CHST6-related macular corneal dystrophy: a matter of endothelium

Penetrating keratoplasty contributing to the lowest recurrence rate

A total of 118 MCD patients with 150 eyes who obtained surgical treatment in Qingdao Eye Hospital were included in the study. Recurrence was defined as typical MCD haze extended in the graft button under slit lamp examination regardless of the visual acuity change. As was shown in Table 1, 25% (4/16) of the MCD patients obtaining PTK, 40.91% (9/22) obtaining DALK, and 13.75% (11/80) obtaining PKP were diagnosed as recurrence. The MCD recurrent rate was found associated with the surgical method (P = 0.017, χ² test). The average time for recurrence was 7.00 ± 0.82 years, 2.56 ± 2.71 years, and 6.38 ± 3.34 years for PTK, DALK, PKP, respectively (Table 1). PKP showed the lowest recurrence rate among the three types of surgeries. More than half of the PKP recurrent cases (54.5%) were still normal at 5 years post-surgery while 87.5% of the DALK recurrent cases showed MCD signs again in the first four years post-surgery (Fig. 1A). Slit lamp photo of a typical recurrent case after PKP was presented (Fig. 1B). OCT indicating the PKP case was more likely to gain recurrence from the peripheral host cornea (Fig. 1C).

• Download figure
• Open in new tab

Fig. 1.

Macular corneal dystrophy recurrent cases. (A) Recurrent cases graft survival curve after PTK, DALK, or PKP. A total of 4 PTK recurrent cases, 9 DALK recurrent cases, and 11 PKP recurrent cases were presented. (B) Typical recurrent eye after PKP. Slit lamp photos of the right eye of an MCD patient before PKP in 2007 and after recurrence in 2021. Recurrent corneal opacities could be observed at the edges of the graft. (C) OCT image of the right eye after the recurrence of the same patient.

Expression pattern of CHST6 in human cornea

After searching our human corneal single cell sequencing dataset (Fig. 2A), we found CHST6 was highly expressed in the endothelial cluster rather than in other clusters (Fig. 2B), with the gene detected in around 49.9% of cells in the endothelial cluster and around 2.2% in other cell types. Consistent with the mRNA expression pattern, dark staining of CHST6 protein was present in the endothelium and only sparsely stained CHST6 in the stromal area (Fig. 2C). There were also condensed signals in the basal layer of the epithelium (Fig. 2C). In contrast, MCD cornea gave little expression of CHST6 protein product (Fig. 2D), suggesting the patient carrying a loss-of-function mutation in CHST6. In all, CHST6 is enriched in the human corneal endothelium at both mRNA and protein levels rather than in the stroma.

• Download figure
• Open in new tab

Fig. 2.

CHST6 is enriched in corneal endothelium. (A) Single-cell transcriptomic atlas of the human cornea. Four clusters were separated by unbiased t-SNE clustering and cell types were annotated. (B) t-SNE visualization of CHST6 in the whole corneal atlas. CHST6 was positively marked in most of the endothelial cluster and a small number of stromal cells. (C) Immunohistology staining of CHST6 in normal human cornea. Dark brown staining could be detected in the endothelial layer and the basal epithelia. Stroma contained sparsely distributed light brown staining. (D) Immunohistology staining of CHST6 in MCD patient cornea. No brown staining could be detected in all layers of the diseased cornea.

