Synergism of TNF and IL-1 in the Induction of Matrix Metalloproteinase-3 in Trabecular Meshwork

Abstract

purpose. TNF and IL-1 increase matrix metalloproteinase-3 (MMP-3) expression in the trabecular meshwork (TM). TNF-α, in combination with IL-1α or IL-1β, produces highly synergistic MMP-3 increases. Possible mechanisms for this synergism in TM cells were investigated.

methods. Porcine and human TM cells were treated with TNF-α, IL-1α, IL-1β and their combinations. Western immunoblots were used to evaluate MMP-3, MMP-9, MMP-12, TNF-α, IL-1α, IL-1β, IL-6, TNF receptor I (RI), IL-1 RI, and IL-1 RII levels and the phosphorylation of Erk, JNK, and p38 MAP kinases. Dose–response effects for TNF-α, IL-1α and IL-1β on MMP-3 were evaluated. Microarray and quantitative RT-PCR were used to determine mRNA levels. MMP-3 transcription rate was assessed by transfecting TM cells with an MMP-3 promoter/reporter construct. Combined cytokine effects on outflow facility were appraised in perfused anterior segment organ culture.

results. TNF-α, IL-1α, and IL-1β each individually increased MMP-3 levels, whereas TNF-α in combination with IL-1α or IL-1β produced highly synergistic increases. MMP-9 and MMP-12 levels were also elevated, but only MMP-12 showed synergism. IL-1α, IL-1β, and IL-6, but not TNF-α mRNA or protein level, were elevated by these cytokines. Maximum MMP-3 production for individual cytokines, even at high doses, was far less than with dual cytokine doses. Erk 1 and 2, JNK 1 and 2, and p38 α and β phosphorylation increased, but not synergistically. However, phosphorylation of novel isoforms of JNK and p38 δ and γ did show synergism. MMP-3 mRNA levels and transcription rates also demonstrated synergism. TNF-α significantly increased IL-1 RI levels. Synergism in outflow facility was observed with TNF-α and IL-1α.

conclusions. TNF-α, in combination with IL-1α or IL-1β, produced intense synergistic increases in MMP-3 and MMP-12 but not in MMP-9. Induction of IL-1 RI by TNF-α partially explains the synergism. Responses of novel JNK and p38 MAP kinase δ and γ isoforms also partially account for the synergism. Understanding this strong synergistic effect may provide useful insight into optimizing therapeutic regulation of intraocular pressure in glaucoma.

Glaucoma is a leading cause of irreversible blindness. ^{1 2} A primary risk factor for glaucoma is elevated intraocular pressure (IOP), which can contribute to significant optic nerve damage and vision loss. Laser trabeculoplasty (LTP), a common treatment to reduce IOP in patients with glaucoma, appears to owe its efficacy to increased extracellular matrix (ECM) turnover triggered by matrix metalloproteinase-3 (MMP-3) within the juxtacanalicular region of the trabecular meshwork (TM). ^{3 4} Addition or induction of MMP-3 in perfused human anterior segment organ culture increases aqueous humor outflow facility. ⁵ Blocking the endogenous MMP activity in the TM reduces outflow facility. Consequently, ongoing ECM turnover, initiated by one or more MMPs, appears to be essential to maintaining IOP homeostasis. ⁵ MMP-2 and MMP-14 are important in IOP homeostasis. ⁶ MMP-3, with possible contributions from MMP-9, appears to be critical to ECM turnover, facilitating the therapeutic effects of LTP in ameliorating glaucomatous IOP elevations. ⁴ The MMP-3 increase produced in the TM by LTP is mediated by the cytokines IL-1β and TNF-α. ⁷ Both are secreted in response to LTP, and blocking the action of either dramatically reduces but does not completely eliminate the MMP-3 increase. ⁷ IL-1α also increases dramatically with LTP but is not secreted. ⁷ IL-1α treatment increases outflow facility, and this increase is antagonized by treatment with a tissue inhibitor of metalloproteinase (TIMP). ⁵ Protein kinase C (PKC)μ and the MAP kinases, extracellular-regulated kinase (Erk), c-Jun N-terminal kinase (JNK), and p38 were found to be necessary, but not sufficient, to transduce the increases in MMP-3 produced by TNF-α or IL-1 treatment of TM cells. ^{8 9 10 11 12 13}  

While investigating the participation of the MAP kinases in the signal transduction of TNF- and IL-1 in the TM, we observed that combinations of TNF-α with either IL-1α or IL-1β produced strong synergistic increases in MMP-3 expression. Although synergism between these cytokines had been reported by a number of investigators in several biological systems, little mechanistic information is available to explain these dramatic effects. ^{14 15 16 17 18 19 20} To understand this synergistic effect in the TM, where these cytokines can have therapeutic and pathologic effects, ^{7 21 22} we evaluated the contribution of several possible regulatory systems to this phenomenon. 

