Abstract

As a specific pearl mussel in China, Hyriopsis cumingii has enormous economic value. However, the organism damage caused by pearl insertion is immeasurable. TGF-β/Smad signal transduction pathways are involved in all phases of wound healing. We have previously reported on two cytoplasmic signal transduction factors, Smad3 and Smad5 in mussel H. cumingii (named HcSmads), suggesting their involvements in wound healing. Here, Smad4 was cloned and described. The full length cDNA of HcSmad4 was 2543 bp encoded 515 amino acids. Deduced HcSmad4 protein possessed conserved MH1 and MH2 domains, nuclear location signals (NLS), nuclear exput signals (NES) and Smad activation domain (SAD). Transcripts of Smad3, 4 and 5 were constitutively expressed in all detected tissues, at highest levels in muscles. Furthermore, HcSmad4 mRNA levels were significantly increased at incision site post wounding, and expression of downstream target genes of Smad4,such as HcMMP1, HcMMP19, HcTIMP1 and HcTIMP2 were upregulated to a certain extent. Whatever knocked down HcSmad3/4 or treated by specific inhibitors of Smad 3 (SIS3), expression levels of these genes displayed a significantly downregulated tendency compared with the wound group. In addition, histological evaluation suggested that Smad3 knockdown or SIS3 treatment was accelerated wound healing, and then Smad4 knockdown delayed the process of wound healing in mussels. These data implicate that Smad3/4 play an important role in tissue repair in mollusks.

1. Introduction

Wound healing is a complex and precise process in living organisms, the process can be divided into three main overlapping phases, including inflammation, proliferation and remodeling phase [1]. These phases are involved in complicated interactions between various cell types and the participation of multiple cytokines, growth factors and proteases [2]. Among these cytokines and growth factors, transforming growth factor beta (TGF-β) plays a fundamental role in wound healing [3,4]. For instance, excessive expression of TGF-β1 can exacerbate inflammation during wound healing [5]. TGF-β1 was necessary for skin wound healing by causing inflammation, promoting epithelial cell migration, stimulating skin fiber cell proliferation, collagen synthesis and extracellular matrix (ECM) remodeling [6]. TGF-β2 was required for expression and organization of collagen and other key ECM components during the healing process [7]. Exogenous addition of TGF-β3 could reduce the formation of scars in mouse skin wounds [8]. TGF-β signals are mediated by the classical Smad pathway, ligands first bind to the TGF-β II type receptor (TβRII) and then recruits, phosphorylates and activates I type receptor (TβRI) to form receptor complexes. Then, activated receptor complexes phosphorylated the receptor regulated Smad (R-Smad, Smad2 and Smad3). The activated R-Smad forms a heterologous complex together with common mediator Smad (co-Smad, Smad4). The resulting Smad complex is translocate into nucleus, in which it becomes a transcription factor that binds directly to Smadspecific DNA-binding motifs and modulates target gene transcription [9-11].

The Smad DNA-binding motif was first identified in the human type VII collagen gene (COL7A1) promoter, which binds activated Smad3 complex [12]. Subsequently, it was identified in promoter regions of massive TGF-β regulatory genes, including plasminogen activator inhibitor-1 [13], c-Jun [14], Smad7 [15] and many ECM encoding genes, such as COL1A2, COL3A1, COL5A2, COL6A1, COL6A3 [16], matrix metalloproteinase1 (MMP1), MMP7, MMP9, MMP13, membrane type MMP1, tissue inhibitors of metalloproteinase 1 (TIMP1), TIMP2 and TIMP3 [16,17], suggesting them as direct Smad target genes. Synthesis and deposition of ECM are a critical feature in the wounds healing. MMPs and TIMPs are major components involved in ECM metabolism,which play vital roles in wound healing [18]. For example, during human skin wound healing, MMP-1 released by keratinocytes promoted cell migration by binding to integrin and collagen I, thereby promoting re-epithelialization [19]. MMP 8 was upregulated in tissue remodeling processes [20]. Overexpress multiple MMPs contributed to re-epithelialization in humans and experimental animals wound tissues [21]. TIMP1 could be detected in the granulation tissue at 5th day post wounding [22]. The combination of MMP and TIMP regulated the synthesis and degradation of ECM at wound sites, their imbalance led to abnormal wound healing [23]. The ratio of MMP/TIMP contributes to the precise regulation of wound healing process.

