Measurement of mRNAs for TGFβ and Extracellular Matrix Proteins in Corneas of Rats after PRK

Abstract

purpose. To assess the role of the transforming growth factor (TGF)β system in formation of corneal haze after excimer laser photorefractive keratectomy (PRK), levels of mRNAs for three TGFβ isoforms (TGFβ1, TGFβ2, and TGFβ3), the TGFβ type II receptor (TβRII), and extracellular matrix (ECM) genes including fibronectin (FN), collagen I, collagen III, and collagen IV were measured in rat corneas.

methods. Corneas were graded for corneal haze at 0, 1.5, 7, 21, 42, and 91 days after PRK. Total RNA was isolated from pooled corneas, and the levels of mRNAs were measured using competition-based quantitative reverse transcription–polymerase chain reaction (RT-PCR).

results. Severe corneal haze developed by day 42 and persisted to day 91. Levels of TGFβ1 mRNA were high in rat corneas before PRK and remained relatively constant. In contrast, levels of TGFβ2 and TGFβ3 mRNAs were very low in normal corneas, increased 300-fold and 25-fold, respectively, on day 21, and remained elevated on day 91. Levels of mRNA for TβRII increased, with a peak elevation of 50-fold on day 42 after PRK. Levels of mRNAs for ECM proteins also increased. Fibronectin mRNA was nondetectable in normal corneas but rapidly increased to 675 copies/cell on day 7 and remained elevated to day 91. Collagen III mRNA levels peaked on day 21 with a 700-fold increase compared with a very low level of expression in normal cornea, and then decreased on day 91. Expression of collagen I mRNA lagged expression of collagen III mRNA and peaked at day 42 after PRK with a 1200-fold increase over normal cornea. In contrast, mRNA for collagen α(1)IV, a major component in basement membranes, remained relatively stable through day 21 and then increased slightly on days 42 and 91.

conclusions. The synchronized increase in mRNA synthesis for both the TGFβ system and key ECM genes supports the hypothesis that TGFβ is a key growth factor promoting stromal haze formation in corneas after PRK and suggests that limiting TGFβ system may reduce corneal scarring after excimer laser ablation.

The potential use of the excimer laser keratectomy in ophthalmology was first proposed in 1983. ¹ It was suggested that ablation of the anterior cornea surface could modify the corneal profile, and thus induce a desired refractive change, remove corneal scars, or smooth surface irregularities. However, complications that may occur after photorefractive keratectomy (PRK) include subepithelial haze and regression of refractive effect. ^{2 3} Previous histologic studies indicated that subepithelial haze was restricted to the area directly beneath the ablated area and contained some atypical extracellular matrix (ECM) components including types III, IV, and VII collagens; fibronectin; laminin; and tenascin, along with proteoglycans and proliferating keratocytes. ^{4 5 6 7 8} The subepithelial haze that is observed clinically in the scar region is thought to result primarily from the nonorthogonal arrangement of fibrillar collagen molecules (types I and III) and the presence of nonfibrillar type IV collagen that normally are not present in high amounts in this region of corneal stroma. In contrast, collagen in normal clear cornea is arranged in a repeating orthogonal arrangement. Consistent with these histologic data, a reverse transcription–polymerase chain reaction (RT-PCR) analysis of rat corneas up to 6 weeks after PRK showed upregulation of the relative levels of mRNAs for types I, III, and V fibrillar collagens and for type IV basement membrane collagen. ⁹ These data led us to hypothesize that factors that increase corneal scarring contribute to development of corneal subepithelial haze and regression after PRK. In addition, the data emphasize the need to identify these factors and develop methods to reduce their action to control corneal scarring after PRK. 

Many growth factors and cytokines have been shown to be involved in corneal wound healing. ^{10 11} One growth factor in particular, transforming growth factor (TGF)β, is a major regulator of scar formation and is involved in fibrosis in many other tissues. ¹² For example, TGFβ directly induces transcription of collagen genes, elastin and lysyloxidase, by skin fibroblasts. ^{13 14 15} Furthermore, addition of exogenous TGFβ increases tensile strength of incisions, and inhibition of TGFβ with neutralizing antibodies reduces fibrosis in models of lung fibrosis and liver cirrhosis. ^{16 17}  

