Author Correction: The ABD on the nascent polypeptide and PH domain are required for the precise Anillin localization in Drosophila syncytial blastoderm

Targeting proteins to regions where they are required is essential for proper development of organisms. For achievement of this, subcellular mRNA localization is one of the critical mechanisms. Subcellular mRNA localization is an evolutionarily conserved phenomenon from E. coli to human and contributes to limiting the regions at which its products function and efficiently supplies substrates for protein translation. During early Drosophila embryogenesis, while 71% of the 3370 mRNAs analyzed have shown prominent subcellular localization, the underlying molecular mechanisms have not been elucidated. Here, we reveal that anillin mRNA, one of the localized mRNAs in early Drosophila embryo, localizes to the tip of the pseudo-cleavage furrow in the Drosophila syncytial blastoderm using in situ hybridization combined with immunohistochemistry. Localization analyses with transgenic fly lines carrying a series of deletion mRNAs indicate that this localization is dependent on its own nascent polypeptides including the actin binding domain (ABD). In addition to the mRNA localization, it is revealed that the pleckstrin homology (PH) domain of Anillin protein is also required for its proper localization. Thus, we indicate that the precise localization of Anillin protein is tightly regulated by the ABD on the nascent polypeptide and PH domain in the Drosophila syncytial blastoderm.

Targeting proteins to regions where they are required is essential for proper development of organisms. For achievement of this, subcellular mRNA localization is one of the critical mechanisms. Subcellular mRNA localization is an evolutionarily conserved phenomenon from E. coli to human and contributes to limiting the regions at which its products function and efficiently supplies substrates for protein translation. During early Drosophila embryogenesis, while 71% of the 3370 mRNAs analyzed have shown prominent subcellular localization, the underlying molecular mechanisms have not been elucidated. Here, we reveal that anillin mRNA, one of the localized mRNAs in early Drosophila embryo, localizes to the tip of the pseudo-cleavage furrow in the Drosophila syncytial blastoderm using in situ hybridization combined with immunohistochemistry. Localization analyses with transgenic fly lines carrying a series of deletion mRNAs indicate that this localization is dependent on its own nascent polypeptides including the actin binding domain (ABD). In addition to the mRNA localization, it is revealed that the pleckstrin homology (PH) domain of Anillin protein is also required for its proper localization. Thus, we indicate that the precise localization of Anillin protein is tightly regulated by the ABD on the nascent polypeptide and PH domain in the Drosophila syncytial blastoderm.
In early Drosophila embryogenesis, the first thirteen nuclear divisions occur without accompanying cytokinesis 1 . In the embryo between nuclear division cycle 10 and 14, which is called the syncytial blastoderm, thousands of nuclei at the embryo surface rapidly and synchronously divide four times. In order to prevent the collision of neighboring mitotic spindles, a transient membrane invagination called the pseudo-cleavage furrow (PCF) or metaphase furrow occurs between nuclei 2,3 . This furrow progressively ingresses from prophase to metaphase, and then starts to retract at anaphase (Fig. 1A-E) 4 . Most of the components in the PCF are shared with those of the cleavage furrow during cytokinesis [5][6][7][8][9][10] . The Anillin protein is one of the main components in the cleavage furrow and functions as a scaffold protein to maintain the contractile ring during cytokinesis and also localizes to the PCF in the syncytial blastoderm with strong accumulation at the tip of the PCF 7 . However, the mechanisms by which the Anillin protein is localized to the location at which it functions remain unclear.
Subcellular mRNA localization provides the appropriate amount of a protein at the region at which it is required. Since subcellularly localized mRNA was initially visualized in Ascidian embryos in 1983 11 , a large number of localized mRNAs have been reported in various species and types of cells (Holt and Bullock, 2009). The Drosophila embryo is an attractive model in analyses of subcellular mRNA localization, and has been utilized for more than three decades. A number of developmentally important genes have been identified and characterized in the Drosophila embryo, and, thus, the subcellular localization of their mRNAs may be directly linked to their functions, as is the case for bicoid and nanos mRNAs (Ephrussi et al., 1991;Gavis and Lehmann, 1992). Furthermore, a large-scale analysis of mRNA localization in the Drosophila early embryo by fluorescence in situ hybridization has been conducted and several types of prominent mRNA localization patterns have been reported, such as accumulation in the posterior yolk or pole plasm, apical localization above nuclei of the embryo surface, and nuclear division-related localization 12 .
