The intervening zone (IZ) is a pool of progenitor cells located at the midbrain-hindbrain boundary (MHB) and important for MHB maintenance,midbrain-hindbrain growth and the generation of midbrain-hindbrain neurons. Recently, we implicated the Hairy/E(spl) transcription factor Her5 in the formation of the medial (most basal) part of the IZ (MIZ) in zebrafish; the molecular bases for lateral IZ (LIZ) formation, however, remain unknown. We now demonstrate that her5 is physically linked to a new family member, him, displaying an identical MHB expression pattern. Using single and double knockdowns of him and her5, as well as a him+her5 deletion mutant background (b404), we demonstrate that Him and Her5 are equally necessary for MIZ formation, and that they act redundantly in LIZ formation in vivo. We show that these processes do not involve cross-regulation between Him and Her5 expression or activities,although Him and Her5 can heterodimerize with high affinity. Increasing the function of one factor when the other is depleted further shows that Him and Her5 are functionally interchangeable. Together, our results demonstrate that patterning and neurogenesis are integrated by the her5-himgene pair to maintain a progenitor pool at the embryonic MHB. We propose a molecular mechanism for this process where the global `Him+Her5' activity inhibits ngn1 expression in a dose-dependent manner and through different sensitivity thresholds along the medio-lateral axis of the neural plate.
Development of the vertebrate central nervous system (CNS) is a complex process that needs to connect patterning and neurogenesis. Crucial to this process are local events of neurogenesis inhibition, which maintain pools of progenitors in defined locations of the neural tube. The delayed differentiation of these progenitor zones permits the generation of large as well as spatiotemporally patterned structures, such as the layered cortex in higher vertebrates, the polarized optic tectum in birds, or the retina. The molecular events controlling the generation or maintenance of neural progenitor pools remain largely hypothetical.
The embryonic midbrain-hindbrain domain (MH) is characterized by the maintenance of a zone of delayed differentiation at the midbrain-hindbrain boundary (MHB). This zone, called `intervening zone' (IZ), separates midbrain from anterior hindbrain neuronal clusters and has been described in all vertebrates (Bally-Cuif et al.,1993; Geling et al.,2003; Vaage,1969). Its functional importance is attested by genetic ablation experiments. For instance, in mouse embryos lacking the function of the two bHLH E(spl)-like transcription factors Hes1 and Hes3, premature differentiation of the IZ occurs, leading to the lack of several MH neuronal populations, and to the collapse of MH structures(Hirata et al., 2001). These defects may primarily result from a disruption of the isthmic organizer, an inducing cell population located at the MHB and involved in MH maintenance(Hirata et al., 2001; Martinez, 2001; Rhinn and Brand, 2001; Wurst and Bally-Cuif, 2001). Recent results in zebrafish have permitted dissection of the mechanisms of IZ formation in more detail. There, expression of the hairy/E(spl) gene her5 (Muller et al.,1996) precisely delineates the IZ at all embryonic stages(Geling et al., 2003). At the onset of neurogenesis (tail-bud stage), her5 expression separates the early midbrain ventrocaudal proneural cluster (vcc) from the anterior hindbrain proneural clusters of rhombomere 2 (presumptive motorneurons -r2MN-and lateral neurons -r2LN-). In the absence of Her5 function, ectopic neurogenesis occurs in the medial (future basal) part of the IZ, as revealed by the ectopic expression of the proneural genes neurogenin1(ngn1) and coe2 and the later differentiation of neurons across the basal MHB, bridging the vcc and r2MN(Geling et al., 2003; Geling et al., 2004). Conversely, forced expression of ngn1 within the MH domain leads to a partial downregulation of MHB markers' expression(Geling et al., 2003). These results have two implications. First, they confirm that the IZ is necessary to maintain MHB integrity. Second, they demonstrate that the IZ is composed of at least two domains along the mediolateral axis, which differ in their requirement for Her5 function: the medial IZ domain (MIZ), which crucially depends on Her5 for neurogenesis inhibition, and the lateral (future alar) IZ domain (LIZ), which forms even in the absence of Her5. Within the MIZ, Her5 acts as a prepattern factor that prevents the formation of a proneural cluster, and inhibits expression of ngn1 and coe2 upstream of Notch signaling (Geling et al.,2004).
Our study of Her5 function did not address the formation of the LIZ, in spite of its crucial role in controlling midbrain and anterior hindbrain alar neurogenesis. To now approach this issue, we reasoned that other Hairy/E(spl)factors might be expressed within this domain and act redundantly with Her5. Because there are examples of physically linked E(spl) genes in Drosophila (E(spl) complex)(Klambt et al., 1989; Knust et al., 1992) and zebrafish (her1 and her7)(Henry et al., 2002), and because linked genes are more likely to share spatiotemporal characteristics of expression, we searched for new Hairy/E(spl) genes in the vicinity of the her5 locus. Sequencing a her5-containing PAC revealed a new her-like gene, him, adjacent to her5 and in opposite orientation (hence him for her5image), identically expressed across the IZ. We report here that Him is the hypothetical factor cooperating with Her5 to control LIZ formation in vivo, and that Him also crucially contributes to MIZ formation. Together,our results unravel the genetic combination preventing neurogenesis across the MHB.
Materials and methods
Cloning and phylogenic analysis of him
Systematic sequencing of genomic DNA surrounding known her genes was performed using PAC clones obtained from RZPD. A new ORF was detected close to the her5 gene locus in PAC BUSMP706H15152Q2. 5′ RACE experiments were done to identify full-length cDNA, according to the manufacturer's recommendations (Invitrogen). The newly cloned gene, him, corresponds to ENSDARG00000002707 now predicted by The Welcome Trust Sanger Institute, with the exception that in the prediction the first exon is missed and the last exon is truncated.