The TGFβ superfamily of proteins contains many multifunctional proteins, including TGFβs, activin-inhibin, and bone morphogenic proteins. ^{18 19} In mammals, there are three isoforms of TGFβ, designated TGFβ1, TGFβ2, and TGFβ3. The TGFβs are homodimers of approximately 28,000 molecular weight, and they often have similar biologic effects in vivo. The TGFβ isoforms all mediate their effects on cells through a membrane receptor system that consists of three distinct transmembrane proteins. Both the type I receptor (TβRI) and the type II receptor (TβRII) are serine or threonine kinases, and both are required for signal transduction. The type III receptor (TβRIII) does not have kinase activity, which suggests that it is not required for signal transduction, and its function is unclear. Signal transduction by the TGFβ receptor system is complex and is thought to be initiated by TGFβs binding directly to TβRII followed by association with TβRI proteins. The trimer complex of TGFβ, TβRI, and TβRII proteins initiates phosphorylation of TβRI by TβRII. Phosphorylation of TβRI activates the serine-threonine kinase active of the TβRI which in turn phosphorylates selected members of the Smad protein family. ^{20 21} The phosphorylated Smad proteins are translocated to the nucleus and recruit other proteins into a transcription factor complex that regulates transcription of different genes such as collagens, fibronectin, and type 1 plasminogen activator inhibitor. ^{22 23 24 25}  

To further investigate the involvement of the TGFβ system in corneal scarring after excimer PRK, we developed a competition-based quantitative RT-PCR assay that can measure the levels of low-abundance mRNAs. ²⁶ We used this quantitative RT-PCR technique to measure levels of mRNAs for TGFβ2, TGFβ3, TβRII, and ECM proteins in rat corneas at multiple time points after excimer laser PRK. 

Materials and Methods

Animal Models

Animal procedures were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The animal protocol was approved by the University of South Florida Animal Care and Use Committee. Adult Sprague–Dawley male rats (300 g) with normal eyes were anesthetized with intraperitoneal injection of ketamine (90 mg/kg) and xylazine (10 mg/kg). The area around the eyes was trimmed to remove eyelashes or whiskers in the visual field. The eyelids and surrounding ocular areas were disinfected by scrubbing with 1% povidone-iodine solution. Anesthetized rats were placed under the laser on a contoured vacuum pillow to prevent minor movements during treatment. A drop of proparacaine-HCl (0.05%) was applied to the eye, and the cornea was centered under the laser microscope. Bilateral ablation of the corneas was performed in a 3-mm treatment zone with an excimer laser (model 20/20B; Visx, Santa Clara, CA) using the laser in phototherapeutic keratectomy mode and a laser fluence between 158 and 162 mJ/cm². The corneal epithelium was ablated to a depth of between 28 and 34 μm, followed by ablation of the stroma to a depth of 20 μm for a total ablation depth of between 48 and 54μ m. After laser treatment, tobramycin (0.3%) ointment was applied to the corneal surface to prevent infection. No postoperative topical steroid was used. Corneas were stained with fluorescein daily to monitor corneal re-epithelialization. At 1.5, 7, 21, 42, or 91 days after excimer laser ablation, rats were killed by peritoneal injection of pentobarbital. Under an operating microscope, a 3-mm disposable biopsy punch was used to excise the ablated corneal area. The corneal buttons were snapped frozen in liquid nitrogen and stored at −84°C until analyzed. Corneas from non–excimer-treated normal rats served as control tissue. 

Evaluation of Corneal Haze, Edema, and Epithelial Healing

Rat corneas were graded for the amount of corneal haze on days 1, 3, 7, 21, 42, and 91 after excimer ablation using a 0-to-4 scale similar to that using in the U. S. Food and Drug Administration’s clinical evaluation trial of the Visx laser: 0, clear cornea; 1, faint haze; 2, haze present but pupil visible; 3, most of iris vessels not visible; 4, iris and pupil completely obscured. Edema was also graded using a 0-to-4 scale: 0, no stromal or epithelial edema; 1, slight stromal thickness; 2, diffuse stromal edema; 3, diffuse stromal edema with microcystic edema of the epithelium; and 4, bullous keratopathy. The time to closure of the epithelial defect in was also assessed by standard fluorescein staining, and corneas were graded as either healed or not healed. 

RNA Extraction from Rat Corneas

At each of the time points, four rats were killed, the eight corneas were excised and pooled, and total RNA was prepared using guanidine isothiocyanate and phenol-chloroform extraction (TRIzol reagent, Gibco–Life Technologies, Gaithersburg, MD) according to the manufacturer’s protocol. Briefly, tissue was homogenized in 1 ml TRIzol solution using a frosted glass-on-glass tissue grinder (Duall 20), RNA was extracted with chloroform, precipitated with isopropanol, washed with 80% ethanol, and dissolved in RNase-free water (0.1% diethylpyrocarbonate [DEPC]). Concentration and purity of RNA were measured spectrophotometrically at 260 nm (GeneQuant; Amersham Pharmacia Biotech, Uppsala, Sweden). It is important to note that eight individual corneas from four rats were pooled for each time point, and five separate RT reactions were performed to generate the competition PCR curve that was used to calculate the number of mRNA molecules for each gene (described below). The large number of individual corneas that were pooled to create the sample for each time point effectively converts the data into the biological average of the tissue. 

