Induction of the Ubiquitin–Proteasome Pathway during the Keratocyte Transition to the Repair Fibroblast Phenotype

Abstract

purpose. To examine dynamics and function of the ubiquitin (Ub)-proteasome pathway (UPP) during corneal stromal cell acquisition of the repair fibroblast phenotype.

methods. An established cell culture model was used in which freshly isolated rabbit corneal stromal cells acquire a repair fibroblast phenotype, thereby mimicking injury-induced stromal cell activation.

results. Transition to the repair fibroblast phenotype during the 72 hours after initial plating was coincident with progressive UPP induction. Levels of Ub, Ub-conjugated proteins, ubiquitinylating enzymes E1 and E2-25K, and 26 S proteasome increased two- to fivefold in activated stromal cells. These increases were associated with enhanced (>10-fold) capacity for Ub-dependent proteolysis of ¹²⁵I-labeled H2A and with progressive (>6-fold) increases in the UPP substrate, inhibitor of κBα (IκBα). Because IκBα expression is induced by nuclear factor (NF)-κB, this finding suggests that rates of constitutive NF-κB activation, and thus IκBα degradation, are elevated in activated stromal cells. Both freshly isolated and activated stromal cells degraded IκBα in response to IL-1α; yet, only activated stromal cells maintained autocrine IL-1α expression after 24 hours. UPP induction was coincident with a more than 90% loss of tissue transketolase (TKT) and aldehyde dehydrogenase (ALDH) class 1. TKT was stabilized during the repair phenotype transition by proteasome inhibition and was degraded (>30%/h) by the UPP in cell-free assays.

conclusions. Coordinate induction of the UPP during stromal cell activation alters levels of IκBα and TKT, two UPP substrates that are implicated in the loss of tissue stasis and corneal clarity after injury.

Corneal transparency is maintained in part by the resistance of corneal stromal cells (keratocytes) to stimuli that typically induce proliferative, inflammatory, and fibrotic responses in other cell types. ¹ However, extensive injury to the corneal stroma can activate in keratocytes at the wound edge a transformation to a repair fibroblast phenotype. ¹ These activated fibroblasts migrate into the wound, proliferate, and deposit a disordered, opaque extracellular matrix, thereby compromising corneal clarity (Fig. 1) . The transition of quiescent keratocytes to the repair phenotype is associated with profound molecular, biochemical, and morphologic changes. One important change is the establishment of an IL-1α autocrine loop, which maintains synthesis and secretion of the tissue-remodeling protease, collagenase. ^{2 3} Induction of the IL-1α autocrine loop is one of many responses of keratocytes to injury that are mediated through the transcription factor, nuclear factor (NF)-κB. ⁴ We have proposed that stromal cell acquisition of competence to activate NF-κB is a critical step in the loss of corneal stasis and the establishment of the repair fibroblast phenotype after injury. ^{1 5} Corneal repair fibroblasts are further distinguished from stromal cell progenitors by the relative absence of corneal crystallins, ^{6 7} the abundant soluble proteins that are hypothesized to protect corneal clarity by minimizing refractive index fluctuations between the cytoplasm and the extracellular milieu. ^{6 7 8} The specific proteins vary with species. ⁷ Stromal cell loss of corneal crystallins after corneal injury may therefore compromise the optical properties of the cornea. 

At least three hallmarks of the repair phenotype transition—cell cycle progression, ligand-induced activation of NF-κB, and the selective loss of bulk protein(s)—have been individually shown, in physiological contexts other than wound healing, to be regulated by the ubiquitin (Ub)-proteasome pathway (UPP). ^{9 10} The UPP is a highly conserved pathway of selective protein modification (ubiquitinylation) and degradation. Substrates of the pathway are covalently ligated (conjugated) at internal lysine(s) to one or more monomers of ubiquitin, an 8.5-kDa protein, by the sequential activities of three families of enzymes: adenosine triphosphate (ATP)-dependent Ub-activating enzymes (E1s), Ub-conjugating enzymes (Ubcs/E2s), and Ub-isopeptide ligases (E3s). Substrate selectivity is achieved at the level of Ubcs and E3s. The sequential ligation of multiple Ubs as a polyubiquitin chain dramatically increases the apparent mass of the substrate and targets it for rapid degradation by the 26 S proteasome, a multicatalytic, ATP-dependent protease. ¹¹ Ub-dependent protein degradation rapidly promotes new cellular steady states in response to changes in the external environment. Thus, the UPP controls cell cycle by degradation of cyclins and cylin-dependent kinases, ¹² activates NF-κB–dependent gene transcription by degradation of inhibitors of κB (IκBs), ¹³ and facilitates cellular remodeling during embryogenesis and terminal differentiation by degradation of obsolete bulk proteins. ^{14 15 16 17 18 19}  

