DAZL Relieves miRNA-Mediated Repression of Germline mRNAs by Controlling Poly(A) Tail Length in Zebrafish

Background During zebrafish embryogenesis, microRNA (miRNA) miR-430 contributes to restrict Nanos1 and TDRD7 to primordial germ cells (PGCs) by inducing mRNA deadenylation, mRNA degradation, and translational repression of nanos1 and tdrd7 mRNAs in somatic cells. The nanos1 and tdrd7 3′UTRs include cis-acting elements that allow activity in PGCs even in the presence of miRNA-mediated repression. Methodology/Principal Findings Using a GFP reporter mRNA that was fused with tdrd7 3′UTR, we show that a germline-specific RNA-binding protein DAZ-like (DAZL) can relieve the miR-430-mediated repression of tdrd7 mRNA by inducing poly(A) tail elongation (polyadenylation) in zebrafish. We also show that DAZL enhances protein synthesis via the 3′UTR of dazl mRNA, another germline mRNA targeted by miR-430. Conclusions/Significance Our present study indicated that DAZL acts as an “anti-miRNA factor” during vertebrate germ cell development. Our data also suggested that miRNA-mediated regulation can be modulated on specific target mRNAs through the poly(A) tail control.


Introduction
Post-transcrptional regulation plays a crucial role in germ cell development. In zebrafish, primordial germ cells (PGCs) are determined by the incorporation of a specific maternal cytoplasm called germ plasm, which forms at the ends of cleavage planes in four-cell stage embryos [1,2]. Several maternal mRNAs, such as vasa [1], nanos1 [3], dead-end (dnd) [4], and Tudor domain-containing protein 7 (tdrd7) [5], are known to be localized to the germ plasm (hereinafter called germ plasm mRNAs). However, a fraction of germ plasm mRNAs is detected throughout the blastomeres during cleavage stages. Such unlocalized mRNAs are subsequently incorporated into somatic cells as well as PGCs. It has been shown that vasa, nanos1, and tdrd7 mRNAs are rapidly degraded in somatic cells but are stabilized in PGCs in a process mediated by cis-acting elements in their 39 untranslated regions (39 UTRs) [3,5,6,7]. Translation control also contributes to the restriction of NANOS expression to PGCs [3]. Our previous study showed that microRNA (miRNA) miR-430, which is abundantly expressed from the onset of zygotic gene expression [8], targets the 39UTRs of nanos1 and tdrd7 to induce mRNA deadenylation, mRNA degradation, and translational repression in somatic cells [5]. Importantly, miR-430 is also expressed in PGCs, and other target mRNAs of miR-430 are equally susceptible to repression in somatic cells and PGCs [5,9]. The nanos1 and tdrd7 39UTRs include cis-acting elements that allow activity in PGCs even in the presence of miRNA-mediated repression [5]. Thus miR-430mediated repression and the activation of germ plasm mRNAs play important roles in germline/somatic cell distinctions in zebrafish embryos.
Recently, Kedde et al. reported that a germline-specific RNAbinding protein DND counteracts the function of several miRNAs by binding mRNAs and prohibiting them from associating with their target sites [10]. In zebrafish PGCs, DND alleviates miR-430 repression of nanos1 and tdrd7. In the present study, we assumed that another RNA-binding protein, DAZL, can also relieve the miR-430-mediated repression of germ plasm mRNAs in zebrafish, for two reasons. First, as an evolutionarily conserved RNA-binding protein, DAZL plays a key role in germ cell development in animals such as Drosophila, C. elegans, Xenopus, and mouse [11]. In Xenopus and zebrafish, dazl mRNA is present in the germ plasm and in PGCs during early embryogenesis [2,12]. Second, DAZL activates the translation efficiency of the target mRNAs through direct binding to ciselements in their 39UTRs [13]. Moreover, Xenopus DAZL interacts with poly(A)-binding proteins (PABPs), which are critical for the initiation of translation [14].
Using a GFP reporter mRNA that was fused with tdrd7 39UTR, we show here that DAZL antagonizes miR-430-mediated repression of the tdrd7 mRNA in zebrafish embryos. Moreover, the addition of DAZL-binding elements to the synthetic miR-430 target mRNA led to mRNA stabilization in a PGC-specific manner in embryos. We also show that DAZL can enhance protein synthesis via the 39UTR of dazl mRNA which is localized to PGCs and targeted by miR-430. To our surprise, we found that DAZL induces polyadenylation of the reporter mRNA irrespective of the function of miRNA. Taken together, these results indicated that DAZL acts as an ''anti-miRNA factor'' during vertebrate germ cell development. Our data also suggested that miRNAmediated regulation can be modulated on specific target mRNAs through the poly(A) tail control.

