NC1 Long and NC3 Short Splice Variants of Type XII Collagen Are Overexpressed during Corneal Scarring

Abstract

Purpose.: To investigate type XII collagen expression in corneal scars in vivo.

Methods.: Type XII collagen protein expression was evaluated by immunohistochemistry in human corneal scars and in a mouse model of corneal scarring at several time points (from day 7 to day 210) after full-thickness excision. Alternative splice variants of the NC3 and NC1 domains of type XII collagen were investigated in the mouse wound-healing model using RT-PCR.

Results.: Type XII collagen was overexpressed in human corneal scars in areas that were also positive for alpha–smooth muscle actin staining. In a mouse model of corneal wound injury we found that at 14 and 21 days postexcision, type XII collagen was largely concentrated in the subepithelial region of the cornea, especially in and near the wound bed. By 28 days postexcision, expression of type XII collagen decreased but remained higher than that in controls. NC3 short form is the main form expressed in the cornea during the wound-healing process. After injury, the NC1 long splice variant mRNA was the most highly overexpressed variant in the cornea, especially in the epithelium (×2.7, 3.72, and 5.57 at days 7, 14, and 21, respectively, P < 0.01 to 0.001 compared with uninjured samples). Corneal scars from a 7-month-old mouse revealed an overexpression of type XII collagen in the wound area similar to what we observed in human corneal scars.

Conclusions.: Type XII collagen is overexpressed in permanent human and mouse corneal scars and could represent a new target to treat corneal scarring.

Introduction

Corneal scarring represents the third cause of blindness in the world (5.1%). ¹ Corneal visual impairment encompasses a variety of factors such as infectious, inflammatory, trauma, and chemical injury. The only available long-term curative treatment is corneal transplantation. However, although this represents a viable treatment, several side effects make it necessary to research new molecular targets. 

Corneal wound healing is a complex process involving many cell types, signaling pathways, and proteins. To date, precise molecular mechanisms and key effectors that determine the success of the wound-healing process are still unknown. When corneal wound healing fails to repair the tissue and restore transparency, a permanent corneal scar is formed. Previous strategies targeting activated stromal cells, such as myofibroblasts, have been disappointing and were either ineffective or accompanied by several side effects. ^2–7 Recently, our group ⁸ and others such as Mohan et al. ⁹ have demonstrated that modulation of the extracellular environment could target the disorganization and excessive accumulation of connective tissue, which are common traits of corneal scar. We hypothesized that these approaches activate a feedback loop, which by remodeling extracellular matrix (ECM) has an impact on myofibroblast transformation and thus partly reduces corneal scar severity. Greater knowledge of corneal scar ECM composition is needed to identify potential ECM targets. 

Corneal transparency mostly depends on the architecture of the stromal ECM, which includes collagen fibril diameter, packing, and lamellar organization. Collagens are the most abundant proteins of the stroma and appear mainly in the form of heterotypic fibrils that consist of collagen types I and V. The stroma also contains a lesser amount of type VI collagen, the nonfibrillar collagen types XIII, ¹⁰ XVIII, ¹¹ and type XII collagen, a member of the FACIT (fibril-associated collagens with interrupted triple helices) subfamily. Since type XII collagen expression is found to be increased during bleomycin-induced lung fibrosis ¹² and during axolotl limb regeneration, ¹³ we hypothesized a potential involvement of this collagen during corneal scar formation. 

Type XII collagen is a homotrimeric molecule composed of two collagenous domains (COL1–COL2) and three noncollagenous domains (NC1 to NC3) ¹⁴ (Fig. 1). This collagen is found to be associated with type I collagen fibrils ^12,15 via its collagenous domain. Its large N-terminal noncollagenous domain (NC3) then interacts with other matrix components. ¹⁶ This domain may have an attached glycosaminoglycan chain, ¹⁷ and functions by interacting with molecules such as decorin, fibromodulin, ^18,19 or tenascin X. ²⁰ Both the NC1 and NC3 domains of type XII collagen are subject to alternative splicing, which could generate up to four different isoforms of type XII collagen ²¹ (Fig. 1). These isoforms differ from each other in terms of length and the presence of a glycosaminoglycan amino acid linker or ECM interaction domains. Recently, using type XII collagen-null mice, it was demonstrated that type XII collagen could be involved in bone formation and altered cell–cell interactions in osteoblasts. ²² Moreover, type XII collagen could be associated with a disorganized collagen fiber arrangement. ²² Thus, type XII collagen could also help to stabilize fibril organization by forming interfibrillar bridges ¹⁵ along with other ECM components. In the cornea, type XII collagen is expressed in both epithelial basement membrane, Descemet's membrane, and along stromal collagen fibers. ^23–28 In humans, the NC3 long isoform is the most abundant, ^25,27 whereas in mouse both NC3 isoforms are present. ^22,26  

Figure 1. 