Knockdown of endothelial Chst5 contributing to mice corneal opacification

Chst5 is the mouse homolog of human CHST6 and the gene showed an expression pattern resemble CHST6 in human cornea, that the endothelial Chst5 was more than 120-fold of that in other corneal layers (n=30, P < 0.01, unpaired t-test) (Fig. 3A). We adopted an intracameral microinjection method (AH) to knockdown the endothelial Chst5 (Fig. 3B). Injection of AAV9-shRNA into the anterior chamber could effectively knockdown the endothelial Chst5 (Fig. 3C). A control shRNA sequence targeting no gene was packaged to the same AAV9 vector serving as the control plasmid. The expression of endothelial Chst5 in the AH knockdown (AH KD) mice was around 36% of the AH control (AH Con) mice. When adjusted to the other parts of the cornea (epithelium+stroma), the efficacy increased to knockdown 70% more endothelial Chst5 in AH KD than in AH Con (n=30, P = 0.015, unpaired t-test). During intracameral injection, the needle went through the corneal stroma as well. To exclude the possibility that the phenotype observed was caused by the leakage of shRNA into the stroma, we constructed an intrastromal control model by directly injecting the AAV9-shRNA into the corneal stroma (Str) (Fig. 3B), which displayed effective knockdown of Chst5 in the stroma (Str KD), but not in the endothelium (Fig. 3C). The endothelial Chst5 was 8.26-fold higher in Str KD than AH KD group while the stromal Chst5 in Str KD was only 15% of AH KD group. When adjusted endothelial Chst5 expression with other corneal layers, Str KD showed the highest while AH KD showed the lowest relative Chst5 expression (Fig. 3C).

• Download figure
• Open in new tab

Fig. 3.

Endothelial Chst5 was related to corneal opacification. For (A) and (B), each dot contains 10 mice and a total of 30 mice were included in each group. (A) The relative expression level of Chst5 in endothelial layer and epithelial + stromal layers of normal mice. (B) Schematic diagram of two ways of AAV9-shRNA plasmids injection. Pink: intracameral injection route; Blue: intrastromal injection route. (C) The fold change of endothelial Chst5 relative to epithelial and stromal Chst5 in four treatment groups. A nonparametric Mann-Whitney U test was applied due to the large standard deviation in Str groups. P values equal to 0.077, 0.017 between AH KD and Str KD, AH KD and Str Con, respectively. P = 0.015 between AH KD and AH Con with unpaired t-test. (D) Broad-beam illumination slit lamp photos of a representative mouse in each treatment group. Only AH KD mice showed corneal opacification while corneas in the other three groups kept transparency. (E) Slit lamp and anterior segment OCT photos of the (D) AH KD mouse. White deposits could be observed in the full thickness of the cornea. Hyper-reflective area could be observed in the central stroma using OCT horizontal line scanning.

Consistent with our hypothesis that MCD corneal haze was related with endothelial Chst5, there were 11.3% (6/53) mice developed MCD-like opacities in the AH KD group, whereas none of the AH Con (0/32), Str KD (0/14), Str Con (0/9) mice developed such opacity (Fig. 3D). The opacity diffused to the full thickness of the corneal stroma (Fig. 3E), similar to that observed in the advanced MCD patients.

Cornea characteristics after endothelial Chst5 knockdown

KS is the major GAG in the cornea and highly sulphated KS is essential to corneal transparency. cGn6ST encoded by CHST6 mediates KS sulphation and sulphated KS is detectable by the monoclonal antibody (18). We stained the KS in our Chst5 knockdown mice. Sulphated KS signal was barely detectable at the opaque area in the AH KD mouse while homogenous staining could be seen in Str KD mice cornea (Fig 4A). Alcian blue is an amphoteric dye reacting with acid mucopolysaccharide and has been used as a classic method to detect the accumulation of acid mucopolysaccharide in MCD stroma (19). Consistent with the antibody staining, AH knockdown cornea but not the Str knockdown cornea showed alcian blue positive staining clusters (Fig 4B). Neither of the control AAV9 plasmids displayed abnormal GAG deposits (Fig 4B). As an alteration of acid mucopolysaccharide content could occur in keratopathy with calcification, we examined the same mice corneas with von Kossa staining for calcified deposits as well. No von Kossa positive staining was identified in the alcian blue positive cornea (Fig 4C). A calcific band keratopathy slide was set as the positive control for the staining (Fig 4C). Finally, we applied transmission electron microscopy (TEM) to characterize the structural changes in the cornea. TEM recognized collagen fibrils with larger diameters with dark staining in the AH KD mouse compared with other groups (Fig 4E), which was also presented in the stroma of MCD patient (Fig 4D). This type of abnormal large collagen fibril was also observed by another group in the posterior stroma of MCD patients (16), indicating our AH KD mouse shared similar corneal stromal characteristics with MCD patients. The average collagen fibril diameter was 32.22 ± 2.13 nm in normal mouse, 34.47 ± 2.28 nm in Str KD mouse, and 39.23 ± 3.69 nm in AH KD mouse, respectively (Fig 4F). Though no corneal opacification was observed in the Str KD group, there were alterations in the stromal collagen fibrils.

