Characterization of N-Myristoyltransferases in Vertebrate Embryos by Using Zebrafish: Appearance of Low Molecular Weight N-Myristoyltransferase 1 in Early Development

Characterization of N-Myristoyltransferases in Vertebrate Embryos by Using Zebrafish: Appearance of Low Molecular Weight N-Myristoyltransferase 1 in Early Development Hajime Nakatani1,2, Natsuki Nukada2, Eriko Avşar-Ban2, Hisayoshi Ishikawa2, Shin’ichi Akiyama1,2 and Yutaka Tamaru2* 1Wakayama Industrial Promotion Foundation, 60 Ogura, Wakayama-shi, Japan 2Department of Life Sciences, Graduate School of Bioresources, Mie University, Japan


Introduction
The covalent lipid modifications on proteins are an important aspect of its functional regulation. The acylation by long fatty acid chain on amino acid residues in proteins alters biochemical characteristic of proteins, that is, conjugation of hydrophobic group heighten its affinity to the lipid membrane structures such as organelles or plasma membrane [1,2]. The lipid modified proteins targeting to plasma membrane appears their functions by interacting to other proteins nearly located the membrane. For example, recruiting of proteins involving specific intracellular signaling on cell membrane is often achieved by lipid modifications, and the effect enables rapid and effective signal transduction depending on the external environment. Due to this function, the cells finally achieve their proper cellular functions [1,2].
N-myristoyltransferases (NMTs) are group of enzyme that mediates protein amino terminal myristoylation. The NMTs recognize amino terminal consensus amino acid sequences like Gly-X-X-X-Thr/Ser and catalyze the covalent transfer of myristic acid from myristoyl-CoA to amino group of glycine residue. Orthologues are widely recorded in the database of eukaryote genome, found in the organisms including human [16], mouse [17], rat [18], bovine [19] and yeast [20]. The wide distribution of NMTs in eucaryotes suggests its importance for basic cellular function. There exist at least two genetically distinct forms but structurally similar NMTs, NMT1 and NMT2, in mammalian [16,21]. Although those two NMTs seem to exist in wide variety of vertebrates, its biochemical characteristics or functions are not much revealed. In case of mammalian, there is one report that NMT1 but not NMT2 is important for the development of mouse. However, the detail roles of NMT1during embryogenesis did not addressed in that study, because the NMT1 knockout mice died between embryonic days 3.5 and 7.5 [22].
For the first step to explore the biological characteristics of NMTs during vertebrate's embryogenesis, we concentrated on zebrafish as a useful model of higher organisms. This model animal has been used commonly in many researches for developmental biology. In this study, we first confirmed the existence of zebrafish NMTs (zNMTs) which has high homology to mammalian's NMTs, and examined their gene expression during embryogenesis. We next analyzed an effect of inhibition of zNMT1 on their embryogenesis. Finally, we analyzed expression of zNMT proteins at several embryonic stages. The results suggest that zNMTs like mammalian NMT are necessary for proper embryonic development of zebrafish. It was also suggested that novel low molecular weight zNMT1 appears specifically during embryogenesis. We think that those findings would be useful information for understanding NMT's function during development of higher organisms.

Zebrafish care and collection of eggs
Adult zebrafish were purchase from local provider and maintained in aquarium under artificial condition (28ºC, light period for 14 hrs and dark period for 10 hrs). For collection of fertilized eggs, two female and three male fish were mated. Spawning was induced by switching from dark period to light period.

