JNK MAPK Signaling Contributes in vivo to Injury-Induced Corneal Epithelial Migration
Abstract
Purpose: Injury-mediated corneal epithelial wound healing in vivo is mediated through different cell signaling pathways depending on whether or not the basement membrane is removed. Given this dependence, we ascertained if c-jun N- terminal kinase (JNK/SAPK) mitogen-activated protein ki- nase (MAPK) cell signaling mediates this response in vivo and in vitro, irrespective of the presence or absence of the basement membrane. Furthermore, in vitro the relative con- tribution was determined by the JNK/SAPK pathway to that of its p38 and ERK MAPK counterparts in mediating injury- induced increases in cell migration. Methods: Corneal epi- thelial debridement was performed in C57BL/6 mice and their organ-cultured eyes without removal of the basement membrane. In rabbits, following basement membrane re- moval by keratectomy, fluorescein-staining monitored reep- ithelialization was performed as in the mice. Immunohisto- chemistry evaluated changes in JNK phosphorylation status and localization. JNK inhibitor I and its inactive analogue de- termined if JNK signaling activation contributes to wound healing. BrdU staining assessed cell proliferation. A scratch wound assay of healing rates in SV40-immortalized human corneal epithelial cell line (HCEC) evaluated the relative con- tributions by p38 and ERK and JNK MAPK signaling activa- tion to wound healing. A TUNEL assay probed for apoptosis after wound closure of HCEC. MTT assay evaluated corneal epithelial viability. Results: Two hours following mice cor- neal epithelial debridement, phospho-JNK was transiently upregulated in the nucleus, whereas total JNK was consti- tutively expressed. JNK inhibitor I suppressed epithelial spreading in organ-cultured mouse eyes and rabbit corneal blocks, irrespective of the presence or absence of basement membrane. No proliferation was detected at the wound edges. In HCEC, a p38 (SB203580) and a JNK pathway inhibi- tor (JNK inhibitor I) inhibited migration rates more than U0126-induced ERK, whereas the JNK inhibitor I inactive an- alogue had no effect. JNK pathway inhibition wound closure in this region was not associated with either any TUNEL or BrdU-positive cells. Cell viability was unaffected by any of these MAPK inhibitors. Conclusion: JNK/SAPK pathway acti- vation stimulates wound healing in vitro and in vivo, irre- spective of the presence or absence of the basement mem- brane. Therefore, studies on how wound closure is elicited in HCEC are relevant for identifying how MAPK signaling medi- ates this response in vivo and in organ-cultured eyes. This realization suggests that the JNK signaling system has a role in vivo that is intermediate to those of ERK and p38 in medi- ating increases in cell migration.
Introduction
Corneal transparency is essential for normal vision. This function is in part dependent on the maintenance of tight junctional and epithelial integrity. Furthermore, in- jury-induced losses in corneal epithelial integrity must be rapidly restored to avoid infection, reduce the risk of cor- neal scar formation and prevent a possible loss of vision. To identify novel strategies for hastening reepithelializa- tion in a clinical setting, different animal injury models have been used to study the receptor-linked signaling events underlying this response. These studies have shown that the first step in the healing of a corneal epi- thelial wound entails single-layer resurfacing of the de- fect through cell migration across the denuded basement membrane surface. Only after this response has been ini- tiated, do the cells at the wound periphery start to prolif- erate in order to reestablish epithelial stratification [1–7]. This epithelial healing process is modulated by a variety of cytokines that are upregulated in response to injury [8–10]. They, in turn, activate their cognate receptors to induce downstream stimulation of signaling events that result in increases in migration and proliferation through stimulation of either mitogen-activated protein kinase (MAPK) signaling and/or the TGFβ-receptor-linked pathway [11, 12]. The MAPK signaling cassette is com- posed of the ERK, p38 and JNK/SAPK pathways, which operate in parallel with one another. In mice corneas, rabbit corneal epithelial cells and human corneal epithe- lial cells (HCEC), as well as organ-cultured rabbit cells, the p38 pathway mediates increases in corneal epithelial cell migration [11, 13, 14]. The ERK system stimulates proliferation in human and rabbit corneal epithelial cells. On the other hand, p38 involvement in mediating TGFβ control of wound closure is dependent on the type of ep- ithelial injury. Following rat corneal epithelial removal by keratectomy, Smad signaling is selectively activated to elicit basement membrane renewal [7]. However, follow- ing mouse corneal epithelial debridement, TGFβ acti- vates p38 signaling, leading to cell migration stimulation [11]. Therefore, delineating the role of signaling pathway control in mediating wound healing in vivo may be de- pendent on the type of injury inflicted.
