Inhibition of Gap Junction–Mediated Intercellular Communication by TNF-α in Cultured Human Corneal Fibroblasts

Abstract

purpose. Keratocytes are connected to each other by gap junctions, which mediate intercellular communication and contribute to maintenance of corneal homeostasis. The possible effect of tumor necrosis factor (TNF)-α, a proinflammatory cytokine, on gap junctional intercellular communication (GJIC) in cultured human corneal fibroblasts was examined.

methods. GJIC activity was measured by observing the intercellular diffusion of the fluorescent dye Lucifer yellow. The expression of the gap junction protein connexin43 (Cx43) was evaluated by immunofluorescence and immunoblot analyses with a specific monoclonal antibody. The abundance of Cx43 mRNA was determined by quantitative reverse transcription and polymerase chain reaction analysis.

results. TNF-α induced a time- and concentration-dependent decrease in GJIC activity in human corneal fibroblasts. Immunofluorescence analysis revealed that TNF-α reduced the level of specific staining for Cx43 at sites of contact between adjacent cells. Immunoblot analysis detected four specific Cx43 bands, one corresponding to the nonphosphorylated form of the protein and three corresponding to phosphorylated forms. Exposure of cells to TNF-α reduced the relative abundance of the three phosphorylated forms of Cx43. The amount of Cx43 mRNA was not affected by TNF-α.

conclusions. TNF-α inhibited GJIC in cultured human corneal fibroblasts, an effect that was possibly mediated by dephosphorylation and consequent degradation of Cx43. The downregulation of GJIC among keratocytes in response to TNF-α may contribute to the breakdown of corneal homeostasis during corneal inflammation.

The corneal stroma consists of keratocytes, extracellular matrix, and nerve fibers. Both transmission and scanning electron microscopy have revealed that keratocytes are connected to each other by gap junctions and form a three-dimensional cellular network. ^{1 2} Gap junctions mediate the intercellular diffusion of ions or molecules, such as cAMP, with a size of <1 kDa, and thereby contribute to the regulation of tissue differentiation and the maintenance of homeostasis. ³ A dye-coupling technique has revealed gap junctional intercellular communication (GJIC) to be operative among keratocytes in both normal and wounded rabbit corneas and in the human cornea ex vivo, ⁴ as well as among rabbit corneal fibroblasts and myofibroblasts in vitro. ⁵ Gap junctions are formed by a pair of connexons, each containing six connexin molecules, that are situated in apposing membranes of the connected cells. At least 18 mammalian members of the connexin protein family have been identified to date by cDNA cloning, and different cell types express different combinations of these proteins. ⁶ Gap junctions in corneal fibroblasts have been found to contain connexin (Cx)43, but not Cx26 or -32. ^{5 7}  

Inflammation is a host response to various insults, including injury and infection, and is mediated by both resident cells of a tissue and infiltrating cells. Both types of cells secrete various cytokines and growth factors and thereby communicate with each other by autocrine and paracrine mechanisms. Tumor necrosis factor (TNF)-α is a proinflammatory cytokine ⁸ that contributes to ocular inflammation. ⁹ The cornea responds to TNF-α under pathologic conditions such as injury, ^{10 11} allergy, ^{12 13} and infection. ¹⁴ TNF-α induces the expression in corneal fibroblasts of granulocyte-macrophage colony-stimulating factor ¹⁵ as well as of the chemokines eotaxin ¹² and thymus- and activation-regulated chemokine, ¹⁶ which promote the infiltration of immune cells such as eosinophils and T-helper-2 cells. TNF-α thus plays an important role in the pathogenesis of inflammation in the corneal stroma. 

To provide insight into the mechanisms of corneal stromal inflammation, we investigated the effect of TNF-α on GJIC among corneal fibroblasts. We determined GJIC activity in cultured human corneal fibroblasts by measuring the intercellular transfer of the fluorescent dye Lucifer yellow. Furthermore, we examined the effect of TNF-α on Cx43 expression at both the protein level, with immunofluorescence and immunoblot analyses, and the mRNA level, with quantitative reverse transcription–polymerase chain reaction (RT-PCR) analysis. 

