Advertisement
Research Article

MiR-10 Represses HoxB1a and HoxB3a in Zebrafish

  • Joost M. Woltering mail,

    To whom correspondence should be addressed. E-mail: jmwoltering@hotmail.com

    Affiliation: Institute of Biology, Leiden University, Leiden, The Netherlands

    X
  • Antony J. Durston

    Affiliation: Institute of Biology, Leiden University, Leiden, The Netherlands

    X
  • Published: January 02, 2008
  • DOI: 10.1371/journal.pone.0001396

Abstract

Background

The Hox genes are involved in patterning the anterior-posterior axis. In addition to the protein coding Hox genes, the miR-10, miR-196 and miR-615 families of microRNA genes are conserved within the vertebrate Hox clusters. The members of the miR-10 family are located at positions associated with Hox-4 paralogues. No function is yet known for this microRNA family but the genomic positions of its members suggest a role in anterior-posterior patterning.

Methodology/Principal Findings

Using sensor constructs, overexpression and morpholino knockdown, we show in Zebrafish that miR-10 targets HoxB1a and HoxB3a and synergizes with HoxB4 in the repression of these target genes. Overexpression of miR-10 also induces specific phenotypes related to the loss of function of these targets. HoxB1a and HoxB3a have a dominant hindbrain expression domain anterior to that of miR-10 but overlap in a weaker expression domain in the spinal cord. In this latter domain, miR-10 knockdown results in upregulation of the target genes. In the case of a HoxB3a splice variant that includes miR-10c within its primary transcript, we show that the microRNA acts in an autoregulatory fashion.

Conclusions/Significance

We find that miR-10 acts to repress HoxB1a and HoxB3a within the spinal cord and show that this repression works cooperatively with HoxB4. As with the previously described interactions between miR-196 and HoxA7 and Hox-8 paralogues, the target genes are located in close proximity to the microRNA. We present a model in which we postulate a link between the clustering of Hox genes and post-transcriptional gene regulation. We speculate that the high density of transcription units and enhancers within the Hox clusters places constraints on the precision of the transcriptional control that can be achieved within these clusters and requires the involvement of post-transcriptional gene silencing to define functional domains of genes appropriately.

Introduction

Hox genes participate in regionalizing the anterior-posterior body axis in metazoan animals. In mammals, this gene family comprises 39 closely related genes for homeodomain transcription factors, organized in 4 homologous clusters (A, B, C, D) [1][4]. The genes are expressed along the body axis in a sequence that corresponds to their genomic sequence within the Hox clusters. The more 3′ a gene is located in a Hox cluster the more anterior is its expression domain. This feature is commonly referred to as ‘spatial colinearity’. The Hox genes have sharply defined anterior expression boundaries but their posterior boundaries are typically less clear and overlap with the expression of more posterior Hox genes.

The 4 mammalian Hox clusters arose from an ancestral Hox cluster via two genome duplications. In Teleost fish, an additional genome duplication generated 8 Hox clusters (named Aa, Ab, etc.), there being 7 clusters containing 49 genes in Zebrafish [5], [6], [7] due to loss of one cluster during evolution.

In addition to the Hox coding genes, the miR-10, miR-196 and miR-615 microRNA gene families have been identified within the vertebrate Hox clusters [8][11]. MicroRNAs are small (~22 nt) non-coding RNAs which are derived from stemloop forming precursor transcripts, through processing by the RNAse III enzymes Dicer [12] and Drosha [13]. MicroRNAs function in post-transcriptional gene silencing by binding to imperfect target sites in messengerRNAs. They thereby induce translational inhibition and RNA destabilization [14], [15].

In the Hox clusters, miR-10 genes are closely associated with the positions of Hox-4 paralogue members, miR-196 is located 5′ of Hox-9 paralogues and the more recently cloned miR-615 is located in the HoxC5 intron in mammals but appears to be absent from Teleosts and Xenopus tropicalis. The latter microRNA could therefore be restricted either to mammals or to amniotes. In the Zebrafish genome, miR-10 is present in 5 paralogues representing 4 different isoforms (a, b, c and d), which differ from each other at 1 to 3 positions [16], [17]. We previously showed that the genomic location of the miR-10d microRNA corresponds to the degenerated HoxDb cluster [16].

Mouse knockouts and a Zebrafish germline Dicer mutant have revealed important functions for microRNAs in the coordination of normal embryonic development [18], [19], but the individual vertebrate microRNAs are in general still enigmatic genetic objects and only a few have been characterized at a functional level. In Zebrafish, miR-430 [20] has been shown to silence maternal RNAs, miR-214 is involved in proper somite specification [21] and miR-375 is necessary for the maintenance of embryonic pancreas integrity [22].

With respect to the Hox related microRNAs, HoxA7 and Hox-8 paralogues have been identified as targets of miR-196 [23], [24], [25]. In chicken the interaction with HoxB8 has been implicated in the mechanism that abolishes the competence of posterior lateral plate mesoderm for limb induction by retinoic acid [25]. In Drosophila, a conserved or possibly convergent interaction exists for the miR-196 homologue IAB-4 which targets the Ubx Hox gene [26]. Until now, the function of the miR-10 microRNA family has remained unclear but, based on its evolutionary conservation within the anterior part of the Hox clusters, an associated role in anterior-posterior patterning seems likely.

The anterior Hox genes are strongly expressed in the central nervous system and play an important role in patterning the hindbrain, spinal cord and branchial arches [27], [28], [29]. In the hindbrain, the expression of these Hox genes follows the rhombomeric boundaries. HoxB1a and HoxB1b are expressed in rhombomere (r) 4, HoxB2a defies the rule of colinearity and is expressed more anteriorly, in r 3 and r 4, HoxB3a and HoxA3a are expressed most strongly in r 5 and 6 with a weaker domain extending more posteriorly in the spinal cord. The Hox-4 paralogues are expressed from r 7 onwards and throughout the spinal cord (reviewed 30). The pattern of Hox expression in the hindbrain contributes to the formation of localized neuronal structures like the rhombomere 4 specific Mauthner neurons and the distinct patterns of cranial motor nerves in different regions of the hindbrain.

Here, we address the function of miR-10 with relation to a possible role in anterior-posterior patterning. We show that miR-10 represses the nearby HoxB1a and HoxB3a genes and that its overexpression also induces the associated loss of function phenotypes for both genes. MiR-10 morphant embryos show upregulation of these target genes within the normal miR-10 expression domain, indicating that active repression occurs in the embryo. In overexpression experiments, miR-10 synergizes with HoxB4 in the repression of these target genes. In the case of a long range HoxB3a transcript that includes miR-10c within its primary transcript, we show that the microRNA acts in an essentially auto regulatory fashion. In addition, we present a model in which we explain the need for post-transcriptional regulatory interactions within the Hox clusters on basis of the high density of enhancers and transcription units within the clusters.

Results

MiR-10 is expressed in a Hox-4 like pattern

The MiR-10 paralogues are associated with the 5′ genomic region of Hox-4 genes and microRNA specific Locked Nucleic Acid (LNA) in situ hybridization [16], [31] and transgenic sensor lines [24] have revealed similar patterns of expression as for the Hox-4 paralogues. RT-PCR with primers located 5′ of miR-10c and in the HoxB4a coding sequence shows that miR-10c and HoxB4a are located on the same primary transcript (figure 1A) and RT-PCR for the individual genes shows the same temporal expression pattern (figure 1B).

thumbnail

Figure 1. Spatial and temporal expression profile of miR-10c.