The phylogenetic tree of zebrafish Her family was done using the phylodendrone software(www.es.embnet.org/Doc/phylodendron/treeprint-form.html). To construct the phylogenetic tree, the VectorNTI software and full-length sequences were used. The accession numbers of compared proteins are: Her5:NP_571152, Her1: NP_571153, Her7: NP_571684, Her4: NP_571165, Her2: NP_571164,Her3: NP_571155, Her9: NP_571948 and Her6: NP_571154. Proteins predicted by The Wellcome Trust Sanger Institute are: Her11 (ENSDARP00000012990), Her13(ENSDARP0000008307), Hes6 (ENSDARP00000021078) and Her12 (ENSDARP00000038100)(http:www.ensembl.org,release from 08-02-2004).
Zebrafish strains and transgenic lines
Embryos obtained from natural spawning of AB wild-type or transgenic fish,her5PAC::egfp (Tallafuss and Bally-Cuif,2003) and -8.4ngn1::egfp(Blader et al., 2003), were raised and staged according to Kimmel et al.(Kimmel et al., 1995). no isthmus (noitu29a), acerebellar(aceti282a) (Brand et al., 1996) and knypek (b404, knym119)(Solnica-Krezel et al., 1996; Topczewski et al., 2001)mutants were obtained by pairwise mating of heterozygous adult carriers, as described previously.
Protein expression interference assays
Morpholino antisense oligonucleotides (her5MOATG) were purchased from Gene-Tools, Inc. (Oregon, USA). The morpholino was dissolved to a stock concentration of 5 mM in H2O and injected into one-cell stage embryos at 1 mM. himMOs lead to non-specific cell death (not shown). Thus, GripNA antisense oligonucleotides preventing him translation, specific for the him ATG region(himGripNAATG) or acceptor site of the second himexon (himGripNASPL), were purchased from Active Motif(Belgium). GripNAs were dissolved to stock solution of 1 mM in H2O and injected into one-cell stage embryos at 0.5 mM. At this dose, the effect of himGripNAs on ngn1 expression was maximal. Sequences of antisence oligonucleotides were as follows: her5MOATG: 5′-TTGGTTCGCTCATTTTGTGTATTCC-3′; himGripNAATG: 5′-ATTCGGTGTGCTCTTCAT-3′ and himGripNASPL: 5′-TACTCACAGTGTCTGCAG-3′. All injection experiments were repeated at least three times.
In situ hybridization and immunohistochemistry
Probe synthesis, in situ hybridization and immunohistochemistry were carried out as previously described(Hammerschmidt et al., 1996). The following in situ antisense RNA probes were used: her5(Muller et al., 1996), him (this paper), ngn1(Korzh et al., 1998), pax2.1 (Lun and Brand,1998) and egfp (Clontech). Primary antibodies used for immunohistochemistry were rabbit anti-GFP (ams biotechnology Europe, TP401)used at a final dilution of 1/500 and mouse anti-human neural protein HuC/HuD(MoBiTec A-21271) (1/300). They were revealed by using FITC-conjugated goat anti-rabbit secondary antibody (Jackson ImmunoResearch Laboratories,111-095-003) or Cy3-conjugated goat anti-mouse secondary antibody (Jackson ImmunoResearch Laboratories, 115-165-044) (1/200), as appropriate. Embryos were scored and photographed under a Zeiss SV 11 stereomicroscope or a Zeiss Axioplan photomicroscope.
knypek capped RNA was synthesized using Ambion mMessage mMachine kit following the recommended procedure. Capped RNA was injected at the concentration of 60 ng/μl into the embryos at the one-cell stage.
Protein interaction assays
For two-hybrid assays, The MATCHMAKER GAL4 Two-Hybrid System 3 (Clontech)was used following procedures described by the manufacturer. The `bait' and`AD' plasmids were constructed by fusing in-frame the complete ORFs of her5, him and ngn1 to either pGBKT7 (encoding the GAL4 DNA-binding domain) or pGADT7 (encoding the GAL4 activation domain). The relative stringency of Her5 homodimerization versus its heterodimerization with Him was quantified by β-galactosidase assay according to the manufacturer's recommendation (Clontech). The β-galactosidase activity was quantified according to Lazo et al.(Lazo et al., 1978).
Co-immunoprecipitation and western blot analysis
Transformed yeast cells expressing the two proteins of interest were lysed in 0.5 ml of lysis buffer (50 mM Tris-HCl pH 8, 150 mM NaCl, 0.1% NP40, 0.1 mM DTT, 0.1 mg/ml pepstatin A, 0.03 mM leupeptin, 145 mM benzamidine, 0.37 mg/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride) with 0.25 g of glass beads(425-600 μl, Sigma) for one hour at 4°C with shaking. The extracts were then centrifuged at 15 000 g for 10 minutes at 4°C to eliminate cell debris, and the supernatant was collected. For each immunoprecipitation, 0.4 ml aliquots of lysate were precleaned by incubation with 150 μl of pre-immune rabbit serum and 100 μl of 1:1 slurry of Protein A Sepharose™ CL-4B (Amersham Bioscience AB) for 30 minutes at 4°C. Precleaned extracts were immunoprecipitated with 10 μl of rabbit anti-HA antibody (dilution 1/1000) (Sigma) and 100 μl of 1:1 slurry of Protein A Sepharose CL-4B (Amersham Bioscience AB) for 30 minutes at 4°C. The sepharose beads were washed three times with 1 ml of lysis buffer. The precipitates were fractionated on SDS-PAGE and subsequent western blot analysis was performed according to standard protocols, by using mouse anti-c-Myc antibodies (1/1000) (Sigma). The primary antibodies were revealed using HRP-coupled secondary antibodies (Jackson Laboratories) diluted to 1/200 and enhanced chemiluminescence (Amersham Bioscience AB).