The pearl mussel Hyriopsis cumingii, is widely used to cultivate highquality freshwater pearls, which has important economic value in China [24,25]. However, artificial nucleus insertion causes a high mortality rate in mussels, and wounds are prone to secondary infection [26]. Therefore, understanding the mechanisms of wound healing in pearl mussel are conducive to promoting wound healing and controlling bacterial infection and even contribute to the sustainable development of pearl industry. In previous work, Smad3 and 5 genes have been cloned from the H.cumingii, suggesting it was involved in wound healing [26]. Several MMPs and TIMPs genes also have been cloned from cDNA library of hemocytes of H.cumingii. In present study, we characterized a full-length of Smad4 from H. cumingii. Histological changes of wounds and the gene’s expression dynamics were evaluated and investigated after knockdown Smad3 and 4 and inhibition of Smad3. Cumulatively, these data will help us to explore the role of HcSmads in wound healing.

2. Materials and methods
2.1. Animal collection and wound model

Healthy mussels, weight 436 ± 15.28 g, length 17.13 ± 0.23 cm, were collected in a pearl farm in Duchang, Jiangxi province, China, and maintained in a plastic container in the laboratory with water temperature 25 ± 2 °C, pH 7.2 ± 0.1, continuous oxygen supplied using an oxygen pump, and were fed with Chlorella algae powder once every two days.To establish a wound healing model, the shells of live mussels were opened 0.8– 1 cm with a shell opener and supported with cork stoppers. Six u-shaped wounds were quickly scratched at the edge of the bilateral mantle (three per side) of each mussel with sterile surgical scissors, and then the cork was removed and the mussels were soaked in penicillinstreptomycin solution (ampicillin 100 U/mL; streptomycin sulfate 0.1 mg/mL) for 2 h, followed by put back to plastic container. After 1, 3, 5, 10 and 15 d, the tissues of 1 cm2 area around the incision was isolated with surgical scissors, three wound tissues were used to histological examination and three wound tissues were used to total RNA extraction.

2.2. Cloning of the Smad4 cDNA

Total RNA was extracted from mussel’s hemocytes by Trizol reagent (Invitrogen, USA). Total RNA (1 μg) was reverse transcribed to firststrand cDNA by smart cDNA synthesis kit (Clontech, USA). A cDNA fragment of HcSmad4 was obtained by mining transcriptome library of mussel hemocytes in our laboratory. A pair of specific primers HcSmad4-F and HcSmad4-R (Table 1) was utilized to verify the putative fragment of HcSmad4. RACE-PCR (rapid amplification of cDNA ends) combined with nested PCR was used to obtain 5′ and 3′ terminals of HcSmad4. The sequences of primers were showed in Table 1. All PCR reaction systems contained 18.4 μL of H2O, 2.5 μL of 10 × ExTaq buffer, 2 μL of dNTP mixture, 0.6 μL of each primer, 0.6 μL of cDNA, 0.3 μL of ExTaq (Takara, China, Dalian), The PCR conditions was 5 min denaturation at 94 °C, 35 cycles at 94 °C for 30 s, annealing at 56 °C (or 54 °C) for 30 s, 72 °C for 1 min, and final extension at 72 °C for 10 min.The PCR products were recovered by GenElute™ Gel Extraction Kit (Promega, USA), and then subcloned into pMD-18T vector (Takala, China, Dalian). PCR positive clones were sent to Sangon Biotech (Shanghai) Co., Ltd. for sequencing.

2.3. Sequence analysis

Amino acid sequence homologous search was performed using the Blastn program at the NCBI website (http://www.ncbi.nlm.nih.gov/ blast). Open reading frame (ORF), molecular mass and theoretical isoelectric point were predicted using corresponding procedures of Expert Protein Analysis System (http://www.expasy.org/). The Simple Modular Architecture Research Tool (SMART) (http://smart.emblheidelberg.de/) was used to predict type 2 immune diseases conserved domains. Multiple sequence alignment was performed with ClustalW 2.0 program (https://
www.ebi.ac.uk/Tools/msa/clustalw2/). The phylogenetic tree was constructed using the Neighbor-Joining (NJ) algorithm within MEGA version 6.0. The number at each branch represented the bootstrap (1000 replications) ratio.

