Serum Albumin in Mammalian Cornea: Implications for Clinical Application

Abstract

purpose. To compare the abundance and spatial distribution of serum albumin in the mouse and bovine cornea.

methods. Serum albumin from cornea was separated from transketolase by SDS-PAGE (±dithiothreitol [DTT]) and identified by peptide sequencing and immunoblot analyses. The fractional content of serum albumin was determined in water-soluble extracts of cornea by imaging analyses after SDS-PAGE. Serum albumin was localized in cornea by immunohistochemistry and by SDS-PAGE analyses of samples from separated epithelium and stroma.

results. SDS-PAGE (−DTT) resolved mouse serum albumin and transketolase and indicated that serum albumin was 13% of the water-soluble protein in whole mouse corneas. By contrast, corneal epithelial fractions contained little (<1%) serum albumin. Immunohistochemistry indicated that mouse serum albumin was present throughout the stroma between collagen lamellae. Immunohistochemical analyses of bovine cornea yielded similar results. In addition, immunohistochemistry for serum albumin revealed positive staining in a small number of basal epithelial cells next to Bowman’s membrane, and greater staining in the anterior–peripheral stroma as well as immediately adjacent to Descemet’s membrane.

conclusions. Mouse and bovine cornea have a similar content and spatial distribution of serum albumin. The appreciable serum albumin in the cornea documented here and elsewhere raise the possibility that it contributes to the physiological or optical functions of the cornea. Moreover, serum albumin’s ability to bind drugs suggests that mice corneas could be exploited to study drug-serum albumin interactions in vivo and to test the usefulness of serum albumin as a drug carrier for corneal disorders.

Serum albumin diffuses from peripheral blood vessels around the cornea toward its center ^{1 2} ; however, inconsistencies in the literature have arisen with regard to its abundance and spatial distribution. Although in one study human serum albumin appeared more predominant in the corneal epithelium, ³ another study found that both human and bovine serum albumin in the cornea are situated primarily within the stroma, with peripheral to central and sharp anterior to posterior concentration gradients. ⁴ The peripheral gradient fits with the evidence that serum albumin diffuses from peripheral blood vessels into the extracellular collagen matrix of the rabbit corneal stroma. ^{1 2} Previous reports have indicated that serum albumin represents the most abundant water-soluble protein in the human cornea. ^{3 5} One report found bovine serum albumin (BSA) to be 12% of the water-soluble stromal protein, ⁶ whereas another found a 62-kDa species, speculated to be BSA, to be 23% of the water-soluble corneal protein. ⁷ Despite several studies characterizing the abundant corneal proteins in the mouse, ^{8 9 10 11 12} there has been no report indicating the presence of mouse serum albumin (MSA). 

In the current study, MSA represented approximately 13% of the water-soluble protein in the mouse cornea, and it comigrated precisely with the abundant transketolase (TKT) when examined by SDS-PAGE under reduced conditions. Further, we show that corneal MSA and BSA are located almost entirely in the stroma. The abundance and spatial distribution of serum albumin could be important keys to understanding its function in cornea and may have particular relevance for the activity of hormones and drugs that it binds. ^{13 14} Further, our finding that the concentration and spatial distribution of serum albumin in the mouse cornea are similar to that of other mammals, including humans, raises the possibility of using mice as an animal model to obtain better understanding of drug–serum albumin interactions ^{14 15} in the cornea when designing new drugs, drug combinations, and delivery protocols. In particular, these data indicate that mice could be exploited to test serum albumin as a drug carrier ¹⁶ to the cornea or other novel technologies that might take advantage of the abundance of this extracellular protein in the cornea. 

