Cornea The Role of -Like Binding 2 (IGFBP2) in the Regulation of Corneal Fibroblast Differentiation

Soo Hyun Park, Kyoung Woo Kim, and Jae Chan Kim

Department of Ophthalmology, College of Medicine, Chung-Ang University Hospital, Seoul, Korea

Correspondence: Jae Chan Kim, De- PURPOSE. Previously, we reported that keratocyte-conditioned medium (KCM) facilitates the partment of Ophthalmology, Chung- differentiation of human mesenchymal stem cells (hMSCs) into corneal keratocyte–like cells. Ang University Hospital, 224-1, This study is designed to investigate the roles of insulin-like growth factor binding protein 2 Heukseok-dong, Dongjak-gu, Seoul (IGFBP2) for the regulation of corneal fibroblast differentiation as a newly unveiled 156-755, Korea; component of KCM. [email protected]. Submitted: February 4, 2015 METHODS. Immunodot blot analysis was performed to identify the factors that are highly Accepted: October 5, 2015 secreted, especially in KCM. Then, we investigated whether IGFBP2 differentiates hMSCs into keratocyte-like cells and whether maintains the phenotypes of keratocyte in human corneal Citation: Park SH, Kim KW, Kim JC. fibroblasts (HCFs) by analyzing expression patterns of alpha-smooth muscle actin (a-SMA) and The role of insulin-like growth factor binding protein 2 (IGFBP2) in the keratocyte markers including keratocan, lumican and aldehyde dehydrogenase 1 family regulation of corneal fibroblast differ- member A1 (ALDH1A1). Furthermore, to specify the role of IGFBP2, the expression of a-SMA entiation. Invest Ophthalmol Vis Sci. and keratocyte markers was determined in transforming growth factor b 1 (TGFb1)-induced 2015;56:7293–7302. DOI:10.1167/ corneal myofibroblast and in HCFs after knockdown of IGFBP2. iovs.15-16616 RESULTS. The most prominent factor in both KCM and amniotic membrane extract was IGFBP2. Insulin-like growth factor binding protein 2 increased the expression of IGFBP2, keratocan, and ALDH1A1, and decreased a-SMA expression in hMSCs and HCFs. Insulin-like growth factor binding protein 2 inhibited TGFb1-induced upregulation of a-SMA and increased expressions of keratocan and ALDH1A1 in HCFs. Furthermore, the knockdown of IGFBP2 increased a-SMA expression and decreased ALDH1A1 level in HCFs.

CONCLUSIONS. Insulin-like growth factor binding protein 2 is strongly associated with restoration of keratocyte phenotype in HCFs. Our results show an important novel role of IGFBP2 in regulation of corneal fibroblast differentiation and suggest that IGFBP2 can be a therapeutic candidate for corneal antifibrotic strategy. Keywords: insulin-like growth factor binding protein 2, alpha-smooth muscle actin, keratocytes, corneal fibroblasts

nsulin-like growth factor (IGF)-1 and -2 are polypeptides that factor binding 1, 2, 3, and 5 have been reported to I exhibit mitogenic, metabolic, and differentiative effects on a bind to the cell surface or ECM.3,7,8 The binding affinity of IGFs variety of cell types. Because of the relative abundance of both to IGFBPs decreases when IGFBPs are bound to the cell surface IGF-1 and IGF-2 during development, the IGFs are thought to or ECM.5,9 play an especially important role in the proliferation and Recently, the complex actions of the IGFBPs in skeletal differentiation of embryonic tissues.1,2 The insulin-like growth muscle have become more apparent, with IGFBP2 implicated factor binding proteins (IGFBP1–6) are a family of circulating in skeletal muscle cell proliferation and differentiation.10,11 proteins that were initially defined by their capacity to Additionally, IGFBPs are thought to have an inhibitory effect on differentially modulate (positively or negatively) the actions of both IGF-1 and -2.12–14 Insulin-like growth factor binding IGF ligands. Insulin-like growth factor binding proteins are protein 2 binds to IGF-1 or -2 with high affinity and can present in serum and in a variety of biological fluids, including manipulate their binding to the IGF-1 receptor. This activity is amniotic, follicular, cerebrospinal, and seminal fluid, as well as modulated by the interaction of the binding protein with milk.3 Insulin-like growth factor binding proteins have also .15 Furthermore, IGFBP2 is expressed in fetal tissues been identified in the extracellular environment and inside that are highly proliferative, and its expression significantly cells, and play distinct physiological roles in growth and decreases after birth.16 development. Insulin-like growth factor binding proteins may Several components of an IGF autocrine–paracrine sys- be differentially targeted to different tissues depending on both tem,17–19 including several different IGFBPs,20–25 have been their primary structure and their posttranslational modifica- identified in ocular tissues. Some studies have reported that tions.4 It has been postulated that a number of IGFBPs can IGFBPs in the vitreous humor exhibit an expression pattern interact with the extracellular matrix (ECM) or cell surface via different to those in serum. This suggests the possibility of local glycoproteins, collagens, and .5,6 Insulin-like growth synthesis of IGFBPs in the eye rather than uptake from the

Copyright 2015 The Association for Research in Vision and Ophthalmology, Inc. iovs.arvojournals.org j ISSN: 1552-5783 7293