In this study, we used an in vitro rabbit model of the corneal repair fibroblast transition to investigate UPP dynamics and contribution to the loss of corneal stasis after injury. In this model, ^{1 2 3 5} keratocytes that are freshly isolated from the corneal stroma initially remain quiescent, mimicking the in vivo setting. However, after exposure to serum, cultured cells begin to acquire the morphologic and biochemical attributes of repair fibroblasts. ^{20 21} These attributes are maintained after serum withdrawal, indicating that the repair phenotype transition is a differentiation process. ¹ The capacity to establish the IL-1α autocrine loop in response to IL-1α is observed after 48 hours, and additional features of the repair phenotype are manifest by 72 hours. ^{3 5} Dynamics of the rabbit corneal crystallins ⁶ —tissue transketolase (TKT) and aldehyde dehydrogenase class (ALDH) I—have not been examined in this model. Once acquired, the repair fibroblast phenotype is maintained in primary culture and during subculturing. ^{3 5} In the present study the repair phenotype transition was coincident with the coordinate induction of multiple UPP components and with enhanced capacity for protein ubiquitinylation and Ub-dependent proteolysis. Based on these findings, we tested the hypothesis that the inability of keratocytes to establish the IL-1α autocrine loop is due to a deficiency in IκBα degradation. We demonstrated in addition that the loss of TKT during the repair phenotype transition reflected Ub-dependent proteolysis, thereby directly implicating the UPP in the loss of corneal clarity after injury. 

Materials and Methods

Materials

Materials for electrophoresis were obtained from Bio-Rad Laboratories (Hercules, CA), polyvinylidene difluoride (PVDF) membrane from Millipore (Bedford, MA), Coomassie Plus protein assay reagent from Pierce (Rockford, IL), Na¹²⁵I from DuPont NEN (Boston, MA), chemiluminescence kits from Amersham Life Science (Arlington Heights, IL) or from Pierce, MG132 from Calbiochem (La Jolla, CA), and Clasto-lactacystin β-lactone from Boston Biochem, Inc (Cambridge, MA). Whole rabbit eyes and reticulocytes were purchased from Pel Freez (Rogers, AR). Reticulocyte lysate was prepared by standard techniques. ²² All other materials were purchased from Sigma Chemical Co. (St. Louis, MO) and were the highest grade available. Histone 2A was iodinated by reaction with chloramine T. ²²  

Cell Culture

Stromal cells were obtained from trephined corneas of New Zealand White rabbits by collagenase digestion, as previously described. ⁵ All procedures were in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Harvested cells were cultured in MEM containing 10% calf serum. Freshly isolated (day 0) stromal cells were maintained for up to 5 days in primary culture. Cells were designated as day 1 after 24 hours in culture. Cells that were subcultured after acquiring the repair fibroblast phenotype were used between the second and fourth passages. For establishment of the IL-1α autocrine loop, cells were treated for 24 hours with human IL-1α (10 ng/ml), as previously described. ⁵ To assess the role of the UPP in TKT turnover, day-1 cells were cultured for an additional 24 hours (day 2) in complete medium with the cell-permeable proteasome inhibitor, clasto-lactacystin β-lactone, or were sham treated with dimethyl sulfoxide (DMSO) carrier (≤0.2% final concentration). Cell proliferation was assessed by cell counts of triplicate wells using a hemocytometer and by scintillation counting of[ ³H]thymidine uptake. Cell counts are expressed as mean ± SEM. 

Preparation of Cell Lysates, Supernatants, and Conditioned Medium

For analysis of levels of Ub, Ub-protein conjugates, E1s, E2s, and proteasome subunits, cell lysates were prepared by scraping PBS-washed cells into 150 μl lysis buffer (5 mM Tris-HCl, 4% sodium dodecyl-sulfate [SDS], 20 mM N-ethylmaleimide, [pH 7.6]) followed by boiling. Insoluble material was removed by centrifugation (15,000g, 10 minutes 4°C). For analysis of IκBα, TKT, and IL-1α, cells were solubilized (30 minutes, 4°C) in PBS containing 1% Nonidet-P40 (NP-40), 0.5% sodium deoxycholate, 0.1% SDS, 10 μg/ml phenylmethylsulfonyl fluoride, 20 μg/ml aprotinin, and 20 mM dithiothreitol (DTT). In some experiments, lysates also contained phosphatase inhibitors (0.1 mM sodium orthovanadate and 50 mM NaF). Lysates were boiled after addition of 2× SDS-PAGE loading buffer. 

For cell-free assessment of Ub-dependent proteolytic capacity, stromal cell supernatants containing an active UPP were obtained from scraped cells after resuspension and homogenization by hand in ice-cold 5 mM Tris-HCl-0.5 mM DTT (pH 7.8) and centrifugation (85,000g, 20 minutes, 2°C). 

Twenty-four-hour conditioned medium was collected from sham- and IL-1α–treated cultures, and cell debris was removed by centrifugation (5000g, 2 minutes, 4°C). Cell lysates, supernatants, and conditioned medium were aliquoted and stored at− 80°C. 