Materials and Methods

Animal and Eye Sources

Adult 129/SvEvTac mice came from a small colony derived from purchased breeding pairs (Taconic, Germantown, NY). Bovine eyes from animals 18 to 24 months old were purchased from a local abattoir (J. W. Treath and Sons Inc., Baltimore, MD). Maintenance and treatment of 129/SvEvTac mice were in accordance with National Institutes of Health guidelines and with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Protein Analyses

To generate mouse protein fractions, whole corneas from 32-week-old mice were put into 50 μL of a wash that loosens the epithelium from the stroma (Cellstripper; Mediatech Inc., Herndon, VA) and incubated at 37°C for 45 minutes. This stromal and epithelial wash was designated as fraction SE. The epithelium was then separated from the stroma with forceps, placed into a 1.5-mL minicentrifuge tube containing extraction buffer that consisted of 50 μL of 0.067 M Tris-HCl (pH 7.5) with 10% glycerol and protease inhibitors (1 tablet of a protease inhibitor cocktail MiniComplete per 10 mL of buffer; Roche Diagnostics Corp., Indianapolis, IN). Cells were lysed with five freeze-thaw cycles and the sample ground with a minipellet tissue grinder (Kimble/Kontes, Vineland, NJ) followed by centrifugation at room temperature for 2 to 3 minutes at 14,000g. The aqueous phase was used as the water-soluble protein fraction and designated as fraction E. The stroma was put into a separate tube with 50 μL of 1× PBS and incubated for 2 days at 4°C to allow all other water-soluble proteins to diffuse from between collagen layers. This fraction was designated S. After each extraction, samples were frozen at −20°C until gel analyses could be performed. Analysis was performed on 12 μL of each fraction as will be described. 

For bovine cornea, either whole corneas or epithelia that had been scraped from corneas were put into extraction buffer, cells lysed with five freeze-thaw cycles, debris removed by centrifugation at 14,000g, and the aqueous phase taken. In some cases, the epithelium was removed from the bovine cornea as described for the mouse and extracted. To separate anterior and posterior stromal regions surgically, bovine corneas were dissected from the eye, a center to peripheral cut made to flatten the cornea, the epithelium was gently scraped away, and the stroma was frozen flat on a wax surface cooled with dry ice to solidify the tissue for cutting. The stroma then was clamped with a hemostat prechilled in dry ice, excess stroma was cut away from the clamp edges, and the anterior half was sliced with a razor blade away from the posterior half using the clamp edges as guides. These samples went into preweighed tubes and stromal protein was extracted in a volume (microliters) three times the tissue weight (micrograms). The soluble buffer was removed amd the stroma dried under vacuum and weighed. The water weight of the original stromal sample was calculated by subtracting the weight of the dried stroma from that of the original stroma before extraction. 

SDS-PAGE and immunoblot transfers were performed as described by the manufacturer for 10% Bis-Tris gels using 3-(N-morpholino)propanesulfonic acid (MOPS) running buffer (Novex; Invitrogen, Carlsbad, CA). To identify the 58-kDa protein species in whole corneal extracts from mouse as MSA, SDS-PAGE (−DTT) was performed, the gel was stained (Gel Code Blue; Pierce, Rockford, IL), and a gel slice was cut that contained the 58-kDa band. Mass spectra analysis of trypsinized fragments of this species was performed as a service (Harvard Microsequencing Center, Cambridge, MA). Digital images were obtained under UV light (ChemiImager 4000; Alpha Innotech Corp., San Leandro, CA) from fluorescence-stained immunoblots or red-stained (SyproRuby or SyproRed; Molecular Probes, Inc., Eugene, OR) gels. One of the red stains (SyproRed; Molecular Probes) was used because it generates a linear signal over a more extensive protein concentration range with less dependence on amino acid composition than Coomassie blue staining. ¹⁷ For quantitative analysis, bracketed exposures were taken, and the exposure with the strongest signal that still appeared to be linear with respect to exposure length was used. Band signals from these red-stained gels were quantitated with the imaging software (ChemiImager; Alpha Innotech Corp.) in an average of six experiments. For immunoblots, membranes were blocked (SeaBlock; Pierce) for anti-serum albumin antibodies or 5% dry milk and 0.1% Tween-20 in 1× PBS for anti-TKT ¹¹ and anti-aldehyde dehydrogenase 3a1 (ALDH3) antibodies (the kind gift of Ronald Lindahl, Department of Biochemistry, University of South Dakota, Vermillon, SD). Albumin antibodies included two polyclonal rabbit anti-MSAs (RDI-MSALBabr; Research Diagnostics, Inc., Flanders, NJ, and 55442; ICN Pharmaceuticals, Inc., Costa Mesa, CA), a monoclonal mouse anti-BSA (B2901; Sigma-Aldrich, Inc., St. Louis, MO) and a polyclonal rabbit anti-BSA (56963; ICN Pharmaceuticals, Inc). Secondary antibody incubations with horseradish peroxidase linked to donkey anti-rabbit or anti-mouse Ig (Amersham Pharmacia Biotech, Piscataway, NJ) were diluted 1:3000. Enhanced chemiluminescence (ECL) detection was as described by the manufacturer (Amersham Pharamcia Biotech). 