Cloning of zNMTs and plasmid construction for their expression in Escherichia coli
Total RNA from the zebrafish embryo at 6 hpf was reversetranscribed with oligo-dT primer in a volume of 20 µl. Aliquots of 1 µl were used for amplification of zNMT1 or zNMT2 cDNA. Polymerase chain reaction was carried out with following forward and reveres primers: 5'-CCACCATGGCGGATGAG-3' and 5'-TCACTGCAGAACCAATCC-3' for zNMT1, and 5'-ATGGCGGAGGACAGCGAGTCCGC-3' and 5'-TTACTGTAAAACAAGGCCAAC-3' for zNMT2, respectively. Amplification was done by 35 cycles of denaturation at 95ºC for 30 sec, annealing at 50ºC for 30 sec, and extension at 72ºC for 90 sec. Amplified DNA fragments were purified by agarose gel electrophoresis and cloned into pT7Blue-T vector (Novagen, San Diego CA). The constructs (pT7blue-zNMT1 and -zNMT2) were analyzed by DNA sequencing and confirmed its identity to the sequences on public database. For expression in Escherichia coli (E. coli), the zNMT genes were amplified with primer set as described in Table 1 by polymerase chain reaction with template plasmids (pT7blue-zNMT1 and -zNMT2), and was sub-cloned into pT7Blue-T vector. Generating pT7Blue-Sal I-zNMT1-Not I, pT7Blue-Eco RI-zNMT1s (full, deltaC1, deltaC2, deltaN1, deltaN2)-Not I and pT7Blue-Eco RI-zNMT2s (full, deltaC1, deltaC2, deltaN1, deltaN2)-Not I constructs were digested with two restriction enzymes, Sal I and Not I or Eco RI and Not I (TaKaRa Otsu, Japan). Resulting DNA fragments were separated and purified by agarose gel electrophoresis and the fragments were sub-cloned between the Sal I and Not I sites of pET 50b (+) vector (Novagen, San Diego CA) or the Eco RI and Not I sites of pGEX-6p-1 (GE Healthcare UK Ltd) vector, generated pET50-zNMT1, pGEX-zNMT1s and pGEX-zNMT2s constructs. Correct insertion of those DNA fragments were confirmed by DNA sequencing.

Expression of recombinant zNMTs
E. coli BL21 strain was transformed with pET50b-zNMT1, pGEX-zNMT1 and pGEX-zNMT2 vectors and cultured on the LB plate supplemented with 50 µg/ml of kanamycin (Km) or 50 µg/ml of ampicillin (Amp) at 37ºC overnight. Several colonies were obtained, and one of them was transferred to 4 ml of LB broth with 50 µg/ml of Km or Amp. After incubation at 37ºC for 16 hrs, aliquots of 20 µl was transferred to 2 ml of 2×YT broth, further cultured approximately for 2 hrs at 37ºC until OD 600nm reached 0.6. The expression of the recombinant protein (NUS-zNMT1, GST-zNMT1s and GST-zNMT2s) in BL21 was induced by addition of 100 µM of IPTG followed by culture for 16 hrs at 20ºC or 37ºC. E.coli were then collected from the medium by centrifugation at 5000×g for 10 minutes, washed once with 1×PBS and bacterial cells were collected again by centrifugation. The target protein was extracted from the cells by ultra-sonication in 50 µl of 1% Triton-X 100 in PBS or 2×NMT reaction buffer [50 mM Tris-HCL (pH 7.5), 0.5 mM EDTA, 0.45 mM 2-mercaptoethanol, 1%TritonX-100] or direct addition of 5×SDS-PAGE loading buffer [0.1 M Tris-HCl (pH 6.8), 5% SDS, 30% glycerol, 0.06% BPB, 5% 2-mercaptoethanol]. The protein extract was then centrifuged at 15000×g for 15 min at 4ºC, and separated into soluble and insoluble fractions. Each fraction was used for further analysis.