Downloaded from iovs.arvojournals.org on 09/28/2021 Regulatory Role of Fibroblast Lineage of IGFBP2 IOVS j November 2015 j Vol. 56 j No. 12 j 7294

systemic circulation.26,27 The unique expression of IGFBP2 in mL on 2.5 cm 3 2.5 cm denuded AM Cells were cultured on the the eye suggests that it could be involved in the regulation of AM for 15 days in a medium containing DMEM/F12 supple- ocular growth and differentiation as well as in homeostasis in mented with 10% fetal bovine serum (FBS), and the medium the mature eye.28 However, there is little research focused on was replaced every 2 to 3 days. intraocular IGFBP, and its role remains unclear. The amniotic membrane (AM) is the innermost layer of the Collection of KCM fetal membrane. Studies have demonstrated that human29,30 and murine31 keratocyte, as judged by their characteristic Conditioned medium was obtained by culture of human dendritic morphology as well as expression of corneal stroma- corneal keratocytes on an AM as described above. Confluent specific keratocan, can maintain their phenotype without stromal matrix cells growing on an AM were washed with PBS differentiation into alpha-smooth muscle actin (a-SMA)–ex- before adding 13 mL of DMEM/F12 supplemented with 10% pressing myofibroblasts. This can occur when the keratocytes FBS. The medium was then harvested after 2 days and are cultured on the AM stromal surface even when TGF-b is centrifuged. The supernatant was collected by filtration added in a serum-containing medium.31 Additionally, AM through a 0.22-lm filter and used as KCM, which was directly stromal extract not only helps maintain the fibroblast transferred onto MSC cultures. phenotype of AM stromal cells (AMSC; isolated mesenchymal cells from human AM stromal matrix) in vitro, but can also Primary Culture of HCFs reverse differentiated myofibroblasts back to fibroblasts.32 Moreover, we previously reported that keratocyte-conditioned For fibroblast isolation, corneal stromal tissue was cut into 6 to medium (KCM) has the capacity to facilitate the differentiation 8 pieces and placed in 6-well plates. After 10 minutes of of MSCs into corneal keratocyte–like cells.33 However, the adhesion, each explant was covered with DMEM/F12 supple- factors controlling the differentiation of the keratocyte and mented with 10% FBS and 100 units/mL penicillin/streptomy- fibroblast lineages remain unclear. cin (WelGENE), and then placed in a humidified incubator We hypothesized that factors present in KCM or AM extract (378C, 5%, CO2). The medium was changed every 4 to 5 days. might play an important role in differentiation and mainte- nance of keratocyte characteristics. In this study, IGFBP2 Culture of hMSCs showed distinct expression in both KCM and AM extract compared with other IGFBP family proteins, IGF-1 and -2; Bone marrow–derived MSCs (BM-MSCs, BM3.B10), which were therefore, we investigated the involvement of IGFBP2 in the obtained from human fetal spinal vertebrae at 12 to 15 weeks regulation of the differentiation of human corneal fibroblasts of gestation with an amphotropic, replication-incompetent (HCFs). retroviral vector containing v-myc as previously reported,34 were provided by Seung U. Kim (Professor Emeritus of Neurology, University of British Columbia, Vancouver, Canada). MATERIALS AND METHODS Bone marrow–derived MSCs were cultured in alpha-minimum essential medium (a-MEM; Invitrogen-Gibco) supplemented Study Design with L-glutamine, deoxyribonucleosides, ribonucleosides, 10% FBS, and 1% penicillin/streptomycin. Our study was designed as follows:

1. Composition analysis of conditioned medium (from In Vitro Drug Treatment keratocytes grown on AM; from corneal fibroblasts Cultured HCFs were treated with IGFBP2 (R&D Systems, Inc., grown on plastic dishes; from AM) and AM extract; Minneapolis, MN, USA) at a concentration of 100 to 500 ng/mL 2. Analysis of effects of IGFBP2 on keratocyte differentia- for various durations ranging from 24 to 72 hours. TGFb1 tion; investigation of specific markers in human (ProSpec-Tany Technogene Ltd., Rehovot, Israel) was used for mesenchymal stem cells (hMSCs) and HCFs induced by chemically-induced fibrosis of HCFs. treatment with IGFBP2 or cultured in KCM; 3. Investigation of IGFBP2-mediated inhibition of TGFb1- induced corneal myofibroblast transformation by mea- Quantitative RT-PCR (qRT-PCR) suring alteration of a-SMA expression; and RNA isolation was performed using RNAiso plus (Takara Bio, 4. Investigation of the change of corneal phenotype after Inc., Otsu, Japan) according to the manufacturer’s instructions. knockdown of IGFBP2 in HCFs. For semiquantitative RT-PCR, total RNA was reverse transcribed into complementary DNA (cDNA synthesis ; Takara Bio, Primary Culture of Human Keratocytes on AM Inc.). Equal amounts of samples were used for PCR amplifica- tion of cDNA with primers specific for human a-SMA or the Human donor corneal tissue was obtained and stored in IGFBP family. Real-time quantitative RT-PCR (qRT-PCR) was Optisol-GC (Bausch & Lomb, Rochester, NY, USA) for less than performed using SYBR Premix ExTaq (Takara Bio, Inc.). 3 days. Human keratocytes were isolated from the corneal SybrGreen fluorescence of the amplified cDNA products was stroma by sequential collagenase digestion as described quantified using the CFX96 Real-Time PCR Detection System previously.33 (BioRad, Hercules, CA, USA) and an appropriate standard curve Human AM preserved in Dulbecco’s modified Eagle’s from autonomous qPCR assay reactions. Relative quanti- medium (DMEM; WelGENE, Daegu, South Korea) and pure ties were obtained using the comparative cycle threshold (Ct) glycerol (1:1) at 808C was thawed and incubated in a solution method after normalization to a reference gene (glyceralde- of versene and trypsin-EDTA (1:1; Invitrogen-Gibco, Carlsbad, hyde-3-phosphate dehydrogenase [GAPDH]). The results of the CA, USA) for 30 minutes at 378C, and the amniotic epithelium qRT-PCR analysis are presented as the average amount of each was removed from the AM using a scraper. Epithelium-free AM gene expressed relative to average GAPDH expression. The was placed on a 3 3 3-cm piece of stainless mesh with the specific primers that were used for a-SMA, IGFBP1, IGFBP2, stromal matrix facing upward. The suspended keratocytes IGFBP3, IGFBP4, IGFBP5, IGFBP6, IGF-1, IGF-2, and GAPDH extracted from the corneal tissues were seeded at 1 3 105 cells/ are shown in the Table.