Immunohistochemistry

Immediately after dissection, eyes of 22-week-old mice were cryosectioned 10 μm thick in optimal cutting temperature (OCT) compound (Tissue-Tek; Sakura Finetek USA., Inc., Torrance, CA) at −20°C. Cryosections were fixed in 4% neutral paraformaldehyde for 10 minutes, washed thoroughly, and blocked for 15 minutes with 0.1 M glycine. Primary anti-MSA antibody incubations were performed at a 1:100 dilution in 1× PBS with 0.1% Triton X-100. Sections were washed, incubated with Cy-3 (red) secondary antibodies (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) in the same buffer as used with primary antibodies, washed thoroughly again, and counterstained with 4′,6′-diamino-2-phenylindole (DAPI; 1 μg/mL). For negative controls, primary antibodies were omitted. For analyses of bovine cornea, whole eyes were fixed in neutral 4% paraformaldehyde in 1× PBS within 2 hours of death. A slice of the cornea passing through the center was frozen in OCT, and 14 μm sections cut. Primary anti-BSA antibody incubations were performed in the same fashion as described for mouse sections. Stained sections were imaged on a laser scanning confocal microscope (SP2; Leica, Deerfield, IL). 

Results

Comigration of MSA and TKT

SDS-PAGE (−DTT) revealed three major protein bands (51, 58, and 69 kDa) in water-soluble fractions of the mouse cornea (Fig. 1A) . The 51-kDa and 69-kDa species had been characterized previously as ALDH3 and TKT, respectively. ^{8 10 11} However, the 58-kDa species had not been observed earlier, consistent with its absence when examined by SDS-PAGE (+DTT) in the present investigation (Fig. 1A) . The 58-kDa band was extracted from an SDS-PAGE (−DTT) gel and digested with trypsin, and the peptide sequences were examined by mass spectrometry. This analysis revealed sequences that covered 43% of mouse preserum albumin (Fig. 1B) . To verify that the 58-kDa species in SDS-PAGE (−DTT) was serum albumin, immunoblot analysis using a polyclonal rabbit anti-MSA antibody was performed on purified MSA and whole-corneal extracts (Fig. 2 , Anti-MSA). Both the purified and corneal MSA migrated as a 58-kDa species detected by SDS-PAGE (−DTT), but as a 69-kDa species (as does TKT ¹¹ ) by SDS-PAGE (+DTT; Fig. 2 , Anti-MSA). An immunoblot performed on the water-soluble extract from mouse corneal epithelium confirmed that TKT migrated to the 69-kDa position. Two-dimensional gel analysis, using isoelectric focusing in the first dimension and SDS-PAGE (+DTT) in the second, confirmed that there were two prominent 69-kDa species (data not shown). Thus, TKT and MSA comigrate in SDS-PAGE (+DTT). 