DAZL activates tdrd7 expression through mRNA stabilization
To clarify whether or not DAZL enhances tdrd7 expression, we performed injection experiments of GFP reporter mRNA containing tdrd7 39UTR (GFP-tdrd7) [5] (Fig. 1). As an internal control, DsRed reporter mRNA was co-injected (Fig. 1A). When GFP-tdrd7 and DsRed mRNAs were injected, GFP expression was barely detectable in somatic cells at 24 hr post-fertilization (hpf), due to the repression by miR-430 [5] (Fig. 1A, panel e). In contrast, when wild-type DAZL was co-expressed (see Supplementary Fig. S1), we observed strong enhancement of GFP expression (4.2-fold compared to mock control) throughout the embryos, although DsRed expression was not affected (Fig. 1A, panel a). The mutant DAZL protein, DAZL F91A, which lacks RNA-binding activity because of the amino acid substitution in the RNA-recognition motif (RRM) [15], did not promote GFP-tdrd7 mRNA expression (Fig. 1A, panel b). In addition, we tested two types of truncated DAZL protein that cannot activate translation of target mRNAs: DAZL RRM possesses only the N-terminal portion containing RRM, whereas DAZL DDAZ lacks a conserved DAZ motif [15] (Fig. S1). We found that these truncated DAZL proteins enhanced GFP-tdrd7 expression only weakly (,1.6-fold, Fig. 1A, panels c and d). These results showed that DAZL can activate protein synthesis through the 39UTR of tdrd7 mRNA targeted by miR-430, and that the RRM and DAZ motifs are involved in the activation.
Efficiency of protein synthesis is often controlled by mRNA stability (stabilization/degradation) and/or translation efficiency (activation/repression) [16,17]. In zebrafish embryos, miR-430 enhances the degradation of target mRNAs [9]. Therefore, we examined whether or not DAZL protein affects the stability of GFP-tdrd7 mRNA. In situ hybridization of the embryos injected with GFP-tdrd7 mRNA was performed at 80%-epiboly (,8 hpf), since miR-430-mediated mRNA degradation can be observed immediately after the onset of zygotic expression (around 4-8 hpf) [5]. As a result, we found that the reporter mRNA abounded in the presence of DAZL protein, as compared with mock control (Fig. 1B). Northern analysis revealed that the amount of injected GFP-tdrd7 mRNA was increased more than 3-fold by DAZL expression (Fig. 1C). These results indicated that DAZL activates tdrd7 expression at least in part by enhancing mRNA stability.
The tdrd7 39UTR contains cis-elements for the activation by DAZL protein We examined whether DAZL's effect is specific to the miR-430 target site in the context of tdrd7 39UTR. The miR-204 target site (IPT miR204 ) was inserted into the 39UTR of the GFP-tdrd7 construct, in which the miR-430 target sites [5] had been disrupted by base substitutions (Fig. S2, panels A and B). miR-204 is not strongly expressed during early embryogenesis and therefore we could mimic the ubiquitous expression of miR-430 by injecting miR-204 duplex [5]. We found that somatic GFP expression from the reporter mRNA was inhibited by miR-204 co-injected into the embryo. In contrast, DAZL overexpression caused strong GFP expression from the reporter mRNA even in the presence of miR-204 (3.3-fold, Fig. S2, panel C). These results suggested that the primary sequence of the miRNA target site or the local secondary structure containing the target site is not important for the relief of miRNA function, and that the tdrd7 39UTR contains some cis-acting element(s) responsible for the regulation by DAZL.
Previous studies showed that the DAZ family proteins bind to the cis elements in 39UTR to activate the translation of target mRNAs: zebrafish DAZL binds to the GUUC element [15], while human and mouse DAZL bind to U-rich elements such as U2-10(G/C)U2-10 and (G/C)Un [18,19,20,21]. We examined whether tdrd7 39UTR has such a cis-regulatory element responsible for the enhanced expression induced by DAZL. When a GFP reporter mRNA having the 59 half of tdrd7 39UTR (corresponding to the nucleotides +1-147 relative to the stop codon) was tested, DAZL strongly activated GFP expression (4.6-fold, Fig. 2B), suggesting that this region includes a cis-element(s) for DAZL. Furthermore, DAZL activated the expression of GFP-IPT miR-430 91-123 mRNA that contains a portion of the tdrd7 39UTR (corresponding to the nucleoteides +91-123 relative to the stop codon) fused with a single copy of the exogenous miR-430 target site (IPT miR-430 ) (4.1-fold, Fig. 2B). This portion of the 39UTR contains the GUUC element and its related sequence, GUUA. When we examined a reporter mRNA, GFP-IPT miR-430 91-123 mt, in which we had introduced the base substitutions into the GUUC and GUUA elements, GFP expression was not enhanced even in the presence of DAZL (1 .1-fold, Fig. 2B). These results strongly suggested that DAZL activates tdrd7 expression via binding to the cis-elements in the 39UTR. The tdrd7 39UTR contains additional copies of the GUUC element as well as U-rich sequences outside region +91-123 ( Fig. 2A). Although we have not examined the role of these elements, it is possible that DAZL acts on such elements to enhance the activation of tdrd7 expression.
DAZL-binding element is sufficient for the PGC-specific activation of miR-430 target mRNA Expression of the reporter mRNA, GFP-3xIPT miR-430 , containing three copies of synthetic miR-430 target site (IPT miR-430 ) that is derived from nanos1 39UTR is completely repressed by miR-430 not only in somatic cells but also in PGCs [5,9]. To further examine whether or not direct binding of DAZL to the cis-element is sufficient for the relief of miRNA-mediated repression, we tested a reporter mRNA, GFP-3xIPT miR-430 6x#16, containing three copies of the miR-430 target site fused with six copies of the in vitro selected DAZL-binding sequence, #16, that possessed the GUUC element [15] (Fig. 3A). We found that DAZL enhanced GFP expression from the reporter mRNA (4.3-fold, Fig. 3B). In contrast, GFP expression of a reporter mRNA (GFP-3xIPT miR-430 6xmt2), containing miR-430 target sites with the mutant elements of #16 that do not bind to DAZL [15], was not largely affected by DAZL (1.0-fold, Fig. 3B). These results led us to conclude that DAZL binds to the cis-elements in the 39UTR and cancels the miR-430-mediated repression of tdrd7.
To determine whether miR-430-mediated repression is relieved endogenously through the DAZL-binding sequence, we performed in situ hybridization of embryos injected with GFP-3xIPT miR-430 6x#16 mRNA. As a result, the mRNA was detected in a PGC-specific manner in the absence of exogenous DAZL, while GFP-3xIPT miR-430 +6xmt2 mRNA was undetectable throughout the embryo (Fig. 3C). These results suggested that endogenous DAZL protein cancels the repressive effect of miR-430 on tdrd7 mRNA in PGCs.