A) RT-PCR with primers located 5′ of miR-10c and within the coding region of exon 1 of HoxB4a shows inclusion of miR-10c and HoxB4a on the same transcript. PCR 35 cycles, -RT: no reverse transcriptase added. B) RT-PCR shows similar temporal expression during development of HoxB4a (28 cycles) and miR-10c pre-miRNA (35 cycles). C) Whole mount in situ hybridization on different stage Zebrafish embryos shows mutually exclusive expression of the HoxB3a rhombomere 5/6 domain (red) with miR-10c (purple).

doi:10.1371/journal.pone.0001396.g001

The exact anterior boundary of the neural expression of miR-10c was determined in double in situ hybridization together with the anterior neighboring gene HoxB3a (figure 1C). Consistent with the transcriptionally implied co-regulation, miR-10c has the same anterior boundary of expression as described for HoxB4a and is expressed in a mutually exclusive domain with the anterior strong r 5/6 expression domain of HoxB3a (single in situ, figure S1A).

Under some circumstances, LNA probes are known to exhibit single nucleotide resolution [16], [32]. In situ hybridization with probes matching each of the miR-10 isoforms (figure S2) excludes that other miR-10 isoforms, which are possibly not detected by the miR-10c LNA probe, are expressed in domains overlapping with or anterior to the main expression domain of HoxB3a. The expression patterns of the miR-10 isoforms differ in that the probes for miR-10b and miR-10d show a more posterior rostral boundary, with highest intensity staining caudal to the hindbrain, while miR-10a and miR-10c probes have an anterior boundary at r6/7.

MiR-10 target sites are present in HoxB1a and HoxB3a

MicroRNAs bind through a complementary fuzzy match to target sites in messengerRNAs. The specificity of this interaction resides in the sequence of nucleotides 2-7 of the microRNA, called ‘the seed’, which does not differ between the different isoforms of a microRNA family. This sequence forms the minimal requirement for a target site and is usually flanked at position 1 by an adenosine or a perfect match [33]. Accordingly, sequence information permits the prediction of target genes. In silico target analysis in Teleosts has predicted the presence of miR-10 target sites within the Hox clusters [34]. Inspection of the Zebrafish HoxBa cluster with the miR-10 seed sequence (nucleotide1-7) indicates the presence of putative target sites primarily in the 3′ part of the cluster. Candidate target sequences are associated with two Hox transcripts: 2 sites are located in the 3′ UTR of HoxB1a, 2 sites in the 3′ UTR of HoxB3a and 3 sites are present in the HoxB3a open reading frame (figure 2A).

thumbnail

Figure 2. miR-10 target sites within the HoxBa cluster.

A) Schematic representation of the zebrafish HoxBa cluster with MiR-10 seed sequences (nucleotide 1-7) within the sense strand indicated as orange bars. Known and EST database inferred mature Hox transcripts are indicated in blue. The miR-10c microRNA gene is indicated in green. B) Schematic representation of the HoxB1a and HoxB3a E-YFP sensor constructs and the HoxB3a overexpression construct. Red boxes indicate the position of the seed sequences. The light red box in the HoxB1a 3′ UTR is a target site flanked at position one by a T instead of an A. C) Validation of the HoxB1a and HoxB3a E-YFP sensor constructs by injection of wildtype (WT) and seed mutant (mut) constructs in presence and absence of miR-10 siRNA. E-CFP was co-injected as a loading control. D) Phenotypic sensor assay to validate the HoxB3a ORF miR-10 target sites. Overexpression of 40pg HoxB3a results in severe anterior and posterior truncations that are rescued by co-injection with miR-10 siRNA

doi:10.1371/journal.pone.0001396.g002

E-YFP sensor constructs containing the HoxB1a and HoxB3a 3′UTR target sequences were tested for their sensitivity to silencing by miR-10 (figure 2B). Constructs containing point mutations in the seed sequence were used as negative controls. Sensor construct RNA was injected with or without miR-10 siRNA. Co-injection with E-CFP RNA was used as a loading control. Embryos were analyzed for fluorescence at blastula/early gastrula stages. Both HoxB1a and HoxB3a wildtype sensor constructs are strongly repressed by the microRNA (figure 2C), while the seed point mutant construct proves insensitive to repression by miR-10. The seed mutant construct for HoxB1a in which the two target sites are mutated is still partially silenced however after co-injection with miR-10 (data not shown). Closer inspection revealed a 3rd possible target sequence corresponding to nucleotide 2-7 flanked by a T at position 1. After introduction of a point mutation into this seed sequence the construct is no longer repressed by injection of miR-10 (figure 2C).

A phenotypic sensor assay was used to validate the target sites located in the HoxB3a ORF. Overexpression of 40pg HoxB3a RNA induces a very strong phenotype with both anterior and posterior truncations of the embryo (figure 2D). These defects are completely rescued by co-injection of miR-10 siRNA, indicating absence of overexpressed HoxB3a protein. These experiments identify the predicted HoxB1a and HoxB3a 3′UTR and ORF target sites as mediators of miR-10 repression.

Response of endogenous Hox genes to miR-10 gain and loss of function

To investigate the role of miR-10 in the regulation of endogenous Hox genes, we performed miR-10 gain and loss of function experiments. For gain of function, the injection of a siRNA into the zygote is an effective way to overexpress microRNAs [35]. Loss of function can be achieved via the injection of antisense morpholinos. Processing and production of a mature miRNA are effectively blocked by a morpholino directed against a microRNA precursor [22]. As morpholinos allow mismatches with the target sequence, it is possible to target several miRNA isoforms using fewer morpholino sequences. ClustalW alignment of the 5 miR-10 paralogue precursor sequences reveals a region of extended conservation in the stem loop 5′ to the mature microRNA (figure 3A), which allows the design of two morpholino reagents with only minor overlap to control against off-target effects. Morpholino reagent 1 (MO1) consists of a mix of two morpholinos directed against the miR-10a and miR-10b mature sequences. Morpholino reagent 2 (MO2) is a single morpholino directed against the upstream conserved sequence (figure 3A). Either morpholino has maximally one nucleotide mismatch with any of the 5 miR-10 paralogues. Injection of either MO1 or MO2 leads to absence or very strong reduction of the signal for each of the 4 miR-10 isoforms in northern blots (figure 3B), showing that their processing is efficiently inhibited. For MO2 we observe a slight recovery of the signal at 48 and 72 hpf (hours post fertilization), but overall, injection results in a very strong decrease of mature miR-10 levels. In situ hybridization with miR-10 LNA probes also shows no signal in morpholino injected embryos (figure 3C).

thumbnail

Figure 3. Morpholino knockdown of miR-10.

A) ClustalW alignment of the 5 Zebrafish miR-10 precursor sequences. Indicated are the positions of the mature microRNA, the hairloop and the miR-10* (antisense pairing sequence in the hairpin). The target sequences for both morpholino reagent 1 and 2 are indicated with yellow bars (MO1 and MO2). B) Northern blot for all 4 different miR-10 isoforms in morpholino injected embryos at 24, 48 and 72 hpf. There is an absence or very strong downregulation of the mature microRNA in the morpholino injected samples. C) LNA in situ hybridization for miR-10b and miR-10c in 72 hpf miR-10 morphants. The endogenous expression of miR-10b and miR-10c is no longer detected.

doi:10.1371/journal.pone.0001396.g003

Besides interfering with translational processes, targeting by microRNAs leads to reduced transcript stability and decreases the amounts of transcript present [20], [36]. RNA levels can therefore function as read out of a transcript/microRNA interaction [20].

