Background/Aims To investigate the role of Wnt signalling in adipogenesis using an in vitro model of Graves’ orbitopathy (GO).
Methods Orbital fat was obtained from patients with GO and non-GO participants for primary orbital fibroblast (OF) culture. Expression levels of Wnt5a, Wnt10b, β-catenin, phospho-β-catenin and cyclin D1 were compared between GO and non-GO OFs. These expression levels were also determined during adipogenesis of GO and non-GO OFs. The effects of a stimulator and inhibitor of Wnt signalling on adipogenesis of GO and non-GO OFs were investigated.
Results Western blotting analysis showed significant reductions in β-catenin and cyclin D1 and significant enhancement of phospho-β-catenin in OFs from patients with GO, compared with OFs from non-GO participants (p<0.05). Expression of Wnt5a, Wnt10b, β-catenin and cyclin D1 in OFs was highest on day 0, and then gradually declined after induction of adipogenic differentiation. The expression levels of PPARγ, C/EBPα and C/EBPβ were reduced in Wnt stimulator-treated OFs in a dose-dependent manner. Oil red O staining confirmed that a stimulator of Wnt inhibited adipogenesis in GO OFs.
Conclusion These results indicate that Wnt signalling inhibits adipogenesis in OFs from patients with GO and non-GO participants. Further studies are required to examine the potential of Wnt signalling as a target for therapeutic strategies.
- experimental and laboratory
Data availability statement
Data are available upon reasonable request.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
The key pathogenic processes of Graves’ orbitopathy (GO) are inflammation, adipogenesis, oedema caused by excess production of hydrophilic glycosaminoglycan (GAG) and fibrosis.1–4 Adipogenesis and oedema can cause symptoms of GO by increasing the volumes of intraorbital components such as extraocular muscles and intraorbital adipose tissues within the bony orbit.5 6 Because orbital fibroblasts (OFs) can differentiate into adipocytes and secrete GAG, OFs are presumed to be central players in the development of GO.3 7 8
Wnt signalling pathways are classified into the β-catenin-dependent (canonical) pathway and the independent (non-canonical) pathway.9–11 The canonical Wnt signalling pathway is extensively studied in development, tumorigenesis and regenerative research.10 The non-canonical Wnt signalling pathway is less well understood.12 Wnt proteins belong to a family of secreted glycoproteins that are involved in a variety of developmental and physiological processes. In particular, these are important inhibitory factors in adipogenesis.13–15
In the canonical Wnt signalling pathway, Wnt represses adipogenesis by blocking the induction of peroxisome proliferator-activated receptor (PPAR) γ and CCAAT-enhancer-binding protein (C/EBP)α.15 16 Briefly, Wnt binds to Frizzled (FZD) receptors17 and low-density lipoprotein receptor-related protein (LRP) coreceptors.18 Wnt binding to FZD and LRP5/6 initiates canonical signalling, leading to the release of β-catenin. Activation of β-catenin induces cyclin D1 gene expression. Cyclin D1 then inhibits PPARγ-mediated adipogenesis through histone deacetylase recruitment.13 14 19–21 A likely candidate for a potential antiadipogenic Wnt signal is Wnt10b. Overexpression of Wnt10b in 3T3-L1 preadipocytes stabilises β-catenin and blocks adipogenesis.19 Addition of Wnt10b antibody to 3T3-L1 medium promotes adipocyte differentiation.22 Because adipogenesis is a main pathogenic pathway in GO, we presumed that Wnt10b might participate in the regulation of OF differentiation into adipocytes. To the best of our knowledge, there are no reports concerning the association between Wnt10b and GO in the current literature.
Indeed, there have been few reports regarding Wnt signalling dysregulation in the pathogenesis of GO. Most recently, Ezra et al 23 demonstrated downregulation of Wnt5a, Dickkopf3 (Dkk3), sFRP4 and frizzled 7 (FZD7) in patients with active thyroid-associated orbitopathy, indicating that Wnt signalling may be involved in the pathogenesis of thyroid-associated orbitopathy. Wnt5a is classified as a non-canonical Wnt family member. Previous studies concerning the role of non-canonical Wnt proteins in adipogenesis have reached conflicting conclusions.12 24–26 Some have indicated that Wnt5a promotes adipogenesis,24 25 while others have suggested that Wnt5a inhibits adipogenesis.12 26
Because adipogenesis is one of the main pathogenic pathways in GO, we sought to elucidate the link between the Wnt signalling pathway and adipogenesis in OFs obtained from patients with GO. In this study, we investigated the role of the Wnt signalling pathway, especially Wnt10b and Wnt5a, in adipogenesis of GO OFs. This research will help to gain further insights into GO and its associated changes at both the cellular and molecular levels.