Quantification of him mRNA in the her5PAC::egfp line
Total RNA was isolated from her5PAC::egfp embryos and WT siblings at the five-somite stage and reverse-transcribed before real-time PCR amplification. Real-time PCR was done by using LightCycler FastStart DNA Master SYBR Green I kit (Roche, Germany) and Light Cycler Instrument (Roche,Germany). Quantitative values were obtained from the threshold cycle number at which the increase in the signal associated with exponential growth of the PCR products begins to be detected using the LightCycler Software, according to the manufacturer's recommendations. The precise amount of total RNA added to each reaction mix (based on optical density) and its quality (lack of extensive degradation) are both difficult to assess precisely. We therefore also quantified the transcript of the pax6 gene as the endogenous RNA control, and both samples were normalized to the basis of pax6content (R-value on the graph). The nucleotide sequences of the specific primers used are shown in Table 1. The thermal cycling conditions comprised an initial denaturation step at 95°C for 10 minutes and 65 cycles at 95°C for 15 seconds, 55°C for 10 seconds and 72°C for 15 seconds. The quantifications were performed in triplicate on a pool of 50 embryos for each line and results represent the mean value±s.e.m.
A new hairy/E(spl) gene, him, is physically linked to her5 in a head-to-head orientation
We searched for genes physically linked to her5 by sequencing a her5-containing PAC (Tallafuss and Bally-Cuif, 2003). This revealed the presence of an open reading frame encoding a new Hairy/E(spl) factor located 3.3 kb upstream of her5 and in a head-to-head orientation(Fig. 1A). RT-PCR experiments demonstrated the presence of the corresponding transcript in embryos (not shown), and blasting against the zebrafish genome (release from 08-02-04)confirmed the association of this gene with her5. We named this new gene him (for `her5image') (GenBank accession number AY705671). The genomic structure of him was determined by aligning cDNA and genomic sequences and shows several alterations compared to the zebrafish genome prediction, notably the presence of an additional 5′ coding exon (true exon 1) as well as a different position and size of the last exon (exon 4), coding for 195 amino acids instead of the 22 predicted (Fig. 1A).
The head-to-head association of the gene pair him-her5 is reminiscent of the genomic organization of her7-her1(Henry et al., 2002). Further analysis of the zebrafish genome suggests at least another additional gene pair, her4 and a new predicted her gene that we named her12 (ENSDARG00000028110) (Fig. 1B). Therefore, we propose that a paired organization might be a conserved feature among zebrafish her genes.
The Him protein, translated from our full-length cDNA sequence, consists of 297 amino acids and exhibits all structural features of an Hairy/E(Spl) bHLH factor acting as transcriptional repressor(Davis and Turner, 2001): a conserved proline residue in the basic domain, an `orange' domain(Dawson et al., 1995) and a WRPW tetrapeptide in the C-terminus(Fisher et al., 1996)(Fig. 1C). Within the zebrafish Her family, Him shows the highest similarity to Her1 with 28.6% identical and 35.2% conserved amino acid (aa) residues. Similarity between Him and Her5 is slightly weaker (20.1% identical and 29.2% conserved aa residues)(Fig. 1B, red lines, and Fig. 1C). Comparison restricted to the functional bHLH domain reveals 66% identity to Her1 and 50% identity to Her5.
him expression within the presumptive midbrain-hindbrain is identical to her5 and marks the intervening zone
We analyzed him expression by RT-PCR and in situ hybridization. him, like her5, is maternally expressed(Fig. 2A). Early zygotic him expression is ubiquitous (data not shown) but rapidly resolves in a first, transient, profile within the presumptive dorsal endoderm and mesoderm (Fig. 2B) at 30%epiboly: him is expressed in deep scattered cells of the dorsal embryonic margin and in the deep layer of the dorsal mesendoderm(Fig. 2B and 2B′, red arrows). From mid-gastrulation onwards (70% epiboly), him expression in the presumptive endo- and mesoderm becomes undetectable(Fig. 2C). At that stage, him becomes transcribed in the anterior neural plate, in a V-shaped domain interrupted at the midline (Fig. 2C, red arrowhead). Expression in the lateral aspects of this domain is slightly broader and stronger than medially. At the three-somite stage, this expression fuses medially and, by anatomical landmarks, is clearly located within the presumptive MH domain(Fig. 2E,F). himexpression is maintained at the MHB later on until 36 hpf(Fig. 2H,I). Starting at late gastrulation, him is also expressed in the presomitic mesoderm(Fig. 2D-F, blue arrows). Expression in this territory is detectable until late somitogenesis(Fig. 2H).
To determine whether and to what extent him and her5share expressing cells within the presumptive MH, we compared their expression profiles by double ISH. At three somites(Fig. 2F), observation at high resolution of double-stained embryos showed that the MH expression of him and her5 are exactly identical. Thus, himexpression, like her5, precisely delineates the MH primordium(Tallafuss and Bally-Cuif,2003). Like her5(Geling et al., 2003), it is also a permanent marker of the IZ, separating the early midbrain proneural cluster vcc from rhombomere 2 presumptive motor- and sensory neurons at late gastrulation (Fig. 2J). him expression is detected in this domain slightly earlier than her5 (not shown).