Competition-Based Quantitative RT-PCR

Competition-based quantitative RT-PCR (Q-RT-PCR) was performed as described previously using two synthetic multiprimer external RNA templates. ²⁶ For Q-RT-PCR a synthetic RNA sequence is used that contains the same nucleotide sequence as the authentic mRNA for the gene of interest in the regions where the 5′ and 3′ PCR primers bind. When the synthetic and authentic mRNA molecules are reverse transcribed together into cDNA molecules and then amplified by PCR, they compete with each other for a limiting amount of PCR primers in direct proportion to their relative concentrations. Because the number of synthetic RNA molecules added to the RT reaction is known, the number of authentic mRNA molecules can be calculated from the point at which the competition is calculated to be equal. 

Preparation of Competitor RNA

Construction of the two synthetic multiprimer external RNA template plasmids, designated plasmid epidermal growth factor (pEGF)/TGF and pMatrix, has been described previously. ²⁶ The pEGF/TGF template contains the complementary sequences corresponding to the 3′ primers and the 5′ primers for nine genes including rat and mouse TGFβ1, TGFβ2, TGFβ3, and TβRII. Similarly, the pMatrix template contains the sequences for the 3′ primers and the 5′ primers for genes of nine rat and mouse ECM proteins, including collagen I, collagen III, collagen IV, and fibronectin. Table 1 contains the PCR primer sequences and expected sizes of amplicons from both synthetic template and authentic RNA. Template RNA was transcribed in vitro from the linearized plasmids using T7 RNA polymerase (Ribomax; Promega Corp, Madison, WI), extracted with phenol-chloroform-isoamyl alcohol, and precipitated with sodium acetate. PolyA⁺ RNA was purified with oligo(dT) chromatography (MicroPolyA; Ambion Inc, Austin, TX), and the concentration was calculated by absorption at 260 and 280 nm. 

RT of Template RNAs and Corneal RNAs

First-strand cDNA was synthesized in a series of standard reactions using reverse transcriptase beads (Amersham Pharmacia Biotech, Piscataway, NJ). For each of the six time points, five separate RT reactions were performed. Each RT reaction tube contained 1μ g of total cellular RNA prepared from the pooled rat corneas and known amounts of polyA⁺ RNA from the two templates. For the pEGF/TGF template 10³, 10⁴, 10⁵, 10⁶, and 10⁷ copies were added, and for the pMatrix template, 10⁴, 10⁵, 10⁶, 10⁷, and 10⁸ copies were added. Mixtures of corneal RNA and template RNAs were brought to a volume of 33 μl using 0.1% DEPC-treated water, heated at 65°C for 10 minutes, chilled on ice for 2 minutes, and then transferred to a tube containing the RT reaction mix beads. After 1 μl of 0.5μ g/μl oligo dT ^{12 13 14 15 16 17 18} was added, the RT reaction mixture was gently vortexed and incubated at 37°C for 60 minutes followed by heating to 95°C for 5 minutes. 

PCR Amplification of cDNAs

PCR amplification of cDNAs was performed in a total reaction volume of 25 μl and contained 2.5 μl of the RT reaction product, 1.25 μl of 1 U/μl DNA polymerase (RedTaq; Sigma, St. Louis, MO), 2.5 μl of 10 × PCR buffer, 1 μl of 10 mM dNTPs, 17.25 μl of distilled water, and 25 picomoles of 3′ primer and 5′ primer of target gene. PCR amplification was initiated by one cycle of 94°C for 5 minutes followed by 35 sequential cycles of denaturation at 94°C for 45 seconds, annealing at 59°C for 1.5 minutes, and extension at 72°C for 2 minutes and a final extension cycle at 72°C for 10 minutes in a thermocycler (Ericomp; San Diego, CA). 

