":"XM_001336887","term_id":"528478686","term_text":"XM_001336887"}}XM_001336887) (Calarco et al., 2009)
syt4 ({"type":"entrez-nucleotide","attrs":{"text":"NM_199948","term_id":"41152407","term_text":"NM_199948"}}NM_199948) for 5′-GTGACGACAGAAGAAGCACATTT-3′
rev 5′-ATTACTCGCGACATAAGTCCACA-3′
Results
Maternally provided Rest is required for normal FBMN migration
Previously, we reported characterization of FBMN migration in a zebrafish rest mutant generated using zinc-finger nucleases (Kok et al., 2012). The restsbu29 allele has a 7 bp deletion in the first exon that produces a frameshift upstream of the DNA binding domain to generate a predicted null mutant. Our analysis of homozygous mutant embryos produced from crosses of heterozygous restsbu29/+ adults revealed that FBMN migration is partially disrupted, with a subset of neurons remaining in r4 and r5 rather than completing migration to r6/r7. This mutant phenotype was consistent with the phenotype of rest knockdown embryos generated using a splice-blocking morpholino (MO) (Mapp et al., 2011). Since Rest is also maternally provided, we hypothesized that the additional loss of maternal Rest would further disrupt FBMN migration. To collect embryos devoid of both maternal and zygotic rest, we crossed heterozygous restsbu29/+ males with homozygous restsbu29/sbu29 females to produce maternal-zygotic (MZ) heterozygous and homozygous mutant embryos, which were individually genotyped using PCR (Kok et al., 2012) (Fig. 1A). We found that as Rest is progressively depleted, mutant embryos show increasing disruption of FBMN migration (Fig. 1B–F), with MZrestsbu29/sbu29 mutants showing the most severe abrogation of FBMN migration. To quantify FBMN migration phenotypes, we used EphA4a immunolabeling to visualize r5, allowing us to count the FBMNs in each rhombomere (Fig. 1B′-F′, G). We find that fewer FBMNs migrate to r6 as rest transcript is depleted, with roughly two-thirds of FBMNs failing to migrate to r6 in homozygous MZrestsbu29/sbu29 mutants. As a splice-blocking MO is not expected to knockdown maternal transcript, we designed a translation-blocking MO targeted against the Rest transcriptional start site. We found that the MZrestsbu29/sbu29 mutant is phenocopied by injection of a combination of translation and splice-blocking MOs (Fig. 1H–I). We also confirmed that the disruption to FBMN migration in MZrestsbu29/sbu29 mutants is not merely due to a developmental delay, as the phenotype persists in 72 hpf embryos (Fig. 1J–K). In summary, by removing maternal Rest in addition to the zygotic contribution, we have demonstrated that Rest plays a more significant role in FBMN migration than previously recognized.
(A) PCR genotyping of mating crosses to produce rest zygotic and maternal-zygotic (MZ) mutant embryos. The restsbu29 allele introduces a 7 bp deletion resulting in a smaller PCR product (red arrowhead). Homozygous mutant females (red box) produced eggs lacking maternal rest transcript. (B–F) Average intensity projection confocal images (dorsal views) of Tg(islet1:GFP); restsbu29 mutant embryos at 48 hpf, immunolabeled with EphA4a to allow visualization of r5 (red). Scale bar = 20 μm. Decreasing levels of rest from B to F correlate with increasing disruption in FBMN migration. (B′–F′) Pie charts represent the average number of FBMNs counted in r4 (black), r5 (checkered), and r6 (white) as a percentage of total FBMNs in rest mutant embryos. The total number of FBMNs was not significantly different in embryos of different genotypes. (G) Statistical analysis of FBMN migration in restsbu29 mutants. The percentages of FBMNs that fully migrate to r6 (white) were compared using unpaired t test. Parentheses indicate number of embryos analyzed per group. Mean ± SEM; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; n.s. = not significant. (H,I) Tg(islet1:GFP) embryos co-injected with both splice- and translation-blocking morpholinos targeted against Rest recapitulate the MZrestsbu29/sbu29 mutant phenotype at 48 hpf. Scale bar = 20 μm. (J,K) MZrestsbu29/sbu29 FBMNs fail to reach r6 at 72 hpf, indicating that the mutant phenotype is not merely a consequence of developmental delay.
Rest functions within FBMNs during their migration
Rest is present in non-neural cells and neuronal precursors, where it represses genes important for mature neuron function (Ballas and Mandel, 2005; Qureshi and Mehler, 2009). We have previously shown that Rest localizes within migrating FBMNs as well as in surrounding neuroepithelial cells (Mapp et al., 2011). To test whether Rest functions in the FBMNs to mediate their migration, as opposed to within the surrounding neuroepithelial cells, we specifically disrupted Rest function only in the FBMNs. For these experiments we used two different variants of Rest that interfere with its ability to repress targets (Fig. 2A). The first is a dominant-negative version of zebrafish Rest (Mapp et al., 2011), designed based on the sequence of the naturally occurring REST4 human splice variant, which disrupts Rest function and derepresses Rest targets (Hersh and Shimojo, 2003; Shimojo and Hersh, 2004; Tabuchi et al., 2002). Note that there is no evidence for an endogenous zebrafish Rest4 (Raj et al., 2011). The second is a variant of human REST in which both corepressor interaction domains have been removed and the DNA-binding domain is fused with the VP16 transcriptional activator domain (REST-VP16). This construct removes all repressive activity and transforms REST into an activator of its targets (Immaneni et al., 2000). The two Rest constructs were tagged with mCherry, and expressed under control of zCREST1, an enhancer from the islet1 regulatory elements (Uemura et al., 2005) and a minimal promoter to drive expression in cranial branchiomotor neurons (CBMNs) (Fig. 2B). These constructs were cloned between Tol2 sequences to allow random insertion into the genome via transient transgenesis using the Tol2-transposase system (Kawakami, 2007; Kawakami and Shima, 1999; Mapp et al., 2011). Each construct was coinjected with transposase mRNA into Tg(islet1:GFP) embryos such that all FBMNs were labeled with GFP, but only FBMNs in which the transgene was expressed were labeled with mCherry. As this transgenesis method creates mosaic embryos with random efficiency, we found variable numbers of FBMNs expressing mCherry between individual specimens. To analyze the effect of disrupted Rest function on FBMN migration, we grouped embryos into classes according to the number of mCherry+ FBMNs they possessed, using those with zero mCherry+ cells as controls. As the number of FBMNs containing mCherry-dnRest4 increased, we observed that neuron migration was progressively disrupted (Fig. 2C–G). Quantitative analysis of the locations of all FBMNs showed that nearly half failed to migrate to r6 in embryos containing over 30 mCherry+ FBMNs (Fig. 2C′-G′). Expressing the REST-VP16 activator construct within FBMNs caused an even more dramatic disruption to migration (Fig. 2H–L); nearly three-fourths of FBMNs failed to migrate to r6, and many more failed to migrate out of r4 in comparison to the dnRest4-expressing embryos (Fig. 2H′-L′). Interestingly, when we analyzed the location of mCherry+ FBMNs, we found that these transgene-expressing FBMNs were distributed across all rhombomeres (Fig. 2D″–G″, I″–L″). We suggest this reflects a community effect, also termed “collective migration”, in which FBMNs of one genotype influence migration of neurons of another genotype, presumably via neuron-neuron interactions (Rohrschneider et al., 2007; Walsh et al., 2011). Consistent with this hypothesis, in a previous study using a transient transgenic approach to generate mosaic FBMNs (Mapp et al., 2011), we similarly found that neurons of different genotypes influence one another’s behavior.
(A) Schematic of Rest protein domains, including eight zinc finger DNA binding domains, and corepressor binding domains at amino and carboxy termini. Dominant negative Rest4 (dnRest4) is a truncation of zebrafish Rest with ECDVLG (human sequence) added at the C-terminus (Mapp et al., 2011). REST-VP16 comprises human REST DNA-binding domain fused to VP16 activator with both corepressor binding domains removed (Immaneni et al., 2000). (B) Schematic of Tol2-mediated transgenesis to express mCherry-tagged constructs specifically in FBMNs. Coinjection of DNA constructs with transposase mRNA produces mosaic expression in Tg(islet1:GFP) FBMNs; migration was analyzed at 48 hpf, with r5 marked by EphA4a (red). Embryos classed according to number of mCherry+ FBMNs into five groups: zero (controls, C and H), 1–10, 11–20, 21–30, and >30; n = # embryos analyzed. (D–G) Increasing number of dnRest4 expressing FBMNs progressively disrupts migration. (I–L) Increasing number of REST-VP16 expressing FBMNs causes a greater disruption in migration. (C′–L′) Counts of Tg(islet1:GFP) FBMNs (green) in r4 (black), r5 (checkered) and r6 (white). A greater number of FBMNs remain in r4 and r5 as more FBMNs express each construct. (D″–L″) Counts of mCherry+ FBMNs (red/yellow) in each group of mCherry-expressing embryos. mCherry+ FBMNs are distributed in all three rhombomeres, suggesting FBMNs migrate as a collective group. Split channel images confirmed that all mCherry+ cells were also GFP+ as expected.
The results of our transient transgenic experiments support the hypothesis that Rest function is required within the FBMNs for their proper migration, as mosaic expression of these variants in FBMNs impedes collective migration. To test this hypothesis further we performed a limited set of cell transplantation experiments, generating genetic chimeras in which MZrestsbu29/sbu29 FBMNs were transplanted into wild type hindbrains (Cooper et al., 2003; Rohrschneider et al., 2007; Supplemental Fig. 1A). Consistent with Rest function being required in the FBMNs, 42% of transplanted MZrestsbu29/sbu29 FBMNs (10/24) failed to migrate to r6, although in control transplants between wild type specimens we also observed occasional donor-derived cells in r4 (similar to our previous reports, Rohrschneider et al., 2007; Supplemental Fig. 1F). In 6/8 chimeric specimens donor-derived MZrestsbu29/sbu29 FBMNs (red) were localized in r4 and/or r5, sent out long projections, and in some cases appeared to impede the migration of wild type host-derived FBMNs (green), consistent with Rest affecting collective migration (Supplemental Fig. 1B–D).