Induction of mRNA polyadenylation by DAZL
How does DAZL cancel the miRNA-mediated repression? In zebrafish embryos, miRNA induces deadenylation of target mRNA [5,9]. To test whether or not DAZL affects poly(A) tail length, we measured the poly(A) tail length of reporter mRNAs injected into embryos. After injection, GFP reporter mRNA was subjected to RNase H-digestion in the presence of a specific oligonucleotide complementary to the 39 end of GFP sequence, and the resulting short RNA fragment containing the 39UTR and poly(A) tail was detected by Northern blotting [5]. We found that the injected GFP-tdrd7 mRNA lost its poly(A) tail at 3-6 hpf, as expected [5] (Fig. 4A, and data not shown). In contrast, the mRNA possessed a long poly(A) tail in the presence of DAZL (Fig. 4A). The mutant DAZL did not affect poly(A) tail length (Fig. 4A). Therefore, it is likely that DAZL somehow induces polyadenylation, or represses deadenylation, of tdrd7 mRNA.
We could think of three possible mechanisms by which DAZL might induce polyadenylation of tdrd7 mRNA. The first possibility is that DAZL activates tdrd7 translation without affecting poly(A) tail length, and subsequently active translation results in the polyadenylation of tdrd7 mRNA. The second is that DAZL blocks miR-430 function on tdrd7 mRNA or its access to tdrd7 mRNA, and thereby tdrd7 mRNA is free from the deadenylation mediated by miR-430. The third is that DAZL directly promotes polyadenylation of tdrd7 mRNA. To evaluate these possibilities, we performed two kinds of experiments.
First, we injected the A-capped reporter mRNA instead of the normal m 7 G-capped mRNA. As the A-capped mRNA cannot be recognized by translation initiation factor (cap binding protein), active translation does not occur while the mRNA is stable in embryos [5]. We found that A-capped GFP-tdrd7 mRNA was deadenylated, like the m 7 G-capped one, in the absence of DAZL protein; this is consistent with our previous result with GFP-nanos mRNA [5] (Fig. 4C). In contrast, DAZL induced the poly(A) tail elongation of A-capped GFP-tdrd7 mRNA (Fig. 4C). These results exclude the first possibility that active translation induced by DAZL results in the polyadenylation of tdrd7 mRNA. Second, we measured the poly(A) tail length of a mutant GFP-tdrd7 mRNA that lacks the miR-430 target sites, which is not subjected to the repression mediated by miR-430 [5]. As a result, the injected reporter mRNA possessed a long poy(A) tail approximately 150 nt long in the absence of functional DAZL protein (Fig. 4B). In contrast, we found that the poly(A) tail of the reporter mRNA in the presence of DAZL protein was much longer (approximately 200-250 nt long) (Fig. 4B). The results did not support the second possibility that DAZL only blocks deadenylation mediated by miR-430. Rather, it is likely that DAZL promotes the polyadenylation of tdrd7 mRNA. DAZL also controls another germline mRNA, dazl, targeted by miR-430 Our previous study suggested that DAZL enhanced protein synthesis via the 39UTR of dazl mRNA [15]. So we examined if dazl mRNA is also controlled by a combination of DAZL and miR-430. Injection experiments of the GFP mRNA fused with dazl 39UTR (GFP-dazl) showed that the 39UTR restricted DAZL expression to PGCs (Fig. 5A). Overexpression of DAZL enhanced GFP expression from GFP-dazl mRNA throughout the embryo (Fig. 5B). Moreover, the reporter mRNA was strongly stabilized in the presence of DAZL (Fig. S3). The dazl 39UTR contains a putative miR-430 target site (corresponding to the nucleoteides +813-835 relative to the stop codon), which has the octamer sequence complementary for the miR-430 seed sequence. As expected, GFP-dazl mRNA was subjected to deadenylation (Fig. 5C). A mutant mRNA, GFP-dazl DmiR-430 site, in which the octamer target site had been base-substituted, had a long poly(A) tail approximately 150 nt long (Fig. 5C). Overexpression of DAZL protein clearly induced the poly(A) elongation (up to 200-250 nt long) of both GFP-dazl and GFP-dazl DmiR-430 site mRNAs (Fig. 5C). Taken together, we concluded that DAZL also counteracts miR-430-mediated repression of dazl mRNA, by inducing polyadenylation. This suggests that DAZL promotes its own expression in PGCs, leading to the enhanced expression of the germline-specific proteins, such as TDRD7.

