Defensins HNP1 and HBD2 Stimulation of Wound-Associated Responses in Human Conjunctival Fibroblasts

Jing Li,1 Michael Raghunath,2,3 Donald Tan,1,4 Ricky R. Lareu,2,4 ZhenCheng Chen,5 and Roger W. Beuerman1,4
PURPOSE. To study the responses of human conjunctival broblasts (HCFs) to stimulation by human neutrophil defensin 1 (HNP1) and defensin2 (HBD2). METHODS. Defensin-stimulated gene expression in primary cultures of HCFs was analyzed by real-time PCR after exposure to various concentrations of HNP1 or HBD2. Gene and protein expression for selected collagens, matrix metalloproteinases, and tissue inhibitors of metalloproteinases were determined by real-time PCR and ELISA analysis. Activation of p42/44 mitogen-activated protein (MAP) and Akt was analyzed by Western blot. RESULTS. HCFs did not express signicant levels of the genes for HNP1 or HBD13. However, HNP1 and HBD2 stimulated HCF proliferation, the activation of p42/44 MAP kinase, and Akt kinase in a dose-dependent manner. HNP1 and HBD2 were not found to be chemotoxic for HCFs. It was demonstrated with the use of U0126 and wortmannin that the activation of p42/44 MAP kinase and Akt was responsible for the increased HCF proliferation observed under HNP1 and HBD2 stimulation. HNP1 stimulated the expression of the genes for collagen I, III, VI, and VIII. In addition, it reduced the secretion of collagen I protein but increased its intracellular retention. HNP1 and HBD2 upregulated the transcription and translation of MMP1. Small increases were observed in MMP14 gene expression after HNP1 stimulation and MMP2 gene expression after HBD2 stimulation. CONCLUSIONS. The results of this study suggest that HNP1 and HBD2 have a potential role in the biosynthetic and tissue remodeling responses of conjunctival broblasts. (Invest Ophthalmol Vis Sci. 2006;47:38113819) DOI:10.1167/iovs.05-1360 he mucosal surface of the eye is exposed and, consequently, at risk for microbial infections. Innate immunity provides a rst line of defense against pathogens and a route to initiating host immunity. Defensin peptides, an important component of innate immunity, protect the underlying uid-covered cell membranes of the ocular surface by direct killing of a broad range of pathogens.1,2 Defensins are widely distributed across species. Based on differences in the tertiary structure, defensins are further categorized into three families: , , and .1 In humans, -defensins 1 4 (HNPs) are mainly secreted by neutrophils, whereas defensins 5 6 are the products of Paneth cells.1,3 -Defensins (HBDs) are mainly found in epithelial cells, including corneal and conjunctival epithelial cells.4 7 With the use of mass spectrometry, HNP13 were found in tears of patients before surgery, just at the level of detection. However, their concentrations increased to approximately 15 g/mL 2 to 3 days after ocular surface surgery and in the absence of any infectious events.8 Similarly, an experimental model of corneal wound in the rabbit eye also revealed the selective upregulation of defensins with a time course that paralleled the reepithelialization of the wound.9 Defensins have also been shown in the inamed corneal stroma, but the source of the protein was not well dened.6 Defensins as antimicrobial agents have been studied intensively for their killing ability against bacteria, enveloped viruses, and fungi.1,2,10,11 In recent years, additional roles of defensins on host eukaryotic cells have come under investigation. Defensins were found as regulators of adaptive immunity and chemoattractants for phagocytic and mast cells.12,13 Defensins stimulate epithelial cell proliferation at low concentrations, but cytotoxicity has been a concern at higher concentrations.14,15 Despite an early report15 indicating that defensins are mitogenic for 3T3 broblasts, little information is available about the effect of defensins on corneal or conjunctival broblasts. Several groups have suggested a role for defensins in wound healing based on defensin-stimulated cell proliferation and increased wound closure in vitro.16,17 In the present study, we examined the expression of HNP1 and HBD13 in primary cultured human conjunctival broblasts. We used HNP1 and HBD2 at concentrations that were close to those found in human tears after ocular surface surgery (15 g/mL) to study the effects of defensins on conjunctival stromal broblast proliferation and migration, and on the expression of matrix metalloproteinases (MMPs), tissue inhibitors of metalloproteinases (TIMPs), and collagen. Results of this study suggest that defensins may contribute to ocular surface wound healing responses by modulating conjunctival broblasts.

From the 1Singapore Eye Research Institute, the 2Division of Bioengineering, the 3Department of Biochemistry, the 4Department of Ophthalmology, Yong Loo Lin School of Medicine, and the 5Tissue Engineering Programme (NUSTEP), National University of Singapore, Singapore. Supported by National Medical Research Council of Singapore (NMRC-IBG) Grants CPG/007/2004 (RWB) and CPG/003/2004 (MR), by the NUS Tissue Engineering Programme (NUSTEP), and Faculty of Engineering, Ofce of Research FRC Grant R-397 000-017112. Submitted for publication October 17, 2005; revised April 21 and May 21, 2006; accepted July 14, 2006. Disclosure: J. Li, None; M. Raghunath, None; D. Tan, None; R.R. Lareu, None; Z. Chen, None; R.W. Beuerman, None The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked advertisement in accordance with 18 U.S.C. 1734 solely to indicate this fact. Corresponding author: Roger W. Beuerman, Singapore Eye Research Institute, 11 Third Hospital Avenue, Level 6, Singapore 168751; rbeuer@pacic.net.sg.
Investigative Ophthalmology & Visual Science, September 2006, Vol. 47, No. 9 Copyright Association for Research in Vision and Ophthalmology

MATERIALS

AND

METHODS

Isolation of Primary Human Conjunctival Fibroblasts and Epithelial Cells

Human cadaver conjunctival tissue was obtained from the Singapore Eye Bank. The protocol was approved by the institutional review board (IRB) of the Singapore Eye Research Institute, and it followed the tenets of the Declaration of Helsinki. Donor (n 4) ages ranged between 41 and 65, with a mean of 54 years of age and a mean time after death of 20 hours. Harvested conjunctival tissues were rinsed in

3811

3812

Li et al.