2.4. RNA interference assay

The specific small interfering RNA (siRNA) targeting Smad3 (siRNA 42, siRNA 627, siRNA 1017) and Smad4 (siRNA 287, siRNA 589, siRNA 1086) and negative control oligonucleotide (NC) were designed and synthesized by Shanghai Gene Pharma Co., Ltd. (Table 2). All RNA oligos were dissolved in nuclease-free water to a final concentration of 20 μM. RNA interference efficiencyand effective interference time were tested before formal testing. Forty eight healthy mussels were randomly divided into three Smad3 interference groups, three Smad4 interference groups and two control groups, each group consisted of 6 individuals in per tank. A total of 125 μL different siRNA solution was injected into the adductor muscle of each mussel in experimental groups using microinjector. The same amount of NC were injected each individual in one group used as the negative control. Another group was left untreated and used as the blank control. At 3rd and 5th day after injection, three individuals were randomly sampled, and 0.1 g mantle tissues of each individual was cut with sterile scissors and stored in liquid nitrogen. Total RNA was extracted and cDNA was synthesized from each sample as aforementioned. Interference efficiency of each specific siRNA was evaluated by real-time PCR. According to Ref. [27]; a 70% inhibition efficiency was considered to be effective RNA interference. Therefore, siRNA with inhibition efficiency higher than this threshold was used in formal experiments. At the same time, when the target gene expression level returned to more than 30% of the original expression level, the interference was considered ineffective, and a booster injection was needed.

Forty five mussels were utilized for formal experiments, which were randomly divided into wound group, Smad3 interference plus wound group and Smad4 interference plus wound group, each group included fifteen individuals in per tank. Wound preparation was described as section 2.1. In interference plus wound groups, 125 μL validated effective small siRNA (Smad3-siRNA 627 or Smad4-siRNA 589) solution was injected into the anterior adductor muscle of the mussels using micro syringe, respectively. Untreated mussels were used as a blank control. Samples were taken on the 1st, 3rd, 5th, 10th and 15th day, respectively. According to effective interference time, the residual individuals were performed an enhanced injection with the same dose after the 5th and 10th day of sampling. Three mussels randomly were taken out at each time quantum from each group, 1 cm2 tissues around each wound were cut and three wound tissues from the same body were mixed together as a sample and placed in the liquid nitrogen. Total RNA of samples were extracted and used to synthesized cDNA as templates for further qRT-PCR. Another three wound tissues were used to histological examination to evaluate the effect on wound healing.

2.5. Inhibitors blocking

To further study the role of Smad3 deletion in wound healing, the specific inhibitor of Smad3 (SIS3, from Med. Chem. Express, USA) was used to block signaling pathway. SIS3 is the inhibitor of Smad3 Selleck Corn Oil which can blockades TGF-β/Smads signal transduction via the selective suppression of Smad3 phosphorylation, DNA binding, and the interaction of Smad3 with Smad4 [28]. RNA interference assayed and inhibitors blocking experiments were conducted simultaneously. The group contained 15 individuals with wounding were injected into the adductor muscle with 200 μL of SIS3 solution (10 mM, dissolved in dimethyl sulfoxide, DMSO) every two day as the experimental group. The control group was the same as the RNA interference experiment. Meanwhile, three untreated individuals were used as a blank control. After 1, 3, 5, 10 and 15 days, preparation of tissue samples, extraction of RNA and synthesis of cDNA was performed according to 2.4.

2.6. Quantitative real-time PCR

Expression change of Smads in different tissues after mussel wounded, temporal expression of Smads, MMPs and TIMPs in wound healing process after knockdown of Smad3 or Smad4 and inhibitors blocking were determined by quantitative real-time-PCR (qRT-PCR).Total RNA extraction and cDNA synthesis was conducted as described above. Housekeeping gene GAPDH from H.cumingii was selected as an internal control. Primers employed to amplify GAPDH, HcSmads, HcMMPs and HcTIMPs genes were listed in Table 1. QRT-PCR was performed on a CFX96 Real-time Detection System (Bio-Rad Laboratories, USA) by using the SYBR® Premix Ex Taq™ II (Tli RNaseH Plus) kit (Takara, Dalian, China). Amplification reactions volume contained 2 × SYBR Green Premix ExTaq 10 μL, each primer 0.4 μL, cDNA template 1.0 μL, RNA-free water 8.2 μL. The qRT-PCR cycling parameters was as follows: 95 °C denaturation for 5 min, followed by 40 cycles of 95 °C for 30s, 55 °C for 30s, 72 °C for 30 s. Dissociation curves were performed to evaluate whether each amplified product was single. Prior to the testing of each cDNA samples, the amplification efficiency of each of the primer pairs was validated (Table 1). All primer efficiencies were found to be within ± 10% of the efficiency of the reference gene (GAPDH, 96.4% efficiency). Each sample was run in triplicate in a 96-well plate along with the internal gene. Relative expression levels of each gene were calculated with the 2 −ΔΔCT method. All data were expressed as the mean ± s.d, and were analyzed by an unpaired Student’s t-test after normalization. The significant level was set at P < 0.05.