Taken together, our data suggest a more complex picture than expected from the genomic organization of him and her5. him and her5 share expression within the MH domain, but differ elsewhere in her5-specific versus him-specific domains (pharyngeal precursors versus presomitic mesoderm at late gastrulation, respectively). In the neural plate however, Him is, together with Her5, the earliest marker of the MH and IZ, prompting us to analyze its function in this domain in more detail, in relation to MH patterning and Her5 activity.
Like Her5, and in contrast to most MH factors, Him does not control patterning events within the MH region
Refined regionalization and maintenance of the MH domain at somitogenesis stages depends on a positive cross-regulatory loop involving Fgf8 and Pax2.1(Brand et al., 1996; Lun and Brand, 1998; Reifers et al., 1998; Tallafuss and Bally-Cuif,2003). To determine whether him was part of this loop, we analyzed its expression in ace/fgf8 and noi/pax2.1 mutants(Brand et al., 1996; Lun and Brand, 1998; Reifers et al., 1998). Expression of him in ace mutant embryos is initiated normally (data not shown) but, from mid-somitogenesis stages onwards,gradually narrows to persist at the MHB only in a dorsal patch(Fig. 3A,B). At 24 hpf him expression in ace is undetectable (data not shown). Similarly, in noi mutants, a downregulation of himexpression can be observed from mid-segmentation stages onwards. In contrast to ace, him expression in noi later remains restricted to the ventral MHB (Fig. 3C,D),like her5 (Lun and Brand,1998; Reifers et al.,1998). Together, these observations demonstrate that the maintenance of him expression is, like that of other MH genes and within a similar time-window, under control of the MH regulatory loop.
Because Him is one of the earliest selective markers of the MH primordium,we asked in turn whether Him activity was involved in controlling aspects of MH regionalization. Loss of Him function, performed by injection of an antisense GripNA oligonucleotide specific for the ATG region of him (himGripNAATG) into one-celled embryos (see below for results demonstrating the functionality and specificity of this GripNA), however affected neither the expression of MH patterning markers (e.g. pax2.1, eng2, eng3), nor that of IsO activity markers(e.g. wnt1, fgf8) at the 3- and 15-somite stages (data not shown).
Thus, although him expression depends on the MH maintenance loop,Him activity itself does not appear to impinge on this loop to influence MH patterning, like that of Her5.
Him activity is crucial for the formation of the medial IZ at early neurogenesis stages
We demonstrated previously that Her5 is crucially necessary to prevent neurogenesis across the MIZ (Geling et al., 2003). Because of the similar expression of him, and its encoding a related Hairy/E(Spl) factor, we explored a potential involvement of Him in IZ formation. At early neurogenesis stages, the IZ is also the major site of pax2.1 expression, which we used as landmark in future experiments (see Fig. 4D).
We tested a requirement for Him in IZ formation in loss-of-function experiments where him mRNA translation was blocked by specific antisense GripNAs. Two GripNAs were used: himGripNAATG (as above), and a second GripNA targeting the acceptor splice site of him exon2(himGripNASPL). Undistinguishable phenotypes (see below)were obtained with the two GripNAs, but not with a control GripNA of unrelated sequence, indicating that the phenotypes observed are a specific consequence of Him dysfunction. Specifically, himGripNAs-injected embryos assayed at the three-somite stage for ngn1 expression displayed a complete lack of the MIZ, with bridging of the vcc and r2MN clusters by ectopic ngn1-expressing cells (76% of cases, n=21 for himGripNAATG, 83% of cases, n=24 for himGripNASPL) (Fig. 4B,C, compare to 4D). This phenotype is followed by the development of ectopic differentiated neurons across the basal MHB at later stages (data not shown). It is in all respects similar to that triggered by loss of Her5 function in her5MOATG morphants(Fig. 4A)(Geling et al., 2003). Thus,loss-of-function of either Her5 or Him results in the same failure to form and maintain the MIZ.
Him and Her5 are independently required for medial IZ formation
The above results are compatible with a simple model where Him and/or Her5 would act in a common regulatory cascade, one factor positively regulating expression of the other gene. Thus, loss of Him function would cause loss of her5 expression, or the reverse. Alternatively, Him and Her5 might be independently necessary for MIZ formation. To address this question, we studied him and her5 expression in embryos where Her5 or Him activity, respectively, was blocked. We observed that him expression was unchanged in Her5 morphants, in both the MH region and the presomitic mesoderm, under conditions where ngn1 expression was strongly induced in place of the MIZ (Fig. 4E,F,and data not shown). Thus him expression is not under immediate control of Her5. Likewise, injection of GripNAATG into wild-type embryos did not produce alterations in her5 transcription,although the MIZ was lost (Fig. 4I compared to 4K, and data not shown). Furthermore, injection of GripNAATG did not affect the production of the fusion Her5-GFP protein, driven under control of all her5 regulatory elements in her5PAC::egfp transgenics(Tallafuss and Bally-Cuif,2003) (Fig. 4J,L). Thus, Him does not influence her5 transcription or translation.
We conclude that Him and Her5 do not act in a simple cascade of cross-regulation of expression. Rather, the two genes are expressed independently of each other and are both essential to MIZ formation.
The crucial determinant of MIZ formation is the total dose of Him +Her5 inhibitory activities
Several hypotheses could account for the above results. First, Him and Her5 might both be required for MIZ formation because they need to heterodimerize with each other to be active. Alternatively, these factors do not have unique essential activities, but rather are required to reach together a threshold level of Hairy/E(spl) activity necessary to prevent proneural gene expression. Finally, both factors might exert distinct and/or complementary functions necessary for MIZ formation.