 

Schematic representation of the modular structure of type XII collagen and the different NC3 and NC1 alternatively spliced transcripts.

During corneal wound healing, a provisional matrix made of type I/III collagen is deposited. ^8,29,30 This temporary matrix is mostly responsible for corneal opacity at this stage. ³¹ When the cornea does not heal properly, the ECM associated with the scar still contains a high amount of type I/III collagen. ^8,30 Whether type XII collagen is associated with this specific ECM is unknown. A previous study has shown type XII collagen expression in subepithelial and retrocorneal fibrous proliferative areas of bullous keratopathy corneas, ³² whereas its mRNA was enhanced in a photorefractive keratectomy rabbit model. ¹⁶ Altogether, these observations lead us to hypothesize a potential involvement of this collagen during corneal scar formation. Thus, we investigated five human permanent corneal scars and three control human corneas; moreover, we showed that type XII collagen protein was overexpressed in stromal scar areas containing myofibroblasts. We further analyzed type XII collagen protein expression and mRNA splice variants regulation in a mouse model of corneal wound healing and we discuss its putative role in corneal wound healing and scarring. 

Materials and Methods

Biological Samples

Eight corneas were included in this study: five human corneas presenting permanent stromal scars were collected after transplantation, two normal uninjured human corneas rejected for transplantation (low endothelial cell density), and one normal human cornea obtained after enucleation for melanoma were also collected. Control corneas were processed within 6 hours after death and immediately after enucleation. Table 1 summarizes clinical conditions and parameters of the different corneas. 

Table 1. 

 

Description of Human Corneas

Number	Age	Sex	Etiology	Age of Scar, y	Scar Characteristics
P1	56	M	Physical trauma (glass), neo-vascularization	2	Central opacity, half of the stromal depth
P2	45	F	Physical trauma (vegetal trauma), partial ulceration	2	Complete stromal opacity, no perforation
P3	40	M	Physical trauma (barbed wire)	1.5	Central opacity, all stromal depth, perforating trauma
P4	22	F	Herpetic keratitis	5	Central anterior stroma opacity
P5	74	M	Herpetic keratitis	2	Complete stromal opacity
C1	75	M	No history of corneal pathology
C2	55	F	No history of corneal pathology
C3	34	M	No history of corneal pathology

Corneas were embedded in optimal cutting temperature (O.C.T.) compound (Tissue-Tek; Sakura Finetek Europe B.V., Zoeterwoude, The Netherlands) just after surgery. Corneal samples were intended for destruction after surgery. This study respects the tenets of the Declaration of Helsinki. 

Female C57BL/6 mice at 10 to 12 weeks of age, obtained from Charles River Laboratory (Arbresle, France), were used in these experiments. All experimental procedures were approved by the Ethical Committee of the Centre of Physiopathology Toulouse Purpan and conducted in accordance with the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research. 

Surgical Technique

Surgery was performed on the left eyes of mice by the same surgeon (P.F.) under general (2.78 mg/kg of ketamine + 1.67 mL/kg of Rompun 2% xylazine by IP injection; Bayer Animal Health, Leverkusen, Germany) and topical (drop administration of 1.6 mg Oxybuprocaine Faure/0.4 mL Collyre [eye ointment]; Novartis Ophthalmics AG, Basel, Switzerland) anesthesia. Mice were treated with atropine sulfate 1% ophthalmic solution (Alcon Laboratories, Inc., Fort Worth, TX) before surgery and then for twice a day for 1 week. We performed a full-thickness 1-mm excision of the central cornea using a trephine and ophthalmologic microsurgical scissors under a stereomicroscope, as previously described. ⁸ After surgery, antibacterial ointment (Fucithalmic Ointment; Leo Pharma, Ballerup, Denmark) was topically applied to prevent bacterial infection. 

The animals were then euthanized at specific intervals of healing (day [D] 7, D14, D21, D28, and D210). Each left eye was removed, embedded in optimal cutting temperature compound (Tissue-Tek; Sakura Finetek Europe B.V.) for histology procedures. Alternatively, the left eyes were dissected to collect separately the corneal epithelium and stroma for RNA extraction and subsequent RT-qPCR analysis (Table 2).1552  

Table 2. 

 

Number of Samples per Experiment

Methods	Time Point	Number of Samples
Immunohistochemistry	Uninjured	3
	D7	7
	D14	7
	D21	7
	D28	10
	7 months uninjured	2
	7 months injured	5
RNA (stroma)	Uninjured	41
	D7	19
	D14	20
	D21	21
	D28	12
RNA (epithelium)	Uninjured	36
	D7	10
	D14	9
	D21	10
	D28	11

Immunohistochemistry

Type XII Collagen Staining.