For comparison, bovine corneal proteins were analyzed with immunoblots in similar fashion. An immunoblot using the rabbit polyclonal anti-mouse TKT reacted with the major band migrating at 65 kDa. In this case, 100 μg of water-soluble protein from corneal epithelium were placed on the gel to allow for reduced recognition of the antibody due to cross-species differences, and this resulted in a slight shift in band migration from the usual 68-kDa position (Fig. 2 , Anti-TKT). Immunoblot analysis was performed on water-soluble protein from bovine cornea with a rabbit poly-clonal anti-BSA antibody for samples with and without reducing agent. In SDS-PAGE (+DTT), BSA migrated as a 60-kDa species, separate from the larger TKT (Fig. 2 , Anti-BSA). However, similar to MSA, BSA had an altered migration pattern in SDS-PAGE (−DTT), instead migrating as a 53-kDa species. 

Stromal Localization and Corneal Abundance of Serum Albumin

After tissue dissection of mouse corneas, the water-soluble fractions were analyzed by SDS-PAGE (−DTT). From this dissection, three water-soluble fractions were prepared: (1) fraction SE, containing a wash of both epithelium and stroma; (2) fraction E, containing extracted epithelial proteins; and (3) fraction S, containing the extracted stromal proteins. SDS-PAGE (−DTT) analysis of equivalent amounts revealed TKT and ALDH3 in each fraction (the 69-kDa species and the 51-kDa species, respectively; Fig. 3A ). However, MSA, the 58-kDa species, appeared as a major protein in fractions containing stromal proteins, SE and S, but was barely discernible in fraction E (Fig. 3A) . Similarly, for epithelial extracts of bovine cornea, SDS-PAGE (+DTT) analysis showed that the 50-kDa and 45-kDa species were abundant and BSA was undetectable (Fig. 3B) . However, in the analyses of the water-soluble fraction of whole bovine cornea by SDS-PAGE (+DTT) the 60-kDa BSA clearly was present as a major species (Fig. 3B) . Other reports have identified the 50-kDa protein as ALDH3 ¹⁸ and the 45-kDa species as isocitrate dehydrogenase (ICDH). ¹⁹  

These major species in mouse and bovine corneal extracts were quantified by digital scans of the red-stained (SpyroRed; Molecular Probes) gels (Fig. 3) . A compilation of the signal for the lanes with extracts SE, E, and S revealed that MSA was 13% and TKT was 19% of the total water-soluble protein of the mouse corneas. Further, the digital scans revealed that TKT represented 23% and ALDH3 47% of fraction E—the latter number being similar to that determined previously in the epithelium. ¹⁰ In whole bovine cornea water-soluble protein comprised 10% BSA, 32% ALDH3, and 9% ICDH. In contrast, ALDH3 and ICDH were 58% and 14% of the water-soluble epithelial fraction, respectively. Besides serum albumin, only bovine TKT was present in a higher fractional content in whole cornea, at 6%, than in the epithelial fraction, where it was 3% of the water-soluble protein. At most, trace levels of serum albumin were detected in the bovine and mouse corneal epithelial fractions by SDS-PAGE (Fig. 3) . 

MSA and BSA Spatial Distribution in Stroma

When stained using an anti-MSA primary antibody (Research Diagnostics, Inc.), a typical confocal image of a mouse cornea showed signal for MSA at the epithelial surface, and in a striated pattern throughout the corneal stroma (Fig. 4A , red channel). Results were identical with a different anti-MSA antibody (ICN Pharmaceuticals, Inc; data not shown). Although normally included, we omitted a serum block to avoid nonspecific absorption of anti-albumin antibodies. However, 0.1 M glycine was used to block any unreacted paraformaldehyde from the fixation step. This method yielded no signal when the secondary antibody was used alone on mouse cryosections (Fig. 4B , red channel). A DAPI nuclear counterstain was included to visualize cell nuclei to distinguish between epithelium, stroma and endothelium (Fig. 4 , blue channel). 