Detection and Measurement of RNA

PCR products were separated on Tris-acetate-EDTA 1.5% agarose gels containing 25 ng/ml ethidium bromide, photographed under UV illumination (Foto/Prep I; Fotodyne, New Berlin, WI) with a digital camera (DC120; Eastman Kodak, Rochester, NY) and stored as tagged information file format (TIFF) files. Band intensities were measured by computer (Image ver. 1.54; National Institutes of Health, Bethesda, MD) and normalized based on the molecular weight of the products. ^{27 28} The ratio of band intensity (template/sample) for each RT-PCR reaction was plotted against the number of RNA template molecules added to the RT reaction in the logarithm. The number of the authentic mRNA molecules in a sample was determined when the ratio of template–sample band intensity equaled 1. ^{26 28} Levels of mRNAs were expressed as the number of copies per cell using the constant of 26 pg of total RNA per cell. ²⁹ We reported previously that the amplicons generated with these primers for rat TGFβ genes and the ECM genes correspond to the correct target genes. ^{30 31} The method we used previously to establish the identity of the amplicons was endonuclease digestion of the PCR products. In this procedure, the amplicon generated for each gene was digested with a selected endonuclease that would theoretically produce two fragments with unique sizes that were predicted from the nucleotide sequence of the mRNAs. The endonuclease digestion of the amplicon for each gene generated the predicted size fragments, which demonstrated that the PCR product contained the predicted nucleotide sequence at the correct point in the product. 

Results

Levels of Corneal Haze, Edema, and Epithelial Healing

The levels of corneal haze, edema, and epithelial healing of the rat corneas at time points after excimer ablation are shown in Table 2 . Faint corneal haze developed by day 7 in all the rat corneas and increased to an average grade of 2.5 ± 1.0 at day 91 (corresponding to a level described as haze being present but pupil visible and most of iris vessels not visible). Slight stromal edema developed very rapidly in approximately half the rat corneas and resolved by day 7 in all rats. The 3-mm diameter epithelial defect usually healed in 3 days and was healed in all rats by 7 days after excimer ablation. None of the corneas became infected. 

Levels of mRNAs for TGFβ System

Figure 1A shows the agarose gel electrophoresis pattern of the competitive Q-RT-PCR amplifications generated using primers for TGFβ2. The lower band migrating at 342 bp corresponds to the size predicted for the synthetic template RNA product. The intensity of this band decreased from left to right due to the decreasing number of template RNA molecules added to the RT reaction. The upper band migrating at 604 bp corresponds to the size predicted for the PCR amplicon of the authentic mRNA product. The intensity of this band changed inversely to the template band, indicating competition during the PCR amplification between the authentic and synthetic cDNA molecules. 

By plotting the ratios of the band intensities versus the number of competitor RNA molecules added to the RT reaction on double-logarithm scales, band intensity data were transformed into the graphs shown in Figure 1B . The parallel lines with slopes of approximately 1 indicate similar amplification efficiencies of the authentic and synthetic RNA. The number of copies of authentic mRNA in the corneal samples was determined when the ratio equaled 1. The number of molecules of mRNA for TGFβ2 in corneas before PRK and at different times after PRK is shown in Figure 2 . 

The data indicate that, in normal corneas, the level of mRNA for TGFβ2 was very low with an average of less than 0.1 molecule per cell. The level increased rapidly after PRK with a 10-fold increase at 1.5 days and a 100-fold increase at 7 days after PRK. Levels reached a maximum expression of approximately 30 copies per cell at 21 days after PRK, which is approximately a 1000-fold increase compared with normal cornea. Expression of TGFβ2 mRNA decreased slightly at days 42 and 91 to an average of approximately 10 copies per cell. 

Measurement of TGFβ type II receptor was performed in a similar manner. As shown in Figure 1C , the RT-PCR generated two amplicon bands of the predicted sizes whose intensities varied inversely. Transformation of the ratios of the band intensities again generated linear lines (Fig. 1D) . Calculation of the average number of TβRII mRNA molecules per cell indicated that there was low expression in normal corneas (two copies/cell) and in corneas 1.5 days after excimer ablation (three copies/cell) then rapidly increased 10-fold at 7 days and 50-fold at 21 days after PRK. Peak expression occurred at 42 days (104 copies/cell) then slightly decreased at 91 days after PRK (70 copies/cell; Fig. 2 ). 

Competitive Q-RT-PCR analyses for TGFβ3 and TGFβ1 mRNAs were performed in the same way (agar electrophoresis gels and ratio plots not shown). Calculated levels of mRNAs before and after PRK are shown in Figure 2 . TGFβ3 was expressed at approximately three copies/cell in normal cornea, progressively increased to a maximum of 25-fold elevation (75 copies/cell) by 21 days, and then slightly decreased (58 copies/cell) at 42 and 91 days after ablation. In contrast to the increases observed for mRNAs for TGFβ2, TGFβ3, and TβRII II, the levels of mRNA for TGFβ1 were high in normal corneas (23 copies/cell) and did not change substantially during the 91 days after ablation (39 copies/cell). 