Protein Electrophoresis and Immunoblot Analysis

Cell lysates, proteolysis assay mixtures (described later), and conditioned medium were subjected to reducing SDS-PAGE, and the electrophoresed proteins were transferred to PVDF membrane. For detection of monomeric Ub, blots were autoclaved (30 minutes) before immunodetection. Blots were probed with the following antibodies: (1) rabbit IgG that recognizes both free and conjugated ubiquitin ²² ; (2) rabbit IgG that recognizes the two isoforms of Ub-activating enzyme, E1A and E1B ²³ ; (3) rabbit IgG raised against recombinant bovine E2-25K (the latter generously provided by Cecile Pickart, Johns Hopkins University, Baltimore, MD); (4) rabbit serum raised against the Trip1 subunit of the 26 S proteasome (a generous gift of Richard Young, Massachusetts Institute of Technology, Cambridge, MA); (5) rabbit serum raised against the human 20 S proteasome (generously provided by George DeMartino, University of Texas Southwestern Medical Center, Dallas, TX); (6) rabbit IgG raised against a carboxyl-terminal peptide of human IκBα (number Sc-371; Santa Cruz Biotechnology, La Jolla, CA); (7) goat IgG that specifically recognizes rabbit IL-1α but not human IL-1α (Endogen, Woburn, MA), (8) rabbit serum raised against human TKT, (a generous gift from Joram Piatigorsky, National Eye Institute, Bethesda, MD); (9) rabbit serum raised against mouse ALDH1 (a generous gift from Gregg Duester, Burnham Institute, La Jolla, CA); or (10) appropriate preimmune sera or control IgG. Specific binding was detected by enhanced chemiluminescence (ECL), visualized by autoradiography, and quantified by densitometry (Molecular Dynamics, Sunnyvale, CA). 

Cell-Free Proteolysis Assays

¹²⁵I-labeled histone 2A was incubated (1 hour, 37°C) in ATP-depleted and ATP-supplemented assays ²² containing 6 mg/ml (final concentration) supernatant from day 1 or subcultured stromal cells. Some ATP-supplemented assays also contained the proteasome inhibitor MG132 (80 μM final concentration). Proteolysis was quantified byγ -counting of acid-precipitable radioactivity. Degradation of TKT by the UPP of reticulocyte lysate was assessed by incubating 10 μl of day-1 supernatant (containing ∼1 μg TKT) with 15 μl ATP-depleted or ATP-supplemented rabbit reticulocyte lysate. Some ATP-supplemented assays also contained MG132 (80 μM). Proteolysis was terminated after 1 hour with SDS-PAGE sample buffer, and loss of TKT was assessed by Western blot analysis using anti-TKT serum. 

Results

Increase in Cellular Pools of Free and Conjugated Ub in Corneal Repair Fibroblasts during the Repair Phenotype Transition and Thereafter

Consistent with previous studies, ^{3 5} corneal stromal cells acquired the repair fibroblast phenotype by day 3 (72 hours after plating), based on spindle-shaped morphology and the elaboration of stress fibers (data not shown). We initially asked whether the acquisition of the repair fibroblast phenotype was associated with alterations in levels of free (nonconjugated) Ub and Ub-protein conjugates. Immunoblots of cell lysates from days 1, 2, and 3 stromal cell cultures indicate that levels of free Ub increased two- to threefold between days 1 and 2 and between days 2 and 3 (Figs. 2A 3B ). Thus, levels of free Ub increased five- to sixfold between days 1 and 3 (Figs. 2A 2B) . The cellular pool of conjugated Ub also increased (2.5-fold) between days 1 and 2 and attained maximum levels on day 3 (Figs. 2A 2B) . This increase in conjugated Ub suggests that protein ubiquitinylation is upregulated in day-2 and day-3 stromal cell cultures compared with day-1 cultures. Coordinate increases in free and conjugated Ub between days 1 and 3 (Figs. 2A 2B) indicate that the equilibrium between free and conjugated Ub pools is essentially maintained coincident with enhanced Ub expression. The high mass of the enhanced pool of Ub-protein conjugates in day-2 and day-3 cultures (Fig. 2A) suggests that these conjugates contain polyubiquitin chains and are therefore targeted to the 26 S proteasome for degradation. Together, these results provide evidence that the repair phenotype transition is associated with increased Ub synthesis and increased protein ubiquitinylation. Elevated levels of Ub and Ub-protein conjugates persisted after the acquisition of the repair phenotype on day 3, based on immunoblots of day-4 and day-5 cultures (data not shown). 

The repair fibroblast phenotype is maintained when cells are subcultured. ^{3 5} Consistent with data obtained with primary cultures (Figs. 2A 2B) , levels of free Ub (not shown) and conjugated Ub were four to five times greater in subcultured cells compared with day-1 cells (Fig. 2C ; compare lanes 4–6 with lanes 1–3). Increases in conjugated Ub in subcultured cells were apparent across the entire molecular mass range of ubiquitinylated protein (50 to >200 kDa), suggesting that protein ubiquitinylation is globally upregulated in subcultured corneal stromal cells. Considered together, our data indicate that increased pools of free Ub and ubiquitinylated protein were characteristic of the transition to the repair fibroblast phenotype. 

Elevated Levels of E1s and E2-25K in Corneal Repair Fibroblasts

The initial and obligate enzymatic step in protein ubiquitinylation is the activation of Ub as a thiolester by the Ub-activating enzyme, E1. The two E1 isoforms, E1A and E1B, constitute the nuclear and cytoplasmic E1 pools, respectively, ^{23 24} and they appear to be enzymatically redundant. ²³ Immunoblots indicate that both day-1 and subcultured stromal cells contained approximately equimolar amounts of E1A (117 kDa) and E1B (110 kDa; Fig. 3 , top). However, levels of both E1A and E1B were, on average, two times higher in subcultured cells compared with day-1 cells (Fig. 3 , top; compare lanes 1–3 with lanes 4–6). In conjunction with increased levels of free Ub (Figs. 2A 2B) , elevated expression of both E1 isoforms suggests that rates of Ub activation are increased in both the nucleus and cytoplasm of subcultured cells compared with day-1 cells. Because activated Ub is required for the activity of all Ub-conjugating enzymes (Ubcs), elevated levels of E1A and E1B are consistent with the apparent nonselective increase in Ub-protein conjugates in subcultured cells (Fig. 2 , and the Discussion section). 