Materials and methods
Participants and cell culture protocol
Orbital fat tissues were obtained from 15 patients with GO and from 15 age-matched and sex-matched control participants with no history of GO. All patients with GO underwent orbital decompression for proptosis correction in the inactive GO phase, while control participants underwent cosmetic upper and lower blepharoplasty. We obtained orbital fat during ageing upper and lower blepharoplasty in order to primary-culture OFs from normal control, in accordance with previous reports.27–31 All patients with GO had euthyroid status at the time of surgery and had not been treated with steroids or radiation therapy for at least 6 months, to minimise the effect of prior treatments such as radiotherapy and intravenous methylprednisolone.
OF cultures were conducted in accordance with the previously described methods.27 28 Dulbecco’s modified Eagle’s medium (DMEM; GenDEPOT, Katy, Texas, USA), 20% fetal bovine serum (FBS; GenDEPOT) and penicillin-streptomycin (GenDEPOT) were used for primary cell cultures. The OFs were grown from explants, and the monolayers were passaged serially by gentle treatment with trypsin/EDTA. The cells were incubated in DMEM with 10% FBS and antibiotics and were grown in a humidified 5% CO2 incubator at 37°C. The cells were stored in liquid N2 until needed and used between the second and seventh passages.
Confluent GO or normal OFs were subjected to the differentiation protocol for 4‒10 days, grown to confluence in six-well plates, and then exposed to differentiation medium for 10 days, in accordance with a previously described protocol.31–33 The culture medium was changed to DMEM-high glucose supplemented with 10% FBS, 33 µM biotin (Sigma-Aldrich, St. Louis, Missouri, USA), 17 µM pantothenic acid (Sigma-Aldrich), 10 µg/mL transferrin (Sigma-Aldrich), 0.2 nM T3 (Sigma-Aldrich) and 1 µM insulin (Roche Diagnostics, Mannheim, Germany).32 33 For the first 4 days, 10 µM dexamethasone (Sigma-Aldrich), 0.1 mM isobutylmethylxanthine (Sigma-Aldrich) and differentiation medium were used. After the first 4 days, dexamethasone and isobutylmethylxanthine were excluded from the medium. The differentiation protocol was continued for 10 days, during which the medium was replaced every 2‒3 days.
To evaluate the effects of a Wnt/β-catenin signal stimulator and inhibitor on adipogenesis, we exposed the culture to each stimulator and inhibitor (0, 0.5, 1.0 and 5.0 µM) for the entire 10-day differentiation period. Dickkopf-related protein 1 (DKK1) (Sigma-Aldrich)34 and CHIR99021 (Abcam, Cambridge, UK)35 were used as the Wnt/β-catenin signal inhibitor and stimulator, respectively. To evaluate the effects of a Wnt5a stimulator and inhibitor on adipogenesis, Wnt5a agonist, Foxy-5 (Tocris Bioscience, Bristol, UK) and Wnt antagonist III, Box5 (Calbiochem, San Diego, California, USA) were used as the Wnt5a signal stimulator and inhibitor, respectively.
Quantitative PCR analysis
To examine the effects of stimulation and inhibition of Wnt signalling on adipogenesis in GO orbital cells, we measured the mRNA levels of molecular markers for adipogenesis such as PPARγ, C/EBPα and C/EBPβ on day 10 of adipogenesis after treatments with different doses.
For quantitative PCR (qPCR) analysis, the total RNA (1 µg) of adipogenesis on days 0 and 10 were isolated and reverse-transcribed into complementary DNA using a reverse transcription kit (Applied Biosystems, Foster City, California, USA). The resulting cDNA was amplified using a thermocycler (ABI Step One Plus Real Time PCR; Applied Biosystems) with a universal PCR master mix (Applied Biosystems) using the recommended PCR conditions for quantitative assessment of gene transcript levels in the cell samples. All PCR assays were performed in triplicate. β-actin expression was used for normalisation; the results are expressed as relative fold changes in the threshold cycle values relative to the control group using the 2-ΔΔCt method.
