The American Journal of Pathology, Vol. 186, No. 4, April 2016
ajp.amjpathol.org
EPITHELIAL AND MESENCHYMAL CELL BIOLOGY
aB-Crystallin Regulates Subretinal Fibrosis byModulation of Epithelial-Mesenchymal Transition
Keijiro Ishikawa,*y Parameswaran G. Sreekumar,* Christine Spee,y Hossein Nazari,y Danhong Zhu,z Ram Kannan,* andDavid R. HintonyzFrom the Arnold and Mabel Beckman Macular Research Center,* Doheny Eye Institute, Los Angeles; and the Departments of Ophthalmologyy andPathology,z Keck School of Medicine of the University of Southern California, Los Angeles, California
Accepted for publication
C
h
November 19, 2015.
Address correspondence toDavid R. Hinton, M.D.,Departments of Ophthalmologyand Pathology, Keck School ofMedicine of the University ofSouthern California, 2011Zonal Ave, Hoffman MedicalResearch 209, Los Angeles, CA90033. E-mail: [email protected].
opyright ª 2016 American Society for Inve
ttp://dx.doi.org/10.1016/j.ajpath.2015.11.014
Subretinal fibrosis is an end stage of neovascular age-related macular degeneration, characterized byfibrous membrane formation after choroidal neovascularization. An initial step of the pathogenesis is anepithelial-mesenchymal transition (EMT) of retinal pigment epithelium cells. aB-crystallin plays multipleroles in age-related macular degeneration, including cytoprotection and angiogenesis. However, the role ofaB-crystallin in subretinal EMT and fibrosis is unknown. Herein, we showed attenuation of subretinalfibrosis after regression of laser-induced choroidal neovascularization and a decrease in mesenchymalretinal pigment epithelium cells in aB-crystallin knockout mice compared with wild-type mice. aB-crystallinwas prominently expressed in subretinal fibrotic lesions in mice. In vitro, overexpression of aB-crystallininduced EMT, whereas suppression of aB-crystallin induced a mesenchymal-epithelial transition. Trans-forming growth factor-b2einduced EMT was further enhanced by overexpression of aB-crystallin but wasinhibited by suppression of aB-crystallin. Silencing of aB-crystallin inhibited multiple fibrotic processes,including cell proliferation, migration, and fibronectin production. Bone morphogenetic protein 4 up-regulated aB-crystallin, and its EMT induction was inhibited by knockdown of aB-crystallin. Further-more, inhibition of aB-crystallin enhanced monotetraubiquitination of SMAD4, which can impair itsnuclear localization. Overexpression of aB-crystallin enhanced nuclear translocation and accumulationof SMAD4 and SMAD5. Thus, aB-crystallin is an important regulator of EMT, acting as a molecularchaperone for SMAD4 and as its potential therapeutic target for preventing subretinal fibrosis devel-opment in neovascular age-related macular degeneration. (Am J Pathol 2016, 186: 859e873; http://dx.doi.org/10.1016/j.ajpath.2015.11.014)
Supported by NIH grants EY03040 and EY01545, the Arnold and MabelBeckman Foundation, the Japanese Society for the Promotion of Science,and a Japan Society for the Promotion of Science Postdoctoral Fellowshipsfor Research Abroad fellowship (K.I.).
R.K. and D.R.H. contributed equally to this work as senior authors.Disclosures: None declared.Current address of K.I., Department of Ophthalmology, Kyushu Uni-
versity Graduate School of Medical Sciences, Fukuoka, Japan.
Age-related macular degeneration (AMD) is a leading causeof blindness because of progressive degeneration of themacular region of the retina that is responsible for visualacuity and color vision. The natural history of AMD is aprogression from its early stage to the two forms of latestage of AMD: geographic atrophy and neovascular AMD(nAMD).1
The visual prognosis for nAMD is poor; the conditionprogresses rapidly with the development of choroidalneovascularization (CNV) and subsequent subretinalfibrosis. Although the commonly used treatment withanti-vascular endothelial growth factor (VEGF) drugsimproves visual acuity in nAMD patients, the subretinalscarring (fibrosis) that may develop in approximately halfof all antieVEGF-treated eyes within 2 years has been
stigative Pathology. Published by Elsevier Inc
identified as a cause of unsuccessful outcomes.2 Thus,therapeutic strategies for the inhibition of subretinalfibrosis are currently an active area of investigation.Among the critical growth factors involved in subretinalfibrosis, platelet-derived growth factor (PDGF) is a po-tential therapeutic target. Currently, several clinical trialsfor the treatment for nAMD have been evaluating the
. All rights reserved.
Ishikawa et al
efficacy of dual VEGF/PDGF inhibitors, such as thefollowing: E10030 (Ophthotech, New York, NY), an anti-PDGF pegylated aptamer as an adjunct to anti-VEGFtherapy; sorafenib, an inhibitor of VEGF receptor,PDGF receptor, and Raf kinases; and pazopanib, aninhibitor of VEGF receptor, PDGF receptor, and c-kit.3
In addition, it has been shown that nucleotide-bindingoligomerization domain-, leucine-rich repeat domain-, andpyrin domainecontaining 3 inflammasome activation isimplicated in the pathogenesis of nAMD.4 The recentimplication of inflammasome activation in the pathogen-esis of hepatic fibrosis5 suggests that the inflammasomeshould be further evaluated for its potential role in sub-retinal fibrosis. Important inflammasome effector cyto-kines, IL-1b and IL-18, can be potential therapeutic targetsfor the treatment of subretinal fibrosis in nAMD. In supportof this contention, inhibition of IL-1b has been shown toinhibit the development of experimental CNV in mice.6
Subretinal fibrosis results from an excessive woundhealing response, characterized by fibrous membrane for-mation after CNV. In fibrous membranes, the main cellularcomponents are myofibroblasts, the cells immunoreactivefor a-smooth muscle actin (a-SMA). Previous histologicalstudies imply that the source of myofibroblasts can be bothbone marrowederived cells and retinal pigment epithelium(RPE) cells.7,8 After injury to RPE, the cells undergoepithelial-mesenchymal transition (EMT), which enablestransdifferentiation, resulting in the conversion of epithe-lial cells to myofibroblasts. CNV induction can result in therecruitment of more inflammatory cells and fibroblasts,which can be a direct or indirect source of additionalmyofibroblasts. Myofibroblasts play important roles in thedevelopment of subretinal fibrosis, such as proliferation,migration, and extracellular matrix remodeling.8 Althoughprevious studies have indicated the involvement of severalgrowth factors and cytokines in EMT,8 the precisemolecular mechanism and the critical regulators of thisprocess remain to be determined.
The soluble cytoplasmic protein aB-crystallin is aprominent member of the small heat shock protein family.The small heat shock proteins exert diverse biologicalactivities in both normal and stressed cells. They can act asmolecular chaperones by binding misfolded proteins toprevent their denaturation and aggregation.9 aB-crystallincan bind to and stabilize cytoskeleton proteins, such asdesmin and actin, and help to maintain cytoskeletalintegrity.10 The role of aB-crystallin in EMT in liver andlung fibrosis has been recently reported.11,12
Our previous work has suggested an important rolefor aB-crystallin in both the early and late stages ofAMD. The early stage of AMD is characterized by theaccumulation of drusen between the RPE and Bruch’smembrane, accompanied by RPE cell death and syn-aptic dysfunction.13 Geographic atrophy is caused byextensive atrophy and loss of the RPE and the over-lying photoreceptors that rely on the RPE for trophic
860
support.1 aB-crystallin can be seen in RPE, associatedwith drusen and identified as one of the components ofdrusen.14e16 Our laboratory has shown that RPE cellslacking aB-crystallin are more susceptible to oxidativeand endoplasmic reticulum stress compared with normalRPE.17e20 Furthermore, RPE-overexpressing aB-crystallinshows resistance to apoptosis.21 These findings suggestthat aB-crystallin plays a cytoprotective role againstmultiple stress stimuli that can cause RPE cell death,resulting in drusen formation and geographic atrophy. Inaddition, we previously demonstrated that aB-crystallinplays a regulatory role by functioning as a chaperone forVEGF in ocular angiogenesis and may play a part inCNV formation in nAMD.22 However, the involvementof aB-crystallin in subretinal fibrosis in nAMD has notbeen studied.Herein, we examined the pathogenesis of subretinal
fibrosis in aB-crystallin�/� and wild-type (WT) mice; wefurther investigated the role of aB-crystallin in EMT andfibrotic process in cultured RPE cells.