Together, the results of both our transient transgenic and cell transplantation experiments confirm that Rest function within the FBMNs is required for their proper migration. From these data, together with our previous analysis of Rest protein sub-cellular localization (Mapp et al., 2011), we hypothesize that Rest functions within the nuclei of FBMNs, as they migrate through r4 and r5, to repress maturation genes and maintain these neurons in a migratory state.
Aberrant neurite extension in Rest-deficient FBMNs
Migrating FBMNs in r4 and r5 display exploratory behavior, sending out protrusions that are quickly retracted (Mapp et al., 2010). Upon completion of neuron migration to r6, longer projections (neurites) become stabilized and extend both laterally and across the midline at 48 hpf (Fig. 3A, arrows). In MZrestsbu29/sbu29 embryos, however, we observed ectopic neurite outgrowth both laterally and across the midline from FBMNs localized in r4 and r5 at 48 hpf (Fig. 3B, arrows). Moreover, embryos expressing the mCherry-REST-VP16 construct in >30 FBMNs showed pervasive extension of ectopic neurites both laterally and across the midline (Fig. 3C–D′). The average lengths (Supplemental Table 1) and numbers of neurites extended in r4 (black), r5 (checkered), and r6 (white) were quantified for each rest mutant genotype (Fig. 3E) as well as for embryos with >30 FBMNs expressing the mCherry-dnRest4 (Fig. 3F) or mCherry-REST-VP16 (Fig. 3G) construct. Embryos with zero mCherry FBMNs were used as controls, and projections from GFP+ versus mCherry+ FBMNs were quantified independently, similar to analysis presented in Fig. 2. Overall we observed an increase in the number of r4 and r5 localized neurites as Rest function, and hence repression of Rest targets, was reduced. The greatest numbers of ectopic neurites occurred in mCherry-REST-VP16 mosaic specimens, in which Rest targets should be prematurely activated in the FBMNs. The number of mCherry-dnRest4+ FBMN projections localized in r4 and r5 was greater than the number of GFP+ projections from neurons that lack expression of dnRest4 (Fig. 3F). In contrast, we observed ectopic neurite extension even from GFP+ FBMNs in specimens expressing the mCherry-REST-VP16 construct (Fig. 3G), suggesting mCherry+ FBMNs can influence behavior of neighboring FBMNs. Many of the neurites we observed were Map2-positive, indicating that they are dendrites (Fig. 3H–J). We also quantified the numbers of lateral versus medial projections from FBMNs in specific rhombomeres. We found that similar to ectopic neurite numbers, increasing proportions of lateral versus medial projections correlated with reduced Rest function (Supplemental Table 1). Specifically, wild type FBMNs possess more medial than lateral projections during their migration, with the proportion of lateral projections increasing markedly as FBMNs reach r6. By contrast, in rest mutants and in specimens expressing dnRest4 or REST-VP16, lateral FBMN protrusions outnumber medial protrusions even for neurons localized in r4 or r5. Lastly, we found that during stages of active FBMN migration (24 hpf), long stable neurites are absent in wild type but present in MZrestsbu29/sbu29 embryos, suggesting their formation may inhibit proper migration (Fig. 3K–L). These findings are also consistent with previous reports that Rest negatively regulates neurite outgrowth (Lepagnol-Bestel et al., 2007). Since wild type FBMNs form stable neurites only when they have reached r6 (Fig. 3A), neurite extension from Rest-deficient FBMNs in r4 and r5 reveals that mis-localized Rest-deficient FBMNs possess aspects of mature neuronal morphology despite their inappropriate anteroposterior location in the hindbrain.
(A,B) Average intensity projection confocal images (dorsal views) of Tg(islet1:GFP)-expressing FBMNs at 48 hpf. Scale bar = 10 μm. (A) In wild type specimens protrusions across the midline and laterally (arrows) occur from FBMNs localized in r6. By contrast, in MZrestsbu29/sbu29 mutant specimens protrusions laterally and across the midline frequently occur from FBMNs localized in r4 and r5 (B). (C–D′) Ectopic protrusions (arrows) in mCherry-REST-VP16 mosaic embryos are primarily sent out by FBMNs expressing the construct, as indicated by mCherry expression (D′). (E–G) Average number of neurites extended in r4 (black), r5 (checkered), and r6 (white) by FBMNs in rest mutant embryos (E) and those mosaically expressing dnRest4 (F) or REST-VP16 (G). (H–J) FBMNs expressing mosaic mCherry-REST-VP16 send out ectopic protrusions in r4 and r5 that are also Map2-positive, showing that dendritic branching occurs in inappropriate locations (arrows). (K,L) Lateral views of Tg(zCREST1:membRFP)-expressing FBMNs at 24 hpf. FBMNs in MZrestsbu29/sbu29 embryos send out long protrusions during migration (arrow). Scale bar = 10 μm.
Rest-deficient FBMNs show precocious morphological maturation
As Rest functions to repress neuronal maturation genes, we hypothesized that the partial abrogation of FBMN migration observed in Rest mutants might be due to precocious neuronal maturation. Migrating neurons are typically elongated in the direction of travel, and send out few short protrusions into their environment (Komuro and Rakic, 1998; Mapp et al., 2010; Marin et al., 2006). Once neurons reach their destination they undergo morphological changes as they complete their maturation program: the cytoplasmic volume increases and the neurons take on a more circular shape while neurites are extended to integrate with surrounding neural circuitry (Hatten, 1999; Jobe et al., 2012; Komuro and Rakic, 1998). We hypothesized that Rest-deficient FBMNs might show characteristics of mature neuron morphology at stages and positions when wild type FBMNs have immature morphology, indicating their precocious maturation. Wild type FBMNs mature over developmental time as they migrate, showing morphological differences as they move from r4 to r5 and r6 (Mapp et al., 2010). To analyze morphological maturation, we used the Tg(zCREST1:membRFP) line to visualize details of FBMN morphology (Fig. 4). We compared the FBMNs of MZrestsbu29/sbu29 embryos with those of wild type embryos collected from crosses between wild type siblings of rest mutants. We measured length-to-width ratios of cell bodies to reveal how “round” neurons were, as well as the number and length of protrusions they extended. As migration of MZrestsbu29/sbu29 mutant FBMNs is abrogated, we cannot compare mutant versus wild type neurons in specific rhombomeric locations, we therefore elected to analyze all FBMNs at two different stages of migration to ensure that the only variable in the analysis is time. The analysis was performed on FBMNs at 20 hpf and at 28 hpf; in each case we averaged the measurements of all FBMNs from five embryos of each genotype.
(A–G) Analysis of FBMNs at 20 and 28 hpf in Tg(zCREST1:membRFP) wild type embryos (A, B) and MZrestsbu29/sbu29 embryos (C, D). Average measurements from 5 embryos are shown for wild type and MZrestsbu29/sbu29 (E–G) embryos at each stage. The wild type length-to-width ratio decreases from 20 to 28 hpf, as neurons become rounder; by contrast MZrestsbu29/sbu29 FBMNs are round at both 20 and 28 hpf (E). Both number and length of protrusions increase over time in wild type and MZrestsbu29/sbu29 embryos (F,G). MZrestsbu29/sbu29 FBMNs have significantly more and longer protrusions than wild type FBMNs. (H–J) Volume measurements of FBMNs at 22 hpf. Whole cell and nuclear volume were measured for 10 “immature” neurons in r4 (yellow dots) versus 10 “mature” neurons in r6 (red dots) for 5 wild type (H) and 5 MZrestsbu29/sbu29 (I) embryos. Wild type FBMN whole cell volume increases significantly as neurons migrate from r4 to r6; MZrestsbu29/sbu29 FBMN whole cell volume does not alter as neurons migrate from r4 to r6 and is similar to that of wild type neurons in r6. Nuclear volume does not change significantly. Mean ± SEM; *P < 0.05; ***P < 0.001, ****P < 0.0001, n.s. = not significant.
Consistent with our previous descriptions, we find that wild type FBMNs are elongated at 20 hpf (Fig. 4A) and become more circular by 28 hpf (Fig. 4B,E). The neurons also send out a greater number of protrusions, and longer protrusions, at the later stage compared to early migrating neurons (Fig. 4F–G). By contrast, in Rest-deficient embryos FBMNs are relatively circular at both early and late stages (Fig. 4C–E), and both the number and length of protrusions are greater than in wild type neurons (Fig. 4F–G).
We next measured cytoplasmic volume as the neurons matured using 3D confocal reconstructions. This analysis focused on comparing immature r4-localized FBMNs to mature r6-localized FBMNs at a single stage of migration, 22 hpf. We selected this stage because it is before radial migration commences in r6, after which neuron accumulation along the dorsoventral axis obscures imaging. We measured the volume of TO-PRO-3-labeled nuclei as well as the membRFP-labeled whole cell volume in the ten most anterior and ten most posterior FBMNs in five embryos (yellow and red dots, respectively; Fig 4H–I). While nuclear volume remained unchanged in all neurons, the whole cell (or cytoplasmic) volume of wild type neurons was significantly larger in r6-localized FBMNs compared to r4-localized FBMNs, consistent with mature neurons that have completed migration increasing in size (Fig. 4J). In Rest-deficient embryos, FBMNs in r4 are significantly larger than in wild type, and are comparable in size to mature r6 FBMNs. This suggests that synthesis of new proteins and subsequent cytoplasmic swelling occurs prematurely in r4-localized FBMNs in the absence of Rest repression. While a subset of the Rest-deficient neurons located in r4 might arguably be “older” neurons that have failed to migrate, our quantification reveals that average volumes of r4 neurons vary little, implying that even recently born neurons are unusually large. The question arises of whether FBMNs in Rest-deficient specimens are already unusually large at the moment of their birth; however, current technical limitations prevent us from addressing this point, as we are unable to visualize newborn FBMNs until sufficient fluorescent reporter protein has accrued. In summary, our results are consistent with the hypothesis that Rest-deficient FBMNs precociously develop mature morphology at a premature developmental stage relative to wild type FBMNs. From this analysis we conclude that functional Rest is required to maintain the appropriate shape and size of FBMNs during migratory stages, suggesting in turn that the lack of “immature” morphology in Rest-deficient FBMNs contributes to the disruption of their migration.