IOVS, September 2006, Vol. 47, No. 9

TABLE 1. Primer and Reporter Sequences of Defensin Genes Used by Real-Time PCR Analysis Gene HNP1 HDB1 HDB2 HDB3 Access Code NM_004084 NM_005218 AF071216 AF516673 Forward Primer 5 -3 TGCATCTACCAGGGAAGACTCT TTGGCCACAGATCTGATCATTACAA GATCCTGTTACCTGCCTTAAGAGT GTCATGGAGGAATCATAAACACATTACAG Reverse Primer 5 -3 TTCCCTGTAGCTCTCAAAGCAAATT GTAACAGGTGCCTTGAATTTTGGT CCACAGGTGCCAATTTGTTTATACC CCGATCTGTTCCTCCTTTGGA Reporter Sequence 5 -3 ATTCTGCTGCTGAGCTTG CAGAATAGAGACATTGCCC ACTGGATGACATATGGCTCC CACACCGGCCGCCTC

antibiotic solution three times for 5 minutes before digestion in Dispase II (1.2 U/mL in Hanks balanced salt solution) at 37C for 3 hours. Epithelial cell sheets were scraped from the digested tissue, collected by centrifugation at 900g for 4 minutes, further dissociated into single cells by trypsin, resuspended in keratinocyte growth medium (KGM; Cambrex, Walkersville, MD), and plated at a density of 5000 cells/ cm2.18 Tissue remaining after epithelial cell scraping was cut into small pieces, placed in a tissue culture dish, and submerged in Dulbecco modied Eagle medium (DMEM) with 10% fetal bovine serum (FBS) for the outgrowth of conjunctival broblasts. Explants were discarded after 7 days, and broblasts were trypsinized and replated in DMEM with 10% FBS. Cells from passages 1 to 3 were used in this study. No contamination by epithelial cells was observed in broblast cultures or broblasts in epithelial cell cultures.

presence of different concentrations of HNP1 or HBD2 and were incubated for 24 hours. Cell viability was analyzed (CellTiter-Blue; Promega, Singapore). The intensity of uorescence was measured at 560/590 nm at the end of the incubation. A cell/uorescence standard curve was generated each time to ensure the linearity of uorescence in the range used with the cells.

Cell Migration Analysis

Fibroblast migration was studied with a chemotaxis 24-well cell migration system (QCM; Chemicon International) with 8- m pores. After 4 hours of serum starvation in DMEM, 104 cells/well were seeded in serum-free DMEM in the presence or absence of defensins. At the end of the incubation, cells retained in the inner chamber were scraped off with a cotton swab. Cells that migrated to the outer chamber of the membrane were detached and lysed. The number of migrating cells was quantied by the incorporation of green uorescent dye and was measured at 480/520 nm. To ensure the linearity of the measurement, a cell/uorescence curve was developed in pilot experiments.

Defensin Gene Expression Analysis

Total RNA was extracted from cultured conjunctival epithelial cells and broblasts with the use of a reagent (TRIzol; Invitrogen, Singapore) and was digested with DNAse I to remove possible genomic DNA contamination. Reverse transcription (RT) was performed (RTII; Invitrogen). Real-time PCR primers (TaqMan Assay-by-Design; Applied Biosystems, Singapore) were used for the detection of HNP1 and HBD13. 18S ribosomal RNA was used as internal control to calculate Ct. Additional RT controls (RNA mixture with all the RT reaction reagents except reverse transcriptase) were used as control to further exclude the inuence of genomic DNA carryover. Sequences of the primers and probes are listed in Table 1.

Western Blot Analysis

Western blot analysis was carried out to analyze the activation of p42/44 MAP kinase, Akt, p38, and JUN kinases. Cultures of human conjunctival broblasts in six-well plates at 80% conuence were serum starved in DMEM for 4 hours before the addition of defensin. Defensin stimulation of p42/p44 MAP, Akt, p38, and Jun kinase activation was assayed at 15, 45, and 60 minutes after the addition of HNP1 or HBD2. Lysis buffer consisted of 50 mM Tris, 100 mM KCl, 1 mM EDTA, 1% NP40, 1 mM NaOV4, 100 mM NaF, and a proteinase inhibitor cocktail (Roche Diagnostics Asia Pacic, Singapore).19 An equal amount of protein was loaded (Coomassie Plus protein assay; Pierce, Rockford, IL). All antibodies were purchased (Cell Signaling Technology, Beverly, MA) and used according to the manufacturers instructions.

Defensin Stimulation
Synthetic human HNP1 (Sigma-Aldrich, Singapore), purity greater than 91%, and recombinant human HBD2 (Chemicon International, Temecula, CA), purity greater than 98%, were obtained. Both were dissolved in sterile PBS at a concentration of 100 g/mL before use and were further diluted in serum-free DMEM at the indicated concentration for each experiment.

Cell Proliferation Analysis

Cell proliferation was analyzed by means of bromodeoxyuridine (BrdU) incorporation with the cell proliferation kit (catalog no. RPN 250; Amersham Biosciences, Singapore). Passage 1 to passage 2 broblasts were seeded in 96-well plates at a density of 2 104cells/well and were allowed to settle overnight in DMEM plus 10% FBS. Afterward, cells were gently washed in PBS, serum starved in serum-free DMEM for 18 to 20 hours before the addition of defensins, and incubated for another 24 hours. For U0126- and wortmannin-treated cell proliferation, the two drugs were added 1 hour before the addition of defensin and were maintained throughout the stimulation. BrdU was added 2 to 3 hours before the end of defensin stimulation. Optical density was read at 450 nm (GeniosPro microplate reader; Tecan Asia, Singapore). Triple or quadruplicate wells were used for each condition in all experiments, and the OD readings were subtracted from the blank, which contained the same numbers of cells without BrdU.