Purification of recombinant zNMT1 and preparation of the polyclonal antibodies against zNMT1
Antibodies against zNMT1 were obtained by immunization with purified recombinant NUS-zNMT1 fusion protein. The recombinant NUS-zNMT1 fusion protein was expressed in E. coli BL21 strain as described above, and collected as insoluble fraction. NUS-zNMT1 fusion was extracted from inclusion body with 3 M urea supplemented with 1% 2-mercaptoethanol at 4°C. NUS-zNMT1 protein was purified by Ni affinity chromatography with HisTrap TM HP column (GE healthcare Amersham Place, UK). Purified NUS-zNMT1 including its degraded peptides were collected and concentrated with Amicon R Ultra-15 centrifugal filter devices (Ultracel-30k) (Millipore) until protein concentration of the sample reached to approximately 1 mg/ml. Immunization of rabbits with purified NUS-NMT1 was requested to Kerry (Wakayama Japan). As for titer check of anti-zNMT antibodies, the recombinant GST-zNMT1s and GST-zNMT2s were used. GST-zNMT1s and GST-zNMT2s were expressed as above and extracted from insoluble fraction with 5×SDS-PAGE sample buffer. The samples were accessed by SDS-PAGE, followed by immunoblotting with anti-zNMT antibodies.

Detection of NMT activity by MS analysis
The total volume of soluble fraction of E.coli protein extracts in 2×NMT reaction buffer was adjusted to 100 µl with water, then myristoyl-CoA Li salt (Sigma) and substrate peptide (GARASVLSKbiotin: MW1113.604275, synthesized by Invitrogen) were added to the reaction samples at the final concentration of 180 µM and 90 µM, respectively. After incubation at 30 º C for 1 hour, the reacted peptide were separated and collected with Amicon R Ultrfree R -MC centrifugal filter devices 5000 NMWL filter unit (Millipore). The peptides were then absorbed and desalted with ZipTip C18 column (Millipore). The column was washed 10 times with 0.1% TFA, finally peptides were eluted and directly spotted onto MALDI-Plate with 1-2 µl of 2.5 mg/ ml CHCA in 70% AC/0.1% TFA. MS spectra were acquired using a 4700 Proteomic analyzer (Applied Biosystems, Foster City, CA, USA) in positive ion reflector mode over the 700-4000 m/z mass range, with 6000 laser shots per spot and external calibration with standard peptide [Bradykinin, Angiotensin II, P14R, ACTH (18-39) and Insulin B chain]. MS/MS spectra were acquired in CID mode and externally calibrated with ACTH (18-39) degraded peptide.

Construction of the expression plasmid for myc-and His-tagged zebrafish NMT1
The nmt1a gene was amplified from pT7blue-zNMT1 template by polymerase chain reaction with forward and reveres primer set, EcoRINMTs (5'-CGAGAATTCCACCATGGCGGATGAG-3') and Not INMTas (5'-ATAGCGGCCGCCTGCAGAACCAATCC-3'). The amplicon was purified and digested with Eco RI and Not I restriction enzymes, sub-cloned into pEF1-myc-his C vector (Invitrogen) which was opened by same restriction enzymes, pEF1-zNMT1myc-his construct was generated. Then, Xenopus laevis ef1alpha promoter was amplified from pXI-EGFP [23] template by polymerase chain reaction with forward and reveres primer set, MluIpxiproF (5'-GCATACGCGTCTCGAGCAGGG-3') and SpeIpxiproR (5'-GCATACTAGTGGATCCGTCGAGG-3'). The amplicon was purified and digested with Mlu I and Spe I restriction enzymes, subcloned into pEF1-zNMT1-myc-his and pEF1-myc-his C vector which were opened by same restriction enzymes, pXI pro-zNMT1-myc-his pEF and pXI pro-myc-his pEF construct were generated.

Extraction of Proteins from Zebrafish Embryos
Zebrafish embryos (30 embryos) were collected and yolk was removed by pipetting and washing in cold PBS supplemented with protease inhibitors (Nacalai Tesque Kyoto, Japan) and 0.05% Tween 20. The embryos were then lysed with 10 µl/10 embryos of SDS-PAGE sample buffer or 1% Triton-X 100 in PBS supplemented with proteinase inhibitor cocktail. After ultra-sonication, the insoluble materials were removed by centrifugation. Resultant supernatants were retained as protein extract.

