Bruch’s membrane abnormalities in PRDM5-related brittle cornea syndrome

Informed written consent was obtained and investigations conducted in accordance with the principles of the Declaration of Helsinki, with Local Ethics Committee approval (NHS Research Ethics Committee reference 06/Q1406/52). Patients with BCS P1 and P2, with PRDM5 Δ9–14; and P3, with PRDM5 p.Arg590* have been previously described [4]. Diagnosis of BCS in P4, with PRDM5 p.Glu134*, was based on clinical examination and confirmed by mutation analysis of ZNF469 and PRDM5. Detailed ophthalmic examinations included anterior segment examination by slit lamp biomicroscopy, measurement of corneal thickness using pachymetry, retinoscopy, color photography, fluorescein angiography, and optical coherence tomography (OCT). A systemic workup including full blood count, coagulation screen and renal function analyses was performed.

BCS-affected ocular tissue was obtained from the Department of Histopathology, Manchester Royal Infirmary. Human ocular tissue samples from control individuals were obtained from the Manchester Eye Bank (Table 1). Informed written consent and ethics committee approval was granted (14/NW/1495).

The open reading frames of ZNF469 and PRDM5 were sequenced as described [3, 4]. Variants identified in PRDM5 were checked against control data sets including dbSNP (Build 137) (http://​www.​ncbi.​nlm.​nih.​gov/​SNP), the 1000 Genomes Project (May 2012 release) (http://​browser.​1000genomes.​org/​index.​html), and the NHLBI Exome Sequencing Project (http://​evs.​gs.​washington.​edu/​EVS).

Fibroblast cell lysis and preparation of nuclear extracts was performed according to Schnitzler GR [7]. Total protein content was quantified using a BioRad protein quantification BCA assay (BioRad Laboratories). Skin fibroblasts nuclear extracts were subjected to standard SDS-PAGE using a custom-made antibody PRDM5 Ab2 [6, 8] at a concentration of 1 μg/ml, and GAPDH at a concentration of 2 μg/ml (Santa Cruz sc-47724) on equal amounts of nuclear fraction protein. Membranes were blocked with TBST (0.1 % Tween 20) containing 5 % non-fat dry milk, and incubated with primary antibodies overnight. Visualization was performed with an enhanced chemiluminescence western blotting kit (Cell Signalling Technologies #7003).

Histological analysis was carried out in accordance with standard diagnostic protocols. 4 μm paraffin-embedded slides were stained with hematoxylin and eosin and elastin with van Gieson. Immunohistochemistry was performed using PRDM5 Ab2 [6, 8] and mouse monoclonal antibodies against collagen I (ab90395, Abcam); collagen III (ab6310, Abcam) and collagen IV (ab6311, Abcam). PRDM5 AB2 epitope is situated within the region corresponding to N- terminal amino acids 60–142. Staining was performed on a Ventana Benchmark XT Automated Immunostaining Module (Ventana Medical Systems) together with the XT ultraView Universal Red Alkaline Phosphatase detection system for all antibodies except PRDM5, where DAB was used as the chromogen. Antigen retrieval was performed separately using heat-induced antigen retrieval for PRDM5, and no pre-treatment for collagens I, III and IV. Primary antibodies were diluted in Dako REAL™ Antibody Diluent (Dako, Agilent Technologies, UK) to the indicated optimal dilutions of 3.5 μg/ml for PRDM5; 3 μg/ml for collagen I; 10 μg/ml for collagen III; 2.5 μg/ml for collagen IV. Sections of patient eye tissue were processed in parallel with the control tissue and were collected, sorted and fixed in an identical manner. Tissue section slides were masked for origin and scored for detection of cells showing nuclear PRDM5 staining subjectively by an independent human observer using a binary scale (positive or negative). Tissues were considered positive when >20 % of the cells displayed nuclear PRDM5 staining [8].

Polyclonal rabbit anti-fibronectin (FN) antibody, mouse anti-tenascin monoclonal antibody (clone BC-24), recognizing all the tenascins (TNs), and TRITC- conjugated rabbit anti-goat antibody were from Sigma Aldrich; mouse anti-α5β1 (clone JBS5) and anti-α2β1 (clone BHA.2) integrin monoclonal antibodies; and goat anti-type I collagen, anti-type III collagen and anti-type V collagen antibodies were from Millipore Chemicon Int. Inc. (Billerica, MA). FITC- and TRITC-conjugated goat anti-rabbit and anti-mouse secondary antibodies were from Calbiochem-Novabiochem Int. (San Diego, CA, USA). Antibody dilutions were: anti-tenascin and anti-α5β1: 2 μg/ml; anti-α2β1: 4 μg/ml; anti-FN and anti-type I collagen, 10 μg/ml; anti-type III and type V collagen: 20 μg/ml. Primary dermal fibroblast cultures were established from skin biopsies by routine procedures, maintained and harvested as described [9, 10]. 1.0 × 10⁵ cells were grown for 48 h on glass coverslips, fixed in methanol and incubated with the specified antibodies as reported [9, 10]. For analysis of integrins, cells were fixed in 3 % paraformaldehyde and 60 mM sucrose, and permeabilized in 0.5 % Triton X-100. Cells were reacted for 1 h at room temperature with 1 μg/ml anti-α5β1 and anti-α2β1 integrin monoclonal antibodies. Cells were subsequently incubated with 10 μg/ml FITC- or TRITC-conjugated secondary antibodies. IF signals were acquired by a CCD black/white TV camera (SensiCam-PCO Computer Optics GmbH, Germany) mounted on a Zeiss fluorescence-Axiovert microscope, and digitalized by Image Pro Plus program (Media Cybernetics, Silver Spring, MD).