Rest target genes are precociously expressed in Rest and Pk1b mutants
In addition to precocious morphological maturation, we hypothesized that FBMNs might undergo precocious molecular maturation in the absence of Rest. We therefore tested whether Rest target genes, which are normally repressed, are prematurely expressed in Rest-deficient FBMNs during stages when the neurons should be migrating. To validate the hypothesis that Rest must be localized to the nucleus to act, we also analyzed Rest target gene expression under conditions where Rest fails to localize to the nucleus. We previously established that nuclear localization of Rest in zebrafish FBMNs is dependent on the function of Prickle1b (Pk1b), revealing a planar cell polarity (PCP)-independent role for Pk1b at the nuclear membrane (Mapp et al., 2011). Significantly, in Pk1b-deficient embryos Rest is present but not localized to the nucleus where it can act.
We began this analysis by taking a candidate approach, initially screening expression data available on the ZFIN database (zfin.org) for Rest target genes identified in previous studies (Bruce et al., 2004; Johnson et al., 2007; Johnson et al., 2008; Mortazavi et al., 2006; Sun et al., 2005). We then used in situ hybridization to identify zebrafish Rest target genes that lack expression in FBMNs at 24 hpf, when the migration process is still underway, but that are expressed in FBMNs at 48 hpf, when migration is complete and the Rest repressor has been depleted from the nucleus (Mapp et al., 2011). Having identified such genes, we went on to compare their expression in wild type rest siblings with expression in MZrestsbu29/sbu29 mutants and pk1bfh122/fh122 mutants at 24 hpf. We analyzed four Rest target genes: adam23a (Fig. 5A–A′), snap25b (Fig. 5D–D′), srrm4 (Fig. 5G–G′), and syt4 (Fig. 5J–J′). In 24 hpf MZrestsbu29/sbu29 embryos we found inappropriate expression of all four genes in the FBMNs (Fig. 5B–B′, E–E′, H–H′, K–K′, arrows). Interestingly, a slight elevation in expression levels was also seen in surrounding tissues, suggesting that Rest may additionally regulate expression of these targets in other cell types. In 24 hpf pk1bfh122/fh122 mutant embryos, in which all FBMNs remain in r4 (Mapp et al., 2011), we observed robust expression of all four Rest targets in the FBMNs (Fig. 5C–C′, F–F′, I–I′, L–L′, arrows), in accord with our model that Rest nuclear localization is required for its function. This precocious molecular maturation of pk1bfh122/fh122 FBMNs is also consistent with our previous report that the non-migratory FBMNs display morphological and behavioral characteristics of wild type neurons that have reached r6 (Mapp et al., 2010). While Rest target gene expression levels appear higher in pk1bfh122/fh122 mutant FBMNs than in MZrestsbu29/sbu29 mutant FBMNs, this may primarily be a consequence of the FBMNs being clustered together in r4 of Pk1b-deficient embryos; while we performed in situ hybridization experiments in parallel the technique is inherently non-quantitative. We also found premature expression of FBMN-expressed genes that have not been identified as direct targets of Rest, such as cmet (Fig. 5M-O′; Elsen et al., 2009) and nrp1a/nrp2b (data not shown; Yu et al., 2004). These data suggest that changes in gene expression are not limited to Rest targets, and indirect activation of multiple genes may contribute to the precocious maturation of Rest-deficient FBMNs. In summary, we confirm that in the absence of Rest function FBMNs show expression of neuronal maturation genes at inappropriately early developmental stages; this precocious molecular maturation may in turn drive observed changes in morphology and prevent full migration of many neurons to r6.
In situ hybridization analysis of Rest target gene expression in Tg(islet1:GFP) embryos at 24 hpf in wild type, MZrestsbu29/sbu29, and pk1bfh122/fh122 specimens. Rest targets selected for analysis are adam23a (A–C′), snap25b (D–F′), srrm4 (G–I′), and syt4 (J–L′). We also analyzed the non-Rest target and FBMN maturation marker cmet (M–O′). All five genes are expressed in fully migrated FBMNs at 48 hpf. Upper panels show co-expression with islet1:GFP to label FBMNs, with gene expression alone shown below. At this stage (24 hpf), the selected Rest targets, as well as the maturation marker cmet, are not expressed in wild type FBMNs, but are expressed in both rest and pk1b mutant FBMNs (arrows).
Rest target Srrm4 functions downstream of Rest in FBMN maturation
Rest target Srrm4, also known as nSR100, is an alternative splicing factor required for the inclusion of neuron-specific exons in the mature mRNAs of multiple neural-specific transcripts (Calarco et al., 2009). The srrm4 gene is a direct target of Rest repression (Calarco et al., 2009), and in mammals is also proposed to act in a feedback loop to negatively regulate Rest repression while promoting neuron-specific pre-mRNA processing (Raj et al., 2011). Based on its function, we hypothesized that Srrm4 might function as a key downstream target of Rest to drive neuronal maturation in zebrafish FBMNs. According to this model (Fig. 6A), in wild type migrating neurons Rest functions in the FBMN nucleus to actively repress srrm4 expression, and lack of Srrm4 prevents neuronal maturation to help maintain neurons in an immature state compatible with migration. As Srrm4 is not normally expressed in migrating neurons, we predicted no deficits in FBMN migration in response to Srrm4 deficiency; this prediction was confirmed by knockdown of srrm4 (Fig. 6B–C). By contrast, in Rest-deficient embryos srrm4 is prematurely expressed in migrating FBMNs (Fig. 5G–H′), where we hypothesize it plays a key role in promoting neuronal maturation to cause incomplete neuronal migration (Fig. 6A, D). To test this hypothesis, we investigated whether the precocious maturation and partial migration of Rest mutant FBMNs could be rescued by knocking down Srrm4. We found that there was indeed a partial rescue of FBMN migration in MZrestsbu29/sbu29 specimens injected with Srrm4 MO (as described, Calarco et al., 2009), in comparison to uninjected siblings (Fig. 6D–E). FBMN migration was again quantified using the r5 marker EphA4a, showing that significantly more Srrm4-deficient neurons completed migration to r6 (Fig. 6F).
(A) Model of Rest/Srrm4 regulation of FBMN maturation and the effect on FBMN migration. (B–E) Average projection dorsal views of 48 hpf FBMNs (green) in Tg(islet1:GFP) embryos immunolabeled with EphA4a to mark r5 (red). FBMN migration is unaffected in Srrm4 morphants (C) as compared to uninjected controls (B). MZrestsbu29/sbu29 mutant FBMN migration (D) is partially rescued with knockdown of Srrm4 (E). (F) Quantitative counts of FBMN migration in r4 (black), r5 (checkered), and r6 (white) as % of all FBMNs. A significantly increased number of FBMNs migrate to r6 in MZrestsbu29/sbu29 mutants injected with Srrm4 MO as compared to uninjected mutants. Mean ± SEM; ****P < 0.0001. (G–J) Analysis of FBMNs at 22 and 30 hpf in Tg(zCREST1:membRFP);MZrestsbu29/sbu29 embryos (G,H), and Tg(zCREST1:membRFP);MZrestsbu29/sbu29 embryos injected with Srrm4 MO (I,J). Average morphological measurements from 5 embryos are shown at each stage (K–N). Mutants injected with Srrm4 MO have FBMNs that are more elongated than those injected with Control MO, and remain so at 30 hpf (K). They also send out fewer and shorter protrusions than MZrestsbu29/sbu29 at both early and late stages (L–M). Mutants injected with Srrm4 MO have a smaller cytoplasmic volume compared to mutant controls at 22 hpf, which swells significantly at the later stage compared to the unchanged volume in MZrestsbu29/sbu29 mutants (N).
We further analyzed the morphology of FBMNs in Tg(zCREST1:membRFP);MZrestsbu29/sbu29 mutants injected with Srrm4 MO compared with mutants injected with a standard control MO. Analysis was performed at early and late stages of FBMN migration: 22 and 30 hpf (Fig. 6G–J). While FBMNs of MZrest mutants are round in shape, whether in r4 or r6 (Fig. 6K, and see also Fig. 4E), the FBMNs of mutants injected with Srrm4 MO are more elongated at both stages (Fig. 6K). Additionally, Srrm4 MO-injected mutant embryos have FBMNs that send out fewer and shorter protrusions compared to controls (Fig. 6L–M), and notably, these behaviors persist at the later stage. The cytoplasmic volume of FBMNs in control mutant specimens is elevated at both stages (Fig. 6N; as also shown in Fig. 4J). However, the cytoplasmic volume of r4 localized FBMNs in Srrm4 MO-injected mutant embryos is similar to that observed in wild type specimens (compare Fig. 6N with Fig. 4J); this volume increases as FBMNs migrate from r4 to r6, again similar to wild type (Fig. 6N; Fig. 4J). Our observation that depletion of Srrm4 function allows MZrest mutant FBMNs to maintain an elongated cell shape and low protrusion number and length through later developmental stages, strongly suggests that Srrm4 is an important regulator of neuronal morphology. In summary, our findings suggest that processing of neuron-specific pre-mRNAs by Srrm4 is important for FBMN maturation, and further support our model that appropriate control of the neuronal maturation process is essential for proper migration.