Expression and Purification of zNMT1-myc-his Protein
The pXI pro-zNMT1-myc-his pEF and pXI pro-myc-his pEF constructs were injected into embryos at one cell stage, the embryos were incubated for 6 hours after fertilization. Protein was extracted from 60 embryos, the extract was then mixed with Ni-Sepharose beads (GE healthcare) and incubated for 15 minutes at room temperature with stirring. The beads were then collected by centrifugation at 3000×g for 3 minutes, washed three times with 30 mM imidazole/PBS solution (pH 7.0). Finally, the proteins bound to the beads were eluted with 250 mM imidazole/PBS solution (pH 7.0). The samples were analyzed by SDS-PAGE followed by immunoblotting with anti-myc antibody.

SDS-PAGE, CBB staining and immunoblotting
The protein extracts were boiled with 2.5-5×SDS-PAGE loading buffer for 5 minutes. The samples were then separated on the 10-12.5% polyacrylamide gel with Tris-glycine SDS buffer (25 mM Tris, 192 mM glycine, 0.1% SDS). For CBB staining, the gels were then stained with Coomassie R Brilliant Blue R-250 (CBB R-250 Wako Japan) for several minutes, removed excess CBB with decolorization buffer (25% methanol, 10% Acetic acid in water) until separated proteins were detected.
For immunoblotting, separated proteins by SDS-PAGE were transferred to PVDF membrane. The membranes was reacted with indicated antibodies as described previously [24]. For reaction with primary antibodies, anti-zNMT antibodies (No.1 and No.2) were diluted 1:1000 in blocking solution, respectively and used. Peroxidase conjugated anti-rabbit IgG (GE healthcare) was used as secondary antibody. Bound antibodies were detected with DAB substrate or ECL plus detection reagents (GE healthcare) and chemiluminescence was captured by Light Capture II cooled CCD camera system (ATTO Tokyo, Japan).

Challenging with 2-hydoroxymyristic acid
Approximately 10 fertilized eggs were distributed in 24 well plate. 20 μM 2-hydoroxymyristic acid (MP biomidicals, LLC) or myristic acid (Sigma-Aldrich) was added to each well at several time points. The embryos were incubated at 27ºC until 24 hpf and living and dead embryos were counted under microscope.

Morpholino Experiments
Antisense morpholino oligo against zNMT1 was purchased from Gene Tools LIC (Philomath, OR). The MO sequences of zNMT1 and control were 5'-tgctgtctcattctcatccgccatg-3' and 5'-gtaccgcctactcttactctgtcgt-3', respectively. MOs at a concentration of 50 ng/μl were injected into embryos at the one-to two-cell stage.

RNA Blotting
The NMT1 specific sense and anti-sense digoxigenin labeled RNA probes were prepared by in vitro transcription kid (DIG RNA labeling kid, Roche Diagnostics, Basel, Switzerland). The pT7blue vectors encoding 410-975 or 1212-1477 nucleotides region of zNMT1 cDNA and 177-443 or 912-1200 nucleotides region of zNMT2 cDNA were used as template for in vitro transcription. Total RNA was prepared from zebrafish embryos at 6 hpf and aliquots of 30 μg were dissolved in 50% formamide, denatured at 75°C for 10 minutes. After immediate cooling on ice, RNA was separated by electrophoresis on 1% agarose gel containing 2% formaldehyde in MOPS buffer [20 mM MOPS, 5 mM sodium acetate 660 mM formaldehyde (pH 7.0)] . The separated RNA was transferred to nylon membrane (Hybound-N GE healthcare) by capillary blotting with 10×SSC buffer (Sigma) for 19 hrs. The membrane with RNA was then backed at 80°C for 2hrs, then prehybridization was performed in hybridization buffer [750 mM NaCl, 20 mM Tris-HCl (pH 7.5), 2.5 mM EDTA, 50% formamide, 5×Denhardt solution (Wako Osaka, Japan) ] at 70°C for 2 hrs. The membrane was then hybridized with zNMT1 and zNMT2 specific sense or anti-sense RNA probe in hybridization buffer at 70°C for 19 hrs. The membrane was washed 3 times with 1×SSC buffer for 10 minutes, blocked with 2×SSC buffer containing 2% BSA and 0.05% Triton-X 100. Alkaline phosphatase conjugated anti-digoxigenin antibody (Roche Diagnostics) was diluted 1:10000 in blocking solution and added to the membrane, incubated at room temperature for 1 hour. After wash 3 times with 2×SSC containing 0.05% Triton-X 100, the membrane was incubated with BM-purple AP substrate (Roche Diagnostics) at 37°C until zNMT1 and zNMT2 RNA was detected.

