TM4SF5 (transmembrane 4 L six family member 5) is involved in EMT (epithelial–mesenchymal transition) for liver fibrosis and cancer metastasis; however, the function(s) of TM4SF5 during embryogenesis remains unknown. In the present study the effects of TM4SF5 on embryogenesis of zebrafish were investigated. tm4sf5 mRNA was expressed in the posterior somites during somitogenesis and in whole myotome 1 dpf (day post-fertilization). tm4sf5 suppression impaired development of the trunk with aberrant morphology of muscle fibres and altered expression of integrin α5. The arrangement and adhesion of muscle cells were abnormally disorganized in tm4sf5 morphants with reduced muscle fibre masses, where integrin α5-related signalling molecules, including fibronectin, FAK (focal adhesion kinase), vinculin and actin were aberrantly localized, compared with those in control fish. Aberrant muscle developments in tm4sf5 morphants were recovered by additional tm4sf5 or integrin α5 mRNA injection. Such a role for TM4SF5 was observed in the differentiation of C2C12 mouse myoblast cells to multinuclear muscle cells. Taken together, the results show that TM4SF5 controls muscle differentiation via co-operation with integrin α5-related signalling.
TM4SF5 (transmembrane 4 L six family member 5) is a transmembrane protein with four transmembrane domains and a member of the tetraspanin L6 superfamily, which is characterized by different cysteine residues and sequence homology in the EC2 (extracellular loop 2) between genuine tetraspanins and the L6 superfamily . TM4SF5 is up-regulated in diverse cancer types including hepatic cancers  and in the fibrotic livers of CCl4-administered mice . The TM4SF5-mediated development of liver diseases appears to involve EMT (epithelial–mesenchymal transition). TM4SF5 induces EMT in hepatocytes through regulation of RhoA GTPase activity for elongation of the cellular morphology . In addition, TM4SF5 enhances cell migration via direct interaction-mediated activation of FAK (focal adhesion kinase), a downstream effector of integrins, which are cell adhesion receptors . TM4SF5 is also involved in cell adhesion-related signalling and was shown to be associated with integrins . Therefore TM4SF5-mediated induction of EMT and, consequently, cellular migration can play roles in homoeostatic physiological processes, such as developmental embryogenesis, and pathological processes, such as fibrosis and cancer metastasis, because EMT is a well-known process that is involved in embryogenesis, organ fibrosis and cancer metastasis . It is also plausible that TM4SF5 may play a role in embryogenesis by regulating cellular migration, because correct migration patterns are critical for the control of development in multicellular organisms . However, the role of TM4SF5 in embryonic development has not been explored.
During embryonic morphogenesis, cells undergo a series of complicated movements and cell-shape changes to form highly ordered structures at exact times and positions . During these processes, diverse intracellular signalling cascades are activated to regulate migration, adhesion and morphological changes . At the molecular level, the cell migration and morphological changes occur simultaneously and closely interconnect . A well-known example can be the formation of the vertebrate myotome, including segmentation of somites, which are mesodermal derivatives generated as transient structures along the anterior-to-posterior body axis during the segmentation period . Somitic cells undergo a wide range of migration and morphological changes and produce the dermatome, sclerotome and myotome . Most somitic cells in zebrafish differentiate into muscle, forming a comparatively large myotome .
Presently zebrafish, Danio rerio, is widely used for scientific research, because it has many advantages over any other vertebrates. Zebrafish embryos have a transparent body enabling various methods of microscopic observation, undergo rapid development, can be raised with a lower cost of maintenance and are easy to manipulate genetically . During zebrafish embryonic development, somitogenesis is almost complete and the myotome begins to mature by 1 dpf (day post-fertilization) . Within the segmented somites, cells comprising the posterior compartment express myogenic myoD (myogenic differentiation) or myf5 transcription factors . These cells are adaxial cells located in the mesoderm adjacent to the notochord, and they differentiate into slow muscle fibres shortly after somite formation . Most of the adaxial muscle progenitor cells migrate laterally across the myotome and elongate to form the superficial slow muscle fibres, whereas a few cells remain medially as muscle pioneers . In the lateral part of adaxial cells, called the paraxial mesoderm, cells differentiate into fast muscle fibres . The elongation of slow muscle fibres is necessary for the elongation of fast muscle fibres . The fibres elongate, anchoring somite boundaries composed of epithelial somitic border cells; thus, the proper formation of somite boundaries is critical for the normal elongation of muscle fibres . The somite boundaries mature into V-shaped myotome boundaries with accumulation of various ECM (extracellular matrix) proteins (e.g. fibronectin) and focal adhesion components (e.g. FAK, integrin and paxillin) . In particular, the fibronectin matrix is assembled depending on integrin α5, which is a crucial component for boundary maintenance through its functions in the epithelialization of somitic border cells [19,20]. Regarding these molecular events related to myotome morphogenesis, the functional and/or physical interactions between focal adhesion proteins and membrane proteins or receptors may be important, although the exact signalling mechanisms remain poorly understood.
Similar to tetraspanins located at TERMs (tetraspanin-enriched microdomains) of the cell membrane [21,22], TM4SF5 collaborates with integrins during cell adhesion and migration . As a member of the tetraspan(in) family and a regulator of cell migration and EMT, it is reasonable to hypothesize that TM4SF5 might be involved in zebrafish morphogenesis.