We compared the expression levels of Wnt5a (#ab72583; Abcam), Wnt10b (#ab7086; Abcam), β-catenin (#ab32572; Abcam), phospho-β-catenin (#ab81305, Abcam) and cyclin D1 (#orb10496, Biorbyt, Cambridge, UK) between GO and non-GO OFs using western blotting analyses. We also determined changes in expression levels of Wnt signalling products in GO OFs during adipogenesis using western blotting analyses. To examine the effects of stimulation and inhibition of Wnt signalling on adipogenesis in GO orbital cells, we measured the protein levels of molecular markers for adipogenesis such as PPARγ, C/EBPα and C/EBPβ on day 10 of adipogenesis using western blotting analyses.
Cells were washed with ice-cold phosphate-buffered saline and whole cell lysates were obtained by incubation on ice for 30 min in RIPA II cell lysis buffer (GenDEPOT) containing a protease inhibitor cocktail (Roche Diagnostics). The lysates were centrifuged at 12 000×g for 10 min and the cell homogenate fractions were stored at −70°C until use. Protein concentrations were determined using the Bradford assay. Equal amounts of protein (50 µg) were boiled in sample buffer and resolved by 10% (w/v) SDS-PAGE. The proteins were transferred onto polyvinylidene difluoride membranes (Immobilon; Millipore, Billerica, Massachusetts, USA). Then, the samples were probed overnight with primary antibodies (ie, antibodies against β-catenin, phospho-β-catenin, Wnt5a, Wnt10b, PPARγ, C/EBPα, C/EBPβ and cyclin D1) in Tris-buffered saline containing Tween 20 (TBST), and washed three times with TBST. Immunoreactive bands were detected with horseradish peroxidase-conjugated secondary antibody and developed using an enhanced chemiluminescence kit (Abcam). The immunoreactive bands were quantified by densitometry using a gel documentation system (c280; Azure Biosystems, Dublin, California, USA) and normalised relative to β-actin levels in the same sample.
The effects of the adipogenesis process on secreted Wnt5a and Wnt10b expression were analysed using a human cytokine ELISA kit (AVITA Systems) according to the manufacturer’s protocol. Differentiation medium taken on adipogenesis days 0, 4 and 10 was examined to determine the soluble levels of Wnt5a and Wnt10b by ELISA.
Oil red O staining
OF cultures were plated in six-well culture dishes with DMEM containing 10% FBS and penicillin-streptomycin, then grown to confluence and subjected to the differentiation protocol. Then, the cells were stained with oil red O and processed using an oil red O staining kit (ScienCell, Carlsbad, California, USA), in accordance with the manufacturer’s guidelines. The dishes were rinsed with distilled water before visualisation and photography at 40× and 100× using a Nikon Eclipse Ci-L upright microscope (Nikon, Tokyo, Japan).
All experiments were performed at least three times independently, using at least three cell cultures harvested from different individuals. The results are presented as mean±SDs calculated from normalised measurements. Differences between groups were assessed using independent t-tests and one-way analysis of variance with post-hoc analysis using SPSS V.26.0 (IBM Corporation). In all analyses, p<0.05 was considered to indicate statistical significance.
Orbital fat tissues were obtained from 15 patients with GO and 15 age-matched and sex-matched control participants with no history of GO. Demographic data are shown in table 1. The mean age of all patients with GO was 53.8±13.8 years, and seven patients (77.8%) were women. The mean age of the control group was 59.8±16.3 years, and six participants (66.7%) were women. There were three smokers and three ex-smokers in both groups.
Comparison of Wnt signalling protein expression between GO and non-GO OFs
To determine whether the Wnt pathway is involved in the pathogenesis of GO, we compared the expression levels of Wnt5a, Wnt10b, β-catenin, phospho-β-catenin and cyclin D1 between GO (n=10) and non-GO (n=10) OFs using western blotting analyses. This showed significant reductions in β-catenin, cyclin D1, Wnt5a and Wnt10b, as well as a significant enhancement in phospho-β-catenin, in OFs from GO tissues, compared with those from non-GO tissues (p<0.05, figure 1).