Ubcs-E2s catalyze the irreversible transfer of Ub to target proteins, frequently in association with Ub ligases (E3s). Immunoblots confirmed that levels of at least one Ubc (i.e., E2-25K) were higher (range, 2.3–3.0-fold) in subcultured cells than in day-1 cells (Fig. 3 , lower panel; compare lanes 1–3 with lanes 4–6). Enhanced E2-25K activity may contribute to the widespread (i.e., nonselective) increases in Ub-protein conjugates in subcultured cells (see the Discussion section). Antibodies raised against rat Ubc 2, yeast Ubc 3, and rat Ubc 4-1 failed to unambiguously detect their respective rabbit homologues in corneal stromal cells (data not shown). 

Upregulation of Capacity for Ub-Dependent Proteolysis in Corneal Repair Fibroblasts

We assessed relative levels of the 26 S proteasome in cell lysates of day-1 and subcultured corneal stromal cells by determining relative levels of 26 S proteasome subunits. Immunoblots were probed with an antibody to Trip1-SUG1-p45, ²⁵ an ATPase component of the 26 S proteasome, or with serum that recognizes a common ∼32-kDa subunit of the 20 S proteasome core particle. ²⁶ Levels of these proteasome subunits were elevated 3.1- to 5.3-fold and 3.5- to 4.2-fold, respectively, in subcultured compared with day-1 cells (Fig. 4A ; compare lanes 3 and 4 with lanes 1 and 2). These results indicate that levels of the 26 S proteasome were elevated in subcultured corneal stromal cells compared with day-1 cells. 

We directly assessed the capacity for Ub-dependent proteolysis in day-1 and subcultured cells, by using the exogenous UPP substrate ¹²⁵I-labeled H2A. Supernatant from day-1 cells failed to significantly degrade ¹²⁵I-labeled H2A, whereas supernatant from subcultured cells degraded ¹²⁵I-labeled H2A (∼10% per hour) by an ATP-dependent mechanism (Fig. 4B) . This ATP-dependent degradation was completely abrogated by the proteasome inhibitor, MG132, thereby confirming that H2A proteolysis was Ub dependent. These results demonstrate that the capacity for protein degradation by the UPP is enhanced in corneal stromal cells that have made the transition to the repair fibroblast phenotype. This enhanced proteolytic capacity is consistent with elevated levels of 26 S proteasomes in these cells (Fig. 4A) and probably also reflects increased cellular capacity for protein ubiquitinylation (Figs. 2 3) . 

Dynamics of IκBα during the Repair Phenotype Transition and in Response to Il-1α

Release of the transcription factor NF-κB to the nucleus requires the phosphorylation and Ub-dependent degradation of inhibitor(s) of κB, the IκBs. ¹³ In a surprising observation, steady state levels of IκBα became progressively elevated during the repair phenotype transition (Fig. 5A ). IκBα levels increased 2.3- to 4.0-fold between days 1 and 2 (Fig. 5A ; compare lanes 1 and 2) and continued to increase (1.6–2.2-fold) between days 2 and 3 (Fig. 5A ; compare lanes 2 and 3). These elevated levels of IκBα were maintained when cells were subcultured (described later). As previously reported, ⁵ subcultured cells did not express elevated levels of p50 or p65 NF-κB subunits (data not shown). Thus, constitutive levels of IκBα increased substantially in the absence of increased p50-p65. 

To reconcile elevated levels of IκBα with the acquisition of competence for IL-1α–dependent NF-κB activation, we evaluated the capacity for IL-1α–induced IκBα degradation in day-1 and subcultured stromal cells. Cells were treated either with IL-1α (10 ng/ml, 30 minutes) or with buffer (sham) and were then harvested for immunoblot analysis in the presence of phosphatase inhibitors. Both native IκBα and constitutively phosphorylated IκBα were detected in day-1 cultures, with phosphorylated IκBα the more prevalent species (Fig. 5B , lanes 1 and 2). Treatment with IL-1α resulted in rapid loss of more than 90% of IκBα (Fig. 5B ; compare lanes 3 and 4 with lanes 1 and 2). Levels of phosphorylated IκBα were preferentially diminished compared with levels of native IκBα, consistent with phosphorylation-dependent targeting of IκBα for degradation. However, consistent with previous studies, ⁵ day-1 cells failed to express IL-1α 24 hours after exposure to exogenous IL-1α (Fig. 5C , lanes 1 and 2). Thus, day-1 cells initially activated NF-κB in response to IL-1α, but did not establish a persistent IL-1α autocrine loop. 