Materials and Methods

Porcine eyes were obtained (Carlton Packing Company, Carlton, OR) 2 to 5 hours postmortem, and human eyes were obtained from the Oregon Lions Eye Bank (Portland, OR). Antibodies and other agents used were as follows: recombinant human TNF-α and human and porcine IL-1α, IL-1β, IL-6, and IL-1 RII antibodies (R&D Systems, Minneapolis, MN); MMP-9, JNK 1, IL-1 RI, TNF-α, IL-1α, IL-1β, and p38 β antibodies (Santa Cruz Biotechnology, Santa Cruz, CA); JNK 2, TNF RI, TNF-α, and MMP-12 antibodies (Abcam, Cambridge, MA); MMP-3 and MMP-9 antibodies (TriplePoint Biologics, Forest Grove, OR); phosphospecific Erk (T202/Y204), JNK (T183/Y185), p38 (T180/Y182) antibodies and Erk 1 and 2 protein antibodies (Cell Signaling, Beverley, MA); MMP-9 antibody (Chemicon/Millipore, Billerica, MA); JNK 3, p38 δ, and p38 γ antibodies (Upstate/Millipore, Billerica, MA); p38 α antibody (Calbiochem, San Diego, CA); SB203580 (Biosource, Camarillo, CA); high- and low-glucose Dulbecco modified Eagle medium (DMEM), antibiotics, and antimycotic (Invitrogen-Gibco, Grand Island, NY); fetal bovine serum (FBS; Hyclone, Logan, UT); conjugated secondary antibodies and DNA assay kits (Alexa Fluor 680 and Picogreen, respectively; Molecular Probes, Eugene, OR); conjugated secondary antibodies (IRDye 800; Rockland, Gilbertsville, PA); Escherichia coli cells (One-Shot; Invitrogen); purification protocol (EndoFree Plasmid Maxi Kit; Qiagen, Valencia, CA); pcDNA 3.1 (Invitrogen-Gibco); and reporter vector and chemiluminescence detection kits (SEAP-Basic and Great EscAPe SEAP, respectively; Clontech, Palo Alto, CA). 

Cell Culture, Treatment, and Protein Extraction

Porcine and human TM cells were cultured as previously detailed ^{8 9 23 24 25 26} in medium-glucose (a 1:1 mix of high- and low-glucose) DMEM supplemented with 10% FBS and 1% antibiotic/antimycotic mix. Each separate experiment, designated by n in the figure legends, was conducted with pooled TM cells from 10 to 15 porcine eyes and was used by passage 5. When a range of n values are given, the IL-1β and TNF-α + IL-1β treatment data reflect the lower n value, and the other treatments reflect the higher n value. Confluent cells were serum starved for 48 hours before and during treatment with recombinant human TNF-α (10 ng/mL for human or porcine cells), recombinant human IL-1α (10 ng/mL for human cells), recombinant porcine IL-1α (10 ng/mL for porcine cells), recombinant human IL-1β (10 or 25 ng/mL for human cells), or recombinant porcine IL-1β (10, 25, or 50 ng/mL for porcine cells) for 5, 10, 15, 20, and 30 minutes and 1, 4, 24, 48, and 72 hours, as indicated. Cellular proteins were extracted with a modified radioimmunoprecipitation assay (RIPA) buffer (2 mM EDTA, 1% NP-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 50 mM NaF, 2 mM dithiothreitol, 1 mM sodium orthovanadate, 10 mM NaP₄O₇, 1 mM PMSF, 20 μg/mL leupeptin, 20 μg/mL aprotinin, 20 μg/mL pepstatin, and 50 mM Tris, pH 7.5) on ice, flash frozen in liquid nitrogen, and kept at –80°C until use. Culture media aliquots for MMP analysis were frozen and stored at –20°C. Thawed media aliquots were concentrated 4× with concentration columns (Centricon YM-10; Millipore, Bedford, MA). 

Perfused Anterior Segment Organ Culture

Porcine anterior segment organ culture was conducted essentially as previously described. ^{5 6} Anterior segments were perfused using a constant pressure head of 10 cm until flow stabilized, usually at approximately 24 hours and with flow rates between 2.5 and 5 μL/min, before experiments were initiated. Perfusions were with medium-glucose DMEM with 1% antibiotic/antimycotic mixture and 1% fetal bovine serum. Recombinant human TNF-α (0.5 ng/mL), recombinant porcine IL-1α (0.5 ng/mL), or both at 0.25 ng/mL each were added at approximately 30 hours. Flow rates were measured gravimetrically and were normalized based on pretreatment measurements taken after flow had stabilized. After 72 hours of treatment with cytokines, anterior segments were immersion fixed overnight in 10% neutral-buffered formalin, decalcified, and embedded in paraffin. Slides containing approximately 20 sections, each measuring 5-μm, from each anterior segment were stained with hematoxylin and eosin and were evaluated microscopically to determine the degree of cellularity and structural condition of the TM and Schlemm’s canal regions. Scoring was conducted on coded slides to avoid bias, and anterior segments that scored poor or worse were removed from the flow analysis. 

