TGF‐β isoforms inhibit hepatitis C virus propagation in transforming growth factor beta/SMAD protein signalling pathway dependent and independent manners

Abstract Transforming growth factor beta (TGF‐β) plays an important role in the viral liver disease progression via controlling viral propagation and mediating inflammation‐associated responses. However, the antiviral activities and mechanisms of TGF‐β isoforms, including TGF‐β1, TGF‐β2 and TGF‐β3, remain unclear. Here, we demonstrated that all of the three TGF‐β isoforms were increased in Huh7.5 cells infected by hepatitis C virus (HCV), but in turn, the elevated TGF‐β isoforms could inhibit HCV propagation with different potency in infectious HCV cell culture system. TGF‐β isoforms suppressed HCV propagation through interrupting several different stages in the whole HCV life cycle, including virus entry and intracellular replication, in TGF‐β/SMAD signalling pathway–dependent and TGF‐β/SMAD signalling pathway–independent manners. TGF‐β isoforms showed additional anti‐HCV activities when combined with each other. However, the elevated TGF‐β1 and TGF‐β2, not TGF‐β3, could also induce liver fibrosis with a high expression of type I collagen alpha‐1 and α‐smooth muscle actin in LX‐2 cells. Our results showed a new insight into TGF‐β isoforms in the HCV‐related liver disease progression.

but also promotes SMAD-dependent cascades by phosphorylating receptor-regulated SMADs (including SMAD2 and SMAD3), which subsequently form active SMAD complexes with SMAD4 and regulate the transcription of target genes, such as type I collagen alpha-1 (COL1A1), vascular endothelial growth factor A (VEGFA) and interleukin-17A (IL-17A). [1][2][3][4] Recently, accumulated evidence indicated that activated TGFβ signalling is involved in various cellular processes, such as cell recognition, differentiation, proliferation and apoptosis. 3 However, the role of different TGFβ isoforms in the development of liver disease induced by hepatitis C virus (HCV) infection and the underlying mechanisms remain largely unclear.
In the liver tissue, TGFβ isoforms are produced and secreted not only by non-parenchymal cells but also by hepatocytes. The expressions of TGFβ isoforms are promoted by HCV infection in cultured hepatocytes and in livers of HCV-infected patients, which might be caused by HCV-induced endoplasmic reticulum stress, unfolded protein response activation or HCV core expression. [5][6][7][8] Although TGFβ plays essential roles in the liver disease progression, including initial liver injury, fibrosis, cirrhosis and hepatocellular carcinoma, 9 it is demonstrated that TGFβ suppressed HCV RNA replication and protein expression depending on TGFβ/SMAD signalling pathway in HCV sub-genomic replicon system, and TGF-β1 could directly suppress hepatitis B virus (HBV) replication in the cell culture system. 10,11 Furthermore, though the three isoforms of TGFβ exhibit significant sequence homology (71%-79% identity, Figure 1A), they have opposite effects on fibrosis disease progression, with TGF-β1 promoting fibrosis while TGF-β3 having anti-fibrotic effect. 1,2,6,12 Here, we investigated the antiviral activities and the underlying mechanisms of TGFβ isoforms by which they inhibited HCV propagation.