In the present study, we focus on anillin mRNA, which shows a hexagonal pattern that is one of the distinguishing patterns in the syncytial blastoderm, and this category includes several mRNAs such as canoe, Diaphanous, Polychaetoid, and CG43427 12 . In order to investigate the molecular mechanisms underlying the subcellular localization of anillin mRNA in the syncytial blastoderm, we established transgenic fly lines expressing anillin mRNAs carrying a set of deletions and analyzed their localization using in situ hybridization analysis. In the present study, we reveal that anillin mRNA localized to the tip of the PCF in the syncytial blastoderm in its own nascent polypeptide-dependent manner. In addition to the mRNA localization, the pleckstrin homology domain of Anillin protein is required for its proper localization at the tip of the PCF in the syncytial blastoderm.

Results
anillin mRNA localizes to the tip of the PCF throughout the cell cycle in the syncytial blastoderm.
The diagnostic localization of the anillin mRNA like honeycomb-like pattern in the Drosophila syncytial blastoderm 12 closely resembles the pattern of the PCF, which is an incomplete and temporal membrane invagination 2,3,13 . In addition, anillin mRNA partially colocalizes with the Anillin protein in the plasma membrane, particularly at the tip of the PCF in the syncytial blastoderm ( Supplementary Fig. 1, arrowheads in lower panels) 7,12 . These findings strongly suggest that anillin mRNA localizes to the PCF. In order to investigate more precisely whether anillin mRNA localizes to the PCF, we performed in situ hybridization combined with immunostaining with anti-discs  10 . In the syncytial blastoderm, the cell cycle consists of interphase and a mitotic phase without gap phases and progresses quickly (Zalokar and Erk, 1976; Foe and Alberts, 1983). The results of double staining showed that anillin mRNA partially colocalized to the front edge of the dlg signal throughout the cell cycle in the syncytial blastoderm ( Fig. 1, arrowheads in the lower panels of A"-E").
The coding region of a splicing variant of anillin mRNA, anillin SV-A is necessary for its localization at the tip of the PCF. According to the database (Flybase: http://flybase.org/reports/FBgn0261385.html), three different splice variants (SV-A, B and C) of anillin mRNA have been predicted to exist ( Fig. 2A). In order to identify the PCF localization element in anillin mRNA, we initially investigated which SV of anillin mRNA was expressed and localized to the PCF in the syncytial blastoderm. We measured the expression levels of three SVs using RT-qPCR. Since it was not possible to detect SV-A or B independently ( Fig. 2A), we calculated the expression level of SV-A or B based on differences between the expression levels of SV-A, B or SV-B, C and SV-C, respectively. RT-qPCR revealed that the expression level of SV-A in the syncytial blastoderm was 42-fold higher than those of SV-B and C, which were close to the limit of detection by this method (Fig. 2B). In addition to the quantitative analysis, we examined the localization of each SV by in situ hybridization. We were unable to detect any signals of SV-C (Fig. 2C). Similar to the quantitative analysis, we were unable to find a probe with the ability to detect SV-A and B separately ( Fig. 2A). However, the signals observed appeared to represent SV-A signals because SV-A was dominantly expressed in the syncytial blastoderm (Fig. 2B). SV-C was expressed as weakly as SV-B and, thus, a signal for SV-C was not detected (Fig. 2C). These results indicate that most of the anillin mRNA signals at the tip of the PCF are derived from SV-A signals ( Fig. 2C and Supplementary Fig. 2A).