• Download figure
• Open in new tab

Fig. 4.

Characterization of Chst5 knockdown corneas. (A) Keratan sulfate staining of the AH KD and Str KD corneas. There was a depletion of KS in the AH KD cornea especially at the central corneal region (white arrowhead, white dash line box), where the AAV9-shRNA plasmids were released in the anterior chamber. No clear lamellar structure existed in the dark area of the AH KD cornea. There was a slightly decline in KS content in some of the Str KD cornea (white dash line box) but the lamellar structure was still intact. (B) Alcian blue staining of mice corneas. Blue deposits were spread throughout the full thickness of the AH KD cornea. No such concentrated positive staining could be observed in the AH Con, Str KD, and Str Con groups. (C) von Kossa staining for calcification in an AH KD cornea and a positive control cornea. No calcified deposit could be observed in the cornea, indicating the positive alcian blue was not caused by calcification. For positive calcification staining, brown-black calcium deposits were aggregated in the anterior stroma. (D) Transmission electron microscopy of MCD patient’s corneal stroma. Abnormally large collagen fibrils could be spotted (white arrowhead). (E) Transmission electron microscopy of mice corneal stroma. (F) Quantification of corneal stromal collagen fibril diameter. AH KD mouse had significant larger collagen fibrils than the other two groups (P < 0.01, One-Way ANOVA, LSD for post hoc tests).

Overexpression of CHST6 R50H in the anterior chamber induced MCD-like phenotype

CHST6 has several key domains, including a sulfotransferase domain spanning amino acid residue 42 to 356, and two sulfate donor (PAPS) binding domains (amino acid 49-55 and 202-210) (7). To examine whether intracameral overexpression of Chst5 with compromised enzyme function would induce MCD phenotype, we selected a CHST6 point mutation identified from our patient that resides within one of the PAPS binding domains (Fig 5A-C). The patient showed bilateral MCD phenotype and OCT indicated a full-thickness abnormality in her corneas (Fig 5A). The patient carried a homozygous CHST6 R50H mutation and both of the patient’s parents carried a heterozygous mutation at the same point (Fig 5B). The R50H was conserved among human CHST6, CHST5 and mouse Chst5 (Fig 5D). Overexpression of the R50H Chst5 by intracameral injection-induced MCD-like stromal opacification similar with Chst5 AH KD mice (Fig 5D), with positive alcian blue staining diffused over the corneal stroma (Fig 5E). Again, no calcification staining was observed (Fig 5E). In summary, both intracamerally knockdown of endogenous Chst5 and overexpression of Chst5 mutation could lead to MCD-like phenotype in the cornea, suggesting a potential role of the endothelium in MCD development.

• Download figure
• Open in new tab

Fig. 5.

Construction of the CHST6 mutant overexpression model. (A) Eye manifestations of a patient with bilateral MCD presented. (B) CHST6 mutation point status in the pedigree. Exome sequencing identified the patient carrying a homozygous point mutation (c.G148A, p. R50H) in CHST6. The patient’s parents carrying heterozygous mutation at the same point and displayed no MCD phenotype. (C) Schematic diagram of the CHST6 protein structure. Two PAPS (3’-phospho-5’-adenylyl sulfate) domains are residing at amino acid positions 49-55 and 202-210 (Marked green). The mutation point R50 was inside the first PAPS domain. (D) Conservation analysis of the amino acid sequence around the mutation point. The amino acid arginine at CHST6 position 50 is conserved among human CHST6, human CHST5, and mouse Chst5 proteins. (E) Eye phenotypes of the mouse with intracameral injection of Chst5 R50H. White deposit area could be observed in the central corneal region and OCT indicated the opacification involved the full thickness of the cornea. (F) Alcian blue and calcification staining of the cornea in (E). Positive alcian blue staining spreading at the anterior and posterior stroma. No calcification was detected.