JNK pathway activation has only been reported to oc- cur in rabbit and HCEC during exposure to stressors that include either a hypertonic challenge, UV light or scratch wounding [15–19]. There are no reports on whether or not JNK signaling activation by scrape injury is dependent on the presence or absence of the basement membrane. We show here that following injury, irrespective of the presence or absence of the basement membrane in vitro or in vivo, JNK pathway activation helps mediate increases in cell migration and wound closure. This agreement sug- gests that wound closure studies on HCEC are relevant for understanding how in vivo MAPK signaling mediates this response following basement membrane removal.
Materials and Methods
Reagents
The JNK pathway inhibitor (JNK inhibitor I), its inactive ana- logue, and the ERK pathway inhibitor (U0126) were obtained from Calbiochem (Darmstadt, Germany). The p38 pathway inhibitor, SB203580, was obtained from Sigma-Aldrich (St. Louis, Mo., USA). Anti-phospho-JNK antibody was obtained from Cell Signaling (Beverly, Mass., USA); the antibody to total JNK was purchased from Santa Cruz Biotechnology (Santa Cruz, Calif., USA). All ex- periments and procedures used in this study followed the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were performed according to an institutionally approved pro- tocol. Mice (C57BL/6) were anesthetized with ketamine (50 mg/kg) and xylazine (20 mg/kg) intramuscularly, and a drop of tetracaine- HCl (0.5%) was applied to the eye to deliver local corneal anesthesia before the animals were subjected to injury. Wound closure (reepi- thelialization) was monitored at the indicated times after injury using the fluorescein staining technique. Mice were humanely killed at the indicated times, and the globes were removed.
Immunohistochemistry
All experiments were performed in accordance with the ARVO Resolution on the Use of Animals in Research. In 27 male C57BL/6 mice under general anesthesia, a corneal epithelial defect was cir- cumscribed in their right eyes with a 2.0-mm trephine. The delim- ited area was then removed with a scalpel without disrupting the basement membrane. At 2 and 6 h, the animals were sacrificed us- ing anesthesia. The eyes were then enucleated, embedded in OCT compound and quickly frozen with powdered dry ice. Frozen sec- tions (5 µm) were cut with a cryostat, followed by mounting on glass slides coated with silane and cold acetone. The sections were incubated with primary antibody (monoclonal antibody against JNK; Santa Cruz Biotechnology; diluted 1:200) or monoclonal an- tibody against phospho-JNK (Santa Cruz Biotechnology; diluted 1:200) with 5% normal goat serum for 24 h at 4°C. After washing in PBS, the sections were further incubated with biotinylated goat anti-rabbit IgG (1:250) for 1 h at room temperature, then with av- idin-biotin peroxidase complex (Vectastain ABC Kit; Vector, Bur- lingame, Calif., USA) for 1 h at room temperature. The peroxidase in the tissue sections was visualized with 0.05% DAB (3,3′-diami- nobenzidine) in 0.1 M Tris-buffered saline (pH 7.5) containing 0.01% hydrogen peroxide. The specimens were dehydrated, mounted in balsam and examined under a light microscope.
Corneal Epithelial Wound Healing in Organ-Cultured Mouse Eyes
In mouse eye organ cultures, a round corneal epithelial defect was created with a 2-mm trephine in the right eyes of 44 male C57BL/6 mice. This procedure leaves the basement membrane intact. The eyes were immediately enucleated and incubated in Eagle’s minimum essential medium supplemented with 2% fetal calf serum with either 5 µM JNK inhibitor I or 5 µM of its inactive analogue. The closure of the epithelial defect was determined by fluorescein staining after a 2-hour labeling with BrdU (1:1,000) followed by 6, 12 and 48 h of culture. Eyes were then fixed in 4% paraformaldehyde after BrdU-labeling and embedded in paraf- fin. Deparaffinized sections (5-µm thick) were processed for his- tology and BrdU immunostaining. For BrdU immunostaining, paraffin sections were treated with 2 N HCl for 1 h at 37 °C and then washed 11 times in PBS before the application of anti-BrdU antibody (PBS; Roche Diagnostics, Mannheim, Germany).