KR33, a rabbit anti-mouse antibody recognizing NC3 short and long forms of type XII collagen; KG52, a guinea pig anti-mouse antibody recognizing NC3 long form type XII collagen; and KR75, a rabbit anti-human antibody recognizing NC3 short and long forms of type XII collagen used in these experiments were previously generated by Manuel Koch (University of Cologne, Cologne, Germany). Cryostat sections (7 μm thick) of cornea (human and mice) were immunoreacted using an indirect immunofluorescence technique. Briefly, samples were fixed in methanol at −20°C for 15 minutes. All antibodies were used at a concentration of 0.1 ng/mL. Incubation was performed in a moistened chamber at room temperature for 120 minutes. Corresponding secondary antibodies were used (goat anti-rabbit IgG, goat anti-guinea pig IgG, coupled to AlexaFluor 546; Invitrogen/Molecular Probes, Eugene, OR) at 1:400 dilution, then incubated at room temperature in a moistened dark chamber for 60 minutes. Antibodies were diluted using phosphate-buffered saline (PBS), 0.3% bovine serum albumin (1× PBS, 0.3% BSA). Rabbit and guinea-pig isotype control (normal immunoglobulin G [IgG]) were used as the first antibody in control samples. All experiments were carried out under the same conditions: antibody controls and staining of wounded cornea were carried out at the same times. Final mounting of tissue sections was done with antifade reagent with DAPI (4′,6-diamidino-2-phenylindole) (ProLong Gold; Invitrogen/Molecular Probes). Sections were observed 24 hours later, using ×20 or ×40 objective (Leica DMR microscope; Leica Microsystems GmbH, Wetzlar, Germany) and images were acquired and processed (Leica DFC 300 FX camera and IM50 Software; Leica Microsystems). Exposure times were determined from the control conditions (nonwounded) and applied to the other samples (mouse sample: ×20 objective, DAPI 70 ms, red channel 600 ms; human sample: ×20 objective, DAPI 150 ms; red channel 150 ms). A negative control condition was performed on each slide. 

Alpha–Smooth Muscle Actin (αSMA) Staining.

Cornea immunostaining was performed with the same adapted protocol on 7-μm-thick cryostat sections. However, samples were fixed in 4% buffer paraformaldehyde for 20 minutes (32% stock solution; Electron Microscopy Sciences, Hatfield, PA). The antibody, used at 1:300 dilution, was a rabbit monoclonal IgG antibody (1184‐1; Epitomics, Burlingame, CA) revealed with secondary antibody (Alexa Fluor, 546 nm goat anti-rabbit IgG; Invitrogen/Molecular Probes) at dilution 1:400. Exposure times were: mouse sample: ×20 objective, DAPI 70 ms, red channel 200 ms; human sample: ×20 objective, DAPI 150 ms; red channel: 150 ms. 

Immunoblotting Analysis

Whole unwounded corneas were removed and lysed as previously described. ²² KR33 (1:5000 dilution) and KG52 (1:1000 dilution) antibodies were compared using 5% SDS-PAGE under reducing condition; 15 μg of total protein extract was used. Anti-rabbit and anti-guinea pig (Cell Signaling Technology, Danvers, MA, and Thermo Scientific, Rockford, IL, respectively) horseradish peroxidase–conjugated secondary antibodies were used at 1:1000 dilution. 

Reverse Transcription and Quantitative Real-Time Polymerase Chain Reaction (RT-qPCR)

Total RNAs from mouse corneal epithelium and corneal stroma samples were extracted using a commercial kit (RNeasy Micro Kit; Qiagen, Valencia, CA) according to the manufacturer's protocol. The quantity and quality of extracted RNA were verified using a commercial kit (Agilent RNA 6000 Pico Kit; Agilent Technologies, Waldbronn, Germany). Purified RNA was stored at −80°C until analysis. Reverse transcription was performed with a commercial cDNA synthesis kit (SuperScript VILO cDNA Synthesis Kit; Invitrogen, Carlsbad, CA) according to the manufacturer's recommendations. The resulting cDNA was stored at −20°C until analysis. qPCR was performed using a real-time system (LightCycler 480 Real-Time System; Roche Diagnostics Belgium, Vilvoorde, Belgium) with the following protocol: 5 minutes at 95°C (initial denaturation), followed by 40 amplification cycles consisting of 15 seconds at 95°C (denaturation), 10 seconds at 60°C (hybridization), and 45 seconds at 72°C (elongation), then 40 seconds for melting temperature determination and 30 seconds at 40°C for cooling down. Primers used are listed in Table 3 and were designed on the basis of previous publications ^21,33 (Primers Express software; PE Applied Biosystems, Foster City, CA). The long NC1 domain was flanked by NC1-L primers and the short NC1 domain by NC1-S primers. NC3-L primers target the long NC3 domain, whereas NC3-S+L target both the long and the short forms of the NC3 domain of type XII collagen. PCR efficiency was determined for each primer set to calculate the expression ratio. Normalization was performed using three housekeeping genes, as previously described for the stroma. ⁸ Concerning the epithelium, we used three other housekeeping genes: ubiquitin C (Forward [F] primer: AAC CCA CAG TAT ATC TTT GGCG; Reverse [R] primer: CCC TCA CTA GGT TCG ATG ACT TC), tatabox-binding protein (F: TGC CGA AAG ATG CAC AGA TGA; R: TGT TGT CAC ATA TCG GAA GGC), and beta actin (F: CGG TCC ACC CGC CAC CAG TTC GCCA; R: TCC CAC CAT CAC ACC CTG GTG CCTA). The fold-change in gene expression was calculated using the 2^ΔΔCT ratio using a previously described method. ³⁴ All PCR products were checked by sequencing (Millegen, Toulouse, France).1552  

