The homeodomain-containing transcription factor Nkx2.9 is expressed in the ventralmost neural progenitor domain of the neural tube together with the related protein Nkx2.2 during early mouse embryogenesis. Cells within this region give rise to V3 interneurons and visceral motoneurons in spinal cord and hindbrain, respectively. To investigate the role of the Nkx2.9gene, we generated a mutant mouse by targeted gene disruption. Homozygous mutant animals lacking Nkx2.9 were viable and fertile with no apparent morphological or behavioral phenotype. The distribution of neuronal progenitor cells and differentiated neurons in spinal cord was unaffected inNkx2.9-deficient animals. This finding is in contrast toNkx2.2-null mutants, which have been shown to exhibit ventral to dorsal transformation of neuronal cell fates in spinal cord. Our results suggest that specification of V3 interneurons in the posterior CNS does not require Nkx2.9, most probably because of functional redundancy with the co-expressed Nkx2.2 protein. In hindbrain, however, absence of Nkx2.9 resulted in a significantly altered morphology of the spinal accessory nerve (XIth),which appeared considerably shorter and thinner than in wild-type animals. Consistent with this phenotype, immature branchial motoneurons of the spinal accessory nerve, which normally migrate from a ventromedial to a dorsolateral position within the neural tube, were markedly reduced in Nkx2.9-deficient embryos at E10.5, while ventromedial motor column cells were increased in numbers. In addition, the vagal and glossopharyngeal nerves appeared abnormal in approximately 50% of mutant embryos, which may be related to the observed reduction of Phox2b expression in the nucleus ambiguus of adult mutant mice. From these observations, we conclude that Nkx2.9 has a specific function in the hindbrain as determinant of the branchial motoneuron precursor cells for the spinal accessory nerve and possibly other nerves of the branchial-motor column. Like other Nkx genes expressed in the CNS, Nkx2.9 seems to be involved in converting positional information into cell fate decisions.
INTRODUCTION
Development of the nervous system involves specification of distinct classes of neurons at defined locations within the central nervous system(CNS). In the ventral neural tube, patterning of neuronal progenitor cells depends on notochord and floorplate acting as local organizing centers through secretion of the signaling molecule sonic hedgehog (Shh)(Chiang et al., 1996;Ericson et al., 1997a;Marti et al., 1995;Yamada et al., 1991). Shh forms a concentration gradient, and correct formation of interneurons and motoneurons on the dorsoventral axis of the neural tube requires the graded activity of Shh (Briscoe et al.,1999; Ericson et al.,1997b). It has been proposed that graded Shh activity in ventral neural tube leads to the expression of distinct sets of homeodomain transcription factors. The combinatorial expression of these transcription factors within the various domains of the spinal cord then is believed to specify the principle classes of neurons, referred to as V1 and V2 neurons,motoneurons (MN) and V3 neurons in dorsal-to-ventral order(Briscoe et al., 2000).
According to this model, progenitor cells that give rise to the most ventral population of V3 interneurons are exposed to higher Shh concentrations and express the transcription factors Nkx2.2, Nkx2.9 and Nkx6.1(Barth and Wilson, 1995;Hartigan and Rubenstein, 1996;Pabst et al., 1998;Qiu et al., 1998;Shimamura et al., 1995), while the more dorsal progenitor cells of motoneurons encounter lower Shh signaling activity and express Pax6. In mice that lack Pax6, these progenitor cells generate neurons that are typical of higher Shh concentrations and are located more ventrally (Ericson et al.,1997b; Osumi et al.,1997). In Nkx2.2-deficient mice, differentiated V3 interneurons are missing and the dorsally adjacent population of motoneuron precursors expands ventrally, indicating that Nkx2.2 has a primary role in ventral neuronal patterning(Briscoe et al., 1999).Nkx2.2-expressing progenitors in the spinal cord generate a class of neurons that express the basic helix-loop-helix transcription factors Sim1 and Ngn3 (Fan et al., 1996;Sommer et al., 1996), and give rise to oligodendrocytes at later stages of development(Qi et al., 2001). Accordingly, lack of Nkx2.2 also results in delayed development of oligodendrocytes (Qi et al.,2001). Transformation of neuronal cell fate was further observed in mice that lack the Nkx6.1 gene. These mutant mice contain increased numbers of V1 neurons at the expense of MNs and V2 neurons(Sander et al., 2000). In line with these observations, inactivation of Nkx2.1 that is expressed in ventral domains of the telencephalon also results in ventral to dorsal transformation of brain structures (Sussel et al.,1999). More recently, a function of Nkx2.1 in hypothalamus has been demonstrated (Marin et al.,2002). Taken together, these various results suggest that Nkx transcription factors are important for regional patterning and cell fate determination within the ventral CNS in response to graded Shh signaling.
Interestingly, the NK2/vnd gene, a founding member of the NK gene family in Drosophila (Kim and Nirenberg, 1989), also determines neuronal identity similar to Nkx2 genes in the mouse. It has been shown that NK2/vnd in the fly is necessary and sufficient to generate ventral cell fates, as mutants that lack vnd function fail to form specific ventral neuroblasts(McDonald et al., 1998). Thus,the role of NK/Nkx genes in dorsoventral patterning of the CNS seems to be conserved during evolution.
Identification of the murine Nkx2.9 gene, a novel member of the mammalian Nkx2 family, and its expression pattern in the ventral CNS during mouse embryogenesis has been described previously(Pabst et al., 1998).Nkx2.9 is structurally most closely related to Nkx2.2 and both genes are expressed in largely overlapping domains, depending on Shh during early stages of embryogenesis(Pabst et al., 2000). To investigate the role of Nkx2.9, we generated a null mutation in mouse by targeted gene disruption, and analyzed the resulting phenotype. HomozygousNkx2.9-deficient mutants were viable and fertile without overt morphological or behavioral abnormalities. In contrast to Nkx2.2mutant animals, the distribution of neuronal precursors and mature neurons appeared unaffected in spinal cord of homozygous Nkx2.9 mutant mice,whereas in the hindbrain the accessory nerve (XIth) was markedly reduced in size compared with wild-type animals, and some mutants exhibited morphologically abnormal vagal and glossopharyngeal nerves. Consistent with these nerve defects, Nkx2.9 mutant embryos lacked most of the migratory neuronal precursor cells that represent branchial motoneuron progenitors of the spinal accessory nerve but contained more median motor column neurons instead. These results are in line with the current model of a combinatorial action of Nkx homeobox genes in specifying neuronal cell fates along the dorsoventral axis within the CNS.