TUNEL Assay
Apoptosis was assessed following wound closure by perform- ing TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP end labeling of fragmented DNA) assays. In brief, cells were digested with proteinase K (10 µg/ml; Sigma) for 5 min at room temperature. After the cells were washed, they were then treated with 1x TdT buffer (TdT; Gibco BRL, Grand Island, N.Y., USA) and biotinylated dUTP (Boehringer Mannheim, Mannheim, Germany) for 45 min at 37 °C. The cells were then washed with PBS, followed by a reaction with strepto-avidin peroxidase and DAB. As a negative control, cells were stained without biotinyl- ated dUTP. Creating a positive control entailed treating the cells with 10 µg/ml DNase I for 10 min [20].
Cell Viability
MTT assay was performed according to the manufacturer’s instructions. A standard curve was constructed in which a linear region was identified between changes in cell density and optical absorbance. For each condition of interest, 300,000 cells/ml were loaded in 100 µl per well on a 96-well plate in triplicate. The tet- razolium compound MTT [3-(4,5-dimethylthiazol-2-yl) 2,5 di- phenyltetrazolium bromide] was added 24 h after exposure to an inhibitor. The detergent reagent provided was added following 2 h exposure to MTT, followed by incubation at 37 °C for another 24 h. The optical density was read at 570 nm using a reference wavelength of 650 nm.
Organ-Cultured Rabbit Corneal Epithelial Migration
Cell migration along corneal blocks in the absence of a base- ment membrane was evaluated with a minor modification [21– 22]. Adult albino rabbits were sacrificed by an intravenous injec- tion of an overdose of pentobarbital sodium. Central corneal but- tons without the limbal zone were obtained from each eye. The blocks (2 x 4 mm) were incubated for 24 h in 24-well culture plates (Becton Dickinson, Lincoln Park, N.J., USA) in Eagle’s minimum essential medium with JNK inhibitor I (5 µM) or its inactive analogue (5 µM). Four samples were used per condition. Following incubation, a block was fixed in 10% formaldehyde so- lution, embedded in OCT compound and quickly frozen with powdered dry ice. Frozen sections (5 µm) were cut with a cryostat and then mounted on glass slides coated with silane. The sections were stained with HE and photographed under light microscopy. The epithelial migratory distance was measured within 50-µm segments. The distance traversed by the cell layer front beyond the block edge was used to evaluate the role of JNK activation in mediating epithelial migration. The average of 2 measurements per block determined the epithelial migratory distance.
HCEC Wound Healing
SV40-immortalized HCEC were obtained from Riken Labo- ratories, Tsukuba, Japan. The phenotype of this cell line retains many of the properties identified in their primary cell counter- part [23]. They were grown to confluence in DMEM/F12 medium containing 200 U/ml of penicillin and streptomycin, 5% FBS, 0.1 µg/ml cholera toxin, 5 µg/ml insulin, and 10 ng/ml human epi- dermal growth factor (EGF). The cells were cultured without EGF and serum-starved for 24 h prior to experimentation. Two linear defects were produced using a 27-gauge silicone-coated needle (Alcon Surgical, Fort Worth, Tex., USA). Healing kinetics were evaluated in the presence and absence of the following relatively selective MAPK inhibitors: JNK inhibitor I (JNK; 5 µM); SB203580 (p38; 10 µM); U0126 (Mek1/2; 10 µM). Closure of the defect on each of the 4 slides was observed by light microscopy, and evalu- ated at 4 different points along the wound margins. The points chosen in each image included those in which the encroaching jagged margins protruding into the wound were closest and fur- thest apart from one another. Two other points were chosen whose separation was intermediate between these 2 points. The distanc- es were measured with a caliper, and the mean was calculated for each image.
Results
In situ JNK Activation
Figure 1 compares the time-dependent changes in live mice of their total JNK and phosphorylated JNK levels in the uninjured corneal sections with those obtained at 2 and 6 h subsequent to wounding. Nonphosphorylated JNK was detected at all times in both types of sections. Little or no phospho-JNK immunoreactivity was detect- ed in the uninjured epithelium, whereas immediately af- ter epithelial injury and up until 6 h phospho-JNK im- munoreactivity was detected in the cytosol. In contrast, its nuclear localization was only detectable after 2 h. For- ty-eight hours after injury, phospho-JNK was no longer detectable in either of these regions. There was no detect- able p-JNK in uninjured sections at any time after any of these periods. Appearance of nuclear phospho-JNK im- munoreactivity is reflective of JNK/SAPK pathway in- volvement in the wound healing process.