To unravel the relevance of each hypothesis in vivo, we first tested whether Him and Her5 could interact in a yeast two-hybrid system. bHLH factors have the capacity to dimerize via their HLH domain, however their affinity for hetero- versus homodimerization cannot be a priori predicted, and some instances of DNA binding as oligomers have also been reported(Firulli et al., 2000; Iso et al., 2001; Wainwright and Ish-Horowicz,1992). We observed that Her5 can homodimerize as well as bind Ngn1, while Him and Ngn1 failed to interact. In addition, heterodimers of Him and Her5 were produced. Whether Him is also able to homodimerize could not be tested due to unexplained toxicity of the him-expressing constructs(Table 2). All these interactions were confirmed by coimmunoprecipitation(Fig. 5A). Moreover, the affinity for heterodimerization between Him and Her5, based on beta-galactosidase activity (Lazo et al.,1978), appeared six-fold higher than the affinity of Her5 for homodimerization (Fig. 5B),suggesting that the Him-Her5 configuration predominates in vivo if the amount of proteins is equal. Thus, the requirement for Him and Her5 for MIZ formation in vivo might indeed be explained by the necessity for these factors to heterodimerize.
Our finding that Her5 can homodimerize, however, as well as reports of functional bHLH oligomers (Firulli et al.,2000; Iso et al.,2001; Wainwright and Ish-Horowicz, 1992), suggest that Her5 and/or Him alone could permit MIZ formation, provided the dose of this factor is sufficient. To test this hypothesis we studied neurogenesis in her5PAC::egfp transgenic embryos (Tallafuss and Bally-Cuif,2003), where Him is in excess. This transgenic line was obtained by recombination of egfp into the second exon of her5 in a her5-containing PAC. Thus, in this transgene her5 is not functional. However, because of the small genetic distance separating the her5 and him locus, and because the entire recombinant PAC was used for germ line transformation(Tallafuss and Bally-Cuif,2003), these transgenics carry an additional copy of himtogether with the her5-egfp fusion (see Fig. 5C for a scheme of the transgene). Real-time PCR confirmed a 1.5-fold increase in the amount of him messenger in embryos heterozygous for the her5PAC::egfptransgene, while the amount of otx1 mRNA, used as a control, was unchanged (Fig. 5D). Because him regulatory elements are also contained within the recombined PAC,additional him transcripts produced from the transgene display the endogenous him profile, restricted to the IZ within the neural plate(not shown). Therefore, her5PAC::egfp transgenics provide an ideal background to measure MIZ formation in a context where her5expression is normal but the dose of Him is increased across the IZ. Most interestingly, we observed that her5PAC::egfp transgenic embryos where Her5 activity was blocked by her5MOATG formed a normal MIZ (Fig. 5E, compare to Fig. 4D). In all cases, we verified that Her5 activity was completely abolished, by monitoring the lack of EGFP protein expression (data not shown). Thus Him, when present in sufficient amount (in a minimum of three doses, as in her5PAC::egfp/+heterozygote embryos), is capable of replacing Her5 activity to prevent neurogenesis across the MIZ.
We conclude from these observations that the crucial component of MIZ formation and maintenance is a threshold level of `Him + Her5' inhibitory activity. In the normal embryo, this level is probably achieved by Her5-Him heterodimers, although a possible contribution of homodimers and/or oligomers from each factor separately cannot be excluded.
Formation of the lateral IZ also relies on the level of `Him + Her5 activity' but with a lower threshold than the medial IZ
The LIZ is preserved in both her5 and him single knockdown embryos, suggesting that it might require other factors than Him and Her5 for its formation. Alternatively, the LIZ might primarily differ from the MIZ in requiring a lower threshold of `Him + Her5' activity, the endogenous level of one factor alone (two doses) being sufficient to block neurogenesis in this location. To address these hypotheses we assayed for lateral ngn1 expression in double knockdown embryos obtained by the co-injection of her5MOATG and himGripNAATG. Strikingly, we observed that the simultaneous interference with both Her5 and Him activities results in ectopic ngn1 expression in place of the entire IZ, i.e. including the LIZ(88% of cases, n=25) (Fig. 6A,A′ compared to 6D,D′), in striking contrast to single knockdowns (0% of cases for her5 knockdowns, n=21, 0% of cases for him knockdowns, n=24)(Fig. 6B,C).
To confirm these results we analyzed b404 mutants(Topczewski et al., 2001),which we found to carry a deletion encompassing the her5 and him locus in addition to knypek (kny). As expected,expression of both him and her5 in these mutants is completely absent at all stages (Fig. 7A-D). Mutations in zebrafish kny impair gastrulation movements of convergence and extension that normally narrow the embryonic body and elongate it from head to tail, resulting in shorter and broader embryos(see Fig. 7B,D)(Henry et al., 2000; Topczewski et al., 2001). In addition to this phenotype, assaying b404 mutants for neurogenesis revealed ectopic ngn1-positive cells in place of both the MIZ(Fig. 7F, asterisk) and LIZ(Fig. 7F, arrow, enlarged in F′), a phenotype never observed in control siblings(Fig. 7E, enlarged in E′)or in embryos for a null point mutation in the knypek gene(knym119 allele, not shown)(Solnica-Krezel et al., 1996). The number of ngn1-positive cells in the lateral aspect of the MH domain (white box in Fig. 7E-G)is increased by 40% in b404 embryos compared to control siblings(n=13 mutants and 10 wild-type embryos), while other populations of neuronal precursors, such as trigeminal ganglia neurons, are unaffected (black box in Fig. 7E-G), further supporting the specificity of this phenotype. Further, this phenotype was maintained upon rescuing Kny function in b404 homozygous embryos injected with kny mRNA at the one-cell stage(Fig. 7H,H′).