MATERIALS AND METHODS
Construction of the gene targeting vector and generation of mutant mice
Genomic DNA encompassing the Nkx2.9 gene was isolated from a spotted 129Sv mouse cosmid library using a full-length Nkx2.9 cDNA as hybridization probe (Pabst et al.,1998). Clone MPMGc121F08427Q3 containing the gene was provided by RZPD (Resource Center/Primary Database, Berlin, Germany). The locus organization was analyzed by extensive mapping of restriction sites and nucleotide sequence analysis. The targeting vector was designed to replace the entire coding region by the lacZ reporter gene and the PGK-neomycin resistance cassette. TBV2 ES-cells (kindly provided by W. Wurst, MPI Munich,Germany) were electroporated with linearized targeting vector DNA and growing colonies were selected in medium containing 400 μg/ml G418 for 8 days. Homologous recombination events within individual clones were identified on Southern blots of DNA, digested with EcoRI and hybridized with the 3′-flanking probe as indicated inFig. 1. Wild-type and mutant alleles were detected by 7 kb and 4 kb restriction fragments, respectively. Approximately 5% of the neomycin-resistant clones revealed vector integrations by homologous recombination. Two independently derived clones were used for injections into blastocysts of C57/Bl6 mice.
The targeting strategy to inactivate the Nkx2.9 gene. (A)Illustrates the genomic organization of the Nkx2.9 locus (wt), the targeting vector, and the mutated allele after homologous recombination (mut). Most of the exons (red boxes), including the homeodomain contained in exon 2,were replaced by the IRES-lacZ reporter gene (small green/blue box)and the neomycin selection cassette (large green box). IRES-lacZ-coding sequence was inserted at the translation initiation codon. The transcriptional orientations of both inserted genes are indicated by arrows. The following sites for restriction endonucleases have been mapped:H, HindIII; C, ClaI; E, EcoRI; S, SacI; K,KpnI. Hybridization probes used for genotyping were obtained from sequences flanking the vector insert on both sites as indicated. (B) A typical Southern blot analysis with DNA taken from a litter produced by heterozygous parents. DNA was digested with EcoRI and hybridized with the 3′probe. Wild type and mutant alleles are represented by 7.0 (top, arrowhead)and 4.0 (bottom, arrowhead) kb restriction fragments, respectively. (C) RNA isolated from brain and muscle of wild-type, heterozygous and homozygous adult mice was analyzed using RT-PCR (top). Nkx2.9-specific transcripts were detected in brain of wild-type and heterozygous mutant mice but not in homozygous mutants. RNA from muscle contained no Nkx2.9 transcripts and served as control for PCR specificity (bottom). RNA loading was controlled by RT-PCR for the constitutively expressed ribosomal protein L7.
The targeting strategy to inactivate the Nkx2.9 gene. (A)Illustrates the genomic organization of the Nkx2.9 locus (wt), the targeting vector, and the mutated allele after homologous recombination (mut). Most of the exons (red boxes), including the homeodomain contained in exon 2,were replaced by the IRES-lacZ reporter gene (small green/blue box)and the neomycin selection cassette (large green box). IRES-lacZ-coding sequence was inserted at the translation initiation codon. The transcriptional orientations of both inserted genes are indicated by arrows. The following sites for restriction endonucleases have been mapped:H, HindIII; C, ClaI; E, EcoRI; S, SacI; K,KpnI. Hybridization probes used for genotyping were obtained from sequences flanking the vector insert on both sites as indicated. (B) A typical Southern blot analysis with DNA taken from a litter produced by heterozygous parents. DNA was digested with EcoRI and hybridized with the 3′probe. Wild type and mutant alleles are represented by 7.0 (top, arrowhead)and 4.0 (bottom, arrowhead) kb restriction fragments, respectively. (C) RNA isolated from brain and muscle of wild-type, heterozygous and homozygous adult mice was analyzed using RT-PCR (top). Nkx2.9-specific transcripts were detected in brain of wild-type and heterozygous mutant mice but not in homozygous mutants. RNA from muscle contained no Nkx2.9 transcripts and served as control for PCR specificity (bottom). RNA loading was controlled by RT-PCR for the constitutively expressed ribosomal protein L7.
Genotyping of mice was routinely performed by PCR using the following primers in standard reactions: NkxMUTsense, AGCTCATTCCTCCCACTCATG; NkxWTsense,ACCACCGCTACAAGCTGAAGC; Nkx antisense, GGTGGTGCTAAGTGCTGGTAG. PCR conditions were 94°C for 2 minutes followed by 38 cycles of 94°C for 1 minute,60°C for 45 seconds and 72°C for 1 minute. Wild-type and mutant alleles were identified as 289 bp and 400 bp DNA fragments, respectively.
β-Galactosidase and neurofilament staining, and in situ hybridization on whole-mount embryos
Isolated embryos were briefly rinsed with phosphate-buffered saline (PBS). For β-galactosidase staining, embryos were fixed in 0.2% glutaraldehyde dissolved in PBS containing 5 mM EGTA and 2 mM MgCl2 for 15 minutes, washed three times for 15 minutes each in PBS containing 5 mM EGTA, 2 mM MgCl2, 0.01% NP-40 and 0.1% sodium desoxycholate. Staining was performed in the same buffer with 10 mM K3Fe(CN)6, 10 mM K4Fe(CN)6 and 0.5 mg/ml X-Gal at 37°C. Tissue staining was evaluated after dehydration in graded ethanol and clearing the embryos in benzylethanol/benzaldehyde in 2:1 ratio. For whole-mount in situ hybridization and antibody staining, embryos were fixed in 4% paraformaldehyde(PFA) at 4°C overnight, dehydrated in graded methanol and stored at–20°C. For the actual experiments, embryos were rehydrated, bleached with 6% H2O2 in PBT (PBS containing 0.01% Tween-20),treated with Proteinase K dissolved in PBT (1 μg/ml) for 10 minutes, washed in PBT containing 10 mg/ml glycine, and rewashed three times in PBT. Embryos were then refixed in 4% PFA/0.2% glutaraldehyde dissolved in PBT for 20 minutes followed by three washes in PBT. For whole-mount staining of neurofilaments, embryos were blocked with 10% horse serum in PBT overnight at 4°C and then incubated with the first antibody (partially purified IgG clone 2H3 from Developmental Studies Hybridoma Bank of Iowa, DSHB) diluted 1:900 in 10% horse serum/PBT. After incubation, embryos were washed five times for 1 hour each with 10% horse serum/PBT and then overnight in PBT alone. Incubations with second antibody (HRP anti mouse from Vector Laboratories) and washing were carried out as described above (1:300 anti-serum dilution in PBT). The staining solution contained 0.1 mg DAB and 0.5 μl H2O2 per ml PTB. Whole-mount in situ hybridization was performed as described previously (Bober et al., 1994). Riboprobes for Shh, Nkx2.2, Nkx2.9, Dbx2, Nkx6.1,Pax3, Pax6 and Pax7 have been described previously(Pabst et al., 2000;Sander et al., 2000).