Table 3. 

 

Primer Sequences Used for the Detection of Type XII Collagen Splice Variant mRNA

Name (Ensembl Name)	Transcript ID ENSMUST	Forward Primer	Position Domain (Exon)	Reverse Primer	Position Domain (Exon)
NC1-L (Col12a1-003)	00000135009	GGTCCACCAGGGTCTACAGG	COL1 (Exon 65)	GGCACATAAGGCTCTGGATAGC	NC1 (Exon 65–66)
NC1-S (Col12a1-001)	00000071750	GGTCCACCAGGGTCTACAGG	COL1 (Exon 65)	TTAGCCGGAACCTGGATAGC	NC1 (Exon 65–66)
NC3-L (Col12a1-001)	00000071750	AAGTTTACCACTCGTTGGGCA	FN8 (Exon 17)	GAGAAATGAAGCTTCGCACAGT	vWA3 (Exon 18)
NC3-S+L (Col12a1-001)	00000071750	AGACATTGTGTTGCTGGTGGA	vWA3 (Exon 18)	GAGAAATGAAGCTTCGCACAGT	vWA3 (Exon 18)

 Protein domain positions of these primers are indicated.

Statistical Analysis

For all experiments, group-to-group comparisons were performed using a nonparametric Wilcoxon test. Significance was set at P < 0.05. 

Results

Expression of Type XII Collagen Protein in Human Corneal Scars

We investigated type XII collagen expression in five human permanent stromal scars and in three normal uninjured corneas. Antibody KR75 detected both short and long NC3 forms of type XII collagen. αSMA staining was performed at the same time to distinguish the wound area. αSMA apparent nonspecific staining could be observed in the epithelial layer of some samples. 

All observations performed on the human control corneas were the same; thus, we present pictures only of the C1 sample. Control uninjured corneas were negatively stained for αSMA (Figs. 2A, 2C). Type XII collagen was found expressed in the epithelial basement membrane (Fig. 2B, arrow) and all over the stroma (Figs. 2B, 2D). Type XII collagen basement membrane staining was also found in all traumatic corneas in which preserved Bowman's layer was found (data not shown). 

Figure 2. 

 

Immunofluorescence staining of KG75 (recognizing long and short NC3 isoforms of type XII collagen) in control human cornea. αSMA staining (red, [A, C]). KG75 staining (red, [B, D]). Nuclear DAPI counterstaining in blue. (A, B) Epithelium and anterior stroma; arrow points toward basement membrane, arrowhead points toward Bowman's layer. (C, D) Central stroma. Bar: 50 μm.

P1, P2, and P3 corneas suffered from mechanical traumas that induced partial or total opacities. 

P1 cornea suffered from nonperforating central wound, and thus presented a central opacity. We observed an αSMA staining in this central area, from superficial (Fig. 3A) to deep stroma (Fig. 3C). These regions were strongly stained for type XII collagen (Figs. 3B, 3D), with a high reactivity in the neobasement membrane area (Fig. 3B, arrow), characterized by loss of Bowman's layer. Others areas of the stroma, negatively stained for αSMA (Fig. 3E), were also less reactive to type collagen XII than the wound area (Fig. 3F). 

Figure 3. 

 

Immunofluorescence staining of KG75 (recognizing long and short NC3 isoforms of type XII collagen) in Patient 1 cornea (physical trauma). αSMA staining (red, [A, C, E]). KG75 staining (red, [B, D, F]). Nuclear DAPI counterstaining in blue. (A, B) Epithelium and anterior stroma; arrow points to neobasement membrane. (C–F) Central stroma. Bar: 50 μm.

P2 cornea was characterized by a global stromal opacity. αSMA staining was found throughout the stroma (Figs. 4A, 4C), as well as type XII collagen staining (Figs. 4B, 4D). The epithelial basement membrane was regularly stained by type XII collagen antibodies (Fig. 4B, arrowhead), as well as the reformed basement membrane, which was fragmented and irregular (Fig. 4B, arrow). 