Materials and Methods
Study Approval
All procedures used in these studies were conducted inaccordance with applicable regulatory guidelines at theUniversity of Southern California (Los Angeles, CA),principles of human research subject protection in theDeclaration of Helsinki, and principles of animal researchin the Association for Research in Vision and Ophthal-mology Statement for the Use of Animals in Ophthalmicand Vision Research. Human fetal eyes (16 to 18 weeks ofgestation) were obtained from Advanced Bioscience Re-sources Inc. (Alameda, CA) and Novogenix LaboratoriesLLC (Los Angeles, CA), and written informed consent wasobtained from all donors.
RPE Cell Culture
Human RPE cells were isolated from human fetal eyes(gestational age, 16 to 18 weeks) obtained from AdvancedBioscience Resources Inc. and Novogenix LaboratoriesLLC. Primary cultures of RPE cells were established, asdescribed previously.23 The experiments used cultured RPEcells at two to four passages with normal morphology.
Mice
aB-crystallin�/� mice on the 129 S6/SvEvTac back-ground were generated at the National Eye Institute(Bethesda, MD).24 The 129 S6/SvEvTac WT mice werepurchased from Taconic Farms (Germantown, NY). Malemice, aged between 6 and 8 weeks, were used in allstudies to eliminate sex-related bias in size of laser-induced CNV lesions.25
ajp.amjpathol.org - The American Journal of Pathology
Table 1 siRNA Sequences
aB-crystallin S r(50-CCAGGGAGUUCCACAGGAA-30) dTdTAS r(50-UUCCUGUGGAACUCCCUGG-30) dTdT
Scrambledcontrol
S r(50-UUCUCCGAACGUGUCACGU-30) dTdTAS r(50-ACGUGACACGUUCGGAGAA-30) dTdT
AS, antisense; S, sense.
aB-Crystallin Regulates EMT
Laser-Induced CNV
CNVs were generated as described previously.22 Briefly,laser photocoagulation (532 nm, 150 mW, 50 milliseconds,75 mm) was applied to each fundus using a coverslip as acontact lens on day 0. We made four laser spots per eye forimmunohistochemistry and 20 laser spots for extractingprotein for enzyme-linked immunosorbent assay (ELISA)analysis. The production of a subretinal bubble at the timeof laser application indicated a rupture of Bruch’s mem-brane. We excluded animals that developed burns withbleeding.
CNV and Choroidal Fibrosis Volume Measurement
The volumes of the CNV and choroidal fibrous tissue weremeasured in choroidal flat mounts on days 7, 21, and 35after laser photocoagulation. Mouse eye cups were fixed in4% paraformaldehyde and then permeabilized in 1% TritonX-100 (ICN Biomedicals, Irvine, CA) for 2 hours. Afterremoval of the anterior segment and neural retina, fluorescein-labeled isolectin-B4 (Vector Laboratories, Burlingame, CA)and collagen type I antibody (Rockland ImmunochemicalsInc., Limerick, PA) were added to the mouse eye cup andincubated at 4�C overnight to detect CNV and choroidal
Table 2 List of Antibodies Used in This Study
Target molecule Antibody type
Human aB-crystallin Mouse monoclonalMouse aB-crystallin Rabbit polyclonalPan-cytokeratin Mouse monoclonala-SMA Mouse monoclonalE-cadherin (24E10) Rabbit monoclonalSNAIL Rabbit polyclonalSLUG Rabbit polyclonalSMAD4 Rabbit polyclonalSMAD5 Rabbit polyclonalSMAD2/3 Rabbit polyclonalb-Catenin Rabbit monoclonalNF-kB p65 Rabbit monoclonalNotch2 Rabbit monoclonalFibronectin Mouse monoclonalCollagen type I Rabbit polyclonalUbiquitin Rabbit polyclonalHistone H3 Rabbit monoclonalGAPDH Mouse monoclonalb-Actin Mouse monoclonal
a-SMA, a-smooth muscle actin; GAPDH, glyceraldehyde-3-phosphate dehydrog
The American Journal of Pathology - ajp.amjpathol.org
fibrous tissue, respectively. The secondary antibodyagainst collagen type I was goat anti-rabbit IgG secondaryantibody, Alexa Fluor 647 conjugate (Life Technologies,Grand Island, NY). Samples were coverslipped withVectashield medium (Vector Laboratories) and examinedby use of the Perkin Elmer spinning disk confocal mi-croscope (Perkin Elmer, Waltham, MA). Fluorescencevolume measurements were recorded by generating imagestacks of optical slices within lesions, as previouslydescribed.26
ELISA
Total protein was isolated from the sonicated mouse RPE-choroid complexes using tissue protein extraction reagentwith protease inhibitor (Thermo Scientific, Waltham,MA). Each experiment, consisting of two pooled RPE-choroid complexes, was performed in triplicate (biolog-ical replicates). The concentrations of aB-crystallin andbone morphogenetic protein 4 (BMP4) in the mouseRPE-choroid complexes were measured with ImmunoSetaB-Crystallin ELISA kit (ENZO Life Sciences, Farm-ingdale, NY) and BMP4 BioAssay ELISA Kit (US Bio-logical, Salem, MA), respectively, according to themanufacturer’s instructions.
Transfection
RPE cells were cultured for 24 hours in 6- or 12-well platesor 4-well chamber slides to a confluence of 70% to 90%.siRNA targeting aB-crystallin or control siRNA (Qiagen,Valencia, CA) (sequences included in Table 1) were mixed
Application Source
WB, IHC ENZO Life SciencesIHC ENZO Life SciencesIHC Santa Cruz Biotechnology (Dallas, TX)WB, IHC Sigma-Aldrich (St. Louis, MO)WB, IHC Cell SignalingWB Abgent (San Diego, CA)WB Santa Cruz BiotechnologyIHC Santa Cruz BiotechnologyWB Cell SignalingWB Cell SignalingWB Cell SignalingWB Cell SignalingWB Cell SignalingIHC R&D SystemsWB, IHC Rockland ImmunochemicalsWB Thermo ScientificWB Cell SignalingWB EMD Millipore (Billerica, MA)WB Santa Cruz Biotechnology
enase; IHC, immunohistochemistry; WB, Western blotting.
861
Table 3 Primer Sequences Used for Real-Time RT-PCR
Target molecule Forward primer sequence Reverse primer sequence
aB-crystallin 50-TCCCCAGAGGAACTCAAAGTTAAG-30 50-GGCGCTCTTCATGTTTCCA-30
a-SMA 50-TCTGTAAGGCCGGCTTTGC-30 50-TGTCCCATTCCCACCATCA-30
E-cadherin 50-ATTTTTCCCTCGACACCCGAT-30 50-TCCCAGGCGTAGACCAAGA-30
SNAIL 50-TCGGAAGCCTAACTACAGCGA-30 50-AGATGAGCATTGGCAGCGAG-30
SLUG 50-CGAACTGGACACACATACAGTG-30 50-CTGAGGATCTCTGGTTGTGGT-30
GAPDH 50-ACAGTCGCCGCATCTTCTT-30 50-ACGACCAAATCCGTTGACTC-30
a-SMA, a-smooth muscle actin; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.
Ishikawa et al
(10 nmol/L at final concentration) with RNA transfectionreagent (Lipofectamine RNAiMAX; Life Technologies). Tooverexpress aB-crystallin, pCMV6-XL5 plasmid encodinghuman aB-crystallin (OriGene Technologies, Rockville,MD) or empty vector was mixed with LipofectamineLTX (Life Technologies) or X-tremeGENE DNA (RocheApplied Science, Indianapolis, IN), following the manu-facturer’s protocol. Twenty-four hours after transfection, thecomposite transfection mixture was removed and replacedwith Dulbecco’s modified Eagle’s medium containing 3%fetal bovine serum or serum-free medium for 24 hours,followed with recombinant transforming growth factor(TGF)-b2 (Sigma-Aldrich, St. Louis, MO) or BMP4 (R&DSystems, Minneapolis, MN) treatment before each assay.
Western Blot Analysis
Western blot analysis was performed as described previ-ously.17 After a 48-hour incubation in mediumwith or withoutrecombinant TGF-b2 or BMP4, total cell lysates wereextracted with lysis buffer. In separate experiments, emptyvector or aB-crystallin transfected cells were treated with10 ng TGF-b2 for 2 hours and nuclear and cytosolic frac-tions were separated. The fractionation of the cells intocytosolic and nuclear fractions was performed using anuclear/cytosol fractionation kit (BioVision, Milpitas,CA). The extracted cell lysates were subjected to Tris-HClpolyacrylamide gels, and the blots were incubated withantibodies (Abs) (Table 2) and visualized using an enhancedchemiluminescence (Amersham Pharmacia Biotech, Cleve-land, OH) detection system. Glyceraldehyde-3-phosphatedehydrogenase or histone H3 was used as the loadingcontrol.