The above results demonstrate that ectopic neurogenesis in b404 is selective of the IZ and unrelated to the lack of Kny activity itself. It remains possible however that the b404 deletion encompasses other genes than him and her5 that contribute to this phenotype. To address this issue, we rescued Him activity in b404 homozygotes(Fig. 7J) by crossing them into the transgenic her5PAC::egfp background. b404/b404;her5PAC::egfp embryos display a normal expression of him in time and space (Fig. 7J), but no expression of her5 (not shown). We observed that, under these conditions, the LIZ was preserved(Fig. 7I, white box, enlarged in I′). Thus, restoring Him activity is sufficient to rescue formation of the LIZ in b404 mutants, strongly arguing that the neurogenic phenotype in the lateral MH of these mutants results from the lack of Her5 and Him function. In addition, these results demonstrate that one dose of Him activity provides a level of inhibition sufficient for LIZ formation. In contrast, loss of the MIZ was maintained in b404/b404;her5PAC::egfpembryos (Fig. 7I), in keeping with our finding that MIZ formation requires more than one or two copies of him or her5 (Fig. 4A-C, 5E).
Him and Her5 are equally potent neurogenesis inhibitors in the lateral IZ
We demonstrated above that the total amount of Him + Her5 inhibitory activity is the crucial determinant for IZ formation, and that increased levels of Him could compensate for loss of Her5 both in the medial and lateral IZ. These experiments however did not address the relative contribution of Her5 to the total inhibitory activity required for IZ formation. To determine whether Him and Her5 contribute equally to this activity, we analyzed LIZ formation in b404/+; knym119/+ embryos where the function of either Her5 (Fig. 7K) or Him (Fig. 7L) was abolished. This background, obtained by crossing b404/+ with knym119/+ heterozygote adults, allows immediate identification of the embryos carrying one single copy of each gene him and her5, since such embryos display the knypekphenotype. Blocking the activity of Him or Her5 b404/+; knym119/+ embryos leaves only one functional copy of either her5 or him, respectively. Assaying for ngn1expression revealed that the LIZ forms normally in such embryos (Fig. 7K,K′ and 7L,L′ compared to 7M,M′, white arrowhead)(while, as expected, the MIZ is lost, Fig. 7K, and 7L, compared to 7M, white arrow). These results are in agreement with an equal potency of Him and Her5 to inhibit lateral neurogenesis at the MHB and, like for the MIZ, we propose that a crucial determinant of LIZ formation is the threshold of Him + Her5 activities rather than the specific presence of both factors.
Maintaining a progenitor pool at the embryonic MHB is crucial to MH growth and IsO maintenance, and we report here a new molecular player and its associated mechanism preventing neurogenesis in this territory. Looking for genes physically linked to her5, we unraveled a previously unknown,paired and divergently transcribed hairy/E(spl) gene, him,which shares with her5 expression across the IZ. We demonstrate that blocking either Her5 or Him function results in the same failure to form and maintain the MIZ, and that interfering concomitantly with the function of both factors prevents formation of the LIZ. In both domains, we demonstrate that sufficient levels of one factor alone are sufficient to compensate for the lack of the other. Together, our results are most compatible with a model where the molecular basis of IZ formation is the total Him + Her5 inhibitory activity, at different thresholds along the mediolateral axis. They highlight a new mechanism, relying on paired Hairy/E(spl) factors, for the maintenance of a non-differentiating signaling boundary during embryonic development.
him and her5 are a new co-functional gene pair
him and her5 are located 3 kb apart in a head-to-head, in a manner reminiscent of the her7/her1 gene pair(Henry et al., 2002). Our search through the zebrafish genome revealed an additional similarly organized pair of her genes, her4/her12, located on chromosome fragment ctg10516. ESTs BM023698 and AL716753 match with 100% to the cDNA sequence deducted from ENSDARG00000028110, suggesting that her12 is a real transcribed gene. Thus, our results suggest that a paired and divergently transcribed configuration is a frequent organization of zebrafish her genes. Our search through other vertebrate genomes,however, including mouse and Fugu rubripes (Fugu genome, The Welcome Trust Sanger Institute, release 08-02-04), failed to reveal a similar organization of hairy/E(spl)-like genes in these species, suggesting that the molecular process(es) generating her pairs took place along the lineage leading to zebrafish.
Our results do not suggest a simple evolutionary model leading to the generation of zebrafish her pairs. Indeed, the six her genes involved belong in sequence to two groups of orthologs (group 1: him/her5/her1 versus group 2: her7/her4/her12), but only the her7/her1 pair contains one gene from each group. Thus the situation is not comparable to dlx gene pairs, interpreted to result from a tandem duplication followed by a cluster duplication. Because her4and her12 have very similar coding sequences, it is possible that these two genes underwent a recent event of gene conversion, facilitated by the formation of intrachromosomal hairpins(Hickey et al., 1991). We failed to detect indications supporting gene conversion within the him/her5 gene pair, but other recombination events might have occurred (D. Chourrout, J.N. and L.B.-C., unpublished observations).
Linked genes sharing sequence similarity have been reported for a variety of genetic functions in several organisms.(Akam, 1989; Alonso and Cabrera, 1988; Bober et al., 1994; Coleman et al., 1987; Kmita and Duboule, 2003; Knust et al., 1992; Stein et al., 1996). Duplication events resulting in linked arrays of related genes generate copies that often share cis-acting regulatory sequences. Whether him and her5 expression across the IZ are coregulated remains to be directly demonstrated but is highly likely, given that the enhancer driving MH expression of her5 extends into the him locus(Tallafuss and Bally-Cuif,2003). In addition, him and her5 differ in some aspects of their expression profiles (in the shield and presomitic mesoderm,versus presumptive pharyngeal endoderm, respectively). The regulatory elements controlling endodermal expression of her5 are located closer to the her5 ATG than the MH expression elements(Tallafuss and Bally-Cuif,2003). Thus expression of the him/her5 pair may be controlled by a combination of proximal and gene-specific elements (accounting for the differential expression sites of the two genes) and distal and probably common elements (driving IZ expression). It is possible that the proximal elements are new modifications in the evolution of the gene pair,extending genetic functions by the acquisition of new expression domains(Ohta, 2000). It will be interesting to determine whether such cis-regulatory organization is involved in generating different expression sites within other gene pairs.