Figure 4. 

 

Immunofluorescence staining of KG75 (recognizing long and short NC3 isoforms of type XII collagen) in Patient 2 cornea (physical trauma). αSMA staining (red, [A, C]). KG75 staining (red, [B, D]). Nuclear DAPI counterstaining in blue. (A, B) Epithelium and anterior stroma, head arrow points to preserved basement membrane, arrow points to reformed irregular basement membrane. (C, D) Central stroma. Bar: 50 μm.

P3 cornea suffered from a central perforating wound, and presented with a central opacity, which was positively stained for both αSMA and type XII collagen from top to deep stroma (Figs. 5A, 5B), whereas uninjured areas were negative to αSMA and less reactive to type XII collagen. As in previous cases, increased type XII collagen staining could be observed in the epithelial neobasement membrane area (Fig. 5B, arrow). 

Figure 5. 

 

Immunofluorescence staining of KG75 (recognizing long and short NC3 isoforms of type XII collagen) in Patient 3 cornea (physical trauma). αSMA staining (red, [A]). KG75 staining (red, [B]). Nuclear DAPI counterstaining in blue. Arrow points to neobasement membrane. Bar: 100 μm.

P4 and P5 corneas suffered from herpetic keratitis. Although of a different etiology from that of traumatic scars, postinfectious healing processes left the corneas with scars similar to those induced by physical trauma. P4 cornea presented a central opacity just underneath the epithelium; we observed restricted central anterior stromal staining for αSMA (Fig. 6A), which colocalized with a type XII collagen staining. P5 cornea presented with a full-thickness scar; we noticed a homogeneous and bright αSMA staining localized in all stroma layers. The stroma was strongly stained by type XII collagen antibodies (Figs. 6B, 6D). 

Figure 6. 

 

Immunofluorescence staining of KG75 (recognizing long and short NC3 isoforms of type XII collagen) in Patients 4 and 5 (herpetic keratitis). αSMA staining (red, P4: [A]; P5: [C]). KG75 staining (red, P4: [B], P5: [D]). Nuclear DAPI counterstaining in blue. Bar: 50 μm.

Spatiotemporal Variation of Expression of FACIT Type XII Collagen Protein in Adult Mouse Cornea after 1-mm Central Excision

We assessed the localization of type XII collagen protein in normal adult mouse cornea and investigated its variation during corneal wound healing by immunohistochemistry analysis. Two antibodies directed toward NC3 domains were used: one that recognized both isoforms of NC3 long and short type XII collagen (KR33) and, as such, could be considered to stain all isoforms of type XII collagen, and one that recognized only the long NC3 isoform (KG52). These antibodies have been validated by immunoblotting, which revealed that both NC3 long and short isoforms were expressed in unwounded mouse cornea (see Supplementary Material and Supplementary Fig. S1). The injured area was identified by αSMA staining (Figs. 7A, 7C, 7E, 7G, 7I). 

Figure 7. 

 

Immunofluorescence staining of KR33 (long and short NC3 isoforms of type XII collagen) during murine corneal wound repair. αSMA staining (red, [A, C, E, G, I]), and KR33 staining (red, [B, D, F, H, J]). Nuclear DAPI counterstaining in blue. (A, B) Unwounded cornea. (C, D) D7 postexcision. (E, F) D14 postexcision. (G, H) D21 postexcision. (I, J) D28 postexcision. (K) αSMA negative isotype control. (L) KR33 negative isotype control. Arrowheads point toward neo-Descemet's membrane. (B′, D′, F′, H′, J′) View of the boxed region from corresponding panel. Arrows point toward neoepithelial basement membrane. Bar: 25 μm.

We first present staining obtained with the KR33 antibody. Unwounded cornea expressed a low level of type XII collagen (Fig. 7B). At D7 after injury, type XII collagen staining was slightly stronger in the wound area, and associated with stromal cells (Figs. 7D, 7D′). Close to the wound area, the staining was also a bit stronger compared with control cornea (see Supplementary Material and Supplementary Fig. S2B). At D14, half of injured mouse corneas still exhibited the same profile as at D7, whereas the other mice started to show a reconstructed region (Fig. 7F). Both in the wound area and near the wound bed, we could still observe a greater staining than that in the uninjured mice, but not stronger than that at D7 (Fig. 7F and Supplementary Fig. S2C). Neobasement membrane staining began to appear in both epithelial basement membrane (Fig. 7F′, arrow) and Descemet's membrane (Fig. 7F, arrowheads). These observations were confirmed at D21: almost all mice exhibited a strong type XII collagen staining in the stroma (Fig. 7H). Neobasement membrane areas were clearly outlined by the antibodies (Figs. 7H, arrowheads, and 7H′, arrows). In the same way, stromal staining near the wound bed was also strongly increased compared with control and D7 and D14 (see Supplementary Material and Supplementary Fig. S2D). At D28, most of the mice showed a decreased expression of type XII collagen, but this was still stronger than control samples both in the wound bed (Fig. 7J) and close to it (see Supplementary Material and Supplementary Fig. S2E). However, neobasement membrane staining was now barely visible, and more diffuse (Figs. 7J, 7J′, arrows). A few mice also exhibited a weak but homogeneous αSMA staining (see Supplementary Material and Supplementary Fig. S3A) associated with a reduced type XII collagen staining (see Supplementary Material and Supplementary Figs. S3B, S3C), whereas one mouse exhibited low αSMA staining (see Supplementary Material and Supplementary Fig. S3D) associated also with a reduced type XII collagen staining (see Supplementary Material and Supplementary Figs. S3E, S3F). 