Nuclear-localized Rest can rescue migration in Rest-deficient but not Pk1b-deficient embryos
Central to our overall model of FBMN migration is the prediction that Rest must be localized to FBMN nuclei to repress its target genes and prevent precocious maturation. To test this prediction, we examined whether nuclear-localized Rest can rescue migration of FBMNs in Rest-deficient embryos. To localize Rest to the nucleus we engineered a variant of Rest carrying the nuclear localization sequence (NLS) PKKKRKV from the SV40 Large T antigen (Rai et al., 2007) at its C-terminus (Rest-NLS; Fig. 7A). Analysis of blastula-stage embryos injected with mRNA encoding mCherry-tagged Rest-NLS confirmed the expected localization of Rest protein to the cell nucleus (Fig. 7B–D). To test whether mCherry-Rest-NLS could rescue neuronal migration, we expressed the construct specifically in FBMNs using the zCREST1 enhancer and Tol2 transgenesis (as described above; Fig. 2B). As in our previous experiments, we analyzed groups of embryos with similar numbers of mCherry+ FBMNs expressing Rest-NLS (Fig. 7E–H); embryos with zero mCherry+ FBMNs functioned as controls. We found that as the number of MZrestsbu29/sbu29 mutant FBMNs expressing mCherry-Rest-NLS increased, a greater number of FBMNs migrated to r6 (Fig. 7E′–H′). In embryos with over 20 FBMNs expressing mCherry-Rest-NLS the majority of neurons migrated to r6 (Fig. 7F″–H″), again suggesting that neurons migrate in a collective manner. Essentially, collective migration allows migratory FBMNs containing nuclear-localized Rest to “pull along” their Rest-deficient and otherwise non-migratory neighbors to r6. In summary, we have effectively targeted Rest to the nuclei of FBMNs, and confirmed that nuclear function of Rest promotes FBMN migration through the hindbrain.
(A) Schematic of Rest protein domains with the added NLS sequence PKKKRKV at the C-terminus (Rest-NLS). This construct was tagged with mCherry and cloned into both pCS2+ and Tol2 vectors. (B–D) Expression of mCherry-tagged Rest-NLS (injected mRNA) in blastula stage embryos (5 hpf), compared with expression of Rest protein expression using immunohistochemistry. Rest-NLS is correctly localized to the nucleus while endogenous Rest is expressed in the cytoplasm and nucleus (Kok et al., 2012). (E–H) Expression of mCherry-Rest-NLS specifically in FBMNs using Tol2 transgenesis (as described in Fig. 2) in Tg(islet1:GFP);MZrestsbu29/sbu29 mutant embryos. Embryos were classed into four groups based on the number of mCherry+ FBMNs present: zero (controls, E), 1–10, 11–20, and 21–30. (F–H) Increasing the number of FBMNs expressing Rest-NLS rescues FBMN migration to r6. (E′–H′) Counts of Tg(islet1:GFP) FBMNs (green) in r4 (black), r5 (checkered) and r6 (white). (F″–H″) Counts of mCherry+ FBMNs (red/yellow) in each group of mCherry-expressing embryos. mCherry+ FBMNs are more likely to reside in r6, suggesting their ability to “pull along” mutant FBMNs lacking the construct to complete migration. Split channel images confirmed that mCherry+ cells were also GFP+ as expected. (I–K) Rest-NLS expression within Pk1b mutant FBMNs using Tol2 transgenesis does not rescue migration out of r4.
Finally, to examine to what extent Rest is responsible for mediating the function of Pk1b in FBMN migration, we tested whether nuclear-localized Rest could rescue FBMN migration in Pk1b-deficient embryos. We found that even high numbers of mCherry-Rest-NLS-expressing FBMNs failed to rescue neuronal migration in pk1bfh122/fh122 mutants (Fig. 7I–J′). Thus nuclear-localized Rest is not sufficient to rescue FBMN migration in the absence of Pk1b function. This finding indicates that Pk1b is playing two separable roles in FBMN migration: one role is to properly localize Rest to the nucleus, thus preventing premature maturation of FBMNs, while the second role is independent of Rest and is likely critical to initiation of FBMN migration out of r4.
Maternally provided Rest is required for normal FBMN migration
Previously, we reported characterization of FBMN migration in a zebrafish rest mutant generated using zinc-finger nucleases (Kok et al., 2012). The restsbu29 allele has a 7 bp deletion in the first exon that produces a frameshift upstream of the DNA binding domain to generate a predicted null mutant. Our analysis of homozygous mutant embryos produced from crosses of heterozygous restsbu29/+ adults revealed that FBMN migration is partially disrupted, with a subset of neurons remaining in r4 and r5 rather than completing migration to r6/r7. This mutant phenotype was consistent with the phenotype of rest knockdown embryos generated using a splice-blocking morpholino (MO) (Mapp et al., 2011). Since Rest is also maternally provided, we hypothesized that the additional loss of maternal Rest would further disrupt FBMN migration. To collect embryos devoid of both maternal and zygotic rest, we crossed heterozygous restsbu29/+ males with homozygous restsbu29/sbu29 females to produce maternal-zygotic (MZ) heterozygous and homozygous mutant embryos, which were individually genotyped using PCR (Kok et al., 2012) (Fig. 1A). We found that as Rest is progressively depleted, mutant embryos show increasing disruption of FBMN migration (Fig. 1B–F), with MZrestsbu29/sbu29 mutants showing the most severe abrogation of FBMN migration. To quantify FBMN migration phenotypes, we used EphA4a immunolabeling to visualize r5, allowing us to count the FBMNs in each rhombomere (Fig. 1B′-F′, G). We find that fewer FBMNs migrate to r6 as rest transcript is depleted, with roughly two-thirds of FBMNs failing to migrate to r6 in homozygous MZrestsbu29/sbu29 mutants. As a splice-blocking MO is not expected to knockdown maternal transcript, we designed a translation-blocking MO targeted against the Rest transcriptional start site. We found that the MZrestsbu29/sbu29 mutant is phenocopied by injection of a combination of translation and splice-blocking MOs (Fig. 1H–I). We also confirmed that the disruption to FBMN migration in MZrestsbu29/sbu29 mutants is not merely due to a developmental delay, as the phenotype persists in 72 hpf embryos (Fig. 1J–K). In summary, by removing maternal Rest in addition to the zygotic contribution, we have demonstrated that Rest plays a more significant role in FBMN migration than previously recognized.
(A) PCR genotyping of mating crosses to produce rest zygotic and maternal-zygotic (MZ) mutant embryos. The restsbu29 allele introduces a 7 bp deletion resulting in a smaller PCR product (red arrowhead). Homozygous mutant females (red box) produced eggs lacking maternal rest transcript. (B–F) Average intensity projection confocal images (dorsal views) of Tg(islet1:GFP); restsbu29 mutant embryos at 48 hpf, immunolabeled with EphA4a to allow visualization of r5 (red). Scale bar = 20 μm. Decreasing levels of rest from B to F correlate with increasing disruption in FBMN migration. (B′–F′) Pie charts represent the average number of FBMNs counted in r4 (black), r5 (checkered), and r6 (white) as a percentage of total FBMNs in rest mutant embryos. The total number of FBMNs was not significantly different in embryos of different genotypes. (G) Statistical analysis of FBMN migration in restsbu29 mutants. The percentages of FBMNs that fully migrate to r6 (white) were compared using unpaired t test. Parentheses indicate number of embryos analyzed per group. Mean ± SEM; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; n.s. = not significant. (H,I) Tg(islet1:GFP) embryos co-injected with both splice- and translation-blocking morpholinos targeted against Rest recapitulate the MZrestsbu29/sbu29 mutant phenotype at 48 hpf. Scale bar = 20 μm. (J,K) MZrestsbu29/sbu29 FBMNs fail to reach r6 at 72 hpf, indicating that the mutant phenotype is not merely a consequence of developmental delay.
Rest functions within FBMNs during their migration
Rest is present in non-neural cells and neuronal precursors, where it represses genes important for mature neuron function (Ballas and Mandel, 2005; Qureshi and Mehler, 2009). We have previously shown that Rest localizes within migrating FBMNs as well as in surrounding neuroepithelial cells (Mapp et al., 2011). To test whether Rest functions in the FBMNs to mediate their migration, as opposed to within the surrounding neuroepithelial cells, we specifically disrupted Rest function only in the FBMNs. For these experiments we used two different variants of Rest that interfere with its ability to repress targets (Fig. 2A). The first is a dominant-negative version of zebrafish Rest (Mapp et al., 2011), designed based on the sequence of the naturally occurring REST4 human splice variant, which disrupts Rest function and derepresses Rest targets (Hersh and Shimojo, 2003; Shimojo and Hersh, 2004; Tabuchi et al., 2002). Note that there is no evidence for an endogenous zebrafish Rest4 (Raj et al., 2011). The second is a variant of human REST in which both corepressor interaction domains have been removed and the DNA-binding domain is fused with the VP16 transcriptional activator domain (REST-VP16). This construct removes all repressive activity and transforms REST into an activator of its targets (Immaneni et al., 2000). The two Rest constructs were tagged with mCherry, and expressed under control of zCREST1, an enhancer from the islet1 regulatory elements (Uemura et al., 2005) and a minimal promoter to drive expression in cranial branchiomotor neurons (CBMNs) (Fig. 2B). These constructs were cloned between Tol2 sequences to allow random insertion into the genome via transient transgenesis using the Tol2-transposase system (Kawakami, 2007; Kawakami and Shima, 1999; Mapp et al., 2011). Each construct was coinjected with transposase mRNA into Tg(islet1:GFP) embryos such that all FBMNs were labeled with GFP, but only FBMNs in which the transgene was expressed were labeled with mCherry. As this transgenesis method creates mosaic embryos with random efficiency, we found variable numbers of FBMNs expressing mCherry between individual specimens. To analyze the effect of disrupted Rest function on FBMN migration, we grouped embryos into classes according to the number of mCherry+ FBMNs they possessed, using those with zero mCherry+ cells as controls. As the number of FBMNs containing mCherry-dnRest4 increased, we observed that neuron migration was progressively disrupted (Fig. 2C–G). Quantitative analysis of the locations of all FBMNs showed that nearly half failed to migrate to r6 in embryos containing over 30 mCherry+ FBMNs (Fig. 2C′-G′). Expressing the REST-VP16 activator construct within FBMNs caused an even more dramatic disruption to migration (Fig. 2H–L); nearly three-fourths of FBMNs failed to migrate to r6, and many more failed to migrate out of r4 in comparison to the dnRest4-expressing embryos (Fig. 2H′-L′). Interestingly, when we analyzed the location of mCherry+ FBMNs, we found that these transgene-expressing FBMNs were distributed across all rhombomeres (Fig. 2D″–G″, I″–L″). We suggest this reflects a community effect, also termed “collective migration”, in which FBMNs of one genotype influence migration of neurons of another genotype, presumably via neuron-neuron interactions (Rohrschneider et al., 2007; Walsh et al., 2011). Consistent with this hypothesis, in a previous study using a transient transgenic approach to generate mosaic FBMNs (Mapp et al., 2011), we similarly found that neurons of different genotypes influence one another’s behavior.