A typical confocal image of cryosectioned bovine cornea appeared similar to the mouse cryosections when stained using the polyclonal rabbit anti-BSA (data not shown) or the monoclonal mouse anti-BSA antibody (Fig. 5 , red channel). Again, DAPI nuclear staining was performed to reveal corneal structures (Fig. 5 , blue channel). Peripheral cornea (Fig. 5A) revealed higher levels of staining for BSA at the anterior stroma compared with central cornea (Fig. 5B) . Staining was greater in the stroma immediately adjacent to the Descemet’s membrane in the central (Fig. 5C) and peripheral (not shown) stroma. The endothelium region stained as well (Fig. 5C) ; however, it was not possible in these sections to ascertain whether this stain was intracellular or extracellular. In the negative control experiments, sections treated with only the secondary antibody were not stained (Fig. 5D , red channel). A small number of basal epithelial cells were stained, giving them a rocket-shaped appearance (Figs. 5E 5F) . The staining for serum albumin at the surface of both bovine and mouse corneas (Figs. 4 5) is consistent with the presence of serum albumin in tear film. ²⁰  

To confirm the localization within bovine corneal stroma by an independent method, eight anterior and posterior stromal regions were surgically separated and analyzed for BSA content by immunoblot analysis after normalizing samples to the preextraction water content of the stroma. A typical immunoblot analysis detected a nearly equivalent content of BSA within the anterior and posterior stromal regions (data not shown). However, this method probably did not have the resolving power of the immunohistochemical procedure. Thus, this data did not rule out the existence of a shallow anterior–posterior serum albumin gradient in the stroma of the central cornea (Fig. 5B) . 

Discussion

This study demonstrated that bovine and mouse corneas have a similar abundance and spatial distribution of serum albumin. We found BSA at 10% and MSA at 13% of the water-soluble protein in the respective animals’ corneas. Serum albumin is also a major water-soluble protein in the human cornea. ^{3 5} An earlier study reported an abundant 62-kDa water-soluble protein in bovine cornea, ⁷ which we show in the present study to be BSA, confirming earlier speculation on its identity. Thus, serum albumin is abundant in diverse mammalian corneas. 

The present study provides one reason for the absence of reports on serum albumin in the mouse cornea—namely, that it comigrates with TKT in SDS-PAGE (+DTT). Our findings show that their separation can occur by SDS-PAGE (−DTT), because of a mobility change for serum albumin that has been noted before. ²⁰ Although previous estimates for TKT included MSA, ¹¹ the more rigorous method applied in the current study to estimate TKT proportion confirms that it represents an abundant water-soluble protein in mouse cornea. Unlike serum albumin, TKT and ICDH differed in abundance significantly between bovine and mouse corneas. Our finding that TKT constitutes 23% of the water-soluble protein in mouse corneal epithelial cells, as opposed to 3% in bovine corneal epithelial cells, is consistent with the taxon-specific expression commonly found among abundant corneal water-soluble proteins. ^{21 22} The compensating protein in the bovine corneal epithelial cells appears to be ICDH. ¹⁹  

The present immunohistochemistry and SDS-PAGE analyses confirmed the predominantly stromal localization of serum albumin in bovine cornea ⁴ and extended this observation to mouse cornea. Although a stromal localization for serum albumin is consistent with most earlier reports for mammalian corneas, ^{1 2 4 23} in one immunohistochemical study of human cornea, it was reported that the predominant stain for serum albumin was in the epithelium. ¹ Although our data do not include a human cornea, we have noted a predominantly epithelial stain in both mouse and rat corneas for serum albumin when using paraffin-embedded sections and horseradish peroxidase–amplified staining systems (data not shown). Therefore, we did not use paraffin-embedded sections or amplify signals with enzymatic or biotin-avidin systems in the present study. Because this epithelial staining in paraffin-embedded sections was not dependent on the choice of primary antibody or the species of cornea, it seems unlikely to represent an artifact of antibody specificity. Although staining of the paraffin-embedded sections does not reflect the appropriate relative levels of serum albumin between corneal epithelium and stroma, as confirmed by SDS-PAGE on isolated epithelia and stroma, we cannot exclude low levels of serum albumin in the epithelium that are difficult to detect by the methods we used in this study. 