Levels of mRNAs for ECM Proteins

Competitive Q-RT-PCR also was performed to measure the levels of four mRNAs of ECM components that are common in scars. These were the type I and type III fibrillar collagens, the type IV basement membrane collagen, and fibronectin, a protein important for cell attachment and migration. Figures 3A and 3C show the agar gel electrophoresis of competitive RT-PCR products for collagen III and collagen IV, respectively. The amplicons for synthetic and authentic RNAs were of the predicted sizes, and their intensities varied inversely. Plots of the ratios of the band intensities generated the graphs shown in Figures 3B and 3D . Levels of mRNAs for collagen III were very low in normal corneas (3 copies/cell), increased at 1.5 days (27 copies/cell) and 7 days (1530 copies/cell), reached a maximum at 21 days (2080 copies/cell), decreased at 42 days (1430 copies/cell), and sharply decreased at 91 days (39 copies/cell). In contrast, the levels of mRNA for collagen IV were detected in normal corneas (104 copies/cell) and remained relatively similar up to 21 days after PRK (from 70 to 180 copies/cell), after which levels increased slightly at 42 days (490 copies/cell) and 91 days (675 copies/cell). Levels of mRNA for collagen I were similar in normal cornea (25 copies/cell) and in corneas at 1.5 days after PRK (26 copies/cell). By 7 days and 21 days after PRK, however, mRNA levels had risen to 416 copies/cell and 5450 copies/cell, respectively, and then peaked on day 42 (31,200 copies/cell) and decreased sixfold at day 91 (5320 copies/cell) after PRK. 

Levels of mRNA for fibronectin showed the most dramatic change of the genes studied. Fibronectin mRNA was not detected in normal cornea and was barely detectable 1.5 days after PRK, but sharply increased to 675 copies/cell at day 7 and remained elevated on day 21 (494 copies/cell), day 42 (779 copies/cell), and day 91 (1060 copies/cell) after PRK. 

Discussion

The molecules that are responsible for the subepithelial haze and regulate the formation and regression of haze after excimer PRK are not fully understood. Previous immunohistochemical studies indicate that several ECM proteins including type III collagen, type VII collagen, and fibronectin appear beneath the ablated zone in primates and are removed in different time frames after PRK. ⁷ For example, both type VII collagen and fibronectin appear within 7 days after PRK, but type III collagen is not present until 3 weeks after PRK. At later times, type VII collagen becomes concentrated in the basement membrane and persists to 18 months, whereas fibronectin begins to disappear from the subepithelial stroma at 6 weeks and had virtually disappears by 3 months. Type III collagen staining is very intense in the anterior stroma at 6 weeks and persists at 18 months, although at a lower intensity. 

In the current study of rat corneas after PRK, levels of mRNAs for collagen I, collagen III, and fibronectin all showed dramatic increases after ablation, whereas levels of mRNA for type IV collagen remained almost unchanged until 21 days after PRK then increased slowly through 91 days. Fibronectin mRNA was the first to peak at 7 days, followed by collagen III mRNA, which reached a plateau at 7 days, and finally collagen I mRNA, which peaked at 42 days. Levels of fibronectin mRNA remained elevated at 91 days, whereas both type III and type I collagen mRNAs began to decrease at 91 days. The temporal pattern of mRNA synthesis for fibronectin, which increased from an essentially undetectable level in noninjured corneas to more than 600 copies/cell within a week after PRK, suggests that induction of fibronectin is important for initiating healing of PRK wounds and is consistent with the role of fibronectin in promoting migration of corneal epithelial cells. ^{4 32} Unlike the primate model of PRK, ^{6 7} in this rat model, synthesis of fibronectin mRNA was sustained through 91 days. This may reflect the deeper ablation of the PRK wound in the rats which may cause prolonged synthesis of fibronectin by keratocytes as they modify the provisional wound matrix. ³³  

The low level of type III collagen mRNA measured in the uninjured adult rat corneas on day 0 (3 copies/cell) is consistent with the low level of collagen III protein reported for normal adult corneas (in contrast to the high level of collagen III protein observed in fetal cornea). ^{34 35} Furthermore, the rapid increase and plateau in the level of type III collagen mRNA from days 7 to 42 after PRK ablation followed by a decrease at 91 days after PRK is consistent with the appearance and disappearance of type III collagen protein detected by immunostaining in the primate model of PRK, ^{6 7} and the detection of type III collagen mRNA in rat corneas shortly after PRK. ⁹  

The moderate level of type I collagen mRNA in normal corneas (25 copies/cell) probably reflects a continuous low turnover and remodeling of the type I collagen that comprises most of the corneal stroma matrix. ^{36 37} In addition, type I collagen mRNA levels peaked late in healing (at 42 days) and also reached the highest level (5300 copies/cell) of all the ECM genes after PRK. This is congruent with the observation that type I collagen eventually replaces type III collagen in most scars and becomes the dominant type of collagen in scar tissue. 