Western Immunoblots

Cell signal transduction proteins or cytokine receptors were extracted with modified RIPA buffer and subjected to standard SDS-PAGE on 8% or 12% separating gels. ²⁷ Culture media proteins were subjected to similar SDS-PAGE separation. Proteins were then transferred from gels to polyvinylidene difluoride (PVDF) or nitrocellulose membranes and were blocked (Odyssey Blocking Buffer; Li-Cor Biosciences, Lincoln, NE) before probing with the primary antibody. To verify uniform total cell numbers in experiments, extraction efficiency, processing, protein loading and transfer, some blots were stained with Ponceau stain (Sigma-Aldrich, St. Louis, MO) after transfer and before the addition of the blocking agent. Primary antibody detection was performed with conjugated secondary antibodies (Alexa Fluor 680 [Molecular Probes] or IRDye 800 [Rockland]), blots were scanned, and relative band density was determined (Odyssey Infrared Imaging System; Li-Cor Biosciences). 

Promoter Construct, Transfections, and Chemiluminescent SEAP Reporter Assay

A 2.3-kb DNA fragment containing the human MMP-3 promoter (hMMP3p) was amplified from human genomic DNA by PCR and subcloned into MluI/BglII restriction sites upstream of the secreted alkaline phosphatase (SEAP) gene in the reporter vector (SEAP-Basic; Clontech), as previously described. ¹⁰ Correct insertion and sequence of the MMP-3 promoter in the hMMP3p-SEAP construct was confirmed by sequencing. Vectors and constructs were amplified in E. coli cells (One Shot; Invitrogen) and were extracted before transfection (EndoFree Plasmid Maxi Kit; Qiagen). TM cells (500,000 cells/cuvette) were transfected with 2 μg hMMP3p-SEAP or 2 μg control construct (SEAP-Basic, Clontech) using a kit and program (Basic Endothelial Cell Kit and program T-23; Amaxa, Gaithersburg, MD) in a nucleofector (Amaxa). After transfection, cells were seeded in 12-well plates and allowed to recover overnight in 2 mL DMEM supplemented with 10% serum. Cells were then serum starved for 24 hours before and during treatment with recombinant human TNF-α (20 ng/mL) or recombinant porcine IL-1α (10 ng/mL), separately or in combination. Conditioned medium was collected on days 3, 4, 5, and 6 after treatment, and promoter activity was determined using chemiluminescence detection kits (Great EscAPe SEAP; Clontech) according to the manufacturer’s directions. DNA analysis with assay kits (Picogreen; Molecular Probes) was used in some cases after the analysis was completed to verify that the various treatments did not change TM cell numbers. Optimization of transfection efficiency and establishment of transfection uniformity were determined by parallel transfection with a green fluorescence protein (GFP)–pcDNA3 construct. 

Microarrays and Quantitative RT-PCR

Microarrays were conducted and analyzed as previously described. ²⁸ Confluent porcine TM cells were maintained serum free for 48 hours before and during 12-, 24-, and 48-hour treatment with recombinant human TNF-α (10 ng/mL), recombinant porcine IL-1α (10 ng/mL), or control vehicle. Total RNA was extracted and processed for cDNA microarrays. ²⁸ Two separate arrays were used, SMC8400D and SMC8400E (http://www.ohsu.edu/gmsr/smc/index.html), each of which contain approximately 8000 sequence-verified human cDNAs. Complete analysis of these data will be presented elsewhere (Chen Y, et al., manuscript in preparation). Genes of interest to the current study were identified, and the effects of TNF-α or IL-1α were determined. 

Real-time quantitative RT-PCR (qRT-PCR) was conducted to determine relative MMP-3 mRNA levels using methods previously detailed. ²⁸ Separately or in combination, TNF-α (10 ng/mL), IL-1α (10 ng/mL), and IL-1β (25 ng/mL) were used to treat confluent serum-free porcine TM cells for 24 hours. Triplicate analyses were conducted on samples from three separate experiments. Total RNA was extracted, and qRT-PCR was conducted. Primers were selected in separate exons to allow identification of possible genomic contamination. 

Statistical significance comparing groups subjected to different treatments was performed with Student’s t-test or Mann–Whitney U analysis. 