Downloaded from iovs.arvojournals.org on 09/28/2021 Regulatory Role of Fibroblast Lineage of IGFBP2 IOVS j November 2015 j Vol. 56 j No. 12 j 7295

TABLE. Sequences of PCR Primers paraformaldehyde for 15 minutes at room temperature (RT) and then washed three times, 5 minutes each time, with PBS. Product The fixed cells were permeabilized by incubation with 0.2% Gene Primer Sequence Size, bp Triton X-100 for 15 minutes at RT and then rinsed three times a-SMA with PBS. To prevent nonspecific binding, the slides were 0 0 incubated with a blocking agent (2% bovine serum albumin in Sense 5 -CCGTGATCTCCTTCTGCATT-3 175 PBS) for 30 minutes at RT. The slides were then incubated with Antisense 50- -30 CTGTTCCAGCCATCCTTCAT anti–a-SMA (1:50; Merck Millipore) antibodies. After three IGFBP1 rinses in PBS for 5 minutes each time, the slides were Sense 50-TCAAAAAATGGAAGGAGCCCT-30 127 incubated with secondary antibodies conjugated with fluores- Antisense 50-AATCCATTCTTGTTGCAGTTT-30 cein isothiocyanate (FITC; 1:100, Bethyl Laboratories, Mont- gomery, TX, USA) for 1 hour in the dark. After washing with IGFBP2 PBS (5 minutes each time, three times for each slide), Sense 50-CACCGGCAGATGGGCAA-30 136 0 0 0 coverslips were mounted using Fluoroshield with 4 ,6-diami- Antisense 5 -GAAGGCGCATGGTGGAGAT-3 dino-2-phenylindole (DAPI; Sigma-Aldrich Corp.) to visualize IGFBP3 nuclei and combat fading of the immunolabeling. Sense 50-CTACAAAGTTGACTACGAGTC-30 139 Antisense 50-ACTCAGCACATTGAGGAACTT-30 Immunodot Blot Analysis IGFBP4 The levels of growth factors in the conditioned medium were Sense 50-TCGAGGCCATCCAGGAAA-30 165 evaluated using a custom cytokine antibody array kit (RayBio- Antisense 50-CCCCATTGACCTTCATCTT-30 tech, Inc., Norcross, GA, USA). Cytokine array membranes IGFBP5 were analyzed according to manufacturer’s protocols. The membranes were detected using an ECL Plus detection system Sense 50-AGCAAGTCAAGATCGAGAGAGA-30 154 included with the kit and signals were directly digitized using Antisense 50-TTCTTTCTGCGGTCCTTCTTCA-30 ChemiDoc XRS (BioRad). Blot density measurements were IGFBP6 obtained with a Personal Molecular Imager FX (BioRad) Sense 50-AGAGGAGAATCCTAAGGAGAGT-30 229 supported by imaging analysis software (Quantity One, Antisense 50-ATTGGGCACGTAGAGTGTTTGA-30 Imaging Research, Inc., Ontario, Canada). IGF-I Sense 50-ATCAGCAGTCTTCCAACCCAA-30 102 RNA Interference 0 0 Antisense 5 -CAGCGCCAGGTAGAAGAGAT-3 Small interfering RNAs (siRNAs) targeting the human mRNA IGF-II sequences of IGFBP2 were purchased from Dharmacon Sense 50-ACACCCTCCAGTTCGTCTGTG-30 248 (SMARTpool: siGENOME IGFBP2 siRNA). A siGENOME Non- Antisense 50-CTGCTTCCAGGTGTCATATT-30 targeting siRNA pools (Dharmacon Products, Lafayette, CO, USA) was used as the negative control. Human corneal GAPDH fibroblasts with 70% to 80% confluence were transfected with Sense 50-TGTGGTCATGAGTCCTTCCA-30 294 IGFBP2 or control siRNA by using lipofectamine (RNA iMAX; Antisense 50-CGAGATCCCTCCAAAATCAA-30 Invitrogen-Gibco) following manufacturer’s instructions. The efficacy of knockdown of IGFBP2 was assessed by qRT-PCR and Western blotting.