RNA Isolation and Real-Time Quantitative RT-PCR
Total RNA was isolated from cells using the RNeasy kit(Qiagen), according to the manufacturer’s instructions. Real-time quantitative RT-PCR was performed, as describedpreviously.22 To eliminate contamination of genomic orplasmid DNA, we pretreated total RNA with DNase (LifeTechnologies) before proceeding to reverse transcription.The primer sequences are shown in Table 3. Gene expressionlevels were normalized relative to glyceraldehyde-3-phosphate dehydrogenase mRNA and reported as fold-change over controls.
862
Immunohistochemistry and ImmunofluorescenceStaining
RPE cells were cultured in 4-well multichamber slides(Thermo Scientific) in serum-free medium for 12 hoursand then stimulated with recombinant TGF-b2 for up to2 days. The cells were rinsed in phosphate-bufferedsaline (PBS) for 3 minutes and fixed and permeabilizedwith ice-cold methanol for 20 minutes. After a 30-minute blocking step with 10% goat serum, the cellswere incubated with primary Abs (Table 2) for 24 hoursat 4�C and then incubated with secondary Abs. Cryostatsections (6 mm thick) of snap-frozen mouse eyes were ob-tained from four WT mice and four aB-crystallin�/� mice.For immunohistochemistry, the sections were treated with0.3% hydrogen peroxide for 10 minutes to block endoge-nous peroxide. The sections were then blocked in 10%normal goat serum for 30 minutes and incubated with Abs(Table 2) overnight at 4�C. Slides were washed with PBSand incubated with biotinylated IgG (Vector Laboratories)for 1 hour. Immunoperoxidase was detected with amino-ethylcarbazole as the chromogen, using an amino-ethylcarbazole kit (Life Technologies), according to themanufacturer’s protocol. Immunofluorescence staining forthe sections was performed as described previously.22
Pan-cytokeratin-positive cells or cells positive for a-SMAor E-cadherin were calculated as the number of theimmunoreactive cells per mm2 of CNV area at day 35 afterlaser, according to the method previously reported.7 Theaverage percentage of the cells immunoreactive for a-SMAor E-cadherin in pan-cytokeratin-positive cells was calcu-lated from four eyes. The specimens were mounted inmounting medium, including DAPI (Vector Laboratories),and examined with a Perkin Elmer spinning disk confocalmicroscope.
Immunofluorescence Localization of SMAD4 andSMAD5
RPE cells transfected with empty vector or vector withaB-crystallin grown on 4-well chamber slides were serumstarved overnight and then treated with 10 ng/mL TGF-b2for 2 hours in serum-free medium. After treatment, cellswere washed, fixed in 4% paraformaldehyde, blocked in5% goat serum, and incubated with primary antibody foranti-rabbit SMAD4 or anti-rabbit SMAD5 (1:100 dilution;
ajp.amjpathol.org - The American Journal of Pathology
Figure 1 aB-crystallin (aB) knockout (KO;aB�/�) mice have less subretinal fibrosisdevelopment after laser injury. Wild-type (WT)and aB�/� mice had laser photocoagulation(four lesions per eye) on day 0. A: Represen-tative images of choroidal neovascularization(CNV; stained by isolectin B4) and fibrosis(stained by collagen type I) at days 7, 21, and35 after laser in WT and aB�/� mice. B: Themean volume of the CNV and fibrosis at days7, 21, and 35 after laser. C: Histological sec-tion through a CNV membrane at day 35 afterlaser in WT and aB�/� mice. Fibrous tissuecan be seen (dotted circles) in the subretinalspace. **P < 0.01. n Z 8 per group (B).Scale bars: 100 mm (A); 50 mm (C). INL, innernuclear layer; ONL, outer nuclear layer; RPE,retinal pigment epithelium.
aB-Crystallin Regulates EMT
Cell Signaling Technology, Danvers, MA) overnight at4�C. Cells were washed in PBS, incubated with fluores-cein isothiocyanateeconjugated anti-rabbit secondaryantibody, mounted with DAPI-containing medium (VectorLaboratories), and viewed with a spinning disk confocalmicroscope.
Isolation of Ubiquitin Conjugates
Monotetraubiquitin- and polyubiquitin-conjugated proteinwas isolated from cultured RPE using the Pierce UbiquitinEnrichment kit (Thermo Scientific). Briefly, 150 mL ofcell lysates (1 mg/mL) was applied to 150 mL of sampledilution buffer and 20 mL of ubiquitin affinity resin. Themixtures were incubated at 4�C overnight and thencentrifuged using a spin column. A resin in this spincolumn binds polyubiquitin containing four or moreubiquitin subunits. Thus, the flow-through is harvestedas monotetraubiquitin-conjugated protein, whereas thepolyubiquitinated proteins were recovered in the elutionbuffer. Laemmli sample buffer (Bio-Rad LaboratoriesInc., Irvine, CA) was added to the column and heated at95�C for 5 minutes.
The American Journal of Pathology - ajp.amjpathol.org
Cell Proliferation Assay
Proliferation in RPE cells was measured using 5-bromo-20-deoxyuridine ELISA (Roche Applied Science), according tothe manufacturer’s instructions, with a 2-hour 5-bromo-20-deoxyuridine incubation.
Cell Migration Assay
A cell migration assay was performed using an Oris 96-wellcell migration assay kit (Platypus Technologies, Madison,WI), according to the manufacturer’s instructions, asdescribed previously.27 Briefly, 5 � 104 cells were seeded ineach well and transfected with or without siRNAs. After 1hour of pretreatment with Dulbecco’s modified Eagle’smedium with 3% fetal bovine serum containing recombi-nant TGF-b2 with 5 mmol/L aphidicolin (Sigma-Aldrich) toinhibit cell division, the stoppers were removed to allowcells to migrate into the detection zone. The cells were incu-bated for 48 hours after initiating migration and stained withPBS containing calcein AM (Life Technologies) for 1 hour.The area of cell migration was determined using Photoshopsoftware version CS3 (Adobe Systems, San Jose, CA).
863
Figure 2 aB-crystallin (aB) is up-regulated and has an impact on epithelial-mesenchymal transition markers in retinalpigment epithelium (RPE) cells in sub-retinal fibrous tissue. A: Levels of aB andbone morphogenetic protein (BMP)-4 inprotein extracted from the RPE-choroidalcomplexes after 20 laser lesions. B: Histo-logical section of the choroidal neo-vascularization (CNV) membrane at day 21after laser. Immunoreactivity to aB can beseen in the subretinal space (arrows).C andD: Triple immunofluorescence staining fora-smooth muscle actin (a-SMA; mesen-chymal marker), E-cadherin (epithelialmarker), and pan-cytokeratin (a marker forRPE cells) in eyes without laser treatment(C) and eyes with subretinal fibrosis (D)of wild-type (WT) and aB knockout (KO;aB�/�) mice. Nuclei are counterstainedwith DAPI (blue). E: Bar graph showsaverage number of cells immunoreactivefor both a-SMA and pan-cytokeratin andaverage number of the cells solelyimmunoreactive for pan-cytokeratin inthe subretinal lesion. The proportionsof cells immunoreactive for a-SMA inthe cells immunoreactive for pan-cytokeratin are calculated in the eyesof WT andaB�/�mice at day 35 after laser.F: Bar graph shows average number of thecells immunoreactive for both E-cadherinand pan-cytokeratin and average numberof cells solely immunoreactive for pan-cytokeratin is in the subretinal lesion.The proportions of cells immunoreactivefor E-cadherin in the cells immunoreac-tive for pan-cytokeratin are calculatedin the eyes of WT and aB�/� mice atday 35 after laser. *P < 0.05 versuscontrol (Ctrl) without laser treatment.n Z 3 per group (A); n Z 4 per group(E and F). Scale bars: 50 mm (B); 100 mm(C and D).
Ishikawa et al
864 ajp.amjpathol.org - The American Journal of Pathology
Figure 3 aB-crystallin (aB) induces epithelial-mesenchymal transition through regulation of SNAIL and SLUG expression. After transfection withcontrol (Ctrl) siRNA, aB-crystallin siRNA, empty (Ctrl) vector, and aB-encoding (aBþ) vector, retinal pigment epithelium (RPE) cells were stimulated with10 ng/mL transforming growth factor (TGF)-b2 for 48 hours. A: Western blot analysis of E-cadherin, a-smooth muscle actin (a-SMA), SNAIL, SLUG, aB,and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in the cell lysates of RPE cells. Quantification shown in Supplemental Figure S1. B: mRNAexpression of E-cadherin, a-SMA, SNAIL, SLUG, and aB is shown as relative fold to control siRNA normalized to GAPDH. C: Triple immunofluorescencestaining for a-SMA and E-cadherin in RPE cells. Nuclei are stained blue. Data are presented as means � SEM (B). n Z 4 per group (B). *P < 0.05,**P < 0.01. Scale bar Z 10 mm (C).
aB-Crystallin Regulates EMT
Statistical Analysis
All results are expressed as means � SEM. The sta-tistical significance of differences between groups wasanalyzed using the Tukey’s (honest significant difference)test or the two-tailed t-test. In comparison with thecontrol group, Dunnett’s t-test was applied. Differenceswere considered significant at P < 0.05. Statistical analyseswere performed using JMP version 7.0.1 (SAS Institute,Cary, NC).