Extracted total RNA was reverse-transcribed into single-stranded cDNA using a High Capacity RNA-to-cDNA Kit (Life Technologies, Paisley, UK), according to the manufacturer’s instructions. RT-PCR and data analysis was performed as previously described [4]. The assay numbers for the mRNA endogenous control (GAPDH) and target gene were: GAPDH (Hs02758991_g1*) and ITA8 (Hs00233321_m1*) (Life Technologies). Cycles to threshold (CT) values were determined for each sample and its matched control and relative mRNA expression levels determined by the 2^−ΔΔCt method, providing the fold change value [11]. Error bars representing 95 % confidence intervals around the mean are represented for all experiments. P-values were derived using the 2-tailed T-test with significance level set at 0.01 to compare results between mutant and wild-type cells. One-way ANOVA and Dunnett’s multiple comparison posttest using mean values and standard error were also performed on fold change means in all groups assessed.

A summary of clinical samples used in this study is shown in Table 1. The mutation PRDM5 Δ9–14, carried by P1 and P2 (whose clinical details are described in Burkitt-Wright et al. [4]), is an in-frame deletion mutation that we show here results in the production of a smaller, internally deleted protein confirmed by western blot analysis on skin fibroblasts of P1 (Fig. 1). The mutation p.Arg590* (P3) (whose description is also provided in Burkitt-Wright et al. [4]), also results in a protein truncation confirmed by western blot analysis on skin fibroblasts (Fig. 1). P4 is a 26-year-old woman who presented with decreased vision in her left eye (LE) in the context of normal birth, developmental and family history. Of note she had been seen four years previously in an identical setting with similar symptoms in her right eye (RE), resulting in RE permanent visual loss. On examination best-corrected visual acuity measured 0.16 RE and 0.4 LE (Snellen decimal equivalent). Blue sclera, corneal thinning with central corneal thicknesses of 276 μm RE and 281 μm LE (Fig. 2a), and high myopia were present (spherical equivalent refraction: −12.50 RE and −13.75 LE). Axial lengths were increased in both eyes (28 mm) (IOLMaster, Carl Zeiss Meditec Inc, Jena, Germany). Dilated retinal examination demonstrated a disciform-like scar in her RE and evidence of active choroidal neovascularization (CNV) in her LE (Fig. 2b and c, arrows), leading to retinal exudation (Fig. 2d,*). Management of CNV consisted of three consecutive monthly intravitreal injections of the anti-vascular endothelial growth factor (VEGF) agent ranibizumab (Lucentis, Novartis, Basel, Switzerland), achieving a complete resolution of the exudation and significant VA improvement to 0.70. In a follow-up period of 5 years there has been no disease recurrence. Systemic examination showed marfanoid habitus, scoliosis, arachnodactyly and hyperextensible joints. Genetic analyses revealed a novel homozygous nonsense mutation in PRDM5, c.400G > T p.Glu134*, confirming a diagnosis of BCS. The mutated nucleotide is in exon 4 (Fig. 2e) and is predicted to result in nonsense-mediated decay. This variant was not present in dbSNP132, the Exome Variant Server, or the 1000 Genomes Project (all accessed January 5^th 2015). Western blot analysis on skin fibroblasts of P4 did not detect a protein product, however overlap between the location of the mutation, and location of the antibody epitope was present.

Investigation of PRDM5 expression by immunohistochemistry shows nuclear PRDM5 expression in the corneal epithelium (Fig. 3a) and extensive nuclear and cytoplasmic PRDM5 expression in the retina (Fig. 3c), with absent expression in the retinal pigment epithelium (Fig. 3e, arrow) (control post-mortem eye #2). PRDM5 staining was also not apparent in Bruch’s membrane. Expression was seen in cells of neuroectodermal origin particularly the retina, with the exception of nuclei of the smooth muscle of the ciliary body (mesodermal origin) and nuclei of the corneal epithelium (surface ectodermal origin) (Fig. 3 and Additional file 1: Table S1). Immunohistochemistry using PRDM5 Ab2 in P1 demonstrates loss of PRDM5 retinal nuclear and cytoplasmic staining (Fig. 3e).

Two patients, cousins, aged 10 (P1) and 21 (P2) with PRDM5-associated disease (previously described by Burkitt-Wright et al. [4]) were studied in order to assess whether there are any functional consequences to their retinas. As collagen expression is associated with PRDM5 levels, we studied the expression of PRDM5, collagens I, III, IV and elastin in Bruch’s membranes of eyes from these two subjects. Both patients carried a large homozygous deletion (Δ exons 9–14) in PRDM5. The axial lengths of the two eyes were 20 mm (P1) and 21.8 mm (P2), determined histologically. We found absent or decreased staining in Bruch’s membrane for collagens type I, III and IV in both BCS samples versus a control sample (Fig. 4). Van Gieson staining demonstrated normal elastin staining of Bruch’s membrane in both samples (Additional file 1: Figure S1). No histological evidence of breaks in Bruch’s membrane was evident.

Indirect immunofluorescence performed on skin fibroblasts from patient P4 showed downregulation of structural collagens I and III, fibronectin, all the tenascins (detected by a single antibody), and integrin receptors α2β1 and α5β1. The expression of type V collagen was disorganized in comparison to the control (Fig. 5).