Compared with day-1 cells, subcultured cells contained increased steady state levels of IκBα (Fig. 5B ; compare lanes 5 and 6 with lanes 1 and 2). This result confirmed our observation that IκBα levels increased during the repair phenotype transition (Fig. 5A) . In contrast with day-1 cells, only a small proportion of IκBα in subcultured cells was phosphorylated (Fig. 5B ; compare lanes 5 and 6 with lanes 1 and 2). Exposure of subcultured cells to IL-1α resulted in almost total loss of both forms of IκBα (Fig. 5B ; compare lanes 7 and 8 with lanes 5 and 6). These results suggest an association between upregulated UPP function (Figs. 2 3 4) and the capacity of repair fibroblasts to efficiently degrade elevated levels of IκBα in response to IL-1α (Fig. 5B) . In contrast to day-1 cells (Fig. 5C , lanes 1 and 2), subcultured cells expressed IL-1α protein 24 hours after exposure to IL-1α (Fig. 5D , lanes 3 and 4), thereby confirming their acquisition of competence for the IL-1α autocrine loop. 

Loss of TKT during the Repair Phenotype Transition: Evidence for TKT Degradation by the UPP

Day-1 cells contain abundant levels of ∼70-kDa and ∼50-kDa proteins, which are almost entirely absent from subcultured cells (Fig. 6A ; compare lanes 1 and 2 with lanes 3 and 4). The molecular masses of these two proteins suggest that they were TKT (69 kDa) and ALDH1 (54 kDa), abundant water-soluble proteins that are lost from the rabbit corneal stroma after injury. ⁶ Immunoblots confirmed that the abundant ∼70 kDa and ∼50 kDa proteins in day-1 cells were TKT and ALDH1, respectively, and that neither protein was detectable in subcultured corneal stromal cells (Fig. 6B) . TKT levels declined 77% on average (range, 72%–83%) either between days 1 and 2 (Fig. 6C , left) or between days 2 and 3 (Fig. 6C , right). TKT levels continued to diminish, so that by day 5, levels were reduced by more than 90%, compared with day-1 levels (data not shown). These results establish that TKT levels declined in vitro coincident with transition of stromal cells to the repair fibroblast phenotype and coincident with UPP upregulation (Figs. 2 3 4) . 

Although these data suggest that TKT is degraded during the repair phenotype transition, declines in detectable TKT could reflect downregulation of TKT synthesis coincident with cell proliferation (i.e., dilution). Cell counts and [³H]thymidine uptake studies indicate, however, that cell number increased 4% or less between days 1 (2.00 ± 0.12 × 10⁵ cells per well) and 2 (2.07 ± 0.13 × 10⁵ cells per well). Thus, declines in TKT levels between days 1 and 2 reflect actual loss of TKT, presumably due to proteolysis. Cell number increased approximately 50%, however, between days 2 and 3 (3.07 ± 0.09 × 10⁵ cells per well). It is therefore possible that a dilution effect contributes to reductions in measurable TKT during this period. 

To investigate the potential role of the UPP in TKT regulation, we assessed TKT loss in stromal cell cultures that were cultured in the presence and absence of the cell-permeable proteasome inhibitor, clasto-lactacystin β-lactone. In the absence of proteasome inhibitor, TKT levels dropped approximately 50% between days 1 and 2 (Fig. 7A , compare lanes 1 and 2). Treatment of cells with 5 μM or 10 μM clasto-lactacystin β-lactone significantly increased TKT levels in day-2 cultures (Fig. 7A ; compare lane 2 with lanes 3 and 4). Treatment of day-1 cells with 1 μm clasto-lactacystinβ -lactone did not significantly stabilize TKT in day-2 cells (Fig. 7A ; compare lanes 2 and 5). Moreover, whereas levels of Ub-protein conjugates were stabilized at a threefold or more increase in cells treated with 5 μM or 10 μM lactone, Ub-protein conjugates were stabilized at only a 1.3-fold increase in cells treated with 1 μM lactone (data not shown). Together, these results implicate proteasome activity and degradation of Ub-protein conjugates in the loss of TKT during the repair phenotype transition in vitro (Figs. 6A 6B) . Note also that TKT levels in day-2 cultures that were treated with either 5 or 10 μM lactone were enhanced over levels detected in day-1 cultures (Fig. 7A , compare lane 1 with lanes 3 and 4). This result indicates that proteasome-dependent degradation of TKT occurs in day-1 cultures as well. 

As with many other UPP substrates, TKT immunoblots failed to detect ubiquitinylated (i.e., higher mass) forms of TKT in lactone-treated cells (data not shown), presumably due to the masking of TKT epitopes by polyubiquitinylation. Consequently, these experiments do not rule out the possibility that TKT is degraded by the Ub-independent proteolytic activity of the 20 S proteasome core particle. ¹¹ To determine whether TKT proteolysis is Ub dependent, we exploited the strict requirement for ATP in Ub-dependent proteolysis. We incubated lysate from day-1 stromal cells (a source of TKT) for 2 hours in ATP-depleted and ATP-supplemented reticulocyte lysate (a source of Ub and UPP enzymes). Immunoblots revealed that more than 60% of TKT was degraded in the presence of ATP (Fig. 7B ; compare lanes 1 and 2). TKT levels remained at preincubation levels in the absence of ATP (data not shown), indicating that TKT degradation was almost exclusively ATP dependent. This ATP-dependent proteolysis of TKT was entirely blocked by the proteasome inhibitor MG132 (Fig. 7B ; compare lanes 2 and 3). Thus, TKT degradation by reticulocyte lysate was both ATP and proteasome dependent—two diagnostic hallmarks of Ub-dependent proteolysis. These cell-free data, in conjunction with elevated UPP activities in day-2 and day-3 stromal cell cultures (Figs. 2 3 4) and the stabilization of TKT levels by proteasome inhibition (Fig. 7A) suggest that TKT is degraded by Ub-dependent proteolysis during the repair phenotype transition. 