Results

Effects of TNF-α, IL-1α, IL-1β, and Combinations on MMP-3 Levels

Incubation of porcine TM cells with TNF-α, IL-1α, or IL-1β for 24, 48, or 72 hours produced large increases in MMP-3 protein levels in the culture medium (Fig. 1) . Combinations of TNF-α with IL-1α or IL-1β produced increases in MMP-3 that were strongly synergistic rather than simply additive (Fig. 1) . At these different time points, TNF-α with IL-1α increased MMP-3 between 2.7 and 3.6 times more than the sum of their individual effects. TNF-α with IL-1β was between 2.1 and 13.7 times more effective than their sums separately. 

When these studies were conducted using cultured human TM cells, similar but more pronounced synergism was observed (data not shown). For example, in one experiment with human TM cells at 72 hours, TNF-α and IL-1α together were well over 100 times more effective at increasing MMP-3 protein levels than the sum of their separate effects. However, the combination of IL-1α with IL-1β, in either human or porcine TM, produced less than additive effects on MMP-3 levels (data not shown). 

Effects of TNF-α, IL-1α, and Combinations on MMP-9 and MMP-12 Levels

To determine whether the combined cytokine synergy was unique to MMP-3, the responses of other genes were analyzed. Similar Western immunoblot analyses for MMP-9 and MMP-12 showed that both MMP-9 (Fig. 2A)and MMP-12 (Fig. 2B)were stimulated significantly by TNF-α or IL-1α. However, for the combined treatments, the MMP-9 response was less than additive. The MMP-12 response was highly synergistic and produced a nearly sixfold greater response for the combination than their additive responses separately. 

Effects of TNF-α, IL-1α, IL-1β, and Combinations on Perfused Anterior Segment Outflow Facility

Previously, we showed that IL-1α increased outflow facility when added to human perfused anterior segment organ cultures. ⁵ When TNF-α or IL-1α was added to perfused porcine anterior segment organ culture, each increased outflow facility (Fig. 3)This occurred even at very low concentrations (0.5 ng/mL). TNF-α and IL-1α together, each at 0.25 ng/mL, exhibited a synergistic effect on outflow rates (Fig. 3) . At higher concentrations, the magnitude of the response saturates, masking the synergy (data not shown). In postflow microscopic scoring of two experiments, all the anterior segments were rated between good and excellent in terms of cellularity and morphology, except for one of the TNF-α treatments, which was rated poor in terms of cellularity. Flow data for this anterior segment was thus eliminated from analysis (Fig. 3) . 

Effects of TNF-α, IL-1α, IL-1β, and Combinations on TM Cell Cytokine Production

Because cytokines can induce each other or themselves, the effects of individual and combined cytokine treatments on TNF-α, IL-1α, and IL-1β were assessed (Fig. 4) . TNF-α media levels were low without the treatments and were not increased by any of the treatments or combinations (data not shown). IL-1β media levels were low and were not increased appreciably by any of the individual treatments or by TNF-α + IL-1α combined treatment (Fig. 4A) . However, media levels of IL-1β were increased dramatically in response to TNF-α + IL-1β treatment. The horizontal bar shows the amount of IL-1β that was added as treatment, and the amount above the bar was the IL-1β produced in response to this treatment. IL-1α has no signal peptide and is generally not secreted from TM or most other cell types. The level of IL-1α in TM cell lysates, however, was increased in response to cytokine treatment and showed synergism (Fig. 4B) . The size of the latent, inactive form was approximately 31 kDa. Based on other experiments, we estimate that the highest band density for this figure represents roughly 6 ng IL-1α. Microarray analysis showed IL-1α mRNA levels elevated approximately 50-fold, with IL-1β mRNA levels elevated approximately 2-fold at 24 hours after IL-1α treatment (data not shown). TNF-α did not significantly change IL-1α or IL-1β mRNA levels (data not shown). 

Because IL-6 is often induced by, and can mediate, various cytokine effects, its presence in the media was evaluated after these treatments. IL-6 protein levels in the media were increased at 24 and 48 hours. On Western immunoblot, relative IL-6 protein levels were 0.15 ± 0.05, 11.7 ± 7.2, 12.3 ± 4.1, and 29 ± 9.2 for 48-hour control, TNF-α, IL-1α, and TNF-α + IL-1α, respectively. With n = 3, all treatments were significantly different from control by P = 0.0049, P = 0.008, and P = 0.0055, respectively. The microarrays showed IL-6 mRNA levels increased between 6- and 19-fold at 12, 24, and 48 hours in response to IL-1α treatment and increased 1.9-fold at 12 hours by TNF-α treatment (data not shown). However, MMP-3 levels were not significantly changed when IL-6, alone or in combination with these other cytokines, was added to TM cells (data not shown). 