In order to identify the region in SV-A that is necessary for localization to the tip of the PCF, we established transgenic fly lines expressing a series of deletion derivatives of SV-A fused to DsRed-monomer (Fig. 3A). We performed in situ hybridization with the probe for DsRed-monomer to identify exogenous RNAs without detecting  Supplementary Fig. 2B). RT-qPCR analyses revealed that ΔCDS RNA was sufficiently expressed to be detectable even if it was not specifically localized because the expression level of ΔCDS RNA was found to be 4.0-fold higher than that of the PCF-localized FL (Fig. 3C). These results suggest that CDS is crucial for the furrow tip localization of anillin mRNA in the syncytial blastoderm. In order to further demonstrate the importance of the CDS region of anillin mRNA for localization to the tip of the PCF, we investigated the localization of GFP-fused anillin CDS 14 using in situ hybridization with the probe for GFP. We used the transgenic strain carrying GFP-fused SV-B (CDS), which is available from a stock center. CDS is derived from SV-B, which includes all SV-A CDS and additional 81-bp sequences ( Fig. 2A), and the SV-B protein has been reported to properly localize at the PCF in the syncytial blastoderm 14 . The results obtained showed that CDS mRNA localized to the tip of the PCF (Fig. 3B). Collectively, these results suggest that the CDS region of anillin mRNA is necessary and sufficient for localization to the tip of the PCF in the syncytial blastoderm (Fig. 3C).
In order to further narrow down the region required for the furrow tip localization of anillin CDS, we divided SV-A CDS into three parts ( Supplementary Fig. 3A) and generated transgenic fly lines expressing each divided CDS. None of the SV-A carrying each of the three parts of CDS mRNAs localized to the PCF (Supplementary Figs 3C-E and 2C), despite all of them being sufficiently expressed for detection by RT-qPCR ( Supplementary  Fig. 3F). These results indicate that none of the segments for CDS mRNA were sufficient for localization to the furrow tip.
However, the proteins encoded by each divided CDS segment showed different localization patterns, such as on the plasma membrane, in the nucleus, or colocalization with Peanut ( Supplementary Fig. 3C"-E"). These localization patterns appeared to depend on the domain feature carried in each CDS segment. There are well-known domains in each CDS segment of Anillin, such as the myosin binding domain (MBD) and actin binding domain anillin mRNA localizes to the tip of the PCF in a manner that is dependent on its nascent polypeptides.
In divided CDS segment localization analyses, we were unable to identify any obvious RNA localization element within anillin CDS. Previous studies reported that the mRNAs encoding transmembrane or secreted proteins localized to the endoplasmic reticulum (ER) in a manner that was mediated by their own nascent polypeptides 17 . Furthermore, as is the case with these mRNAs, some of the mRNAs encoding cytoplasmic or nuclear proteins, such as X-box binding protein 1, Diaphanous 1, and ABP140 mRNAs, also localized subcellularly depending on their nascent polypeptides [18][19][20] . In order to investigate whether the localization of anillin mRNA depends on its own nascent polypeptides, we established transgenic fly lines harboring full-length anillin mRNA in which a stable stem-loop structure (∆G = −42.10 kcal/mol) was inserted between 5′UTR and CDS (SL-FL) to inhibit protein translation (Fig. 4A). This stem-loop structure is commonly used to prevent translation by impairing ribosome loading onto CDS 21,22 . We found that while the expression levels of SL-FL were 2-fold higher level than those of FL mRNA (Fig. 4B), no predicted protein was detected in the SL-FL fly early embryo (Fig. 4C). Under these conditions, SL-FL did not localize to the PCF (Fig. 4D). These results indicate that the nascent polypeptides translated from anillin mRNA are important for its mRNA localization. However, since CDS 1, 2, and 3 are translated into a part of the Anillin protein, translation itself may not have been sufficient for the proper localization of anillin mRNA ( Supplementary Figs 2C and 3). Therefore, we examined whether the specific domain of the Anillin protein is required for the localization of anillin mRNA to the PCF. Anillin contains three known domains: the MBD, ABD, and PH domain (Fig. 5A) 7,14,23 . In order to investigate the role of each domain region of the Anillin protein in its mRNA localization, we generated transgenic fly lines expressing each domain-deleted Anillin, namely, ΔMBD, ΔABD, and ΔPH (Fig. 5A). In addition to these deletions, we investigated the effects of base substitutions corresponding to each amino acid in the PH domain (T1087I, G1094E) that are suggested to be responsible for PI(4,5)P 2 binding 16 (Fig. 5A). PI(4,5)P 2 has been predicted to function in cell cleavage during cytokinesis 16 . Domain-deleted or base-substituted mRNAs, except for ΔABD, localized to the tip of the PCF (Fig. 5B, arrowheads and Supplementary Fig. 2D). Based on RT-qPCR data, the expression level of ΔABD appeared to be sufficient for detection even if its mRNA was not specifically localized to the PCF (Fig. 5D). These results demonstrate that ABD is necessary for the proper localization of anillin mRNA.