MMP and TIMP Expression Analysis

Cultured conjunctival broblasts at P2 to P3 were plated in 24-well plates and incubated with 17.5 g/mL HNP1 or HBD2 in serum-free DMEM. At indicated times, cells were lysed and total RNA was extracted and reverse transcribed into cDNA for relative comparative real-time PCR analysis (TaqMan gene expression system; Applied Biosystems). 18S ribosomal RNA was used as an internal control for the calculation of Ct. Control cells without defensin stimulation were used as the calibrator for relative quantication. Primer and probe sequences for MMP and TIMP were proprietary (ABI Prism, Faireld, CT) and unavailable. Aliquots of supernatant were used for the analysis of MMP proteins. A human MMP array that included MMP1, MMP2, MMP3, MMP8, MMP9, MMP10, MMP13, TIMP1, and TIMP2 (Pierce SearchLight, Rockford, IL) was used to measure total protein levels of secreted MMPs and TIMPs. This array used multiplex sandwich ELISA with conjugated streptavidin-horseradish peroxidase to generate luminescent signals. The signal was detected by an 8-byte cooled charge-coupled device (CCD) camera, and the density of each spot was quantied (Multigenius Bioimagine System; Syngene, Cambridge, UK). Total MMP1 content was further determined by activity assay (Biotrak; Amersham). With the addition of p-aminophenylmercuric acetate (APMA), total

Cell Viability Analysis

Conjunctival broblasts were trypsinized and seeded at a density of 2 104 cells/well in serum-free DMEM in a black 96-well plate (Corning Life Sciences, Acton, MA) with a transparent bottom in the

IOVS, September 2006, Vol. 47, No. 9

Defensin Stimulation of Wound-Associated Responses

3813

TABLE 2. Real-time PCR Analysis of HNP1 and HBD13 Gene Expression in Cultured HCEs and HCFs HBD1 Ct HCE Ctrl LPS TNFHCF Ctrl LPS TNFFold Ct HBD2 Fold HBD3 Ct Fold HNP1 Ct

18.35 19.02 18.01 ND ND ND

0.84 1.73 1.69

1 0.77 1.37

22.05 20.35 18.91 ND ND ND

0.58 0.92* 2.13*

1 3.48* 8.81*

15.62 15.23 15.92 ND ND ND

3.65 1.04 3.02

1 1.31 0.81

ND ND ND ND ND ND

18S ribosomal RNA was used as internal control to calculate Ct. Experiments were repeated ve times under control conditions and three times under 500 ng/mL LPS and 50 ng/mL TNF- stimulation. ND, not detectable. * P 0.05.

MMP1 protein activity was measured, and the resultant color was read at 405 nm in a microplate reader (GeniosPro; Tecan).

Immunouorescence Staining and Analysis of Collagen I Protein

Monoclonal antibody against human collagen I was purchased (Abcam, Cambridge, UK). Fibroblasts grown in eight-well chamber slides were cultured in serum-free medium supplemented with 50 g/mL fresh L-ascorbic acid with or without (as control) 17.5 g/mL HNP1 for 24 hours before they were xed with methanol for 10 minutes at 20C. Slides were then blocked in 3% BSA and incubated with or without (conjugate control) primary antibody overnight at 4C. Cells were visualized with a secondary conjugated antibody (Alexa 488; Invitrogen). For quantitative analysis, the slides were imaged on a bioimaging station (TE600 microscope plus Xenon illuminator [Nikon, Tokyo, Japan] and CoolSNAP camera [Photometrics, Tucson, AZ]). Images were analyzed at 40 magnication with imaging system software (Metamorph; Molecular Devices, Eugene, OR). The x- and y-axes of the image were calibrated to 0.162481 m/pixel at 40 magnication, and images were taken at an exposure of 800 msec. Images of the samples were normalized for background uorescence by subtracting the uorescence intensities of conjugate control. To investigate the pattern and intensity distribution for the collagen I signal in intracellular granules, threshold uorescence intensity (greater than 180) was selected to dene the desired regions of more intense uorescence. Areas of these identied regions were measured with an analysis module (Metamorph Integrated Morphometry Analysis; Molecular Devices).

Collagen Gene Expression Analysis

Real-time PCR primers (TaqMan; Applied Biosystems) specic for the 1 chain of human collagen types I, III, VI, and VIII genes were purchased. Conjunctival broblasts cultured in 24-well plates at a density of 5000 cells/well were allowed to settle overnight in DMEM plus 10% FBS before treatment with different concentrations of HNP1 in serum-free medium. At the indicated times, cells were lysed in reagent (TRIzol; Invitrogen), and RNA was extracted for real-time PCR analysis. 18S ribosomal RNA was used as internal control for the calculation of Ct. Control cells without defensin stimulation were used as the calibrator for relative quantication.

Biochemical Analyses of Collagen I Production

Cell cultures were seeded at 5000 cells/well in 24-well plates and were allowed to settle overnight in DMEM plus 10% FBS before treatment for 24 hours with 17.5 g/mL HNP1 in the presence of freshly made L-ascorbic acid at 50 g/mL in serum-free medium. Medium was harvested, and cell layers were washed three times with Hanks balanced salt solution. If necessary, media and washed cell layers (without buffer on top) were stored at 20C before analysis. For collagen analyses, media and cell layers were subjected to peptic digestion, as described.20 Porcine gastric mucosa pepsin (2500 U/mg; Roche Diagnostics Asia Pacic, Singapore) was dissolved in 0.1 N HCl to yield a stock solution of 1 mg/mL (wt/vol). This solution was added directly to culture medium to a nal concentration of 100 g/mL for 2 hours at room temperature (0.1 vol analyte), followed by neutralization with an equal volume of 0.1 N NaOH. Aliquots of digested and neutralized samples were prepared for gel electrophoresis by the addition of 4 SLS sample buffer (Invitrogen) in appropriate amounts and were heated for 5 minutes at 100C. Protein samples were separated using pre-cast 3% to 8% Tris-acetate polyacrylamide gels (NuPage; Invitrogen), each 1-mm thick, and were run with the appropriate apparatus (XCell Surelock Mini-Cell; Invitrogen) with their respective loading and running buffers at 150 V for 70 to 90 minutes. Protein bands on gels were stained with silver staining (SilverQuest; Invitrogen) in accordance with the manufacturers protocol. Densitometric analysis of silver-stained protein bands was carried out on image analysis software (GS-800 Calibrated Densitometer with Quantity One, v4.5.2; Bio-Rad, Hercules, CA). Gels were scanned in their wet state, and collagen bands were quantitated by dening each band with the rectangular tool with background subtraction. Densitometric values for the 1(I) and 2(I) chains were added to the values obtained for the bands to obtain the total amount of collagen per lane.

Statistical Analysis
Student t test was used to determine the difference between means of groups. P 0.05 was accepted as signicantly different. Data are presented as mean SD.