Staining for β-galactosidase and acetylcholinesterase on tissue sections
Dissected tissues were cryoprotected in 30% sucrose/PBS without prior fixation and frozen on dry ice in OCT. Cryosections (14 μm) were collected on coated glass slides and dried at 50°C for 1 hour. β-Galactosidase staining was performed as described above for whole-mount preparations except that fixation and washing steps lasted only 5 minutes. Sections for acetylcholinesterase staining were hydrated in PBS and incubated in 65 mM sodium acetate (pH 6.0), 5 mM sodium citrate, 3 mM copper sulfate, 0.5 mM potassium ferricyanide and 0.5 mg/ml acetylthiocholine iodide for 2 hours at 37°C.
Immunohistochemistry
Embryos were fixed as described above, dehydrated in graded ethanol,treated with xylene alone, followed by a 1:1 mixture of xylene and paraffin wax, and then embedded in paraffin wax. Paraffin blocks were cut at 6 μm and then collected on Vectabond-coated glass slides. Each slide was photographed to select sections of the identical the anteroposterior level from different embryos. Paired sections were then used for double immunofluorescence staining with two monoclonal mouse antibodies. Slides were dewaxed and endogenous peroxidase activity was abolished by incubation in a 9:1 mixture of methanol and H2O2 for 1 hour. Sections were then incubated in blocking solution (0.1% Triton-X100, 10% horse serum in PBS) for 2 hours. The first antibody was applied over night at 4°C. HRP-conjugated anti-mouse antibody was then added, diluted 1:200 in blocking solution and the slides were developed using FITC-labeled tyramide as substrate according to the instructions of the manufacturer (NEN). To prevent crossreactivity with the next antibody, pre-stained sections were incubated in a 1:20 dilution of sheep anti-mouse Fab fragments (Dianova) in blocking solution for 1 hour. Subsequently the second primary antibody was applied in blocking solution for 2 hours at room temperature and visualized with Cyanin3-coupled anti-mouse antibody (Sigma). Mouse monoclonal antibodies against Nkx2.2, Isl1(clone 2D6), Pax6 and Lim3 were obtained from DSHB. Dilutions (1:500) of partially purified IgG fractions were used for the first and 1:100 dilutions for the second reaction. Phox2b detection with rabbit Phox2b antiserum was performed as described previously(Pattyn et al., 1997).
RESULTS
Targeted disruption of the mouse Nkx2.9 gene
To investigate the role of Nkx2.9 during mouse development and in adult animals, we inactivated the gene by homologous recombination in ES cells as outlined in Fig. 1A. TheNkx2.9 gene consists of two exons separated by an intron of 800 bp. The targeting vector was designed to delete the first and most of the second exon, including the homeodomain, and to introduce the lacZ reporter gene at the endogenous translation start site, followed by the PGK-neomycin resistance cassette in reverse orientation. Among 400 analyzed ES cell clones,22 had integrated the vector by homologous recombination as determined by Southern blot analysis using the hybridization probes shown inFig. 1A. Two of these ES cell clones were injected into blastocysts and yielded male chimeras that transmitted the mutant allele to their offspring. The mutant allele was crossed into C57/BL6 and 129Sv mouse strains. No difference in phenotypes in the two genetic backgrounds was apparent. Data presented were all derived from mice on mixed C57/BL6/129Sv background.
Crosses of heterozygous Nkx2.9 parents produced homozygous offspring with the expected Mendelian frequency (Fig. 1B). The mutants appeared morphologically normal and showed no signs of grossly altered behavior over the observation period of more than 1 year. Using RT-PCR, Nkx2.9 transcripts were readily detectable in brains of wild-type animals and at a reduced level in heterozygous but not in homozygous mutants (Fig. 1C). The same results were obtained by whole-mount in situ hybridization of embryos using a Nkx2.9-specific riboprobe (data not shown). These observations confirmed that the gene disruption has caused a Nkx2.9-null mutation.
Expression of the Nkx2.9/lacZ reporter gene in heterozygous and homozygous mutants
First, we analyzed the expression pattern of theNkx2.9/lacZ reporter gene in heterozygous embryos in order to test whether it recapitulated endogenous Nkx2.9 expression, as previously determined by in situ hybridization(Pabst et al., 1998). Weakβ-gal activity was first detected in E8.0 embryos in the most ventral aspect of the neural tube along the entire neural axis. Expression persisted until E9.5, mimicking the endogenous Nkx2.9 pattern(Fig. 2A,B). In E11.5 and E12.5 embryos, lacZ activity was still present along the anteroposterior axis (Fig. 2C,D) continuing until E17.5 in the caudal spinal cord (data not shown). The developmental timecourse of this pattern was not entirely consistent with the previously reported downregulation of Nkx2.9 transcripts in neural tube beginning at E10(Briscoe et al., 1999;Pabst et al., 1998). Whether the difference in temporal expression reflects high stability of β-gal transcripts or protein, or disturbance of regulatory elements within the mutated Nkx2.9 locus remains to be clarified. In late embryonic and early postnatal stages (E17.5 to P3), lacZ activity was found in hypothalamus and in the third ventricle, prominently in the subfornical organ. Expression in the subfornical organ and the median eminence remained detectable in brains of adult mice. Additional lacZ activity was observed in lung epithelium of embryos and adult animals (data not shown).