| Quantification of HCV RNA and mRNA
The cellular RNAs harvested by RNeasy Mini Kit (QIAGEN) from Huh7.5 cells were detected using the AgPath-ID One-Step RT-PCR Kit (Applied Biosystems). Fluorescent signals were identified with F I G U R E 1 TGFβ isoforms inhibit HCV propagation in HCVcc system. A, Sequence alignment of C-terminal domains of mature human TGFβ isoforms. Green shading represented the amino acid residues contacting with TβRII; grey shading represented the cysteine residues that formed the inter-chain disulphide; open box designated the residues of α-helix 3 in TGFβ isoforms; the homology sequence of TGFβ isoforms was indicated in red text. B, Huh7.5 cells were treated with HCV at different MOI for 72 h or at MOI = 0.2 for 0, 3 days or 6 days, and then TGFβ isoforms mRNA or protein were quantified by qRT-PCR or WB analysis. Huh7.5 cells were infected with HCV for 72 h in the presence of TGFβ isoforms or sofosbuvir. Intracellular proteins were extracted and detected by WB (C) and the cytotoxicity without HCV infection was identified using an MTT assay (D). E, Huh7.5 cells were incubated with HCV viral stock at 10-fold dilutions ranged from 10 −1 to 10 −8 , and simultaneously treated with 0.5 nmol/L of TGF-β1, TGF-β2 and TGF-β3. At 72 h, HCV-positive wells were examined using a microscope after immunostaining against HCV core protein (left). Huh7.5 cells were inoculated with HCV viral stock and 0.5 nmol/L of TGF-β1, TGF-β2 or TGF-β3 for 24 h. The newly released infectious virus particles were collected at 48 h and then quantified by TCID 50 assay (right). Data represent the mean ± SD of three independent experiments. ANOVA analysis followed by the Student's t test method was used. *P < 0.05, **P < 0.01, ***P < 0.001 vs the control group

| Determination of anti-HCV activity
Huh7.5 cells were inoculated with HCV viral stock (MOI = 0.5) and simultaneously treated with TGFβ isoforms or sofosbuvir. At 72 hours, intracellular proteins were harvested by the CytoBuster Protein Extraction Reagent (PER, Novagen) containing 1 mmol/L protease inhibitor cocktail (Roche Applied Science) and HCV core protein level in cells was determined by WB analysis.
At 72 hours post-infection, HCV-positive wells were examined using a microscope after immunostaining against HCV core protein. 16 To identify the effects of TGFβ isoforms on de novo HCV particle production, Huh7.5 cells were inoculated with HCV viral stock (MOI = 0.7) and 0.5 nmol/L of TGF-β1, TGF-β2 or TGF-β3 for 24 hours. The newly released infectious virus particles were collected at 48 hours and then quantified by 50% tissue culture infective dose (TCID 50 ) assay. The infectivity titre was shown as TCID 50 per mL. [17][18][19] GS4.3 cells were treated with 0.5 nmol/L of TGFβ isoforms or sofosbuvir (0.5 μmol/L) for 72 hours. Then, intracellular proteins were extracted by PER and HCV NS3 protein level was detected by WB analysis.

| HCV entry inhibition and time-ofaddition assays
To evaluate the antiviral effective stages of TGFβ isoforms, Huh7.5 cells were pre-treated with TGFβ isoforms (0.5 nmol/L) for 2 hours followed by HCV infection for 2 hours, simultaneously treated with TGFβ isoforms (0.5 nmol/L) and HCV for 2 hours, or infected with HCV (MOI = 0.5) for 2 hours followed by TGFβ isoforms (0.5 nmol/L) 2 hours treatment.
At 72 hours, intracellular HCV core protein level was detected by WB analysis. In order to further characterize the inhibition stages, a time-ofaddition assay was performed. In brief, Huh7.5 cells were inoculated with HCV (MOI = 0.5) at 37°C for 2, 4, 6, 8 hours or 10 hours, and then, the HCV-containing medium was replaced by TGFβ isoforms (0.5 nmol/L) or sofosbuvir (0.5 μmol/L) containing medium. After 2 hours of treatment, the medium was replaced by fresh medium. At 72 hours, HCV core protein level in cells was determined by WB analysis. were harvested and HCV core protein level was determined by WB.

| The inhibitory activities of TGFβ isoforms on HCV particle release
Huh7.5 cells were incubated with HCV viral stock (MOI = 0.5) for 16 hours, and then, the cells were treated with TGFβ isoforms (0.5 nmol/L), sofosbuvir (0.5 μmol/L) or 0.1% BSA for 48 hours. The culture medium was replaced by adding fresh culture medium. After 24 hours, the cell viability was tested using MTT assay, the culture medium was collected for incubation to naïve Huh7.5 cells, and intracellular RNA was extracted and quantified by qRT-PCR. Meanwhile, naïve Huh7.5 cells were incubated with above culture supernatants for 72 hours. Then, the intracellular RNA of Huh7.5 cells was extracted and quantified by qRT-PCR.