Discussion
In this study, we showed that DAZL binds tdrd7 and dazl mRNAs and counteracts the miRNA-mediated repession by inducing polyadenylation in zebrafish (Fig. 6). The DAZ family is evolutionarily conserved and expressed in germline cells. Members of the DAZ family may control germline gene expression by counteracting the repression mediated by miRNAs in various animals.
In Xenopus oocytes, it has been shown that DAZL activates the translation of target mRNAs with short poly(A) tails via the recruitment of PABPs [14]. Here we showed that DAZL induces polyadenylation of target mRNA, enhancing mRNA stability and/ or translatability (Fig. 6). Thus it raises the possibility that DAZL plays several distinct roles in the target mRNAs. Alternatively, it is possible that DAZL and PABP may cooperatively function to induce polyadenylation on tdrd7 and dazl mRNAs. Future studies will be necessary to clarify the mechanism of how DAZL induces polyadenylation of the germ plasm mRNAs.
In zebrafish and human germline cells, DND suppresses miR-430 function through blocking miRNA accessibility by binding to U-rich mRNA regions (URRs) [10]. The loss of DND function or its target sequences make nanos and tdrd7 39UTRs susceptible to miR-430-mediated repression in zebrafish. However, those 39UTRs appeared to be able to direct PGC-specific protein expression weakly even in the absence of DND function [10]. Moreover, DAZL overexpression strongly induces somatic protein expression from the GFP reporter mRNA fused with the tdrd7 39UTR lacking the DND-binding element, URR (our unpublished observation). Although we cannot exclude the possibility that DAZL can block the association of miRNA with target mRNAs, the induction of polyadenylation by DAZL strongly suggests that DAZL relieves miRNA-mediated repression in a distinct mode from DND. Interestingly, we found that DAZL is not sufficient to promote protein expression through the intact 39UTR of nanos1 mRNA (our unpublished observation). Thus, it is likely that DAZL and DND proteins function additively to exclude the miR-430 function on the various germline mRNAs. Loss of function experiments of DAZL will be important to further clarify the interplay of DAZL and DND during germ cell development.
It has been shown that an AU-rich RNA-binding protein HuR relieves CAT-1 mRNA from miR-122 repression upon stress in human liver cells [22]. TRIM-NHL proteins modulate miRNA activity [23,24]. In addition, some miRNA-mediated repression is relieved by synaptic stimulation of neuron cells [25,26]. Biogenesis of miRNA is also controlled [27,28,29]. A variety of RNA-binding proteins may regulate miRNA function in a spatiotemporal manner or in response to a specific signal during animal development.