In silico analysis shows existence of zebrafish N-myristoyltransferases (zNMTs) with high homology to mammalian's NMTs
Knock out of mouse N-myristoyltransferase 1 (NMT1) gene resulted in embryonic lethal [22], suggesting importance of NMT1 for development of mammalian. However, due to the embryogenesis of mammalian occurs in uterus, detail observation and analysis of NMTs in embryos is difficult. Thus, we have concentrated on a model animal, zebrafish, which can easily observe and analyze their embryonic development.
When we study an organism as a model of mammalian, most important thing, however, is whether the organism has a similar gene to mammalian. To identify the zebrafish N-myristoyltransferase (zNMT) which has homology to mammalian's one, we first searched NMTs on several databases, and found three N-myristoyltransferases (NP_001018316, NP_001156321 and NP_001186683). NP_001018316 (N-myristoyltransferase 1a) and NP_001186683 (N-myristoyltransferase 2) proteins were predicted to be orthologues of human NMT1 and NMT2, respectively. On the other hand, NP_001156321 (N-myristoyltransferase 1b) does not exist in mammalian, its orthologues proteins only exist in fish and invertebrates. We next curried out an alignment studies of the amino acid sequences of NP_001018316 (zNMT1), NP_001186683 (zNMT2) and other NMTs orthologues (human, mouse, fruit fly and yeast) which were found in protein data base. The result showed that zNMT1 and zNMT2 had high homology to human and mouse NMT1 or NMT2, respectively and also has moderate and low homology to fruit fly and yeast NMT (Figure 1). Phylogenic tree drown by identity of amino acid sequences showed that zNMT1 and zNMT2 were closely related protein to mammalian's NMT1 and NMT2, respectively ( Figure 2). Those data indicate that zebrafish can be a good tool for analyzing NMTs substitute for mammalian.

Gene expression of zNMTs in the embryos at early developmental stage
To investigate the expression of zNMT1 and zNMT2 gene on zebrafish development, we performed RT-PCR analysis and RNA blotting. In RT-PCR analysis, the expression of zNMT1 and zNMT2 was observed even in the embryos at 4 hours post-fertilization (4 hpf), both expression continued to at least 72 hpf ( Figure 3A). More detail analysis of zNMTs' expression by RT-PCR showed that expression of zNMT1 and zNMT2 was observed even after 2 hpf and the expression was kept during early development. In initial stages (1-3 hpf) nmt2 expression seems to be more evident than that of nmt1a (Supplemental Figure 1). RNA blotting with RNA from 6 hpf embryos showed that zNMT1 and zNMT2 RNA was detected as single band, suggested that both zNMT1 and zNMT2 RNA did not have splicing variants. The signal of zNMT2 RNA was weaker than that of zNMT1 RNA ( Figure 3B). These data suggest that expression of zNMTs starts initial developmental stage. Therefore, zNMTs, like mammalian's one, are suggested to play an important role in early stage of embryogenesis.