Results
Attenuation of Subretinal Fibrosis in aB-Crystallin�/�
Mice
To evaluate a time-dependent alteration of CNV and fibrosisdevelopment after laser treatment, we labeled RPE/choroid
The American Journal of Pathology - ajp.amjpathol.org
flat mounts with Abs against isolectin B4 and collagen type Ito represent CNV and fibrosis, respectively. CNV reaches amaximum on day 7; thereafter, it begins to regress, dis-appearing almost completely within 35 days after laser.Fibrosis continues to increase for up to 35 days after laser(Figure 1A). As shown previously,22 CNV volume at day 7after laser was reduced in aB-crystallin�/� mice comparedwith the WT mice. At days 21 and 35 after laser, no signif-icant difference was seen in CNV volume between WT andaB-crystallin�/� mice. Fibrosis volume was significantly(P < 0.01) reduced in aB-crystallin�/� mice compared withtheWT at days 7, 21, and 35 after laser (Figure 1, A and B). InWT mice, prominent subretinal fibrosis was seen at 35 daysafter laser with extensive deposition of collagen type I in thelesions; however, fibrosis in aB-crystallin�/� mice wasmarkedly reduced (Figure 1C). These findings demonstratedsignificant attenuation of subretinal fibrosis that occurs sub-sequent to CNV in aB-crystallin�/� mice.
865
Figure 4 Bone morphogenetic protein (BMP)-4 induces epithelial-mesenchymal transition through up-regulation of aB-crystallin (aB). A: Western blotanalysis of aB and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in retinal pigment epithelium (RPE) cells stimulated with the indicated concen-trations of BMP4 for 48 hours. Quantification shown in Supplemental Figure S2A. B: Expression of aB mRNA shown as relative fold to control (Ctrl) normalizedto GAPDH. C: After transfection with control siRNA and aB siRNA, RPE cells were stimulated by 50 ng/mL BMP4 for 48 hours. Western blot analysis of E-cadherin, a-smooth muscle actin (a-SMA), SNAIL, SLUG, aB, and GAPDH in the cell lysates of RPE cells. Quantification shown in Supplemental Figure S2B. D:mRNA expression of E-cadherin, a-SMA, SNAIL, SLUG, and aB shown as relative fold to control normalized to GAPDH. Data are presented as means � SEM (Band D). n Z 4 per group (B and D). *P < 0.05, **P < 0.01.
Ishikawa et al
Impact of aB-Crystallin on EMT in Subretinal Fibrosisof Mouse Model
BMP4, a member of the TGF-b superfamily, is expressed insubretinal fibrosis lesions in AMD patients28 and up-regulatesaB-crystallin in microvascular endothelial cells.29 To examinethe expression changes of aB-crystallin and BMP4 in thedevelopment of subretinal fibrosis in mice, we measuredaB-crystallin and BMP4 in the RPE-choroid complexes afterlaser injury of WT mice.30 The concentration of aB-crystallinand BMP4 gradually increased, peaking at day 21 after laser(Figure 2A). Prominent expression of aB-crystallin can be seenin the lesion at day 21 after laser, whereas expression ofaB-crystallin was weak in control eyes (Figure 2B).
To investigate the difference in the cellular expression ofEMT markers in the subretinal fibrosis lesion between WTand aB-crystallin�/� mice, we stained the sections with Absagainst a-SMA (mesenchymal marker), E-cadherin(epithelial marker), and pan-cytokeratin (a marker for RPEcells). As a control, we evaluated expression of EMT-relatedmarkers in the RPE layer in mice without laser treatment.No significant difference was seen between WT andaB-crystallin�/� mice in the expression of EMT-relatedmarkers in RPE. In both WT and aB-crystallin�/� mice,RPE were positive for pan-cytokeratin and E-cadherin, butwere negative for a-SMA (Figure 2C). In the subretinalfibrosis lesions, RPE cells in WT mice expressed high levelsof a-SMA and low E-cadherin in the lesion at day 35 afterlaser. In contrast, RPE cells in aB-crystallin�/� miceexpressed less a-SMA and more E-cadherin compared with
866
WT mice (Figure 2D). Average numbers of RPE cellsimmunoreactive for a-SMA were lower in aB-crystallin�/�
mice compared with WT mice. The proportion of the RPEcells immunoreactive for a-SMA in the total RPE populationwas significantly reduced in aB-crystallin�/� mice comparedwith WT mice (Figure 2E). Average numbers of RPE cellsimmunoreactive for E-cadherin were higher in aB-crystallin�/�
mice compared withWTmice. The proportion of the RPE cellsimmunoreactive for E-cadherin in the total RPE population wassignificantly increased in aB-crystallin�/� mice compared withWT mice (Figure 2F). Taken together, these findings indicatethat in the absence ofaB-crystallin, RPE inCNV lesions show aprominent inhibition of the EMT phenotype.
aB-Crystallin Induces EMT through Regulation ofSNAIL and SLUG Expression
To test whether modulation of aB-crystallin expression canchange EMT inRPE cells, we examined the expression changesof E-cadherin, a-SMA, SNAIL, and SLUG in RPE cells trans-fected with aB-crystallin siRNA and aB-crystallineencodingvector. SNAIL and SLUG are transcription factors that canrepress E-cadherin while stimulating a-SMA31; SNAIL isknown to be expressed in CNV membranes from patients withnAMD and to induce EMT in RPE cells in vitro.32,33 TGF-b2 isthe predominant TGF-b isoform in the posterior segment of theeye and a crucial EMT inducer for RPE cells.34 TGF-b2 treat-ment at 10 ng/mL for 48 hours significantly reduced E-cadherinand increased a-SMA, SNAIL, and SLUG but did not alter aB-crystallin at the protein or mRNA levels (Figure 3, A andB, and
ajp.amjpathol.org - The American Journal of Pathology
Figure 5 aB-crystallin (aB) silencing impairs nuclear translocation of SMAD4 mediated by its monotetraubiquitination. Retinal pigment epithelium (RPE) cellswere transfected with control (Ctrl) siRNA and aB siRNA. A: Western blot (WB) analysis of SMAD4, glyceraldehyde-3-phosphate dehydrogenase (GAPDH; loadingcontrol for cytosolic protein), histone H3 (loading control for nuclear protein), and aB in nuclear and cytoplasmic fractions extracted from RPE cells with or without10 ng/mL transforming growth factor (TGF)-b2 stimulation for 24 hours. B: Triple immunofluorescence staining for SMAD4 and aB in RPE cells, with or without 10ng/mL TGF-b2 stimulation for 24 hours. Nuclei are counterstained with DAPI (blue). C: Western blot analysis of ubiquitin and b-actin in monotetraubiquitinated andpolyubiquitinated protein isolated from RPE cells. D: Monotetraubiquitinated SMAD4 and polyubiquitinated SMAD4 detected by Western blot of ubiquitinatedproteins; densitometric quantitation of blots from three independent experiments is shown. *P < 0.05. Scale bar Z 10 mm (B).
aB-Crystallin Regulates EMT
Supplemental Figure S1). Suppression of aB-crystallin bysiRNA significantly increased E-cadherin and decreaseda-SMA, SNAIL, and SLUG, whereas overexpression ofaB-crystallin by DNA vector decreased E-cadherin andincreased a-SMA, SNAIL, and SLUG in protein and mRNAlevels (Figure 3, A and B, and Supplemental Figure S1). TheTGF-b2einduced decrease of E-cadherin and the increase of a-SMA, SNAIL, and SLUG could be inhibited by suppression ofaB-crystallin and enhanced by overexpression of aB-crystallin(Figure 3, A and B, and Supplemental Figure S1). These resultssuggest that suppression of aB-crystallin induced amesenchymal-epithelial transition (increased E-cadherin anddecreased a-SMA), whereas overexpression of aB-crystallininduced EMT (decreased E-cadherin and increased a-SMA).
