RNA polymerase II subunit D is essential for zebrafish development

DNA-directed RNA polymerase II (pol II) is composed of ten core and two dissociable subunits. The dissociable subcomplex is a heterodimer of Rpb4/Polr2d and Rpb7/Polr2g, which are encoded by RPB4/polr2d and RPB7/polr2g genes, respectively. Functional studies of Rpb4/Polr2d in yeast have revealed that Rpb4 plays a role primarily in pol II-mediated RNA synthesis and partly in various mRNA regulations including pre-mRNA splicing, nuclear export of mRNAs and decay of mRNAs. Although Rpb4 is evolutionally highly conserved from yeast to human, it is dispensable for survival in budding yeast S. cerevisiae, whereas it was indispensable for survival in fission yeast S. pombe, slime molds and fruit fly. To elucidate whether Rpb4/Polr2d is necessary for development and survival of vertebrate animals, we generated polr2d-deficient zebrafish. The polr2d mutant embryos exhibited progressive delay of somitogenesis at the onset of 11 h postfertilization (hpf). Mutant embryos then showed increased cell death at 15 hpf, displayed hypoplasia such as small eye and cardiac edema by 48 hpf and prematurely died by 60 hpf. In accordance with these developmental defects, our RT-qPCR revealed that expression of housekeeping and zygotic genes was diminished in mutants. Collectively, we conclude that Rpb4/Polr2d is indispensable for vertebrate development.

To analyze physiologically-relevant function of Polr2d in vertebrate development, we employed CRISPR/Cas9 method in zebrafish and generated polr2d-deficient zebrafish. The polr2d mutants initially showed retardation of somitogenesis, eventually exhibited increased cell death and finally displayed severe hypoplasia, resulting in premature death. Our RNA analysis revealed that mRNA levels were significantly affected in polr2d mutant embryos. Taken together, we demonstrated that Rpb4/Polr2d contributes to mRNA regulation of housekeeping and zygotic genes and is essential for vertebrate development.

Results
Generation of polr2d-deficient zebrafish. To investigate polr2d gene in zebrafish, we cloned full length polr2d cDNA and identified 141 amino acid sequence. An amino acid alignment of Polr2d/Rpb4 protein from many organisms showed that Polr2d is conserved from yeast to human, especially among vertebrate animals (Fig. 1a). We designed a spacer and a PAM for CRISPR/Cas9-mediated genome editing in the conserved helix-1 domain of polr2d gene, injected guide RNA along with Cas9 mRNA into fertilized zebrafish embryos and then obtained an 8-bp deletion allele of polr2d (Fig. 1b). This deletion caused a frameshift in the helix-1 domain, thereafter adding 30 amino acids followed by a stop codon. Genomic PCR enabled reliable genotyping of wildtype (+ / +), heterozygous mutant (+/−) and homozygous mutant (−/−) embryos (Fig. 1c).
To assess spatial and temporal expression of polr2d, we performed whole mount in situ hybridization using full length polr2d cRNA as a probe. At 24 and 48 hpf, polr2d was ubiquitously expressed with intense signals in the brain and the spinal cord (Fig. 1d,e) in accordance with a publicly available expression data 17,18 .
Heterozygous mutant embryos showed no developmental defects and grew up to become normal adults. By crossing heterozygous mutant carrier fish, we obtained homozygous mutant embryos at a ratio of one quarter (27.1%, 26/96). Hereafter, we simply refer to homozygous mutants as mutants. The mutants showed apparent delay of embryogenesis at 24 hpf and then exhibited developmental hypoplasia such as small eye and cardiac edema at 48 hpf ( Fig. 1f-i). All of the mutant embryos died between 48 and 60 hpf. These hypoplasia at 48 hpf were cancelled by injecting wild-type polr2d mRNA into mutants at 1-2 cell stage (81% rescue, 17/21), thereby confirming that developmental defects in mutants were attributable to the loss of Polr2d function. These results indicate that polr2d is essential for zebrafish embryogenesis.