JNK Induces Mouse Corneal Epithelial Wound Closure
In organ-cultured mice eyes, JNK-signaling-mediated corneal epithelial wound closure was determined by comparing fluorescein-stained areas at 18, 24 and 36 h after injury during exposure to either 5 µM JNK inhibitor I or its inactive analogue at the same concentration. The inactive analogue served as a negative control for assess- ing JNK-inhibitor I effects since wound closure rates were the same with this compound as those in nontreated eye- balls. Figure 2a shows that with JNK inhibitor I healing was somewhat delayed at 18 and 24 h as compared to its inactive analogue. Figure 2b is a summary comparing the remaining wounded areas in these images. Prior to 18 h, the effect of JNK inhibitor I was not significant. However, after 18 h, the difference in healing rates between the in- active analogue and JNK-inhibitor-I-treated eyes became more pronounced. At 18 h, with the inactive analogue, the defect had further decreased to 1.13 ± 0.21 mm2 (n = 4), whereas in those globes cultured with JNK inhibitor I the remaining wound area was 1.71 ± 0.40 mm2. This difference represents an inhibition of wound closure by JNK inhibitor I of 34% (p ! 0.05). In another group of eyes cultured for 24 h, the wound area in the presence of this inhibitor decreased to 1.23 ± 0.11 mm2, whereas with its inactive analogue it shrank more to 0.83 ± 0.20 mm2 (n = 4). In this case, JNK inhibitor I suppressed closure by 33% (p ! 0.05). At 36 h, wound closure was nearly com- plete despite the presence of JNK inhibitor I. Figure 3 shows that BrdU staining (i.e. proliferation) in the pe- riphery was unchanged in the presence and absence of the JNK inhibitor I, suggesting a nonselective JNK in- hibitor I effect.
JNK Mediated Cell Migration across Rabbit Corneal Blocks
In 12 isolated rabbit corneal blocks, following expo- sure to 5 µM JNK inhibitor I for 24 h, epithelial migration was suppressed from 808 ± 363 to 525 ± 271 µm (i.e. 34%, p ! 0.01). This decline suggests that JNK signaling contributes to corneal epithelial migration during injury- induced wound healing.
MAPK Pathway Contributions to HCEC Migration
Figure 4a shows the images of wound healing in the presence and absence of JNK, p38 and ERK pathway MAPK inhibitors as well as the inactive analogue of a JNK inhibitor. None of these inhibitors were cytotoxic, based on negative results obtained with the MTT assay. Figure 4b summarizes the kinetics of wound closure. Un- der control conditions, wound closure was complete after about 14 h, which is 10 h quicker than the time reported with the same cell line in another recent study [18]. With 5 µM JNK inhibitor I, wound closure was delayed about 13 h from the time required for the control and the inac- tive analogue. The largest decline occurred at about 14 h, relative to its inactive analogue. The JNK inhibitor I sup- pression of closure was smaller than with 10 µM SB203580, since even at 27 h closure had not yet occurred with this p38 MAPK inhibitor. With 10 µM U0126, wound closure was delayed by about 10 h and at all times migration was slower than in the untreated cells. However, these de- clines were less pronounced than those obtained with ei- ther the p38 inhibitor or JNK inhibitor I.
Discussion
To delineate the mechanisms underlying corneal epi- thelial wound healing control, it is necessary to take into account the complexity of receptor-mediated control of wound healing. Following corneal epithelial injury, the wound healing response is elicited through a host of cy- tokines that undergo upregulation. In addition, injury exposes the extracellular membrane and matrix proteins, which can mediate increases in cell migration and prolif- eration through some of the same signaling pathways that elicit cytokine receptor stimulation of these respons- es [24]. As numerous mediators activate a myriad of cell signaling pathways, which can interact with one another to modulate their activation, it is important to validate in the intact cornea whether or not a specific signaling path- way is activated in the same manner as in a less complex cell culture system.
To delineate the influence of the extracellular milieu upon TGFβ-induced wound healing, the cell signaling pathways activated by this cytokine were identified fol- lowing corneal epithelial debridement and keratectomy in vivo and in vitro. In one case, it was shown that the cell signaling pathways induced by exogenous TGFβ are dif- ferent depending on whether or not the wound removed the basement membrane [25]. This result supports the notion that merely studying corneal epithelial wound healing in vitro with a scratch wound assay may not ad- equately mimic the complexity of this process. Therefore, we chose to determine if the reported involvement of JNK/SAPK pathway activation by injury of HCEC also induces increases in migration irrespective of the pres- ence or absence of the basement membrane.