(A) Schematic of Rest protein domains, including eight zinc finger DNA binding domains, and corepressor binding domains at amino and carboxy termini. Dominant negative Rest4 (dnRest4) is a truncation of zebrafish Rest with ECDVLG (human sequence) added at the C-terminus (Mapp et al., 2011). REST-VP16 comprises human REST DNA-binding domain fused to VP16 activator with both corepressor binding domains removed (Immaneni et al., 2000). (B) Schematic of Tol2-mediated transgenesis to express mCherry-tagged constructs specifically in FBMNs. Coinjection of DNA constructs with transposase mRNA produces mosaic expression in Tg(islet1:GFP) FBMNs; migration was analyzed at 48 hpf, with r5 marked by EphA4a (red). Embryos classed according to number of mCherry+ FBMNs into five groups: zero (controls, C and H), 1–10, 11–20, 21–30, and >30; n = # embryos analyzed. (D–G) Increasing number of dnRest4 expressing FBMNs progressively disrupts migration. (I–L) Increasing number of REST-VP16 expressing FBMNs causes a greater disruption in migration. (C′–L′) Counts of Tg(islet1:GFP) FBMNs (green) in r4 (black), r5 (checkered) and r6 (white). A greater number of FBMNs remain in r4 and r5 as more FBMNs express each construct. (D″–L″) Counts of mCherry+ FBMNs (red/yellow) in each group of mCherry-expressing embryos. mCherry+ FBMNs are distributed in all three rhombomeres, suggesting FBMNs migrate as a collective group. Split channel images confirmed that all mCherry+ cells were also GFP+ as expected.
The results of our transient transgenic experiments support the hypothesis that Rest function is required within the FBMNs for their proper migration, as mosaic expression of these variants in FBMNs impedes collective migration. To test this hypothesis further we performed a limited set of cell transplantation experiments, generating genetic chimeras in which MZrestsbu29/sbu29 FBMNs were transplanted into wild type hindbrains (Cooper et al., 2003; Rohrschneider et al., 2007; Supplemental Fig. 1A). Consistent with Rest function being required in the FBMNs, 42% of transplanted MZrestsbu29/sbu29 FBMNs (10/24) failed to migrate to r6, although in control transplants between wild type specimens we also observed occasional donor-derived cells in r4 (similar to our previous reports, Rohrschneider et al., 2007; Supplemental Fig. 1F). In 6/8 chimeric specimens donor-derived MZrestsbu29/sbu29 FBMNs (red) were localized in r4 and/or r5, sent out long projections, and in some cases appeared to impede the migration of wild type host-derived FBMNs (green), consistent with Rest affecting collective migration (Supplemental Fig. 1B–D).
Together, the results of both our transient transgenic and cell transplantation experiments confirm that Rest function within the FBMNs is required for their proper migration. From these data, together with our previous analysis of Rest protein sub-cellular localization (Mapp et al., 2011), we hypothesize that Rest functions within the nuclei of FBMNs, as they migrate through r4 and r5, to repress maturation genes and maintain these neurons in a migratory state.
Aberrant neurite extension in Rest-deficient FBMNs
Migrating FBMNs in r4 and r5 display exploratory behavior, sending out protrusions that are quickly retracted (Mapp et al., 2010). Upon completion of neuron migration to r6, longer projections (neurites) become stabilized and extend both laterally and across the midline at 48 hpf (Fig. 3A, arrows). In MZrestsbu29/sbu29 embryos, however, we observed ectopic neurite outgrowth both laterally and across the midline from FBMNs localized in r4 and r5 at 48 hpf (Fig. 3B, arrows). Moreover, embryos expressing the mCherry-REST-VP16 construct in >30 FBMNs showed pervasive extension of ectopic neurites both laterally and across the midline (Fig. 3C–D′). The average lengths (Supplemental Table 1) and numbers of neurites extended in r4 (black), r5 (checkered), and r6 (white) were quantified for each rest mutant genotype (Fig. 3E) as well as for embryos with >30 FBMNs expressing the mCherry-dnRest4 (Fig. 3F) or mCherry-REST-VP16 (Fig. 3G) construct. Embryos with zero mCherry FBMNs were used as controls, and projections from GFP+ versus mCherry+ FBMNs were quantified independently, similar to analysis presented in Fig. 2. Overall we observed an increase in the number of r4 and r5 localized neurites as Rest function, and hence repression of Rest targets, was reduced. The greatest numbers of ectopic neurites occurred in mCherry-REST-VP16 mosaic specimens, in which Rest targets should be prematurely activated in the FBMNs. The number of mCherry-dnRest4+ FBMN projections localized in r4 and r5 was greater than the number of GFP+ projections from neurons that lack expression of dnRest4 (Fig. 3F). In contrast, we observed ectopic neurite extension even from GFP+ FBMNs in specimens expressing the mCherry-REST-VP16 construct (Fig. 3G), suggesting mCherry+ FBMNs can influence behavior of neighboring FBMNs. Many of the neurites we observed were Map2-positive, indicating that they are dendrites (Fig. 3H–J). We also quantified the numbers of lateral versus medial projections from FBMNs in specific rhombomeres. We found that similar to ectopic neurite numbers, increasing proportions of lateral versus medial projections correlated with reduced Rest function (Supplemental Table 1). Specifically, wild type FBMNs possess more medial than lateral projections during their migration, with the proportion of lateral projections increasing markedly as FBMNs reach r6. By contrast, in rest mutants and in specimens expressing dnRest4 or REST-VP16, lateral FBMN protrusions outnumber medial protrusions even for neurons localized in r4 or r5. Lastly, we found that during stages of active FBMN migration (24 hpf), long stable neurites are absent in wild type but present in MZrestsbu29/sbu29 embryos, suggesting their formation may inhibit proper migration (Fig. 3K–L). These findings are also consistent with previous reports that Rest negatively regulates neurite outgrowth (Lepagnol-Bestel et al., 2007). Since wild type FBMNs form stable neurites only when they have reached r6 (Fig. 3A), neurite extension from Rest-deficient FBMNs in r4 and r5 reveals that mis-localized Rest-deficient FBMNs possess aspects of mature neuronal morphology despite their inappropriate anteroposterior location in the hindbrain.
(A,B) Average intensity projection confocal images (dorsal views) of Tg(islet1:GFP)-expressing FBMNs at 48 hpf. Scale bar = 10 μm. (A) In wild type specimens protrusions across the midline and laterally (arrows) occur from FBMNs localized in r6. By contrast, in MZrestsbu29/sbu29 mutant specimens protrusions laterally and across the midline frequently occur from FBMNs localized in r4 and r5 (B). (C–D′) Ectopic protrusions (arrows) in mCherry-REST-VP16 mosaic embryos are primarily sent out by FBMNs expressing the construct, as indicated by mCherry expression (D′). (E–G) Average number of neurites extended in r4 (black), r5 (checkered), and r6 (white) by FBMNs in rest mutant embryos (E) and those mosaically expressing dnRest4 (F) or REST-VP16 (G). (H–J) FBMNs expressing mosaic mCherry-REST-VP16 send out ectopic protrusions in r4 and r5 that are also Map2-positive, showing that dendritic branching occurs in inappropriate locations (arrows). (K,L) Lateral views of Tg(zCREST1:membRFP)-expressing FBMNs at 24 hpf. FBMNs in MZrestsbu29/sbu29 embryos send out long protrusions during migration (arrow). Scale bar = 10 μm.
Rest-deficient FBMNs show precocious morphological maturation
As Rest functions to repress neuronal maturation genes, we hypothesized that the partial abrogation of FBMN migration observed in Rest mutants might be due to precocious neuronal maturation. Migrating neurons are typically elongated in the direction of travel, and send out few short protrusions into their environment (Komuro and Rakic, 1998; Mapp et al., 2010; Marin et al., 2006). Once neurons reach their destination they undergo morphological changes as they complete their maturation program: the cytoplasmic volume increases and the neurons take on a more circular shape while neurites are extended to integrate with surrounding neural circuitry (Hatten, 1999; Jobe et al., 2012; Komuro and Rakic, 1998). We hypothesized that Rest-deficient FBMNs might show characteristics of mature neuron morphology at stages and positions when wild type FBMNs have immature morphology, indicating their precocious maturation. Wild type FBMNs mature over developmental time as they migrate, showing morphological differences as they move from r4 to r5 and r6 (Mapp et al., 2010). To analyze morphological maturation, we used the Tg(zCREST1:membRFP) line to visualize details of FBMN morphology (Fig. 4). We compared the FBMNs of MZrestsbu29/sbu29 embryos with those of wild type embryos collected from crosses between wild type siblings of rest mutants. We measured length-to-width ratios of cell bodies to reveal how “round” neurons were, as well as the number and length of protrusions they extended. As migration of MZrestsbu29/sbu29 mutant FBMNs is abrogated, we cannot compare mutant versus wild type neurons in specific rhombomeric locations, we therefore elected to analyze all FBMNs at two different stages of migration to ensure that the only variable in the analysis is time. The analysis was performed on FBMNs at 20 hpf and at 28 hpf; in each case we averaged the measurements of all FBMNs from five embryos of each genotype.