Expression of NC3 long isoform of type XII collagen was performed using KG52 antibody. The corneal structures revealed by the antibody were the same as those revealed with the KR33 antibody. Weak expression was observed in uninjured samples (see Supplementary Material and Supplementary Fig. S4A). Maximal expression was observed at D7 (see Supplementary Material and Supplementary Figs. S4B, S4C), then it started to decrease at D14 and D21 (see Supplementary Material and Supplementary Figs. S4D–G). By D28, NC3 long isoform of type XII collagen expression was returned to preinjured levels (see Supplementary Material and Supplementary Figs. S4H, S4I). 

Expression of the Four Type XII Collagen mRNA Splice Variants in Mouse Cornea during Wound Healing

Four splice variants of type XII collagen were previously described. Thus, during this study we searched for the presence of corresponding mRNAs to these splice variants in adult mouse corneal epithelium and corneal stroma. We also assessed how their expression levels changed in these regions during corneal wound healing. Total mRNA was isolated from corneal epithelium and stroma on day (D) 7, D14, D21, and D28 after injury and the mRNA expression of the four type XII collagen splice variants was examined by RT-qPCR. Expression levels were normalized using three housekeeping genes and expressed as fold changes in expression compared with noninjured samples. 

Our data revealed that all of these four collagen XII mRNAs splice variants were detected in uninjured mouse corneal epithelium and corneal stroma samples (Fig. 8). 

Figure 8. 

 

Epithelial (A) and stromal (B) fold change in expression (±SE) of NC3 and NC1 splice variants of type XII collagen during murine corneal wound repair. Fold changes in gene expression were calculated in comparison with levels observed in unwounded epithelia (A) or unwounded stroma (B). *P < 0.05

Expression of Type XII Collagen NC1 Splice Variants (NC1-L and NC1-S) in Mouse Corneal Epithelium and Stroma after Injury

NC1-L expression was significantly increased in the epithelium by a factor of 2.68 ± 0.77, 3.72 ± 1.23, and 5.57 ± 2.06, respectively, for D7, D14, and D21 after excision (P < 0.01, Fig. 8A). 

In the stroma, NC1-L was significantly increased by a factor of 1.68 ± 0.21 and 1.68 ± 0.25, respectively, for D21 and D28 (P < 0.05, Fig. 8B). 

Concerning expression of the NC1-S variant in the epithelium, we found it increased by a factor of only 2.53 ± 2.06 at D21 postinjury (P < 0.05, Fig. 8A). 

In the stroma, we did not observe any statistically significant changes in expression of the NC1-S mRNA variant (Fig. 8B). 

Expression of Type XII Collagen NC3 Splice Variants (NC3-L and NC3-L+S) in Mouse Corneal Epithelium and Stroma after Excision

We observed no differences in the expression of the NC3-L variant either in the epithelium (variation was almost statistically significant at D14 and D21, P < 0.1), or in the stroma during corneal wound healing (Figs. 8A, 8B). In contrast, the NC3-S+L variant demonstrated a significant increase in the epithelium on D14 and D21, respectively, by a factor of 2.35 ± 0.76 and 2.29 ± 0.62 (P < 0.05, Fig. 8A). In the stroma, we observed a significant decrease in its expression on D21 and D28, respectively, by a factor of 0.81 ± 0.09 and 0.83 ± 0.09 (P < 0.05, Fig. 8B). 

Expression of Type XII Collagen Protein in Permanent Murine Corneal Scar

We collected corneas from mice 7 months after injury. These mice carried permanent scars. Except for corneal opacity, these mice did not present any other clinical signs (no synechia, no cataract). We evaluated mice presenting grade 1 and grade 3 opacities as previously described. ⁸ To avoid potential artifactual staining due to aging, we also investigated two corneas from mice of the same group, which were not wounded. Nonwounded corneas did not express αSMA and expressed low levels of type XII collagen (Figs. 9A, 9B). In grade 1 opacity, low αSMA was found (Fig. 9C), whereas type XII collagen staining (KR33) was slightly increased in the upper part of the stroma. Subepithelial (arrow) and endothelial (arrowhead) areas were also positive (Fig. 9D). In grade 3 opacity, we observed a stronger staining than that in grade 1 for both αSMA (Fig. 9E) and type XII collagen (Fig. 9F). Subepithelial (arrows) and endothelial (arrowhead) areas were also positive (Fig. 9F). 