Embryos were injected with either miR-10 siRNA or miR-10 morpholino and then analyzed at 24 hpf for the expression of the target genes HoxB1a and HoxB3a, and of HoxB2a, HoxB4a and HoxB5a, genes that are predicted not to be targets.

Overexpression of miR-10 leads to downregulation of HoxB1a and HoxB3a in their strong anterior hindbrain expression domains but does not influence the expression of the other Hox genes (figure 4A marked miR-10 siRNA). In situ hybridization with the ‘sensor part’ of HoxB3a shows that this region responds identically to overexpression of miR-10 (figure S1B). In the morpholino injected embryos, increased HoxB1a expression is observed in the hindbrain/spinal cord transition (figure 4A). HoxB3a expression shows a similar posterior upregulation in morphant embryos, although to a lesser extent (figure 4A). The other Hox genes, which are also not affected by the overexpression of miR-10 siRNA, show no change in expression levels in morphant embryos. Relative quantitativity of the method was assessed by control double in situ hybridization using HoxB1a and HoxB4a, which shows that it is possible to visualize the different responses of these genes (figure 4B).

thumbnail

Figure 4. Effect of miR-10 knockdown and overexpression on endogenous Hox target transcripts.

A) Whole mount in situ hybridization with probes for hoxB1a, B2a, B3a, B4a and B5a on 24hpf embryos injected with morpholino reagent 1 or 2 (MO1 or MO2), miR-10 siRNA or non injected controls (NIC). To allow quantitative detection, embryos hybridized with the same probe were stained equally long and staining was continuously monitored and stopped before reaching signal saturation. HoxB1a and HoxB3a respond to both gain and loss of function (arrows) of miR-10 with a decrease and increase in expression levels respectively. HoxB2a, HoxB4a and HoxB5a are unresponsive to miR-10 overexpression or knockdown. B) Double whole mount in situ hybridization on 24 hr embryos using probes for hoxB1a and hoxB4a showing the different responses of the genes. Embryos were stained equally long till adequate staining was obtained for the hoxB4a probe. C) In situ hybridization with HoxA1a and HoxA3a on 24 hpf embryos injected with MO1, MO2 or miR-10 siRNA. HoxA1a responds to both overexpression and knockdown. The transcript level of HoxA3a does not respond to either overexpression or knockdown. D) In situ hybridization with HoxB1b on 24 hpf embryos morphant and overexpression embryos. There is strong upregulation of HoxB1b expression in the morphants but no downregulation is observed in the miR-10 siRNA injected embryos.

doi:10.1371/journal.pone.0001396.g004

Additional Hox-1 and Hox-3 paralogue members are located in the HoxAa, HoxBb, HoxCa and HoxDa clusters. In HoxA3a, one putative miR-10 target site is present in the HoxA3a 3′ UTR (777 nt downstream of the ORF), in HoxB1b one putative target site is located in the 3′ UTR (125 nt downstream of the ORF) and in HoxA1a, a seed sequence is located 5472 nt downstream of its ORF.

No seed sequences are associated with the HoxC1a, HoxC3a or HoxD3a coding regions or 3′ UTRs. We also determined the responses of HoxA3a, HoxA1a and HoxB1b to overexpression and knockdown of miR-10. Surprisingly, both HoxA1a and HoxB1b are strongly upregulated in the spinal cord in miR-10 morphant embryos (figure 4C, E), suggesting either direct de-repression by miR-10 or activation by the now de-repressed HoxB1a gene. In addition HoxA1a also responds strongly to overexpression of the miR-10 siRNA. We don't observe any changes in the expression of HoxA3a resulting from overexpression or knockdown of miR-10 (figure 4C).

HoxB1a upregulation by retinoic acid is elevated in miR-10 morphants

Anterior Hox genes are regulated in early developing central nervous system by retinoic acid (RA) [27] and possess cis-acting retinoid response elements [37]. The co-expression and implied co-regulation of miR-10c with HoxB4a suggest that it could be regulated in the same way. In situ hybridization shows that treatment with 10−6 M RA leads to upregulation of miR-10c (figure 5A), in a manner similar to that of HoxB4a, showing that transcription of miR-10c is indeed activated by RA.

thumbnail

Figure 5. Retinoid induction of miR-10c and upregulation of HoxB1a in miR-10 morphants.

A) LNA in situ hybridization for miR-10c in wildtype (WT) and 10−6M retinoic acid (RA) treated embryos. RA treatement results in miR-10c upregulation. B) Presence of a DR[2] type retinoic acid response element (RARE) 1kb 3′ of the Zebrafish HoxB1a gene. This sequence is conserved in the mouse in which it has been shown to mediate the neural response of HoxB1 to RA [38] C) Different response of HoxB1a to RA stimulation in wildtype or miR-10 morphant embryos. HoxB1a is strongly upregulated in miR-10 morphants. Injection with the miR-10 siRNA has no effect. HoxB4a responds similar to all conditions.

doi:10.1371/journal.pone.0001396.g005

In mouse, HoxB1 possesses several retinoid response elements and retinoids play a role in the establishment of the endogenous neural expression pattern [38], [39], [40], [41]. The 3′ DR[2] type retinoid response element, which has been shown to regulate neural RA responsiveness of the mouse HoxB1 gene [38], is conserved in the Zebrafish HoxB1a gene (figure 5B). In RA stimulated Zebrafish embryos, HoxB1a expression is indeed no longer restricted to a single strong domain (figure 5C) and this gene is expressed much more extensively throughout the embryo, suggesting activation. There is, however, a decrease in expression level, which is remarkable in the light of the presumed activating activity of RA. We investigated whether the co-activation of miR-10 plays a role in this. When miR-10 knockdown embryos are stimulated with 10−6 RA, an increased upregulation of HoxB1a but not HoxB4a is observed (figure 5C) compared to wildtype treated embryos. Apparently RA simultaneously activates expression of both HoxB1a and miR-10 and miR-10 modulates the downstream response to retinoid signaling.

Overexpression of miR-10 induces phenotypes associated with loss of HoxB1a and HoxB3a but not HoxB1b

The phenotypes of morphant and overexpression embryos are remarkably normal during the first 5 days of development and they have no apparent defects (figure 6A, 72 hpf embryos shown). However, as miR-10 appears to target HoxB1a and HoxB3a and possibly other 1 and 3 paralogue genes, the phenotype of miR-10 overexpression is expected to combine at least the loss of function phenotypes for these genes.

thumbnail

Figure 6. Overexpression of miR-10 induces HoxB1a and HoxB3a loss of function phenotypes.

A) Wildtype (WT), miR-10 morphant (MO1, MO2) and miR-10 siRNA overexpression embryos at 72 hpf show no apparent developmental differences. B) Mauthner neuron development as visualized by 3A10 neurofilament immunostaining in 72 hpf embryos shows no differences between miR-10 siRNA injected embryos and controls. C) Confocal images of reticulospinal hindbrain neurons in retrograde labeled, 5 day old embryos. Wildtype and miR-10 siRNA injected embryos are similar. D) Islet-1 and tag-1 in situ hybridization on 30 hpf wildtype and miR-10 siRNA injected embryos. Flatmounts of head regions are shown. In wildtype embryos the VIIth cranial nerve migrates into rhombomere 5/6 at the level of the otic vesicle. In miR-10 siRNA injected embryos the VIIth nerve does no longer migrate out of rhombomere 4. E) Co-injection of 5pg HoxB1a RNA rescues the miR-10 siRNA induced migration defect of the VIIth cranial nerve as shown by islet-1 and tag-1 in situ hybridization. F) Gcm-2 expression is downregulated in miR-10 siRNA injected embryos, which is consistent with repression of HoxB3a.

doi:10.1371/journal.pone.0001396.g006

In Zebrafish, morpholino studies have shown that HoxB1a is required for the correct patterning of rhombomere 4 together with HoxB1b [42]. Single knockdown of HoxB1a results in a failure of the branchiomotor neurons of the VIIth cranial nerve to migrate out of rhombomere 4. In the double knockdown of HoxB1a and HoxB1b, there is additional absence of the rhombomere 4 primary Mauthner neurons.