2.7. Histological staining

Wound tissues from section 2.1 and 2.4 were immediately fixed overnight with formaldehyde, dehydrated with gradient ethanol, cleared ethanol with the xylene and embedded in paraffin wax. Five micrometer sections were cut with automatic slicer and stained using a standard hematoxylin and eosin (H&E) method. Stained sections were then analyzed using a Nikon Eclipse E100 microscope connected to the Nikon DS-U3 digital camera and linked to a computer for digital image scanning analysis.

3. Result
3.1. Nucleic acid and amino acid sequence of HcSmad4

The full length of HcSmad4 cDNA sequence was 2543 bp, which contained a 5′ terminal untranslated region (UTR) of 119 bp, a 3’ UTR of 876 bp without polyadenylation signals (AATAAA) and poly (A) tail and an open reading frame (ORF) of 1548 bp. The sequence of HcSmad4 was deposited in GenBank under accession of MK043083. Predictive encoded peptide of HcSmad4 was 515 amino acids with a theoretical isoelectric point of 6.90 and molecular weight of 56.73 kDa (Fig. s1). It contained conserved N-terminal MH1 domain (18-138 residues), C-terminal MH2 domain (287-513 residues), a nuclear localisation signal (NLS, 31-49 residues), a nuclear export signals (NES, 137-144 residues), a Smad4 activation domain (SAD, 233-280 residues) and a DNA binding motif (DBM, 72-82 residues) (Fig. 1).

Fig. 1. Multiple sequences alignment of HcSmad4 and other Smad4. GenBank accession numbers for the analyzed sequences were as follows: MmSmad4, Mus musculus (AAM74472), HsSmad4, Homo sapiens (NP_005350), XlSmad4, Xenopus laevis (NP_001090536), DrSmad4, Danio rerio (ACA58502), CgSmad4, C. gigas (AHB37077). The MH1 and MH2 domain of HcSmad4 are marked with black boxes, respectively. Nuclear localisation (NLS), nuclear export (NES) signals and DNA binding motif (DBM) are marked with dashed boxes. HcSmad4 activation domain (SAD) was shown between brackets with proline residues typed in gray background.

The sequence of HcSmad4 showed highest identities with Mizuhopecten yessoensis, Crassostrea gigas and Lottia gigantea (85%, 72% and 74%, respectively) and a higher identity to Octopus bimaculoides and Branchiostoma floridae Smad4 (80% and 64%, respectively). Identities with mammalian Smad4 ranged from 59% to 60% (Fig. 1). Phylogenetic analysis showed that all selected Smad4 were clustered into two vertebrates and invertebrates branches. And it showed that all Smad4 from mollusks clustered together as well as diverged from their counterparts in other species (Fig. 2).

3.2. Tissue distribution and expression changes of HcSmads in diferent tissues after wounding

Transcripts of HcSmad3, 4 and 5 were constitutively expressed in all examined tissues. In healthy mussels (0d), the highest expression of these genes was in muscle, the lowest level was in hemocytes, the expression levels in hepatopancreas, gill and mantle were basically the same (Fig. 3). After mussels wounded, the expression of Smad3 was upregulated at 1st and 3rd day in hemocytes, at 3rd day in mantle and at 5th in gill (P < 0.05), a slight upregulation was found in hepatopancreas and muscle (P > 0.05) (Fig. 3A). The expression of Smad4 was significantlyupregulated at 5th day in hemocytes, at 3rd and 5th day in the mantle, and at 3rd, 5th and 10th day in gill (P < 0.05) Fig. 3B). The expression of Smad5 was significantlyupregulated at 1st, 3rd, 5th and 10th day in hemocytes (P < 0.05), and slightly upregulated in other tissues (P > 0.05), and then it recovered to the initial level at 15th day (Fig. 3C).