Materials and Methods Ethics and subjects

This study was conducted under the tenets of the Declaration of Helsinki and approved by the Ethics Committee of Qingdao Eye Hospital with informed consent obtained from the participants. The data used in this manuscript has been approved by the Review Board of Qingdao Eye Institute.

Patients diagnosed as MCD and underwent LKP or PKP between 1992 and 2021 or underwent PTK between 2011 and 2021 at Qingdao Eye Hospital of Shandong First Medical University were reviewed retrospectively.

Single-cell sequencing analysis

Three human whole corneas were acquired and digested as previously described (25). The single-cell suspensions were converted to barcoded scRNA-seq libraries using the Chromium Single Cell 3’ Library and Single Cell 3’ v3 Gel Beads (10× Genomics). Libraries were sequenced on Illumina NovaSeq and mapped to the human genome (build GRCh38) using Cell Ranger software (10x Genomics, version 3.1.0) (26). The feature-barcode matrices were further analyzed by the Seurat R package (version 3.2.2) (27). For each library, cells with detected gene numbers less than 200, and the ratio of mitochondria higher than 0.2 were filtered, and genes with at least one feature count in more than five cells were used for the following analysis. Doublets identified by DoubletFinder (28) were removed. Seurat “NormalizeData” and “ScaleData” were used to normalize and scale the data. 4,000 high variable genes were used to performed linear dimensional reduction. Different libraries were integrated through the identification of cell pairwise correspondences between single cells across datasets. Cells were clustered using the ‘‘FindClusters’’ function at 0.4 resolution. The non-linear dimensional reduction technique, tSNE, was employed to learn the underlying manifold and visualization. Cell types were discriminated by the expression of canonical marker genes for each cluster. Gene expression was visualized on a dimensional reduction plot with “FeaturePlot”.

AAV9 plasmids design and packaging

AAV9 plasmids were packaged by Shanghai GeneChemCo., Ltd. For the knockdown plasmid, shRNA oligos targeting three positions of mouse Chst5 (NM_019950) were designed and inserted into the GV478 vector (U6-MCS-CAG-EGFP) with the BsmBI restriction site. 5’-CGCTGAGTACTTCGAAATGTC-3’ was designed as the knockdown control sequence.

Both the shRNAs plasmids and a Chst5 overexpression plasmid used for knockdown efficiency test were transfected into HEK 293T cells. Proteins were collected and the gene knockdown effects were examined with Western blot against the FLAG tag with anti-FLAG M2 antibody (1:2000, F1804, Sigma-Aldrich) and goat anti-mouse IgG-HRP antibody (1:4000, sc-2005, Santa Cruz). A pair of oligos with the highest Chst5 knockdown efficiency was selected for virus packaging (5’-accggACGCTGCCCTTTGCCAAGATTttcaagagaAATCTTGGCAAAGGGCAGCGTttttt-3’).

For the mutant overexpression plasmid (Chst5 R50H), two primers carrying the mutant nucleotide were designed to generate point mutagenesis (Fm: TCCTCGTGGCACTCGGGCTCGTCCTTCGTGG; Rm: GAGCCCGAGTGCCACGAGGACAGTACCAGC). The full-length Chst5 carrying R50H was inserted into the GV411 vector (CMV-beta globin-MCS-3Flag-SV40 PolyA) between BsmBI and NheI restriction sites.

Construction of the macular corneal dystrophy mice

C57BL/6J mice were randomly allocated for AAV-shRNA injection. Two injection routes were designed: intrastromal and intracameral. 1 µL virus was injected with a 36G NanoFil Syringe (World Precision Instruments Inc, FL, USA).