Delay of development in polr2d mutant zebrafish. To detail the delay of embryogenesis in polr2d mutants, we assessed precise developmental stages of mutant embryos by counting somite numbers. In standard zebrafish development at 28.5 °C, somite formation is initiated at the onset of 10 hpf with the rate of 3 somites per hour. After 12 hpf (6 somite), the rate becomes 2 somites per hour and somitogenesis continues up to 24 hpf (30 somites). We obtained embryos by crossing heterozygous mutant carrier fish, counted the number of somite in each embryo every 2 h from 9 to 21 hpf and checked genotype for each. At 9 hpf when somitogenesis is not yet initiated, wild-type and mutant embryos were indistinguishable in appearance (Fig. 2a,b). Although wild-type and mutant embryos looked similar at 11 hpf, the number of somite in mutants was significantly lower than that in wild-type (wild-type: 3.5 ± 0.1; mutant: 2.9 ± 0.2; P < 0.001; Fig. 2c Fig. 2e-n). The pace of developmental in heterozygous mutant embryos was indistinguishable from that in wild-type embryos (data not shown). These results indicate that progressive retardation of embryogenesis appears at the onset of 11 hpf in polr2d mutants. We also noticed that head region in mutants became less transparent after 19 hpf. This typically suggests that many neuronal cells undergo cell death in polr2d mutants.
Increased cell death in polr2d mutants. To address whether cell death is accelerated in mutants, we counted the number of cells that undergo cell death during embryogenesis. By immersing live zebrafish  www.nature.com/scientificreports/ embryos in acridine orange-containing solution, we could fluorescently label and count dead cells at each time point 19 . Samples were then subjected to genotyping. At 13 hpf, the numbers of dead cells in the head (anterior to midbrain/hindbrain boundary) and the anterior trunk (somite 1-4) regions were comparable between wild-type (head 52.8 ± 5.0; anterior trunk 21.2 ± 3.7; Fig. 3a,b) and mutants (head 65.6 ± 11.9, P = 0.30; anterior trunk 25.0 ± 4.1, P = 0.51; Fig Fig. 3k,l) was larger than that in wild-type (head 54.3 ± 10.7, P < 0.01; anterior trunk 41.9 ± 8.8, P < 0.01; Fig. 3i,j) at both head and anterior trunk regions. The cell death in heterozygous mutant embryos was similarly seen as in wild-type embryos (data not shown). These results indicate that cell death is accelerated in polr2d mutants during embryogenesis.
Gene expression is affected in polr2d mutants. Since Polr2d/Rpb4 plays important roles in Pol IIdependent synthesis of mRNA in yeast, we assessed mRNA levels of several genes in wild-type and polr2d mutants. We collected total RNAs and genomic DNAs from individual embryos of heterozygous carrier crosses and performed RT-qPCR and genotyping. We then compared mRNA levels relative to 18S rRNA levels, since 18S rRNA is transcribed by pol I, which does not require Polr2d/Rpb4 for rRNA synthesis, and is suitable for the internal standard in this case 20 . The mRNA levels of polr2d in mutants were diminished at 9, 24 and 48 hpf compared to those in wild-type and heterozygous mutant embryos (Fig. 4a). Expression of housekeeping genes such as β-actin (actb1) and ribosomal protein L13a (rpl13a) in mutants were initially comparable to those in wildtype siblings at 9 hpf (Fig. 4b,c). But they eventually decreased compared to those in wild-type and heterozygous mutant embryos at 24 and 48 hpf. Similarly, expression of zygotic genes such as glycine receptor α4a subunit (glra4a) and keratin 4 (krt4) in mutants was unaffected by 9 hpf but then reduced in comparison to wild-type and heterozygous mutant embryos by 48 hpf (Fig. 4d, e). The levels of maternal transcripts such as DEAD-box RNA helicase vasa (vasa) and RNA-binding protein dead end (dnd) were comparable between wild-type siblings and mutant embryos at 9 and 24 hpf (Fig. 4f,g). In both cases, mRNA levels in mutants did not become lower and rather became higher than those in wild-type and heterozygous mutant embryos at 48 hpf. These results confirm that Polr2d/Rpb4 is involved in Pol II-mediated transcription of housekeeping and zygotic genes during embryogenesis.

Discussion
In this study, we generated polr2d knockout vertebrate animals for the first time and demonstrated that polr2d mutant zebrafish showed initially the delay of somite formation, eventually the increase of cell death and finally developmental hypoplasia and premature death. Expression of housekeeping and zygotic genes was severely affected in mutants. Taken together, we conclude that Polr2d/Rpb4, which is encoded by polr2d/RPB4 gene, is essential for mRNA regulation and embryogenesis in vertebrates.

Loss of polr2d and lethality.