3.3. Efficiency of RNA interference

Interference efficiency of each specific siRNA was evaluated by qRTPCR. As shown in Fig. 4, the transcripts of HcSmad3 had decreased 21%, 15% and not even reduced at 3 days post-injection of siRNA 42 and siRNA 1017. And the mRNA level of HcSmad3 only decreased 27%, not even reduced at 5 days post-injection of siRNA 42 or siRNA 1017. But a mean significant reduction of 85% were observed in HcSmad3 mRNA level, relative to GAPDH mRNA level at day 5 post-injection of siRNA 627 compared with negative control RNA (NC) oligo. There had no reduction in HcSmad4 mRNA levels after injection of those siRNAs by day 3. However, the HcSmad4 mRNA levels reduced 73%, 70% and 50% at day 5 after injection of siRNA 589, siRNA 287 and siRNA 1086 compared with NC. Therefore, siRNA 627 and siRNA 589 were used further study.

Fig. 2. Phylogenetic tree of HcSmad4 and related proteins. Tree was constructed based on amino acid sequences by MEGA 6.0 software with a neighbor-joining method. The number at each branch represents the bootstrap (1000 replications) ratio.

The sequences numbers of the selected species were as follows: Ovis aries (NP_001254815), Bos taurus (NP_001069677), Corvus cornix cornix (XP_020439875), Lottia gigantea (BAQ19237), Crassostrea virginica (XP_022311791), Oncorhynchus kisutch (XP_020322945), Mizuhopecten yessoensis (XP_021379469), Octopus bimaculoides Smad4 like isoform X6 (XP_014780400.1), O. bimaculoides Smad4 like isoform X7 (XP_014780401.1), the sequences numbers of unlisted species were shown in Fig. 1.

3.4. Gene expression analysis after interference Smad3/4 and inhibitor SIS3 treated

Expression trends of these genes in wounded, knockdown and inhibitor blocking experiments were shown in supplement Table 1. Post wounded, compared with blank control, transcripts of HcSmad3 were upregulated 3.1-fold at day 10 (p < 0.05) (Fig. 5A); Smad4 mRNA levels significantly increased to 2.08 and 2.17 fold at day 3 and 10 (p < 0.05) (Fig. 5B); Smad5 mRNA were significantly increased to 2.71, 2.15, 1.98 and 2.4 fold at day 1, 3, 5, 10 (p < 0.05), and then it restored to the initial level at day 15 (p > 0.05) (Fig. 5C). Transcripts of HcMMP1, HcMMP19, HcTIMP1 were significant upregulated at all time points (p < 0.05) (Fig. 5D, E, F). HcTIMP2 mRNA level was 1.5 and 2.5 fold at day 1 and 5 that of control group, respectively (p < 0.05) (Fig. 5G).

After Smad3 knocked down, compared with wounded group, expression levels of HcSmad4 had no significant change at the corresponding time point (P > 0.05) (Fig. 5B). HcSmad5 expression was reduced 1.59 and 2.11-fold at day 3 and day5 (P < 0.05) (Fig. 5C). HcMMP1 mRNA expression was decreased 1.83 and 4.53-fold at day 3 and 10, but with a 6.66 fold increase at day 15 (p < 0.05) (Fig. 5D). Transcripts of HcMMP19 were significantly decreased 1.52, 2.68 and 1.96-fold at day 1, 5 and 10, and increased 1.41 and 6.30-fold at day 3 and 15 (p < 0.05) (Fig. 5E). TIMP1 mRNA levels were downregulated 2.15 and 2.03-fold at day 1 and 5 (p < 0.05), however, it was increased 10.01-fold at day 15 (p < 0.05) (Fig. 5F). HcTIMP2 mRNA decreased 1.65-fold at day 5 and increased 5.16-fold at day 15 (p < 0.05). No significant changes were observed at other time points (p > 0.05) (Fig. 5G).