Inhibition of zNMT1 causes development arrest and decreases viability of embryos in early developmental stages
To examine a function of zNMTs during embryogenesis, we first analyzed the effects of NMT inhibitor, 2-hydroxymyristic acid (2-OHMyr) on development. The 2-OHMyr or myristic acid (Myr) was challenged to the embryos in several early developmental stages and the number of dead embryos per total embryos was analyzed at 24 hpf. The macroscopic observation of embryos treated with those fatty acids at 6 hpf revealed that most of the embryos were killed by 2-OHMyr treatment (Figure 4). The viability of 2-OHMyr challenged embryos was significantly decreased compared with that of Myr challenged one (Table 2). Especially, the challenge before 10 hpf increased the number of death, significantly (Table 2). We next performed inhibition of nmt1a mRNA translation with morpholino antisense oligo (MO) against zNMT1. As shown in Figure 5, The development of the embryos injected nmt1a (MO) was arrested at dome or early epiboly stage compared with no-injection or control MO injection. These data suggests that zNMT1a is essential for development of zebrafish; other NMTs (zNMT2 or zNMT1b) could not rescue the function of zNMT1a.

Recombinant zNMT1 exhibited myristoyltransferase activity
For the purpose of confirming enzymatic activity of nmt1a gene product (zNMT1), we next carried out production of recombinant zNMT1 in E. coli. As shown in Supplemental Figure 2, NUS-zNMT1 fusion protein (about 120 kDa) was induced in E. coli with IPTG. The NUS-NMT1 partially existed in soluble fraction when induced at 25ºC for 6 hrs. Up to 50% of recombinant zNMT1 were produced as soluble protein when induced at 20ºC O/N (data not shown).
We next attempted to detect the NMT1 activity in unpurified bacterial lysate ( Figure 6A). Peptide substrate (GARASVLSK-biotin), E.coli lysate and myristoyl-CoA were mixed and reacted at 30ºC for 1 hour. Reacted peptide was then analyzed by MS spectrum. Mass spectrum of substrate peptide reacted with bacterial lysate containing NUS tag showed one major peak (m/z 1114.6617) derived from substrate peptide ( Figure 6A). On the other hand, MS of reacted peptide with the lysate containing NUS-zNMT1 fusion protein contained another extra peak (m/z 1324.8723) corresponding to the MS of myristoyl-GARASVLSK-biotin peptide. Amino acid sequences of these peaks were further analyzed by MS/MS analysis and confirmed they were derived from substrate peptide (data not shown).
Purification of soluble zNMT1 was then attempted. However, degradation of recombinant protein was observed during purification process, small amount of zNMT1 was finally obtained ( Figure 6B). With this purified protein, we attempted to detect substrate binding activity of recombinant zNMT1. As shown in Figure 6C , the recombinant protein was binding its substrate peptide without depending on the existence of myristoyl-CoA. In contrast, pre-myristoylated substrate peptide did not bind to the recombinant protein in the presence of myristoyl-CoA. These data suggest that NUS-zNMT1 has enzymatic activity characteristic of NMT, thus nmt1a gene encodes functional protein.

Myr 2-OHMyr
Considering from the reactivity of zNMT1 antiserum, we expected that the 29 kDa zNMT1 detected in early developmental stages included an amino-terminal part of zNMT1 peptide which is probably produced from full length zNMT1 gene. Therefore, we next expressed tagged zNMT1 in early developmental stage ( Figure 7B). Full length zNMT1 fused to myc-His tag at its C-terminal was expressed in the embryos at 6 hpf, the protein extracted from the embryos was purified with Ni-Sepharose beads. Immunoblotting of purified zNMT1-myc-His with anti-myc antibody revealed that its molecular weight became smaller (35 kDa) than expected from the full length zNMT1s amino acid sequence (58 kDa). Intact NMT in protein extract from 6 hpf embryos was then detected by substrate binding assay and found that main bands were detected in low molecular weight (30 kDa and 33 kDa). The NMTs in early embryo could be detected by using other potential NMT substrate peptides originated from EF-1α, MARCKS, MARCKSL, Src (Supplemental Figure 5). These findings suggest that low molecular weight zNMTs with enzymatic activity are produced specifically in early developmental stage from nmt1a gene.