Expression of the Nkx2.9-lacZ reporter gene in heterozygous (A-E) and homozygous (F) Nkx2.9 mutants. lacZstaining of whole-mount embryos at E8.5 (A), E9.5 (B), E11.5 (C) and E12.5(D). Insets in A and B show whole-mount in situ hybridization withNkx2.9-specific probe. Note the slightly delayed onset oflacZ activity compared with endogenous Nkx2.9 transcription in the E8.5 embryo (A) but a comparable expression in the E9.5 embryo (B).lacZ activity continues to be present in Nkx2.9 domains of embryos at E11.5 (C) and E12.5 (D). Comparison of lacZ staining in E10.5 heterozygous (E) and homozygous (F) mutants reveals identical activity patterns except for the enlarged domain in hindbrain of homozygousNkx2.9 mutants at the level of rhombomeres 3 and 4 (arrows in E,F). Inset in E,F shows lacZ staining on transverse sections of spinal cord at the forelimb level.
Expression of the Nkx2.9-lacZ reporter gene in heterozygous (A-E) and homozygous (F) Nkx2.9 mutants. lacZstaining of whole-mount embryos at E8.5 (A), E9.5 (B), E11.5 (C) and E12.5(D). Insets in A and B show whole-mount in situ hybridization withNkx2.9-specific probe. Note the slightly delayed onset oflacZ activity compared with endogenous Nkx2.9 transcription in the E8.5 embryo (A) but a comparable expression in the E9.5 embryo (B).lacZ activity continues to be present in Nkx2.9 domains of embryos at E11.5 (C) and E12.5 (D). Comparison of lacZ staining in E10.5 heterozygous (E) and homozygous (F) mutants reveals identical activity patterns except for the enlarged domain in hindbrain of homozygousNkx2.9 mutants at the level of rhombomeres 3 and 4 (arrows in E,F). Inset in E,F shows lacZ staining on transverse sections of spinal cord at the forelimb level.
The lacZ expression pattern in spinal cord of heterozygous and homozygous embryos at E10.5 was essentially the same, suggesting that Nkx2.9 activity is not required to establish progenitors of V3 interneurons that express the gene in wild-type animals (Fig. 2E,F and data not shown). In the hindbrain, however, a subtle change in the lacZ pattern of expression was apparent(Fig. 2E,F). Although a region of very low β-gal staining at the level of rhombomeres 3 and 4 was seen in heterozygous animals, the corresponding area in homozygous embryos showed intensified activity in a dorsally expanded domain. This may indicate that a pool of distinct neuronal progenitor cells was expanded or alternatively the reporter gene was aberrantly expressed in the absence of Nkx2.9 protein. During later developmental stages (E13.5 to E17.5), and in early postnatal and adult brains, however, this difference in Nkx2.9 expression was no longer observed. Taken together the data of the reporter gene expression suggested that Nkx2.9 is not essential to establish or maintain the early neuronal progenitor cells from which V3 neurons and branchial motoneurons are generated.
Neuronal differentiation in the spinal cord is not affected by theNkx2.9 mutation
In spinal cord, Nkx2.9, like Nkx2.2, is expressed in the ventral domain of neuronal progenitors that give rise to V3 interneurons. This domain has been referred to as `x'-region or p3-domain(Yamada et al., 1991). To search for a potential phenotype of the Nkx2.9-deficient mouse, we analyzed the distribution of neuronal cell types within the spinal cord at forelimb and hindlimb levels of heterozygous and homozygous E10.5 mutant embryos, using whole-mount in situ hybridization and antibody staining for the appropriate marker molecules.
The most ventral cell population in the neural tube forms the floorplate that expresses Shh and the transcription factor HNF3β. Expression of both floorplate markers was unchanged in Nkx2.9 mutants, suggesting that Nkx2.9 has no role in the formation and maintenance of floorplate(Fig. 3A,B; data not shown). The dorsally adjacent p3 domain of V3 progenitor cells is characterized by co-expression of Nkx2.9, Nkx2.2 and Nkx6.1, and the lack of Pax6 that is expressed dorsolateral to the p3 domain throughout the neural tube. Immunofluorescence staining showed that the pattern of Nkx2.2expression was indistinguishable in heterozygous and homozygousNkx2.9 mutant mice (Fig. 3C,D,G,H). Likewise, the expression domains of theNkx2.9-lacZ reporter gene and Nkx6.1 were unaltered in mutant embryos(Fig. 2E,F; data not shown). Moreover, the number of cells within the p3-domain appeared unchanged(Fig. 3). The ventral boundary of the Pax6-expressing domain was maintained at its normal position(Fig. 3G,H), and more dorsally located progenitor regions, as exemplified by the expression pattern of Dbx2 for V1 neuronal precursors, were also unaffected in Nkx2.9 mutants(Fig. 3K,L). In situ hybridization with a variety of additional probes for marker transcripts within the neural tube, including Pax3, Pax6, Pax7, Nkx6.1 and Nkx2.2,confirmed the normal patterning of spinal cord in Nkx2.9 mutant embryos (data not shown). In Nkx2.2-deficient mice, V3 progenitors failed to differentiate and the number of somatic motoneurons was increased with many of them located next to the floorplate within the p3 domain(Briscoe et al., 1999). To test thoroughly whether a similar cell fate transformation occurred inNkx2.9 mutants, we investigated the distribution of both neuronal cell types in E10.5 and E11.0 embryos by in situ hybridization using a Sim1-specific probe and immunofluorescence staining using neurogenin 3 (Ngn3),islet 1 (Isl1) and Nkx2.2 antibodies. The Sim1 expression domain was unchanged in the mutant, indicating that V3 neurons had been established normally(Fig. 3I,J). This was confirmed by the normal staining pattern for Ngn3, which also labels V3 neurons (data not shown). There was also no difference in the relative location or number of Isl1-positive and Nkx2.2-expressing cells between heterozygous and homozygous Nkx2.9 mutants, indicating that somatic motoneurons were formed correctly in their defined territory(Fig. 3C-F). It should be mentioned, however, that on sections through more-caudal segments of spinal cord (at and posterior to hindlimb level), we occasionally but consistently found single Isl1-positive cells within the Nkx2.2 domain ofNkx2.9 mutants (Fig. 3F). This was never observed in wild-type embryos(Fig. 3E). Significantly, the individual Isl1-positive cells within the Nkx2.2 domain failed to express Nkx2.2, suggesting that expression of both genes within one cell is mutually exclusive. Whether in the absence of Nkx2.9 these putative motoneurons were generated within the wrong domain or became misplaced by migration is not clear. All observations taken together indicate that Nkx2.9 function is not essential to establish or maintain the p3-domain in spinal cord nor to generateV3 neurons. Dorsoventral patterning of the spinal cord and the determination of neuronal cell subtypes occurred correctly in the absence of Nkx2.9, with the exception of few motoneurons that occasionally appeared within the Nkx2.2 domain in posterior segments of the spinal cord.