Western Blot Analysis Statistical Analysis Western blot analysis of cultured HCFs was performed as Data are presented as the mean 6 SD. Analysis of variance previously described.10 Primary mouse monoclonal antibodies (ANOVA) followed by Bonferroni’s post hoc analysis and two- against human keratocan, lumican (1:1000; MD Biosciences, tailed Student’s t-test were used for statistical analysis of Inc., St. Paul, MN, USA), ALDH1A1 (1:1000; Cell Signaling multiple groups and pairwise comparison, respectively. All Technology, Inc., Danvers, MA, USA), and a a-SMA (1:1000; experiments in this study were repeated in three or more Merck Millipore, Darmstadt, Germany) were diluted in tris- separate trials. SPSS software version 20.0 (SPSS, Chicago, IL, buffered saline (TBS), applied to the membrane, and incubated USA) was used for all statistical analyses and P less than 0.05 overnight at 48C. Secondary antibodies were then diluted in indicated statistical significance. TBS (1:2000), applied to the membrane, and incubated for 1 hour at room temperature. The protein signal after the application of secondary antibody was visualized using an RESULTS enhanced chemiluminescence (ECL) Western blotting detec- tion kit (Pierce Biotechnology, Inc., Rockfold, IL, USA), and b- Expression of IGFBP Family Proteins, IGF-1, and -2 actin (Sigma-Aldrich Corp., St. Louis, MO, USA) was used as a in Conditioned Medium of Various Conditions and loading control. Image analysis of the immunobands was in AM Extracts performed using ImageJ software ver. 1.46 (http://imagej.nih. gov/ij/; provided in the public domain by the National We intended to identify the factors, which are expressed Institutes of Health, Bethesda, MD, USA). significantly higher in both of KCM and AM extracts compared with corneal fibroblast conditioned medium (CFCM) from CFs Immunocytochemistry which were grown on plastic dishes. Moreover, AM condi- tioned medium (AMCM) without keratocytes was analyzed to Human corneal fibroblasts were cultured on coverslips. The exclude the AM-related effects in KCM, which is conditioned coverslips were briefly washed with PBS, fixed in 4% medium from keratocytes grown on AM. Thus, the immunodot

Downloaded from iovs.arvojournals.org on 09/28/2021 Regulatory Role of Fibroblast Lineage of IGFBP2 IOVS j November 2015 j Vol. 56 j No. 12 j 7296

FIGURE 1. Secreted level of IGFBP family proteins IGF-I and -II in diverse conditioned media and amniotic membrane (AM) extract. (A) Immunodot blot analysis of conditioned media derived from keratocytes, corneal fibroblasts and AM, and of AM extracts. (B) Custom human growth factor antibody array map used in immunodot blot analysis. (C) The level of IGFBP1, 2, 4, 6, and IGF-II significantly differed according to four groups. The IGFBP2 level revealed higher in both keratocyte conditioned medium (KCM) and AM extract than in corneal fibroblast conditioned medium (CFCM), while IGFBP1, 4, and 6 were highly expressed in KCM or AM extract only, when compared with CFCM. ***P < 0.001 (ANOVA among four groups; KCM, CFCM, AMCM, and AM extract); ###P < 0.001, #P < 0.05 (Bonferroni’s post hoc, KCM, or AM extract versus CFCM).

blot assay was performed to detect growth factors in shown). Furthermore, in order to clarify the effects of IGFBP2 conditioned medium derived from keratocytes, corneal fibro- in hMSC cultured in KCM, MSCs were treated with IGFBP2 blasts, AM and in AM extracts (Figs. 1A, 1B). protein and we analyzed the expression pattern of markers of Among IGFBP family proteins, IGF-1 and -2, only IGFBP2 MSC and keratocyte. After treating the hMSCs with various showed significantly higher expression in both KCM (P < concentrations (100–500 ng/mL) of the IGFBP2 for 24 hours, 0.001) and AM extracts (P < 0.001) compared with in CFCM. mRNA level of a-SMA was decreased (Fig. 2B). A similar effect Moreover, the difference of IGFBP2 level between in KCM and was observed when 100 ng/mL of IGFBP2 was added for in conditioned medium from AM was significant (P < 0.001) various durations (24–72 hours; Fig. 2C). indicating that prominent existence of IGFBP2 in KCM may not We confirmed the downregulation of a-SMA, an MSC be derived from AM Although IGFBP1 revealed high expression marker, by IGFBP2 in hMSCs using Western blot analysis. in AM extracts (P < 0.001, in KCM versus in CFCM), IGFBP1 Additionally, expression of keratocyte markers, including level between in KCM and CFCM showed no significant keratocan and ALDH1A1, was strongly increased in hMSCs by difference (P ¼ 1.000). Likewise, IGFBP4 and IGFBP6 levels IGFBP2, especially at the concentration of 500 ng/mL were higher in KCM than in CFCM (P ¼ 0.015 and P < 0.001, compared with lower concentrations of IGFBP2. However, respectively), but no significant difference was noted between lumican expression was not clearly influenced by IGFBP2 (Fig. in CFCM and in AM extract (P ¼ 1.000 and P ¼ 0.059, 2D). respectively). Insulin-like growth factor 2 expression was most prominent in CFCM and levels of IGFBP and IGF-2 showed no Maintenance of Corneal Phenotype by IGFBP2 in difference according to mediums (Fig. 1C). HCFs Acquisition of Corneal Phenotype by IGFBP2 in Stromal cells grown in AM are good at maintaining their cell 29 HMSCS shape and marker expression. However, when cultured in plastic dishes, corneal stromal cells rapidly lose their dendritic Human mesenchymal stem cells were seeded on a culture dish, morphology and acquire a fibroblastic shape.35,36 Therefore, and then cultured with KCM. When MSCs were grown in KCM the former is called a keratocyte, while the latter is referred to for two passages, mRNA expression of IGFBP2 was significant- as a corneal fibroblast. ly increased (Fig. 2A), but other IGFBP family proteins, IGF-1 Keratocytes strongly expressed keratocan, lumican, ALD- and -2 were not changed by real-time PCR analysis (data not H1A1, and IGFBP2, but a-SMA expression was negligible.