After knocking down Smad4, compared with wound group, there was a small increase (p > 0.05) in the expression of Smad3 at day 1 and 3, and it was basically same as the wound group at day 5, 10 and 15 (Fig. 5A). The expression of HcSmad5 was declined 1.50 and 2.17-fold at day 3 and day 5 (P < 0.05) (Fig. 5C). Expression of MMP1 were decreased at all time points except the 3rd daywhen it increased to 5.7-fold (P < 0.05) (Fig. 5D). MMP19 expression also decreased significantly at day 3, 5 and 10 (P < 0.05), but it rose 3.6-fold at day 15 (P < 0.05) (Fig. 5E). HcTIMP1 mRNA was significantly downregulated 1.85 and 1.52-fold at day 5 and 10, however, it was significantly increased to 4.6 and 2.9-fold at day 3 and 10 (p < 0.05) (Fig. 5F). In addition to a significant 2.58-fold drop at day 1, expression of TIMP2 increased 3.94, 1.72, 1.73-fold at day 3, 5, 15, respectively (Fig. 5G).

After inhibitor SIS3 treatment, mRNA expression of HcSmad4 was decreased 4.80 and 3.77 at day 3 and 10 compared with wound group (p < 0.05) (Fig. 5B). Similarly, HcSmad5 mRNA was significantly downregulated 4.18, 5.15, 1.61 and 3.87-fold at day 1, 3, 5, 10 (p < 0.05) (Fig. 5C). HcMMP1 significantly reduced 18.61, 3.26, 5.10, 4.81 and 2.90-fold at day 1, 3, 5, 10 and 15 (p < 0.05) (Fig. 5D). HcMMP19 was decreased 34.92, 3.60, 4.10 and 2.69-fold at day 1, 5, 10, and 15 (p < 0.05) (Fig. 5E). HcTIMP1 mRNA was downregulated 23.50, 1.56, 2.03 and 2.74-fold at day 1, 3, 5, 10 and then increased 2.66-fold at day 15 (p < 0.05) (Fig. 5F), HcTIMP2 mRNA was reduced to 4.46 and 3.83-fold at day 1 and 10 compared with wound group (p < 0.05) (Fig. 5G).

3.5. Histological evaluation

In the wound group, a small amount of hemocytes along with pink eosinophilic cells were aggregated at the incision site on the 1st day (Fig. 6A). The 3rd day of wounding, the number of aggregated hemocytes were increased, a little pink eosinophilic cells and longitudinal muscles were found at the wound site (Fig. 6B). At 5th day, more hemocytes were observed at the excision site (Fig. 6C). At day 10, the newly formed hemocyte cells began to be integrated in the incision, longitudinal muscles and connective tissues were continued to accumulate (Fig. 6D). At day 15, newly generated epithelial cells completely sealed the wound (Fig. 6E); Compare with wound group, more hemocytes were gathered at the edge of wounds in HcSmad3-siRNA group at the first day, freshly generated pink eosinophilic cells and muscles also occurred at the wound site (Fig. 6F), At day 3, a large number of hemocytes continued to aggregate in HcSmad3-siRNA group (Fig. 6G), By 5 days, the wound site was completely plugged by hemocytes, forming a distinct layer of blood cells in HcSmad3-siRNA group (Fig. 6H), At day 10, the newly formed epithelial cells had not yet reached the columnar shape, but they are similar to columnar epithelial cells and have oval or cylindrical nuclei in Smad3 knockdown group (Fig. 6I). On day 15, The length and area of the healing site increased, which contained more connective tissue, strong staining of columnar epithelial cells and slender nucleus in the Smad3 knockdown group (Fig. 6J); In SIS3 inhibition group, more haemocytes accumulated around the wound, pink eosinophilic cells and muscles also occurred which similar to the Smad3-knockdown group after one day (Fig. 6K). Hemocytes had a sharp increased trend near the margin of the wound on day 3 (Fig. 6L). Meanwhile, a layer of hemocytes could be obvious seen in SIS3 inhibition group; In addition, connective tissues and pink eosinophilic cells continue to increase by 5 days after excision (Fig. 6M). Newly generated epithelial cells completely aggregated at the incision site, and pink eosinophilic cells continue to grow on day 10 (Fig. 6N). After 15 days, the epithelial cells assumed a more regular shape in the SIS3 inhibition group (Fig. 6O). In Smad4-siRNA group, hemocytes and muscles were less than those of wound group, HcSmad3siRNA group and the SIS3 inhibition group on 1st day (Fig. 6P). Hemocyteshad a significant trend of rising in HcSmad4-siRNA group after 3 days (Fig. 6Q). Hemocytes and muscles kept on increasing by 5 days in Smad4 knockdown group (Fig. 6R). After 10 days, a great many of hemocytesaccumulated toward the edge of the wound (Fig. 6S). But it only formed epithelial cell layer in the Smad4 knockdown group on day 15 (Fig. 6T); Generally, the healing rates of the Smad3 knockdown group and the SIS3 inhibition group were faster than the Smad4 knockdown group.