Dorsoventral patterning in spinal cord is not affected by theNkx2.9 mutation. Immunohistochemistry on transverse sections of spinal cord at forelimb (A-D,G,H) and hindlimb level (E,F) demonstrates the expression of marker genes for distinct subpopulations of neuronal progenitors in E10.5 heterozygous (A,C,E,G) and homozygous (B,D,F,H) Nkx2.9mutant embryos. Immunofluorescence staining for HNF3β marks the floorplate (A,B), Nkx2.2 (red in C-H) and Isl1 (green in C-F) label V3 interneurons and motoneurons, respectively. Pax6 (green in G,H) labels all neuronal precursors except V3 neurons and floorplate. No major differences are seen between heterozygous and homozygous animals. Note, however, individual Isl1-positive cells within the Nkx2.2 domain of the neural tube at hindlimb level (E,F). In situ hybridization with Sim1- (I,J) and Dbx2- (K,L) specific probes on transverse sections illustrates normal production of V3 neurons and dorsoventral patterning of the V1 progenitor domain in neural tube ofNkx2.9-deficient embryos, respectively. (I-L) Heterozygous (I,K) and homozygous (J,L) mutant embryos.
Dorsoventral patterning in spinal cord is not affected by theNkx2.9 mutation. Immunohistochemistry on transverse sections of spinal cord at forelimb (A-D,G,H) and hindlimb level (E,F) demonstrates the expression of marker genes for distinct subpopulations of neuronal progenitors in E10.5 heterozygous (A,C,E,G) and homozygous (B,D,F,H) Nkx2.9mutant embryos. Immunofluorescence staining for HNF3β marks the floorplate (A,B), Nkx2.2 (red in C-H) and Isl1 (green in C-F) label V3 interneurons and motoneurons, respectively. Pax6 (green in G,H) labels all neuronal precursors except V3 neurons and floorplate. No major differences are seen between heterozygous and homozygous animals. Note, however, individual Isl1-positive cells within the Nkx2.2 domain of the neural tube at hindlimb level (E,F). In situ hybridization with Sim1- (I,J) and Dbx2- (K,L) specific probes on transverse sections illustrates normal production of V3 neurons and dorsoventral patterning of the V1 progenitor domain in neural tube ofNkx2.9-deficient embryos, respectively. (I-L) Heterozygous (I,K) and homozygous (J,L) mutant embryos.
Nkx2.9 mutants show defects of the spinal accessory nerve
Progenitors in the p3 domain of the hindbrain also co-expressNkx2.9 and Nkx2.2 but give rise to branchio-visceral motoneurons rather than to V3 neurons, as they do in the spinal cord(Ericson et al., 1997b;Tanabe et al., 1998). Interestingly, the correct formation of branchial, visceral and somatic motoneurons in hindbrain of Nkx2.2-deficient mice was not affected,probably because of the overlapping expression of Nkx2.9 and functional redundancy of both transcription factors(Briscoe et al., 1999). To test whether Nkx2.9 may have a role in specifying neuronal cell types in hindbrain,we first analyzed the nerve pattern in the cranial region of E10.5 and E11.5 embryos by whole-mount immunostaining using the anti-neurofilament antibody 2H3. Although most cranial nerves appeared unaffected in homozygous mutant embryos, the spinal accessory nerve (XIth) containing axons of purely branchial motoneurons was abnormal. The bundle of axon trajectories of this nerve was consistently thinner and shorter in homozygous Nkx2.9mutants than in wild-type animals (Fig. 4A-D). A significant fraction of mutant embryos showed additional abnormalities of the glossopharyngeal (IXth) and vagal (Xth) nerves, which appeared to be partially fused and axons, particularly in the vagal nerve,were reduced compared with wild-type animals(Fig. 4A-D). The truncation of the spinal accessory nerve was even more pronounced in E11.5 mutant embryos(Fig. 4E-H). The hypoglossal nerve (XIIth), a somatic motor nerve, and the cervical motor nerves were morphologically unaltered with normal ventral axon projections. These observations were a first indication that the Nkx2.9 null-mutation affected the normal formation of some cranial nerves, most notably those containing axons of the branchial motoneuron subtype. By contrast, nerves of the somatic motoneuron type appeared to be normal. It is important to note,however, that the Nkx2.9 mutation did not cause complete absence of any brainstem nerves but rather seemed to result in a partial loss of neurons,presumably of a particular subtype.
Neurofilament staining reveals nerve defects in hindbrain ofNkx2.9 mutants. Whole-mount preparations of wild type (A,C,E,G) and homozygous Nkx2.9 mutant (B,D,F,H) embryos at E10.5 (A-D) and E11.5(E-H) were stained with anti-neurofilament antibody. Note the considerably reduced length and thickness of the spinal accessory nerve in mutants (compare arrows in C,D and G,H). (D) Partial fusion of the N. vagus (X) with the N. glossopharyngeus (IX) were frequently observed in homozygous mutant animals(arrowhead). In addition, the axon bundle of the vagal nerve appears reduced.
Neurofilament staining reveals nerve defects in hindbrain ofNkx2.9 mutants. Whole-mount preparations of wild type (A,C,E,G) and homozygous Nkx2.9 mutant (B,D,F,H) embryos at E10.5 (A-D) and E11.5(E-H) were stained with anti-neurofilament antibody. Note the considerably reduced length and thickness of the spinal accessory nerve in mutants (compare arrows in C,D and G,H). (D) Partial fusion of the N. vagus (X) with the N. glossopharyngeus (IX) were frequently observed in homozygous mutant animals(arrowhead). In addition, the axon bundle of the vagal nerve appears reduced.