The combined activities of Her5 and Him determine LIZ formation
We previously identified Her5 as the first determinant of MIZ formation in zebrafish (Geling et al.,2003). However, although her5 expression covers the whole IZ, and ectopic her5 expression can inhibit ngn1 in the lateral MH area, we failed to implicate Her5 alone in LIZ formation in vivo. A main advance of our present work is to provide an interpretation for this finding, by identifying a new Hairy/E(spl) factor, Him, as the partner for Her5 in LIZ formation. Our arguments rely on the phenotype of embryos where the functions of Her5 and Him are concomitantly blocked in a non-genetic interference approach, and of embryos carrying the b404 deletion,where both him and her5 genes are absent. In both cases,ectopic ngn1-positive cells replace the LIZ. This phenotype is not found by blocking the function of either Him or Her5 alone, and is rescued by selectively reintroducing endogenous levels and profile of Him function into the b404 background, arguing for its specificity. Further, we show that one copy of either him or her5 (as in b404/+;knym119/+ heterozygote embryos where Her5 or Him function is blocked) similarly preserves the LIZ, demonstrating that Him and Her5 are equally potent at inhibiting ngn1 expression in that location. Thus, our results identify Him and Her5 as truly redundant factors that play equivalent roles, and are together the only determinant, in LIZ formation.
The molecular cascade downstream of Him remains unknown. Because Him is sufficient for LIZ formation in the absence of Her5, we can exclude a mechanism where Him would primarily promote Her5 activity. Rather, because of the similar sequences of Him and Her5, it is more likely that both factors act together on common targets controlling neurogenesis. Within the MIZ, Her5 and Him (J.N. and L.B.C., unpublished) act upstream of Notch to inhibit expression of ngn1 and coe2, but not other early MH proneural genes such as asha, ashb and ato3(Geling et al., 2003). It will be important to determine whether the molecular cascade(s) and mechanisms downstream of Her5 and Him are conserved in the MIZ and LIZ. Compared to the MIZ, the LIZ exhibits an additional block, still molecularly unknown, that prevents neurogenesis downstream of Ngn1 activity(Geling et al., 2003). Whether Him and Her5 also take part in this second block remains to be tested.
A unified model for IZ formation along the entire mediolateral extent of the neural plate
The absence of Her5 leads to disappearance of the entire MIZ and its replacement by ngn1-expressing cells, which later differentiate into Hu-, HNK1- and acetylated-tubulin-positive neurons(Geling et al., 2003). Surprisingly, our results now demonstrate that Him plays an equally important role in MIZ formation, since an exactly identical phenotype is triggered by lack of Him activity (this paper, and data not shown). We further rule out an interdependent regulation of him and her5 expression(Fig. 4E-L). Thus, another important implication of our work is that MIZ formation relies on prepatterning by both Him and Her5.
A priori, the finding that loss of Him or Her5 function result in identical phenotypes can have three different molecular interpretations: first, Him and Her5 might act in distinct pathways that converge on and are both necessary for neurogenesis control at the MIZ; second, the activities of Him and Her5 might be interdependent; third, Him and Her5 might have equivalent functions,a minimal dose of `Him + Her5' activity being required for MIZ formation. The first two mechanisms are unlikely, given the observation that increased levels of Him alone to three doses (as in her5PAC::egfp/+ heterozygote transgenic embryos injected with her5MO, Fig. 5E) can compensate for the lack of Her5 function within the MIZ. We do not have genetic means of assessing whether a high dose of Her5 alone would also suffice for MIZ formation. However, our findings that Him and Her5 are equally potent to prevent lateral neurogenesis strongly suggest that this is the case. Thus, we propose that the crucial determinant of MIZ formation, is a total level of`Him + Her5' inhibitory activity. Hence, above a threshold of Him + Her5, ngn1 expression is prevented medially and the MIZ is formed, while ngn1 expression is induced below this threshold(Fig. 8). As discussed above,our results indicate that three doses of one factor alone is the minimum level of inhibitory activity required for MIZ formation. Interestingly, however, two doses are sufficient when both Him and Her5 are present, as in b404/+heterozygote embryos. This result might be related to the higher propensity of Him and Her5 to hetero- than homodimerize, or to an increased activity of heterodimers versus homodimers or oligomers. Because the same factors Him and Her5 account for LIZ formation, and can functionally replace each other in this domain as well, a parsimonious interpretation of our findings is to implicate the same dose-dependent mechanism within the LIZ, albeit with a lower threshold level (Fig. 8). The LIZ minimal level of inhibition would be achieved with one dose of Him or Her5 alone. Together, our results thus lead to a unified model where the maintenance of a pool of progenitor cells at the MHB is orchestrated by a variable dose-dependency to the Him/Her5 pair.
Even in the absence of Him and Her5, we failed to induce ngn1expression within a small intermediate field located between the MIZ and LIZ(see Fig. 7F, and grey triangle in Fig. 8). In this domain, an additional (as yet unknown) factor might increase the total inhibitory activity and/or prevent neurogenesis in addition to Him and Her5. We failed to recover additional IZ-expressed E(spl) genes following our search through the zebrafish genome and expression studies (J.N., C.L. and L.B.-C.,unpublished). Because the intermediate field is aligned with the longitudinal domains of non-differentiation in the hindbrain and spinal cord, it is perhaps more likely that this factor is expressed along the AP axis of the neural plate, like other known neurogenesis inhibitors(Bally-Cuif and Hammerschmidt,2003).