Expression of Wnt signalling protein during adipogenesis in GO and non-GO OFs
We investigated changes in the expression levels of Wnt signalling proteins in non-GO and GO OFs during adipogenesis using western blotting analyses. Wnt5a, Wnt10b, β-catenin and cyclin D1 were at their highest levels in GO OFs on day 0. They then declined gradually after induction of adipogenic differentiation, such that there was a significant difference (p<0.05, figure 2A,B). In contrast, phospho-β-catenin protein expression gradually increased during adipogenesis (p<0.05, figure 2A,B).
In non-GO OFs, the expression levels of Wnt5a, Wnt10b, β-catenin and cyclin D1 gradually decreased during adipogenic differentiation (p<0.05, figure 2A,C). However, the expression levels of phospho-β-catenin protein were significantly elevated only on day 4, not on day 10 (figure 2A,C).
We investigated changes in the expression levels of secreted Wnt5a and Wnt10b in non-GO and GO OFs during adipogenesis using ELISA. Adipogenesis reduced secreted Wnt5a and Wnt10b ligands in supernatants of GO and non-GO samples (figure 2D,E). Wnt5a and Wnt10b levels were at their highest level on adipogenesis day 0 and then declined gradually after induction of adipogenic differentiation (p<0.05, figure 2D,E).
Effects of a stimulator and inhibitor of Wnt/β-catenin signalling on adipogenesis of OFs
To examine the effects of stimulation and inhibition of Wnt/β-catenin signalling on adipogenesis in non-GO and GO OFs, we measured the mRNA and protein expression levels of molecular markers for adipogenesis such as PPARγ, C/EBPα and C/EBPβ on day 10 of adipogenesis, following treatment with different doses of a stimulator or inhibitor of Wnt/β-catenin signalling.
The mRNA expression levels of PPARγ, C/EBPα and C/EBPβ gradually decreased in a dose-dependent manner following treatment with CHIR99021, a Wnt/β-catenin stimulator, in non-GO and GO OFs (p<0.05, figure 3A,C). The mRNA expression levels of PPARγ, C/EBPα and C/EBPβ tended to increase in a dose-dependent manner following treatment with DKK1, a Wnt/β-catenin inhibitor, in non-GO and GO OFs. However, these differences were not statistically significant (figure 3B,D).
Western blotting analyses showed that the protein expression levels of PPARγ gradually decreased in a dose-dependent manner following treatment with CHIR99021 in non-GO and GO OFs (p<0.05, figure 4A,C). However, the protein expression levels of C/EBPα were not affected by CHIR99021 pretreatment in GO OFs (figure 4A,C). The protein expression levels of C/EBPβ were significantly reduced following treatment with 5.0 µM CHIR990921 in non-GO and GO OFs (figure 4A,C).
Furthermore, western blotting analyses showed that the protein expression levels of PPARγ, C/EBPα and C/EBPβ tended to increase by DKK1 pretreatment in non-GO and GO OFs. However, the differences were not statistically significant. These findings were consistent with the results of qPCR analysis (figure 4B,D). Therefore, these results indicate that treatment with a stimulator of Wnt/β-catenin signalling inhibited adipogenesis in OFs from both patients with GO and non-GO participants. However, the results of pretreatment with an inhibitor of Wnt/β-catenin signalling were inconsistent.
We performed oil red O staining to confirm these findings. Oil red O staining showed greater accumulation of lipid droplets in DKK1-treated non-GO and GO cells, whereas it showed reduced accumulation of lipid droplets in CHIR99021-treated non-GO and GO cells (figure 5A). Oil red O staining demonstrated that CHIR990921 treatment reduced adipogenesis, but it was not apparent whether DKK1 enhanced adipogenesis (figure 5B).
Effects of a stimulator and inhibitor of Wnt5a signalling on adipogenesis of OFs
To examine the effects of stimulation and inhibition of Wnt5a signalling on adipogenesis in non-GO and GO OFs, we measured the protein levels of molecular markers for adipogenesis (ie, PPARγ, C/EBPα and C/EBPβ) on day 10 of adipogenesis, following treatment with various doses of a stimulator or inhibitor of Wnt5a signalling.
Western blotting analyses showed that the expression levels of PPARγ, C/EBPα and C/EBPβ gradually decreased in a dose-dependent manner following treatment with Foxy-5, a Wnt5a agonist (figure 6A,C). In contrast, these expression levels increased in a dose-dependent manner following treatment with BOX5, a Wnt5a antagonist (figure 6B,D). Overall, these results indicate that Wnt5a signalling inhibited adipogenesis in OFs from patients with GO.