Six genetically distinct chains of type IV collagen have been described, and the spatial distributions of the resultant isoforms in the anterior segment are distinct. ^{38 39 40} In normal human corneas, α3 and α4 chains are present in the epithelial basement membrane of the central cornea, around stromal keratocytes, and on the endothelial face of Descemet’s membrane. In contrast, α1, and α2 chains are present in the epithelial basement membrane of the limbus and conjunctiva and on the stromal face of Descemet’s membrane. Theα 5 and α6 chains colocate with the α3 and α4 chains and are also found in the epithelial basement membrane of the limbus. These spatial distinctions in localizations of type IV chains probably contribute substantially to the different structural and functional properties of the basement membranes in the central cornea, limbus, conjunctiva, and Descemet’s membrane. In addition, an abnormal limbal-like distribution of the type IV collagen chains (i.e., the presence of α1 chains and the absence of α3 chains) was observed in the basement membranes of human radial keratotomy scars as long as 3 years after surgery. ⁴¹  

In the current study, the central 3 mm area of the cornea was excised, including the epithelium, stroma, Descemet’s membrane, and endothelium. The detection of mRNA for α1(IV) gene in normal rat corneas most likely reflects normal turnover of α1(IV) collagen by endothelial cells in Descemet’s membrane. ³⁷ The very gradual increase in α1(IV) mRNA levels through day 91 may represent a combination of synthesis of α1(IV) mRNA and protein by epithelial cells, stromal fibroblasts, and endothelial cells. All three cell types were observed to express α1(IV) mRNA in lacerated rabbit corneas in an apparent attempt to construct a basal lamina-like structure in the corneal wound. ³⁷ Unfortunately, the sequences of ratα 3(IV) and α4(IV) genes are not known, and that prevents analysis of the mRNA levels of these type IV collagen chains by Q-RT-PCR. 

Overall, the temporal patterns and levels of mRNAs for ECM proteins measured in rat corneas after PRK presents a picture of an integrated process with sequential but overlapping phases that combine to heal the corneal wound. The healing process, however, results in a scar that repairs the injury rather than regenerates the structures of the original clear cornea. The variations in the expression of these ECM genes in rat corneas after PRK implies that there is a complex pattern of expression of the molecules that regulate their synthesis. This is because, in general, the regulation of expression of ECM proteins occurs at the level of mRNA transcription and stability of the mRNA, and not at the level of translation. For example, TGFβ treatment markedly increased the levels of mRNAs for type 1 collagen, fibronectin, and thrombospondin in cultures of mouse 3T3 fibroblasts. ⁴² Thus, changes in the levels of ECM proteins are generally reflected in changes in levels of mRNAs. 

In this study, we examined the expression of the three isoforms of TGFβ and the TGFβ type II receptor. Of the four TGFβ system genes, only levels of TGFβ1 mRNA did not vary dramatically with time after excimer ablation (23 copies/cell in normal cornea, and levels did not increase more than twofold after ablation). This finding is consistent with reports that levels of TGFβ1 mRNA remained constant in other cell types, even after stimulation. For example, the level of TGFβ1 mRNA was similar in unstimulated monocytes and in activated macrophages. ⁴³ However, TGFβ1 protein was secreted only by activated macrophages, suggesting that synthesis of TGFβ1 protein was controlled at the level of translation. Furthermore, PC-3 human prostate adenocarcinoma cells were reported to contain high levels of TGFβ1 mRNA, but they mainly secreted TGFβ2 protein in spite of containing low levels of TGFβ2 mRNA. ⁴⁴ Thus, regulation of TGFβ1 protein synthesis may occur predominantly at the posttranscriptional level, and the relatively constant levels of TGFβ1 mRNA measured in the rat cornea may not reflect levels of TGFβ1 protein. 