| Western blot analysis
The Western blot analysis was performed as previously described. 21 In brief, equal amounts of protein from cellular lysates were subjected to SDS-PAGE and electroblotted onto transmembrane (Millipore). Proteins were probed with antibodies against HCV core, TGF-β1, TGF-β3, His tag, TβRI or SMAD2/3, respectively. β-Actin antibody was used as an internal control. Proteins were detected by chemiluminescence with Immobilon Western Chemiluminescent HRP Substrate (Millipore) and visualized by ChemiDo XRS gel imager system (Bio-Rad). Protein band intensity was analysed by Gelpro32 software. The ratio of interested protein to internal control protein β-Actin was calculated and normalized as 1.00 for the control group.

| Statistical analysis
The data were presented as means ± standard deviation (SD) more than three independent experiments. Statistical analyses were performed in GraphPad Prism 7 (GraphPad) using ANOVA analysis followed by the Student's t test; P < 0.05 was considered as statistically significant.

| TGFβ isoforms inhibit HCV propagation in vitro
In infectious HCV cell culture (HCVcc) system, the three TGFβ iso- and protein levels were relatively higher compared to TGF-β1 and TGF-β3 ( Figure 1B), which was consistent with previous studies. [4][5][6][7][8]22 Then, we confirmed whether TGFβ isoforms still have the anti-HCV activities in HCVcc system. Following HCV infection and treatment of different concentrations of TGFβ isoforms for 72 hours, the level of HCV core protein in Huh7.5 cells was quantified with WB analysis. The results showed that all of the three TGFβ isoforms significantly reduced HCV core protein level in a dose-dependent manner with TGF-β1 possessing the best anti-HCV effect ( Figure 1C), though a slightly detectable cytotoxicity was detected with an MTT assay without a significant dose-dependent manner ( Figure 1D), which might be induced by the inhibition of TGFβ to tumour cells. 23 To further confirm the anti-HCV activities of TGFβ isoforms, we analysed the HCV infectivity after TGFβ isoforms treatment using im-