Branchial motoneuron progenitors are reduced in Nkx2.9mutants
Significantly, the most visibly affected nerve in mutant embryos, the nervus accessorius (XIth) consists of branchial motoneurons in line with the expression of Nkx2.9 in progenitor cells of this neuronal subtype. The branchial motoneurons of the spinal accessory nucleus (SAN) are generated in the ventral neural tube at the C4-C3 level(Krammer et al., 1987;Liinamaa et al., 1997)together with somatic motoneurons of the median motor column (MMC)(Callister et al., 1987) and phrenic motoneurons (Goshgarian and Rafols, 1981). At midgestation, cell bodies of MMC neurons are located in a ventromedial position and project axons ventrally through segmental motor nerves, while SAN neurons migrate to a dorsolateral position and project axons via the spinal accessory nerve. These spinal accessory neurons express Isl1 but not Isl2 and Lim3/Lhx3, whereas MMCs express Isl1 together with Isl2 and Lim3 (Ericson et al., 1997b).
We analyzed the neuronal progenitor cell populations at different hindbrain levels and in the anterior spinal cord. Immunostaining of E10.5 embryos with Isl1-specific antibody revealed no significant difference in total numbers of Isl1-positive cells in the ventromedial position between wild-type and homozygous mutant embryos, suggesting that motoneurons were generated in the absence of Nkx2.9, occupying the correct domain dorsally adjacent to theNkx2.2-expressing cells (Fig. 5A,B). If anything, a slight but statistically not significant increase of Isl1-positive cells was observed in the mutant. On serial transverse sections through the spinal cord at C4-C3 level, however, the number of laterally migrating Isl1-positive cells representing immature SANs was drastically reduced (Fig. 5). Cell counts showed 60 to 70% loss of these cells inNkx2.9-null mutant embryos compared with wild-type or heterozygous animals. We also determined the number of Isl1/Lim3 co-expressing somatic motoneurons of the MMC and found a marked increase of these cells in mutant embryos by approximately the same margin by which branchial motoneurons of the SAN were decreased (Fig. 5C,D). From these results, we conclude that lack of Nkx2.9 causes the formation of supernumerary somatic motoneurons of the MMC at the expense of branchial motoneurons at the level of C3-C4, consistent with fewer axons projecting from the SAN into the spinal accessory nerve.
Branchial motoneurons of the spinal accessory nerve are reduced inNkx2.9-deficient mouse embryos. Transversal serial sections through the neural tube at C4-C3 level of E 10.5 wild-type (A,C) and homozygous mutant(B,D) embryos were stained with Isl1-specific antibody (green) together with Nkx2.2 antibody (red in A,B) or Lim3 antibody (red in C,D). Note the drastic reduction of dorsolaterally migrating Isl1-positive cells (compare arrows in A and B) and the markedly increased number of Isl1/Lim3-positive somatic motoneurons (yellow cells indicated by arrows in C,D) in the mutant. Quantification of migrating Isl1-positive and Isl1/Lim3 double-positive cells was obtained by computer-aided determination of total pixel area with color information for the respective cell types. (C′,D′) A typical example of at least five independent determinations on level-matched sections of embryos of both genotypes. Changes in cell numbers of approximately 60%were consistently found.
Branchial motoneurons of the spinal accessory nerve are reduced inNkx2.9-deficient mouse embryos. Transversal serial sections through the neural tube at C4-C3 level of E 10.5 wild-type (A,C) and homozygous mutant(B,D) embryos were stained with Isl1-specific antibody (green) together with Nkx2.2 antibody (red in A,B) or Lim3 antibody (red in C,D). Note the drastic reduction of dorsolaterally migrating Isl1-positive cells (compare arrows in A and B) and the markedly increased number of Isl1/Lim3-positive somatic motoneurons (yellow cells indicated by arrows in C,D) in the mutant. Quantification of migrating Isl1-positive and Isl1/Lim3 double-positive cells was obtained by computer-aided determination of total pixel area with color information for the respective cell types. (C′,D′) A typical example of at least five independent determinations on level-matched sections of embryos of both genotypes. Changes in cell numbers of approximately 60%were consistently found.
In order to test whether this also applies to other progenitors of the branchial motor column, we performed immunostaining on serial transverse sections through hindbrain of E10.5 embryos, employing Phox2b-specific antibodies that mark visceral and branchial motoneurons and Isl1-specific antibodies that label postmitotic motoneurons(Pattyn et al., 2000;Pattyn et al., 1997). At the level of rhombomeres 4/5, both wild-type and mutant embryos displayed the normal pattern of Phox2b-positive cells in the ventral, lateral and dorsal domains, and in a dorsoventral string of cells at the lateral aspect of the neural tube that probably correspond to the presumptive motoneurons of the facial nucleus (Fig. 6A,E). Likewise, Isl1-positive cells appeared unchanged in the mutant(Fig. 6B,F). In keeping with the normal appearance of the facial nerve in mutants, this result suggests that lack of Nkx2.9 does not affect all branchial motoneurons but rather a subset. Similar immunostaining at more caudal level (r7), where progenitors of the nucleus ambiguus arise, also did not show statistically significant differences of Phox2b- and Isl1-positive cells between wild-type and mutant embryos, suggesting that normal numbers of motoneurons have been born(Fig. 6C,D,G,H).
Phox2b-expressing progenitors of visceral and branchial motoneurons in hindbrain of E10.5 wild-type and Nkx2.9 mutant mouse embryos. Transverse sections through hindbrain at the level of rhombomere 4 (A,B,E,F)and rhombomere 7 (C,D,G,H) were immunostained with Phox2b-(A,E,C,G) and Isl1-specific antibodies (B,F,D,H). Sections of wild-type (A-D) and mutant embryos (E-H) reveal similar patterns of Phox2b expression in mitotic and postmitotic cells of the ventral and lateral domain, and in the mantel layer of the dorsal domain at r4 level. At r7 only postmitotic cells of the ventral domain and a dorsoventral stripe of presumably migratory cells express Phox2b. Isl1-positive cells are found only in postmitotic cells of the ventral domain both at r4 and r7 levels. Note that the expression of both markers appears essentially unchanged in wild-type and mutant embryos.