The American Journal of Pathology - ajp.amjpathol.org
TGF-b2einduced EMT was inhibited by suppression ofaB-crystallin.
Immunofluorescence staining validated the effect ofaB-crystallin modulation on the expression changes ofE-cadherin and a-SMA, with or without TGF-b2 treat-ment (Figure 3C). These results suggest that aB-crystallinplays a significant role in EMT of RPE cells.
BMP4 Up-Regulates aB-Crystallin, Resulting in EMT
Because BMP4 was induced in the CNV lesion after lasertreatment in mice and up-regulates aB-crystallin in micro-vascular endothelial cells,29 we investigated whether BMP4up-regulates aB-crystallin in RPE cells. Expression levels of
867
Figure 6 aB-crystallin (aB) overexpressionincreases nuclear translocation of SMAD4 in retinalpigment epithelium (RPE) cells. RPE cells weretransfected with aB-encoding vector or emptyvector, followed by treatment with 10 ng/mLtransforming growth factor (TGF)-b2 for 2 hours. A:RPE cells transfected with green fluorescent protein(GFP)eencoding plasmid using X-tremeGENE HP DNAtransfection reagent showing transfection efficiency.Immunoblot analysis of aB in vector-only and aBtransfected RPE cells; densitometric analysis of theblots from three independent experiments is shown.B: Immunofluorescence staining of SMAD4 in RPEcells transfected with empty vector or with aB-encoded vector. aB overexpression enhances nuclearexpression of SMAD4. C: Immunoblot analysis of thenuclear fraction of SMAD4 after cell fractionation ofRPE cells transfected with aB or empty vector;densitometric analysis of the blots from three inde-pendent experiments is shown. **P < 0.01. Scalebar Z 25 mm (A and B). GAPDH, glyceraldehye-3-phosphate dehydrogenase.
Ishikawa et al
aB-crystallin protein and mRNA were strikingly up-regulated at 48 hours after BMP4 stimulation at 25, 50,and 100 ng/mL in a dose-dependent manner (Figure 4, Aand B, and Supplemental Figure S2A). BMP4 treatment (50ng/mL) significantly reduced E-cadherin and increaseda-SMA, SNAIL, and SLUG in protein and mRNA levels;under these experimental conditions, aB-crystallin proteinlevels increased 4.5-fold. Next, we explored the potentialrole of aB-crystallin in BMP4-inducible EMT. We inhibitedaB-crystallin expression by siRNA in RPE cells treatedwith BMP4. In aB-crystallin siRNA-transfected cells,BMP4-induced expression changes of E-cadherin, a-SMA,SNAIL, and SLUG were inhibited at the protein and mRNAlevels (Figure 4, C and D, and Supplemental Figure S2B).These findings indicate that BMP4 up-regulates aB-crystallin,resulting in the induction of EMT in RPE cells.
Suppression of aB-Crystallin Inhibits NuclearTranslocation of SMAD4 Mediated by ItsMonotetraubiquitination
aB-crystallin is a molecular chaperone that facilitatestranslocation of proteins by binding and stabilizing the target
868
protein.14 EMT can occur through nuclear translocation ofproteins, such as b-catenin, NF-kB, Notch, R-SMADs, andSMAD4, which regulate SNAIL gene expression.35 We hy-pothesized that the effect of aB-crystallin modulation on EMTcan be mediated by a change in nuclear translocation of theseproteins. Nuclear and cytosolic fractions from RPE cells wereused to investigate nuclear translocation of b-catenin, NF-kB,Notch, R-SMADs, and SMAD4. Suppression of aB-crystallinby siRNA did not change levels of b-catenin, NF-kB, Notch2,or SMAD2/3 in nuclear and cytosolic fractions (SupplementalFigure S3). In contrast, the SMAD4 level was decreased in thenuclear fraction and increased in the cytosolic fraction by aB-crystallin suppression. SMAD4 was increased in the nuclearfraction and decreased in the cytosolic fraction at 24 hoursafter treatment by TGF-b2 at 10 ng/mL. Suppression of aB-crystallin inhibited the TGF-b2einduced expression changeof SMAD4 in the cytosolic and nuclear fractions (Figure 5A).Next, we performed immunofluorescence staining to confirm
the Western blot findings. Treatment with aB-crystallin siRNAmarkedly suppressed aB-crystallin expression in the cytosol. InRPE cells treated with control siRNA, SMAD4 is present inboth cytosol and nucleus, and TGF-b2 treatment (10 ng/mL for24 hours) induced accumulation of SMAD4 in the nucleus.
ajp.amjpathol.org - The American Journal of Pathology
Figure 7 aB-crystallin (aB) induces nucleartranslocation of SMAD5. Retinal pigmentepithelium (RPE) cells were transfected with aB-encoding vector or empty vector, followed bytreatment with 10 ng/mL transforming growthfactor (TGF)-b2 for 2 hours. A: Immunofluores-cence staining showing increased nuclear trans-location of SMAD5 in aB-overexpressed RPE cellsversus vector-only control cells. TGF-b2 treat-ment enhances nuclear translocation of SMAD5in both vector-only and aB-overexpressed cells.B: Immunoblot analysis of SMAD5 in the nuclearextracts from the vector and aB-overexpressedcells with and without TGF-b2 treatment;densitometric analysis of the blots from threeindependent experiments is shown. *P < 0.05.Scale bar Z 25 mm (A).
aB-Crystallin Regulates EMT
SMAD4 expression was not seen in the nucleus after sup-pression of aB-crystallin by siRNA with or without TGF-b2treatment (Figure 5B). These findings clearly demonstrated thatsuppression of aB-crystallin inhibits SMAD4 nuclear trans-location in RPE cells.
Monoubiquitin is a regulator of the location and activityof diverse cellular proteins.36 Recently, a regulatory role ofaB-crystallin in SMAD4 monoubiquitination was re-ported.12 We first isolated monotetraubiquitinated or poly-ubiquitinated protein from RPE cells treated with control oraB-crystallin siRNA. Western blot analysis of ubiquitin inthe whole gels shows a similar pattern of mono-tetraubiquitinated proteins in RPE cells transfected withaB-crystallin siRNA compared with control (Figure 5C)and showed equal loading of b-actin (Figure 5C). The levelof monotetraubiquitinated SMAD4 was increased in theaB-crystallin siRNA-treated cells (Figure 5D). Analysis ofpolyubiquitinated proteins showed they are similarly rep-resented in aB-crystallin siRNA-treated samples and incontrols (Figure 5C) with equal level of polyubiquitinatedSMAD4 (Figure 5D). These results suggest that the inhibitoryeffect of aB-crystallin suppression on SMAD4 nuclear trans-location could be mediated by monoubiquitination.
Overexpression of aB-Crystallin Modulates NuclearTranslocation of SMAD4 and SMAD5
On the basis of our finding that knockdown of aB-crystallinprevented nuclear translocation of SMAD4, we hypothesizedthat an increased nuclear translocation and accumulationof SMAD4 occurs in aB-crystallineoverexpressed RPEcells. To test this hypothesis, we transiently overexpressed
The American Journal of Pathology - ajp.amjpathol.org
aB-crystallin in RPE cells. The efficacy of overexpressionwas verified using a green fluorescent proteineencodingplasmid and immunoblot analysis of total cell extracts probedfor aB-crystallin (Figure 6A). A significant >4.5-foldincrease in aB-crystallin protein expression was obtainedin aB-crystallineoverexpressed RPE cells versus emptyvector control cells (Figure 6A). Immunofluorescencestudies revealed a prominent increase in nuclear accumulationof SMAD4 in aB-crystallineoverexpressed cells (Figure 6B).Immunoblot analysis for SMAD4 in nuclear fractions of thecells showed a significant (P < 0.01) increase in the nuclearlevels of SMAD4 (Figure 6C) in aB-crystallineoverexpressedcells when compared with empty vector control cells. TGF-b2treatment also significantly up-regulated SMAD4 accumulationin the nuclear fraction of empty vector control cells; however,no further accumulation of SMAD4 was observed inaB-crystallineoverexpressed cells (Figure 6C).