TNF and IL-1 Dose–Response Curves for MMP-3 Production

Because IL-1β media and IL-1α cellular levels were elevated by some of these treatments, dose–response curves were run at 24, 48, and 72 hours to determine whether this cytokine induction could contribute to the synergistic effect (Fig. 5) . Individual dose–response curves, even at 50 ng/mL, never approached the synergistic responses of combinations of TNF-α + IL-1α (solid bars shown to the right of the figures for 10 + 10 ng/mL or for 25 + 25 ng/mL of TNF-α + IL-1α, as indicated in Fig. 5 ). They were much smaller than the combined responses. Dose–response curves for IL-1β were also similarly far below the synergistic responses. 

Effects of TNF-α, IL-1α, IL-1β, and Combinations on TM Cell Cytokine Receptors

Cytokines can also affect receptor levels, as assessed after 24 hours of treatment (Fig. 6) . TNF RI was present at moderate levels and was not changed appreciably by any of the cytokine treatments (Fig. 6A) . IL-1 RI, which is the active signaling receptor for both IL-1α and IL-1β, was significantly increased by TNF-α. It was also increased by IL-1α and the combination of TNF-α and IL-1α, but not by IL-1β alone or in combination with TNF-α (Fig. 6B) . With a Bonferroni correction for multiple comparisons, only the change induced by TNF-α alone achieved clear statistical significance. The effect of IL-1α alone was outside the threshold, and the combined TNF-α and IL-1α effect on IL-1 RI was close to threshold, based on the number of actual comparisons. IL-1 RI levels were only approximately half as high for TNF-α and IL-1β together as they were for TNF-α alone. The IL-1 RII level, which is the nonsignaling decoy receptor for IL-1α and IL-1β, was reduced only by the combination of TNF-α and IL-1α (Fig. 6C) . Although this reduction produced P = 0.04 by t-test, after Bonferroni correction for multiple comparisons, the reduction did not reach the threshold for statistical significance. Levels of these receptors at 48 or 72 hours (data not shown) were not dramatically different from the 24-hour levels shown. Microarray analysis showed no significant change in TNF RI mRNA in response to either TNF-α or IL-1α. IL-1 RI mRNA was increased 2- and 4-fold, respectively, by IL-1α and TNF-α treatment after significance level analysis (data not shown). 

Effects of TNF-α and IL-1α, Separate and Combined, on TM Cell MMP-3 mRNA Transcription and Levels

The effect of treatments with TNF-α and IL-1α, separately or in combination on MMP-3 transcription rate, was assessed by transfecting porcine TM cells with a plasmid construct in which the MMP-3 promoter was positioned to drive expression of a reporter protein, SEAP. Alkaline phosphatase enzymatic activity in the culture medium was then assayed over several days after the various treatments (Fig. 7A) . Control SEAP reporter construct without a promoter (Basic-SEAP) or the active construct with the 2.3-kb MMP-3 promoter positioned to drive SEAP expression (2.3-kb hMMP3 Promoter-SEAP) was transfected into TM cells. Cells were recovered from transfection, made serum free, and were treated with no cytokines or with TNF-α, IL-1α, or both, and media were analyzed for SEAP activity (Fig. 7A) . Separately, TNF-α and IL-1α produced strong increases in MMP-3 promoter activity that increased with treatment time (Fig. 7A) . Together, TNF-α and IL-1α exhibited approximately 1.5-fold more MMP-3 promoter activity than the sum of the separate effects. This is clearly a synergistic effect, but it is not large. 

Analysis of TM cell MMP-3 mRNA levels after 24 hours of treatment with TNF-α, IL-1α, and IL-1β, separately or together, was conducted using qRT-PCR (Fig. 7B) . Separately, these cytokines triggered significant increases in MMP-3 mRNA levels. Microarray analysis also showed significant increases in MMP-3 mRNA in response to TNF-α or IL-1α treatment (data not shown). Together in the qRT-PCR analysis, TNF-α and IL-1α exhibited synergy by increasing transcript levels nearly 3-fold more than additive. By contrast, TNF-α and IL-1β together were slightly less than additive. Results from similar analysis at 12 or 48 hours were not significantly different (data not shown). 

Effects of TNF-α, IL-1α, IL-1β, and Combinations on TM Cell MAP Kinase Activation

Because three major MAP kinase pathways have been shown to be critical signal transducers for TNF- and IL-1 induction of MMP-3 in the TM, the possibility that the cytokine combination effect could relate to differences in MAP kinase activation was evaluated. The phosphorylation of Erk 1 and 2, also called p44 and p42, respectively, was assessed in TM cell extracts by Western immunoblot with phosphospecific antibodies (Fig. 8) . TM cells were treated with the cytokines separately or in combination, as indicated, for 15 minutes (Fig. 8A)or 24 hours (Fig. 8B) , and the two Erk isoforms (p44 and p42) were scanned, quantified separately, and plotted as indicated. Profiles were similar for the two isoforms, though phosphoimmunostaining for Erk 2 (p42) was stronger than for Erk 1 (p44). At 15 minutes, IL-1α stimulated more phosphorylation than TNF-α, whereas the reverse was true at 24 hours. Although the response to the combination of TNF-α with IL-1α was larger than to either alone at either time point and for either isoform, it was less than additive and certainly was not synergistic. 