Discussion
Although N-myristoyltransferases (NMTs) are expected to have important roles in development of mammalian [22], its characteristics during deveFlopment have not been revealed. In previous studies, the NMT of Saccharomyces cerevisiae was used as model for NMT and well examined biochemically, probably because the enzyme in yeast can be prepared and purified easily [6,18,20]. Those studies addressed the protein structure and biochemical characteristics of yeast NMT, provided a good understanding of a NMT enzyme [25][26][27]. However, the yeast NMT does not have high homology to mammalian's one (Figure 1 and 2). Some studies described different activity of rat NMT from yeast NMT to same substrate peptide [18,28]. These facts implied distinct characteristics of mammalian's NMTs from yeast NMT, thus it is important to study with enzyme which has close relation to mammalian's one.
To elucidate the physiological functions of mammalian's NMTs, the NMT1 deficient mouse was established previously and their phenotype was analyzed. The report implied that NMT1 but not NMT2 is essential for embryonic development of mouse [22]. The detail roles of NMT1 in development, however, are largely unknown because of lethality of NMT1 KO mice [22] and inaccessibility of mammalian's embryos.
In our study, we concentrated on zebrafish as a useful model for developmental studies to elucidate the functions of NMT1 during embryogenesis. The usefulness of this model is availability of a large number of embryos at one time, external fertilization, rapid development and easier observation of their embryonic processes. In addition to these advantages, the information about zebrafish genome is recently available on the web site such as ZFIN or Ensembl. For the first step to investigate NMTs in zebrafish, we searched database and found three transcript of zebrafish NMTs (zNMTs) (nmt1a, NM_001020480, nmt1b, NM_001162849 and nmt2, NM_001199754). As we expected, nmt1a and nmt2 product (NP_001018316 and NP_001186683) had significant high homology to human or mouse NMT1 and NMT2, respectively ( Figure 1 and Figure 2), and nmt1a product had N-myristoyltransferase activity ( Figure 6). The nmt1a and nmt2 were expressed in embryos at early developmental stages (at least in 2 hpf embryos) and the expression lasted during embryogenesis ( Figure 3 and Supplemental Figure 1). These data suggest that zNMTs play pivotal roles in embryonic development of zebrafish like mammalians' one. Consequently, zebrafish can be a good substitute for mammalian studies for NMT's roles during development. A NMT inhibitor, 2-hydroxymyristic acid (2-OHMyr) impaired the embryonic development ( Figure 4 and Table 2). Inhibition of NMTs revealed that embryos in early developmental stage were affected critically; most of the embryos challenged by it at early stage become lethal ( Table 2). Those results imply that zNMTs are important for basic functions, such as proliferation of cells in early developmental stage. It has not been completely cleared which gene, nmt1a or nmt2, has more contribution to the development. However, morpholino antisense oligo against mnt1a caused development arrest in dome or early epiboly stage and finally those embryos were dead ( Figure 5), suggesting importance of nmt1a gene after early epiboly. From RNA blotting, expression of nmt1a in 6 hpf embryo seems to be major compared with that of nmt2 ( Figure 3B). Semi-quantitative RT-PCR showed, however, that nmt2 expression could be seen even in 1 hpf embryo (Supplemental Figure  1). Therefore, mnt1a expression might be important for development after epiboly stage and mnt2 might work in initial stage.
We next attempted to analyze zNMTs protein expression during development with antibodies which recognize specifically N-terminal part of zNMT1 (supplemental Figure 3). Interestingly, the endogenous zNMTs appears in various molecular weights by immunoblotting with anti-zNMT antibodies ( Figure 7A and supplemental Figure 4). In 48 hpf, at least three zNMTs with different molecular weight (about 49 kDa, 53 kDa, 66 kDa) were detected with antiserum for zNMT1 ( Figure  7A). In mammalian tissue or cell, existence of several NMT isoforms with different molecular weight (mostly>40 kDa) has been reported [19,[29][30][31][32]. In database, molecular weight of zNMT1 produced from nmt1a gene are expected to be about 39 kDa and 58 kDa, and form nmt1b are 54 