Because SMAD4 is the common mediator or co-SMADthat displays continuous shuttling between the nucleus andthe cytoplasm,37,38 we determined whether its accumulationin the nucleus after aB-crystallin overexpression wasassociated with nuclear accumulation of receptor-regulatedR-SMADs. We first evaluated SMAD2/3 because theseare activated in the canonical TGF-b/SMAD signalingpathway. To determine whether aB-crystallin over-expression can alter SMAD2/3 translocation and accumu-lation, we extracted nuclear and cytosolic fractions fromRPE cells transfected with aB-crystallineencoding vector.Overexpression of aB-crystallin did not alter the accumu-lation of SMAD2/3 in the nuclei (Supplemental Figure S3).We next evaluated R-SMAD5 because recent studiesshowed that SMAD5 activation may occur as a noncanonical
869
Figure 8 Inhibition of aB-crystallin (aB) expression prevents cell proliferation, migration, and fibronectin synthesis. Retinal pigment epithelium(RPE) cells were transfected with control (Ctrl) siRNA and aB siRNA. A: 5-Bromo-20-deoxyuridine (BrdU) incorporations were measured to assay pro-liferation. B: Immunofluorescence staining of fibronectin in RPE cells with or without 10 ng/mL transforming growth factor (TGF)-b2 stimulation for 12hours. C: Western blot analysis of procollagen I, aB, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in the cell lysates of RPE cells. D: For themigration assay, RPE cells were incubated after transfection for 48 hours with or without 10 ng/mL TGF-b2. The areas of the cells (stained by calcein AM)that migrated into the detection zone (white dotted circle) were measured. Data are presented as means � SEM (A and D). n Z 4 per group (A and D).**P < 0.01. Scale bars: 10 mm (B); 1 mm (D).
Ishikawa et al
TGF-b signaling pathway in macrophages.39 Interestingly,confocal immunofluorescence studies and immunoblotstudies of nuclear fractions revealed increased SMAD5accumulation in the nuclei of RPE after treatment withTGF-b2 (Figure 7). Just as with SMAD4, analysis of aB-crystallineoverexpressing cells showed increased SMAD5accumulation in the nuclei when compared with vector-onlycontrol cells (Figure 7).
Suppression of aB-Crystallin Inhibits Cell Proliferation,Migration, and Fibronectin Synthesis
EMT of RPE cells is an initial step in subretinal fibrosis thatis followed by cell proliferation, migration, and ECMremodeling.8,40 Fibronectin and collagen I are the mostprominent ECM components in human subretinal fibroticlesions.41 Fibronectin provides a provisional matrix for cellmigration of RPE cells.40,42 Procollagen I synthesis leads tocollagen type I deposition during tissue fibrosis.43 BecauseaB-crystallin induced EMT, we next tested the effect ofaB-crystallin inhibition on cell proliferation, migration, andfibronectin and procollagen I synthesis in RPE cells. Inhi-bition of aB-crystallin expression by siRNA significantlyinhibited cell proliferation, as determined by decreased5-bromo-20-deoxyuridine incorporation at 24 hours after
870
transfection (Figure 8A). Immunofluorescence stainingshowed prominently increased fibronectin expression at 12hours after treatment with TGF-b2. RPE cells treated byaB-crystallin siRNA showed less expression of fibronectinstimulated with or without TGF-b2 (Figure 8B). Procolla-gen I expression and TGF-b2einduced expression of pro-collagen I were decreased after silencing aB-crystallin inRPE cells (Figure 8C). In the cell migration assay, TGF-b2stimulation significantly promoted cell migration. Inhibi-tion of aB-crystallin expression by siRNA significantlysuppressed the cell migration stimulated with or withoutTGF-b2 (Figure 8D).
Discussion
This study demonstrates, for the first time, the functional role ofaB-crystallin in EMT of RPE and its association with subretinalfibrosis. We further show that inhibition of aB-crystallin down-regulates nuclear translocation of SMAD4 in RPE cells un-dergoing EMT through an increase in monotetraubiquitinationthat can impair nuclear localization of SMAD4. We alsodemonstrate that overexpression of aB-crystallin results inincreased nuclear accumulation of co-SMAD4 and R-SMAD5.Our studies thus identify aB-crystallin as a molecular chap-erone for SMAD4, and SMAD5 as a component of a
ajp.amjpathol.org - The American Journal of Pathology
aB-Crystallin Regulates EMT
noncanonical TGF-b2 signaling pathway in RPE that couldbind with SMAD4, allowing their nuclear accumulation.
In the natural history of nAMD, CNV progresses to anend-stage fibrous plaque/disciform scar.44,45 Similarly, inWT mice, after laser-induced CNV reaches a maximum atday 7 after laser and starts to regress, fibrous tissue in-creases for up to 35 days. By contrast, in aB-crystallinknockout mice, fibrous tissue does not increase after day21 after laser, whereas CNV regresses. In addition, theaB-crystallin expression level reaches a peak at day 21,which can imply the facilitating role of aB-crystallin insubretinal fibrosis development. Decreased mesenchymalRPE cells in the subretinal fibrous tissue in aB-crystallinknockout mice is supported by the in vitro data showingthat inhibition of aB-crystallin expression repressed EMTof RPE cells. Because EMT is an essential step of fibroticprocesses, such as cell proliferation, migration, andfibronectin production,40 the effect of aB-crystallin inhi-bition on the process shown in vitro can cause less fibroustissue formation in vivo.
Two recent studies demonstrated a role for aB-crystallinin the EMT process in the pathogenesis of hepatocellularcarcinoma and pulmonary fibrosis.11,12 Huang et al11
showed that aB-crystallin protects 14-3-3z from itsdegradation by the proteasome leading to the activation ofextracellular signaleregulated kinase signaling pathway,which can induce EMT of hepatocellular carcinoma cells.Bellaye et al12,46 found the overexpression of aB-crystallinin human and rodent fibrotic lung tissue. The authorselucidated the causal role of aB-crystallin in EMT bydemonstrating that overexpression of aB-crystallin dis-rupted monoubiquitination of SMAD4 by interacting withtranscriptional intermediary factor 1g, E3eubiquitin ligaseand thus limiting its nuclear export.12,46 Our study in RPEshowed that knockdown of aB-crystallin increased themonotetraubiquitination of SMAD4 by threefold anddecreased its nuclear localization in both TGF-b2 treatedand untreated cells. Interestingly, in RPE, EMT is inducedby both TGF-b2, the major TGF-b isoform in the RPE, andBMP4. BMP4 may be mediating its effects through regulationof aB-crystallin expression. BMP4 induction of EMT wasassociated with a 4.5-fold increase in aB-crystallin, andknocking down aB-crystallin in the BMP4-treated cellsreversed the induction of EMT. Previous histological findingsdemonstrated the expression of aB-crystallin and BMP4 insubretinal disciform scarring of human nAMD,15,28 indicatingthat BMP4-inducible up-regulation of aB-crystallin may play arole in the pathogenesis of subretinal fibrosis.
Overexpression of aB-crystallin led to a prominent in-crease in SMAD4 translocation/accumulation. In canonicalTGF-b signaling, nuclear translocation of R-SMAD2/3facilitates binding to Smad binding elements via maskingthe SMAD4 nuclear export signal.37,38 However, we did notfind evidence that nuclear SMAD2/3 accumulation wasaltered after aB-crystallin knockdown or overexpression.Instead, we considered the possibility that SMAD4
The American Journal of Pathology - ajp.amjpathol.org
stabilization in the nucleus involved other R-SMADs.Herein, we identified an alternate TGF-b2 pathway in whichtreatment of RPE with TGF-b2 increased the nucleartranslocation/accumulation of SMAD5. Although SMAD5is an integral component of the BMP signaling pathway, thenoncanonical TGF-b1emediated activation of SMAD5 hasbeen demonstrated in epithelial cells47 and recently in B-celllymphomas.48 Similarly, TGF-b1, and not BMP, proteinsactivated SMAD1/5 signaling in macrophages.39 Recently,TGF-b1 regulation of posterior lateral line formation inzebrafish was shown to be mediated by SMAD5.49 Skewingof canonical Wnt signaling by TGF-b toward the ALK1/SMAD1/5/8 pathway was reported to cause chondrocytehypertrophy.50 Our new data in RPE cells reveal that aB-crystallin overexpression increases nuclear SMAD5 accu-mulation. In support, overexpression of aA-crystallin,another member of the a-crystallin family, was associatedwith increased SMAD3/5 expression in a pancreatic cell line.51
Thus, SMAD4 and SMAD4/SMAD5 complex play animportant role in aB-crystallinemediated EMT in RPE cells.