Materials and Methods

Eagle’s minimum essential medium (MEM), fetal bovine serum (FBS), and trypsin-EDTA were obtained from Invitrogen-Life Technologies (Rockville, MD); bovine serum albumin (BSA) and fraction V from Nacalai Tesuque (Kyoto, Japan); Lucifer yellow CH (Li⁺ salt) from Aldrich (Dreieich, Germany); and recombinant human TNF-α from Genzyme (Cambridge, MA). Mouse monoclonal antibodies to rat Cx26, -32, or -43 or to α-tubulin were obtained from Chemicon (Temecula, CA); fluorescein isothiocyanate (FITC)–conjugated goat secondary antibodies from ICN (Aurora, OH); horseradish peroxidase–conjugated goat secondary antibodies were from Promega (Madison, WI); rhodamine-phalloidin from Molecular Probes (Eugene, OR); and mouse immunoglobulin G (IgG) from Cedarlane (Hornby, Ontario, Canada). Antibodies to vimentin, cytokeratin, and α-smooth muscle actin (α-SMA) were from Dakocytomation (Glostrup, Denmark); a kit (CytoTox96) for measurement of lactate dehydrogenase (LDH) activity was obtained from Promega; and protease inhibitor cocktail was from Sigma-Aldrich (St. Louis, MO). 

Cells and Cell Culture

A human cornea was obtained from Northwest Lions Eye Bank (Seattle, WA). The donor was a 58-year-old white woman. After the center of the donor cornea was punched out for corneal transplantation surgery, the remaining rim of tissue was used for the present experiments in strict accordance with the tenets of the Declaration of Helsinki. The corneal tissue was digested with bacterial collagenase, and corneal fibroblasts were prepared as previously described. ¹² The cells were cultured in MEM supplemented with 10% FBS and were used for experiments after four to seven passages. As previously described, ¹² the purity of the corneal fibroblast cultures was judged on the basis of both cell morphology and immunoreactivity with antibodies to vimentin (non–epithelial cell marker), to cytokeratin (epithelial cell marker), and to α-SMA (myofibroblast marker). All the cells used in the present study were positive for vimentin and negative for both cytokeratin and α-SMA (Fig. 1) , suggesting that the cultures were not contaminated by epithelial cells. For experiments, cells were first cultured for 3 days in MEM supplemented with 10% FBS. The culture medium was then changed to serum-free MEM for 1 day. The medium was replaced with MEM in the absence and presence of various concentrations of TNF-α (0.1–10 ng/mL), and the cells were incubated at 37°C for the indicated times before being subjected to the various analyses. 

Dye Coupling

Cells (3 × 10⁴) were seeded in four-well chamber slides and GJIC was monitored with Lucifer yellow as described previously. ^{17 18} In brief, after exposure to TNF-α, cells were washed with Ca²⁺-free phosphate-buffered saline (PBS(−)), and Lucifer yellow (10% in 0.33 M LiCl) was injected into two widely separated cells per slide with the use of a microinjector (Micromanipulator and Transjector; Eppendorf, Hamburg, Germany). The cells were observed with a fluorescence inverted microscope (Axioscope; Carl Zeiss Meditec, Hallbergmoos, Germany) and photographed 1 minute after dye injection. The number of cells containing the dye was counted. GJIC activity was expressed as the mean number of cells coupled to each injected cell. 

Immunofluorescence Microscopy

Cells cultured on plastic dishes were fixed with acetone for 20 minutes at −20°C, washed with PBS(−), and incubated with 1% BSA in PBS(−) for 1 hour at room temperature. They were then incubated for 1 hour with monoclonal antibodies to Cx26, -32, -43, vimentin, cytokeratin, or α-SMA (each at a 1:200 dilution in PBS(–) containing 1% BSA), washed with PBS(−), and incubated for 1 hour with FITC-conjugated goat antibodies to mouse IgG (1:500 dilution in PBS(−) containing 1% BSA). For staining of F-actin, the cells were also incubated for 30 minutes at room temperature with rhodamine-phalloidin (1:200 dilution in PBS(−) containing 1% BSA). Cells were examined with a microscope and photographed with a digital camera (Axioscope and Axiocam; Carl Zeiss Meditec). 