Discussion

The present study is part of a broader initiative to understand the molecular bases of stromal cell activation in the injured cornea. ¹ In the study, the in vitro transition of quiescent corneal keratocytes to the activated repair fibroblast phenotype was coincident with coordinate upregulation of the UPP. Further, UPP upregulation was associated with dramatic changes in levels of IκBα and TKT, UPP substrates that are mechanistically implicated in the loss of corneal transparency after injury (discussed later). These results suggest that the UPP is a positive regulator of the repair phenotype transition and its sequelae in the injured cornea. To our knowledge, the present study is the first published analysis of UPP function during acquisition of a repair phenotype. 

Our demonstration of enhanced levels of Ub in repair fibroblasts confirms prior reports of injury- and/or disease-induced Ub gene induction in corneal and cutaneous fibroblasts in vivo. ^{27 28 29} The present study extends these earlier reports, in that Ub induction was coincident with enhanced protein ubiquitinylation and Ub-dependent proteolysis, and in that biochemical mechanisms to account for these increased activities were elucidated. Specifically, enhanced protein ubiquitinylation reflected elevated levels of Ub-activating enzymes (E1s) and Ub-conjugating enzymes (E2s), and increased capacity for Ub-dependent proteolysis reflected enhanced expression of 26 S proteasomes. Increased Ub-conjugating and proteolytic activities in repair fibroblasts could also reflect increased activity of Ub-protein ligases (E3s), ^{15 30} reduced rates of conjugate disassembly by deubiquitinylating enzymes, ³¹ or increased substrate availability. 

Coordinate induction of multiple (yet variable) UPP components is observed in response to stress, ^{32 33} during development and differentiation, ^{14 15 16 17 18 19 34 35 36 37} and in certain pathologic states. ^{30 38 39} Although E1 enzyme activity is posttranslationally upregulated during lens epithelial cell recovery from peroxide stress, ⁴⁰ significant upregulation of E1 protein levels appears to be a unique feature of developmental transitions (Annette Baich and Martin Obin, unpublished data). ^{30 40} Similarly, E2-25K is developmentally regulated, ^{14 35} but is not generally induced by stress or disease. ⁴¹ Induction of E1 and E2-25K may therefore define the repair phenotype transition as a developmental progression and distinguish it from a stress response. At present, the molecular mechanisms controlling the coordinate induction of multiple, yet distinct sets of UPP genes remain unelucidated. 

E2-25K is additionally noteworthy because of its substrates. E2-25K ligates Ub_n+1 to Ub_n, thereby catalyzing the synthesis of unanchored polyUb chains. ⁴² These chains are competent intermediates for en masse conjugation to protein substrates. ⁴² E2-25K could therefore promote rapid and global increases in high-mass Ub-protein conjugates in repair fibroblasts by generating preformed polyUb chains for ligation to cell proteins by other E2s and E3s. E2-25K can also ubiquitinylate p105, the precursor of the p50 subunit of NF-κB. ⁴³ Ubiquitinylation targets p105 for proteasome-dependent endoproteolytic cleavage, which generates p50. This observation suggests potential associations among E2-25K induction, enhanced expression of p50, and the acquisition of competence to activate NF-κB by repair fibroblasts ⁵ (discussed later). However, p50 levels did not appear to increase during the repair fibroblast transition, an observation that argues against upregulated E2-25K–dependent processing of p105 in repair fibroblasts. Because Ubcs are typically rate-limiting for ubiquitinylation, ¹⁹ our proteolysis data suggest that Ubcs other than E2-25K are also likely to be upregulated in repair fibroblasts. Specifically, the enhanced capacity for Ub-dependent proteolysis of ¹²⁵I-H2A suggests that repair fibroblasts contain elevated levels of Ubc2 and/or UbcH7, the Ubcs that catalyze H2A polyubiquitinylation in vitro. ⁴⁴  

The transcription factor NF-κB regulates many cell processes that can disrupt corneal structure, including proliferation, apoptosis, and the expression of inflammatory and degradative proteins. ⁴ Consequently, the resistance of corneal stromal cells to NF-κB activation has been proposed as an important mechanism protecting corneal stasis and function. ^{1 5} Our interest in IκBα derives from its role as an inhibitor of NF-κB nuclear translocation and from the essential role of Ub-dependent IκBα degradation in NF-κB activation. ¹³ In the present study , we identified increases in IκBα levels as an early event (day 2) in the repair phenotype transition and sustained increases in IκBα levels as a phenotypic marker of repair fibroblasts (Figs. 5A 5B) . Because the IκBα gene is itself induced by NF-κB, ⁴⁵ this finding suggests that NF-κB activation, and thus IκBα degradation, are constitutively upregulated in repair fibroblasts. Constitutive upregulation of IκBα degradation induces NF-κB-dependent IκBα expression during the differentiation of pre-B cells to mature B cells. ⁴⁶ Constitutive degradation of IκBα entails phosphorylation, putative ubiquitinylation by an as yet unidentified E2-E3 complex, and degradation by the proteasome. ⁴⁷ Therefore, upregulated UPP activities provide a mechanism to explain both the elevated levels of total IκBα and the relatively low levels of phosphorylated IκBα in repair fibroblasts. Upregulated proteolysis of IκBα by cytosolic calpain(s) ⁴⁸ could similarly regulate IκBα expression. 