In the present study, we investigated whether and how TM4SF5 affected embryogenic developmental processes using zebrafish injected with TM4SF5 morpholino oligonucleotides and how TM4SF5 was involved in myotube differentiation of C2C12 mouse myoblast cells. We found that knockdown of TM4SF5 resulted in an altered expression pattern of integrin α5, abnormal translocation and differentiation of slow and fast muscle cells, and aberrant myotome formation. The TM4SF5 suppression-mediated effects were recovered by additional introduction of integrin α5. TM4SF5-mediated muscle differentiation was also observed in the differentiation of C2C12 cells into multinuclear muscle cells. Regarding the impaired muscle differentiation, the TM4SF5 suppression-mediated abnormal expression pattern of integrin α5 appeared to be linked to aberrant accumulation of focal adhesion molecules, such as vinculin, FAK, Tyr397-phosphorylated FAK and actin, leading to impaired formation of somite boundaries and aberrant formation of muscle fibre stacks.
Protein sequence analysis
TM4SF5 protein sequences for zebrafish (accession number NP_001002372.1), mice (accession number NP_083636.2) and humans (accession number NP_003954.2) were obtained from the NCBI (National Center for Biotechnology Information). Multiple sequence alignment was performed using CLUSTAL Ω and the NCBI HomoloGene service.
Zebrafish strains and maintenance
Wild-type zebrafish (D. rerio) were maintained at 28°C in an oxygenated fish tank on a 14 h/10 h (light/dark) cycle as described previously  and used for the experiments in the present study. Embryos were collected from natural mating and maintained in mineralized water. Ages are given in hpf (hours post-fertilization) or dpf based on the standard developmental stage .
RNA isolation and cDNA synthesis from zebrafish embryos
Embryos (15–30) matched by exact ages were collected in eppendorf tubes. Embryos were lysed completely by adding TRIzol® reagent (Ambion) into the tubes and homogenizing with 1 ml syringes. Standard chloroform/isopropanol/75% ethanol precipitation and washing steps were performed. The total RNA pellet was eluted with diethylpyrocarbonate-treated RNase-free water. All procedures were performed at 4°C in an RNase-free environment. For cDNA synthesis from total RNA, template RNA was mixed with oligo(dT) and incubated at 70°C for 5 min and then was incubated on ice for 5 min. The reactant was mixed with 5×MMLV (Moloney murine leukaemia virus) reverse transcription buffer, 10 mM of a dNTP mixture [(dUTP, dATP, dGTP, dCTP and dTTP (pH 7.0)], ribonuclease inhibitor and MMLV reverse transcriptase (Promega), before incubation at 25°C for 5 min (annealing), followed by incubation at 42°C for 60 min (reverse transcription reaction) and incubation at 70°C for 15 min (inactivation of the enzyme).
PCR of the synthesized cDNA was carried out under the following conditions: initial denaturation at 95°C for 5 min, followed by 30 cycles of 95°C for 50 s, 55°C for 30 s, 72°C for 50 s, and then a final extension at 72°C for 5 min. The following primer pairs during RT (reverse transcription)–PCR analysis to confirm the effect of spl-MO (splice-blocking morpholino) against the tm4sf5 gene were used: forward primer F1 (5′-ATGTGTACCGGAAAGTGTGCT-3′) and reverse primers R1 (5′-TGTTGGATGTATCTAATCAGTTTTGT-3′) giving a product of 363 bp, R2 (5′-CCTTTCCTGTCCTCAAATGGAT-3′) giving a product of 398 bp, and R3 (5′-GATTTTCTTTTCCGGCAGTCT-3′) giving a product of 629 bp. For control experiment, β-actin forward (5′-GCAGAAGGAGATCACATCCC-3′) and β-actin reverse (5′-CATTGCCGTGCACCTTCACCG-3′) giving a product of 323 bp were used. The primer pairs for RT–PCR analysis of gene expression were: tm4sf5 forward (5′-GAT-AAGCTTATCATGTGTACCGGAAAGTGTGCT-3′), tm4sf5 reverse (5′-GATTCTAGAATCCAGACCTCCTTCATCCGA-TTT-3′), integrin α5 forward (5′-CCCAAGCTTTCGGCGTC-TACAAGCTGCTCTC-3′), integrin α5 reverse (5′-TGCTCT-AGAAGGTCCTCCCAGCACCACCTT-3′), fak forward (5′-GCCGGGCTCTGGATTTATTTA-3′), fak reverse (5′-CAGTCCTAGGAGAAGCGTGAGAGT-3′), vinculin forward (5′-CCGAACACGTGCCTTTTTCTC-3′), vinculin reverse (5′-CACAGGCGTTCTCCACCTTA-3′), acta1b (actin α1b) forward (5′-GGAAATGAGCGTTTCCGCTG-3′), acta1b reverse (5′-GAAGACAGATGCAGCCGATG-3′), myhc4 (fast myosin heavy chain 4) forward (5′-AGGCGTCCGCAAATATGAGA-3′), myhc4 reverse (5′-AGTGGAGCTTCTATTCGGGT-3′), β-actin forward (5′-GCAGAAGGAGATCACATCCC-3′), β-actin reverse (5′-CATTGCCGTCACCTTCACCG-3′), elf [an abberantly spliced product of 363 bp including the tm4sf5 exon E1 and part of intron 1 (as shown in Figures 2A and 2B)] alternative splicing product forward (5′-ATGTGTACCGGAAAGTGTGCT-3′) and elf alternative sp-licing product reverse (5′-TGTTGGATGTATCTAATCAGTTT-TGT-3′).