Immunohistochemical staining for serum albumin showed a spatial distribution between collagen lamellae of the bovine and mouse corneal stroma, indicative of an extracellular protein. The staining was visible throughout the entire corneal stroma and was not associated specifically with the keratocytes. Further, the striated staining varied in concert with the direction of the collagenous lamellae (Figs. 4 5) . This pattern of staining was present throughout these corneas, except that bovine cornea had an elevated staining at the anterior peripheral stroma and adjacent to Descemet’s membrane. Higher peripheral concentration of serum albumin is consistent with its diffusing from peripheral blood vessels. ^{1 2} The distribution of serum albumin in the stroma in the current study appeared similar to that of the serum proteins IgG and IgA in human cornea. ^{23 24} It is interesting that confocal microscopy of bovine sections revealed a population of basal epithelial cells that stained for BSA, giving them a rocket-shaped appearance. Thus, some epithelial cells may endocytose stromal serum albumin. Although the frequency of the serum albumin–stained cells in bovine cornea appeared low, such cells may contribute to the epithelial staining for serum albumin in paraffin-embedded sections of rat and mouse, as discussed earlier. Endocytosis is consistent with the rate of serum albumin loss from rabbit cornea that indicates a vesicle transport mechanism. ²⁵  

Although the function of serum albumin in the cornea is unresolved, it seems reasonable to explore its known roles in plasma that include binding of toxic metabolites, osmotic balance, and transport of fatty acids. ²⁶ Also, serum albumin may serve a cornea-specific function. It could filter UV light ⁷ or affect the optical properties of the cornea. ²⁷ Serum albumin may also be an antioxidant reacting with free radicals ⁵ in a function similar to that proposed for ALDH3. ²⁸ Furthermore, it is interesting that serum albumin has a chaperone-like activity. ²⁹ Even trace amounts of serum albumin in reperfusion experiments prevent damage to rat hearts that have been subjected to ischemia. ³⁰ A chaperone function for serum albumin could be very useful in preserving the native state of long-lived extracellular proteins in the stroma under conditions of continual exposure to physiological stress. Indeed, the α-crystallins of the lens are small heat shock proteins with chaperone activity critical for preserving the solubility of partially denatured proteins that accumulate during aging. ³¹  

That serum albumin is an abundant component of the stroma raises several issues for corneal hormone function and drug therapy. First, serum albumin binds many lipophilic hormones and drugs and it has been suggested that it could act as a general hormone carrier or that it protects their receptors by buffering against xenobiotics. ¹³ If serum albumin acts as a hormone carrier, knowledge of its high abundance in corneal stroma and its spatial distribution could be important in the understanding of hormone action and drug–hormone interactions in general. Because the cornea represents a relatively exposed tissue, the need for protection of hormone receptors from xenobiotics also is plausible. Second, only small lipophilic drugs generally are capable of penetrating the corneal epithelium, and some of these drugs encounter their rate-limiting barrier to further penetration of the eye at the stromal layer. ³² The stromal abundance of serum albumin and its capability of binding many small lipophilic drugs may contribute to this stromal barrier. If so, these drugs, on delivery, could concentrate inside the abundant pool of serum albumin in the stroma, even when they have a fairly low affinity. ¹⁴ As a general example of the significance of this issue, the release of bound drug occurs rapidly when a second drug is introduced with a higher affinity for serum albumin, and this sudden drug load can be life threatening with certain anticoagulants in the serum. ^{14 15}  