kDa, respectively. The nmt1b gene product, however, does not have a high homology to nmt1a product. Therefore, the antiserum against zNMT1 (immunized with nmt1a product) hardly reacts with nmt1b product. Those facts indicate that 66 kDa, 53 kDa and 49 kDa proteins might be nmt1a product. In the early developmental stages, 65 kDa and 29 kDa zNMT1 detected specifically with zNMT1 antiserum were not also consistent with the molecular weight of zNMT1 molecules recoded in database. Those NMTs are suggested to have similar amino acid sequence to amino terminal part of zNMT1, because the antiserum recognizes only N-terminal part of zNMT1 (Supplemental Figure 3). Because single band was detected by RNA blotting for nmt1a expression ( Figure 3B), and because the nmt1a splicing variants registered in data base does not produce such low or high molecular weight protein, the 29 kDa and 65 kDa zNMT is not thought to be produced from the splicing variant of nmt1a. We have expressed zNMT1-myc-His protein, which had myc and His tag at the carboxyl terminal of zNMT1, and found that zNMT1-myc-His protein was detected as much smaller molecule than it was expected from the ORF ( Figure 7B). These data implied that the small zNMT1 was produced from full length cDNA. One possibility for those multiple molecular weight NMTs is cleavage by specific protease or degradation of full length NMTs which can be detected as 66 kDa protein in early embryonic stages (Supplemental Figure 5). Another possibility is that multiple potential start codons exist in open reading frame (data not shown) causes translation of zNMT1 with multi-length N-terminal part. Cleavage by specific protease of mammalian NMTs has been reported and those processing would be involved in regulation of the enzymatic activity or its localization in cell. For example, bovine muscle NMT has 'PEST' sequence which is recognized by specific protease [33,34]. Caspases recognize NMTs and cleave them at lysine (K) box near N-terminal, result in alteration of their distribution in the cell [34,35]. It was also reported that N-terminal deletion causes increase of yeast NMT activity [36]. Multiple start codons in one open reading frame were shown in rat NMT1 [21]. These finding implied that posttranslational processing might regulate NMT activity in the cell.
The next question is whether the low molecular weight zNMTs has any functions during embryogenesis of zebrafish. Detection of endogenous NMTs by using a biotinylated peptide substrate recognized by zNMT1 ( Figure 6) showed that there existed 33 kDa and 30 kDa protein which could interacted with the peptide ( Figure 7C and supplemental Figure 5). These results suggest that low molecular weight zNMT1 maintain its activity. Since the C-terminal part of NMTs seems to be critical for their activity [36], thus 29 kDa zNMT detected by immunoblotting with the antiserum might have low activity. On the other hand, 33 kDa zNMT, which could not be detected with the antiserum, seemed to have highest activity in protein extract from 6 hpf embryos. Expression of the zNMT1-myc-his resulted in appearance of 35 kDa zNMT1-myc-his protein ( Figure 7B).
In conclusion, we have shown in this study that zebrafish NMT (zNMTs) has high homology to human or mouse NMTs and those NMTs are essential for their embryonic development like mammalians' case. These findings clearly show that zebrafish can be a substitute (B) The recombinant NUS-zNMT1 protein in E.coli lysate was purified by His tag affinity chromatography. The purified sample was reacted with HRV3C protease to removal of NUS tag. Purification of zNMT1 at each process was confirmed by SDS-PAGE followed by CBB staining. (C) The purified recombinant zNMT1 protein was reacted with substrate peptide labeled with biotin as indicated, after linkage between zNMT1 and its substrate peptide. zNMT1-substrate complex in each samples were analyzed by SDS-PAGE followed by immunoblotting with anti-biotin antibody. model for mammalian when analyzing a function of NMTs during development. We have also found that N-terminal part of zNMT1 (29 kDa) appeared specifically during embryogenesis. These findings suggest that N-terminal part of zNMT1 is cleaved by post-translational processes for specific characteristics like substrate specificity. We believe our findings in this study will provide a key to analyze the characteristics of NMTs during vertebrate's embryogenesis.