Acquisition of competence to efficiently activate an NF-κB–dependent IL-1α autocrine loop (p50-p65) has been proposed as a critical step in the loss of corneal stasis after injury. ^{3 5} In previous work we reported the inability of day-1 stromal cell cultures to establish the IL-1α autocrine loop in response to exogenous IL-1α. Moreover, it was suggested that this inability to establish the IL-1α autocrine loop is due to a deficiency in NF-κB activation. ⁵ Based on the downregulated UPP activities in day-1 cells, we hypothesized that this deficiency in NF-κB activation reflects deficiency in ubiquitinylation and/or Ub-dependent proteolysis of IκBα. Deficiency in IκBα ubiquitinylation has recently been shown to constitutively downregulate NF-κB–dependent inflammatory responses of intestinal epithelial cells. ⁴⁹ However, our data demonstrate that day-1 cells rapidly degrade IκBα in response to IL-1α (Fig. 5B) . Quiescent corneal stromal cells therefore possess a functional pathway for IL-α induced IκBα degradation, including the IκB kinase (IKK) and Ubc5/E3RS^{Iκ B} complexes, ¹³ valosin-containing protein, ⁵⁰ the 26 S proteasome, and perhaps calpain. ⁵¹  

Moreover, in data not shown, we demonstrated that day-1 cells efficiently degraded the IκBα, which was rapidly induced by NF-κB activation, thereby reestablishing steady state levels of NF-κB–associated IκBα. Thus, day-1 cells appear competent to proteolytically regulate IκBα in response to IL-1α. Yet, in contrast to repair fibroblasts, day-1 cells failed to synthesize IL-1α when assayed 24 hours after IL-1α–induced NF-κB activation.(Fig. 5C) . Together, these observations indicate that the inability of keratocytes to establish the IL-1α autocrine loop reflects an inability to maintain persistent IL-1α synthesis in response to IL-1α. This conclusion is supported by the observation that levels of IL-1α mRNA are elevated in day-1 cells 2 hours after IL-1α treatment, but are undetectable after 24 hours. ⁵ Future studies in our cell culture model will examine stromal cell mechanisms that inhibit or promote persistent induction of IL-1α in response to IL-1α. 

Developmental transitions require molecular and structural reorganization of cells, and this reorganization is accomplished in part through programmed protein degradation. ^{15 17 19 35} The present study demonstrated for the first time that TKT is lost during the repair phenotype transition in vitro (Figs. 6A 6B 6C) , consistent with whole-animal studies. ⁶ This result further validates the in vitro rabbit model of keratocyte activation. Notably, TKT loss in vitro reflects degradation by the proteasome (Fig. 7A) and, based on cell-free studies of TKT proteolysis (Fig. 7B) , is likely to be Ub dependent. These data identify TKT as a highly regulated proteasome substrate and mechanistically implicate the UPP in TKT loss in the corneal stroma after injury. 

Loss of TKT (and ALDH1) from the injured cornea in vivo is believed to compromise corneal clarity by promoting backscattering of light in migrating keratocytes and myofibroblasts. ^{6 8} In addition to its proposed role as a corneal crystallin, ⁷ the metabolic function of TKT in the pentose–phosphate pathway suggests that TKT loss after wounding may also predispose corneal cells to oxidative damage, further compromising corneal integrity. Noteworthy in this regard is the report that in vivo abrogation of TKT enzyme activity by thiamine depletion leads to lens fiber cell degeneration and cataract, presumably as a consequence of oxidative stress. ⁵² Irrespective of how TKT promotes corneal clarity, our data suggest that TKT levels, and thus corneal transparency, are directly regulated by the UPP, presumably in conjunction with downregulated TKT transcription. 

In summary, we showed that the UPP was coordinately induced during the corneal repair phenotype transition and we have suggested how UPP upregulation alters levels of IκBα and TKT, two UPP substrates that are implicated in the loss of corneal clarity after injury. Elucidation of the molecular mechanisms regulating the coordinate induction of multiple yet distinct sets of UPP genes will enhance our understanding of the corneal wound response and should provide insight into the molecular basis of tissue stasis in the cornea and elsewhere. ⁵³  

 Supported by National Eye Institute Grants EY06722 (JRC), EY11703 (MO), and EY09828 (MEF); United States Department of Agriculture Contract 53-3K06-0-1 (AT); and the Massachusetts Lions Eye Research Fund. MEF is a Jules and Doris Stein Research to Prevent Blindness Professor.

 Submitted for publication November 27, 2000; accepted February 7, 2001.

 Commercial relationships policy: N.

 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.

 

Figure 1.

 

The corneal repair phenotype transition. Keratocytes in the stroma of the uninjured cornea are quiescent, remaining resistant to proliferative and proinflammatory stimuli. In response to corneal wounding, quiescent keratocytes become activated and undergo transition to the repair fibroblast phenotype. The migration, proliferation, and inflammatory responses of these repair fibroblasts compromise corneal clarity. Epi, corneal epithelium; endo, corneal endothelium.