(A–G) Analysis of FBMNs at 20 and 28 hpf in Tg(zCREST1:membRFP) wild type embryos (A, B) and MZrestsbu29/sbu29 embryos (C, D). Average measurements from 5 embryos are shown for wild type and MZrestsbu29/sbu29 (E–G) embryos at each stage. The wild type length-to-width ratio decreases from 20 to 28 hpf, as neurons become rounder; by contrast MZrestsbu29/sbu29 FBMNs are round at both 20 and 28 hpf (E). Both number and length of protrusions increase over time in wild type and MZrestsbu29/sbu29 embryos (F,G). MZrestsbu29/sbu29 FBMNs have significantly more and longer protrusions than wild type FBMNs. (H–J) Volume measurements of FBMNs at 22 hpf. Whole cell and nuclear volume were measured for 10 “immature” neurons in r4 (yellow dots) versus 10 “mature” neurons in r6 (red dots) for 5 wild type (H) and 5 MZrestsbu29/sbu29 (I) embryos. Wild type FBMN whole cell volume increases significantly as neurons migrate from r4 to r6; MZrestsbu29/sbu29 FBMN whole cell volume does not alter as neurons migrate from r4 to r6 and is similar to that of wild type neurons in r6. Nuclear volume does not change significantly. Mean ± SEM; *P < 0.05; ***P < 0.001, ****P < 0.0001, n.s. = not significant.
Consistent with our previous descriptions, we find that wild type FBMNs are elongated at 20 hpf (Fig. 4A) and become more circular by 28 hpf (Fig. 4B,E). The neurons also send out a greater number of protrusions, and longer protrusions, at the later stage compared to early migrating neurons (Fig. 4F–G). By contrast, in Rest-deficient embryos FBMNs are relatively circular at both early and late stages (Fig. 4C–E), and both the number and length of protrusions are greater than in wild type neurons (Fig. 4F–G).
We next measured cytoplasmic volume as the neurons matured using 3D confocal reconstructions. This analysis focused on comparing immature r4-localized FBMNs to mature r6-localized FBMNs at a single stage of migration, 22 hpf. We selected this stage because it is before radial migration commences in r6, after which neuron accumulation along the dorsoventral axis obscures imaging. We measured the volume of TO-PRO-3-labeled nuclei as well as the membRFP-labeled whole cell volume in the ten most anterior and ten most posterior FBMNs in five embryos (yellow and red dots, respectively; Fig 4H–I). While nuclear volume remained unchanged in all neurons, the whole cell (or cytoplasmic) volume of wild type neurons was significantly larger in r6-localized FBMNs compared to r4-localized FBMNs, consistent with mature neurons that have completed migration increasing in size (Fig. 4J). In Rest-deficient embryos, FBMNs in r4 are significantly larger than in wild type, and are comparable in size to mature r6 FBMNs. This suggests that synthesis of new proteins and subsequent cytoplasmic swelling occurs prematurely in r4-localized FBMNs in the absence of Rest repression. While a subset of the Rest-deficient neurons located in r4 might arguably be “older” neurons that have failed to migrate, our quantification reveals that average volumes of r4 neurons vary little, implying that even recently born neurons are unusually large. The question arises of whether FBMNs in Rest-deficient specimens are already unusually large at the moment of their birth; however, current technical limitations prevent us from addressing this point, as we are unable to visualize newborn FBMNs until sufficient fluorescent reporter protein has accrued. In summary, our results are consistent with the hypothesis that Rest-deficient FBMNs precociously develop mature morphology at a premature developmental stage relative to wild type FBMNs. From this analysis we conclude that functional Rest is required to maintain the appropriate shape and size of FBMNs during migratory stages, suggesting in turn that the lack of “immature” morphology in Rest-deficient FBMNs contributes to the disruption of their migration.
Rest target genes are precociously expressed in Rest and Pk1b mutants
In addition to precocious morphological maturation, we hypothesized that FBMNs might undergo precocious molecular maturation in the absence of Rest. We therefore tested whether Rest target genes, which are normally repressed, are prematurely expressed in Rest-deficient FBMNs during stages when the neurons should be migrating. To validate the hypothesis that Rest must be localized to the nucleus to act, we also analyzed Rest target gene expression under conditions where Rest fails to localize to the nucleus. We previously established that nuclear localization of Rest in zebrafish FBMNs is dependent on the function of Prickle1b (Pk1b), revealing a planar cell polarity (PCP)-independent role for Pk1b at the nuclear membrane (Mapp et al., 2011). Significantly, in Pk1b-deficient embryos Rest is present but not localized to the nucleus where it can act.
We began this analysis by taking a candidate approach, initially screening expression data available on the ZFIN database (zfin.org) for Rest target genes identified in previous studies (Bruce et al., 2004; Johnson et al., 2007; Johnson et al., 2008; Mortazavi et al., 2006; Sun et al., 2005). We then used in situ hybridization to identify zebrafish Rest target genes that lack expression in FBMNs at 24 hpf, when the migration process is still underway, but that are expressed in FBMNs at 48 hpf, when migration is complete and the Rest repressor has been depleted from the nucleus (Mapp et al., 2011). Having identified such genes, we went on to compare their expression in wild type rest siblings with expression in MZrestsbu29/sbu29 mutants and pk1bfh122/fh122 mutants at 24 hpf. We analyzed four Rest target genes: adam23a (Fig. 5A–A′), snap25b (Fig. 5D–D′), srrm4 (Fig. 5G–G′), and syt4 (Fig. 5J–J′). In 24 hpf MZrestsbu29/sbu29 embryos we found inappropriate expression of all four genes in the FBMNs (Fig. 5B–B′, E–E′, H–H′, K–K′, arrows). Interestingly, a slight elevation in expression levels was also seen in surrounding tissues, suggesting that Rest may additionally regulate expression of these targets in other cell types. In 24 hpf pk1bfh122/fh122 mutant embryos, in which all FBMNs remain in r4 (Mapp et al., 2011), we observed robust expression of all four Rest targets in the FBMNs (Fig. 5C–C′, F–F′, I–I′, L–L′, arrows), in accord with our model that Rest nuclear localization is required for its function. This precocious molecular maturation of pk1bfh122/fh122 FBMNs is also consistent with our previous report that the non-migratory FBMNs display morphological and behavioral characteristics of wild type neurons that have reached r6 (Mapp et al., 2010). While Rest target gene expression levels appear higher in pk1bfh122/fh122 mutant FBMNs than in MZrestsbu29/sbu29 mutant FBMNs, this may primarily be a consequence of the FBMNs being clustered together in r4 of Pk1b-deficient embryos; while we performed in situ hybridization experiments in parallel the technique is inherently non-quantitative. We also found premature expression of FBMN-expressed genes that have not been identified as direct targets of Rest, such as cmet (Fig. 5M-O′; Elsen et al., 2009) and nrp1a/nrp2b (data not shown; Yu et al., 2004). These data suggest that changes in gene expression are not limited to Rest targets, and indirect activation of multiple genes may contribute to the precocious maturation of Rest-deficient FBMNs. In summary, we confirm that in the absence of Rest function FBMNs show expression of neuronal maturation genes at inappropriately early developmental stages; this precocious molecular maturation may in turn drive observed changes in morphology and prevent full migration of many neurons to r6.
In situ hybridization analysis of Rest target gene expression in Tg(islet1:GFP) embryos at 24 hpf in wild type, MZrestsbu29/sbu29, and pk1bfh122/fh122 specimens. Rest targets selected for analysis are adam23a (A–C′), snap25b (D–F′), srrm4 (G–I′), and syt4 (J–L′). We also analyzed the non-Rest target and FBMN maturation marker cmet (M–O′). All five genes are expressed in fully migrated FBMNs at 48 hpf. Upper panels show co-expression with islet1:GFP to label FBMNs, with gene expression alone shown below. At this stage (24 hpf), the selected Rest targets, as well as the maturation marker cmet, are not expressed in wild type FBMNs, but are expressed in both rest and pk1b mutant FBMNs (arrows).
Rest target Srrm4 functions downstream of Rest in FBMN maturation
Rest target Srrm4, also known as nSR100, is an alternative splicing factor required for the inclusion of neuron-specific exons in the mature mRNAs of multiple neural-specific transcripts (Calarco et al., 2009). The srrm4 gene is a direct target of Rest repression (Calarco et al., 2009), and in mammals is also proposed to act in a feedback loop to negatively regulate Rest repression while promoting neuron-specific pre-mRNA processing (Raj et al., 2011). Based on its function, we hypothesized that Srrm4 might function as a key downstream target of Rest to drive neuronal maturation in zebrafish FBMNs. According to this model (Fig. 6A), in wild type migrating neurons Rest functions in the FBMN nucleus to actively repress srrm4 expression, and lack of Srrm4 prevents neuronal maturation to help maintain neurons in an immature state compatible with migration. As Srrm4 is not normally expressed in migrating neurons, we predicted no deficits in FBMN migration in response to Srrm4 deficiency; this prediction was confirmed by knockdown of srrm4 (Fig. 6B–C). By contrast, in Rest-deficient embryos srrm4 is prematurely expressed in migrating FBMNs (Fig. 5G–H′), where we hypothesize it plays a key role in promoting neuronal maturation to cause incomplete neuronal migration (Fig. 6A, D). To test this hypothesis, we investigated whether the precocious maturation and partial migration of Rest mutant FBMNs could be rescued by knocking down Srrm4. We found that there was indeed a partial rescue of FBMN migration in MZrestsbu29/sbu29 specimens injected with Srrm4 MO (as described, Calarco et al., 2009), in comparison to uninjected siblings (Fig. 6D–E). FBMN migration was again quantified using the r5 marker EphA4a, showing that significantly more Srrm4-deficient neurons completed migration to r6 (Fig. 6F).