Morpholino knockdown of HoxB3a and HoxA3a has been shown to result in downregulation of the gcm-2 gene in the branchial arches in Zebrafish [43].

Analysis of the miR-10 overexpression phenotype by immunolabeling with the primary neuron specific 3A10 antibody shows that the Mauthner neurons are still present (figure 6B). Retrograde labeling in 5 days old embryos also reveals a normal pattern of reticulospinal neurons projecting from the hindbrain into the spinal cord (figure 6C).

Hindbrain branchiomotorneurons were visualized in in situ hybridization with islet-1 and tag-1. Islet-1 stains the bodies of the Vth, VIIth, IXth and Xth nerves and tag-1 is specifically expressed in the migrating VIIth nerve.

In situ hybridization on 30 hpf miR-10 injected embryos shows that branchiomotor neurons of the VIIth nerve no longer migrate into rhombomere 5 and 6 but remain in rhombomere 4 (figure 6D). The pattern of the Vth, IXth and Xth branchiomotor nerves as visualized by islet-1 appears normal. To show that the VIIth nerve defect is directly due to targeting of HoxB1a by miR-10, we rescued the miR-10 overexpression by co-injecting 5pg HoxB1a RNA from a construct that does not contain any of the target sites.

Injection of 5 pg HoxB1a alone does not induce any phenotype. Co-injection with miR-10 siRNA restores migration of the VIIth nerve as show by both islet-1 and tag-1 in situ hybridization (figure 6E).

To show targeting of HoxB3a, 72 hpf embryos were analyzed for the expression of gcm-2. In miR-10 overexpression embryos we observe downregulation of gcm-2 (figure 6F) in the branchial arch region as would be expected for embryos with impaired HoxA3a and/or HoxB3a expression [43].These analyses show that miR-10 is able to induce specific phenotypes associated with the loss of function of HoxB1a and HoxB3a/HoxA3a genes but not of HoxB1b. Analysis of the same genes in miR-10 morphant embryos shows patterns similar to wildtype embryos (figure S3).

MiR-10 acts synergistically with HoxB4

In Xenopus laevis, overexpression of HoxB4 has been reported to repress the expression of HoxB1 and HoxB3 in neuralized animal caps [44]. Considering that these are the same genes that are targeted by miR-10 and that there exists a close association between the microRNA gene and the HoxB4 open reading frame, this could indicate a synergistic action between miR-10 and HoxB4. To test this hypothesis we overexpressed HoxB4 and miR-10 individually and combined. Embryos were injected with 150pg hoxB4 RNA, miR-10 siRNA or a combination of the two and analyzed for HoxB1a and HoxB3a expression. Injection of 150 pg Xenopus laevis HoxB4 RNA strongly represses the hindbrain rhombomere 4 expression of HoxB1a and rhombomere 5/6 expression of HoxB3a (figure 7A).

thumbnail

Figure 7. MiR-10 acts in synergy with HoxB4.

A) Embryos injected with HoxB4, miR-10 siRNA and HoxB4+miR-10 siRNA analyzed for the expression of HoxB1a and HoxB3a at 24 hpf. Injection of 150pg HoxB4 leads to downregulation of HoxB1a and of downregulation of the hindbrain domain of HoxB3a. The rhombomere 4 expression domain of HoxB1a and the rhombomere 5/6 expression domain of HoxB3a are still discernable though. When 150pg HoxB4 is expressed together with miR-10 siRNA the expression domain of HoxB1a disappears and no clear rhombomere 5/6 stripe of HoxB3a expression can be detected. B) Embryos injected with HoxB4, miR-10 siRNA and HoxB4+miR-10 siRNA analyzed for the expression of endogenous HoxB4a at 48 hpf. The combination of HoxB4 together with the miR-10 siRNA induces a stronger phenotype with more severe anterior and posterior truncations than injection with HoxB4 alone. On the right groups of embryos injected with HoxB4 or the combination of HoxB4 and miR-10 siRNA are shown.

doi:10.1371/journal.pone.0001396.g007

However, for HoxB1a there is still a weak r4 domain detectable and for HoxB3a a discrete r5/6 stripe of expression is present. When coexpressed with miR-10, we observe a complete disappearance of the rhombomere 4 HoxB1a expression stripe (figure 7A). For HoxB3a, there still is expression in the hindbrain but a discrete r5/6 domain is no longer discernable (figure 7A). In addition to this, we observe a stronger phenotype at 48 hpf, with more severe anterior and posterior truncations in embryos injected with the combination of HoxB4 RNA and miR-10 siRNA (figure 7B). In situ hybridization for the endogenous HoxB4a shows that in the HoxB4+miR-10 siRNA co-injected embryos, the parts anterior and posterior to the endogenous HoxB4a domain are reduced more strongly than in embryos injected with HoxB4a only. These experiments indicate that miR-10 synergizes with HoxB4 in the repression of HoxB1a and HoxB3a and also attains a greater posteriorizing activity in the presence of miR-10.

Evolutionary conservation of the target sites in the HoxB cluster

Evolutionary conservation of sequence information is considered a good indicator of functionality and is used in microRNA target prediction programs [34] to assign confidence levels. We searched the anterior part of the HoxB(a) clusters in Medaka, Three spined stickleback, Tetraodon, Takifugu, Xenopus, Oppossum, Mouse, Rat, Cow and Human for the presence of putative miR-10 target sites (seed nucleotide 1-7). In figure 8A, the anterior parts of the HoxB and HoxBa cluster homologues are shown with indication of the identified seed sequences. The conservation of the miR-10 target sites in the 3′UTR and coding regions of HoxB3(a) genes is clear; all species investigated have at least 2 target sites associated with the ORF or 3′ UTR region. The sites in HoxB1(a) show a weaker conservation profile and are most prominently present in Zebrafish. All of the Teleosts for which sequence information could be found (note that the available HoxBa Medaka contig stops 300 nt downstream of HoxB1) posses a candidate miR-10 target site in the 3′ UTR of their HoxB1a gene. This further adds to the implied relevance of the repression of HoxB1a and HoxB3a by miR-10.

thumbnail

Figure 8. Evolutionary conservation of miR-10 targetsites and autoregulation of miR-10c.

A) Putative miR-10 target sites are indicated by seed sequences in the sense strand of the anterior vertebrate HoxB(a) clusters. Seed sequences are shown in green, open reading frames are indicated in light blue. Note conserved association of target sites with the HoxB3(a) ORF and conserved presence of a putative target site in Teleost HoxB1a. B) The HoxB3a splv2 polycistronic transcript includes one exon between HoxB4a and HoxB5a, two exons between HoxB4a and HoxB3a and the main HoxB3a coding sequence. The primary transcript for this isoforms includes miR-10c. The 5′ UTR sequence is shown in orange, this sequence corresponds to the probe used in C and D to specifically detect this splice isoforms. C) Comparison of the HoxB3a exon1 expression (red) and the expression of HoxB3a splv2 (purple). HoxB3a splv2 is expressed posterior to the main rhombomere 5/6 expression domain of HoxB3a as reported previously [46]. The staining reaction for HoxB3a splv2 was developed for much longer than the reaction for the HoxB3a exon1 probe and the HoxB3a splv2 is presumably expressed at a much lower level. D) In situ hybridization with HoxB3a splv2. Expression is upregulated in miR-10 morphant embryos (arrows). E) Semi quantitative RT-PCR for the HoxB3a splv2 5′ UTR, ß-actin is used as loading control. HoxB3a is upregulated in miR-10 morphant embryos. HoxB3a splv2: 31 cycles, ß-actin: 22 cycles.

doi:10.1371/journal.pone.0001396.g008

A polycistronic transcript including both HoxB3a and miR-10c is targeted by miR-10

A transcript, HoxB3a splv2 [45], has been described, which starts 3′ of HoxB5a, has two exons originating 3′of HoxB4a and includes the main HoxB3a open reading frame (figure 8B). The primary unspliced form of this long transcript thus includes both HoxB3a and the miR-10c microRNA.