RESULTS
HNP1 and HBD13 Expression in Human Conjunctival Fibroblasts
Real-time PCR at more than 40 cycles failed to detect messages for either HNP1 or HBD13 in HCFs. In contrast, the expression of HBD13 was consistently found in conjunctival epithelial cells used as a positive control, a result that veried previous ndings (Table 2).4 Given that the expression of HBD2 in ocular surface epithelial cells was previously shown to be upregulated by lipopolysaccharide (LPS) and inammatory cytokines such as TNF- ,4 we challenged HCF with LPS (500 ng/mL), and TNF- (50 ng/mL) and harvested mRNA 18 hours later.21 Upregulation of HBD2 in conjunctival epithelial cells after LPS and TNF- stimulation was conrmed. However, the

3814

Li et al.

IOVS, September 2006, Vol. 47, No. 9

expression of HBD1-3 genes by real-time PCR in broblasts was not detectable at 40 cycles (Table 2). Interestingly, we did not observe changes of HBD3 gene expression in conjunctival epithelial cells after TNF- stimulation though increased expression was reported in human keratinocytes.22

Defensin-Stimulated Fibroblast Proliferation

Cell proliferation is an important component of the woundhealing response. Defensins have been shown to stimulate the proliferation of human airway epithelial cells and mouse 3T3 broblasts.14,15 HCFs at passages 1 and 2 were stimulated with 0.35 g/mL, 3.5 g/mL, 17.5 g/mL, and 30 g/mL HNP1 or HBD2 in serum-free DMEM (Fig. 1A). A signicant increase in BrdU incorporation was observed with HNP1 above the concentration of 3.5 g/mL compared with controls. HBD2 also stimulated the proliferation of broblasts. However, it was less effective than HNP1 at the same concentrations. A signicant increase in BrdU incorporation was observed for HBD2 at a concentration of 17.5 g/mL compared with the control. For HNP1 and HBD2, no further signicant increase was observed at concentrations of 30 and 50 g/mL compared with that of 17.5 g/mL. As a positive control, broblasts were incubated in DMEM with 10% FBS, which induced an average 6.78 2.12-fold increase in BrdU incorporation compared with controls. When HNP1 or HBD2 at a concentration of 17.5 g/mL was added to the medium with 10% FBS, no further increase in BrdU incorporation was observed, indicating that there was no synergistic effect of HNP1 or HBD2 with FBS. These results suggested that HNP1 and HBD2 are potent mitogens for human conjunctival broblasts. It was reported for airway epithelial cells and renal carcinoma cells that the mitogenic effect of defensins was limited to the concentration range of 10 to 20 g/mL, whereas at higher concentrations of 25 to 50 g/mL, substantial cytotoxicity was observed in these cells.14,23 However, HCF cell death was not observed at the concentration of 20 g/mL, 50 g/mL, or 100 g/mL, suggesting cell-type specicity for the cytotoxicity of defensins (Figure 1B).

Defensins and Conjunctival Fibroblast Chemotaxis

HNP1 and HBD2 are chemoattractants for various leukocytes.12,13 To determine whether HNP1 and HBD2 are chemotactic for conjunctival broblasts, we examined defensin-stimulated broblast migration using 24-well inserts with an 8- m pore size. The number of cells migrating to the outside of the chamber was quantied with a uorescent dye that bound to cellular nucleic acids. Cell migration was analyzed 2 to 18 hours after stimulation with HNP1 or HBD2. Random cell movement was subtracted using serum-free DMEM as a control in inner and outer chambers. Signicantly increased broblast migration was observed at 4 hours with approximately 240 cells from the original 105 cells in the presence of HNP1 but not of HBD2. However, when the same concentration of HNP1 was added in the inner and outer chambers, we found that the number of cells in the outer chamber was not signicantly different from that of the rst condition, indicating that the migration resulted from chemokinesis.
FIGURE 1. Effect of HNP1 and HBD2 on HCF proliferation and viability. (A) HNP1- and HBD2-stimulated HCF proliferation, as calculated by BrdU incorporation. Cells were serum starved for 18 to 20 hours, and defensins were added in fresh serum-free medium for another 24 hours. All tests were run in triplicate or quadruplicate. n 6 for experiments with HNP1 at concentrations lower than 30 g/mL; n 3 for HNP1 at 30 and 50 g/mL; n 3 for all HBD2-stimulated experiments. Error bars represent SD. Signicance between stimulated and control groups was assessed with unpaired Student t test. *P 0.05. (B) Effect of HNP1 and HBD2 on HCF viability. HCFs were serum starved for 18 to 20 hours before they were plated onto 96-well plates at 10 103 cells/well per 100 L in serum-free DMEM with HNP1 or HBD2 at the concentrations indicated. The control was cells cultured in serum-free medium without defensins. Shown are the percentages of live cells at different concentrations of HNP1 or HBD2 compared with controls. Each experiment was repeated three times with triplicate wells.

Defensins Affect the Expression of MMPs and TIMPs

The modulation of MMP and TIMP gene expression and protein production is critical for extracellular matrix remodeling and cell migration during wound healing.24,25 Expression of various MMPs and TIMPs has been reported in conjunctival broblasts.24 Real-time PCR analysis showed the expression of MMP1 (collagenase 1), MMP2 (gelatinase A), MMP9 (gelatinase

B), MMP3 (stromelysin), MMP14 (MT1-MMP), TIMP1, TIMP2, TIMP3, and TIMP4 in HCFs. Among those, only MMP1 and TIMP2 gene expression was increased by both HNP1 and HBD2 at a concentration of 17.5 g/mL 24 hours after stimulation in serum-free medium (MMP1: 2.51 0.50-fold increase by HNP1, 2.25 0.30-fold increase by HBD2; TIMP2: 1.23 0.07-fold increase by HNP1, 1.22 0.05-fold increase by HBD2) (Fig. 2A). Small increases in MMP2 gene expression (1.58 0.08-fold) by the same concentration of HBD2 and in

IOVS, September 2006, Vol. 47, No. 9

Defensin Stimulation of Wound-Associated Responses

3815

FIGURE 2. Defensin-stimulated MMP and TIMP expression. (A) MMP1, MMP2, MMP3, MMP9, MMP14, and TIMP1 to TIMP4 gene expression 24 hours after 17.5 g/mL HNP1 and HBD2 stimulation. (B) MMP1 gene expression at 4, 6, 16, and 24 hours after 17.5 g/mL HNP1 and HBD2 stimulation. (C) Dose-response increase of MMP1 gene expression to HNP1 and HBD2 4 hours after stimulation. (D) Total MMP1 protein in the supernatants of cells 24 hours after stimulation with 17.5 g/mL HNP1 or HBD2. Each experiment was repeated at least three times with duplicate wells. Calculation of gene expression was based on TaqMan real-time PCR analysis using control group as the calibrator and 18S as the internal control. *P 0.05 when treated groups were compared with controls by unpaired Student t test.