Immunoblot Analysis

The abundance and phosphorylation state of Cx43 in corneal fibroblasts were examined by SDS–polyacrylamide gel electrophoresis (PAGE) and immunoblot analysis. ^{18 19} Cells cultured in 6-cm dishes were lysed in 0.1 mL of lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1 mM EDTA, 5 mM NaF, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM Na₃VO₄, and 1% protease inhibitor cocktail). The lysate was centrifuged at 15,000g for 5 minutes, and the resultant supernatant (10 μg of protein) was subjected to SDS-PAGE on a 12.5% gel. The separated proteins were transferred to a polyvinylidene difluoride membrane, which was then exposed to 5% skimmed milk for 1 hour at room temperature before incubation for 1 hour with monoclonal antibodies to Cx43 or to α-tubulin (each at a 1:1000 dilution) in buffer (20 mM Tris-HCl [pH 7.4], 5% skimmed milk, 0.1% Tween 20). The membrane was washed in washing buffer, incubated for 1 hour at room temperature with horseradish peroxidase–conjugated goat antibodies to mouse IgG (1:2000 dilution in washing buffer), washed again, incubated with chemiluminescence detection reagents (ECL Plus; Amersham, Buckinghamshire, UK) for 5 minutes, and then exposed to film. The relative amounts of nonphosphorylated and phosphorylated Cx43 were determined with a densitometer (Arcus II; PDI, New York, NY) equipped with analysis software (Quantity One; PDI). 

Alkaline Phosphatase Treatment

For determination of whether a shift in electrophoretic mobility of the Cx43 band on SDS-PAGE gels was related to phosphorylation, cells cultured in 6-cm dishes were lysed in lysis buffer without Na₃VO₄ (an inhibitor of alkaline phosphatase) as described for immunoblot analysis. After centrifugation, cell lysates were incubated for 1 hour at 37°C in the absence or presence of alkaline phosphatase (25 IU/mL) or 10 mM Na₃VO₄. They were then subjected to immunoblot analysis as just described. 

Quantitative RT-PCR Analysis

Total RNA was isolated from corneal fibroblasts and subjected to RT-PCR analysis as previously described. ¹⁶ The sequences of the PCR primers were as follows: Cx43 sense, 5′-CTCGCCTATGTCTCCTCCTG-3′; Cx43 antisense, 5′-GCTGGTCCACAATGGCTAGT-3′; glyceraldehyde-3-phosphate dehydrogenase (GAPDH) sense, 5′-TGAACGGGAAGCTCACTGG-3′; and GAPDH antisense, 5′-TCCACCACCCTGTTGCTGTA-3′. 

The PCR protocol comprised denaturation at 94°C for 15 seconds, annealing at 58°C for 20 seconds, elongation at 72°C for 13 or 15 seconds for the amplification of Cx43 and GAPDH cDNAs, respectively. Real-time PCR data were then analyzed (Light Cycler software 3.01; Roche Molecular Biochemicals, Mannheim, Germany). 

Statistical Analysis

Quantitative data are presented as the mean ± SD. Differences were analyzed by the Dunnett or Scheffé test for comparisons among multiple groups or between two groups, respectively. P < 0.05 was considered statistically significant. 

Results

The effect of TNF-α on GJIC in cultured human corneal fibroblasts was investigated by measurement of dye coupling. Injection of Lucifer yellow into a single cell resulted in transfer of the dye into neighboring cells (Fig. 2A) , demonstrating the presence of functional gap junctions connecting the cultured cells. Incubation of the corneal fibroblasts with various concentrations of TNF-α for 24 hours resulted in a concentration-dependent inhibition of GJIC. The inhibitory effect of TNF-α on GJIC activity was significant at 0.1 ng/mL and maximum at 10 ng/mL (Fig. 2) . Characterization of the time course of this effect of TNF-α revealed that, at a concentration of 1.0 ng/mL, the cytokine had little effect on GJIC activity at 6 or 12 hours; the effect was significant at 24 hours, however (Fig. 3) . 

The possibility that TNF-α exerted a cytotoxic effect in the fibroblast cultures was examined by measurement of the activity of LDH released into the culture medium. Incubation of the cells for 24 hours with TNF-α at 1.0 or 10 ng/mL had no significant effect on the amount of LDH activity in the culture medium (Fig. 4) . TNF-α thus did not appear to exhibit cytotoxicity in our experimental system. 

