Smyd5 plays pivotal roles in both primitive and definitive hematopoiesis during zebrafish embryogenesis

Methylation of histone tails plays a pivotal role in the regulation of a wide range of biological processes. SET and MYND domain-containing protein (SMYD) is a methyltransferase, five family members of which have been identified in humans. SMYD1, SMYD2, SMYD3, and SMYD4 have been found to play critical roles in carcinogenesis and/or the development of heart and skeletal muscle. However, the physiological functions of SMYD5 remain unknown. To investigate the function of Smyd5 in vivo, zebrafish were utilised as a model system. We first examined smyd5 expression patterns in developing zebrafish embryos. Smyd5 transcripts were abundantly expressed at early developmental stages and then gradually decreased. Smyd5 was expressed in all adult tissues examined. Loss-of-function analysis of Smyd5 was then performed in zebrafish embryos using smyd5 morpholino oligonucleotide (MO). Embryos injected with smyd5-MO showed normal gross morphological development, including of heart and skeletal muscle. However, increased expression of both primitive and definitive hematopoietic markers, including pu.1, mpx, l-plastin, and cmyb, were observed. These phenotypes of smyd5-MO zebrafish embryos were also observed when we introduced mutations in smyd5 gene with the CRISPR/Cas9 system. As the expression of myeloid markers was elevated in smyd5 loss-of-function zebrafish, we propose that Smyd5 plays critical roles in hematopoiesis.

Histone modification constitutes one epigenetic mechanism that plays a critical role in the dynamic regulation of chromatin structure and gene expression, and several enzymes that catalyse histone modifications have been identified 1 . Histone lysine residue methylation contributes both positively and negatively to gene transcription, and a family of histone lysine methyltransferase containing the evolutionally conserved catalytic SET domain has been reported 2 . More than 60 SET domain-containing proteins have been identified in the mammalian genome; among them, the SMYD family, which is comprised of five members in humans, SMYD1-5, has been described 3,4 . Members of the SMYD family have been implicated in diverse biological functions in skeletal and cardiac muscle development as well as in cancer progression [5][6][7][8][9] . SMYD1, SMYD2, and SMYD3 show histone H3K4 methyltransferase activity 7,10,11 , as SMYD2 and SMYD3 methylate histones H3K36me2 and K5me1, respectively 12,13 . In addition, SMYD2 mediates the methylation of lysine residues of non-histone proteins such as tumour suppressor p53, retinoblastoma (RB), heat shock protein 90 (HSP90), and poly ADP-ribose polymerase (PARP1) [14][15][16][17] . Moreover, SMYD3 also catalyses non-histone proteins, such as vascular endothelial growth factor receptor (VEGFR) and mitogen-activated protein kinase 3/2 (MAPK3/K2) 18,19 . Unlike other family members, SMYD5 does not contain a C-terminal tetratricopeptide repeat (TPR) domain 20 . SMYD5 trimethylates H4K20 and negatively regulates inflammatory response genes 21 . However, the physiological function of SMYD5 remains largely unknown.
Scientific RepoRts | 6:29157 | DOI: 10.1038/srep29157 Zebrafish (Danio rerio) provide an excellent model system with which to study the biological processes of vertebrates. Similar to mammalian models, zebrafish hematopoiesis consists of both primitive and definitive waves 22 . The primitive hematopoiesis wave occurs in the intermediate cell mass (ICM). Blood cell circulation begins around 24 hours post-fertilisation (hpf), at which time, hematopoiesis shifts from ICM to the posterior blood island (PBI) 22 . The definitive wave occurs in the aorta-gonadmesonephros (AGM) around 30 hpf 23 . There are three hematopoietic stem cell (HSC) migration and colonisation events beginning around 48 hpf. AGM progenitor cells migrate to the caudal hematopoietic tissue (CHT), an intermediate site of hematopoiesis. Next, lymphocyte differentiation occurs in the thymus. Finally, kidney marrow produces all hematopoietic cell types, which corresponds to bone marrow hematopoiesis in mammals 24 .