Fig. 3. Expression change of HcSmad3 (A), 4 (B) and 5 (C) in various tissues after mussels wounded. Samples from 0 h represented tissue distribution, and the mantle were from non-wound sites. GAPDH was used as the reference gene for normalization. The expression levels of HcSmads at 0 h in hemocytes were used to calibrate other tissues. Data were shown as means ± SD (n = 3). The differences were statistically significant at p < 0.05 level. ‘*’ indicated significant difference in comparison to 0 h for the corresponding tissues.

Fig. 4. Analysis of interference efficiency of different siRNAs. The muscles were injected with HcSmad3 siRNAs, HcSmad4 siRNAs and negative control (NC) RNA oligo. Expression of HcSmad3 mRNA (A) and HcSmad4 mRNA (B) were quantified by qRT-PCR. GAPDH was used as an internal control. Vertical bars represented the means ± SD. (n = 3).

4. Discussion

Smad4 plays a central role in signal transduction of TGF-β family members through forming R-Smads transcription complexes [29]. Here, a putative Smad4 gene was identified and characterized from pearl mussel H. Immune landscape cumingii. Several typical structure features were necessary to maintain the function of Smad4, including NLS and NES for nuclearcytoplasmic shuttling, a MH1 domain for DNA binding and MH2 domain for homo or hetero-oligomerization, transcriptional activation and nuclear location [30,31], a proline-rich Smad4 activation domain (SAD) located the linker region maximizes the activation of Smad-dependent transcriptional responses [32,33]. All these features were also present and conserved in HcSmad4. The carboxyl terminal of Co-Smad had no SxxS motif, which was also not found in HcSmad4. In summary, according to the amino acid sequence features and its domain structure, HcSmad4 could be attributed to the Co-Smad family of transcriptional regulators.In mammal, Smad4 mRNA was widely expressed in various tissues [34,35]. Smad4 of zebrafish was widely expressed in the skin, gills, kidneys, ovaries, heart, eyes, fins, liver,brain, swim bladder, muscles, gallbladder, guts and testis, with a high expression in muscles suggesting that smad4 plays an important role in heart and skeletal muscle development [36]. Four types of Smad4 from the common carp exhibited different expression patterns in different tissues suggesting its own function [37]. In the present study, the expression pattern of HcSmad4, like that of HcSmad3 and 5, was also widely expressed in detected tissues, which was largely consistent with findings from previous studies [26], indicating HcSmad4 may be had diverse function in physiological processes of molluscs. The highest express levels in muscles suggested that Smad4 may be related to adductor muscles development and shells formation.

Fig. 5. Expression of Smad3 (A), Smad4 (B), Smad5 (C), MMP1 (D), MMP19 (E), TIMP1 (F) and TIMP2 (G) at different time periods in black group, wound group,Smad3-siRNA plus wound group, Smad4-siRNA plus wound group, SIS3 plus wound group. Expression level of genes were standardized according to reference GAPDH levels. The blank control group represents the expression of each gene in untreated condition. Vertical bars represent the means ± SD (n = 3). Significant differences compared to the control groups was marked with an asterisk “*” at P < 0.05 and an asterisk “**” at P < 0.01.

Fig. 6. Histological observation of mantle in H. cumingii at different time point after wounding. (A-E) wound group, (F-J) HcSmad3 knockdown group, (K-O) SIS3 treatment group, (P-T) HcSmad4 knockdown group. h: hemocytes; es: eosinophilic cells; ws: wound site; hl: hemocyte layer; ep: epithelial cell; ct: connective tissue; lm: longitudinal muscles. Original magnification × 20.