Figure 9. 

 

Immunofluorescence staining of KR33 (long and short NC3 isoforms of type XII collagen) in murine corneal scars. Corneas were investigated 7 months of injury. αSMA staining (red, [A, C, E]). KR33 staining (red, [B, D, F]). Nuclear DAPI counterstaining in blue. Unwounded cornea (A, B). Grade 1 opacity (slight opacity that did not interfere with the visualization of fine iris detail) (C, D). Grade 3 opacity (complete opacification of the stroma in the excision area). Scale bar: 12.5 μm. Arrows point toward neoepithelial basement membrane; arrowheads point toward neo-Descemet's membrane.

Discussion

The purpose of our study was to investigate the expression of a class of collagen protein that could be involved in corneal scar formation and stabilization. We demonstrated that FACIT type XII collagen protein was overexpressed in human corneal scars. 

Besides their association with fibrillar collagens, their putative role in stabilization of these fibers during embryogenesis ^24,28 and their association with a disorganized collagen fiber arrangement phenotype during bone formation, ²² little is known about the function of type XII collagens in adult tissues, especially in the context of wound healing. In the cornea, only one study from El-Shabrawi et al. ¹⁶ describes changes in type XII collagen mRNA expression, which are increased following an incision wound in the rabbit. However, they used whole cornea mRNA and did not investigate the different mRNA splice variants, nor did they perform protein staining. Interestingly it has been reported that type XII collagen was expressed during embryogenesis in part of the stroma where collagen fibers showed thermal and collagenase resistance. ^35–37 These observations suggest different ECM properties where the type XII collagen is abundant. Moreover, type XII collagen expression has also been reported in bleomycin-induced lung fibrosis ¹² and corneal fibrosis. ³² We extrapolated these observations to permanent corneal scars and hypothesized that type XII collagen could contribute to the formation and maintenance of the scar extracellular matrix. 

Using an antibody targeting both short and long NC3 isoforms, we investigated normal uninjured human corneas. We observed a slight staining all over the stroma, similar to what has been previously described in man and other species. ^{23,24,26–28} We also confirmed a linear staining in the epithelial basement membrane, as previously shown in rabbit and man. ^23,27 Furthermore, we analyzed traumatic corneas presenting permanent scars that were positively stained for αSMA, as previously described. ^10,38 In our data set, we observed that old corneal scar areas were overexpressing type XII collagen, compared with unwounded parts of the cornea, which revealed a slight staining throughout the stroma. Interestingly, in epithelial areas where the Bowman's layer has been preserved, we found a similar staining in the basement membrane like that in uninjured corneas. We also observed a staining of the reformed basement membrane area in the epithelial–stromal junction. In addition, we observed an apparently nonspecific labeling of the epithelium using αSMA antibodies, which is probably due to tissue preparation, because it has already been observed by others using frozen sections. ^39,40  

Similar observations have been made in our mouse model of corneal excision. In the unwounded cornea, we observed a weak expression of type XII collagen in the stroma, and no expression in the epithelium or the endothelium, confirming a previous study. ²⁶ Following corneal injury, type XII collagen expression was increased, with a different spatial organization over time. Thus, we observed an increased expression of type XII collagen starting at D7 in all the stromal layers, with a stronger staining in the subepithelial and subendothelial parts. The staining gained in intensity in these interface areas between D14 and D21, and started to decrease at D28. Previous studies have shown that the type XII collagen gene contains a mechanical strain response element ^41,42 ; thus, we can hypothesize that these areas support an increased mechanical load to compensate for the central injury. It has been suggested in corneal development studies that type XII collagen could be involved in collagen fiber compaction. ^24,36,43 Wound repair resolution also implies stromal collagen fiber deposition and compaction in the reconstructed area, similar to what occurs during stroma formation in the developing embryo. ²³ This could explain the increased expression that we observed all over the stroma during the healing process. 