Downloaded from iovs.arvojournals.org on 09/28/2021 Regulatory Role of Fibroblast Lineage of IGFBP2 IOVS j November 2015 j Vol. 56 j No. 12 j 7297

FIGURE 2. Upregulation of IGFBP2 and inhibition of a-SMA in human mesenchymal stem cells (hMSCs). (A) The relative expression of IGFBP2 mRNA in MSCs statistically increased in the presence of KCM. After MSCs were treated with IGFBP2, (B) a-SMA was downregulated at various concentrations (at 24 hours) and (C) at various durations ranging from 24 to 72 hours. There was no significant difference of a-SMA mRNA expression according to the concentration (B) or treating time (C) of IGFBP2. (D) Insulin-like growth factor-BP2 downregulated protein expression of a-SMA and upregulated the expression of IGFBP2 and corneal markers including keratocan and ALDH1A1 with statistical significance. Especially, IGFBP 500 ng/mL showed prominent alteration of expressions of a-SMA, IGFBP2, keratocan, and ALDH1A1 compared with lower concentrations of IGFBP2. **P < 0.01, *P < 0.05, versus control. ##P < 0.01, #P < 0.05.

However, in corneal fibroblasts, a-SMA expression gradually expression of a-SMA in HCFs increased dependent upon increased and expression of IGFBP2, keratocan, and ALDH1A1 treatment duration, but there was no definite change of decreased over the culture passages (Fig. 3A). In addition, the expression pattern of a-SMA according to the concentration of expression of a-SMA was inhibited, and keratocan and TGFb1 used (Fig. 4A). ALDH1A1 were upregulated by IGFBP2 in HCFs, which was The a-SMA expression induced by TGFb1 was decreased by similar to the effect of KCM itself (Figs. 3B, 3C). Moreover, IGFBP2, although IGFBP2 could not reverse the a-SMA treatment with IGFBP2 or culture in KCM increased the expression in HCFs when treating time of TGFb1 was extended expression of intracellular IGFBP2 in HCFs. The expression of to 72 hours. Moreover, IGFBP2 upregulated the expression of keratocan and ALDH1A1 were increased depending on the ALDH1A1 and keratocan in TGFb1-treated HCFs, but IGFBP2 increased expression of IGFBP2. However, there was no could not restore the corneal phenotype of HCFs when they significant change in lumican expression. Human corneal had been treated with TGFb1 for 72 hours (Figs. 4B, 4C). fibroblasts without IGFBP2 or KCM did not alter the expression Expression of lumican was not affected by IGFBP2 (data not level of all of a-SMA, IGFBP2, keratocan, ALDH1A1, and lumican until 72 hours (data not shown). shown). We confirmed this phenomenon again using immunocyto- chemical staining. The formation of prominent stress fibers by Restoration of Corneal Phenotype by IGFBP2 in TGFb1 in HCFs was visualized by a-SMA staining. Nearly all TGFb1-Induced Corneal Myofibroblasts cells obtained a myofibroblast-like shape and showed strong Transforming growth factor–b1 is known to play a central role expression of a-SMA depending on the duration of TGFb1 in fibroblast activation and fibroblast-to-myofibroblast differen- treatment. Alpha-SMA–expressing stress fibers, which were tiation, and induces the expression of a-SMA, a marker of more prominent in TGFb1-treated HCFs than control HCFs, myofibroblasts.37 Therefore, we treated HCFs with various were attenuated by IGFBP2 for 24 to 24 hours. There was no concentrations of TGFb1 (5–20 ng/mL) for various durations change of expression pattern of a-SMA in HCFs after 72 hours (24–72 hours) to prepare the corneal myofibroblasts. The (Fig. 4D).

Downloaded from iovs.arvojournals.org on 09/28/2021 Regulatory Role of Fibroblast Lineage of IGFBP2 IOVS j November 2015 j Vol. 56 j No. 12 j 7298

FIGURE 3. Expression of marker proteins in cultured HCFs according to with or without IGFBP2 or KCM. (A) As the passage of culture progresses, IGFBP2, keratocan, and ALDH1A1 were downregulated, while a-SMA was upregulated in HCFs. (B) However, when incubated with IGFBP2 or KCM, a-SMA expression decreased steadily and the expression of IGFBP2, keratocan, and ALDH1A1 increased. (C) The protein expression pattern by Western blot analysis (A, B) was confirmed statistically based on density measurements. Change of lumican expression by IGFBP2 was not statistically significant. **P < 0.01, *P < 0.05; #, versus keratocytes; ¥, versus control (HCF [P8]).

Attenuation of Corneal Phenotypes by Knockdown DISCUSSION of IGFBP2 in HCFs Corneal stromal cells, keratocytes, secrete cornea-specific ECM To more specifically elucidate the role of IGFBP2 in HCFs, we components, including keratocan sulfate, lumican, and kerato- 38 performed knockdown of IGFBP2 expression in HCFs using can. They play an important role in the maintenance of 39,40 IGFBP2 siRNA. Insulin-like growth factor binding protein 2 corneal transparency. In conditions of disrupted tissue homeostasis, such as after injury, during wound healing, or siRNA–transfected HCFs revealed decreased of IGFBP2 expres- chronic inflammation, keratocytes tend to differentiate into sion in mRNA and protein level to approximately 0.5-fold and fibroblasts and myofibroblasts and deposit a less-organized 0.3-fold, respectively, while nontargeting (negative control) collagen-fibrillar construct in a pattern with similarities to siRNA-transfected cells and vehicle alone (mock) showed no corneal scar tissue due to a lack of cornea-specific ECM difference. Knockdown of IGFBP2 in HCFs cells increased a- components.38 If this differentiation process can be alleviated, SMA expression and inhibited the expression of ALDH1A1 the regulation of corneal scar formation may be possible. In significantly (Figs. 5A, 5B). this study, we analyzed the components of the KCM and AM

Downloaded from iovs.arvojournals.org on 09/28/2021 Regulatory Role of Fibroblast Lineage of IGFBP2 IOVS j November 2015 j Vol. 56 j No. 12 j 7299