MMP14 (1.49 0.14-fold) by HNP1 stimulation were also observed. Expression levels of MMP9, TIMP1, TIMP3, and TIMP4 were not changed by either HNP1 or HBD2. This was followed by the analysis of total protein levels for MMP1, MMP2, MMP3, MMP8, MMP9, MMP10, TIMP1, and TIMP2 in cell culture supernatants 24 hours after stimulation with 17.5 g/mL HNP1 and HBD2 in serum-free medium with a human MMP array. Only an increase in total MMP1 protein was observed in both conditions compared with serum-free medium alone. We further analyzed the defensin-stimulated increase in MMP1 expression. As shown in Figure 2B, an increase of MMP1 gene expression was seen 4 hours after stimulation with 17.5 g/mL HNP1 or HBD2 and persisted at 24 hours. The increase in MMP1 gene transcripts was also dose dependent, as shown in Figure 2C. A signicant increase in MMP1 gene expression was observed with 3.5 g/mL HNP1 or 17.5 g/mL HBD2 and greater concentrations. However, at concentrations higher than 30 g/mL, no further increase in MMP1 transcription was observed under HNP1 or HBD2 stimulation. A more accurate analysis of total MMP1 protein in the supernatant (Biotrak assay; Amersham) showed an increase of 2.86 0.73fold and 2.63 0.49-fold in total MMP1 activity in the supernatant 24 hours after stimulation by HNP1 or HBD2, respectively, both at a concentration of 17.5 g/mL (Fig. 2D).

Collagen Gene Expression and Protein Production

Conjunctival broblasts secrete collagen in response to tissue damage. Collagen I is the most abundant collagen in conjunctiva stroma, followed by collagen III, VI, and VIII.26 The 1 chain of collagen types I, III, VI, and VIII gene expression was studied by real-time PCR after HNP1 or HBD2 stimulation. HNP1 led to modest twofold to fourfold increases in gene expression of all four types of collagen examined at the concentrations of 17.5 g/mL and higher (Fig. 3). The increase in gene expression was observed at 6 hours and lasted to 16 hours but returned to the control level at 24 hours. No further signicant increase of collagen gene expression was observed with 30 g/mL HNP1 compared with 17.5 g/mL. Among those, the collagen VIII 1 gene was most sensitive to the stimulation of HNP1. However, HBD2 did not change gene expression of these collagen types (data not shown). On the protein level, conjunctival broblasts showed ascorbic acid dependent collagen I production. By using gradient SDS-PAGE gel and silver staining in the presence of human collagen standards, we found that HNP1 reduced the presence of collagen I in the culture medium by 41% and 80%, respectively, in two independent experiments (Fig. 4A). In contrast, HNP1 increased the measurable amount of

3816

Li et al.

IOVS, September 2006, Vol. 47, No. 9

FIGURE 4. Biochemical analysis and immunouorescent staining of collagen I in HCF exposed to HNP1. (A) Silver staining of collagen I protein in cell layers and culture medium. Both samples were subjected to peptic digest to destroy noncollagenous proteins, and remaining proteins were separated by denaturing SDS-PAGE. According to the collagen I standard (250 ng/lane), no evidence was observed for other collagen types. (B) Immunouorescent staining of collagen I in cells with and without HNP1 treatment. (a) Control broblasts showed intracellular ne granules and longitudinal structures, indicative of the perinuclear envelope and adjacent cisternae of the rough endoplasmic reticulum. (b, c) Cells exposed to HNP1 showed a different distribution pattern of collagen I, as evidenced by the presence of aggregated coarse granules with a patchy concentration and indicative of a potential association with the Golgi apparatus (arrows). No evidence was seen in treated or untreated cultures for extracellular matrix deposition of collagen.

FIGURE 3. Defensin-stimulated collagen gene expression. Cells were treated with different concentrations of HNP1 ( control, f 0.35 g/mL, 3.5 g/mL, 17.5 g/mL, E 30 g/mL) and were harvested at 4, 6, 16, and 24 hours after stimulation. Calculation of gene expression was based on TaqMan real-time PCR analysis using control group as the calibrator and 18S as the internal control. (A) Expression of collagen I type 1a. (B) Expression of collagen III type 1a. (C) Expression of collagen VI type 1a. (D) Expression of collagen VIII type 1a. Each experiment was repeated three times with duplicate wells. *P 0.05 when treated groups were compared with controls by unpaired Student t test.

collagen I in the cell layer in both experiments. In the rst experiment, collagen was not measurable in the untreated cell layer, but under HNP treatment it amounted to 56% of the collagen in the medium of untreated cells. In the second experiment, a minimal baseline value was obtained for collagen I that increased by 42% after treatment. Therefore, an obvious and complementary shift of collagen I from the culture medium into the cell layer was observed. Morphometric analysis of immunouorescently stained broblasts conrmed that HNP1-treated cells contained granules with twofold to threefold higher immunouorescent signals for collagen I than did controls (Fig. 4B). However, we did not detect collagen III, VI, or VIII protein by SDS-PAGE or silver staining in any of the samples.

IOVS, September 2006, Vol. 47, No. 9

Defensin Stimulation of Wound-Associated Responses

3817

Activation of p42/44 MAP Kinase and Akt Kinase

The activation of p42/p44 MAP kinase (ERK1/2), Akt, p38, and JUN kinase after HNP1 and HBD2 stimulation was examined in HCFs. HNP1 and HBD2 induced the dose-response activation of p42/44 MAP kinase and Akt, which could be detected 15 minutes after stimulation (Fig. 5). Repetitive experiments showed that HCF was more responsive to HNP1 than to HBD2 at the same concentrations in the phosphorylation of p42/44 MAP kinase and Akt. Both defensins led to the phosphorylation of p42/44 MAP kinase at a concentration as low as 0.35 g/mL. However, consistent activation of Akt was only observed with 3.5 g/mL HNP1 and higher. For HBD2, consistent activation of Akt was observed at a concentration of 30 g/mL. No activation of p38 or JUN kinase was observed in any of the experiments (data not shown).