Similar analysis of JNK 1 (46 kDa) and JNK 2 (54 kDa) MAP kinase isoforms was shown after 15-minute and 24-hour treatments, as indicated (Fig. 9) . Phosphoimmunostaining for JNK 1 was stronger than for JNK 2. Dual cytokine treatments stimulated more phosphorylation than separate treatments, but the increases were far from additive, and in no case were they synergistic. 

Phosphorylation of the 38-kDa band of p38 MAP kinase at 15 minutes (Fig. 10A)and at 24 hours (Fig. 10B)was also similar, with a slightly less than additive effect and no synergistic effects for the combinations. For all three MAP kinase pathways, the effect of IL-1β, alone or in combination, was similar to but less dramatic than that for IL-1α (data not shown). 

Recently, several isoforms and alternative splice variants of p38 have been identified. Phosphospecific p38 immunoblots scanned for Figures 10A and 10Balso have a band of phosphorylation that migrates at approximately 42 kDa. We have recently shown ²⁹ that the 38-kDa band of phosphorylation corresponds to p38 α but that the 42-kDa band of phosphorylation corresponds to p38 δ, p38 γ, or both. The δ and γ isoforms are essentially the same size in the TM. When analyzed separately, phosphorylation of this 42-kDa p38 δ or p38 γ band does show strong synergy (Fig. 10C) . In addition, the phosphospecific JNK immunoblots scanned in Figure 9had an intermediate band of phosphorylation that migrated between JNK 1 and JNK 2. This band showed strong synergistic increases with the cytokine combinations (data not shown). Given that several antibodies that recognized JNK 1, 2, or 3 proteins did not show a band coincident with this phosphorylation, we could not identify this band of phosphorylation. 

Discussion

Much has been written about the synergism among these cytokines in systems other than the TM. ^{14 15 16 17 18 19 20 30 31} Thus far, no results provide a clear mechanism to explain this synergy. Individually, TNF and IL-1 generally produce similar, though not identical, cellular and molecular responses. ³² In TM, both human and porcine, the synergistic effects of TNF-α with either IL-1α or IL-1β on MMP-3 expression are dramatic. The MMP-12 response is similarly synergistic, whereas the MMP-9 response to these cytokines is individually strong but is clearly not synergistic. Based on the MMP-3 promoter study in which TNF-α and IL-1α together were approximately 1.3 times more effective than the sum of their separate effects, MMP-3 transcription rates showed modest synergy. Based on qRT-PCR analysis, which reflected mRNA levels, the effect of TNF-α combined with IL-1α was approximately 2.6-fold larger than responses to them separately. TNF-α combined with IL-1β produced approximately an additive response when compared with the sum of their individual responses. Because mRNA levels reflect a balance between transcription rates and mRNA turnover rates, both would appear to be important components of cytokine synergy. However, because protein level synergy was larger than mRNA level synergy, particularly for IL-1β, protein translation and protein half-life may also be significant contributors to synergy. 

Cytokines can induce each other or themselves in some other systems; thus, cotreatments could conceptually increase the effective total cytokine doses. We did see increases in IL-1α and IL-1β protein and mRNA, but not in TNF-α, under some treatment conditions. However, dose–response curves for the individual cytokines eliminate this as an important component. Cytokine dose responses plateau at high levels. In fact, even at the highest doses of the individual cytokines, MMP-3 levels were far below the synergistic levels produced by the cytokines together. Along the same lines, IL-1α and TNF-α increase IL-6 mRNA and protein levels. IL-6 has been shown to mediate some of the effects of these cytokines in other systems, and the involvement of IL-6 in this process seemed possible. However, the addition of IL-6, either alone or with the other cytokines, did not significantly affect MMP-3 production. IL-6 is, therefore, not important in explaining this synergy. 