Immunofluorescence staining

The mouse eye was embedded in the Tissue-Tek O.C.T. compound (Sakura Finetek, Tokyo, Japan), and sectioned into 7 µm thick sections. The sections were stained with KS (Santa Cruz Biotechnology, sc-73518, 1:10) and fluorescein-conjugated secondary antibody (Goat Anti-Mouse IgG H&L, Alexa Fluor 594, Abcam, ab150116, 1:50). Sections were observed under a confocal microscope (Zeiss LSM 880) or Revolve Fluorescence Microscope (Echo) after counterstaining with DAPI.

Paraffin sectioning and immunohistochemistry

Corneas from MCD patients were fixed in 4% paraformaldehyde at 4 °C, embedded in paraffin and sectioned into 4 µm thickness sections. Anti-CHST6 (1:200, ab154332, Abcam) diluted in NCM Universal Antibody Diluent (WB500D, New Cell & Molecular, China) were used as primary antibodies.

Alcian blue staining

The cryosection slides were heated at 60 °C for 15 min and fixed in 4% PFA for 15min, rinsed in 3% acetic acid for 3 min and stained with Alcian Blue solution (pH 2.5, Solarbio, G1560) for 30 min. After washed with distilled water the slides were counterstained with Nuclear Fast Red (Beyotime, C0151) for 5 min, dehydrated in ethanol and mounted.

von Kossa staining

The cryosection slides were heated at 60 °C for 15 min. The slides were covered with von Kossa solution (Servicebio, G1043) for 40 min with exposure to ultraviolet light. After rinsing with distilled water the slides undergoing the hematoxylin and eosin stain procedure.

Transmission electron microscopy

Samples were trimmed to less than 1 mm³ and were fixed with 2.5% glutaraldehyde in 100 mM Sorensen’s phosphate buffer (pH 7.0) at 4 °C for no less than 4 hrs. Samples were then washed with 0.1 M Sorenson’s phosphate buffer pH 7.2 and had a secondary fixation of 1% osmium tetroxide. Samples were dehydrated sequentially in 70%, 80%, 95% and 100% acetone. Samples were then cleared in propylene oxide and infiltrated sequentially in Epoxy 812 embedding medium (Sigma-Aldrich): propylene oxide mixtures at ratios of 2:1 for 0.5 hr at room temperature, of 1:2 for 1.5 hrs at 37 °C and pure embedding medium for 3 hrs at 37 °C. The samples were embedded in a mould and polymerized at 37 °C for 24 hrs, at 45 °C for 24 hrs, and at 60 °C for 24 hrs. Ultrathin sections of 70 nm were made on an ultramicrotome (Reichert-Jung Ultracut E) and placed on a 200 mesh grid. The ultrathin sections were stained with saturated uranyl acetate and lead citrate for 15 min each. The stained sections were examined under a transmission electron microscope (JEM1200-EX, JEOL).

RNA extraction, cDNA synthesis and qPCR

Mouse cornea tissue was collected, and the endothelial layer was carefully dissected. Total RNAs were extracted with TransZol Up Plus RNA Kit (TransGen Biotech, Beijing, China) by following the manufacturer’s instructions. RNA concentrations and purities were measured using NanoDrop One (Thermo Fisher Scientific). Complementary DNA (cDNA) was synthesized using the PrimeScript cDNA Synthesis Kit (Takara Bio, Dalian, China) following the manufacturer’s instructions and stored at -20 º C. Premix Taq (TaKaRa, Beijing, China) was used for PCR reaction. Chst5 (NM_019950) qPCR primers used were: 5’-GCCAGCTCTTCAGCCAACAC-3’, 5’-ATACGTCCATGTCGCATAGGAA-3’.

Exome sequencing for CHST6 mutation detection

2 mL peripheral blood was collected from the MCD pedigree with consent forms received. Genomic DNA was extracted from the peripheral blood (QIAamp, Qiagen) and exome sequencing was conducted on Illumina HiSeq X Ten System. Sequencing and data analysis were conducted by MyGenostics Inc. (Beijing, China).