mRNA synthesis and RNA injection
The GFP open reading frame (ORF) and 39UTR were cloned into the BamHI and XbaI sites of pCS2+ [30]. The resulting plasmid was digested with XhoI, whereas pCS2+ MT DAZL constructs containing Myc-tag [15] and pCS2+ DsRed [5] were digested with Asp718. Using these DNAs as templates, mRNAs were transcribed by the mMessage mMachine SP6 kit (Ambion). To prepare A-capped mRNA, the dinucleotide ApppG (Ambion) was added to the reaction mixture instead of m 7 GpppG [31]. RNA solution contained GFP and DsRed reporter mRNAs at 0.1 mg/ml each, with or without 0.2 mg/ml Myc-DAZL mRNA, was injected into the cytoplasm of one-cell-stage embryos with an IM300 Microinjector (Narishige).

RNase H-poly(A) assay
The assay was performed essentially as described [5]. Total mRNA was extracted from five embryos and then mixed with 25 pmol oligo DNA complementary to the 39 end of the GFP ORF sequence (59-CTCGACCCCGCCTGAC-39). The resultant fragment containing the tdrd7 39UTR was detected by Northern blotting.

Image acquisition
Fluorescent images were captured by a Carl Zeiss Ste REO Lumar V12 microscope and an Axio Cam MRm digital camera. To get images of in situ hybridization embryos, we used a Nikon SMZ1500 microscope and a Nikon DXM 1200F digital camera.

The measurement of fluorescence
The average pixel intensity of a rectangle above the yolk extension (596118 pixels, total 6962 pixels) was measured using Adobe Photoshop CS4 Extended ver. 11.0. Background signal was measured at dark regions next to the embryo. We measured the intensity of GFP and DsRed fluorescence in three embryos injected. For normalization, GFP signal was divided by DsRed signal. To calculate the fold change of GFP expression with DAZL overexpression, the average of GFP/DsRed ratio was normalized with that of embryos injected without any mRNA encoding DAZL (mock control).