These findings were confirmed by oil red O staining, which showed a greater accumulation of lipid droplets in BOX5-treated GO cells than in Foxy-5-treated GO cells (figure 7).
Wnt signalling pathways play important roles in numerous physiological and pathological processes. In 2000, MacDougald et al 36 first reported that Wnt signalling was implicated in the regulation of adipocyte differentiation. They demonstrated that canonical pathway activation in 3T3-L1 preadipocytes through overexpression of Wnt1 led to adipogenesis inhibition, while inhibition of Wnt signalling in preadipocytes induced adipogenesis. They also showed that Wnt10b mRNA expression was reduced on 3T3-L1 adipocyte differentiation, implying that reduced Wnt10b expression is required for adipogenesis to occur. The factors that regulate adipogenesis either promote or block the cascade of transcription factors that coordinate the adipogenic differentiation process. Wnt is presumably involved in some of the negative aspects of adipogenesis.14 As mentioned in the Introduction section, adipogenesis is blocked by secretion of Wnt family members, particularly Wnt5a12 26 37 and Wnt10b.19 36 38 Hence, we hypothesised that Wnt signalling may also inhibit adipogenesis in OFs, and investigated whether Wnt5a and Wnt10b regulate adipogenesis in patients with GO.
First, we examined the expression levels of Wnt signalling components in OFs from patients with GO and non-GO participants to determine whether the Wnt pathway is involved in GO pathogenesis. We found significant reductions in Wnt5a, Wnt10b, β-catenin and cyclin D1 in OFs from patients with GO, compared with OFs from non-GO participants. Levels of phospho-β-catenin were significantly greater in OFs from patients with GO than in those from non-GO participants. Because phosphor-β-catenin is an inactive form of β-catenin, the Wnt signalling pathway is attenuated in patients with GO, compared with non-GO participants. Investigation of the Wnt/β-catenin signalling pathway indicated that Wnt10b binds to an LRP and a membrane receptor complex composed of an FZD G-protein coupled receptor to initiate a signalling cascade. Activation of canonical Wnt/β-catenin signalling results in the stabilisation and activation of β-catenin, which translocates from the cytoplasm into the nucleus and interacts with numerous partners, including the T-cell factor/lymphoid-enhancing factor family. Formation of the β-catenin/T-cell factor transcription factor activates the transcription of Wnt-responsive genes such as cyclin D1.39 Cyclin D1 inhibits PPARγ-mediated adipogenesis through histone deacetylase recruitment.20 21
During adipogenesis, we found that Wnt10b expression was highest in preadipocytes, then decreased after induction of OF differentiation. Furthermore, β-catenin and cyclin D1 were highest in preadipocytes, then decreased after induction of OF differentiation. These results are consistent with previously published findings.36 Levels of Wnt signalling proteins presumably decrease during adipogenesis. Alternatively, reduced levels of Wnt signalling proteins are required for adipogenesis to occur in non-GO and GO OFs. To further explore the association between adipogenesis and Wnt signalling, we examined the effects of DKK1, a Wnt/β-catenin inhibitor, and CHIR99021, a Wnt/β-catenin stimulator,19 on adipogenesis in GO OFs. When Wnt signalling was stimulated by treatment with CHIR99021, the expression levels of PPARγ, C/EBPα and C/EBPβ were reduced. Although inactivation of Wnt signalling with DKK1 appeared to stimulate adipogenesis, these differences were not statistically significant.
Wnt5a is classified as a non-canonical Wnt family member. These proteins act through a different set of coreceptors (eg, receptor tyrosine kinase-like orphan receptor and receptor-like tyrosine kinase) and a different set of downstream effectors (eg, Rho GTPase and c-Jun N-terminal kinases).40 41 Previous studies concerning the roles of non-canonical Wnt proteins in adipogenesis have reached conflicting conclusions.12 24–26 Some reports indicated that Wnt5a promotes adipogenesis,24 25 while others indicated that Wnt5a inhibits adipogenesis.12 26 In the present study, we found that Wnt5a expression was highest in preadipocytes and decreased after induction of OF differentiation with Wnt10b. Moreover, we found that inactivation of Wnt5a signalling with box 5 (a Wnt5a antagonist) stimulated adipogenesis and induction of PPARγ, C/EBPα and C/EBPβ. In contrast, when Wnt signalling was stimulated by Foxy-5 (a Wnt5a agonist), the expression levels of PPARγ, C/EBPα and C/EBPβ were gradually reduced. These results show that Wnt5a may inhibit adipogenesis in OFs.