An interesting open question remains to identify the cues controlling the differential sensitivity of the MIZ versus LIZ to Him + Her5, and their functional significance. The MIZ and LIZ differ in their proliferation rates:the MIZ exhibits more cells in M phase than the LIZ at late gastrulation,based on anti-phosphoH3 immunostaining(Geling et al., 2003). It will be crucial to investigate the possible relationship between MIZ and LIZ cell cycle properties and their response to Him + Her5. Also, several morphogens acting in this region are expressed following a mediolateral gradient. For instance, wnt1 is expressed in a spatio-temporal pattern similar to her5 and him at late gastrulation, thus with initially higher levels laterally than medially, and might enhance cell sensitivity to neurogenesis inhibitors. This might be related to the delay of dorsal differentiation proposed to result from the gradient of Wnt signaling from the spinal cord roof plate (Megason and McMahon, 2002). Conversely Shh signaling from the ventral midline and specifically active at the MHB (Carl and Wittbrodt, 1999; Koster et al., 1997) could increase `neurogenic competence'. These hypotheses will be important to test experimentally to gain insight into the prepatterning of IZ formation.
Biological significance of a redundant process for IZ formation
Redundant factors are generally viewed as `safety' locks, and the biological significance of the Him/Her5 couple might be to secure IZ formation. This case of redundancy is more extreme than observed for Her1/Her7, where the disruption of each gene alone produced distinct (although moderate) somitic defects, indicating partially different activities(Henry et al., 2002). The embryonic MHB progenitor pool serves several vital functions. It generates the large majority of MH neurons and glia, as demonstrated in lineage tracing experiments (Tallafuss and Bally-Cuif,2003) and genetic or surgical ablation(Cowan and Finger, 1982; Hirata et al., 2001). MH neurons form crucial integration centers involved in visual, auditory and motor control and social behavior. By its long-lasting proliferative activity,the IZ also permits the expansion of MH tissue over time. Although the relative importance of MH derivatives varies between species, most vertebrates are characterized by highly developed visual, auditory or locomotor functions,which are paired with enlarged mesencephalic derivatives or cerebellum. Finally, and importantly, the IZ coincides in space with the isthmic organizer, necessary for patterning the entire MH domain and for the subdivision of mid-versus hindbrain structures(Bally-Cuif et al., 2000; Liu and Joyner, 2001; Rhinn and Brand, 2001). In the mouse, IZ formation also relies on the two redundant bHLH factors Hes1 and Hes3 (Hirata et al., 2001). In that case however Hes1 and Hes3 are not genetically linked and their expression profiles are clearly distinct, overlapping only at the MHB (Allen and Lobe, 1999; Lobe, 1997), suggesting that mouse and zebrafish have independently evolved a strategy for the redundant expression and function of Hairy/E(spl) factors at the MHB. A dose dependency and the spatial details of IZ formation in the mouse have not been explored. The fact that one dose of each factor Him and Her5 suffices to maintain the MIZ in zebrafish, while two doses of each single factor do not, probably explains the maintenance of the two genes him and her5 in zebrafish. Whether Hes1 and Hes3, or Him and Her5, exert in addition other and perhaps distinct activities at the MHB remains to be explored.
The IZ is not an isolated case of maintenance of a non-differentiation zone at embryonic signaling boundaries. Such events have been reported, e.g. at the Drosophila wing margin, along the dorsal and ventral midlines of the neural tube (Alexandre and Wassef,2003), as well as between rhombomeres(Cheng et al., 2004). Like the IZ, these boundaries are involved in the progressive building and patterning of their adjacent territories, and the maintenance of their integrity necessitates their remaining undifferentiated. This process is achieved by Notch signaling at the wing margin and inter-rhombomeric boundaries, and Shh signaling along the neural tube ventral midline, while the factors involved along the dorsal midline probably involve Wnt and BMP signaling. Our work demonstrates that a distinct molecular mechanism accounts for non-differentiation at the MHB, namely the differential response of MHB cells to the combined inhibitory activity of two twin and co-regulated Hairy/E(spl)-like factors, independently of Notch. Our findings add to the panel of identified developmental strategies used to build and maintain signaling centers.
Note added in proof
him is identical to her11, which has been recently reported for its role in zebrafish somitogenesis(Sieger et al., 2004). The gene referred to as her11 in the present manuscript(Fig. 1B, ENSDARP00000012990),named before publication by Sieger et al., is a different coding sequence and corresponds to her13 of Sieger et al. The gene referred to as her13 in the present manuscript(Fig. 1B, ENSDARP0000008307) is a new gene, not reported by Sieger et al. We suggest that the latter gene be renamed her16 and that the nomenclature of Sieger et al. be used for all other genes in future work.
We are grateful to U. Strähle, M. Wassef and W. Wurst for their critical reading of the manuscript, and to K. Imai, M. Wahl and L.B.C. lab members for discussions and suggestions throughout this work. We thank A. Folchert and B. Tannhäuser for expert technical assistance, and the GSF fish facility staff for fish care. We acknowledge M. Brand (pax2.1),J. A. Campos-Ortega (her5) and U. Strähle (ngn1) for gifts of probes and constructs, and D. Meyer for comparison of himsequences prior to publication. Work in L.B.-C. laboratory is supported by a VolkswagenStiftung `junior research group' grant.