| The distinct anti-HCV mechanisms of TGFβ isoforms lead to their additional anti-HCV effects
To investigate which HCV viral cycle was interrupted by TGFβ iso- inhibited HCV propagation in the whole viral life cycle, including before, during and after virus entry (Figure 2A), while TGF-β2, except TGF-β3, inhibited HCV propagation after virus entry (Figure 2A).
To further explore the anti-HCV effective stages of TGFβ isoforms, we performed time-of-addition experiments by adding TGFβ isoforms for 2 hours after 2, 4, 6, 8 and 10 hours of HCV infection.
As shown in Figure 2B, TGF-β1 showed similar HCV inhibition effects in all time-of-addition stages, which was consistent with our above results (Figure 2A). However, TGF-β2 and TGF-β3 treatment for 2 hours after 2, 4, 6, 8 and 10 hours of HCV infection had lower anti-HCV activities than those of 72 hours treatment ( Figure 1C).
These results suggest that TGF-β1 might down-regulate HCV propagation at several stages of the whole HCV life cycle, while TGF-β2 and TGF-β3 might mainly interrupt HCV entry into cells.
We then detected whether TGFβ isoforms inhibited HCV entry via direct action on viral particle. HCV virus stock was pre-incubated with TGFβ isoforms for 2 hours, respectively, and then inoculated to Huh7.5 cells after diluted pre-incubation mixture 25 times. As a control, the untreated HCV performed in parallel in the presence of TGFβ isoforms was also inoculated to Huh7.5 cells. The results exhibited TGF-β2 and TGF-β3, except TGF-β1, had higher inhibitory effects following pre-incubation with HCV prior to inoculation ( Figure 2C), suggesting that TGF-β2 and TGF-β3 might directly act on HCV particles.
To validate whether TGFβ isoforms might interrupt intracellular HCV replication after virus entry, we first applied GS4.3 cells, a human hepatoma Huh7 cell line, which carried an HCV sub-genomic replicon I 377-3'del.S. 24 In replicon cells, the three TGFβ isoforms (0.5 nmol/L) inhibited HCV replication at protein level ( Figure 2D) and TGF-β3 showed the best inhibitory effect, which was equal to that of sofosbuvir (0.5 μmol/L), an NS5B polymerase inhibitor. These data demonstrated the TGFβ isoforms also interrupted HCV propagation in viral replication stage, especially TGF-β3 ( Figure 2D), which was similar to the results from other HCV sub-genomic replicon system cured MH14 cells. 10 However, due to the low anti-HCV capacities of TGF-β2 and TGF-β3 after 2 hours treatment, we speculated TGF-β2 and TGF-β3 might indirectly interrupt HCV propagation in viral replication stage through indirect impacts, such as activating signalling pathways.
Then, we analysed the inhibitory effects of TGFβ isoforms on the production of infectious virion. After 16 hours of HCV incubation, the Huh7.5 cells were treated with TGFβ isoforms for 48 hours.
The supernatants were replaced by fresh medium and then transferred to naïve Huh7.5 cells to measure viral infectivity at 72 hours ( Figure 2E, left). Similarly, TGFβ isoforms and sofosbuvir significantly decreased intracellular ( Figure 2E, left, viral propagation) and extracellular ( Figure 2E, left, infectious HCV) HCV levels with no or slightly cytotoxicity ( Figure 2E, left, cell viability). However, when compared the ratio of extracellular and intracellular HCV levels with that of control, they all showed no significance ( Figure 2E, right), suggesting the isoforms could not inhibit HCV particle secretion.
Furthermore, there was an interesting phenomenon. In Figure 2E, we investigated the inhibition effects of TGFβ isoforms on intracellular viral propagation and extracellular infectious HCV by quantifying the HCV mRNA to identify the inhibitory activities of TGFβ isoforms on HCV particle release. The results highlighted TGF-β3 could more strongly inhibit HCV mRNA level than TGF-β1 ( Figure 2E). But in Figure 2A, we investigated which HCV viral cycle was interrupted by TGFβ isoforms by analysing the anti-HCV effects of TGFβ isoforms with treatment for 2 hours. The data confirmed 2 hours pre-treatment of TGF-β1 showed the stronger anti-HCV activities than TGF-β3. The phenomenon revealed distinct anti-HCV mechanisms of TGFβ isoforms and the anti-HCV activity of TGF-β3 might depend on the treatment time.
All the above results suggested that TGFβ isoforms could suppress HCV propagation at different stages of HCV life cycle. To further confirm that TGFβ isoforms inhibiting HCV propagation can be through different mechanisms, we investigated their antiviral activities in combination. The combined treatment with TGFβ isoforms could enhance the anti-HCV effects ( Figure 2F). When calculated with the improved Bürgi formula, 20 the q value was 1.04 for combined treatment with TGF-β1 and TGF-β3, 1.01 with TGF-β1 and TGF-β2, and 0.92 with TGF-β2 and TGF-β3, respectively, suggesting that combined use of the three isoforms could additionally inhibit HCV propagation. Certainly, the anti-HCV activity was also stronger in the group of combined treatment with the three isoforms ( Figure 2F) and the additional concentration of each isoform also increased the anti-HCV activities ( Figure 2F). However, the anti-HCV activity of the additional each isoform was relatively lower than that of the combined treatment ( Figure 2F). Therefore, these data further highlighted that TGFβ isoforms might exhibit their anti-HCV activities through distinct mechanisms and could obtain an additional anti-HCV effect.