Five members of the Smyd family have been identified in zebrafish 25 . In the work described herein, we aimed to determine the physiological function of Smyd5 in zebrafish embryogenesis. Using a morpholino oligonucleotide (MO)-mediated knockdown and CRISPR/Cas9 knockout approach to smyd5 during zebrafish embryonic development, we found that Smyd5 plays a crucial role in hematopoiesis. These results indicate that Smyd5 represents an epigenetic regulator of hematopoiesis during zebrafish embryogenesis.

Results
Expression profile of smyd5 in zebrafish embryogenesis and adult tissues. We first examined the expression pattern of smyd5 during zebrafish embryogenesis by quantitative reverse transcription polymerase chain reaction (qRT-PCR) using RNA extracted from embryos at different developmental stages. smyd5 was abundantly expressed at early developmental stages but decreased slightly when embryos proceeded in development (Fig. 1A). To examine the spatial and temporal expression patterns of smyd5 during embryogenesis, a whole-mount in situ hybridisation (WISH) assay was performed. Smyd5 expression was strongly detected from 0.25 to 3 hpf in whole embryos, but it was only weakly observed at 12 hpf. At 24 and 36 hpf, signals were observed only around the eye with stronger intensities at 24 hpf than 36 hpf (Fig. 1B). The distribution of smyd5 transcripts was also examined in adult tissues by qRT-PCR. smyd5 transcripts were observed in all tissues examined, but the expression levels were different among tissues (i.e., high in the ovary but relatively weak in the skin, gut, heart, and skeletal muscle) (Fig. 1C).
Smyd5 is dispensable for heart and skeletal muscle development. To characterise the physiological functions of smyd5 during embryogenesis, we used smyd5-MOs to knock down the expression of smyd5 in zebrafish embryos. We designed two MOs, smyd5-MO1 and smyd5-MO2, which target different regions of the 5′untranslated region (UTR) of smyd5. The efficiency with which smyd5-MO1 and smyd5-MO2 suppressed smyd5 expression was tested by co-injection of an expression plasmid encoding the 5′ -UTR of smyd5; this was followed by enhanced green fluorescent protein (EGFP) and control MO (Con-MO), MO1, or MO2. MO and EGFP plasmids were co-injected into zebrafish embryos at the one-to two-cell stages, and EGFP expression was observed at 24 hpf. Co-injection of MO1 but not of Con-MO inhibited the EGFP signal at 24 hpf ( Fig. 2A). Similar results were obtained from the injection of MO2 (data not shown), suggesting that MO1 and MO2 efficiently knock down smyd5.
We then injected smyd5-MO into embryos and examined gross morphological phenotypes. At 24, 48, and 72 hpf, embryos showed no abnormality when observed under binoculate (Fig. 2B). Moreover, zebrafish injected with MO2 did not show visible abnormalities of gross morphogenesis until at least 72 hpf (data not shown).
SMYD family genes play crucial roles in the development of heart and skeletal muscle 10,26,27 . Therefore, we next examined whether ablation of smyd5 zebrafish leads to defects in heart and skeletal muscle development by WISH using myogenic and cardiac makers 8 . GATA-binding protein 5 (gata5) and cardiac myosin light chain2 (cmlc2) are markers for early cardiogenesis and cardiac chamber, respectively 28,29 . Myogenic differentiation (myod), myogenic factor 5 (myf5), and myogenin (myog) are myogenic regulatory factors, and muscle creatine kinase (mck) is a marker for terminally differentiated skeletal muscle 30,31 . At 12 hpf, the expression of myf5, myod, myog, and gata5 in embryos injected with smyd5 MO1 was indistinguishable from that in Con-MO-injected embryos or embryos that did not receive an injection (Fig. 3A). These results suggest that Smyd5 is not involved in the early stages of cardiogenesis and myogenesis. Moreover, the intensity of signals and expression patterns of myod, myog, and mck in morphants did not differ from that of control embryos at 24 hpf (Fig. 3B). No difference in the expression of cmlc2 at 12 hpf was observed between smyd5-MO injected morphants and controls at 48 hpf (Fig. 3C). In addition, the structure of sarcomere of heart, fast-and-slow skeletal muscle was indistinguishable between smyd5-MO1 injectedembryos and controls at 48 hpf (Fig. 3D). Taken together, these results indicate that Smyd5 is not responsible for the development of cardiac and skeletal muscle.