The expression of Smad3 and 4 showed a slight change in rabbit ear rim trauma model [38]. After injury, transcripts of Smad3 from P. fucata and Smad4 from Biomphalaria glabrata were upregulated, indicating that cell proliferation was activated [39,40]. In this study, the transcript of HcSmad3 was slightly upregulated only at 10th day, while expression of Smad5 was upregulated at multiple time points after mussels wounding, this was consistent with our previous reported [26]. There was a slight difference in the expression pattern of smad3/4/5 in mantle after wounding in tissue expression and knockdown experiments, which may be related to different sampling sites (Figs. 3 and 4). Previous reports indicated that the expression of diverse MMPs and TIMPs were upregulated after tissue injury [41-43]. The increased expression of two kinds of MMPs (MMP23659 and MMP54089) from P. fucata suggested that they may play a role in ligament regeneration and shell wound repair [44]. The mRNA expression of TIMP from C. gigas was increased in hemocytes after shell injury or bacterial
challenge, suggesting that Cg-timp was involved in wound healing and defense mechanism [45]. In this study, the expression of HcMMP1, HcMMP19, HcTIMP1 and HcTIMP2 were significantly
upregulated after wounding, suggesting the involvement of them in wound healing of H. cumingii.

Recent studies have revealed that knocking down Smads could significantly affect the expression of downstream target genes. For example, deletion of Smad3 completely blocked the increased expression of TIMP1 mRNA induced by TGF-β1 in mice [46]. The expression levels of MMP2, TIMP1 significantly decreased after Smad3 silencing [47]. Specifically, the activity of MMP9 began to decrease at 12 h after knockout of Smad3 and continuing for seven days, while increased at the subsequent time points (21 and 36 days) during wound closure [48]. Smad3-deficient myofibroblasts exhibited significant downregulation of TIMP-1 [49]. Smad4 knockdown via RNAi abolished TIMP-3 augmentation by TGF-β in human chondrocytes [17]. Similarly, MMP3, 12, 13 and 14 were found to be significantlyupregulated in skin prior to wounding and throughout the healing in Smad4 KO skin and wounds [50]. In this study, whether knocked down Smad3 or Smad4, both transcriptional expression of HcMMP1, HcMMP19, HcTIMP1 and HcTIMP2 showed a downward trend in the initial stage (3rd, 5th and 10th days). These results indicated their expression levels were inhibited by Smad transduction pathway. A recent report revealed that SIS3 administration significantly downregulated the phosphorylation of Smad3 and reduced the expression level of TIMP1 in the pulmonary fibrosis of mice [51]. SIS3 can inhibit TGF-β1-induced TIMP-3 expression by inhibition of Smad3 phosphorylation [52]. Our study showed that SIS3 treatment can significantly reduce the expression of HcMMPs and HcTIMPs, which was consistent with the above results.

Healing processes of mantle tissues of H. cumingii have been describedby histological methods after wounding [53]. Our results were basically consistent with the report. The significantly increased hemocytes and connective tissue, as well as the accelerated formation of hemocytes layer and epithelial cell layer, suggested that Smad3 knockout and inhibition had accelerated the wound healing of mussels. Similarly, some wound tissues from Smad3 null mice, including cutaneous, brain and tail wound healing speed of Smad3-null mice was faster than that of wild-type mice [6,48,54,55]. On the contrary, other evidence suggested that exogenous Smad3 could promote wound healing of rabbit ear wounds [56]. These data suggested that normal wound healing was associated with inhibition of Smad3 levels, but the complete loss of Smad3 may further accelerated the wound healing process and an impaired local inflammatory response [54].

Loss of Smad4 could have vast implications during cutaneous wound healing because it mediates all TGF-β superfamily signaling [57]. Smad4-deficient epithelial cells have delayed cell migration [58]. Smad4-null could delay the closure and remodeling of mouse skin wound [50]. However, mutation of Smad4 resulted in proliferation of keratinocyte, acceleration of epithelialization and faster wound healing in the epidermis of mice [59]. In this study, after knockdown Smad4, the number of hemocytes and the increase of connective tissue was less than the wound group. The hemocyteslayer was also formed later than in the wound group (15th vs 10th days), suggesting Smad4-knockdown delayed wound healing and slower re-epithelialization. Loss Smad4 could lead to accumulation of myofibroblasts in the stromal, and the accumulation myofibroblasts could prevent normal ECM molecule deposition during wound reconstruction, leading to delayed scar resolution, this may be one of the mechanisms by which Smad4 silencing delayed wound healing [50,57].