In vitro transcription of digoxygenin-UTP labelled anti-sense RNA probes
Probe DNA myoD (a gift from Dr T. Huh, Kyungpook National University, Korea), tm4sf5 (accession number NM_001002372), and integrin α5 (accession number NM_001004288) were separately amplified to include restriction enzyme sites by PCR and eluted using a PCR purification kit (GeneAll) to be cloned into the pGem-T vector (Promega). Sequences and directions of inserted DNAs were confirmed by direct sequencing and NCBI BLAST analysis. Probe vectors were linearized by appropriate restriction enzymes and then were purified using a gel purification kit (GeneAll). In vitro transcription of digoxygenin-labelled RNA was carried out using a Digoxygenin RNA Labeling kit (Roche). The RNA probe was denatured at 80°C for 5 min before use.
Whole-mount in situ hybridization
Whole-mount in situ hybridization was carried out with digoxygenein-labelled antisense in situ probes as described previously, using RNA-grade solutions and apparatuses .
Whole-mount in situ immunostaining
Embryos at 1 dpf were fixed in 4% paraformaldehyde in PBS at 4°C overnight, rinsed for 2 h with PBT (PBS and 0.1% Tween 20), treated with collagenase (1 mg/ml) in PBT for 10 min, quickly washed twice with PBT, washed twice with PBDTT (1×PBS, 0.1% Tween 20, 1% DMSO, 1% BSA and 0.5% Triton X-100), and then blocked in PBDTT with 5% BSA and 10% sheep serum at room temperature for 3 h. An F59 primary antibody (anti-myosin heavy chain antibody to detect slow muscle cells) and F310 primary antibody (anti-myosin light chain antibody to label fast-twitch muscle myosin) were added at a 1:10 dilution and incubated overnight at 4°C. Embryos were washed for 4 h at room temperature in PBDTT and blocked in PBDTT/1% BSA for 4 h at room temperature. Biotin-labelled anti-mouse IgG (Vector Laboratories) was added at a 1:2000 dilution and incubated overnight at 4°C. After washing with PBDTT for 4 h at room temperature, VECTASTAIN® ABC reagent (Vector Laboratories) was added to embryos and incubated according to the manufacturer's instructions. After the second washing with PBDTT for 2 h, diaminobenzidine and H2O2 were added. After colour detection, embryos were fixed with 4% paraformaldehyde in PBS for 20 min at room temperature. For microscopy and storage, embryos were washed with PBT and stored in PBT with 90% glycerol. Images were collected using a microscope (BX51; Olympus).
Antisense morpholino oligonucleotides
MOs (morpholino-modified oligonucleotides; Gene-tools) were generated to target the splice-donor site of exon 1 of tm4sf5 (MO1–ACAGACTGGAAACTCACAAATAGTC, spl-MO) or translation initiation site of tm4sf5 (MO2–TACACATGGCCTTTCACACGAGCCA, trs-MO). The mismatch MO was 5′-ACAcACTcGAAAgTCAgAAATAcTC-3′ (lower case letters represent the mismatched nucleotides compared with spl-MO). Embryos were microinjected at the 1–4 cell stage with 2, 3 or 4 ng of morpholino/embryo. Standard control or mismatch morpholino was used at the respective concentration.
Recovery of tm4sf5 morphants by tm4sf5 or integrin α5 mRNA
Embryos were also microinjected at the 1–4 cell stage with control MO, mismatch MO, tm4sf5 MO or tm4sf5 MO (2 ng) plus zebrafish tm4sf5 (100 pg/embryo) or human integrin α5 mRNA (25 pg/embryo). Embryos (1 or 3 dpf) were analysed for trunk appearance phenotypes and immunostained against the indicated molecules.
Embryos hybridized with tm4sf5 RNA probe and stored at 4°C in 90% glycerol/PBS were used for cryosection preparation. After removal of glycerol by several washes in TBS/0.1% Tween 20, 5–10 embryos were transferred into a plastic mould filled with frozen section medium (FSC22 Clear; Leica) and oriented appropriately with forceps. Next, the embryos were stored at −80°C overnight. Using a Microm cryostat HM525 (Thermo Scientific), serial sections (10-μm thick) were prepared. Sectioned slices were attached on to microslides. The morphology of embryos was viewed using a Zeiss Axiocam ICc1 microscope (Carl Zeiss) and processed using Axiovision software or was viewed using a BX51 microscope (Olympus).
Embryos at the indicated hpf were processed for immunofluorescence staining as described previously . The primary antibodies included anti-FAK (1:30 dilution), anti-fibronectin (1:50 dilution), anti-vinculin (1:100 dilution), anti-F59 (1:10 dilution), anti-F310 (1:10 dilution; University of Iowa, Developmental Studies Hybridoma Bank) and anti-integrin α5 (1:50 dilution; Chemicon). The antibody incubations were processed overnight at 4°C.
C2C12 cell culture and differentiation
C2C12 mouse myoblast cells  were obtained from Dr T.M. Kang (Sungkyunkwan University, Korea). Subconfluent cells in medium containing 10% FBS or cells transiently transfected with the indicated plasmids for 48 h were harvested for whole-cell lysates. Cells were transiently transfected with GFP–shRNA against control scrambled sequence or TM4SF5 (shTM4SF5). The cells were induced to myotube differentiation by reducing the serum content to 2%. At 2 days after the reduction in FBS, cells were processed either for RT–PCR or indirect immunofluorescence, in parallel with undifferentiated control cells. RNA preparation and cDNA synthesis were performed as above and RT–PCR was carried out for mouse myoD, myog (myogenin) and tm4sf5. The primers used were: myoD forward (5′-AAGACGACTCTCACGGCTTG-3′), myoD reverse (5′-CATTCACTTTGCTCAGGCGG -3′), myog forward (5′-GAAGCGCAGGCTCAAGAAAG-3′), myog reverse (5′-CCACGATGGACGTAAGGGAG-3′), tm4sf5 forward (5′-CTTCCCTGCTCACGTTCACT-3′) and tm4sf5 reverse (5′-GCGCCTCACACAAATTCCAA-3′). Cells were immunostained for myosin heavy chain (red; F59; Developmental Studies Hybridoma Bank), in addition to nuclear staining using DAPI and GFP staining for shRNA.