To investigate the physiological necessity of Polr2d/Rpb4, disruption of polr2d/RPB4 gene has been assayed in several organisms ranging from yeast to invertebrate animals. All of the polr2d/RPB4-deficient organisms were lethal, except for RPB4 mutant S. cerevisiae that could grow at normal temperature 9 . In vertebrates, our polr2d mutant zebrafish were embryonic lethal. A previous study of retrovirusmediated mutagenesis showed that zebrafish embryos homozygously carrying a retroviral insertion in an intron of polr2d gene (hi2718b) exhibited the increase of neuronal cell death 21,22 . Since this apoptotic phenotype in hi2718b embryos is similar to that in our polr2d mutants, hi2718b defects are attributable to the polr2d deficiency, supporting our notion that polr2d is indispensable for zebrafish development. In cultured human cells, cell proliferation was severely impaired by shRNA-mediated POLR2D/RPB4 knockdown 14 . Collectively, we can conclude that POLR2D/RPB4 gene is essential for survival in eukaryotes with an exception of the S. cerevisiae case. Interestingly, only S. cerevisiae have 11-amino acids insertion and 60-amino acids insertion before helix-1 domain and in between helix-1 and helix-2 domains, respectively 9 . Although the significance of these S. cerevisiae-specific insertion is unclear, this organism might have divided Rpb4 function and distributed it to Rpb4 and the other proteins.
mRNA defects in zebrafish polr2d mutants. In the earliest stages of animal development, a fertilized egg initiates cell cleavages. Transcription of zygotic genes is quiescent immediately after fertilization and then is activated in a phase called maternal-to-zygotic transition 23,24 . In zebrafish, this transition occurs at 2-4 hpf 25,26 .
Although we anticipated the impairment of transcription during maternal-to-zygotic transition in mutants, our RT-qPCR analyses revealed that mRNA levels in mutants were not globally affected at 9 hpf. We assume that polr2d is a housekeeping gene that is maternally transcribed in oocytes. Therefore, polr2d mutants likely have maternal reserves of normal polr2d mRNAs that enabled the synthesis of normal Polr2d protein during early embryogenesis. As a consequence, mRNA levels in mutant embryos were mostly unaffected by 9 hpf. But after www.nature.com/scientificreports/ this stage, mutants could no longer sustain Pol II-dependent transcription due to the running out of Polr2d protein, leading to progressive defects of embryogenesis. Despite that mRNA levels of housekeeping and zygotic genes were maintained in mutants at 9 hpf, polr2d mRNA was significantly reduced at the same time. This specific elimination of polr2d mRNA was probably caused not by the impairment of Pol II-dependent transcription but rather by nonsense-mediated mRNA decay. The nonsense-mediated mRNA decay is an active degradation system of abnormal mRNAs that harbor premature nonsense codons 27 . Since polr2d gene has four exons and the frame-shift mutation generating a downstream nonsense codon was located in the second exon, mutant polr2d mRNAs can be the target of nonsense-mediated decay. This accounts for the early specific loss of polr2d mRNA in mutants.
Progressive developmental defects in polr2d mutants. We could observe a significant delay of development in polr2d mutants at the onset of 11 hpf. The degree of the delay was only 0.6 somite that corresponds to 12 min at 11 hpf. The degree of the delay became larger during somite formation, which took place in an anterior to posterior wave (13 hpf: 1.1 somite = 33 min; 15 hpf: 1.4 somite = 42 min; 17 hpf: 3.1 somite = 93 min; 19 h: 3.9 somite = 117 min; 21 hpf: 4.8 somite = 144 min). This progressive delay of developmental was followed by the increase of cell death at the onset of 15 hpf. The cell death accelerated progressively in an anterior to posterior direction, because cell death is triggered by the failure of neuronal differentiation, which proceeds in the same direction along with body extension 28,29 . Then, mutant embryos displayed reduced transparency and developmental hypoplasia such as small eye and cardiac edema at 24 and 48 hpf. Small eye and cardiac edema are typical phenotypes of gross developmental defects in zebrafish 30,31 . Collectively, polr2d mutation caused sequential developmental defects: the delay of development, the increase of cell death and developmental hypoplasia.

Animals. Zebrafish (Danio rerio) AB line was purchased from Zebrafish International Resource Center
(https ://zebra fish.org/home/guide .php). They were reared and maintained at 28.5 °C under a 14 h light and 10 h dark photoperiod and fed twice a day as in the regular care and maintenance protocol 32 . The polr2d mutant allele that harbors 8-bp deletion was generated in this study and designated as agu8 by the Zebrafish Nomenclature Committee in ZFIN (https ://zfin.org). This allele is available from National BioResource Project Zebrafish (https ://shige n.nig.ac.jp/zebra /). All of the polr2d homozygous mutants used in this study was obtained by crossing heterozygous mutant carrier fish that was back-crossed to wild-type zebrafish AB line at least twice.