| TGFβ/SMAD signalling pathway is essential for interrupting HCV propagation by TGF-β1 and TGF-β2, except TGF-β3
Previous data suggested that the antiviral effect of TGFβ might be associated with TGFβ/SMAD signalling in HCV sub-genomic replicon system. 10 To further investigate whether TGFβ/SMAD signalling was involved in anti-HCV effects of TGFβ isoforms in HCVcc system, we first used a serial of inhibitors. LY2109761, an orally active TβRI/II kinase dual inhibitor, dramatically reduced the anti-HCV activities of TGF-β1 and TGF-β2, except TGF-β3 ( Figure 3A). Pretreatment with RepSox, an inhibitor of TβRI, also showed a significant reduction of the anti-HCV effects of TGF-β1 and TGF-β2 with a negligible effect on TGF-β3 ( Figure 3B). Similarly, silencing TβRI in Huh7.5 cells with siRNA for TβRI also inhibited the anti-HCV effects of TGF-β1 and TGF-β2, except TGF-β3 ( Figure 3C). Cumulatively, these data showed TβRII and TβRI, which is one of the key factors associated with TGFβ/SMAD signalling pathway, played a pivotal role in anti-HCV activities of TGF-β1 and TGF-β2 other than TGF-β3.
To gain insight into the effect of TGFβ/SMAD signalling pathway on the anti-HCV activities of TGFβ isoforms, we further used siRNAs against human SMAD2/3 to assess the contribution of SMAD2/3 to the anti-HCV effects. Similarly, down-regulation of SMAD2/3 protein expression significantly decreased the anti-HCV capacities of TGF-β1 and TGF-β2, except TGF-β3 ( Figure 3D).

| Residues of Arg25, Val92 and Arg94 and dimer formation mediated by Cys77 or Cys109 residue are responsible for the anti-HCV activities of TGFβ isoforms independent of TGFβ α-helix 3
In order to investigate which regions or amino acid residues of TGFβ isoforms were responsible for their anti-HCV capacities, we first compared the antiviral activities between overexpressed TGFβ isoforms and with His tag at C-terminal. However, we found the plasmid pcDNA3.1(+)-TGF-β2 could not well express TGF-β2 in Huh7.5 cells as TGF-β1 and TGF-β3 did. Therefore, we only test the ant-HCV effects caused by overexpressing TGF-β1 and TGF-β3 to investigate which regions or amino acid residues of TGFβ isoforms were responsible for their anti-HCV capacities.
Overexpressing TGF-β1 and TGF-β3 with His tag at C-terminal had equal anti-HCV effects as TGF-β1 and TGF-β3, respectively ( Figure 4A), suggesting that fusing His tag did not affect the functions of TGFβ isoforms.
Cys77 or Cys109 in TGFβ isoforms is important to form homodimer for activation of their signalling activities. 25,26 In HCV-infected Huh7.5 cells, Cys77 substituted with serine or Cys109 exchanged by tyrosine completely eliminated the anti-HCV activities of TGF-β1 and TGF-β3 ( Figure 4B,C), indicating that dimer formation mediated by Cys77 or Cys109 residues was responsible for the anti-HCV activities of TGFβ isoforms.
The amino acid residues Arg25, Val92 and Arg94 of TGF-β1 and TGF-β3 are responsible for their high-affinity interaction with TβRII, once they are exchanged with Lys25, Ile92 and Lys94, respectively, TGFβ signalling activities of TGF-β1 and TGF-β3 are reduced. 27 Though TGF-β1 and TGF-β3 significantly inhibited F I G U R E 3 TGFβ/SMAD signalling pathway is essential for anti-HCV activities of TGF-β1 and TGF-β2, except TGF-β3. HCV core level was determined by WB in Huh7.5 cells treated after 20 μmol/L LY2109761 (A) or 50 μmol/L Repsox (B) 6 h treatment prior to coincubation with 0.5 nmol/L of TGFβ isoforms and HCV. At 72 h, intracellular HCV core was detected by WB. Huh7.5 cells were transfected with siRNA for TβRI (C) or siRNA for SMAD2/3 (D). After 48 h, the cells were sub-cultured and then infected with HCV and simultaneously treated with 0.5 nmol/L TGFβ isoforms after 24 h incubation. Intracellular HCV core was detected by WB after 48 h. **P < 0.01 and ***P < 0.001 vs solvent control or siRNA control plus solvent control group; # P < 0.05, ## P < 0.01 and ### P < 0.001 vs LY2109761 (Repsox or siRNA groups) control. Data represent the mean ± SD of three independent experiments. ANOVA analysis followed by the Student's t test method was used HCV propagation, TGFβ isoforms with mutants at R25K, V92I and R94K inversely promoted HCV propagation when the plasmids were introduced into HCV-infected Huh7.5 cells ( Figure 4D).
It was surprised that Cys77 or Cys109 residues, or Arg25, Val92 and Arg94 mutants of TGF-β1 and TGF-β3, on the contrary, promoted HCV propagation ( Figure 4B-D). Because of the significant sequence homology of TGFβ isoforms ( Figure 1A) and their possibly direct interaction with HCV ( Figure 2C), it might be that the conformation of the mutants would contribute to the direct interaction to facilitate HCV entry into cells. However, the mechanism needs to be further investigated.
The α-helix 3 of TGFβ isoforms is response for their binding surface for TβRI and truncating α-helix 3 of TGF-β2 suppresses its signalling activity. 28 However, loss of α-helix 3 of TGF-β1 and TGF-β3 with deletion of residues 52-71 aa did not interfere with their anti-HCV activities ( Figure 4D). These results indicated the truncated α-helix 3 might partly block TGFβ/SMAD signalling activities of TGF-β1 but was insufficient to influence the anti-HCV activities of TGF-β1 and TGF-β3, or there was other antiviral mechanism independent of TGFβ/SMAD signalling.
Interestingly, by blocking the TGFβ receptor kinase activity or silencing SMAD2/3, we presumed the anti-HCV activities of TGF-β1 and TGF-β2 might be through TGFβ/SMAD signalling pathway, while TGF-β3 inhibited HCV replication independent on TGFβ/ SMAD signalling pathway ( Figure 3). However, blocking TGF-β1/3 dimerization or the ligand-receptor II interaction led to abolishment of TGF-β1/3-mediated HCV propagation (Figure 4). This discrepancy might highlight that although TGF-β3 inhibited HCV replication independent on TGFβ/SMAD signalling pathway. The amino acid sequences, which were responsible for TGF-β3 homodimer formation or the ligand-receptor II interaction, were also important to the anti-HCV effects of TGF-β3.