Student's t test was performed for comparisons of mean values to determine statistical significance. A P value less than 0.05 was considered statistically significant.
Comparison of TM4SF5 protein sequence homology among zebrafish and other vertebrates
The zebrafish tm4sf5 gene is located on chromosome 7. The full-length clone of zebrafish tm4sf5 is 1672 bp in size and consists of a reading frame encoding 201 amino acids. We first analysed the sequence homology of TM4SF5 protein among zebrafish and other vertebrates through database screening. Approximately 57% of the amino acids in zebrafish TM4SF5 are identical with human and mouse TM4SF5 (Supplementary Figure S1 at http://www.biochemj.org/bj/462/bj4620089add.htm). Importantly, several regions are highly conserved, including the N-terminal region (Met1–Cys9; eight of the nine amino acids are identical), and the intracellular loop domain (Arg67–Val89; 22 of the 23 amino acids are identical) (Supplementary Figure S1).
The intracellular loop domain (ICL) of TM4SF5 has been known to directly bind FAK, leading to FAK activation, and cell adhesion and migration . During embryogenesis, cells undergo dynamic changes of migratory and adherent states . Thus these similarities in protein sequences indicate that zebrafish TM4SF5 might also have roles in developmental processes through molecular interactions or regulation of cell adhesion and migration in a manner similar to human TM4SF5.
Expression of tm4sf5 during zebrafish development
RT–PCR analysis revealed that the tm4sf5 mRNA level showed two peaks during zebrafish development, including a spike at 17–18 hpf, a decrease until 2 dpf, and another peak at 3 or 4 dpf (Figure 1A). Because most steps of morphogenesis in zebrafish embryos are completed by 3 dpf, we investigated the first peak of tm4sf5 expression more thoroughly. The first peak in expression (17–18 hpf) corresponds to a time period at which somites are formed and develop into the myotome . In the more detailed time analysis, tm4sf5 expression was initiated during the gastrulation period (90% epiboly), increased at the beginning of somitogenesis, and was maintained throughout the somitogenesis stage (Figure 1B). Interestingly, such an expression profile of tm4sf5 (L6H) was quite similar to that of tm4sf1 (L6) (Supplementary Figure S2 at http://www.biochemj.org/bj/462/bj4620089add.htm), which is also involved in cell migration .
Expression pattern of zebrafish tm4sf5 during embryogenesis
We next investigated the spatio-temporal expression pattern of tm4sf5 using whole mount in situ hybridization. At the early time of somitogenesis, tm4sf5 was expressed throughout the whole embryo (Figure 1C, a). In the middle of the somitogenesis stage, however, tm4sf5 expression became more distinct in the posterior somites of the trunk region (Figure 1C, b–d). After 1 dpf, tm4sf5 was expressed in the entire myotome, notochord, neural tubes and head (Figure 1C, e–h). The entire expression level of tm4sf5 was decreased at 2 dpf (not more than 10% of those for 1 dpf embryos) as shown by RT–PCR, but was maintained slightly in the notochord (Figure 1C, i). After 3 and 4 dpf, the expression of tm4sf5 was found in the notochord and head (Figure 1D, a, b, d and e) and was highly expressed particularly in the gut tube (Figure 1D, c and f). Because most of the developmental changes in embryonic morphology occur by 2 dpf , the elevated expression of tm4sf5 until 1 dpf suggested that TM4SF5 might have a role in zebrafish morphogenesis. On the basis of the observations that tm4sf5 transcripts were detected in somites and the entire myotome, we speculated that TM4SF5 functions throughout somitogenesis and myogenesis.
Knockdown of tm4sf5 by antisense morpholino oligonucleotides
To study the function of tm4sf5, we interfered with its expression using two different MOs, which were targeted to the translation initiation site (trs-MO) or the splice-donor site (spl-MO). The spl-MO was designed to target the exon 1/intron 1 boundary (Figure 2A). After microinjection of the spl-MO into the embryos at the 1–4 cell stage, we selected embryos after 4 dpf and performed RT–PCR analysis to assess the efficiency of tm4sf5 knockdown. Using primers to detect the full-length tm4sf5 (F1 and R3; a product of 629 bp) or exon1 to exon2 of tm4sf5 (F1 and R2; a product of 398 bp), we confirmed that the expression of tm4sf5 was suppressed efficiently (Figure 2B). Aberrantly spliced product was also detected using intron-specific primers (F1 and R1; a product of 363 bp) (Figure 2B). We next performed in situ hybridization using 1 dpf control and morphants microinjected with spl-MO to reconfirm the suppression of tm4sf5. The expression of tm4sf5 in morphants was decreased in the head and trunk region, compared with control embryos (Figure 2C). On the basis of these observations, we successfully confirmed that the knockdown of tm4sf5 using antisense MOs was functional.