(A) Model of Rest/Srrm4 regulation of FBMN maturation and the effect on FBMN migration. (B–E) Average projection dorsal views of 48 hpf FBMNs (green) in Tg(islet1:GFP) embryos immunolabeled with EphA4a to mark r5 (red). FBMN migration is unaffected in Srrm4 morphants (C) as compared to uninjected controls (B). MZrestsbu29/sbu29 mutant FBMN migration (D) is partially rescued with knockdown of Srrm4 (E). (F) Quantitative counts of FBMN migration in r4 (black), r5 (checkered), and r6 (white) as % of all FBMNs. A significantly increased number of FBMNs migrate to r6 in MZrestsbu29/sbu29 mutants injected with Srrm4 MO as compared to uninjected mutants. Mean ± SEM; ****P < 0.0001. (G–J) Analysis of FBMNs at 22 and 30 hpf in Tg(zCREST1:membRFP);MZrestsbu29/sbu29 embryos (G,H), and Tg(zCREST1:membRFP);MZrestsbu29/sbu29 embryos injected with Srrm4 MO (I,J). Average morphological measurements from 5 embryos are shown at each stage (K–N). Mutants injected with Srrm4 MO have FBMNs that are more elongated than those injected with Control MO, and remain so at 30 hpf (K). They also send out fewer and shorter protrusions than MZrestsbu29/sbu29 at both early and late stages (L–M). Mutants injected with Srrm4 MO have a smaller cytoplasmic volume compared to mutant controls at 22 hpf, which swells significantly at the later stage compared to the unchanged volume in MZrestsbu29/sbu29 mutants (N).
We further analyzed the morphology of FBMNs in Tg(zCREST1:membRFP);MZrestsbu29/sbu29 mutants injected with Srrm4 MO compared with mutants injected with a standard control MO. Analysis was performed at early and late stages of FBMN migration: 22 and 30 hpf (Fig. 6G–J). While FBMNs of MZrest mutants are round in shape, whether in r4 or r6 (Fig. 6K, and see also Fig. 4E), the FBMNs of mutants injected with Srrm4 MO are more elongated at both stages (Fig. 6K). Additionally, Srrm4 MO-injected mutant embryos have FBMNs that send out fewer and shorter protrusions compared to controls (Fig. 6L–M), and notably, these behaviors persist at the later stage. The cytoplasmic volume of FBMNs in control mutant specimens is elevated at both stages (Fig. 6N; as also shown in Fig. 4J). However, the cytoplasmic volume of r4 localized FBMNs in Srrm4 MO-injected mutant embryos is similar to that observed in wild type specimens (compare Fig. 6N with Fig. 4J); this volume increases as FBMNs migrate from r4 to r6, again similar to wild type (Fig. 6N; Fig. 4J). Our observation that depletion of Srrm4 function allows MZrest mutant FBMNs to maintain an elongated cell shape and low protrusion number and length through later developmental stages, strongly suggests that Srrm4 is an important regulator of neuronal morphology. In summary, our findings suggest that processing of neuron-specific pre-mRNAs by Srrm4 is important for FBMN maturation, and further support our model that appropriate control of the neuronal maturation process is essential for proper migration.
Nuclear-localized Rest can rescue migration in Rest-deficient but not Pk1b-deficient embryos
Central to our overall model of FBMN migration is the prediction that Rest must be localized to FBMN nuclei to repress its target genes and prevent precocious maturation. To test this prediction, we examined whether nuclear-localized Rest can rescue migration of FBMNs in Rest-deficient embryos. To localize Rest to the nucleus we engineered a variant of Rest carrying the nuclear localization sequence (NLS) PKKKRKV from the SV40 Large T antigen (Rai et al., 2007) at its C-terminus (Rest-NLS; Fig. 7A). Analysis of blastula-stage embryos injected with mRNA encoding mCherry-tagged Rest-NLS confirmed the expected localization of Rest protein to the cell nucleus (Fig. 7B–D). To test whether mCherry-Rest-NLS could rescue neuronal migration, we expressed the construct specifically in FBMNs using the zCREST1 enhancer and Tol2 transgenesis (as described above; Fig. 2B). As in our previous experiments, we analyzed groups of embryos with similar numbers of mCherry+ FBMNs expressing Rest-NLS (Fig. 7E–H); embryos with zero mCherry+ FBMNs functioned as controls. We found that as the number of MZrestsbu29/sbu29 mutant FBMNs expressing mCherry-Rest-NLS increased, a greater number of FBMNs migrated to r6 (Fig. 7E′–H′). In embryos with over 20 FBMNs expressing mCherry-Rest-NLS the majority of neurons migrated to r6 (Fig. 7F″–H″), again suggesting that neurons migrate in a collective manner. Essentially, collective migration allows migratory FBMNs containing nuclear-localized Rest to “pull along” their Rest-deficient and otherwise non-migratory neighbors to r6. In summary, we have effectively targeted Rest to the nuclei of FBMNs, and confirmed that nuclear function of Rest promotes FBMN migration through the hindbrain.
(A) Schematic of Rest protein domains with the added NLS sequence PKKKRKV at the C-terminus (Rest-NLS). This construct was tagged with mCherry and cloned into both pCS2+ and Tol2 vectors. (B–D) Expression of mCherry-tagged Rest-NLS (injected mRNA) in blastula stage embryos (5 hpf), compared with expression of Rest protein expression using immunohistochemistry. Rest-NLS is correctly localized to the nucleus while endogenous Rest is expressed in the cytoplasm and nucleus (Kok et al., 2012). (E–H) Expression of mCherry-Rest-NLS specifically in FBMNs using Tol2 transgenesis (as described in Fig. 2) in Tg(islet1:GFP);MZrestsbu29/sbu29 mutant embryos. Embryos were classed into four groups based on the number of mCherry+ FBMNs present: zero (controls, E), 1–10, 11–20, and 21–30. (F–H) Increasing the number of FBMNs expressing Rest-NLS rescues FBMN migration to r6. (E′–H′) Counts of Tg(islet1:GFP) FBMNs (green) in r4 (black), r5 (checkered) and r6 (white). (F″–H″) Counts of mCherry+ FBMNs (red/yellow) in each group of mCherry-expressing embryos. mCherry+ FBMNs are more likely to reside in r6, suggesting their ability to “pull along” mutant FBMNs lacking the construct to complete migration. Split channel images confirmed that mCherry+ cells were also GFP+ as expected. (I–K) Rest-NLS expression within Pk1b mutant FBMNs using Tol2 transgenesis does not rescue migration out of r4.
Finally, to examine to what extent Rest is responsible for mediating the function of Pk1b in FBMN migration, we tested whether nuclear-localized Rest could rescue FBMN migration in Pk1b-deficient embryos. We found that even high numbers of mCherry-Rest-NLS-expressing FBMNs failed to rescue neuronal migration in pk1bfh122/fh122 mutants (Fig. 7I–J′). Thus nuclear-localized Rest is not sufficient to rescue FBMN migration in the absence of Pk1b function. This finding indicates that Pk1b is playing two separable roles in FBMN migration: one role is to properly localize Rest to the nucleus, thus preventing premature maturation of FBMNs, while the second role is independent of Rest and is likely critical to initiation of FBMN migration out of r4.
Discussion
In this study we have established that Rest inhibits the precocious maturation of migrating FBMNs. We show that Rest functions within the nuclei of zebrafish FBMNs, where active repression of Rest target genes is required for complete migration of the neurons to r6/r7. Our data lead to a model (Fig. 8), in which absence of Rest repression results in precocious expression of terminal maturation genes leading to mature neuron morphology (including loss of elongate shape, increased cytoplasm-to-nucleus ratio, and extensive neurite outgrowth) in FBMNs at an inappropriate stage and location within the hindbrain. The morphological changes we have documented in maturing FBMNs are consistent with changes observed as neurons develop from postmitotic progenitors to mature functional neurons (Ayala et al., 2007; Dotti et al., 1988; Le et al., 2007; Marin and Rubenstein, 2001; Metin et al., 2006; Wonders and Anderson, 2006). Our study establishes the Rest transcriptional repressor as a critical regulator in the FBMNs, providing a molecular link between the immature state of neurons and their ability to migrate during development.
(A) In immature FBMNs born in r4, Pk1b translocates Rest into the nucleus to repress its target genes. This gene repression inhibits the maturation of neurons and maintains FBMNs in a migration-competent state. When FBMNs reach r6/r7, Rest is depleted from the nucleus, and Rest targets are expressed. The expression of neuronal maturation genes and the alternative splicing of neuron-specific transcripts allow the FBMNs to adopt mature neuron characteristics in r6. (B) Rest-deficient FBMNs only partially migrate. The absence of Rest allows for the expression of both RE1-containing genes and other maturation genes at stages and locations prior to normal maturation. This precocious expression correlates with morphological characteristics consistent with mature neurons, demonstrating that gene regulation underlies the timing of neuronal maturation and the ability of neurons to migrate. (C) In Pk1b-deficient FBMNs, Rest cannot translocate into the nucleus to repress maturation genes, and FBMNs show precocious gene expression and mature neuron morphology (Mapp et al., 2010).
Rest has been hypothesized to have critical functions in controlling the progress of cells from the progenitor state to mature differentiated neurons, yet most previous studies have analyzed Rest function in cultured cells (Ballas et al., 2005; Ballas and Mandel, 2005; Johnson et al., 2007; Otto et al., 2007). Rest-deficiency in mice leads to early embryonic lethality, precluding a thorough analysis of Rest function in vivo (Chen et al., 1998), yet conditional mutation has revealed that Rest function in the brain is not necessary for neurogenesis to occur (Aoki et al., 2012). Although zebrafish MZrestsbu29/sbu29 mutant embryos devoid of any rest transcript sometimes fail to form a swim bladder and have variable survival to adulthood (our observations, Howard Sirotkin pers. comm.), they reliably survive through embryogenesis (5 days post fertilization), and zygotic restsbu29/sbu29 mutants are homozygous viable (Kok et al., 2012). Thus zebrafish provides a useful model in which to analyze mechanisms of Rest function in a specific in vivo neuronal cell type during embryogenesis. As Rest function has been implicated in several disorders, including aging and Alzheimer’s disease (Lu et al., 2014), Huntington disease (Zuccato et al., 2003), cancers (Negrini et al., 2013), neuropsychiatric disorders (Hsieh and Eisch, 2010), and epilepsy (McClelland et al., 2014), our study helps to establish zebrafish MZrestsbu29/sbu29 mutant FBMNs as a powerful tool to analyze the function of this important protein.