Cytokines can also affect cytokine receptor levels. Upregulation and downregulation of cytokine receptors can occur. TNF RI is not affected, either at the protein or at the mRNA level, by any of the cytokine treatments. However, IL-1 RI protein and mRNA levels are increased strongly by TNF-α. mRNA levels were increased moderately by IL-1α. Protein levels were increased, but probably not significantly, by IL-1α and only slightly by IL-1β. Thus, TNF-α should increase TM cellular responses and signal transduction triggered by IL-1α or IL-1β. IL-1 RII is reduced modestly by TNF-α and IL-1α cotreatment. However, given the Bonferroni correction for multiple testing with P = 0.05 as the threshold significance for a single test, this P value of 0.04 was not significant. This receptor is a negative regulator that serves as a decoy and binds IL-1 but does not transmit the signal to the cellular transduction pathways. Thus, this reduction of the type II receptor would increase the levels of effective IL-1, which could interact with the active receptor, IL-1 RI. However, IL-1α and IL-1β in combination with each other do not show synergy. Furthermore, the simultaneous combination of all three cytokines, i.e., TNF-α, IL-1α, and IL-1β, is not more effective than the double combinations (data not shown). Because IL-1β has a much higher affinity than IL-1α for the decoy receptor IL-1 RII, increasing IL-1β should titrate IL-1 RII and allow more IL-1α to interact with the active IL-1 RI. The two cytokines together, in the presence or the absence of TNF-α, do not show increases in MMP-3 production; hence, IL-1 RII must not play a particularly important role in the individual or the combined processes. Thus, the change in IL-1 RI levels in response to TNF-α, but probably not the other receptor changes, appears to make important mechanistic contributions to this synergistic MMP-3 response. 

Another possible mechanism to generate synergism would be for the cytokines to act through different signal transduction pathways, which converge to produce the MMP-3 increases. We and others ^{8 10 13} have shown that three MAP kinase pathways make necessary but not individually sufficient contributions to the TNF and IL-1 signal transduction to MMP-3 production. ²⁹ When the phosphorylation levels of Erk 1 and 2, JNK 1 and 2, and p38 α and β were compared for the cytokines separately and in combination, no synergism was observed. In fact, except for JNK 2 at 24 hours, none of the cytokine combinations produced even additive changes in these phosphorylation levels. 

Phosphorylation levels of the 42-kDa p38 δ and γ isoforms and of a putative novel JNK isoform, however, showed strong synergism. We were unable to find a JNK antibody that also recognized the novel phospho-JNK gel band; therefore, its identity as a JNK isoform remains conjectural. However, we have shown that in TM, p38 δ and γ are present and migrate at approximately 42 kDa, which is coincident with this phosphospecific p38 band. ²⁹ Thus, we can conclude that p38 δ and γ are critically involved in the mechanism of this synergistic MMP-3 response. TNF-α and IL-1α show some divergence in their usage of p38 δ/γ, ²⁹ but nuances of this regulatory system remain to be elucidated. 

At this time, we do not know which proteins are found downstream of p38 δ/γ in this synergistic signaling cascade. We have shown (Song K, et al. IOVS 2005;46:ARVO E-Abstract 1356) that at least four enhancer/suppressor elements exist in the MMP-3 promoter and are involved in TNF-α and IL-1α induction of MMP-3 transcription. JNK phosphorylates c-Jun on S63 and S73 and ATF-2 on T71 in the TM in response to TNF-α or IL-1α. ¹⁰ Additional specific transcriptional activator proteins involved in activating the MMP-3 promoter in the TM remain to be identified. The other specific molecular sites of action of these MAP kinase pathways in increasing MMP-3 levels are also unknown. PKCμ also plays an important but unidentified role in this signaling. ^{9 13}  

Given that both IL-1 and TNF are elevated by laser trabeculoplasty (LTP) and that blocking either of these only partially blocks the MMP-3 increase involved, this synergy may well have real therapeutic significance. ⁷ In addition, the synergy extends to changes in the facility of perfused anterior segment outflow, implying that the efficacy of LTP in restoring aqueous humor outflow facility in glaucoma may be attributed, at least in part, to the synergistic interaction of these cytokines to induce MMPs. 

 Supported by National Institutes of Health Grants EY003279, EY008247, and EY010572 and by grants from Research to Prevent Blindness (New York, NY) and Alcon Labs (Fort Worth, TX).

 Submitted for publication December 6, 2006; revised January 29, 2007; accepted March 23, 2007.

 Disclosure: M.J. Kelley, None; A.Y. Rose, None; K. Song, None; Y. Chen, None; J.M. Bradley, None; D. Rookhuizen, None; T.S. Acott, Alcon Labs (F)

 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.

 

Effects of TNF-α, IL-1α, IL-1β, and their combinations on MMP-3 production by TM cells. Media levels of MMP-3 were measured by Western immunoblot after treatment of porcine TM cells for (A) 24 hours, (B) 48 hours, or (C) 72 hours with TNF-α (10 ng/mL), IL-1α (10 ng/mL), IL-1β (10 ng/mL) or their combinations, as indicated. Mean ± SD is shown, with t-test significance and sample number as indicated.

Figure 2.