Previous reports also suggest that TGFβs are involved in corneal wound healing. An immunohistochemical study reported that all three TGFβ isoforms are expressed in the regenerating epithelial cells of rats after PRK. ²⁸ Also, systemic treatment of rats by intraperitoneal injection with a panspecific neutralizing antibody to all three isoforms of TGFβ reduces stromal cell density and immunostaining of laminin and fibronectin in the subepithelial stroma during the first 10 days after PRK. ⁴⁵ Systemic treatment of rats with a neutralizing antibody specific to TGFβ1 or TGFβ2 also reduces stromal cell recruitment to the wound site and reduces subepithelial fibrosis, whereas treatment with a neutralizing antibody specific to TGFβ3 is not effective. In addition, topical application of a neutralizing antibody to TGFβ1 reduces stromal fibrosis in rabbits after PRK but does not inhibit or delay stromal rethickening (regression). ^{46 47} Although inhibition studies have not been performed in humans, substantial levels of latent TGFβ1 (38 ng/ml) and TGFβ2 (2 ng/ml), which probably originates from the lacrimal gland epithelial cells, have been detected in tears from normal eyes. ⁴⁸ Furthermore, the rate of release of TGFβ1 in tears after PRK increases approximately 18-fold in the first 2 days after surgery. ⁴⁹  

In contrast to the constant level of mRNA for TGFβ1 measured in the rat corneas, we found that levels of mRNAs for TGFβ2 and TGFβ3 varied substantially after excimer ablation. Normal rat corneas contained very low levels of mRNAs for TGFβ2 (0.1 copy/cell) and TGFβ3 (3 copies/cell). Twenty-one days after excimer ablation, levels of mRNAs for TGFβ2 (500-fold) and TGFβ3 (100-fold) increased dramatically and remained elevated at 91 days. Equally important, was the increase observed in expression of mRNA for the type II TGFβ receptor (from 2 copies/cell to 100 copies/cell) which paralleled the increases in mRNAs for TGFβ2 and TGFβ3. The parallel increases observed in both ligands and their receptor suggests that these genes are predominantly regulated by transcriptional activation in corneas after PRK. In addition, it is possible that the sustained elevated production of TGFβ2 and TGFβ3 mRNAs may be due in part to autoinduction of their own mRNAs. Autostimulation of TGFβ1 gene expression has been demonstrated in many normal and transformed cultures of cells, ⁵⁰ and TGFβ response elements have been found in promoter regions of the TGFβ1 gene. ⁵¹  

In addition to transcriptional and translational regulation of isoforms of TGFβ genes, there is another level of regulation of TGFβ activity, which is the posttranslational activation of the latent TGFβ. During lung fibrosis, β6 integrin can appear in epithelial cells. This integrin can bind TGFβ1 latency–associated peptide, thereby activating latent TGFβ1. ⁵² Therefore, even in the absence of elevated transcription of TGFβ1 gene, the amount of activated growth factor can increase substantially and be a factor in scar formation. 

The correlation observed in this study between the increases in mRNA levels for the TGFβs and the ECM proteins implies that there is a cause-and effect-relationship between induction of TGFβ genes and production of corneal scar components. Perhaps the most direct evidence for a major role of TGFβ in formation of haze after PRK is the report that topical treatment of rabbit corneas with neutralizing antibodies to TGFβs for the first 3 days after PRK reduced the level of haze, when measured by light reflectivity. ⁴² Other data indirectly support a role for the TGFβ system in promoting scar formation in corneas after PRK. A TGFβ response element was identified in the promoter of the α1(I) collagen gene, and TGFβ1 increased the stability of mRNA for α1(I) collagen in confluent cultures of fibroblasts. ⁴² Also, TGFβ2 knockout mice have a reduced corneal stroma layer compared with normal mice, which indicates that TGFβ2 plays an important role in embryonic corneal development, which shares some key process with corneal wound healing. ⁵³ In addition to its effects on collagen synthesis, TGFβ1 increases the synthesis of the chondroitin sulfate proteoglycan core protein and the mass of the glycosaminoglycan side chains. ⁵⁴ Besides its direct effects on ECM gene expression, TGFβs also could have indirect effects on corneal scar formation by altering the levels of the proteases that degrade matrix proteins and their inhibitors. TGFβs have been reported to suppress production of matrix metalloproteinases and increase production of tissue inhibitors of metalloproteinases (TIMPs). ^{55 56} The combination of increasing synthesis of ECM genes while suppressing production of MMPs and increasing production of TIMPs would have the overall effect of increasing deposition of scar matrix. Levels of MMP-9 and MMP-2 and TIMPs 1 and 2 also were reported to increase in rat corneas after PRK. ⁵⁷ TGFβs also could induce other factors that promote ECM formation. One likely candidate factor would be connective tissue growth factor (CTGF) which has been shown to mediate increases in matrix synthesis in cells treated with TGFβs. ⁵⁸ Expression of the CTGF gene in corneas after PRK has not been investigated. 