Figure 2.

 

Upregulation of Ub expression and protein ubiquitinylation in repair fibroblasts. (A) Immunoblot of free Ub (8.5 kDa) and Ub-protein conjugates (>130 kDa) in lysates of individual day(d)-1, -2, and -3 stromal cell cultures. Proteins were separated by SDS-PAGE on a 15% gel. Molecular mass markers are at right. Lanes 2 and 3 contain 2.5 and 1.6 times as much protein as lane 1, respectively. (B) Densitometric quantitation of free Ub (shaded bars, left axis) and conjugated Ub (filled bars, right axis) in day-1, -2, and -3 cell culture lysates as in (A), corrected for protein loading. Data (mean ± SEM) are from three experiments performed with individual cell lysates. (C) Immunoblot of conjugated Ub in lysates of day 1 (lanes 1 to 3) and subcultured (lanes 4 to 6) corneal stromal cells. Proteins were separated by SDS-PAGE on a 4% to 20% gel. Molecular mass markers are at right. One of three experiments performed with triplicate cell lysates is shown.

Figure 3.

 

Expression of Ub-activating enzymes (E1A and -B) and the 25 kDa Ub-conjugating enzyme (E2-25K) are upregulated in repair fibroblasts. Immunoblots of lysates of day(d) 1 (lanes 1 to 3) and subcultured (lanes 4 to 6) stromal cells were probed with anti-E1 IgG (top) or with anti-E2-25K serum (bottom). Proteins were separated as in Figure 2C . The E1 doublet (top) contains the E1 isoforms, E1A (117 kDa) and E1B (110 kDa). Molecular mass markers are at right. One of two experiments performed with triplicate cell lysates is shown.

Figure 4.

 

Capacity for Ub-dependent proteolysis is enhanced in repair fibroblasts. (A) Immunoblots of lysates from day(d) 1 (lanes 1 and 2) and subcultured (lanes 3 and 4) stromal cells were probed with antisera against Trip1 and p32, components of the 26 S-proteasome regulatory complex and 20 S core particle, respectively. One of two experiments performed in duplicate is shown. (B) Ub-dependent proteolysis of ¹²⁵I-labeled H2A by cell-free preparations of day-1 and subcultured corneal stromal cells. Protein degradation is quantified from acid-precipitable counts per minute. Ub-dependent proteolysis is calculated as the proportion of ATP-dependent protein degradation that is inhibited by MG132.

Figure 5.

 

Dynamics of IκBα during the repair phenotype transition and in response to IL-1α. (A) Immunoblot of lysates from day(d)-1, -2, and -3 stromal cell cultures was probed with IgG against IκBα. Lysates were prepared in the absence of phosphatase inhibitors. One of two experiments is shown. (B) IκBα immunoblot, demonstrating steady state levels and IL-1α–stimulated proteolysis of IκBα in day-1 (lanes 1 to 4) and subcultured (lanes 5 to 8) stromal cell cultures. Lysates were harvested with phosphatase inhibitors 30 minutes after treatment with 10 ng/ml IL-1α (+) or buffer sham (−). One of three experiments is shown. (C) Day-1 (lanes 1 and 2) and subcultured (lanes 3 and 4) stromal cells were treated with human IL-1α for 24 hours, and secreted IL-1α in the conditioned medium was detected by immunoblot analysis, using antiserum specific for rabbit IL-1α. One of two experiments performed with duplicate sets of conditioned medium is shown. IκBα∼P, phosphorylated IκBα.

Figure 6.

 

Loss of TKT and ALDH1 during the repair phenotype transition. (A) Day(d)-1 cells (lanes 1 and 2) contained abundant ∼70-kDa and ∼50-kDa proteins (arrows), which were absent from subcultured stromal cells (lanes 3 and 4). Cell lysates were separated by 12% SDS-PAGE and stained with Coomassie blue. One of three experiments performed with duplicate or triplicate cell lysates is shown. Molecular mass markers are at right. (B) Antibodies identify the abundant ∼70-kDa and ∼50-kDa proteins as TKT and ALDH1, respectively. Immunoblot of lysates from day-1 and subcultured corneal stromal cells was stained with amido black and then probed sequentially with anti-TKT and anti-ALDH1 antibodies. One of two representative experiments is shown. (C) TKT immunoblot of lysates from day-1, -2, and -3 cell cultures, showing loss of TKT on day 2 (left) or 3 (right) of the repair phenotype transition.

Figure 7.

 

Proteasome- and ATP-Ub-dependent degradation of TKT. (A) TKT immunoblot showing loss of TKT in day-2 cells (compare lanes 1 and 2), and stabilization of TKT levels by clasto-lactacystin β-lactone (lactone). Note that only higher concentrations of lactone (≥5 μM) were effective. Similar results were achieved in duplicate experiments, using only 10 μM proteasome inhibitor. (B) TKT immunoblot demonstrating the requirement of ATP and the proteasome for TKT degradation by reticulocyte lysate. TKT was quantitatively degraded in the presence (+) of ATP (compare lanes 1 and 2) and was stabilized in the presence (+) of the proteasome inhibitor MG132 (compare lanes 2 and 3). One of three experiments is shown.

The authors thank Xin Gong for technical assistance and Katherine Strissel for providing Figure 1 .