Developmental defects by knockdown of tm4sf5 using antisense MOs
We next investigated the phenotypes of morphants induced by loss of tm4sf5 function (Supplementary Figure S3 at http://www.biochemj.org/bj/462/bj4620089add.htm). Similar phenotypes, including a small head, embryonic heart failure and a curved trunk were induced by injection of trs-MO or spl-MO. Because of the small head, pericardial oedema appeared in most morphants regardless of the dose injected, the phenotypes of tm4sf5 morphants were classified by the degree of the defect provoked in the trunk after 3 dpf (Figure 2D). The severity of the defect in trunk was proportionally increased as the concentration of spl-MO or trs-MO was increased (Figures 2D–2F). From these data we could conclude that knockdown of tm4sf5 inhibited the overall development of the head, trunk and heart in zebrafish embryos. However, because only the tail morphology was induced dose-dependently among these phenotypes, we could infer that tm4sf5 might functionally affect the formation of the trunk and tail. The optimal dose of spl MO used in the present study was fixed at 2 ng/embryo when the defect of embryonic morphology was not too severe. Meanwhile, a mismatch MO did not cause the defect in the trunk nor suppression of tm4sf5 mRNA (Figure 2G, and Supplementary Figure S4 at http://www.biochemj.org/bj/462/bj4620089add.htm), although the spl-MO injection-mediated tm4sf5 suppression caused abnormal trunks (Figures 2D and 2G). Furthermore, microinjection of zebrafish tm4sf5 mRNA together with the spl tm4sf5 MO resulted in a partial recovery from the defected phenotypes (Figure 2G). The recovery effect was partial, presumably indicating that the tm4sf5 mRNA injection could be more optimized.
Expression pattern of myogenic progenitor cell markers in control and tm4sf5 morphant embryos
In adult zebrafish, the trunk structure is composed of muscle, cartilage and bone derived from somites . The somites produce the myotome mainly during somitogenesis in the first day of development . Several data above revealed that tm4sf5 was expressed in the somites during somitogenesis and in the entire myotome at the end of somitogenesis. Additionally, loss of tm4sf5 function affected the trunk morphology of the morphants (Figure 2). Furthermore, the morphants could not swim straight, and were swirling in response to touch (results not shown). Therefore we hypothesized that tm4sf5 might function in the formation of somites and the myotome, including development of muscle fibres.
We next tested whether the differentiation of myogenic progenitor cells was normal in tm4sf5 morphants. myoD is known to be expressed in muscle progenitor cells from the onset of somitogenesis . Using in situ hybridization analysis, we found that expression of myoD in tm4sf5 morphants was not affected during several stages of somitogenesis, compared with that in control MO-injected embryos (Figure 3A). On the basis of these observations, it is likely that tm4sf5 may not affect the formation of somites and that the phenotypes shown in the trunk after tm4sf5 MO injection might be quite specific rather than a result of injection-based artefacts. The expression of EphrinB2, a transmembrane ligand for ephrin receptor tyrosine kinases and a posterior marker involved in somite boundary formation in addition to myoD, was not changed during somitogenesis after tm4sf5 suppression, compared with the control MO-injected embryos (Figure 3B). After 1 dpf, when somite formation was nearly complete, the expression of myoD was altered in morphants and the area of cells expressing myoD was reduced in morphants, compared with that in control MO-injected embryos (Figure 3C). Such aberrant myoD expression patterns after tm4sf5 MO injection were obvious at 3 dpf (Figure 3D). Therefore aberrant trunk and tail morphology in morphants was not correlated with the formation of somites, but were more correlated with the processes in muscle differentiation or morphogenesis.
Impaired slow muscle and fast muscle fibre development in tm4sf5 morphant embryos
Slow muscle and fast muscle development in control and tm4sf5 morphant embryos
During somitogenesis, myoD-expressing muscle progenitor cells begin to differentiate into two different types of muscle fibres . The more medially positioned adaxial cells migrate away from the notochord to become superficial slow muscle fibres, and other cells differentiate into fast muscle fibres in the deep portion of the myotome . Thus we tested whether loss of tm4sf5 function affected formation of muscle fibres. Using antibodies that recognize myosin heavy chain in slow and fast muscle fibres, we found that both muscle types were morphologically abnormal in morphants (Figures 3E and 3F). Although the slow muscle cells migrated properly to the superficial region of the myotome and formed parallel stacks of fibres, there were spaces where the fibres were missing in the dorsal and ventral regions of morphant embryos (Figure 3E, * and square brackets), and the muscle pioneer cells in the midline of the embryos were abnormally distributed or disorganized (Figure 3E, arrows). With regard to the fast muscle fibres, it was shown that fibres were missing in the dorsal and ventral regions of the morphants and the transverse views of this staining seemed to indicate a shift to a more interior location compared with those in the control embryos (Figure 3F). Furthermore, the organization or packing of slow muscle cells visualized using the anti-F9 antibody (i.e. alignment of the fibres in one somite with those in the neighbouring somite) was abnormally interrupted in tm4sf5 morphants compared with that in the control embryos (Figure 3G). In addition, tm4sf5 morphants seemed to have less fibres in some somites compared with the control embryos (Figure 3G). When counted, the slow muscle cell numbers per somite in tm4sf5 morphants were significantly lower compared with those of the control MO-injected embryos (Figure 3H). Thus we could conclude that knockdown of tm4sf5 induced morphogenetic defects of both of the muscle fibre types, including reduced muscle masses, suggesting that tm4sf5 may be involved in the morphogenesis of muscle fibres.