Phox2b-expressing progenitors of visceral and branchial motoneurons in hindbrain of E10.5 wild-type and Nkx2.9 mutant mouse embryos. Transverse sections through hindbrain at the level of rhombomere 4 (A,B,E,F)and rhombomere 7 (C,D,G,H) were immunostained with Phox2b-(A,E,C,G) and Isl1-specific antibodies (B,F,D,H). Sections of wild-type (A-D) and mutant embryos (E-H) reveal similar patterns of Phox2b expression in mitotic and postmitotic cells of the ventral and lateral domain, and in the mantel layer of the dorsal domain at r4 level. At r7 only postmitotic cells of the ventral domain and a dorsoventral stripe of presumably migratory cells express Phox2b. Isl1-positive cells are found only in postmitotic cells of the ventral domain both at r4 and r7 levels. Note that the expression of both markers appears essentially unchanged in wild-type and mutant embryos.
Analysis of brainstem nuclei in adult mutant mice
Given the reduced number of branchial motoneuron progenitors of the SAN in mutant mouse embryos, the altered morphology of the accessory nerve, and the,albeit not fully penetrant, alterations of the vagal and glossopharyngeal nerves, we sought to analyze the corresponding mature neurons in brain stem of adult animals. Nuclei projecting branchio-efferent fibers to the glossopharyngeal (IX), vagus (X) and the cranial region of the spinal accessory (XI) nerves are located in the nucleus ambiguus, while the spinal region of the XIth brainstem nerve is classically said to come from the SAN. Visceral motoneurons of the vagus are located in the dorsal motor nucleus of the vagus (dmnX). We performed cell counts using Nissl staining and immunohistochemistry for acetylcholineesterase on serial transverse sections through hindbrain of adult mice and found no significant differences in the size of the dmnX and the nucleus ambiguus between wild-type and mutant mice(Fig. 7A,C,F,H). Significantly,however, immunostaining of parallel sections for Phox2b revealed a drastic reduction of Phox2b-positive cells in the nucleus ambiguus of mutant mice,whereas the number of these cells appeared unchanged in the dmnX(Fig. 7E,J). These results then suggest that Nkx2.9 is essential for maintenance of Phox2b expression and probably terminal differentiation of branchial motoneurons, while Phox2b expression in visceral motoneurons of the dmnX is apparently not dependent on Nkx2.9. A differentiation defect of branchial motoneurons in the nucleus ambiguus would be consistent with the mutant phenotype of the vagus and glossopharyngeus nerves in the Nkx2.9 mutant. Unfortunately, the SAN located in the rostral spinal cord was not amenable to this type of analysis in our hands, because these neurons do not form a conspicuous nucleus.
Phox2b expression in the dmnX and nucleus ambiguus (NA) of adult wild-type(A,B,F,G) and homozygous Nkx2.9 mutant (C,D,H,I) mice. Nissl-stained transverse sections through hindbrains (A,C,F,H) reveal normal morphology of the dmnX (A,C) and Nucleus ambiguus (F,H). (B,D,G,I) Parallel sections immunostained with Phox2b-specific antibody. Note the drastically reduced number of Phox2b-positive cells in the NA (G,I) but not in the dmnX (B,D). Graphs illustrate the ratio of Nissl-positive cell counts (total) and the ratio of Phox2b-positive cells (Phox2b+) from mutants versus wild-type animals in dmnX (E) and NA (J). Each column represents the mean of five slides counted from three individual animals for each genotype.
Phox2b expression in the dmnX and nucleus ambiguus (NA) of adult wild-type(A,B,F,G) and homozygous Nkx2.9 mutant (C,D,H,I) mice. Nissl-stained transverse sections through hindbrains (A,C,F,H) reveal normal morphology of the dmnX (A,C) and Nucleus ambiguus (F,H). (B,D,G,I) Parallel sections immunostained with Phox2b-specific antibody. Note the drastically reduced number of Phox2b-positive cells in the NA (G,I) but not in the dmnX (B,D). Graphs illustrate the ratio of Nissl-positive cell counts (total) and the ratio of Phox2b-positive cells (Phox2b+) from mutants versus wild-type animals in dmnX (E) and NA (J). Each column represents the mean of five slides counted from three individual animals for each genotype.
DISCUSSION
In this study we demonstrate that the Nkx2.9 gene product is indispensable for the generation of a subset of hindbrain neurons during mouse embryogenesis but plays apparently no role in neuronal differentiation in the spinal cord at trunk level. We have previously shown that Nkx2.9 is most closely related to the Nkx2.2 gene, both in sequence homology and in its spatiotemporal expression pattern(Pabst et al., 1998). Expression of both genes depends on Shh(Pabst et al., 1998;Pabst et al., 2000), and both are likely to be involved in interpreting graded Shh signaling for cell fate decisions (Briscoe et al.,1999). Conserved linkage of the homeobox gene pairsNkx2.2/2.4 and Nkx2.9/2.1 has been reported (Wang et al., 2000),suggesting that duplication of an ancestral gene cluster led to the four closely related proteins in mice. Taken together, these observations tend to suggest that Nkx2.9 and Nkx2.2 fulfil similar functions during mouse development, which may also be true for the Nkx2.1/2.4 gene pair.