In situ hybridization with a 5′ UTR probe shows that expression of this transcript obeys the rules of colinearity and that its rostral expression boundary thus corresponds to the position of its transcriptional start site (i.e. expression similar to that of HoxB5a) (figure 8C) and [46]. This transcript is thus expressed more posteriorly than the main HoxB3a expression domain and completely within the domain of the miR-10c microRNA, a feature also expected from the presence of miR-10c on the HoxB3a splv2 primary transcript. This transcript includes the full HoxB3a open reading frame together with the 3 miR-10 target sites. Morpholino knockdown of miR-10 leads to upregulation of this transcript in both in situ hybridization and RT-PCR (figure 8D, E), confirming that it is indeed targeted by miR-10 in vivo. In this case the miR-10c microRNA apparently acts on parts of its own primary transcript and is therefore autoregulatory.

Discussion

MiR-10 is expressed in the hindbrain and spinal cord posterior to the rhombomere 6/7 boundary and occupies an axial domain similar to those of Hox-4 paralogue genes. We reveal an interaction between miR-10 and the anterior HoxB1a and HoxB3a genes. These Hox genes have strong anterior expression domains in the hindbrain and are expressed at a low level in the spinal cord where their expression overlaps with miR-10 expression (figure 9A). The upregulation of the target genes in the morphant embryos shows that the target genes are indeed repressed by miR-10 within this posterior domain.

thumbnail

Figure 9. Post-transcriptional regulatory interactions within the hox clusters.

A) Schematic representation of miR-10 and target gene expression in the Zebrafish hindbrain. MiR-10 is expressed posterior from the rhombomere 6/7 boundary. The target genes HoxB1a and HoxB3a are expressed in a strong domain (dark colour) anterior in the anterior hindbrain and in a weaker domain (light colour) in the area where they overlap with miR-10. HoxB1a shows a gap in expression in r5 and 6, possibly due to stronger transcriptional repression. B) Schematic representation of the post-transcriptional relations within the hox clusters. MiR-196 is known to represses HoxB8, HoxC8, HoxD8 and HoxA7 and we have identified HoxB1a and HoxB3a as targets for miR-10. The emerging view is that the microRNAs in the hox clusters target more anterior genes in their close proximity. C) Polycistronic transcripts identified from the EST database show inclusion of miR-196 paralogues and HoxB8 and HoxC8 target genes on the same primary transcripts.

doi:10.1371/journal.pone.0001396.g009

Overexpression of miR-10 indeed induces the phenotypes associated with the loss of HoxB1a and HoxB3a. The very specific phenotype induced in the overexpression experiments is striking. The embryos are virtually indistinguishable from wildtypes, apart from the VIIth nerve defect and altered gcm-2 expression. In the target prediction section of miRBase, where the output of the Miranda algorithm (http://microrna.sanger.ac.uk/) is listed, there are however 1969 predicted target genes for Zebrafish miR-10. This high number of predicted target genes seems incompatible with the very specific phenotypic defects observed in the miR-10 overexpression embryos and strongly suggest that, at least for miR-10, there is a high component of false positives in the outcome of target prediction algorithms.

We observe that the HoxB4 overexpression phenotype is significantly enhanced by miR-10. The close genomic association of these two genes and their synergistic activity could indicate that these genes are part of the same genetic unit. The system of Hox regulation is characterized by the phenomenon of posterior prevalence, meaning that there is a hierarchy in the functioning of the Hox genes, such that posterior genes are always dominant in the determination of a regional phenotype over coexpressed anterior genes [2]. In the overexpression experiments it looks as if miR-10 facilitates a full posteriorizing activity of HoxB4. This synergistic action between miR-10 and HoxB4 suggests that post-transcriptional gene regulation by microRNAs plays a role in posterior prevalence.

We find that HoxB1a and HoxB3a are targeted by miR-10. Other Hox-1 and Hox-3 paralogues genes are present in different Zebrafish Hox clusters but are not targeted. On the basis of our experiments and the presence of putative target sites, only HoxA1a, HoxB1a, HoxA3a and HoxB3a genes are candidate miR-10 targets. Whether they are true targets remains to be seen; HoxA1a, despite not being near to a clear target site, responds strongly to the loss and gain of miR-10. HoxB1b possesses a candidate target site and is upregulated in morphant embryos. What however argues strongly against the targeting of HoxB1b is that the overexpression of miR-10 does not induce the same phenotypic changes as observed in the double HoxB1a/HoxB1b knockdown [42]. As there is extensive crossregulation between Hox paralogues it is also possible that the effects observed are a direct result of the derepression of HoxB1a. HoxA3a also possesses one candidate target site but seems unaffected by gain and loss of miR-10. MicroRNAs affect target genes both by inhibition of translation and by degradation of messengerRNA [20], [36]. To what extent these processes are coupled and whether translational inhibition is always accompanied by an increase in messenger RNA decay is not yet known. It is thus theoretically possible that the effect of HoxA3a repression will only be noticeable at the protein level.

There could be several reasons why miR-10 targets particular Hox-1 and Hox-3 genes and not others. One explanation would be that there is a high degree of subfunctionalization within these paralogue groups, as was nicely illustrated for the Zebrafish Hox-1 genes [42]. This could create needs for post-transcriptional silencing that differ from one paralogue member to another.

The inhibition of miR-10 leads to posterior upregulation of the targeted genes. Since microRNAs cause downregulation of their target messenger RNAs, it has been a frequently debated issue whether low or absent levels of target gene expression within the microRNA domain reflect different domains of transcription or whether they are a direct consequence of the downregulation by the microRNA [e.g. 47]. Inhibition of miR-10 leads to posterior target gene upregulation in case of HoxB1a, HoxA1a and HoxB3a but certainly not to the same high level found in their dominant anterior expression domains. It thus appears that restriction to the dominant expression domain occurs primarily at the transcriptional level and that it is within a posterior domain with an already low level of transcription that silencing by the microRNA occurs. This observation is consistent with the identification of rhombomere specific transcriptional Hox enhancers in mouse [e.g. 48], [49]. In case of the microRNA target interactions described in this study, it seems that both different transcriptional domains and a direct repression by the microRNA shape the mRNA expression domains in the embryo. The effects at the transcript level likely reflect an on/off situation at the protein level where all translation is silenced although there is still a significant amount of messengerRNA detectable.

Why are there post-transcriptional gene regulatory interactions within the Hox clusters?