Finally, the data in the present study suggest that serum albumin may serve as a drug carrier to the cornea. Topical eye drops are limited to the delivery of small lipophilic drugs, and they are only partially absorbed because of the corneal barrier function of the epithelium and lachrymal drainage. ^{32 33} Thus, an alternative route of drug delivery could be advantageous. Recently, plasma serum albumin has been developed as a carrier for drug targeting in cancer chemotherapy, taking advantage of the general trait of tumors to amass serum albumin. ¹⁶ We propose that this method may be of value for normal cornea, because it also accumulates serum albumin. The 19-day plasma half-life of serum albumin greatly increases the plasma half-life of conjugated small drugs that are quickly cleared from the body as free molecules. ¹⁶ Similarly, a serum albumin–drug conjugate should extend the drug half-life in the corneal stroma. We calculated the half-life of serum albumin in rabbit cornea to be approximately 1.5 to 8 days, by the kinetics of its loss as determined by others. ^{1 2 25} This suggests the important possibility that drug conjugates to serum albumin would accumulate in the corneal stroma even when injected into the body at a more distal location. Targeting may be enhanced further if the drug is tethered with an appropriate photolytic bond ^{34 35} or if a ligand to a corneal specific receptor is included in the conjugate. Although many technical details remain to be explored, our data suggest that mice may be a favorable species in which to test novel methods that use the abundance of serum albumin in the corneal stroma. 

 Supported by the National Institutes of Health Intramural Program.

 Submitted for publication November 14, 2002; revised February 6 and April 2, 2003; accepted April 2, 2003.

 Disclosure: D.W. Nees, None; R.N. Fariss, None; J. Piatigorsky, 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.

 

SDS-PAGE (−DTT) revealed a species migrating at 58 kDa (✶) that was not evident in SDS-PAGE (+DTT), whereas the positions of ALDH3 and TKT (arrows) appeared unchanged (A). Mass spectrometry of the trypsinized 58-kDa species revealed peptide sequences (bold) that covered 43% of the serum albumin amino acid sequence (B).

Figure 2.

 

Membranes were stained (Stain) and then probed with antibodies (Immnunoprobe). Antibodies were rabbit polyclonal against MSA (Anti-MSA), mouse TKT (Anti-TKT), and BSA (Anti-BSA). Water-soluble protein samples were extracted from tissues and treated with (+) or without (−) the reducing agent DTT in SDS-PAGE for these analyses. Amounts of samples loaded were as indicated in the Methods section.

Figure 3.

 

SDS-PAGE analysis (−DTT) was performed with mouse water-soluble corneal extracts E, SE, and S (A). For comparison, SDS-PAGE of 10 μg of water-soluble protein from bovine corneal epithelium (lane E) and from whole bovine cornea (lane C). (B) Major proteins are indicated: A, ALDH3; B, BSA; I, ICDH; M, MSA; and T, TKT. BSA ran more diffusely in the gel system than other proteins, and its position is shown as a range. The identities of the labeled proteins were made by immunoblot analysis (Fig. 2) or taken from determinations reported in the literature. ^{18 19}

Figure 4.

 

Staining of cryosections for MSA revealed its spatial distribution in mouse cornea (red channel) and DAPI nuclear staining (blue channel) localized the corneal structures (A). Mouse cornea did not stain for MSA (red channel) when treated with secondary antibody alone (B). Ep, epithelium; St, stroma. Scale bar, 40 μm.

Figure 5.

 

Staining of cryosections for BSA (red channel) revealed its spatial distribution in peripheral cornea (A), central cornea (B), and a central posterior region that included the endothelial layer and Descemet’s membrane (C). Bovine cornea did not stain for BSA (red channel) when treated with secondary antibody alone (D). An epithelial region contained basal cells that stained for BSA (E). Magnified images of the region with epithelial cells staining for BSA (F). DAPI nuclear staining (blue channel) localized the corneal structures. Structures are labeled as follows: Ep, epithelium; St, stroma; En, endothelium; Dm, Descemet’s membrane; Bm, Bowman’s membrane.

The authors thank Amanda Bundek for technical help with the fluorescence immunohistochemistry, Dionne Davis and R. Steven Lee for help with the mouse colony, and Zbynek Kozmik and Ali Djalilian for helpful ideas in the project and manuscript.