Cloning of full length zebrafish polr2d cDnA. Total RNA was extracted from embryos of zebrafish AB line using Sepasol RNA II Super (Nacalai Tesque). The Oligo(dT)18 Primer (Thermo Fisher Scientific), SuperScript IV Reverse Transcriptase (Thermo Fisher Scientific), polr2d cloning primers and Phusion DNA polymerase (Thermo Fisher Scientific) were used for RT-PCR as described previously 33 . The polr2d cDNA was cloned into pCR4Blunt-TOPO vector (Thermo Fisher Scientific) and pCS2 + expression vector. The polr2d cloning primer sequences are listed in Table 1. mRnA rescue. Capped, full-length zebrafish polr2d mRNA was synthesized from pCS2 + zPolr2d plasmids using the mMessage mMachine SP6 Transcription Kit (Thermo Fisher Scientific) as described previously 34 . The mRNA was microinjected into 1-2 cell stage embryos and the embryos were subjected for observation at 48 hpf as described previously 35 . Generation of polr2d mutants by CRISPR/Cas9. CRISPR/Cas9 targets were selected in exons 2 of zebrafish polr2d gene using a web-based free software CRISPRscan (https ://www.crisp rscan .org) 36 . To generate a gRNA construct, polr2d CRISPR forward and reverse oligonucleotides (Table 1) were annealed and subcloned into pDR274 37 . The gRNAs were synthesized from Hind III-digested pDR274-zPolr2d plasmids using MEGAshortscript T7 Transcription Kit (Thermo Fisher Scientific). The Cas9 RNA was synthesized from pCS2 + hSpCas9 38 , which contains the human codon-optimized S. pyogenes Cas9 cDNA 39 , using mMESSAGE mMACHINE SP6 Transcription Kit (Thermo Fisher Scientific). A solution containing 20 ng/μl gRNA and 200 ng/μl Cas9 RNA was injected into zebrafish embryos at the one-cell stage. To detect insertion/deletion mutations, the target regions were amplified by PCR and subjected to heteroduplex mobility analysis 40 . The 8-bp deletion was confirmed by sequencing after subcloning of the target region.
Genotyping. The  Reverse transcription quantitative PCR (RT-qPCR). The RT-qPCR was done as described previously 41 .
Total RNA was extracted from zebrafish embryos using Sepasol RNA II Super (Nacalai Tesque). For reverse transcription, SuperScript IV Reverse Transcriptase (Thermo Fisher Scientific) and Random Primer Mix (New England Labs), the latter contains both random 6-mer and oligo-dT oligonucleotides, were used to synthesize cDNAs. The qPCR was carried out using gene-specific primers with a KAPA SYBR Fast qPCR Kit (Kapa Biosystems) on QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific). The 18S rRNA was used as an www.nature.com/scientificreports/ internal control. Each qPCR amplification was performed in triplicate and the data were analyzed using the 2 −ΔΔCt program as described previously 42 . The gene-specific primer sequences are listed in Table 1.
In situ hybridization. In situ hybridization of wholemount zebrafish embryos with a digoxigenin-labeled antisense RNA probe was performed as described previously 43 . A polr2d probe covering the full coding sequence was used as a probe.
Counting the number of somite. Zebrafish embryos obtained by crossing heterozygous mutant carrier fish were incubated in embryo medium in individual wells of 48-well plate (As One) at 28.5 °C. At each time point after fertilization, images of whole embryos were captured under a stereo microscope SZX16 (Olympus) as described previously 44 . The number of somite in each embryo was visually counted. Each embryo was subjected to genotyping later.
Acridine orange labeling of dead cells. Live zebrafish embryos were immersed in embryo medium containing 5 μg/ml acridine orange (Sigma-Aldrich) at 28.5 °C for 60 min and washed in embryo medium for three times. Fluorescent images of labeled embryos were captured using a confocal microscopy TCS SP5 II (Leica) at each time point. The number of fluorescence-positive cells were counted.
Statistics. Quantitative data were given as mean ± SEM. All error bars in graphs represent the SEM values. Statistical significance was determined using the two-tailed Student's t test. Asterisks show P value ranges; *P < 0.05; **P < 0.01; ***P < 0.001. The sample numbers are indicated in Figure legends.