Altered expression pattern of integrin α5 in tm4sf5 morphant embryos
We next investigated the factors that affect the morphogenesis of muscle fibres. It has been proposed that muscle fibres are formed as somitic cells elongate into long fibre type cells, anchoring to intersegmental boundaries . Therefore proper formation of the boundaries is important for morphogenesis of muscle fibres. Importantly, the focal adhesion components, such as FAK and paxillin, have been shown to be accumulated at the intersomitic borders and act as adaptor proteins during cell engagement via an interaction between integrins and the ECM . Particularly, integrin α5-dependent fibronectin accumulation at the boundaries has been shown to be required for the maintenance of intersomitic boundaries of posterior somites . Because previous studies have demonstrated that TM4SF5 retains integrin α5 on the plasma membrane surface to regulate intracellular signalling activities , we hypothesized that tm4sf5 might affect zebrafish myogenesis through regulation of the integrin α5 expression level on the membrane surface. Although many different commercial antibodies against zebrafish and/or human integrin α5 did not show very clear staining, it appeared that tm4sf5 morphants might involve abnormal integrin α5 localization (Supplementary Figure S5 http://www.biochemj.org/bj/462/bj4620089add.htm). Meanwhile, using in situ hybridization analysis, we determined that expression of integrin α5 in tm4sf5 morphant and control embryos. Integrin α5 was expressed in many regions of wild-type embryos, including muscle pioneers (indicated by black arrowheads) and branchial arches after 1 dpf (Figure 4A, upper panel). However, in tm4sf5 morphant embryos, the overall expression of integrin α5 was disrupted and more concentrated in the YSLs (yolk syncytial layers) (Figure 4A, lower panel). The expression of integrin α5 in the pharyngeal arches was reduced in morphants (Figure 4A, red broken triangle). Importantly, the expression of integrin α5 in the muscle pioneer cells was also reduced in the morphant embryos (Figure 4A, inserts). A muscle pioneer cell is a type of slow muscle cell, and the alteration in the expression of integrin α5 in these cells coincided with disruption of slow muscle fibre development, as shown in Figures 3(E) and 3(G). These results support our hypothesis that tm4sf5 can co-operate with integrin α5 during zebrafish myogenesis.
tm4sf5 suppression alters the expression pattern of integrin α5 and its related signalling molecules
Altered, disrupted somite boundaries in tm4sf5 morphant embryos
Because integrin α5 interacts with fibronectin and controls focal adhesion molecules, including FAK and vinculin, we examined their expression levels in either control or tm4sf5 morphant embryos by RT–PCR. First, it was found that tm4sf5 morphants slightly reduced integrin α5 mRNA levels compared with the control embryos (Figure 4B). In addition to myogenic factors in skeletal muscle, including myoD, myhc and acta1b, the integrin-related adhesion molecules fak and vinculin localized at focal adhesions had unaltered expression levels, although tm4sf5 spl-MO injection resulted in efficient suppression of tm4sf5 and a concomitant induction of an aberrant splicing product of e1f, the 363 bp F1R1 product in Figure 2(B) (Figure 4C). The tm4sf5 and integrin α5 mRNA levels in tm4sf5 morphants were significantly lower than those in the control embryos, although the fak mRNA level was not significantly changed (Figure 4D). Next we compared the expression patterns between control and tm4sf5 morphants by immunostaining. When FAK and fibronectin at 1 dpf and vinculin and actin at 3 dpf were immunostained, tm4sf5 morphants showed disrupted somite boundaries with their abnormal localizations compared with the control MO-injected embryos (Figure 4E). These data indicate that TM4SF5 suppression might disturb the expression level and pattern of integrin α5, which was linked to extracellular fibronectin and intracellular FAK, vinculin and actin, leading to aberrant somite boundary formation.
tm4sf5 MO-mediated phenotypes are recovered by integrin α5 mRNA introduction
Since TM4SF5 suppression impaired the expression and localization of integrin α5, we next examined whether tm4sf5 MO-mediated aberrant somite boundary formation could be recovered by a concomitant integrin α5 mRNA microinjection. At 1 or 3 dpf, embryos were analysed for their phenotypes. Interestingly, the abnormal trunk phenotypes of the tm4sf5 morphants became less numerous and severe when integrin α5 mRNA was additionally microinjected (Figure 5A). The recovery effects were partial, presumably because the integrin α5 mRNA was not optimally injected and/or other molecules might further be involved in the effects of TM4SF5. In addition, the shortened dorsal–ventral thickness of the 10th–12th somites (Figure 5B), the missed local packing of slow muscle fibres (white arrow head) or weak and internally dense fast muscle fibres, as shown previously in Figures 3(E) and 3(F) (Figure 5C), and abnormal stacking of slow muscle immunostained by anti-F59 antibody (Figure 5D) upon tm4sf5 MO injection were each recovered to the control level by additional integrin α5 mRNA injection. Furthermore, an abnormal somite boundary, as immunostained by vinculin (Figure 5E), disrupted somite boundary, visualized by FAK staining (Figure 5F), and less efficient differentiation of fast muscles to a multi-nucleated state, visualized by anti-F310 antibody (Figure 5G) in the tm4sf5 morphants, were not shown in the embryos microinjected with both tm4sf5 MO and integrin α5 mRNA. Therefore it appears to be clear that TM4SF5 is involved and linked to integrin α5 in the development of muscle fibres in the zebrafish model.
Abnormal muscle development by tm4sf5 MO was recovered by additional integrin α5 mRNA injection
TM4SF5 expression is required for the differentiation of C2C12 murine myoblast cells into muscle cells
We also aimed to understand the importance of TM4SF5 expression for muscle development in another animal model. Using C2C12 murine myoblast cells in which the reduction of serum in the culture medium leads to differentiation of myotubes , it was examined whether TM4SF5 expression was accompanied by myogenic differentiation of C2C12 cells. Additionally serum reduction of C2C12 cells increased the levels of myoD, myog and tm4sf5 (Figure 6A). Their mRNA levels were significantly enhanced during differentiation processes caused by the serum reduction (Figure 6B). When myosin expression and multinuclear cells were further stained, differentiated C2C12 cells showed enhanced myosin levels with multiple nuclei (Figure 6C, top panels). Cells transfected with GFP-tagged shTM4SF5 did not show an increased myosin level and continued to show mononuclear cells even after serum reduction (Figure 6C, bottom panels); however, cells with GFP–control shRNA still showed an increased myosin level and were multinuclear (Figure 6C, middle panel). Counting of nuclei in the cells transfected with either control shRNA or shTM4SF5 found multinuclear cells after control shRNA transfection and mostly mononuclear cells after shTM4SF5-mediated TM4SF5 suppression (Figure 6D). These experiments also suggest that tm4sf5 may be required for myoblast fusion and this is perhaps the mechanism by which TM4SF5 affects differentiation.