Our findings demonstrate that complete tangential migration of the FBMN population to its final destination in the hindbrain requires active Rest repression. In contrast, it was recently reported that sustained Rest repression inhibits radial migration of neural stem/progenitor (NS/P) cells in the mammalian neocortex (Mandel et al., 2011). Both this study and ours conclude that Rest function is critical for regulating the timing of neuronal maturation, yet suggest opposite roles for Rest function in radial and tangential migration, consistent with the divergent reliance on radial glia for these two modes of migration. Our analysis of migrating MZrestsbu29/sbu29 mutant FBMNs at 24 hpf indicates that the neurons can undergo normal radial migration if they reach r6, and do not migrate radially prior to reaching r6 (Fig. 3K–L). These findings demonstrate that whereas tangential migration of FBMNs is reliant on Rest function, radial migration of FBMNs in r6 must depend on other mechanisms. Thus Rest is required for tangential, but not radial, migration of zebrafish FBMNs, and functions principally to maintain their immature migration-competent state.
We demonstrated that Rest acts within the FBMNs by using a transient transgenic approach to disrupt Rest function specifically in cranial branchiomotor neurons. Consistent with these findings, a limited set of cell transplantation experiments, in which MZrest mutant FBMNs were transplanted into wild type hindbrains, similarly indicated that Rest function is required in the FBMNs for normal migration to occur. We confirmed the FBMN-specific function of Rest by targeting wild type Rest protein exclusively to FBMN nuclei of Rest mutant specimens, which proved sufficient to efficiently rescue their migration. Interestingly, FBMN expression of human REST fused to the potent VP16 transcriptional activator caused more dramatic defects in FBMN migration than did de-repression of Rest using a dominant negative variant. These defects included a robust abrogation of neuronal migration, as well as the projection of ectopic neurites from FBMNs that did not migrate. We suggest that by activating Rest targets in the FBMNs at a stage when those genes should normally be repressed, we have generated a neomorphic activity. These results additionally suggest that full expression of Rest target genes in the zebrafish FBMNs is normally dependent on additional transcriptional activators, which in our experimental situation are being substituted by the VP16 activator. Our findings are also consistent with a previous report of REST-VP16 activation of human neuronal differentiation genes (Immaneni et al., 2000). We conclude that Rest must cooperate with additional regulatory factors to precisely regulate expression of FBMN maturation genes.
Our experiments to confirm that Rest functions specifically within the FBMNs have also provided additional evidence for a “collective mode” of FBMN migration. Similar evidence for collective migration came from our previous investigation of the Rest nuclear translocator Pk1b (Rohrschneider et al., 2007), and from analysis of the Pk1b transcriptional regulator Hoxb1a (Cooper et al., 2003). In all three cases, mosaic analysis has established that the molecule in question functions cell-autonomously within the FBMNs, but when wild type and “deficient” neurons are combined in the same specimen they are able to influence one another’s migratory behavior. Collective migration modes have also been uncovered in other systems using similar experimental approaches. For example, zebrafish lateral line cells lacking the Cxcl12a chemokine receptor, or border cells of the fly egg chamber lacking the slbo transcription factor, can migrate only in the presence of wild type migration-competent cells (Haas and Gilmour, 2006; Rorth et al., 2000). There are interesting parallels between collective migration and the “domineering non-cell autonomy” of PCP signaling described in the fly, in which mutant clones influence the behavior of adjacent wild type cells (reviewed by Adler, 2002). Although PCP signaling plays an important role in FBMN migration (reviewed by Wanner et al., 2013), the collective mode of FBMN migration is nevertheless apparently independent of PCP signaling, given that it does not require the function of key PCP molecules in the rescued neurons (Walsh et al., 2011). The underlying molecular mechanism of collective migration therefore remains elusive, but the process is likely mediated through cell-cell contact-mediated signaling.
The requirement for Rest to localize to the nuclei of migratory FBMNs, where it can function to repress its target genes, is confirmed by our demonstration that nuclear-localized Rest is sufficient to rescue FBMN migration in Rest-deficient embryos. In unmanipulated specimens Rest becomes depleted from FBMN nuclei as they reach r6/r7 (Mapp et al., 2011). However, in this study we have shown that continuous expression of Rest in FBMN nuclei does not interfere with the termination of their tangential migration. This finding indicates that the switch from tangential to radial migration modes must be regulated by Rest-independent mechanisms, and suggests that the observed depletion of Rest from FBMN nuclei is either independent of their altered mode of migration or occurs in response to that switch. One possibility is that the specific local environment triggers the observed change in Rest protein sub-cellular localization; previous work has shown that the Huntingtin protein can sequester Rest in the cytoplasm (Lunyak and Rosenfeld, 2005; Zuccato et al., 2003), suggesting a potential regulator.
Our experiments with nuclear-localized Rest have also revealed its inability to rescue migration of Pk1b-deficient FBMNs. This finding, together with the observation that the block to FBMN migration in Pk1b-deficient specimens is absolute, rather than partial as in Rest-deficient specimens, demonstrates that Pk1b must play one or more roles in FBMN migration that are independent of its function in Rest nuclear localization. Given that Prickle molecules are well-established core components of the PCP pathway, it is likely that Pk1b also has a PCP-dependent role in FBMN migration. For example, Pk1b may play a critical role in establishing aspects of FBMN polarity and thus allowing the neurons to respond to other guidance cues. Alternatively, or in addition, Pk1b may function as a nuclear translocator for other molecules (beyond Rest) that are required for FBMN migration. Future analysis of Pk1b interaction partners should help to uncover its additional role or roles in FBMN migration.
Rest functions by binding directly to the RE1 motif in the regulatory regions of hundreds of neuron-specific genes, and recruiting cofactors to mediate long-term repression of chromatin in that region (Ballas and Mandel, 2005). Initiation of the terminal maturation program requires not only the expression of genes normally repressed by Rest, but also a global switch in transcriptome modifications so that the proteome adopts neuron-specific functions. The Rest target and vertebrate-specific alternative splice factor Srrm4 is a key regulator of pre-mRNA splicing required for neural cell differentiation (Calarco et al., 2009; Nakano et al., 2012). During neuronal differentiation, Srrm4-containing alternative splicing complexes activate the inclusion of neuron-specific exons that were previously excised out of pre-mRNAs (Calarco et al., 2009). The inclusion of neuron-specific alternative exons depends on the binding of Srrm4 and the neuron-specific factor nPTB to a unique sequence motif on the pre-mRNA. This site is present in a wide range of gene products that function throughout the cell, especially those required for cytoskeleton remodeling and the secretory vesicle pathway (Nakano et al., 2012). Here we report a partial rescue of FBMN migration in MZrestsbu29/sbu29 mutants by Srrm4 knockdown, suggesting that Srrm4 is a key activator of FBMN maturation. Consistent with this, we find that Srrm4 function is associated with the establishment of mature neuron morphology, specifically, the loss of elongate cell shape and neurite extension. Importantly, as Srrm4 is a direct target of Rest repression, Rest function within the nucleus is essential to delay the activation of a cascade of transcriptome modifications required for mature neuron function. Further studies will be necessary to understand the role of alternative splicing and Srrm4-nPTB function in FBMN maturation. Functional analysis of other FBMN Rest targets, such as Adam23a, which is involved in neurite outgrowth (Owuor et al., 2009), will also deepen our understanding of FBMN maturation.
Supplementary Material
1
2
3
1
2
3
Acknowledgments
We would like to thank Howard Sirotkin for providing Rest mutant fish and helpful discussions on mutant analyses; Sadhan Majumder for providing the REST-VP16 construct; David Wilkinson for EphA4a antibody; Brian Ciruna for the srrm4 probe; Bernard and Christine Thisse for the snap25b probe; Corinne Houart for nrp probes; Anita Ng for help with Tol2 cloning, immunohistochemistry and fish care; Sarah Wanner for help with statistical analysis and microscopy; Anastasia Beiriger for assistance with cell transplantation experiments; the Integrated Microscopy Core Facility at the University of Chicago for help with data analysis; members of the Prince and Ho labs for discussion and experimental troubleshooting; and Sarah Wanner, Anastasia Beiriger and Cliff Ragsdale for comments on the manuscript. This work benefitted from the resources of the ZFIN database (zfin.org). This work was supported by March of Dimes Grant (FY07-410) to V.E.P., NSF Major Research Instrumentation grant (1040297), and The University of Chicago Clinical and Translational Science Award (CTSA; UL1 RR024999). C.E.L. was supported by NIH T32 {"type":"entrez-nucleotide","attrs":{"text":"HD055164","term_id":"300424876","term_text":"HD055164"}}HD055164.
Abstract
The vertebrate brain arises from the complex organization of millions of neurons. Neurogenesis encompasses not only cell fate specification from neural stem cells, but also the terminal molecular and morphological maturation of neurons at correct positions within the brain. RE1-silencing transcription factor (Rest) is expressed in non-neural tissues and neuronal progenitors where it inhibits the terminal maturation of neurons by repressing hundreds of neuron-specific genes. Here we show that Rest repression of maturation is intimately linked with the migratory capability of zebrafish facial branchiomotor neurons (FBMNs), which undergo a characteristic tangential migration from hindbrain rhombomere (r) 4 to r6/r7 during development. We establish that FBMN migration is increasingly disrupted as Rest is depleted in zebrafish rest mutant embryos, such that around two-thirds of FBMNs fail to complete migration in mutants depleted of both maternal and zygotic Rest. Although Rest is broadly expressed, we show that de-repression or activation of Rest target genes only within FBMNs is sufficient to disrupt their migration. We demonstrate that this migration defect is due to precocious maturation of FBMNs, based on both morphological and molecular criteria. We further show that the Rest target gene and alternative splicing factor srrm4 is a key downstream regulator of maturation; Srrm4 knockdown partially restores the ability of FBMNs to migrate in rest mutants while preventing their precocious morphological maturation. Rest must localize to the nucleus to repress its targets, and its subcellular localization is highly regulated: we show that targeting Rest specifically to FBMN nuclei rescues FBMN migration in Rest-deficient embryos. We conclude that Rest functions in FBMN nuclei to inhibit maturation until the neurons complete their migration.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.