FIGURE 4. The change in protein expression of myofibroblast and keratocyte markers by IGFBP2. (A) Human corneal fibroblasts HCFs were treated with TGFb1 under various concentrations and times. Transforming growth factor–b1 induced expression of a-SMA in HCFs, the myofibroblast transdifferentiation marker. (B) After HCFs were cotreated with TGFb1 and IGFBP2, relative mRNA expression of TGFb1-induced a-SMA was downregulated and keratocan was upregulated by treatment with IGFBP2 for 24 hours (a-SMA) and for 24 to 48 hours (keratocan), respectively, although there was no change when treated for 72 hours. (C) Protein expression of a-SMA, keratocan, and ALDH1A1 by Western blot analysis after treated with TGFb1 and IGFBP2 revealed a pattern similar to the results of qRT-PCR. (D) Immunofluorescence staining of a-SMA expression in HCFs with TGFb1 and IGFBP2 according to treating time (24–72 hours). Expressed a-SMA–positive stress fibers were decreased by IGFBP2 cotreatment for 24 and 48 hours. The proposed mechanism explaining such a phenomenon is also illustrated. **P < 0.01, *P < 0.05.

Downloaded from iovs.arvojournals.org on 09/28/2021 Regulatory Role of Fibroblast Lineage of IGFBP2 IOVS j November 2015 j Vol. 56 j No. 12 j 7300

FIGURE 5. Aggravated myofibroblast transformation and attenuated expression of keratocyte phenotypes induced by IGFBP2 siRNA in HCFs. (A) Knockdown of IGFBP2 increased a-SMA mRNA level and decreased ALDH1A1 mRNA level. (B) Western blot analysis of a-SMA, ALDH1A1, and keratocan revealed that knockdown of IGFBP2 protein in HCFs attenuated ALDH1A1 level and uninhibited a-SMA expression. There was no significant change in keratocan expression before and after knockdown of IGFBP2 similar to in mRNA level. N/C, negative control. **P < 0.01, *P < 0.05.

extract that are likely involved in the homeostatic maintenance the expression of keratocyte markers. The HCFs reacquired the of corneal stroma, and IGFBP2 was identified as a key properties of keratocytes after IGFBP2 treatment. This effect candidate for such a mechanism. was reproduced in the same manner as KCM treatment. In the present study, IGFBP2 was strongly identified in both Furthermore, knockdown of IGFBP2 deprived HCFs of corneal KCM and AM extract among IGFBP family proteins, IGF-1 and phenotype (ALDH1A1) and promoted the acquisition of -2 when compared with in CFCM and AMCM. It is interesting myofibroblast feature. These serial results suggest that IGFBP2 that IGFBP2 is abundant in AM itself as appears by the immune- may normally serve to maintain the corneal phenotype against expression of IGFBP2 in AM extract, while there is lack of to be myofibroblast transformation. IGFBP2 in AMCM although IGFBP2 is highly expressed in KCM A central feature of activated stromal cells is the acquisition from keratocytes cultured on AM. We thought that IGFBP2 in of smooth muscle features, most notably neoformation of KCM was derived not from AM but from keratocytes and contractile stress fibers and expression of a-SMA; hence, the inherited IGFBP2 in AM might interact with or force keratocyte name myofibroblast. The transient acquisition of this pheno- to excrete IGFBP2. type is beneficial for normal tissue repair processes when We confirmed that the hMSCs would differentiate into myofibroblast remodeling activities restore and preserve tissue keratocyte-like cells after treatment with IGFBP2. This is integrity. However, persistence of myofibroblast transformation similar to the effect of KCM treatment as reported in our results in tissue stiffening and deformation. In fibrosis, stiff scar previous study33 in which the differentiation-inducing effect of tissue alters normal organ function; the mechanical and a single treatment of IGFBP2 was confirmed to be equally chemical conditions generated by myofibroblasts promote powerful as KCM. Thereafter, we investigated whether IGFBP2 disease progression.41 Moreover, fibrosis of the cornea can has the effect of maintaining the characteristics of the lead to corneal opacification and subsequent loss of vision.42,43 keratocytes in in vitro culture, similar to its effects in the Therefore, the maintenance of keratocyte characteristics can induction of differentiation in MSCs. In addition, treating HCFs be linked to the maintenance of corneal clarity and homeosta- with IGFBP2 reduced their expression of a-SMA and increased sis. In this context, our results show antifibrotic potential that

Downloaded from iovs.arvojournals.org on 09/28/2021 Regulatory Role of Fibroblast Lineage of IGFBP2 IOVS j November 2015 j Vol. 56 j No. 12 j 7301