In the physiological condition, RPE maintains the epithelialphenotype of highly polarized cells located between the neuralretina and the choroid.23 Physiological RPE cells are mitoti-cally quiescent, with cell-cell contact inhibition mediated bythe homotypic adhesion of cadherins on adjacent cells.52 In thelate phase of AMD, RPE dissociation can occur because ofRPE and retinal detachment subsequent to CNV formation.1
Once these contacts are disrupted and exposed by profibroticgrowth factors (ie, TGF-b and PDGF), RPE cells undergoEMT. This enables RPE to assume a mesenchymal cellphenotype, which includes enhanced migratory capacity andelevated resistance to apoptosis, as well as increased produc-tion of ECM components and proangiogenic factors, includingVEGF.8,53,54 Therefore, the novel property of aB-crystallin asan EMT inducer of RPE might account for the known func-tional roles of aB-crystallin in AMD, such as cytoprotectionand VEGF production in response to stress stimuli.21,22
The physiological RPE monolayer with cell-cell contactshowed no difference in the expression of EMT markersbetween WT and aB-crystallin knockout mice. RPE cellswith cell-cell contact are known to preferentially expressE-cadherin.55 The high expression of E-cadherin can preventcells from undergoing EMT by limiting SNAIL gene tran-scription. Thus, when the epithelial phenotype is maintainedwith adherent junctions, the cells will not be affected by theaction of stimuli triggering EMT.31 These findings mightelucidate the mechanism of the absence of aB-crystallinsilencing effect on EMT markers in the physiological RPEcells. Our results indicate that aB-crystallin inhibition canaffect only RPE cells that have lost cadherin-mediated adhe-sions, but it does not influence the cellular phenotypes of thenormal RPE monolayer.
Although anti-VEGF therapy for nAMD is effective as anearly treatment intervention to prevent consequential fibroticprogression, its direct suppressive effect on fibrosis is yet tobe proved.2 Notably, some studies reported that anti-VEGF
871
Ishikawa et al
therapy is associated with development of fibrosis in nAMDand proliferative diabetic retinopathy.56e58 The presentstudy demonstrated that local aB-crystallin inhibition couldsuppress development of subretinal fibrosis through EMTrepression. Recently, a small-molecule inhibitor that blocksthe interaction between aB-crystallin and VEGF-165 wasidentified59; this provides a proof of concept that perhaps aninhibitor of aB-crystallin and SMAD4 binding could beidentified. Therefore, aB-crystallin would be an attractivetherapeutic target for the treatment of nAMD with anadvantage in controlling both CNV and subretinal fibrosis.
Acknowledgment
We thank Ernesto Barron for technical assistance.
Supplemental Data
Supplemental material for this article can be found athttp://dx.doi.org/10.1016/j.ajpath.2015.11.014.
References
1. Ambati J, Fowler BJ: Mechanisms of age-related macular degenera-tion. Neuron 2012, 75:26e39
2. Daniel E, Toth CA, Grunwald JE, Jaffe GJ, Martin DF, Fine SL,Huang J, Ying GS, Hagstrom SA, Winter K, Maguire MG: Risk ofscar in the comparison of age-related macular degeneration treatmentstrials. Ophthalmology 2014, 121:656e666
3. Kudelka M, Grossniklaus H, Mandell K: Emergence of dualVEGF and PDGF antagonists in the treatment of exudative age-related macular degeneration. Expert Rev Ophthalmol 2013, 8.475e484
4. Ambati J, Atkinson JP, Gelfand BD: Immunology of age-relatedmacular degeneration. Nat Rev Immunol 2013, 13:438e451
5. Wree A, McGeough MD, Pena CA, Schlattjan M, Li H,Inzaugarat ME, Messer K, Canbay A, Hoffman HM, Feldstein AE:NLRP3 inflammasome activation is required for fibrosis developmentin NAFLD. J Mol Med (Berl) 2014, 92:1069e1082
6. Lavalette S, Raoul W, Houssier M, Camelo S, Levy O, Calippe B,Jonet L, Behar-Cohen F, Shemtob S, Guillonneau X, Combadiere C,Sennlaub F: Interleukin-1b inhibition prevents choroidal neo-vascularization and does not exacerbate photoreceptor degeneration.Am J Pathol 2011, 178:2416e2423
7. Espinosa-Heidmann DG, Reinoso MA, Pina Y, Csaky KG,Caicedo A, Cousins SW: Quantitative enumeration of vascularsmooth muscle cells and endothelial cells derived from bone marrowprecursors in experimental choroidal neovascularization. Exp Eye Res2005, 80:369e378
8. Ishikawa K, Kannan R, Hinton DR: Molecular mechanisms of sub-retinal fibrosis in age-related macular degeneration. Exp Eye Res2016, 142:19e25
9. Basha E, O’Neill H, Vierling E: Small heat shock proteins and alpha-crystallins: dynamic proteins with flexible functions. Trends BiochemSci 2012, 37:106e117
10. Wettstein G, Bellaye PS, Micheau O, Bonniaud P: Small heat shockproteins and the cytoskeleton: an essential interplay for cell integrity?Int J Biochem Cell Biol 2012, 44:1680e1686
11. Huang XY, Ke AW, Shi GM, Zhang X, Zhang C, Shi YH, Wang XY,Ding ZB, Xiao YS, Yan J, Qiu SJ, Fan J, Zhou J: alphaB-crystallincomplexes with 14-3-3zeta to induce epithelial-mesenchymal
872
transition and resistance to sorafenib in hepatocellular carcinoma.Hepatology 2013, 57:2235e2247
12. Bellaye PS, Wettstein G, Burgy O, Besnard V, Joannes A, Colas J,Causse S, Marchal-Somme J, Fabre A, Crestani B, Kolb M,Gauldie J, Camus P, Garrido C, Bonniaud P: The small heat-shockprotein alphaB-crystallin is essential for the nuclear localization ofSmad4: impact on pulmonary fibrosis. J Pathol 2014, 232:458e472
13. Johnson PT, Brown MN, Pulliam BC, Anderson DH, Johnson LV:Synaptic pathology, altered gene expression, and degeneration inphotoreceptors impacted by drusen. Invest Ophthalmol Vis Sci 2005,46:4788e4795
14. Kannan R, Sreekumar PG, Hinton DR: Novel roles for alpha-crystallins in retinal function and disease. Prog Retin Eye Res2012, 31:576e604
15. De S, Rabin DM, Salero E, Lederman PL, Temple S, Stern JH:Human retinal pigment epithelium cell changes and expression ofalphaB-crystallin: a biomarker for retinal pigment epithelium cellchange in age-related macular degeneration. Arch Ophthalmol 2007,125:641e645
16. Nakata K, Crabb JW, Hollyfield JG: Crystallin distribution in Bruch’smembrane-choroid complex from AMD and age-matched donor eyes.Exp Eye Res 2005, 80:821e826
17. Zhou P, Kannan R, Spee C, Sreekumar PG, Dou G, Hinton DR:Protection of retina by alphaB crystallin in sodium iodate inducedretinal degeneration. PLoS One 2014, 9:e98275
18. Dou G, Sreekumar PG, Spee C, He S, Ryan SJ, Kannan R,Hinton DR: Deficiency of alphaB crystallin augments ER stress-induced apoptosis by enhancing mitochondrial dysfunction. FreeRadic Biol Med 2012, 53:1111e1122
19. Yaung J, Kannan R, Wawrousek EF, Spee C, Sreekumar PG,Hinton DR: Exacerbation of retinal degeneration in the absence ofalpha crystallins in an in vivo model of chemically induced hypoxia.Exp Eye Res 2008, 86:355e365
20. Yaung J, Jin M, Barron E, Spee C, Wawrousek EF, Kannan R,Hinton DR: alpha-Crystallin distribution in retinal pigment epithe-lium and effect of gene knockouts on sensitivity to oxidative stress.Mol Vis 2007, 13:566e577
21. Sreekumar PG, Spee C, Ryan SJ, Cole SP, Kannan R, Hinton DR:Mechanism of RPE cell death in alpha-crystallin deficient mice: anovel and critical role for MRP1-mediated GSH efflux. PLoS One2012, 7:e33420
22. Kase S, He S, Sonoda S, Kitamura M, Spee C, Wawrousek E,Ryan SJ, Kannan R, Hinton DR: alphaB-crystallin regulation ofangiogenesis by modulation of VEGF. Blood 2010, 115:3398e3406
23. Sonoda S, Spee C, Barron E, Ryan SJ, Kannan R, Hinton DR: Aprotocol for the culture and differentiation of highly polarizedhuman retinal pigment epithelial cells. Nat Protoc 2009, 4:662e673