MAP Kinase and Akt Are Involved in the Mitogenic Effect of HNP1 on HCFs
To further dene the relationship between defensin activation of p42/44 MAP kinase and the increase in cell proliferation, we looked at defensin-stimulated BrdU incorporation in the presence of U0126, a specic inhibitor of p42/44 MAP kinase. U0126 at a concentration of 15 M was added 1 hour

FIGURE 6. Effects of ERK and Akt inhibitors on BrdU incorporation. (A) BrdU incorporation of 15 M U0126 or 5 M wortmannin. Incubation was carried out as described for Figure 1. Both drugs were added 1 hour before the addition of 17.5 g/mL HNP1 or HBD2 or of 10% FBS. Control refers to cells incubated in DMEM. Each experiment was repeated three times in triplicate. *P 0.05 by unpaired Students t test in treated group compared with the corresponding controls. (B) Western blot analysis of phospho-p42/44 MAP kinase and Akt kinase. U indicates U0126 at 15 ; W, wortmannin at 5 M. Activation of p42/44 MAP kinase and Akt was totally inhibited by U0126 and wortmannin, respectively.

FIGURE 5. Defensin-stimulated activation of Akt and p42/44 MAP kinase in HCF. HCFs in six-well plates were briey starved in DMEM before the addition of 17.5 g/mL HNP1 or HBD2. Epithelial growth factor (EGF; 2 ng/mL) was used as a positive control. Stimulation was terminated at 15 minutes for MAP kinase activation and 45 minutes for Akt activation. Total p42/44 MAP kinase and Akt kinase protein were shown as loading control.

before HNP1 or HBD2 in serum-free medium and was maintained throughout the stimulation. The increase in BrdU incorporation in HCF after HNP1 or HBD2 stimulation was abolished in the presence of U0126 (Fig. 6A). Wortmannin, an Akt kinase inhibitor, also reduced the HNP1- and HBD2stimulated BrdU incorporation to control levels (serum-free DMEM) at a concentration of 5 M. At these concentrations, U0126 and wortmannin totally blocked the activation of p42/22 MAP kinase and Akt kinase, respectively (Fig. 6B). Because wortmannin inhibits the activation of Akt by blocking the activation of phosphoinositol-3 (PI3) kinase, the upstream activator of Akt, our results suggested that PI3 kinase and MAP kinase activation are responsible for the mitogenic effects of HNP1 and HBD2 on conjunctival broblasts.

3818

Li et al.

IOVS, September 2006, Vol. 47, No. 9 modestly stimulated gene expression for collagen I, collagen III, collagen VI, and collagen VIII. However, when the protein production for collagen I was studied, we saw no clear indication of increased total collagen I protein production in HNP1treated conjunctival broblasts. On the contrary, soluble collagen I in cell culture medium was clearly reduced by HNP1 treatment. The reduction of secreted collagen I could be the result of increased extracellular degradation, a possibility considering the increased level of MMP1. On the other hand, immunouorescent staining suggested increased intracellular retention of collagen I in these cells. Retention in endoplasmic reticulum is well characterized for mutant and articially retained extracellular matrix proteins.20,3236 In these cases, gel electrophoresis showed slower migration because of overmodication in this compartment. However, we did not observe such slowly migrating collagen bands by SDS-PAGE, nor did we see a microscopic pattern suggestive of intra-endoplasmic retention. It is also known that a signicant amount of newly synthesized collagen is degraded intracellularly rather than secreted.37 Our study suggested a complicated mechanism for the effect of HNP1-stimulated collagen gene expression and protein synthesis. Collagen III, VI, and VIII protein was not identied by silver staining in cell medium or cell layer samples, suggesting low levels of protein content. As part of antimicrobial activity, defensins have been found to be chemoattractants for immune cells, such as phagocytes and mast cells.12,38 Various defensins show specicity for different cell types.38 40 In the cornea, it has been shown that keratocytes exhibit wound orientation after experimental wounds, and, with a wound, defensins in tears would be in close contact with keratocytes.41,42 However, neither HNP1 nor HBD2 was chemotactic for conjunctival broblasts. Thus, defensins may only serve this function for immune cells, which would extend innate immunity to adaptive immunity, whereas chemotaxis and wound orientation exhibited by broblasts are probably not mediated by defensins. We also observed differences between HNP1 and HBD2 with regard to their effects on broblasts. HNP1 was more potent than HBD2 in stimulating the proliferation and activation of Akt at the same concentrations. Additionally, the increase in collagen gene expression was only observed with HNP1 stimulation. These observations suggested a mechanistic difference in the functions of HNP1 and HBD2 on broblasts. Some previous studies have suggested that Toll-like receptor 4 and the chemokine receptor CCR6 could be receptors for -defensins.12,43 Indeed defensin and interleukins bear structural similarities at the C-terminus. In general, the effect of HNP1 on conjunctival broblast proliferation and the stimulation of collagen and MMP gene expression are similar to those of cytokines such as IL-1, IL-4, and TNF- .44 47 Clarication of the HNP1 and HBD2 signaling mechanisms will greatly facilitate understanding of the functions of these natural antimicrobial peptides on conjunctival wound healing and tissue remodeling.