In ΔMBD or T1087I base-substituted transgene-expressing embryos, those mRNAs and proteins were precisely localized (Fig. 5B,C). However, while ΔPH and G1094E base-substituted mRNAs localized to the PCF (Fig. 5B), the localization of their proteins to the PCF was very faint (Fig. 5C). The results of a Western blot analysis and its quantified data revealed that the protein level of ΔABD, ΔPH, or G1094E was similar to that of ΔMBD  Fig. 4). These results suggest that not only ABD, but also the PH domain is required for the proper localization of the Anillin protein in the syncytial blastoderm, as reported for the localization of the human Anillin protein to the cleavage furrow of mammalian cultured cells 23 .

Discussion
It is well known that subcellular mRNA localization is one of the most important processes to express proper protein function in organisms. However, in contrast to the rapidly increasing number of studies describing subcellularly localized mRNAs, our knowledge on the molecular mechanisms underlying mRNA localization is still limited. In an attempt to reduce this gap, we herein focused on anillin mRNA, which shows a characteristic hexagonal localization pattern in the Drosophila syncytial blastoderm 12 , and revealed that anillin mRNA localizes to the tip of the PCF and interior embryo throughout the cell cycle in the syncytial blastoderm (Lower panels of Fig. 1A'-E").
The localization patterns of a series of deficient mRNAs indicated that the CDS of anillin mRNA is necessary and sufficient for its PCF localization ( Fig. 3 and Supplementary Fig. 2B). The in situ hybridization analysis of translationally inactive mRNA-expressing embryos revealed that the PCF localization of anillin mRNA was achieved in its own translation-dependent manner (Fig. 4). Translation-dependent mRNA localization is widely observed in several different types of cells and organisms. In most cases of co-translational mRNA subcellular localization, the translation of these mRNAs is transiently paused 18,20,24,25 . Regarding anillin mRNA, it is important to note that ribosome footprint data with Drosophila early embryos (RPFDB: http://sysbio.gzzoc.com/rpfdb/index.html) 26 match the predicted translational pause or delay occurring downstream of the sequences encoding ABD. Consistent with this finding, 7 successive prolines are downstream of the ABD. Since the polyproline stretch over 6 prolines in the nascent polypeptide was shown to strongly repress its own translation under eIF5A-absent conditions 27 , the translation of anillin mRNA may be paused at the polyproline stretch in anillin mRNA.
We revealed that nascent polypeptides including ABD of SV-A are necessary for the localization of its mRNA to the tip of the PCF, and SV-B mRNA was localized to the furrow tip even when an additional 27 amino acids were inserted into its ABD (CDS in Fig. 3B). A previous study reported that some actin-binding proteins expressed in the syncytial blastoderm were not localized to the PCF 27 . These findings suggest that actin-binding activity may not be sufficient to make its mRNA localize to the tip of the PCF. However, CDS 1 mRNA does not localize to the furrow tip even though it encodes ABD (Supplementary Fig. 3C). In S. cerevisiae, ABP140 mRNA localizes to the distal pole of the mother cell through its own nascent polypeptide, which mainly includes ABD 27 . However, not only ABD, but also additional polypeptides without sequence specificities are needed for the proper localization of its mRNA. As is the case for ABP140 mRNA, additional polypeptides of the Anillin protein besides ABD may be necessary for the localization of anillin mRNA to the tip of the PCF. ΔPH mRNA, which is truncated but carries a longer additional sequence than CDS 1 mRNA, localized to the furrow tip, whereas CDS 1 mRNA did not (Fig. 5B).