Investigating NC3 type XII collagen isoforms involved in the repair process, we concluded that expression of the short NC3 isoform was increased over time with a maximum at D21 postexcision, whereas the long NC3 isoform had its maximum expression at D7, and then progressively regressed to uninjured expression level by D28. mRNA analysis partially confirmed these findings, given that primers targeting both the short and long NC3 variants showed a significantly increased expression in the epithelium on D14 and D21. With a second set of primers targeting only the long NC3 domain, we observed a tendency of this variant to increase in the epithelium on D14 and D21, but this did not reach statistical significance. These results correlate with our immunostaining experiment because we observed an important expression of type XII collagen in the epithelial basement membrane revealed with the antibody KR33 recognizing short and long isoforms. The anti-long NC3 isoform antibody KG52 also stained this area but to a lesser extent. Finally, at D28 postinjury, we observed that the mRNA level was back to control level for both long and short variants of NC3, and that in immunofluorescence, there was no longer any staining of the basement membrane by KG52, whereas KR33 staining was greatly reduced. Interestingly, a similar observation was found during embryonic corneal development. Gordon et al. ²⁴ showed that during avian corneal embryonic development there was a dramatic increase of the short NC3 isoform in the epithelium around day 15. Furthermore, these authors and others ^35,44 have suggested that type XII collagen may play a role in the stabilization of fibrillar collagens localized in the matrix adjacent to the corneal epithelium. 

However, concerning the results on the stroma, some questions still arise. The expression of short and long NC3 variants decreased on D21 and D28, although expression of long NC3 mRNA was unchanged. Thus, whereas mRNA expression is either stable or reduced over the healing process, we observed an increased staining with both KR33 and KG52 in the stroma. Most likely, collagen protein deposition is stable over time. Even if a statistically reduced expression of mRNA was observed; it may not be biologically relevant. Moreover, mRNA half-life changes during the wound-healing process have not been described. Future in vitro experiments will help to elucidate this point. Finally, we observed that some mice expressed low levels of type XII collagen by D28 together with a reduced expression of αSMA. These mice represented a subset that will present low-grade opacities (approximately 25% of the group) or will totally regenerate (<10%). Increased number of samples kept after D28 will help to understand these observations. 

Regarding the NC1 domain, we observed increased expression of the long variant in both the epithelium and the stroma, although expression of the short NC1 variant was mostly unchanged. To date, this study is the first to report a differential expression of these variants in vivo, which is of special interest because the NC1 long variant contains a potential site of interaction with glycosaminoglycans. ²¹ Thus, it will now be interesting to investigate with which other extracellular matrix molecule this domain is interacting. 

In our model, most of the mice did not heal from the wound. Several weeks after injury, we still observed a strong staining revealed with KR33 antibody, which seems to correlate with the intensity of the opacity and αSMA staining. We can ask the question whether this collagen deposit is specific of myofibroblastic cells or if this collagen has not been removed after repair and, as such, may participate in the stromal opacity. 

As a conclusion, in this study, we demonstrated that (1) type XII collagen is overexpressed in human corneal scar areas positively stained for αSMA. Specific human isoforms involved in the scar will need to be determined; (2) following mouse corneal injury, there is a differential expression of type XII collagen isoforms. The NC3 short isoform accumulates in the stroma, whereas the NC1 long variant mRNA is overexpressed in both epithelium and stroma; (3) corneal epithelium could contribute to stromal ECM composition after injury. The precise role of type XII collagen in the neobasement membrane still needs to be understood; and (4) during acute phase of corneal wound repair, a relationship could be proposed with regard to corneal embryogenesis, especially concerning collagen mechanical stability and ECM maturation. However, its applicability to late scar stabilization needs to be investigated. 

This study brings new evidence on the complexity of the role of type XII collagen in modifications to ECM organization in the context of tissue repair and scar formation. The next steps will concern the functional analysis of type XII collagen using mice deficient for collagen XII. Targeting type XII collagen could represent a new strategy that could be developed to treat permanent corneal scars. 

During the revision of this manuscript, a study on proteomic signatures of the desmoplastic invasion front during colorectal cancer metastasis has proposed to consider type XII collagen as a novel candidate marker of myofibroblasts, and/or cancer cells undergoing dedifferentiation. ⁴⁵ This study highlights the link between myofibroblast differentiation and type XII collagen overexpression in pathologic tissue. 

Supplementary Materials

jpg S1 

jpg S2 

jpg S3 

jpg S4 

Acknowledgments

The authors thank David Hulmes for helpful comments on the manuscript; Marie-Andrée Daussion and Jérôme Bernard (Centre de Recherche de Chirurgie Expérimentale Claude Bernard, CHU Purpan, Toulouse, France) for animal care and experiments; and Talal al Saati and Florence Capilla from the histopathology core facility for their technical assistance (Anexplo/GenoToul; UMS 06). 

Footnotes

 Supported by University Paul Sabatier and Région Midi-Pyrénées Grant 08008578 (DM).

Footnotes

⁵  These authors contributed equally to the work presented here and should therefore be regarded as equivalent authors.

Footnotes

 Disclosure: D. Massoudi, None; F. Malecaze, None; V. Soler, None; J. Butterworth, None; A. Erraud, None; P. Fournié, None; M. Koch, None; S.D. Galiacy, None