DISCUSSION
Defensins are secreted peptides with broad-spectrum antimicrobial properties and are produced by several cell types.2,4 However, only a few reports show positive immunostaining of -defensins in diseased corneal and conjunctival stroma related to inammation.6,7,27 It has been speculated that these defensins are produced by inltrating neutrophils.4,6 In the present study, we did not detect the expression of HNP1 or HBD13 in human conjunctival broblasts by real-time PCR. Furthermore, transcription of these genes was not stimulated by exposure of HCFs to LPS or TNF- . These ndings indicate that HCFs are not the source of defensins that have been shown to be upregulated in the tears after ocular surface surgery involving the conjunctiva.8 However, in patients with conjunctival injury or inammation, polymorphonuclear neutrophilsa major source of HNP1accumulate in the damaged area within hours of injury, and their presence can be detected weeks after injury. In fact, we previously reported that an increase in HNP13 in the tears of patients was found 2 to 3 days after ocular surface surgery.8,9,28 In addition to the -defensins, broblasts are exposed to HBDs expressed by the conjunctival epithelial cells. Therefore, we went on to study the biologic effects of two representative defensins, HNP1 and HBD2, on primary cultured human conjunctival broblasts. We found that HNP1 at concentrations of 3.5 g/mL and greater and HBD2 at concentrations of 17.5 g/mL and greater stimulated the proliferation of conjunctival broblasts. Our results are consistent with the mitogenic effects of defensins previously reported for airway epithelial cells and NIH 3T3 broblasts.14,15 Furthermore, the mitogenic effect of HNP1 and HBD2 correlated with the rapid phosphorylation of p42/44 MAP kinase and Akt in HCFs. Proliferation induced by these defensins was inhibited in the presence of specic inhibitors of p42/44 MAP kinase and Akt kinase. Similarly, activation of p42/44 MAP kinase was also reported in lung epithelial cells in the presence of HNP1 to HNP3.17 The rapid onset and low concentrations needed to induce kinase activation also suggested a receptor-mediated mechanism for HNP1 and HBD2 in conjunctival broblasts. HNPs were reported to suppress DNA synthesis and cell proliferation in renal carcinoma cell lines and lung epithelial cells at concentrations higher than 25 g/mL.14,15,23 However, we found no evidence of cytotoxicity for HNP1 or HDB2 at concentrations as high as 100 g/mL. In fact, in an early study using 3T3 broblasts, a mitogenic effect of HNP1 was observed at a concentration of 60 g/mL.15 Therefore, it is possible that some epithelial cells are more sensitive than broblasts to defensins. Fibroblasts are activated during ocular surface wound healing.29,30 Proliferation and activation of selective MMP, TIMP, and collagen gene expression are hallmarks of broblast activation.24 During wound healing and tissue remodeling, the extracellular matrix undergoes dynamic changes that include degradation by various MMPs and the incorporation of new matrix components by broblasts and epithelial cells.24,30,31 Collagen I is the major component of conjunctival stroma secreted by broblasts, though the conjunctival broblasts make other collagens, such as collagen III, VI, and VIII.26 We found that HNP1 and HBD2 stimulated MMP1 gene expression in a time- and dose-dependent manner. The increase in MMP1 expression was also observed at the protein level. In addition to the stimulatory effect on MMP1, HNP1 and HBD2 also selectively increased gene expression for other MMPs and TIMPs, such as TIMP2. MMP2 expression was stimulated by HBD2, and MMP14 expression responded to HNP1. However, the increase in mRNA was small, and no changes in corresponding proteins were identied. We also found that HNP1

Acknowledgments
The authors thank Patty Chen PeiHsin for excellent technical assistant and Dr. Howard Cajucom-Uy from the Singapore Eye Bank for tissue supply.

References
1. Ganz T. Defensins: antimicrobial peptides of innate immunity. Nat Rev Immunol. 2003;3:710 720. 2. Hancock RE, Diamond G. The role of cationic antimicrobial peptides in innate host defences. Trends Microbiol. 2000;8:402 410. 3. Cunliffe RN. Alpha-defensins in the gastrointestinal tract. Mol Immunol. 2003;40:463 467.

IOVS, September 2006, Vol. 47, No. 9

Defensin Stimulation of Wound-Associated Responses

3819

4. McDermott AM. Defensins and other antimicrobial peptides at the ocular surface. Ocular Surface. 2004;2:229 247. 5. Hattenbach LO, Gumbel H, Kippenberger S. Identication of betadefensins in human conjunctiva (letter). Antimicrob Agents Chemother. 1998;42:3332. 6. Gottsch JD, Li Q, Ashraf MF, OBrien TP, Stark WJ, Liu SH. Defensin gene expression in the cornea. Curr Eye Res. 1998;17:10821086. 7. Haynes RJ, Tighe PJ, Dua HS. Antimicrobial defensin peptides of the human ocular surface. Br J Ophthalmol. 1999;83:737741. 8. Zhou L, Huang LQ, Beuerman RW, et al. Proteomic analysis of human tears: defensin expression after ocular surface surgery. J Proteome Res. 2004;3:410 416. 9. Zhou L, Beuerman RW, Barathi A, Tan D. Analysis of rabbit tear proteins by high-pressure liquid chromatography/electrospray ionization mass spectrometry. Rapid Commun Mass Spectrom. 2003; 17:401 412. 10. Yang D, Biragyn A, Hoover DM, Lubkowski J, Oppenheim JJ. Multiple roles of antimicrobial defensins, cathelicidins, and eosinophil-derived neurotoxin in host defense. Annu Rev Immunol. 2004;22:181215. 11. Zhang L, Yu W, He T, et al. Contribution of human alpha-defensin 1, 2, and 3 to the anti-HIV-1 activity of CD8 antiviral factor. Science. 2002;298:9951000. 12. Niyonsaba F, Ogawa H, Nagaoka I. Human beta-defensin-2 functions as a chemotactic agent for tumour necrosis factor-alphatreated human neutrophils. Immunology. 2004;111:273281. 13. Territo MC, Ganz T, Selsted ME, Lehrer R. Monocyte-chemotactic activity of defensins from human neutrophils. J Clin Invest. 1989; 84:20172020. 14. Aarbiou J, Ertmann M, van Wetering S, et al. Human neutrophil defensins induce lung epithelial cell proliferation in vitro. J Leukoc Biol. 2002;72:167174. 15. Murphy CJ, Foster BA, Mannis MJ, Selsted ME, Reid TW. Defensins are mitogenic for epithelial cells and broblasts. J Cell Physiol. 1993;155:408 413. 16. van Wetering S, Sterk PJ, Rabe KF, Hiemstra PS. Defensins: key players or bystanders in infection, injury, and repair in the lung? J Allergy Clin Immunol. 1999;104:11311138. 17. Aarbiou J, Verhoosel RM, Van Wetering S, et al. Neutrophil defensins enhance lung epithelial wound closure and mucin gene expression in vitro. Am J Respir Cell Mol Biol. 2004;30:193201. 18. Ang LP, Tan DT, Phan TT, Li J, Beuerman R, Lavker RM. The in vitro and in vivo proliferative capacity of serum-free cultivated human conjunctival epithelial cells. Curr Eye Res. 2004;28:307 317. 19. Li J, Houseknecht KL, Stenbit AE, Katz EB, Charron MJ. Reduced glucose uptake precedes insulin signaling defects in adipocytes from heterozygous GLUT4 knockout mice. FASEB J. 2000;14: 11171125. 20. Raghunath M, Steinmann B, Delozier-Blanchet C, Extermann P, Superti-Furga A. Prenatal diagnosis of collagen disorders by direct biochemical analysis of chorionic villus biopsies. Pediatr Res. 1994;36:441 448. 21. McDermott AM, Redfern RL, Zhang B, Pei Y, Huang L, Proske RJ. Defensin expression by the cornea: multiple signalling pathways mediate IL-1beta stimulation of hBD-2 expression by human corneal epithelial cells. Invest Ophthalmol Vis Sci. 2003;44:1859 1865. 22. Harder J, Bartels J, Christophers E, Schroder JM. Isolation and characterization of human beta-defensin-3, a novel human inducible peptide antibiotic. J Biol Chem. 2001;276:57075713. 23. Muller CA, Markovic-Lipkovski J, Klatt T, et al. Human alphadefensins HNPs-1, -2, and -3 in renal cell carcinoma: inuences on tumor cell proliferation. Am J Pathol. 2002;160:13111324. 24. Sivak JM, Fini ME. MMPs in the eye: emerging roles for matrix metalloproteinases in ocular physiology. Prog Retin Eye Res. 2002; 21:114. 25. Saika S, Ohnishi Y, Ooshima A, Liu CY, Kao WW. Epithelial repair: roles of extracellular matrix. Cornea. 2002;21(suppl):S23S29. 26. Ihanamaki T, Pelliniemi LJ, Vuorio E. Collagens and collagenrelated matrix components in the human and mouse eye. Prog Retin Eye Res. 2004;23:403 434.