Several connexins have been shown to be expressed in the cornea. ²⁰ The expression of connexin isoforms in our cultures of human corneal fibroblasts was examined by immunofluorescence microscopic analysis with antibodies to either Cx43, -32, or -26. Punctate staining for Cx43 was detected in the region of the cell membrane at sites of cell–cell contact, but no specific fluorescence was observed for Cx26 or -32 (Fig. 5) . Cx43 thus appears to be an important component of gap junctions in human corneal fibroblasts. The extent of punctate staining for Cx43 was markedly decreased after incubation of corneal fibroblasts with TNF-α (10 ng/mL) for 24 hours (Fig. 6) . 

Phosphorylation of connexins contributes to regulation of their function and is mediated by several kinases. ^{21 22} The phosphorylation state of Cx43 in corneal fibroblasts was therefore examined. Cell lysates were incubated in the absence or presence of alkaline phosphatase or its inhibitor Na₃VO₄ before immunoblot analysis with antibodies to Cx43. Four specific signals for Cx43 corresponding to molecular sizes of 43, 47, 48, and 49 kDa were detected in untreated cell lysates (Fig. 7A) . The intensity of the three bands corresponding to molecular sizes of 47, 48, and 49 kDa was reduced, whereas that of the 43-kDa band was increased, after treatment of the cell lysates with alkaline phosphatase. These effects of alkaline phosphatase were prevented by the simultaneous presence of Na₃VO₄. These results show that the 43-kDa band corresponded to nonphosphorylated Cx43 (Cx43-NP) and that the three bands at 47, 48, and 49 kDa corresponded to phosphorylated Cx43 (Cx43-P₁, Cx43-P₂, and Cx43-P₃, respectively). The low intensity of the Cx43-P₃ band in untreated lysates was attributable to the instability of this phosphorylated form of Cx43 during the 1-hour incubation at 37°C in this experiment. 

The intensities of the Cx43-P₁, -P₂, and -P₃ bands were decreased significantly by treatment of the cells with TNF-α (10 ng/mL) for 24 hours (Figs. 7B 7C) . Given that the intensity of the Cx43-NP band was not increased by TNF-α treatment, these results suggest that TNF-α affects not only the phosphorylation of Cx43 but also its abundance, consistent with our immunofluorescence data (Fig. 6) . Indeed, gap junctions are dynamic structures, and many connexin isoforms have been found to turn over rapidly, ^{23 24} which may be essential in the regulation of GJIC. 

Finally, the effect of TNF-α on the abundance of Cx43 mRNA in human corneal fibroblasts was examined by RT-PCR analysis. Incubation of cells with TNF-α (10 ng/mL) for 6, 12, or 24 hours had no significant effect on the amount of Cx43 mRNA (Fig. 8) , suggesting that the downregulation of Cx43 by TNF-α is mediated at the posttranscriptional level. 

Discussion

In the current study, Cx43 was localized to the junctional regions of neighboring cells and probably contributed to the formation of gap junctions in human corneal fibroblasts cultured with serum. Furthermore, we found that TNF-α inhibited GJIC activity in these cells and that this effect appeared to be mediated by downregulation of Cx43 phosphorylation and expression. 

Connexons assemble or disassemble in response to various stimuli that regulate GJIC activity. ²⁴ Various connexins have been shown to be expressed in the cornea by immunohistochemical or quantitative RT-PCR analyses. ²⁰ Cx26 has thus been found to localize to basal cells of the corneal epithelium and Cx43 to the epithelium and stromal keratocytes in the rat cornea. With the use of immunofluorescence analysis, we have now shown that Cx43, but not Cx26 or -32, is expressed in cultured human corneal fibroblasts. These results suggest that the cell type–specific expression of connexin isoforms may subserve specific functions related to intercellular communication. 