24. Brady JP, Garland DL, Green DE, Tamm ER, Giblin FJ,Wawrousek EF: AlphaB-crystallin in lens development and muscleintegrity: a gene knockout approach. Invest Ophthalmol Vis Sci 2001,42:2924e2934
25. Zhu Y, Lu Q, Shen J, Zhang L, Gao Y, Shen X, Xie B: Improvementand optimization of standards for a preclinical animal test model oflaser induced choroidal neovascularization. PLoS One 2014, 9:e94743
26. Sreekumar PG, Zhou J, Sohn J, Spee C, Ryan SJ, Maurer BJ,Kannan R, Hinton DR: N-(4-hydroxyphenyl) retinamide augmentslaser-induced choroidal neovascularization in mice. Invest Oph-thalmol Vis Sci 2008, 49:1210e1220
27. Ishikawa K, Yoshida S, Nakao S, Nakama T, Kita T, Asato R,Sassa Y, Arita R, Miyazaki M, Enaida H, Oshima Y, Murakami N,Niiro H, Ono J, Matsuda A, Goto Y, Akashi K, Izuhara K, Kudo A,Kono T, Hafezi-Moghadam A, Ishibashi T: Periostin promotes thegeneration of fibrous membranes in proliferative vitreoretinopathy.FASEB J 2014, 28:131e142
ajp.amjpathol.org - The American Journal of Pathology
aB-Crystallin Regulates EMT
28. Zhu D, Deng X, Xu J, Hinton DR: What determines the switch be-tween atrophic and neovascular forms of age related maculardegeneration? the role of BMP4 induced senescence. Aging (AlbanyNY) 2009, 1:740e745
29. Ciumas M, Eyries M, Poirier O, Maugenre S, Dierick F,Gambaryan N, Montagne K, Nadaud S, Soubrier F: Bone morpho-genetic proteins protect pulmonary microvascular endothelial cellsfrom apoptosis by upregulating alpha-B-crystallin. ArteriosclerThromb Vasc Biol 2013, 33:2577e2584
30. Xu J, Zhu D, He S, Spee C, Ryan SJ, Hinton DR: Transcriptionalregulation of bone morphogenetic protein 4 by tumor necrosis factorand its relationship with age-related macular degeneration. FASEB J2011, 25:2221e2233
31. de Herreros AG, Peiro S, Nassour M, Savagner P: Snail familyregulation and epithelial mesenchymal transitions in breast cancerprogression. J Mammary Gland Biol Neoplasia 2010, 15:135e147
32. Li H, Li M, Xu D, Zhao C, Liu G, Wang F: Overexpression of Snailin retinal pigment epithelial triggered epithelial-mesenchymal transi-tion. Biochem Biophys Res Commun 2014, 446:347e351
33. Hirasawa M, Noda K, Noda S, Suzuki M, Ozawa Y, Shinoda K,Inoue M, Ogawa Y, Tsubota K, Ishida S: Transcriptional factorsassociated with epithelial-mesenchymal transition in choroidal neo-vascularization. Mol Vis 2011, 17:1222e1230
34. Pfeffer BA, Flanders KC, Guerin CJ, Danielpour D, Anderson DH:Transforming growth factor beta 2 is the predominant isoform in theneural retina, retinal pigment epithelium-choroid and vitreous of themonkey eye. Exp Eye Res 1994, 59:323e333
35. Azmi AS: Unveiling the role of nuclear transport in epithelial-to-mesenchymal transition. Curr Cancer Drug Targets 2013, 13:906e914
36. Hicke L: Protein regulation by monoubiquitin. Nat Rev Mol Cell Biol2001, 2:195e201
37. Pierreux CE, Nicolás FJ, Hill CS: Transforming growth factor beta-independent shuttling of Smad4 between the cytoplasm and nu-cleus. Mol Cell Biol 2000, 20:9041e9054
38. Xiao Z, Latek R, Lodish HF: An extended bipartite nuclear locali-zation signal in Smad4 is required for its nuclear import and tran-scriptional activity. Oncogene 2003, 22:1057e1069
39. Nurgazieva D, Mickley A, Moganti K, Ming W, Ovsyi I, Popova A,Sachindra, Awad K, Wang N, Bieback K, Goerdt S,Kzhyshkowska J, Gratchev A: TGF-b1, but not bone morphogeneticproteins, activates Smad1/5 pathway in primary human macro-phages and induces expression of proatherogenic genes. J Immunol2015, 194:709e718
40. Kent D, Sheridan C: Choroidal neovascularization: a wound healingperspective. Mol Vis 2003, 9:747e755
41. Das A, Puklin JE, Frank RN, Zhang NL: Ultrastructural immuno-cytochemistry of subretinal neovascular membranes in age-relatedmacular degeneration. Ophthalmology 1992, 99:1368e1376
42. Jin M, He S, Worpel V, Ryan SJ, Hinton DR: Promotion of adhesionand migration of RPE cells to provisional extracellular matrices byTNF-alpha. Invest Ophthalmol Vis Sci 2000, 41:4324e4332
43. Cutroneo KR: How is type I procollagen synthesis regulated at thegene level during tissue fibrosis. J Cell Biochem 2003, 90:1e5
44. Lim LS, Mitchell P, Seddon JM, Holz FG, Wong TY: Age-relatedmacular degeneration. Lancet 2012, 379:1728e1738
The American Journal of Pathology - ajp.amjpathol.org
45. Ryan SJ: The development of an experimental model of subretinalneovascularization in disciform macular degeneration. Trans AmOphthalmol Soc 1979, 77:707e745
46. Bellaye PS, Burgy O, Colas J, Fabre A, Marchal-Somme J,Crestani B, Kolb M, Camus P, Garrido C, Bonniaud P: Antifibroticrole of alphaB-crystallin inhibition in pleural and subpleural fibrosis.Am J Respir Cell Mol Biol 2015, 52:244e252
47. Daly AC, Randall RA, Hill CS: Transforming growth factor beta-induced Smad1/5 phosphorylation in epithelial cells is mediated bynovel receptor complexes and is essential for anchorage-independentgrowth. Mol Cell Biol 2008, 28:6889e6902
48. Jiang D, Aguiar RC: MicroRNA-155 controls RB phosphorylation innormal and malignant B lymphocytes via the noncanonical TGF-b1/SMAD5 signaling module. Blood 2014, 123:86e93
49. Xing C, Gong B, Xue Y, Han Y, Wang Y, Meng A, Jia S: TGFb1aregulates zebrafish posterior lateral line formation via Smad5 medi-ated pathway. J Mol Cell Biol 2015, 7:48e61
50. van den Bosch MH, Blom AB, van Lent PL, van Beuningen HM,BlaneyDavidson EN, van der Kraan PM, van den BergWB: CanonicalWnt signaling skews TGF-b signaling in chondrocytes towardssignaling via ALK1 and Smad 1/5/8. Cell Signal 2014, 26:951e958
51. Deng M, Chen PC, Xie S, Zhao J, Gong L, Liu J, Zhang L, Sun S,Liu J, Ma H, Batra SK, Li DW: The small heat shock proteinalphaA-crystallin is expressed in pancreas and acts as a negativeregulator of carcinogenesis. Biochim Biophys Acta 2010, 1802:621e631
52. Binder S, Stanzel BV, Krebs I, Glittenberg C: Transplantation of theRPE in AMD. Prog Retin Eye Res 2007, 26:516e554
53. Tiwari N, Gheldof A, Tatari M, Christofori G: EMT as the ultimate sur-vival mechanism of cancer cells. Semin Cancer Biol 2012, 22:194e207
54. Schlingemann RO: Role of growth factors and the wound healingresponse in age-related macular degeneration. Graefes Arch Clin ExpOphthalmol 2004, 242:91e101
55. Burke JM, Cao F, Irving PE, Skumatz CM: Expression of E-cadherinby human retinal pigment epithelium: delayed expression in vitro.Invest Ophthalmol Vis Sci 1999, 40:2963e2970
56. Barikian A, Mahfoud Z, Abdulaal M, Safar A, Bashshur ZF: In-duction with intravitreal bevacizumab every two weeks in the man-agement of neovascular age-related macular degeneration. Am JOphthalmol 2015, 159:131e137
57. Van Geest RJ, Lesnik-Oberstein SY, Tan HS, Mura M,Goldschmeding R, Van Noorden CJ, Klaassen I, Schlingemann RO:A shift in the balance of vascular endothelial growth factor andconnective tissue growth factor by bevacizumab causes the angiofi-brotic switch in proliferative diabetic retinopathy. Br J Ophthalmol2012, 96:587e590
58. Arevalo JF, Maia M, Flynn HW Jr, Saravia M, Avery RL, Wu L, EidFarah M, Pieramici DJ, Berrocal MH, Sanchez JG: Tractional retinaldetachment following intravitreal bevacizumab (Avastin) in patientswith severe proliferative diabetic retinopathy. Br J Ophthalmol 2008,92:213e216
59. Chen Z, Ruan Q, Han S, Xi L, Jiang W, Jiang H, Ostrov DA, Cai J:Discovery of structure-based small molecule inhibitor ofaB-crystallin against basal-like/triple-negative breast cancerdevelopment in vitro and in vivo. Breast Cancer Res Treat 2014,145:45e59
873