The results of the current study show that JNK activa- tion contributes to the wound healing response irrespec- tive of the presence or absence of the basement mem- brane. Given this agreement, the effects of JNK pathway inhibition on migration by JNK inhibitor I in this study can be accounted for by the previously reported inhibi- tion in HCEC by SP600125 of JNK phosphorylation that occurs transiently 15 min following injury [18]. Since our assessment included use of mice in vivo and organ- cultured mouse eyes as well rabbit corneal blocks, inju- ry-induced JNK activation is also not sensitive to: (1) lack of innervation; (2) absence of stroma and orbital gland secretions; (3) use of a phenotypically modified immor- talized cell line; (4) loss of an intact barrier function. Other potential differences between HCEC in culture and these 3 different tissue preparations used by us that did not impact on JNK activation include the heteroge- neity of receptor upregulation in response to wounding.
Receptor-mediated control of wound healing is a dy- namic process suggesting differential signaling control of this response. Following rat corneal injury in vivo, there is immediate TGFβ receptor upregulation at the edge and downregulation of the EGF receptors, whereas periph- eral to the wound EGF receptors are upregulated [25]. These differences may account for why cells stop prolif- erating at the edge and instead continue to migrate, whereas in the peripheral regions marked increases occur in proliferation. In mouse cornea organ cultures, losses in ERK pathway activation occurred at the wound edge, whereas further away from the wound the converse was found [11]. Nevertheless, none of these differences changed the role of JNK signaling in mediating injury- induced increases in cell migration.
Activation of the JNK pathway is particularly impor- tant during the early phase of wound closure since JNK inhibitor I most markedly delayed wound closure over this period. There are several indications of JNK inhibi- tor I selectivity: (1) its inactive analogue failed to suppress migration; (2) neither the JNK inhibitor I nor its inactive analogue had any effect in organ-cultured mouse eyes on injury-induced increases in cell proliferation; (3) in HCEC, these compounds were not cytotoxic according to MTT assay results. Unlike other stressors, which also in- duce JNK signaling in HCEC, no apoptosis was detected with the TUNEL assay. In those studies, exposure to hy- pertonic media or UV irradiation induced cell death [15– 17]. This difference points to possible signaling complex- ity caused by different receptors and/or mediators being activated by each one of these stressors. Our finding that p38 pathway inhibition in HCEC suppressed migration is consistent with other in vitro and in vivo studies [11, 13].
The fact that ERK pathway inhibition in HCEC also suppressed migration is consistent with numerous re- ports that all three MAPK pathways mediate receptor control of cell migration [26]. In renal epithelial cells and fibroblasts, ERK involvement in migration has been de- scribed [27, 28]. Furthermore, JNK-null fibroblasts ex- hibit less migratory activity. Similarly, JNK-induced in- creases in cell motility were described in human airway epithelial cells [29]. JNK is also required for maximal mi- gratory activity of cultured fibroblasts and brain endo- thelial cells [30, 31]. The downstream effectors of JNK signal activation in the corneal epithelium are unknown. In vascular endothelial cells, JNK-induced increases in angiogenesis involved cytoskeletal reorganization and upregulation of matrix metalloproteinase-2 (MMP-2) ex- pression [32]. During corneal wound healing resulting from anterior keratectomy in rabbits, MMP-2 is one of four MMPs that are upregulated [33]. Three of these MMPs are thought to be involved in remodelling of the stroma and reformation of epithelial basement mem- brane [34]. However, the role of JNK activation in this response remains unclear.
In summary, irrespective of the presence or absence of a basement membrane, JNK pathway inhibition selec- tively suppressed corneal epithelial injury-induced cell migration in mice in vivo and organ-cultured mouse, rabbit and HCEC. This finding suggests that inhibition of JNK activation and cell migration might not be related to either a change in cell matrix adhesion interaction or keratocyte presence. On the other hand, JNK activation may stimulate cell migration through a direct effect on a cytoskeletal element controlling cell motility. Kimura et al. [18] reported that in HCEC JNK activation modifies paxillin through phosphorylation in its complex with fo- cal adhesion kinase. Once the downstream mechanism of JNK control of migration has been more fully eluci- dated, other studies may be indicated to assess whether or not in a clinical setting it will be possible to better treat persistent corneal epithelial defects by transfecting these cells with vectors containing a CC-930 construct that leads to constitutively active JNK gene expression.