Several cytokines are secreted by macrophages and leukocytes during inflammation. One such proinflammatory cytokine is TNF-α, which has also been detected at the base of corneal ulcerations. ²⁵ With the use of a dye-coupling assay, we observed that TNF-α inhibited the function of gap junctions in human corneal fibroblasts. A similar effect of this cytokine has been demonstrated in other cell types, including human umbilical vein endothelial cells, ^{26 27} mouse hepatocytes, ²⁸ human smooth muscle cells, ²⁹ rat Schwann cells, ³⁰ rat cardiac muscle cells, ³¹ and human bronchial Beas2B cells. ³² Although corneal fibroblasts cultured in medium containing serum (as in the present study) differ in certain respects from primary keratocytes cultured in serum-free medium as well as from stromal keratocytes in vivo, ^{33 34} our results suggest that TNF-α likely modulates GJIC among keratocytes during corneal inflammation. 

Connexins are phosphorylated by various kinases, and their phosphorylation level is a determinant of their activity or stability. ^{35 36} We detected three phosphorylated forms of Cx43 in human corneal fibroblasts, and the abundance of these species was reduced by exposure of the cells to TNF-α. Phosphorylation of specific serine residues of Cx43 and -45 has been shown to affect protein stability. ^{37 38} Phosphorylation of Cx43 on Ser²⁵⁵ by the kinase p34cdc2 in Rat1 cells during the G₂–M phase of the cell cycle thus promotes its endocytosis and degradation. ³⁸ In contrast, phosphorylation of serine residues in the carboxyl-terminal region of Cx45 results in marked stabilization of the protein in transfected HeLa cells. ³⁷ The TNF-α–induced decrease in the proportion of phosphorylated forms of Cx43 in corneal fibroblasts observed in the present study may thus contribute to the downregulation of Cx43 elicited by this cytokine. We previously showed that the downregulation of GJIC activity in trabecular meshwork cells induced by phorbol 12-myristate 13-acetate, an activator of protein kinase C, was associated with Cx43 phosphorylation. ¹⁸ This difference in the effects of phosphorylation of Cx43 on its stability between these two cell types is possibly attributable to differences in the residues targeted by the respective extracellular signals. 

The TNF-α–induced downregulation of the expression of Cx43 in human corneal fibroblasts demonstrated by both immunofluorescence and immunoblot analyses was not accompanied by an effect on the amount of Cx43 mRNA, as revealed by quantitative RT-PCR analysis. The regulation of Cx43 abundance by TNF-α in these cells thus appears to be mediated at a posttranscriptional level, consistent with the notion that this cytokine promotes the degradation of Cx43 through an effect on its phosphorylation state. The half-lives of most connexin family proteins have been determined to be between 1 and 5 hours. ^{39 40} The abundance of Cx43 in apposed regions of neighboring cells has been found to be increased by treatment with proteasome inhibitors, such as acetyl-leucyl-leucyl-norleucinal or MG132, ⁴¹ or with inhibitors of lysosome function, such as primaquine or chloroquine. ⁴² It is thus possible that the loss of Cx43 from junctional regions of human corneal fibroblasts induced by TNF-α is attributable to proteasome- or lysosome-mediated degradation. 

Stimulation of corneal fibroblasts with TNF-α induces the expression of adhesion molecules such as intercellular adhesion molecule (ICAM)-1 ¹³ and chemokines such as eotaxin ¹² as well as triggers the release of matrix metalloproteinases such as MMP-2 and -9, ⁴³ which degrade collagen in the corneal stroma. We have now shown that TNF-α also downregulates GJIC activity in these cells. All these effects are likely to contribute to the promotion of corneal inflammation by TNF-α. Further characterization of the mechanisms underlying the regulation of GJIC activity in the corneal stroma may provide the basis for possible therapies to prevent or ameliorate corneal inflammation associated with infection or injury. 

 Supported in part by Grant A14207070 from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. J-LH is a recipient of a China-Japan Scholarship from the government of China.

 Submitted for publication July 17, 2004; revised November 21, 2004; accepted December 9, 2004.

 Disclosure: J.-L. Hao, None; K. Suzuki, None; Y. Lu, None; S. Hirano, None; K. Fukuda, None; N. Kumagai, None; K. Kimura, None; T. Nishida, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Immunofluorescence staining of cultured human corneal fibroblasts with antibodies to vimentin, cytokeratin, or α-SMA or with control mouse IgG. Cell nuclei were stained with propidium iodide. Scale bar, 100 μm.

Figure 2.