In summary, these data show for the first time a synchronized increase in transcription of TGFβ isoforms, the type II receptor gene, and several key ECM genes during the 91 days after PRK. This finding strongly suggests that the subepithelial haze that developed in the rat corneas after PRK was due to chronic, elevated expression of the TGFβs and the type II receptor that caused excessive accumulation of abnormal ECM in the ablated area. Furthermore, the data reinforce the concept that limiting the activity of the TGFβ system is a key objective for controlling corneal scarring after excimer laser PRK. 

 Supported by National Institutes of Health Grant EY05587.

 Submitted for publication June 13, 2000; accepted July 19, 2000.

 Commercial relationships policy: N.

 

Table 1.

 

Primer Sequences for TGFβ, Receptor, and ECM Components

Gene	5′ Primer	3′ Primer	Template	Sample Size (bp)
TGFβ1	CAAAGACATCACACACAGTA	GGTGTTGAGCCCTTTCCAGG	344	440
TGFβ2	GAGCAGGAGTACTACGCCAA	ACTGTTGTGACTCCAGTCTG	342	604
TGFβ3	ATTACCTCTAAGGTTTTCCG	AGGGCCCTCTTCTTCCTCTG	347	527
TβR II	TGTGCTCCTGTAACACAGAG	GATCTTGACAGCCACGGTCT	339	431
COLα2(I)	GGCCCTGGACCAATGGGCTTA	ACCAACAAGTCCTCTGGCACC	334	774
COLα1(III)	GGCTCCTGGTGAGCGAGGAC	CCCATTTGCACCAGGTTCTCC	337	523
COLα1(IV)	GGACAAGCAGGCTTTCCTGGA	TGCTGTCCAGGAAGGCCAGG	336	743
Fibronectin	CCGGGTTCTGAGTACACAGTC	AGGGACCACTTCTCTGGGAGG	345	768

Table 2.

 

Microscopic Evaluation of Rat Corneas after Excimer PRK

Parameters	Day
		0	1.5	7	21	42	91
Haze	0.0	0.0	0.83± 0.40	0.60± 0.33	2.25± 0.53	2.50± 1.00
Epithelial defect	0.0	1.31 ± 0.06	0.0	0.0	0.0	0.0
Edema	0.0	0.47 ± 0.09	0.0	0.0	0.0	0.0

 Data are mean scores ± SD on a scale of 0 to 4 (grades described in the Methods section).

Figure 1.

 

Quantitative RT-PCR for TGFβ2 and TβRII mRNAs in rat corneas after PRK. (A, C) Photographs of ethidium bromide–stained PCR amplicons separated by 1.5% agarose gel electrophoresis. Two different-sized bands were generated during RT-PCR from the specific message in total RNA from pooled rat corneas and serial dilutions of synthetic mRNA from a pEGF/TGF template. Dilutions of template RNA are shown from left to right and correspond to 10⁷ to 10³ copies/reaction. (B, D) The ratios of template to sample amplicon band intensities were calculated from data shown in (A) and (C), respectively, after normalization for their molecular weights. The log of the ratio was plotted versus the log of the copy number of the added template (10⁷–10³ copies/reaction). Levels of TGFβ2 and TβRII RNA molecules in normal and PRK corneas on days 1.5, 7, 21, 42, and 91 were determine from the best fit lines for which the ratio equaled 1.

Figure 2.

 

Levels of mRNA for TGFβs and ECM proteins in pooled rat corneas before and after PRK. Competitive Q-RT-PCR was performed on pools of rat corneas before (day 0) and at 1.5 days, 7 days, 21 days, 42 days, and 91 days after PRK. Levels of mRNAs for TGFβ1, TGFβ2, TGFβ3, TβRII, collagen I, collagen III, collagen IV, and fibronectin mRNA were calculated from best fit lines of plots of the log of ratios of intensities of authentic and template amplicons versus the log of added competitor RNAs.

Figure 3.

 

Quantitative RT-PCR for collagen III and collagen IV mRNAs in rat corneas after PRK. (A, C) Ethidium bromide-stained PCR amplicons separated by 1.5% agarose gel electrophoresis. Two different-sized bands were generated during RT-PCR from the specific message in total RNA from pooled rat corneas and serial dilutions of synthetic mRNA from pMatrix. Dilutions of template RNA are shown from left to right and correspond to 10¹⁰– 10⁶ copies/reaction. (B, D) The ratios of template to sample amplicon band intensities were calculated from data shown in (A) and (C), respectively, after normalization for their molecular weights. The log of the ratio was plotted versus the log of the copy number of the added template (10¹⁰–10⁶ copies/reaction). Levels of collagen III and collagen IV in normal and PRK corneas on days 1.5, 7, 21, 42, and 91 were determined from the best fit lines when the ratio equaled 1.