the keratocyte-fibroblast lineage that can be controlled by migration and binds to the alpha 5 beta 1 by means of IGFBP2. Although IGFBP2 has been previously shown to be its Arg-Gly-Asp sequence. Proc Natl Acad Sci U S A. 1993;90: involved in abrogation of proliferation and of biosynthesis of 10553–10557. collagen, fibulin and fibronectin in liver myofibroblasts against 7. Russo VC, Bach LA, Fosang AJ, Baker NL, Werther GA. Insulin- liver fibrogenesis44 with a similar mechanism to in cornea as like growth factor binding protein-2 binds to cell surface seen in our study, there have been no studies regarding this proteoglycans in the rat brain olfactory bulb. Endocrinology. kind of effect of IGFBP2 in cornea. 1997;138:4858–4867. Inhibition of a-SMA and corneal markers by IGFBP2 was 8. Russo VC, Schutt BS, Andaloro E, et al. Insulin-like growth affected by the status of TGFb1-induced myofibroblast factor binding protein-2 binding to extracellular matrix plays a transformation. When HCFs were treated by TGFb1 for 72 critical role in neuroblastoma cell proliferation, migration, and hours, there was no significant influence of IGFBP2 on the invasion. Endocrinology. 2005;146:4445–4455. reverse of expression of a-SMA, keratocan, and ALDH1A1 9. Conover CA, Powell DR. Insulin-like growth factor (IGF)- unlike at 24 and 48 hours (Fig. 4C). We suggest that the binding protein-3 blocks IGFI-induced receptor down-regula- expression of all those markers is probably variable depending tion and cell desensitization in cultured bovine fibroblasts. on the exposure time to TGFb1, which can affect the severity Endocrinology. 1991;129:710–716. of myofibroblast transformation. Furthermore, it is speculated 10. Sharples AP, Al-Shanti N, Stewart CE. C2 and C2C12 murine that IGFBP2 may regulate the myofibroblast transition as a skeletal myoblast models of atrophic and hypertrophic possible antifibrotic strategy, particularly in the early stages of potential: relevance to disease and ageing? J Cell Physiol. cell transformation. 2010;225:240–250. In analyses of expression alterations of a-SMA by IGFBP2 11. Ernst CW, McCusker RH, White ME. and treatment in hMSCs, there was discordance of expression secretion of insulin-like growth factor-binding proteins during between mRNA and protein levels (Figs. 2B, 2D). Unlike mRNA myoblast differentiation. Endocrinology. 1992;130:607–615. levels, protein levels of a-SMA, IGFBP2, and corneal markers 12. Firth SM, Baxter RC. Cellular actions of the insulin-like growth including keratocan and ALDH1A1 revealed their gradual factor binding proteins. Endocr Rev. 2002;23:824–854. change of expressions toward the increase of IGFBP concen- 13. Hoeflich A, Nedbal S, Blum WF, et al. Growth inhibition in trations. As a possible mechanism, we suggest that there may giant transgenic mice by overexpression of be possible unknown impact of microRNAs on translation or insulin-like growth factor-binding protein-2. Endocrinology. process of posttranscriptional regulation as proposed previ- 2001;142:1889–1898. ously45 and think that this would be an interesting issue of 14. Jones JI, Clemmons DR. Insulin-like growth factors and their future study. binding proteins: biological actions. Endocr Rev. 1995;16:3– In conclusion, we demonstrated the novel effect of IGFBP2; 34. regulation of the differentiation of corneal fibroblasts. Although further studies are necessary to determine the 15. Collett-Solberg PF, Cohen P. The role of the insulin-like growth specific mechanism of IGFBP2, our results suggest the IGFBP2 factor binding proteins and the IGFBP proteases in modulating IGF action. . 1996;25:591– may be considered to be a novel antifibrotic strategy in corneal Endocrinol Metab Clin North Am 614. diseases. 16. Zapf J. Physiological role of the insulin-like growth factor binding proteins. Eur J Endocrinol. 1995;132:645–654. Acknowledgments 17. Cuthbertson RA, Beck F, Senior PV, Haralambidis J, Penschow Supported by grants from the Basic Science Research Program JD, Coghlan JP. Insulin-like growth factor II may play a local through the National Research Foundation of Korea (NRF) funded role in the regulation of ocular size. Development. 1989;107: by the Ministry of Education, Science, and Technology 123–130. (2015R1A2A2A01004643; Daejeon, South Korea), and by a grant 18. Bassnett S, Beebe DC. Localization of insulin-like growth from the Korea Healthcare technology R&D Project, Ministry for factor-1 binding sites in the embryonic chick eye. Invest Health, Welfare & Family Affairs, Republic of Korea (HI12C1376; Ophthalmol Vis Sci. 1990;31:1637–1643. Cheongju, Chungcheongbuk-do, South Korea). 19. Waldbillig RJ, Arnold DR, Fletcher RT, Chader GJ. Insulin and Disclosure: S.H. Park, None; K.W. Kim, None; J.C. Kim, None IGF-I binding in developing chick neural retina and pigment epithelium: a characterization of binding and structural References differences. Exp Eye Res. 1991;53:13–22. 20. Waldbillig RJ, Pfeffer B, Schoen TJ, et al. Evidence for an 1. De Pablo F, Perez-Villami B, Serna J, et al. IGF-I and the IGF-I insulin-like growth factor-I autocrine-paracrine system in the receptor in development of nonmammalian vertebrates. Mol retinal photoreceptor-pigment epithelial cell complex. J Reprod Dev. 1993;35:427–433. Neurochem. 1991;57:1522–1533. 2. Heyner S, Shi C, Garside WT, Smith RM. Functions of the IGFs 21. Ocrant I, Fay CT, Parmelee JT. Expression of insulin and in early mammalian development. Mol Reprod Dev. 1993;35: insulin-like growth factor receptors and binding proteins by 421–426. retinal pigment epithelium. Exp Eye Res. 1991;52:581–589. 3. Mohan S, Baylink DJ. IGF-binding proteins are multifunctional 22. Arnold DR, Moshayedi P, Schoen TJ, Jones BE, Chader GJ, and act via IGF-dependent and -independent mechanisms. J Waldbillig RJ. Distribution of IGF-I and -II, IGF binding proteins Endocrinol. 2002;175:19–31. (IGFBPs) and IGFBP mRNA in ocular fluids and tissues: 4. Binoux M, Hossenlopp P. Insulin-like growth factor (IGF) and Potential sites of synthesis of IGFBPs in aqueous and vitreous. IGF-binding proteins: comparison of human serum and lymph. Exp Eye Res. 1993;56:555–565. J Clin Endocrinol Metab. 1988;67:509–514. 23. Izumi K, Kurosaka D, Iwata T, et al. Involvement of insulin-like 5. Jones JI, Gockerman A, Busby WH Jr, Camacho-Hubner C, growth factor-i and insulin-like growth factor binding protein-3 Clemmons DR. Extracellular matrix contains insulin-like in corneal fibroblasts during corneal wound healing. Invest growth factor binding protein-5: potentiation of the effects Ophthalmol Vis Sci. 2006;47:591–598. of IGF-I. J Cell Biol. 1993;121:679–687. 24. Robertson DM, Ho SI, Hansen BS, Petroll WM, Cavanagh HD. 6. Jones JI, Gockerman A, Busby WH Jr, Wright G, Clemmons DR. Insulin-like growth factor binding protein-3 expression in the Insulin-like growth factor binding protein 1 stimulates cell human corneal epithelium. Exp Eye Res. 2007;85:492–501.