Indeed, few Wnt signalling inhibitors have been approved for clinical use due to the complexity of the cascades triggered by Wnt.41 42 Nonetheless, romosozumab, an inhibitor of LRP5 and LRP6, was recently approved by the Food and Drug Administration for treatment of patients with osteoporosis, who have a high risk of fracture.42–44 The development of such clinically available medicines is the result of extensive studies that demonstrated associations between Wnt signalling and osteogenesis.22 45–47 Several studies have shown associations between Wnt signalling and adipogenesis in people with obesity.48–50 Although adipogenesis is a key pathogenic process involved in GO, only a few studies have characterised the relationship between Wnt signalling and adipogenesis in patients with GO. For example, Kumar et al 8 reported that epigenetic silencing of Wnt signalling led to enhanced differentiation of GO orbital preadipocytes. Ezra et al 23 investigated the gene expression profiles of orbital fat from patients with active GO, and reported dysregulation of Wnt signalling gene expression involving Wnt5a, sFRP and DKK. Overall, the results of the present study support the previous findings that Wnt/β-catenin signalling inhibits adipogenesis in an in vitro model of GO. By understanding the pathogenesis of GO, new potential treatment targets have been identified.51–56 Because our results also indicate that adipogenesis was downregulated by Wnt signalling in OFs from GO tissues, this treatment may represent a new therapeutic target for GO.
The adipogenesis process comprises two main phases: the determination phase and the terminal differentiation phase.57 In the determination phase, multipotent mesenchymal stem cells (MSCs) differentiate into preadipocytes. The preadipocytes then undergo mitosis, in the form of mitotic clonal expansion (MCE), and finally differentiate into adipocytes in the terminal differentiation phase.57–60 Recently, it has been reported that GO OFs also undergo an MCE process in the early stage of adipogenesis.61 In general, it has been suggested that the inhibition of WNT signalling is required to induce MSC adipogenesis.15 The induction of Wnt inhibitors has been shown to be important to ensure a sufficient differentiation process.62 We believe that in future studies, the experiments should focus on the role of WNT signalling in MCE on GO OFs.
In this study, we investigated the role of the Wnt signalling pathway, especially Wnt10b and Wnt5a, in adipogenesis in GO OFs. We presume that this research will provide further insights into GO and its associated changes at the cellular and molecular levels. Nonetheless, limitations of this study should be addressed. First, the Wnt protein family includes 19 members. Although we chose Wnt10b and Wnt5a as candidates for regulation of adipogenesis in patients with GO, the expression patterns of other Wnt proteins may contribute to differences between GO and non-GO tissue. Second, the key pathogenic processes of GO include adipogenesis, as well as inflammation, oedema caused by excess production of hydrophilic GAG, and fibrosis. It is important to determine whether Wnt/β-catenin signalling also affects other pathological processes, such as hyaluronic synthesis and fibrosis, which contribute to exophthalmos. Indeed, hyaluronic acid enhances the proliferation of human amniotic MSCs through the activation of the Wnt/β-catenin signalling pathway.63 Further extensive studies are needed.
In summary, this study provides useful information concerning the Wnt signalling pathway in GO OFs, which can be used as a target for the treatment of GO. Future studies can identify additional molecular and chemical agents that activate the Wnt signalling pathway and may aid in the treatment of patients with GO.
Data availability statement
Data are available upon reasonable request.
Patient consent for publication
Informed written consent was obtained from all participants, and this study was approved by the Institutional Review Board of Soonchunhyang Hospital, Soonchunhyang University College of Medicine.
Contributors SJJ, TKP and SYJ were responsible for the conception and design of the study, as well as the intellectual content. YJC performed experiments. SEW, BYK, J-SY and SYJ revised the article critically for intellectual content. All authors read and approved the final manuscript.
Funding This work was supported by a National Research Foundation of Korea (NRF) grant funded by the government of Korea (MSIT) (No. 2020R1A2C4002095), and was partially supported by the Soonchunhyang University Research Fund.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.