In general it is poorly understood why functional domains of genes are sometimes restricted post-transcriptionally instead of by transcriptional silencing. The emerging view for the microRNAs in the Hox clusters is that they target coding Hox genes (figure 9B) that can even be located within the same clusters. These microRNAs thus seem to be involved in post-transcriptional gene regulatory interactions with genes that are located in their very close vicinity. The short genomic distances between miR-196 and miR-10 and their targets are remarkable; miR-10c is ~25 kb from the target sites in HoxB3a and ~48 kb from those in HoxB1a and (in mammals) a miR-196 paralogue is located at ~18 kb from HoxB8 and HoxC8 and at ~14 kb from HoxA7.

In this light, the presence of miR-10c and its target HoxB3a on a single primary transcript is also interesting. The miR-10c/HoxB3a polycistronic transcript includes both the microRNA and a target gene. In this case, the microRNA thus acts in an autoregulatory fashion on parts of its original precursor. The HoxB3a splv2 transcript itself appears to be expressed exclusively within the expression domain of the microRNA and is thus never expected to be translated into a functional HoxB3a protein. Similar transcripts are present in the EST database for miR-196a-1/HoxB8 and miR-196a-2/HoxC8 (figure 9C). The inclusion of microRNAs and target genes on the same transcription unit is counterintuitive in the sense that one wonders why the target genes are not simply omitted from the transcript as they are silenced by the accompanying microRNA anyway.

A possible explanation for the presence of target gene/microRNA combinations within single transcription units and for the targeting of nearby Hox genes by both miR-10 and miR-196, may lie in the complexity of the Hox regulatory mechanisms which involve multiple global and local transcriptional elements. The high selective pressure to maintain the clustered genomic organization of vertebrate Hox genes probably results from the presence of global enhancers located outside of the clusters and from dependence on sharing of local enhancers [50]. As a result, the Hox clusters consist of closely spaced transcription units and enhancer regions. The high density of transcription units could easily cause them to interfere with one another and make the system prone to inappropriate enhancer sharing, resulting in ectopic expression. The extensive amount of ‘strange’ polycistronic and antisense transcripts being produced from the Hox clusters [51] could result from this. These transcripts do not necessarily have any function but could represent inherent transcriptional ‘noise’. It is interesting that it is specifically the nearest genes that are silenced post- transcriptionally as these are the ones most likely to be influenced by the same enhancers as the microRNA genes. The posterior expression domains of these anterior Hox genes could well be a consequence imposed on the transcriptional process by the clustered nature of the genes and they do not necessarily serve any function. We suggest that an inability to separate the transcriptional controls of several Hox genes is the selective force driving the post-transcriptional gene silencing relationships within the Hox clusters.

In vertebrates, Hox genes have stayed clustered throughout evolution. However, in the sister group of tunicates (Ciona and Oikopleura), the Hox clusters have broken up and are present in separate regions of the genome [52], [53]. It is interesting to note that the miR-10 microRNA has been lost from the tunicates [17], [54]. This observation provides a possible phylogenetic link between post-transcriptional gene silencing and gene clustering.

It would be interesting to see whether it is possible to extrapolate these observations to other microRNA/(predicted) target pairs and see if similar constraints can be identified that possibly account for the involvement of post-transcriptional gene regulation.

Materials and Methods

Zebrafish husbandry and embryo culturing

An AB×TL strain of Zebrafish was used for all experiments; housing and embryo collection was according to standard procedures; embryos were cultured at 28°C.

RT-PCR

Whole embryo RNA was isolated using Tri-pure (Roche #1667165) and reverse transcribed with MuMlv Reverse transcriptase (Promega) using oligo-dT N = 18.

Primer sequences;

  1. miR-10c up (AGCTGGCTTTCTCAATACC)
  2. miR-10c dow n (TACATACTCCCCTAGATACGAA)
  3. HoxB4a exon1 up (ATGGCCATGAGTTCCTATTTG)
  4. HoxB4a exon1 down (TTGGTTCACCCCCTGAATAG)
  5. HoxB4a exon1 5′down (TTGTGGGTAGAACGTGACCTC)
  6. HoxB3a splv2 5′ UTR up (CAGTGCCAGTGTCTAGTCAG)
  7. HoxB3a splv2 5′UTR down (GTAATACGACTCACTATAGGCTCTTTCCAATGGCCTCTT​GG)
  8. β-Actin up (CGAGCAGGAGATGGGAACC)
  9. β-Actin down (CAACGGAAACGCTCATTGC)

DNA oligos were obtained from Biolegio, Malden, The Netherlands.

Micro-injection

Embryos were injected with 1 or 2 nl at the zygote stage; RNAse free phenol red was added as tracer to injection mixtures prior to injections.

Morpholinos were obtained from genetools, OR, USA; miR-10 morpholino reagent 1 corresponds to a mix of miR-10a (CACAAATTCGGATCTACAGGGTA) and miR-10b (CACAAATTCGGTTCTACAGGGTA) antisense morpholino, the sequence of miR-10 morpholino reagent 2 is (TCTACAGGGTATATATAGACGAC).

RNA oligos were obtained from Biolegio, Malden, The Netherlands.

The miR-10 siRNA sense strand corresponds to a mix of miR-10a (UACCCUGUAGAUCCGAAUUUGUGUG) and miR-10b (UACCCUGUAGAACCGAAUUUGUGUG), sequence of the antisense strand is (CACAAAUUCGGAUCUACAGGGGCAU). Note that the antisense sequence has mismatches with the miR-10 sense strand at its 3′ end resulting in the specific incorporation of the sense miR-10 strand in the microRNA silencing complex. Oligos were annealed to siRNAs in by gradually cooling from 98°C to 20°C. in buffer in 500ml H2O beaker glass; 30 ul 50 µM of each oligo, 15 µl annealing buffer (50 mM Tris, pH 7.8, 100 mM NaCl RNAse free) in 75 µl, final concentration of siRNA is 20 µM.

RNA for injection was transcribed using Ambion Sp6 message machine kit (# 1340) and purified using an RNA easy column (Qiagen), from the CS2+ plasmids; CS2+HoxB1a sensor wt, CS2+HoxB1a sensor mut, CS2+HoxB3a sensor wt, CS2+ HoxB3a sensor mut, CS2+Dre-HoxB3a ORF, CS2+Xl-HoxB4-Myc, CS2+E-YFP, CS2+E-CFP.

In situ Hybridization

In situ hybridization was performed according to standard procedures and Kloosterman et al. [32]. Hybridization temperatures were 65°C for normal probes and 56°C for LNA probes. In double in situ hybridization with a LNA probe 56°C was used. In double in situ hybridization DIG and fluorescein labeled probes were used. Embryos were stained using BM-Purple (Roche #11442674001) and Fast Red (Roche #11496549001). Probes were synthesized using T7 and Sp6 polymerase (Promega) in the presence of labeled nucleotides (RNA DIG or fluorescein labeling mix, Roche #11277073910 and #10805221) from pGEM-TE plasmids containing: HoxB1a, HoxB3a, HoxB4a and HoxB5a exon 1 coding sequence, HoxB2a exon 2-3′UTR, HoxB1b exon1-2 coding sequence; HoxB3a splv2 was synthesized from PCR product from a partial cDNA cloned in pGEM-TE.

LNA probes were obtained from Exiqon, Denmark and sequences are:

  1. miR-10a (CACAAATTCGGATCTACAGGGTA),
  2. miR-10b (ACAAATTCGGTTCTACAGGGTA),
  3. miR-10c (CACAAATCCGGATCTACAGGGTA),
  4. miR-10d (ACACATTCGGTTCTACAGGGTA ).

Probes were labeled using the DIG labeling kit (Roche #03353575910) and purified before use over a microspin G-25 column (Amersham #27-5325-01) Our step by step in situ hybrdization protocol is available on request.