 

Effects of TNF-α, IL-1α, and their combination on MMP-9 and MMP-12 levels. (A) Media levels of MMP-9 after treatment of porcine TM cells, as indicated for 48 or 72 hours, were determined by Western immunoblot. Mean ± SD is shown, with t-test significance and sample number as indicated. (B) Media levels of MMP-12 were determined by Western immunoblot after 48 hours of treatments with the cytokines, as indicated. Mean ± SD is shown, with t-test significance and sample sizes as indicated.

Figure 3.

 

Effects of TNF-α, IL-1α, and combination on perfused anterior segment outflow. Porcine anterior segments were perfused until flow stabilized and cytokines were added at the indicated doses, as marked by the arrow. Mean normalized flow rates and SEM are shown for three separate experiments.

Figure 4.

 

Effects of TNF-α, IL-1α, IL-1β, and combinations on IL-1β and IL-1α levels. Western immunoblots of (A) TM cell media IL-1β levels or (B) TM cell lysate IL-1α levels after 48-hour treatment with the indicated cytokines or combinations. Means, SD, sample number, and t-test significance are shown. Bars indicate the pairs compared with the related P values shown over the bars. (A) Horizontal line: the amount of measured IL-1β that was added as the treatment. Above the line is the amount that was produced in response to the treatment. IL-1β in the media (A) was the 17-kDa activated form, whereas IL-1α in the cell lysate was the 31-kDa latent form.

Figure 5.

 

Dose–response curves for MMP-3 levels after TNF-α or IL-1α, separately or in combination. (A) 24-hour, (B) 48-hour, and (C) 72-hour MMP-3 media levels assessed by Western immunoblot after TNF-α and IL-1α treatments. Curves (left): the effects of individual doses of the cytokines. Solid line: TNF-α responses; dashed line: IL-1α responses. Bars (right): MMP-3 levels for combinations of TNF-α and IL-1α at the doses (ng/mL) indicated below the respective bars.

Figure 6.

 

Effects of TNF-α, IL-1α, IL-1β, and combinations on TNF and IL-1 receptor levels. Western immunoblots for TNF RI (A), IL-1 RI (B), and IL-1 RII (C) of TM cell extracts after 24-hour treatment, as indicated. Significance from t-test comparisons with the untreated control is shown above the bars, and sample size is as indicated.

Figure 7.

 

Effects of TNF-α and IL-1α, separately and in combination, on MMP-3 mRNA transcription rates and mRNA levels. (A) MMP-3 transcription rates were assessed after treatment without (–) or with (+) TNF-α, IL-1α, or their combination, as indicated, by using the MMP-3 promoter to drive a SEAP reporter. Cells were transfected with the SEAP reporter vector without a promoter (Basic-SEAP) or with the human MMP-3 promoter (2.3-kb hMMP-3 Promoter-SEAP) inserted to activate expression of SEAP by TM cells. MMP-3 promoter activity was determined by measuring SEAP enzymatic activity secreted into the media at the times indicated. Mean ± SEM for triplicate determinations from three separate experiments are shown (n = 9) and significance, evaluated by t-test, is as indicated. (B) TM cell MMP-3 mRNA levels were determined by quantitative RT-PCR at 24 hours after treatment, as indicated, using TNF-α (10 ng/mL), IL-1α (10 ng/mL), and IL-1β (25 ng/mL). Mean ± SEM for triplicate determinations from three separate experiments are shown (n = 9). Significance, evaluated by paired t-test, is as indicated.

Figure 8.

 

Effects of TNF-α, IL-1α, IL-1β, and combinations on Erk MAP kinase phosphorylation levels. Western immunoblots of TM cell extracts after treatments, as indicated for 15 minutes (A) or 24 hours (B), were probed with Erk phosphospecific antibodies (T202/Y204). The two bands, 42-kDa Erk 2 (solid bars) and 44-kDa Erk 1 (hatched bars), were scanned and quantified separately. Mean relative band density and SEM are plotted.

Figure 9.

 

Effects of TNF-α, IL-1α, and combinations on JNK MAP kinase phosphorylation levels. Western immunoblots of TM cell extracts after treatment, as indicated, for 15 minutes (A) or 24 hours (B) were probed with JNK phosphospecific antibodies (T183/Y185). The two bands, 54-kDa JNK 2 (solid bars) and the 46-kDa JNK 1 (hatched bars), were scanned and quantified separately. Mean relative band density and SEM are plotted.

Figure 10.

 

Effects of TNF-α, IL-1α, and combination on p38 MAP kinase phosphorylation levels. Relative band density from scans of Western immunoblots for 38-kDa band of p38 α MAP kinase phosphorylation after 15-minute (A) or 24-hour (B) treatment of TM cells, as indicated. (C) Phosphorylation levels of 42-kDa band of p38 δ/γ MAP kinase after 24-hour treatment, as indicated. Mean relative band density and SEM are shown (n = 4).

The authors thank Genevieve Long, PhD, for editorial assistance.