TM4SF5 expression is involved in differentiation of C2C12 myoblasts to myotube
The present study presents functional analysis of tm4sf5 in zebrafish development. We observed that tm4sf5 was highly expressed in the myotome during early zebrafish development. The knockdown of tm4sf5 impaired development of the trunk and tail and these embryos showed abnormal morphology of muscle fibres. Introduction of zebrafish tm4sf5 mRNA together with tm4sf5 MO resulted in a partial recovery of the impaired phenotypes, suggesting a regulatory role for tm4sf5 in myogenesis. Suppression of tm4sf5 impaired muscle mass via the aberrant development of slow and fast muscle fibres. Knockdown of tm4sf5 further slightly reduced and disturbed the localization of integrin α5, which could disrupt the localization of extracellular fibronectin and intracellular FAK, vinculin and actin at the somite boundaries. Furthermore, additional introduction of human integrin α5 mRNA recovered the tm4sf5 MO-mediated phenotypes to more normal ones. Thus we speculate that tm4sf5 functions as a regulator of myogenesis via co-operation with integrin α5. Previous studies have revealed that TM4SF5 functions as a key modulator in the communication between tumour cells and the tumour microenvironment, regulating cell migration and EMT via interaction with signalling molecules such as FAK, integrin receptors and growth factor receptors [2,4,31–33].
TM4SF5 retains integrin α5 on cellular surface to co-operate activations of FAK and c-Src-mediated intracellular signalling pathways . Similarly, TM4SF1 (L6-Ag), another member of transmembrane 4 L six family or tetraspanin L6 superfamily, has also shown to retain integrin β1 on the cell surface . Thus TM4SF5 may play a role in regulation of integrin α5 levels on the cell surface either transcriptionally and/or post-transcriptionally leading to mRNA or protein stability, and therefore co-operation between TM4SF5 and integrin α5 might lead to proper signal transduction for cellular adhesion during the muscle developmental processes. Because the proper regulation of cellular adhesion, migration and EMT is crucial for normal development during embryonic morphogenesis , the present study suggests that zebrafish tm4sf5 might also play an important role as a regulator in developmental processes via its involvement in the proper organization and adhesion of muscle progenitor cells with, and cell–cell contacts around, somite boundaries. The immunostaining of slow and fast muscle cells in the tm4sf5 morphants was disrupted or localized in abnormal local densities compared with those in the control fishes (Figures 3 and 5). Furthermore, the decreased fast myosin staining in the tail (Figure 3F, middle panels) may be due to lack of cells or lack of differentiation to activate myosin as seen in TM4SF5-dependent C2C12 cell differentiation.
We identified that TM4SF5 is conserved in zebrafish as well as other vertebrates and co-operates with integrin α5 to transduce intracellular FAK/c-Src signalling for gene regulation in hepatocytes , indicating that the co-operative cross-talk between TM4SF5 and integrin α5 in zebrafish may also be functional. We observed aberrantly reduced and altered integrin α5 mRNA levels by in situ hybridization or abnormal localization pattern of integrin α5 protein (although of poor quality due to inefficient commercial antibodies) in the tm4sf5 morphants compared with those in the control embryos. Furthermore, the intracellular loop domain of TM4SF5 that directly binds FAK is highly conserved in zebrafish TM4SF5, leading to FAK activation and enhanced directional cell migration in hepatocytes . These data suggested that zebrafish TM4SF5 may also play a similar role via its interaction with FAK. fak mRNA is expressed in developing somites and FAK localizes at intersegmental boundaries in zebrafish as the myotome matures . We found that tm4sf5 mRNA expression was elevated during somitogenesis and was also detected in the developing myotome as was fak mRNA. This coincidence supports our hypothesis that tm4sf5 may play potential roles in cell developmental behaviours correlated with fak. Our observation that tm4sf5 mRNA expression was increased until 1 dpf and decreased after 2 dpf also suggests the possibility that tm4sf5 is likely to act during the period important for myogenesis, which usually occurs within 1 dpf.
A role for tm4sf5 in developmental regulation during myotome formation could be reinforced in a knockdown study using antisense MOs. We observed that the knockdown of tm4sf5 impaired the normal development of the head, heart and trunk in zebrafish. tm4sf5 morphants displayed a small head, pericardial oedema and a curved trunk. In particular, only in the trunk, the impaired phenotype became worse in a tm4sf5 MO dose-dependent manner. The morphants displayed signs of myopathy, such as overall muscle mass reduction and disarray of muscle fibres. Moreover, the tm4sf5 MO-mediated phenotypes were recovered by additional injection of zebrafish tm4sf5 mRNA, indicating that the tm4sf5 MO-mediated phenotypes were not artefacts, although the partial recovery could be improved by optimization of the mRNA injection methodology. Similar phenotypes in zebrafish have previously been observed in many other genes acting in different cellular pathways [37–41]. However, these defects were not caused by the early differentiation of myogenic progenitor cells, but by malformation of the myofibres because expression of the myogenic transcription factor myoD was not altered in tm4sf5 morphants.