Smyd5 is required for primitive myelopoiesis. With the aim of defining the function of Smyd5 in zebrafish, we then focused on the development of hematopoietic cells, which derive from the mesoderm as heart and skeletal muscle. Similar to the case in other vertebrates, zebrafish hematopoiesis consists of two stages, primitive and definitive hematopoiesis 22 . To assess the role of smyd5 in primitive hematopoiesis, WISH was performed to examine hematopoietic cell markers. gata1 and pu.1 are markers for erythroid and myeloid progenitors, respectively 32,33 . Hemoglobin beta embryonic 1 (hbbe1), lymphocyte cytosolic protein 1 (l-plastin), and myeloperoxidase (mpx) are markers for erythrocyte, macrophage, and granulocytes, respectively 33,34 . Smyd5-MO1 was injected into zebrafish embryos at the two-to four-cell stages, and WISH was performed at each indicated developmental stage. At 24 hpf, expression patterns and signal intensities of gata1 in embryos injected with smyd5-MO1 were comparable to those in control embryos injected with Con-MO or those that did not receive an injection (Fig. 4A). Similarly, at 26 hpf, no difference in hbbe1 expression was observed between embryos injected with smyd5-MO1 and control embryos injected with Con-MO or those that did not receive an injection (Fig. 4B). These data indicate that primitive erythropoiesis is not affected by smyd5 knockdown. In contrast, although the expression pattern of pu.1 was comparable to that of control embryos, at 24 hpf, pu.1 signal intensity was elevated in embryos injected with smyd5-MO1 (Fig. 4C). In addition, mpx and l-plastin expression was increased without perturbation in the expression patterns of embryos injected with smyd5 MO1 (Fig. 4D,E). Essentially the same results were obtained for smyd5-MO2 (Fig. 4A-E).
To test whether the observed results were specific to smyd5, smyd5-MO1, and smyd5 mRNA were co-injected into embryos at the two-to four-cell stages. Enhanced expression of cmyb, mpx, and l-plastin, through suppression of smyd5 expression by smyd5-MO1, was reversed by injection of smyd5 mRNA (Supplemental Fig. 1). These results indicate that Smyd5 negatively regulates the expression of genes related to primitive myelopoiesis in zebrafish.
Smyd5 is required for definitive myelopoiesis. We then investigated the role of smyd5 in definitive hematopoiesis by examining the expression of genes related to definitive hematopoiesis. v-myb avian myeloblastosis viral oncogene homolog (c-myb), recombination activating gene 1 (rag1), hbbe1, l-plastin, and mpx were examined. By 30 hpf, c-myb was expressed in definitive HSCs of AGM 35 . Rag1 was expressed in lymphocytic lineage 36 . Smyd5-MO1 was injected into embryos at the two-to four-cell stages, and WISH was performed using embryos at 30 hpf. c-myb expression was increased in embryos injected with smyd5-MO1 relative to that of controls, indicating that Smyd5 positively affects definitive hematopoiesis (Fig. 5). In zebrafish, myeloid cells begin to be observed at CHT around 72 hpf, and definitive erythrocytes are detected at PBI at 96 hpf 34,37 . At 72 hpf, l-plastin-and mpx-expressing cells were observed in CHT in both smyd5-MO1-injected and control embryos, and signal intensities were much stronger in smyd5-MO1-injected embryos than in controls (Fig. 5B,C). However, at 96 hpf, the hbbe1 and rag1 expression of embryos injected with smyd5-MO1 was indistinguishable from that of control embryos (Fig. 5D,E), suggesting that Smyd5 does not affect erythropoiesis and lymphopoiesis. Injection of smyd5-MO2 resulted in essentially the same phenotypes as that of smyd5-MO1 (Fig. 5A-E). Co-injection of  Fig. 2). Taken together, these findings suggest that SMYD5 negatively regulates definitive myelopoiesis.