In mice that lack the Nkx2.2 gene, neurons that normally arise in territories of the neural tube that are exposed to lower Shh activity now form ectopically in more ventral domains with higher Shh activity(Briscoe et al., 1999). This observation has been interpreted as ventral-to-dorsal transformation of progenitor cell fates. Lack of Nkx2.2, however, does not affect the establishment of neuronal cells in ventral spinal cord but rather seems to control their position-specific differentiation. Interestingly, a similar switch in neuronal cell fate has been observed in Drosophila that lack the NK2/vnd gene, suggesting that NK2 functions have been conserved during evolution (McDonald et al.,1998). Mice carrying the Nkx2.9 mutation, as reported here, do not show the transformation of neuronal cell fates in the spinal cord with the exception of few individual Isl1-positive putative motoneurons that are located ectopically within the Nkx2.2 domain of the posterior neural tube. Why these rare and misfated cells arise is not clear, but it may either reflect a crucial threshold level of Nkx2.2 or a very small subpopulation of cells which normally express Nkx2.9 only.Nkx2.9 mutant mice develop the normal V3 neuron progenitor domain(p3-domain) but also form the correct compartments of mature V3 neurons and somatic motoneurons, as judged by the expression of Sim1, and maintenance of the normal Nkx2.2 and Isl1-positive cell domains. These results clearly demonstrate that, unlike Nkx2.2, Nkx2.9 is not required to generate V3 neurons from P3-domain progenitors. A likely explanation of this phenotype is the overlapping expression and redundant activity of Nkx2.2 that may substitute for the missing Nkx2.9. Indeed, individual overexpression of Nkx2.2 orNkx2.9 is sufficient to induce the expression of Sim1 and generate V3 neurons throughout the Nkx6.1 domain, and each factor alone can suppress somatic motoneuron fate by repressing Pax6 expression(Briscoe et al., 2000). This of course raises the question of why loss of Nkx2.2 is apparently not rescued byNkx2.9, although the latter gene is expressed in the absence of Nkx2.2 during early embryogenesis (Briscoe et al., 1999). The explanation for the distinct phenotypes ofNkx2.2 and Nkx2.9 mutants may come from the different temporal expression patterns of both genes. While Nkx2.9 in ventral spinal cord at trunk level is rapidly downregulated after establishment of the p3 domain, Nkx2.2 expression persists much longer until V3 neurons are definitively determined. In fact, an early rescue function of Nkx2.9 can be deduced from the observation that few Sim1-positive cells have been observed occasionally in Nkx2.2-deficient mice early on, while these cells were completely lost at later developmental stages(Briscoe et al., 1999). Clarification of whether both Nkx transcription factors exert redundant functions or even participate in the establishment of the p3-domain at early stages awaits the generation of the double mutant mouse that lacks bothNkx2.2 and Nkx2.9 gene products.
Expression of Nkx2.9 in brain parallels that of Nkx2.2and continues until at least E11 of embryogenesis, in contrast to its early repression in spinal cord. In line with the prolonged presence of Nkx2.9 in brain domains and supporting the idea of redundant activities,Nkx2.2-deficient mice exhibit no obvious phenotypic switch in motoneuron identity within the hindbrain(Briscoe et al., 1999). In the reverse situation presented here by the Nkx2.9 knockout mouse, e.g. lack of Nkx2.9 but continuing presence of Nkx2.2, the phenotypic rescue is at least incomplete. In hindbrain neuronal progenitors of the p3 domain that co-express Nkx2.2 and Nkx2.9 give rise to branchiovisceral motoneurons (Ericson et al.,1997b). These cells can be identified by the expression of the transcription factor Phox2b (Dubreuil et al., 2000; Pattyn et al.,2000; Pattyn et al.,1997). In E10.5 mutant embryos, Phox2b expression appears essentially normal in hindbrain at different axial levels, indicating that the formation of branchiovisceral motoneurons is not generally dependent on Nkx2.9 function but rather affects a subset or only some of these cells. Significantly, the population of presumptive branchial motoneurons of the spinal accessory nucleus, which are characterized by expressing Isl1 alone and their dorsolateral position in neural tube, is markedly reduced in the mutant mouse. Moreover, the population of somatic motoneurons of the median motor column, which co-express Isl1, Isl2 and Lim3, is increased. This result somewhat resembles the cell fate switch observed in the spinal cord ofNkx2.2 mutants with respect to the increase of somatic motoneurons at the expense of another neuronal subpopulation originating in theNkx2.2/Nkx2.9 domain. Whether this reflects aberrant specification of neuronal identity in response to graded Shh signaling, as it is the case in Nkx2.2 mutants, cannot be decided easily here, because the precise local relationship of the premigratory SAN progenitors and the median motor column precursors is not clear. Consistent with the reduction of branchial motoneurons in the dorsolateral position of the neural tube at the level of C4-C3 we found a severely abnormal spinal accessory nerve in all mutant mice. The partial phenotype of the glossopharyngeal and the vagal nerves in ∼50% of mutant embryos may be related to the finding that, in the absence of Nkx2.9, Phox2b expression in cells of the nucleus ambiguus is drastically reduced, although the number of cells appears largely unaltered. As Nkx2.9 is only expressed in the progenitor domain and Phox2b-positive progenitors are present in normal numbers in the neuroepithelium of mutant embryos (E10.5), but not in the mature nucleus ambiguus, it seems to have a function in maintaining the phenotypic trait of branchial motoneurons rather than specifying them. Another unknown transcription factor is possible involved acting downstream of Nkx2.9. Whether the alterations in the nucleus ambiguus also contribute to the mutant phenotype of the spinal accessory nerve appears disputable, as the existence of projections from the nucleus ambiguus has been recently questioned, at least in humans(Lachman et al., 2002). The three nerves affected in the mutant belong to the branchial motor column,suggesting that Nkx2.9 in the hindbrain has a unique role in formation of branchial motoneurons, a function that can not be fully substituted for by Nkx2.2. Clearly, visceral motoneurons are not affected by the Nkx2.9mutation, as demonstrated here for the dmnX. It is also interesting to note that the more rostrally located branchial motor nerves, such as the facialis and the trigeminus nerve, appear quite normal in mutants, suggesting a differential requirement for Nkx2.9 along the rostrocaudal axis. Whether the fractional loss of branchial motoneurons in hindbrain of Nkx2.9mutants is due to partial rescue by redundant Nkx2.2 function or,alternatively, reflects the total loss of a neuronal subpopulation whose fate is entirely dependent on Nkx2.9, cannot be decided unequivocally by the available data. The latter possibility, however, seems less likely given the reduced size but not the complete absence of the affected nerves. Taken together, our data provide evidence that Nkx2.9 is a crucial transcription factor for the determination and/or differentiation of at least a subset of branchial motoneurons during hindbrain development. Its early role in establishing the p3-domain in spinal cord remains to be determined inNkx2.2/Nkx2.9 double mutants.
Acknowledgements
We thank C. Goridis for Phox2 reagents, and M.Sander for Dbx2 and Sim1 probes. We also thank the Resource Center/Primary Database (RZPD) in Berlin for provision of the spotted mouse cosmid library, and Developmental Studies Hybridoma Bank of Iowa (DHSB) for antibodies. Help by F. Vauti and S. Willenzon in generating the mouse mutant is gratefully acknowledged. This work was supported by the Deutsche Forschungsgemeinschaft, SFB 271, TPA1 and the Fond der Chemischen Industrie.