F I G U R E 4
Anti-HCV effects are eliminated by residue mutations in TGFβ isoforms, but not influenced by deletion of α-helix 3. Huh7.5 cells transfected with plasmids with His tag at C-terminal (A), C77S mutant (B), C109Y mutant (C), TGFβ α-helix 3 deletion, or R25K, V92I and R94K mutants (D) for 48 h were subcultured and then infected with HCV after 24 h. Intracellular HCV core was detected by WB after 48 h. Data represent the mean ± SD of three independent experiments. ANOVA analysis followed by the Student's t test method was used. *P < 0.05, **P < 0.01 and ***P < 0.001 vs plasmid control 3.5 | TGF-β1 and TGF-β2, except TGF-β3, mediate HCV promoting hepatic stellate cell activation Activation of TGF-β1 initiates liver fibrotic tissue response, a programme of temporary collagen accumulation, and then induces pathogenic fibrosis. 29,30 We thus evaluated whether the TGFβ isoforms promoted hepatic stellate cell activation at their antiviral effective concentrations. Indeed, 0.5 nmol/L of TGF-β1 and TGF-β2 which concentration was active against HCV propagation, significantly increased COL1A1 and α-SMA mRNA expression ( Figure 5), while 0.5 nmol/L of TGF-β3 did not enhance COL1A1 and α-SMA mRNA expression in LX-2 cells ( Figure 5).