To validate the Smyd5 loss-of-function phenotypes, we used CRISPR/Cas9-mediated genome editing system to generate F0 mutants. The guide RNA (gRNA) against smyd5 genomic (coding) region was designed and injected in combination with Cas9 mRNA into embryos. We confirmed mutations in the target region with heteroduplex mobility assay (HMA) and sequencing analysis. HMA revealed that mutagenesis rates reached 40% (2 of 5 embryos showed only heteroduplex DNA: Fig. 6A). The sequencing analysis showed that all the examined sequences had small insertion and/or deletion near the smyd5 target loci (Fig. 6B). We then examined the expression pattern of genes, which were modified in smyd5-MO injected embryos, by whole mount in situ hybridization in embryos injected with smyd5-gRNA and Cas9 mRNA (smyd5-KO F0). Expression of pu.1 was increased in smyd5-KO F0 embryos than to no-injection controls at 24 hpf (6/16, Fig. 6C). In addition, mpx and l-plastin signal intensity was elevated in smyd5-KO F0 embryos at 28 hpf (6/18 and 5/15, respectively; Fig. 6C). cmyb signal intensity was elevated in smyd5-KO F0 embryos at 30 hpf (8/20, Fig. 6C). At 72 hpf, mpx and l-plastin expression was also increased in smyd5-KO F0 embryos (6/17 and 5/14, respectively; Fig. 6D). Taken together, the phenotype observed with Smyd5 knock-down embryos was validated in embryos bearing mutations by CRISPR/ Cas9-mediated genome editing system. Therefore, we concluded that Smyd5 regulates primitive and definitive myelopoiesis.

Discussion
In this report, we showed that zebrafish smyd5 plays pivotal roles in primitive and definitive hematopoiesis; however, we did not observe an apparent phenotype of the cardiac system or skeletal muscle associated with smyd5 downregulation during zebrafish development. Previous studies have shown that SMYD1-4 are involved in the development of heart and skeletal muscle in both vertebrates and invertebrates 5 . Deletion of Smyd1 caused hypoplasia of the right ventricle in mice through disrupted maturation of ventricular cardiomyocytes. Knockdown of smyd1 also led to malfunction of skeletal and cardiac muscles in zebrafish. Knockdown of smyd2 in zebrafish impaired cardiac and skeletal muscle development 26,27 . We also reported that Smyd3 plays a critical role in cardiogenesis and myogenesis in zebrafish 38 . In addition, muscle-specific depletion of Drosophila Smyd4 led to the failure of eclosion, resulting in late pupal death 39 . Therefore, it is conceivable that SMYD proteins have an evolutionally conserved function in the development of cardiac and skeletal muscle. Therefore, we examined the expression patterns of various genes specific to cardiac and myogenic markers, and the structure of sarcomere of heart, fast-and-slow skeletal muscle, but no abnormality was observed in smyd5-knockdown zebrafish embryos. These results indicate that Smyd5 plays physiological functions that are distinct from those played by the other Smyds. This notion is supported by previous reports describing the roles of SMYDs in cancer progression. All SMYD family members, except SMYD5, have been reported to be involved in the proliferation and survival of a variety of tumors [6][7][8] . However, SMYD5 was identified to be critical in cancer metastasis in breast cancer cells during lung colonisation 9 .