During somitogenesis, specification of the slow muscle precursor begins with the onset of myogenesis . Muscle progenitors expressing myogenic transcription factors begin to differentiate into muscle fibres . In this process, a wide range of cell migration and adhesion occurs. Adaxial cell populations that are adjacent to the notochord move towards the lateral surface of the myotome, simultaneously changing the cell shape into a long and skinny rod shape . A few of the adaxial cells remain adjacent to the notochord, comprising muscle pioneers . The more lateral paraxial cell populations differentiate into fast muscle fibres . The signalling molecules involved in specifying slow and fast muscle cell fates are proteins such Sonic Hedgehog and bone morphogenic proteins . However, the morphology of muscle fibres was abnormal in tm4sf5 morphants, as indicated by the disarray of muscle fibres and reduced overall mass with aberrant organizations of fibronectin, integrin α5, FAK, vinculin and actin. Thus TM4SF5 knockdown appeared to mediate the MTJs (discontinuous myotendinous junctions; also called myotome boundaries or myosepta in Figure 4E), similar to a previous study showing that the phenotypes involving a disrupted laminin organization at somite boundaries depend on Nrk2b (nicotinamide riboside kinase 2b)-mediated NAD+ production, even though the somite boundaries are already formed . Therefore these observations suggest that tm4sf5 has a function(s) in the other underlying mechanism in myotome formation, which may appear to involve the adhesion and thereby organization of slow and fast muscle cells, as shown in the hepatocytes leading to enhanced adhesion . Thus it will be important to study migration and fusion of muscle precursor cells by time-lapse imaging analysis.
Once the muscle fibres elongate, they attach strongly to the myotome boundary . The myotome boundary is a more mature structure derived from the initial somite boundary . Improper boundary formation induces the abnormal morphology of muscle fibres . Dystrophin-associated glycoprotein complexes, integrin receptors and focal adhesion molecules function as linkage systems between the muscle cytoskeleton and the ECM at the myotome boundary . When the strong adhesion between muscle cells and the ECM is formed at the boundary sites, activated FAK, as shown by phosphorylation of Tyr397, is accumulated, indicating that integrin signalling is activated at these sites . In previous studies, TM4SF5 was shown to interact with integrins, including integrin α5, and act as a modulator of downstream signalling molecules such as FAK and c-Src [2,4,31,32,47,48]. In the present study, the impaired muscle development due to abnormal myotome boundaries in the tm4sf5 morphants was, interestingly, recovered by additional integrin α5 mRNA injection designed to compensate for the reduced and abnormally distributed integrin α5 mRNA in tm4sf5 morphants. As mentioned above, TM4SF5 binds to FAK and regulates its activity . Thus it could be likely that the function of tm4sf5 in zebrafish may be to modulate these molecules in myotome formation, presumably via collaboration with integrins.
Integrin α5, a typical fibronectin receptor, has been shown to regulate cell migration and adhesion in human cells and has increased retention at the cell surface due to forming a complex with TM4SF5 . It has been proposed that integrin α5-directed assembly of the fibronectin matrix is necessary for epithelialization of somitic border cells and somite boundary maintenance in zebrafish [19,20]. In the present study, we observed that when tm4sf5 was functionally impaired in tm4sf5 morphants the expression of integrin α5 was aberrantly reduced and distributed. These observations support our hypothesis that tm4sf5 functions as a signalling modulator during myotome formation, regulating the function and/or expression of integrin α5 and its related molecules. Because the focal adhesion proteins, integrin receptors and ECM molecules form the myotome boundary in co-ordination with each other, the alteration of integrin α5 expression or function by tm4sf5 suppression may affect other components in focal adhesions and/or integrin-downstream molecules . Furthermore, additional injection of integrin α5 mRNA together with tm4sf5 MO into zebrafish embryos induced lower rates of abnormal muscle development compared with those injected with tm4sf5 MO alone, indicating that integrin α5 downstream of TM4SF5 could, in part, compensate for the loss of TM4SF5 function.
In summary, we found that when tm4sf5 was down-regulated in zebrafish, normal trunk development was impaired dramatically and the morphology of muscle fibres was abnormal, indicating malformation of the myotome with altered expressions of integrin α5 and its functionally related molecules around the somite boundaries following knockdown of tm4sf5. From all of the observations in the present study, we can speculate that tm4sf5 performs a regulatory role via co-operation with integrin α5 in the developmental process, particularly during myotome formation or organization.
focal adhesion kinase
Moloney murine leukaemia virus
fast myosin heavy chain 4
PBS and 0.1% Tween 20
1×PBS, 0.1% Tween 20, 1% DMSO, 1% BSA and 0.5% Triton X-100
transmembrane 4 L six family member 5
yolk syncytial layer
Yoon-Ju Choi and Hyun Ho Kim performed most of the experiments. Jeong-gyun Kim and Hye-Jin Kim performed the recovery experiment during the revision and helped with constructs respectively. Minkyung Kang, Mi-Sook Lee, Jihye Ryu, Haeng Eun Song, Seo Hee Nam and Doohyung Lee helped with the constructs and reagents. Kyu-Won Kim and Jung Weon Lee designed the experiments, analysed the data and discussed the results. Yoon-Ju Choi and Jung Weon Lee wrote the paper.
This work was supported by the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (MEST) through the Global Research Laboratory Program [grant number 2011-0021874 (to K.W.K.)], the Tumor Microenvironment Global Core Research Center (GCRC) from the Korean government (Ministry of Science, ICT and Future Planning) [grant number 2011-0030001 (to J.W.L.)], the Leap Research senior researchers programme [grant number 2013-035235 (to J.W.L.)] and the Medicinal Bioconvergence Research Center [grant number NRF-2012M3A6A4054271 (to J.W.L.)].
These authors contributed equally to this work.