The localization patterns of CDS 1 and 3 revealed that the CDS 1 protein including ABD colocalized with actin at the apical region of the PCF (Supplementary Fig. 3C'-C"'), while the CDS 3 protein may recruit Peanut and narrowly localized to the tip of the PCF, whereas CDS 3 mRNA did not ( Supplementary Fig. 3E-E"'). However, the localization of both of these proteins was not same as that of FL or CDS ( Fig. 5C and Supplementary Fig. 3B' ,E'). The PCF may form in embryos expressing CDS 1 or CDS 3, and endogenous Anillin protein was also expressed and localized normally (data not shown). Nevertheless, actin and Peanut exhibited abnormal localization patterns from those in wild-type embryos ( Supplementary Fig. 3C",E") 15,28 . These results suggest that the expression of CDS 1 or CDS 3 exerts a dominant negative effect on the function of endogenous Anillin protein. Furthermore, these results indicate that the precise localization of the Anillin protein in the syncytial blastoderm requires multiple domains, similar to mammalian Anillin (Oegema et al. 23 ). The localization patterns of ΔABD and ΔPH protein are also consistent with the requirement of multiple domains for the precise Anillin protein localization in the syncytial blastoderm (Fig. 5C). We also revealed that G1094E base-substituted protein shows impaired ability to localize to the tip of the PCF, while T1087I base-substitution does not affect on the Anillin protein localization (Fig. 5C). Although previous study reported that both the glycine at position 1094 and threonine at position 1087 are required for the binding to phosphatidylinositol phosphate lipids including PI(4,5)P 2 or liposome 16 , the glycine at position 1094 might therefore have another function other than the binding to PIPs that may be important for the Anillin protein localization.
In order to localize the Anillin protein precisely in the syncytial blastoderm, we revealed that it is necessary to localize its mRNA to the tip of the PCF in its nascent polypeptide-dependent manner. However, while anillin mRNA in the cellular blastoderm following the syncytial blastoderm does not localize subcellularly to the tip of the invagination, Anillin protein can localize to the furrow tip (Field and Alberts, 1995) (data not shown). These results suggest that other trans-factor(s) is (are) necessary for the localization of anillin protein to the furrow tip in the later cellular blastoderm.
Fixation of Drosophila embryos. Embryos were collected for 2.5 hours. They were washed with H 2 O and dechorionated with 10% sodium hypochlorite solution. After washing with H 2 O, embryos were fixed in 1:1 4% paraformaldehyde in phosphate-buffered saline (PBS): heptane with shaking. After washing three times with methanol, embryos were stored at −30 °C and then used for in situ hybridization or immunohistochemical staining within one month. RNA probe production. PCR was performed to prepare the template for transcription. The PCR products obtained were purified with the High Pure PCR Product Purification Kit (Sigma-Aldrich; 11 732 676 001). Anti-sense RNA probes labeled with digoxigenin (DIG) were synthesized according to the protocol of Lécuyer, E. et al. 12 .
In situ hybridization and immunohistochemistry. The conditions and methods for fluorescence in situ hybridization and fluorescence in situ hybridization combined with immunostaining are described in detail by Lécuyer, E. et al. 12,15,28  RT-qPCR. RNAs were isolated using TRIzol reagent (Thermo Fisher Scientific). cDNA was synthesized using the PrimeScript RT reagent kit (TaKaRa). PCR was performed with SYBR Premix Ex Taq II in the CFX96 Touch TM Real-Time PCR Detection System (BIO-RAD), and data were analyzed with a standard curve-based method calculated with CFX Manager TM software. The specificities of primers were tested with melt curves created by CFX Manager TM software and agarose gel electrophoresis of the amplified fragments. The following primers were used for RT-qPCR: DsRed-monomer (fw 5′-ACGGCACCTTCATCTACAAG-3′/ rev 5′-CGGCAGTCTTCTTCTGCATTA-3′), GAPDH (fw 5′-GGAGCCACCAGACGAAAC-3′/rev 5′-CGAACACAGACGAAGGG-3′).