The structure of SMYD5 also differs from that of other members of the SMYD family. Most SMYD proteins, except SMYD5, possess at least one C-terminal TPR domain, which is critical for its interaction with other proteins 20 . However, the TPR is not present in Smyd5, and SMYD1, SMYD2, and SMYD3 interact with HSP90 5-11 . HSP90 is a homodimeric, ubiquitous, and essential chaperone involved in a variety of biological processes, including myogenesis and cardiogeneis 5,40 . Therefore, a lack of heart and skeletal muscle differentiation may, at least partly, be attributed to the lack of the TPR domain in SMYD5. A proteomics approach to identifying the binding partners of SMYD2, SMYD3, and SMYD5 revealed that SMYD2 and SMYD3 share many interactors, including DNA sliding clamp proliferating cell nuclear antigen (PCNA), replication factors, and mini-chromosome maintenance proteins 11 . These proteins interact with DNA polymerase during the DNA damage response 11,20 , suggesting that SMYD2 and SMYD3 share common physiological roles. On the other hand, some proteins involved in DNA repair and chromatin maintenance during the cell cycle have been identified as common interactors of SMYD2, SMYD3, and SMYD5 20 , reflecting both the common and distinct activities of SMYD members through interacting proteins.
We found that Smyd5 plays critical roles in both primitive and definitive myelopoiesis in zebrafish. SMYD5 interacts with nucleophosmin 1 (NPM1) 20 , which regulates myelopoiesis 41 . In addition, NPM1 plays an important role in the regulation of a number of hematopoietic stem cells 42 . NPM1 mutations are commonly observed (~30%) in acute myeloid leukemia (AML), suggesting that SMYD5 participates in myeloid cell differentiation/ proliferation through its interaction with NPM1. The role of SMYDs in the hematopoietic system was also observed for other members. SMYD2 regulates the differentiation of regulatory T cells (Tregs) and Th-17 cells 43 , whereas SMYD3 controls inducible Tregs 42 . In both cases, the methyltransferase activities of SMYD for specific target gene loci were suggested to play critical roles 43,44 . Based on the current study, we do not have sufficient evidence to prove that SMYD5 exerts its biological function through enzymatic activity, but this is a critical issue that should be addressed in the future. It has been reported that SMYD5 trimethylates histone H4 lysine 20 through its association with nuclear receptor corepressor 1 (NCoR1) complexes, which repress the expression of toll-like receptor 4 (TLR4) target genes 21 . TLR4 signalling promotes granulocyte and macrophage development 45,46 . Tlr4 and NCoR are conserved in zebrafish 47,48 ; based on the current observations, we speculate that enhanced expression of myeloid markers by smyd5 suppression may be due to suppression of the TLR4 signalling pathway. Interestingly, knockdown of SMYD5 or SMYD3 results in reduced activation of TLR4, but the H4K20me3 mark in TLR4-responsive promoters is largely dependent on SMYD5 21 . These results suggest that SMYD5 and SMYD3 act on different substrates/genomic locations through alternative protein complexes 21 .
Taken together, our current results reveal the important roles of Smyd5 in hematopoiesis and indicate that this activity is specific to Smyd5. These findings will aid in the understanding of the epigenetic regulation underlying hematopoiesis. Future studies will be required to reveal the molecular mechanisms of hematopoiesis through Smyd5.

Material and Methods
Maintenance of zebrafish. Zebrafish (Danio rerio) were purchased from a local pet shop and maintained under a 14-h day/10-h night cycle at 28.5˚C. Fertilised eggs were obtained by mating adult fish from outbred colonies soon after the light was turned on. Embryos were staged according to hours post-fertilisation (hpf) and morphological criteria 49 . Quantitative reverse-transcription polymerase chain reaction (qRT-PCR) analysis. Total RNA was extracted from embryos and adult tissues using TRIzol ® reagent (Invitrogen). cDNA was synthesised using Superscript III reverse transcriptase (Invitrogen) and oligo (dT)15 primers (Invitrogen). Real-time PCR was performed using SYBR Green technology with sets of primers (smyd5: forward primer, 5′ -ACTTCCTCCTTCACCTCACG-3′ , reverse primer, 5′ -GTCCAGGTAGCTGATGCAAAT-3′ , ef1a : forward primer, 5′ -CCTTCGTCCCAATTTCAGG-3′ , reverse primer, 5′ -CCTTGAACCAGCCCATGTT-3′ ) for smyd5 on StepOnePlus (Life Technologies). Amounts of transcripts were determined by relative standard curve method, and ef1a was used as internal control.