Nothern Blot

Northern Blot was performed essentially according to Kloosterman et al. [55]. Total RNA was extracted using Tri-Pure (Roche #1667165). 3 µg RNA was separated on a 15% denaturing PAGE gel using a Biorad minigel system and subsequently blotted using a semidry blotter (175 mA constant, 10–20V for 25–30 min.) to a postitively charged nylon membrane (Roche #1417240). Membranes were pre-hybridized at 60°C for 1 hr in hybridization buffer (0.36 M Na2HPO4, 0.14M NaH2PO4, 1 mM EDTA, 7%SDS, 0.1 mg/ml yeast tRNA, 0.04% Blocking reagent (Roche #1096176)) and subsequently hybridized overnight at 60°C in hybridization buffer containing miR-10 LNA probes, labeled and purified as mentioned above, and diluted 1:50.000. The next day blots were washed 1× at 60°C with hybridization buffer, 1× at 50°C with 2× SSC, 0.1% SDS and 1× at 50°C with 0.1xSSC, 0.1%SDS in order to remove excess probe. Blots were incubated 2× for 5 min. at RT in Maleic Acid buffer (0.1 M Maleic Acid, 150mM NaCl, pH7.5 with NaOH, 0.1% Tween-20) to equilibrate and remove residual SDS. Blots were blocked for 30 min. in blocking buffer (Maleic Acid buffer containing 1% Blocking Reagent (Roche #1096176)) and subsequently incubated for 30 min. in blocking buffer containing 1:50.000 Anti-Digoxigenin-AP antibody (Roche #093274). Excess antibody was washed away in 4×15 min. washes with Maleic Acid buffer. Blots were subsequently washed 5 minutes in AP-buffer (0.1 M Tris Base, 0.1 M NaCl, pH9.5) and signal was detected on X-Ray film using CDP-star kit (Roche #12041677001) according to manufacturer's instructions with a typical exposure time of 4 hours.

Retrograde labeling

Anesthetized 5 days old embryos were retrograde labeled by making an incision with a tungsten needle in the spinal cord at the level of the hindgut and injecting 1–5 nl of a concentrated rhodamine-dextran solution. After injection embryos were left to recover for 1–1.5 hrs and fixed in 4% PFA.

Supporting Information

Figure S1.

doi:10.1371/journal.pone.0001396.s001

(0.32 MB PDF)

Figure S2.

doi:10.1371/journal.pone.0001396.s002

(0.38 MB PDF)

Figure S3.

doi:10.1371/journal.pone.0001396.s003

(0.75 MB PDF)

Acknowledgments

We would like to thank Nabila Bardine, Gabby Krens, Hans Jansen, Ferran Lloret Vilaspasa, Max Corredor-Adámez and Martje Jespers for useful discussion and assistance in the lab, Yanju Zhang and Fons Verbeek for help with miRBase, Florian Steiner for advise on the Northern Blotting, Jacqueline Deschamps for useful comments on the manuscript, Victoria Prince for gift of the HoxB1a overexpression construct and two anonymous reviewers for useful criticism and suggestions. This project was initiated at the Hubrecht Institute for Developmental Biology and Stem Cell Research.

Author Contributions

Conceived and designed the experiments: JW. Performed the experiments: JW. Analyzed the data: JW. Contributed reagents/materials/analysis tools: AD JW. Wrote the paper: AD JW. Other: Main author: JW. Lab head: AD. General supervision: AD. Provided facilities: AD. Paid JMW: AD.