27. Ikeda A, Sakimoto T, Shoji J, Sawa M. Expression of alpha- and beta-defensins in human ocular surface tissue. Jpn J Ophthalmol. 2005;49:7378. 28. Wilson SE, Liu JJ, Mohan RR. Stromal-epithelial interactions in the cornea. Prog Retin Eye Res. 1999;18:293309. 29. Lu L, Reinach PS, Kao WW. Corneal epithelial wound healing. Exp Biol Med (Maywood). 2001;226:653 664. 30. Wilson SE, Mohan RR, Ambrosio R Jr, Hong J, Lee J. The corneal wound healing response: cytokine-mediated interaction of the epithelium, stroma, and inammatory cells. Prog Retin Eye Res. 2001;20:625 637. 31. Fini ME. Keratocyte and broblast phenotypes in the repairing cornea. Prog Retin Eye Res. 1999;18:529 551. 32. Raghunath M, Bruckner P, Steinmann B. Delayed triple helix formation of mutant collagen from patients with osteogenesis imperfecta. J Mol Biol. 1994;236:940 949. 33. Raghunath M, Kielty CM, Kainulainen K, Child A, Peltonen L, Steinmann B. Analyses of truncated brillin caused by a 366 bp deletion in the FBN1 gene resulting in Marfan syndrome. Biochem J. 1994;302(pt 3):889 896. 34. Raghunath M, Kielty CM, Steinmann B. Truncated probrillin of a Marfan patient is of apparent similar size as brillin: intracellular retention leads to over-N-glycosylation. J Mol Biol. 1995;248:901 909. 35. Raghunath M, Tschodrich-Rotter M, Sasaki T, Meuli M, Chu ML, Timpl R. Confocal laser scanning analysis of the association of bulin-2 with brillin-1 and bronectin dene different stages of skin regeneration. J Invest Dermatol. 1999;112:97101. 36. Stracke JO, Hutton M, Stewart M, et al. Biochemical characterization of the catalytic domain of human matrix metalloproteinase 19: evidence for a role as a potent basement membrane degrading enzyme. J Biol Chem. 2000;275:14809 14816. 37. Ripley CR, Bienkowski RS. Localization of procollagen I in the lysosome/endosome system of human broblasts. Exp Cell Res. 1997;236:147154. 38. Niyonsaba F, Iwabuchi K, Matsuda H, Ogawa H, Nagaoka I. Epithelial cell-derived human beta-defensin-2 acts as a chemotaxin for mast cells through a pertussis toxin-sensitive and phospholipase C-dependent pathway. Int Immunol. 2002;14:421 426. 39. Yang D, Chen Q, Chertov O, Oppenheim JJ. Human neutrophil defensins selectively chemoattract naive T and immature dendritic cells. J Leukoc Biol. 2000;68:9 14. 40. Agerberth B, Charo J, Werr J, et al. The human antimicrobial and chemotactic peptides LL-37 and alpha-defensins are expressed by specic lymphocyte and monocyte populations. Blood. 2000;96: 3086 3093. 41. Beuerman RW, McDonald MB, Zhang D, Varnell RJ, Thompson HW. Diclofenac sodium attenuates neural activity after photorefractive keratectomy in rabbits. J Refract Surg. 1996;12:783791. 42. Jester JV, Petroll WM, Cavanagh HD. Corneal stromal wound healing in refractive surgery: the role of myobroblasts. Prog Retin Eye Res. 1999;18:311356. 43. Biragyn A, Rufni PA, Leifer CA, et al. Toll-like receptor 4-dependent activation of dendritic cells by beta-defensin 2. Science. 2002; 298:10251029. 44. Meller D, Li DQ, Tseng SC. Regulation of collagenase, stromelysin, and gelatinase B in human conjunctival and conjunctivochalasis broblasts by interleukin-1beta and tumor necrosis factor-alpha. Invest Ophthalmol Vis Sci. 2000;41:29222929. 45. Leonardi A, Cortivo R, Fregona I, Plebani M, Secchi AG, Abatangelo G. Effects of Th2 cytokines on expression of collagen, MMP-1, and TIMP-1 in conjunctival broblasts. Invest Ophthalmol Vis Sci. 2003;44:183189. 46. Fujitsu Y, Fukuda K, Kumagai N, Nishida T. IL-4-induced cell proliferation and production of extracellular matrix proteins in human conjunctival broblasts. Exp Eye Res. 2003;76:107114. 47. Li de Q, Shang TY, Kim HS, Solomon A, Lokeshwar BL, Pugfelder SC. Regulated expression of collagenases MMP-1, -8, and -13 and stromelysins MMP-3, -10, and -11 by human corneal epithelial cells. Invest Ophthalmol Vis Sci. 2003;44:2928 2936.