Microinjection of morpholino-oligonucleotides (MOs), guide RNA and mRNA. All antisense morpholino-oligonucleotides (MOs) were designed and supplied by Gene Tools LCC. Two different MOs, smyd5-MO1 (5′ -CATTTTAACCTCTAACCTCTCAACC) and smyd5-MO2 (5′ -CAACCTGACCAATG AGTGTGCGAGA-3′ ), were designed to hybridise to sequences in the 5′ -UTR of smyd5. A standard control MO (Con-MO) available from the same manufacturer was used as a control and had no effect on embryonic development under our experimental conditions. MOs were diluted to 1 ng/nl with 1x Danieau buffer, and approximately 3nl was injected into fertilised zebrafish eggs at the one-to two-cell stages using a microinjector (IM-300; Narishige). Smyd5-EGFP plasmid was constructed as follows. A fragment containing the 5′ -UTR of smyd5 containing MO target sequences was obtained by RT-PCR using the following primers: 5′ -CCGGAATTCTGTTAAAAAAAGAAAGGCGATC-3′ and 5′ -CCGCTCGAGGTCATCTACGGGGGCCGC -3′ . The fragment was then cloned into pCS2 + EGFP plasmid and subjected to RNA synthesis. Zebrafish smyd5 (NM_001004614) cDNA, including the open reading frame (ORF) of smyd5, was purchased from GENEWIZ, Inc. Zebrafish smyd5 cDNA was subcloned into pCS2 + to synthesise mRNA. mRNAs were synthesised using m7G(5′ ) PPP(5′ ) G (NEB) and SP6 RNA polymerase (Takara). To confirm the effects of smyd5-MO knockdown, zebrafish embryos were injected with 100 pg smyd5-EGFP mRNA and 300 pg smyd5-MO1 or smyd5-MO2. Rescue experiments were performed by injection with 300 pg smyd5 mRNAs and smyd5-MO1. A guide RNA (gRNA) was generated as described 50 . Cloning of a gRNA template was initiated by annealing two oligonucleotides (Forward, TAGGCATTCCACAAGAACTGAG and Reverse, AAACCTCAGTTCTTGTGGAATG), and double strand oligonucleotides were ligated into BsmBI site of the pT7-gRNA vector (Addgene). To generate the gRNA, template DNA was linearised with BamHI, and purified by phenol/chloroform extraction. The gRNA was transcribed in vitro by using MEGA short script T7 kit (Thermo Fisher Scientific). Subsequently, 100 pg of gRNA and 150 pg of Cas9 mRNA (SBI, CAS500A-1) were injected into one-cell stage embryos. Embryos were anesthetised on ice and observed under a macro-zoom microscope (MVX10, Olympus).
Heteroduplex mobility assay (HMA) and sequencing analysis. To prepare the genomic DNA, embryos at 24 hpf were incubated in 30 μ l of 25 mM NaOH, 0.2 mM EDTA at 95 °C for 15 min. Then, 3 μ l of 40 mM Tris-HCl (pH8.0) was added to the resultant solution. Genomic fragments at the target sites were amplified by PCR using the following primers: 5′ -TCAGGGCAAAGGATTATTCG -3′ and 5′ -AAAAAGAACCCAAAATCACCAACA-3′ . PCR amplicons were electrophoresed on a 15%polyacrylamide gel containing 10% glycerol. The PCR products were sub-cloned into the pTAC-2 vector (BioDynamics Laboratoey Inc). The plasmid DNAs containing the genomic fragments were prepared from individual colonies, and then, random sequencing was performed.
Electron microscopies. Electron microscopy analysis was carried out as previously described with modifications 51 . Sodium cacodylate buffer was used as glutaraldehyde buffer, and the samples were sectioned using Ultracut N and analyzed by HITACHI H-7500 electron microscope.