References

  1. 1. Boncinelli E, Mallamaci A, Lavorgna G (1994) Vertebrate homeobox genes. Genetica 94(2–3): 127–40.
  2. 2. Kmita M, Duboule D (2003) Organizing axes in time and space; 25 years of colinear tinkering. Science 301(5631): 331–3.
  3. 3. Krumlauf R (1994) Hox genes in vertebrate development. Cell 78(2): 191–201.
  4. 4. Pearson JC, Lemons D, McGinnis W (2005) Modulating Hox gene functions during animal body patterning. Nat Rev Genet 6(12): 893–904.
  5. 5. Amores A, Force A, Yan YL, Joly L, Amemiya C, et al. (1998) Zebrafish Hox clusters and vertebrate genome evolution. Science 282(5394): 1711–4.
  6. 6. Corredor-Adámez M, Welten MC, Spaink HP, Jeffery JE, Schoon RT, et al. (2005) Genomic annotation and transcriptome analysis of the zebrafish (Danio rerio) Hox complex with description of a novel member, Hox b 13a. Evol Dev 7(5): 362–75.
  7. 7. Hoegg S, Boore JL, Kuehl JV, Meyer A (2007) Comparative phylogenomic analyses of Teleost fish Hox gene clusters: lessons from the cichlid fish Astatotilapia burtoni. BMC Genomics 8: 317.
  8. 8. Lim LP, Glasner ME, Yekta S, Burge CB, Bartel DP (2003) Vertebrate microRNA genes. Science 299(5612): 1540.
  9. 9. Lagos-Quintana M, Rauhut R, Meyer J, Borkhardt A, Tuschl T (2003) New microRNAs from mouse and human. RNA 9(2): 175–9.
  10. 10. Cummins JM, He Y, Leary RJ, Pagliarini R, Diaz LA Jr, et al. (2006) The colorectal microRNAome. Proc Natl Acad Sci USA 103: 3687–92.
  11. 11. Mineno J, Okamoto S, Ando T, Sato M, Chono H, et al. (2006) The expression profile of microRNAs in mouse embryos. Nucleic Acids Res 34(6): 1765–71.
  12. 12. Bernstein E, Caudy AA, Hammond SM, Hannon GJ (2001) Role for a bidentate ribonuclease in the initiation step of RNA interference. Nature 409(6818): 363–6.
  13. 13. Lee Y, Ahn C, Han J, Choi H, Kim J, et al. (2003) The nuclear RNase III Drosha initiates microRNA processing. Nature 425(6956): 415–9.
  14. 14. He L, Hannon GJ (2004) MicroRNAs: small RNAs with a big role in gene regulation. Nat Rev Genet 5(7): 522–31.
  15. 15. Valencia-Sanchez MA, Liu J, Hannon GJ, Parker R (2006) Control of translation and mRNA degradation by miRNAs and siRNAs. Genes Dev 20(5): 515–24.
  16. 16. Woltering JM, Durston AJ (2006) The zebrafish HoxDb cluster has been reduced to a single microRNA. Nat Genet 38(6): 601–2.
  17. 17. Tanzer A, Amemiya CT, Kim CB, Stadler PF (2005) Evolution of microRNAs located within Hox gene clusters. J Exp Zoolog B Mol Dev Evol 304(1): 75–85.
  18. 18. Wienholds E, Koudijs MJ, van Eeden FJ, Cuppen E, Plasterk RH (2003) The microRNA-producing enzyme Dicer1 is essential for zebrafish development. Nat Genet 35(3): 217–8.
  19. 19. Giraldez AJ, Cinalli RM, Glasner ME, Enright AJ, Thomson JM, et al. (2005) MicroRNAs regulate brain morphogenesis in zebrafish. Science 308(5723): 833–8.
  20. 20. Giraldez AJ, Mishima Y, Rihel J, Grocock RJ, Van Dongen S, et al. (2006) Zebrafish MiR-430 promotes deadenylation and clearance of maternal mRNAs. Science 312(5770): 75–9.
  21. 21. Flynt AS, Li N, Thatcher EJ, Solnica-Krezel L, Patton JG (2007) Zebrafish miR-214 modulates Hedgehog signaling to specify muscle cell fate. Nat Genet 39(2): 259–63.
  22. 22. Kloosterman WP, Lagendijk AK, Ketting RF, Moulton JD, Plasterk RH (2007) Targeted inhibition of miRNA maturation with morpholinos reveals a role for miR-375 in pancreatic islet development. PLoS Biol 5(8): e203.
  23. 23. Yekta S, Shih IH, Bartel DP (2004) MicroRNA-directed cleavage of HOXB8 mRNA. Science 304(5670): 594–6.
  24. 24. Mansfield JH, Harfe BD, Nissen R, Obenauer J, Srineel J, et al. (2004) MicroRNA-responsive ‘sensor’ transgenes uncover Hox-like and other developmentally regulated patterns of vertebrate microRNA expression. Nat Genet 36(10): 1079–83.
  25. 25. Hornstein E, Mansfield JH, Yekta S, Hu JK, Harfe BD, et al. (2005) The microRNA miR-196 acts upstream of Hoxb8 and Shh in limb development. Nature 438(7068): 671–4.
  26. 26. Ronshaugen M, Biemar F, Piel J, Levine M, Lai EC (2005) The Drosophila microRNA iab-4 causes a dominant homeotic transformation of halteres to wings. Genes Dev 19(24): 2947–52.
  27. 27. Godsave SF, Koster CH, Getahun A, Mathu M, Hooiveld M, et al. (1998) Graded retinoid responses in the developing hindbrain. Dev Dyn 213(1): 39–49.
  28. 28. Kiecker C, Lumsden A (2005) Compartments and their boundaries in vertebrate brain development. Nat Rev Neurosci 6(7): 553–64.
  29. 29. Dasen JS, Tice BC, Brenner-Morton s, Jessell TM (2005) A Hox regulatory network establishes motor neuron pool identity and target-muscle connectivity. Cell 123(3): 477–91.
  30. 30. Moens CB, Prince VE (2002) Constructing the hindbrain: insights from the zebrafish. Dev Dyn 224(1): 1–17.
  31. 31. Wienholds E, Kloosterman WP, Miska E, Alvarez-Saavedra E, Berezikov E, et al. (2005) MicroRNA expression in zebrafish embryonic development. Science 309(5732): 310–1.
  32. 32. Kloosterman WP, Wienholds E, de Bruijn E, Kauppinen S, Plasterk RH (2006) In situ detection of miRNAs in animal embryos using LNA-modified oligonucleotide probes. Nat Methods 3(1): 27–9.
  33. 33. Lewis BP, Burge CB, Bartel DP (2005) Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell 120(1): 15–20.
  34. 34. John B, Enright AJ, Aravin A, Tuschl T, Sander C, et al. (2004) Human MicroRNA targets. PLoS Biol 2(11): e363.
  35. 35. Kloosterman WP, Wienholds E, Ketting RF, Plasterk RH (2004) Substrate requirements for let-7 function in the developing zebrafish embryo. Nucleic Acids Res 32(21): 6284–91.
  36. 36. Lim LP, Lau NC, Garrett-Engele P, Grimson A, Schelter JM, et al. (2005) Microarray analysis shows that some microRNAs downregulate large numbers of target mRNAs. Nature 2005 433(7027): 769–73.
  37. 37. Mainguy G, In der Rieden PM, Berezikov E, Woltering JM, Plasterk RH, et al. (2003) A position-dependent organisation of retinoid response elements is conserved in the vertebrate Hox clusters. Trends Genet 19(9): 476–9.
  38. 38. Huang D, Chen SW, Gudas LJ (2002) Analysis of two distinct retinoic acid response elements in the homeobox gene Hoxb1 in transgenic mice. Dev Dyn 223(3): 353–70.
  39. 39. Studer M, Gavalas A, Marshall H, Ariza-McNaughton L, Rijli FM, et al. (1998) Genetic interactions between Hoxa1 and Hoxb1 reveal new roles in regulation of early hindbrain patterning. Development 125(6): 1025–36.
  40. 40. Langston AW, Thompson JR, Gudas LJ (1997) Retinoic acid-responsive enhancers located 3′ of the Hox A and Hox B homeobox gene clusters. Functional analysis. J Biol Chem 272(4): 2167–75.
  41. 41. Sirbu IO, Gresh L, Barra J, Duester G (2005) Shifting boundaries of retinoic acid activity control hindbrain segmental gene expression. Development 132(11): 2611–22.
  42. 42. McClintock JM, Kheirbek MA, Prince VE (2002) Knockdown of duplicated zebrafish Hoxb1 genes reveals distinct roles in hindbrain patterning and a novel mechanism of duplicate gene retention. Development 129(10): 2339–54.
  43. 43. Hogan BM, Hunter MP, Oates AC, Crowhurst MO, Hall NE, et al. (2004) Zebrafish gcm-2 is required for gill filament budding from pharyngeal ectoderm. Dev Biol 276(2): 508–22.
  44. 44. Hooiveld MH, Morgan R, in der Rieden P, Houtzager E, Pannese M, et al. (1999) Novel interactions between vertebrate Hox genes. Int J Dev Biol 43(7): 665–74.
  45. 45. Hadrys T, Prince V, Hunter M, Baker R, Rinkwitz S (2004) Comparative genomic analysis of vertebrate Hox3 and Hox4 genes. J Exp Zoolog B Mol Dev Evol 302(2): 147–64.
  46. 46. Hadrys T, Punnamoottil B, Pieper M, Kikuta H, Pezeron G, et al. (2006) Conserved co-regulation and promoter sharing of Hoxb3a and Hoxb4a in zebrafish. Dev Biol 297(1): 26–43.
  47. 47. Plasterk RH (2006) Micro RNAs in animal development. Cell 124(5): 877–81.
  48. 48. Ferretti E, Cambronero F, Tumpel S, Longobardi E, Wiedemann LM, et al. (2005) Hoxb1 enhancer and control of rhombomere 4 expression: complex interplay between PREP1-PBX1-HOXB1 binding sites. Mol Cell Biol 25(19): 8541–52.
  49. 49. Kwan CT, Tsang SL, Krumlauf R, Sham MH (2001) Regulatory analysis of the mouse Hoxb3 gene: multiple elements work in concert to direct temporal and spatial patterns of expression. Dev Biol 232(1): 176–90.
  50. 50. Duboule D (1998) Vertebrate Hox gene regulation: clustering and/or colinearity? Curr Opin Genet Dev 8(5): 514–8.
  51. 51. Mainguy G, Koster J, Woltering J, Jansen H, Durston A (2007) Extensive polycistronism and antisense transcription in the Mammalian Hox clusters. PLoS ONE 2(4): e356.
  52. 52. Ikuta T, Yoshida N, Satoh N, Saiga H (2004) Ciona intestinalis Hox gene cluster: Its dispersed structure and residual colinear expression in development. Proc Natl Acad Sci U S A 101(42): 15118–23.
  53. 53. Seo HC, Edvardsen RB, Maeland AD, Bjordal M, Jensen MF, et al. (2004) Hox cluster disintegration with persistent anteroposterior order of expression in Oikopleura dioica. Nature 431(7004): 67–71.
  54. 54. Prochnik SE, Rokhsar DS, Aboobaker AA (2007) Evidence for a microRNA expansion in the bilaterian ancestor. Dev Genes Evol 217(1): 73–7.
  55. 55. Kloosterman WP, Steiner FA, Berezikov E, de Bruijn E, van de Belt J, et al. (2006) Cloning and expression of new microRNAs from zebrafish. Nucleic Acids Res 34(9): 2558–69.