Downloaded from iovs.arvojournals.org on 09/28/2021 Regulatory Role of Fibroblast Lineage of IGFBP2 IOVS j November 2015 j Vol. 56 j No. 12 j 7302

25. Burren CP, Berka JL, Batch JA. Localization studies of IGFBP2 35. Beals MP, Funderburgh JL, Jester JV, Hassell JR. Proteoglycan and IGFBP-5 in the anterior compartment of the eye. Curr Eye synthesis by bovine keratocytes and corneal fibroblasts: Res. 1997;16:256–262. maintenance of the keratocyte phenotype in culture. Invest 26. Schoen TJ, Beebe DC, Clemmons DR, Chader GJ, Waldbillig RJ. Ophthalmol Vis Sci. 1999;40:1658–1663. Local synthesis and developmental regulation of avian vitreal 36. Jester JV, Barry-Lane PA, Cavanagh HD, Petroll WM. Induction insulin-like growth factor-binding proteins: a model for of a-smooth muscle actin expression and myofibroblast independent regulation in extravascular and vascular com- transformation in cultured corneal keratocytes. Cornea. partments. Endocrinology. 1992;131:2846–2854. 1996;15:505–516. 27. Yang YW, Brown DR, Robcis HL, Rechler MM, De-Pablo F. 37. Desmouliere A, Geinoz A, Gabbiani F, Gabbiani G. Transform- Developmental regulation of insulin-like growth factor binding ing growth factor-beta 1 induces alpha-smooth muscle actin protein-2 in chick embryo serum and vitreous humor. Regul expression in granulation tissue myofibroblasts and in Pept. 1993;48:145–155. quiescent and growing cultured fibroblasts. J Cell Biol. 1993; 28. Schoen TJ, Bondy CA, Zhou J, et al. Differential temporal and 122:103–111. spatial expression of insulin-like growth factor binding 38. Wu J, Du Y, Mann MM, Funderburgh JL, Wagner WR. Corneal protein-2 in developing chick ocular tissues. Invest Ophthal- stromal stem cells versus corneal fibroblasts in generating mol Vis Sci. 1995;36:2652–2662. structurally appropriate corneal stromal tissue. Exp Eye Res. 29. Espana EM, He H, Kawakita T, et al. Human keratocytes 2014;120:71–81. cultured on amniotic membrane stroma preserve morphology 39. Carlson EC, Liu CY, Chikama T, et al. Keratocan, a cornea- and express keratocan. Invest Ophthalmol Vis Sci. 2003;44: specific keratan sulfate proteoglycan, is regulated by lumican. 5136–41. J Biol Chem. 2005;280:25541–25547. 30. Espana EM, Kawakita T, Liu CY, Tseng SCG. CD-34 expression 40. Kao WW, Liu CY. Roles of lumican and keratocan on corneal by cultured human keratocytes is downregulated during transparency. Glycoconj J. 2002;19:275–285. myofibroblast differentiation induced by TGF-beta1. Invest 41. Otranto M, Sarrazy V, Bonte´ F, Hinz B, Gabbiani G, Desmouliere` Ophthalmol Vis Sci. 2004;45:2985–2991. A. The role of the myofibroblast in tumor stroma remodeling. 31. Kawakita T, Espana EM, He H, et al. Keratocan expression of Cell Adh Migr. 2012;6:203–219. murine keratocytes is maintained on amniotic membrane by 42. Friedlander M. Fibrosis and diseases of the eye. J Clin Invest. downregulating TGF-beta signaling. J Biol Chem. 2005;280: 2007;117:576–586. 27085–27092. 43. Wilson SL, El Haj AJ, Yang Y. Control of scar tissue formation in 32. Li W, He H, Chen YT, Hayashida Y, Tseng SC. Reversal of the cornea: strategies in clinical and corneal tissue engineer- myofibroblasts by amniotic membrane stromal extract. J Cell ing. J Funct Biomater. 2012;3:642–687. Physiol. 2008;215:657–64. 44. Pannem R, Hoeflich A, Wolf E, Scharf JG. Functional analysis of 33. Park SH, Kim KW, Chun YS, Kim JC. Human mesenchymal IGFBP2 overexpression in mouse liver myofibroblasts. Z stem cells differentiate into keratocyte-like cells in keratocyte- Gastroenterol. 2009;47:P329. conditioned medium. Exp Eye Res. 2012;101:16–26. 45. Wei YN, Hu HY, Xie GC, et al. Transcript and protein 34. Nagai A, Kim WK, Lee HJ, et al. Multilineage potential of stable expression decoupling reveals RNA binding proteins and human mesenchymal stem cell line derived from fetal marrow. miRNAs as potential modulators of human aging. Genome PLoS One. 2007;2:e1272. Biol. 2015;16:41.

Downloaded from iovs.arvojournals.org on 09/28/2021