 

Inhibitory effect of TNF-α on GJIC in cultured human corneal fibroblasts. (A) Cells were incubated for 24 hours in the absence or presence of TNF-α at concentrations of 0.1, 1.0, or 10 ng/mL, after which an individual cell ( ) was microinjected with Lucifer yellow. After 1 minute, the cells were examined by phase-contrast (left) or fluorescence (right) microscopy. Scale bar, 50 μm. (B) GJIC activity in experiments similar to that shown in (A) was quantitated by counting the number of cells to which the dye spread from the injected cell. Data are the mean ± SD of results obtained from 10 injected cells in a single experiment. *P < 0.05, **P < 0.01 (Dunnett test) versus the corresponding value for cells incubated in the absence of TNF-α. Data are representative of results in three independent experiments.

Figure 3.

 

Time course of the inhibitory effect of TNF-α on GJIC activity in cultured human corneal fibroblasts. Cells were incubated in the absence (○) or presence (•) of TNF-α (1.0 ng/mL) for the indicated times, after which an individual cell was microinjected with Lucifer yellow. The dye was allowed to spread for 1 minute and the number of coupled cells was then determined. Data are the mean ± SD of results obtained from 10 injected cells in a single experiment. *P < 0.01 (Dunnett test) versus the corresponding time point for cells incubated in the absence of TNF-α. Data are representative of three independent experiments.

Figure 4.

 

Lack of TNF-α cytotoxicity in cultured human corneal fibroblasts. Cells were incubated for 24 hours in the absence or presence of TNF-α (1.0 or 10 ng/mL), after which the activity of LDH released into the culture medium was determined. Data are normalized by the amount of activity released from cells after exposure to a cell lysis solution and are the mean ± SD of triplicate results from a single experiment. There was no statistically significant difference between cultures incubated with or without TNF-α. Data are representative of three independent experiments.

Figure 5.

 

Expression of Cx43 in cultured human corneal fibroblasts. Cells were stained with antibodies to Cx43, -26, or -32 or with mouse IgG as a negative control. Immune complexes were detected with FITC-conjugated secondary antibodies (green). Cell shape was visualized by staining for F-actin with rhodamine-phalloidin (red). Data are representative of results in three independent experiments. Scale bar, 100 μm.

Figure 6.

 

Inhibitory effect of TNF-α on the expression of Cx43 in cultured human corneal fibroblasts. Cells were incubated for 24 hours in the absence or presence of TNF-α (10 ng/mL), after which Cx43 was detected by indirect immunofluorescence staining (green), and actin filaments were visualized with rhodamine-phalloidin (red). Data are representative of results of three independent experiments. Scale bar, 50 μm.

Figure 7.

 

Immunoblot analysis of the effects of TNF-α on the phosphorylation and abundance of Cx43 in cultured human corneal fibroblasts. (A) Cell lysates were incubated for 1 hour at 37°C in the absence or presence of alkaline phosphatase (25 IU/mL) (ALP) or 10 mM Na₃VO₄, as indicated. The lysates were then subjected to immunoblot analysis with antibodies to Cx43. The positions of nonphosphorylated (Cx43-NP) and various phosphorylated (Cx43-P₁, Cx43-P₂, Cx43-P₃) forms of Cx43 are indicated. (B) Cells were incubated for 24 hours in the absence or presence of TNF-α (10 ng/mL), after which cell lysates were prepared and subjected to immunoblot analysis with antibodies to Cx43 or to α-tubulin (control). (C) Immunoblots from experiments similar to that shown in (B) were subjected to densitometric analysis for determination of the intensity of the bands corresponding to the nonphosphorylated and phosphorylated forms of Cx43. Data are the mean ± SD of results of four independent experiments. *P < 0.05, **P < 0.01 (Scheffé test) versus the corresponding value for cells incubated in the absence of TNF-α.

Figure 8.

 

Lack of effect of TNF-α on the abundance of Cx43 mRNA in cultured human corneal fibroblasts. Cells were incubated for the indicated times in the presence of TNF-α (10 ng/mL), after which total RNA was isolated from the cells and assayed for Cx43 mRNA by quantitative RT-PCR. The amount of Cx43 mRNA was normalized by that of GAPDH mRNA and is presented in arbitrary units. Data are the mean ± SD of triplicates from a single experiment and are representative of three independent experiments.