| D ISCUSS I ON
In this study, we found all of the three TGFβ isoforms suppressed HCV propagation through interrupting several different stages in the whole life cycle of HCV, including virus entry and intracellular replication stages, in TGFβ/SMAD signalling pathway-dependent and TGFβ/SMAD signalling pathway-independent manners, and therefore, they had additional anti-HCV activities.
Virus infection, particularly at the early stage, triggers the early host antiviral defence and adaptive immunity. 31,32 Hence, it commonly induces chronic inflammation, which is a leading cause of chronic disease, such as HCV infection induced fibrosis, cirrhosis and even liver cancer. 33,34 In the disease progression, TGFβ isoforms play crucial roles in the pro-inflammatory and antiinflammatory response process. 4,35,36 In healthy individuals, the levels of TGFβ isoforms in serum are different. TGF-β1 exhibits the highest level with about dozens of ng/mL, while TGF-β2 and TGF-β3 are relative low, 5,37,38 with the median concentration of TGF-β3 in plasma about 189 pg/mL (0.00756 nmol/L). 39 However, once HCV infection, TGFβ isoforms were up-regulated in both serum and liver, with the concentration of TGF-β1 at 87.5 ng/mL (3.5 nmol/L) and TGF-β2 at 0.3 ng/mL (0.012 nmol/L) in serum. 5,6,37,40 In HCVcc system, we also demonstrated the elevated expression of the three TGFβ isoforms after HCV infection. However, the elevated TGFβ isoforms, in turn, inhibited HCV propagation. Previous studies have reported that TGFβ can inhibit HCV RNA replication in HCV subgenomic replicon system and also directly restrict HBV replication in the cell culture system. 10,11 We also demonstrated that all of the three TGFβ isoforms could suppress HCV propagation in HCVcc system. Most importantly, the three TGFβ isoforms could additionally inhibit HCV propagation under the pathophysiological condition. TGFβ/SMAD signalling pathway activation could contribute to hepatic stellate cells activation, collagen gene transcription and liver fibrosis and was associated with later tumour progression. 41,42 Previous studies confirmed that chronic inflammation associated with HCV infection perturbed hepatic TGFβ signalling and then promoted fibrogenesis, and TGF-β2 released from HCV-infected cells contributed to hepatic fibrogenic responses through TGFβ signalling. 5,43 Our results also demonstrated TGF-β1 and TGF-β2 for 2 hours before HCV infection also inhibited HCV propagation after virus entry stage. Furthermore, we found all of the three TGFβ isoforms could reduce de novo produced HCV without inhibiting HCV release step. The detailed mechanisms showed that the anti-HCV effects of TGF-β1 and TGF-β2, except TGF-β3, were partly dependent on activating TGFβ/SMAD signalling pathway by applying the inhibitors and siRNA focusing on TGFβ/SMAD signalling pathway. The results from amino acids mutant experiments also suggested that TGFβ signalling activated by TGF-β1 was critical for its anti-HCV effect. Although TGF-β1 and TGF-β2 suppressed HCV propagation dependent on TGFβ/SMAD signalling, they inhibited HCV in different HCV infection stages, which might be owing to their different TβRII affinity and subsequently anti-HCV gene expression. 44,45 In addition, TGFβ isoforms might inhibit HCV propagation in a TGFβ/SMAD signalling pathwayindependent manner because they were still active against HCV propagation when the α-helix 3 of TGFβ isoforms was deleted.
Other antiviral mechanisms of TGFβ isoforms remain to be clarified. Furthermore, owing to part of anti-HCV mechanisms of TGFβ isoforms was the same, the additional anti-HCV activities might depend on both of promoting the same mechanism and acting on different mechanisms.
F I G U R E 5 TGF-β1 and TGF-β2, except TGF-β3, promote hepatic stellate cell activation. LX-2 cells were stimulated with 0.5 nmol/L TGFβ isoforms for 24 h and COL1A1 and α-SMA mRNA was quantified with qRT-PCR. Data represent the mean ± SD of three independent experiments. ANOVA analysis followed by the Student's t test method was used. **P < 0.01 vs solvent control In conclusion, HCV infection increases the level of TGFβ isoforms in infected hepatocytes, and thus, the elevated TGF-β1 and TGF-β2 exacerbate the liver disease, such as fibrosis. In turn, the elevated TGFβ isoforms might be one of the host antiviral defenders through interrupting HCV life cycle in TGFβ/SMAD signalling pathwaydependent and TGFβ/SMAD signalling pathway-independent manners.

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support the findings of this study are available from the corresponding author upon request.