Exploring Ocular Fibulin-3 (EFEMP1): Anatomical, Age-Related, and Species Perspectives

ABSTRACT

Fibulin-3 (FBLN3, aka EFEMP1) is a secreted extracellular matrix (ECM) glycoprotein implicated in ocular diseases including glaucoma and age-related macular degeneration. Yet surprisingly, little is known about its native biology, expression patterns, and localization in the eye. To overcome these shortcomings, we conducted gene expression and immunohistochemistry for FBLN3 in ocular tissue from mice, pigs, non-human primates, and humans. Moreover, we evaluated age-related changes in FBLN3 and FBLN3-related ECM remodeling enzymes/inhibitors in aging mice. We found that FBLN3 displayed distinct staining patterns consistent across the mouse retina, particularly in the ganglion cell layer and inner nuclear layer (INL). In contrast, human retinas exhibited a unique staining pattern, with enrichment of FBLN3 in the retinal pigment epithelium (RPE), INL, and outer nuclear layer in the peripheral retina. This staining transitioned to the outer plexiform layer in the central retina/macula, and was accompanied by reduced RPE immunoreactivity approaching the fovea. Surprisingly, we found significant age-related increases in FBLN3 expression and protein abundance in the mouse retina which was paralleled by reduced transcript levels of FBLN3-degrading enzymes (i.e., Mmp2 and Htra1). Our findings highlight important species-dependent, retinal region-specific, and age-related expression and localization patterns of FBLN3 which favor its accumulation during aging. These findings contribute to a better understanding of FBLN3’s role in ocular pathology and provide valuable insights for future FBLN3 research.

Graphical Abstract

INTRODUCTION

Fibulin-3 (FBLN3), or epidermal growth factor (EGF)-containing fibulin-like extracellular matrix protein 1 (EFEMP1), is a secreted extracellular matrix (ECM) glycoprotein, belonging to the fibulin family of proteins¹. FBLN3, like many of the other members of its protein family, is widely expressed throughout the human body and plays an important role in regulating the structural and physiological integrity of basement membranes and the ECM^2–4. Accordingly, loss of FBLN3 expression (or loss of FBLN3 function) is associated with reduced pelvic organ support, inguinal hernias, advanced bone age, ECM disruptions, and development of Marfan-like characteristics^5–10.

Conversely, increases in FBLN3 copy number/expression have been associated with a higher chance of development of ocular diseases such as age-related macular degeneration (AMD)^{11, 12}, the leading cause of irreversible blindness in the elderly in industrialized nations¹³. Moreover, select autosomal dominant mutations in FBLN3 appear to cause a range of diverse and devastating vision disorders including juvenile or primary open angle glaucoma^14–16, generalized retinal degeneration¹⁷, and the rare AMD-like disease, Doyne honeycomb retinal dystrophy/Malattia Leventinese (DHRD/ML)^18–20.

Given the wide range of human diseases in which FBLN3 is involved, it is important for researchers to utilize appropriate model systems that accurately recapitulate FBLN3-related human physiology and pathophysiology. For example, from a systemic disease perspective, FBLN3 knockout mice mimic many of the same phenotypes⁶ observed in humans who lack FBLN3 expression or function^{7, 8}. Moreover, mice expressing the Arg345Trp (R345W) mutation that causes DHRD/ML primarily form age-dependent sub-retinal pigmented epithelium (RPE) basal laminar deposits (BLamDs)^19–21 that are reminiscent of early AMD disruptions^{22, 23}. Yet, these deposits do not resemble the extensive drusen typically observed in DHRD/ML patients²⁴. The reasons for this discrepancy are not clear – whether such differences are due to inherent anatomical dissimilarities, such as a thinner Bruch’s membrane and elastin layer in mice (likely facilitating better transport and diffusion of macromolecules between the RPE and choroid vasculature)²⁵, or whether lower levels of FBLN3 expression in the mouse RPE²⁶ could be a contributing factor. Surprisingly, there are virtually no studies that compare and contrast FBLN3-related ocular expression and protein localization information amongst mice, the most commonly used laboratory model to study visual disorders, and other mammals such as non-human primates (NHPs), or even humans.

To bridge this gap in knowledge, in this study, we have thoroughly and rigorously evaluated the similarities and differences in FBLN3 transcription and localization in ocular tissues implicated in FBLN3-related diseases isolated from mice, pigs, NHPs, and humans. In complementary studies, we paralleled these findings with FBLN3 immunofluorescence localization using a knockout-validated FBLN3 antibody, and determined how FBLN3 production/accumulation changes with advanced age and in diseases such as AMD. Ultimately, our results demonstrate the surprising and dynamic nature of FBLN3 expression/localization which is dependent on the retinal area visualized, species used, and age. Integration of these factors, along with potentially subject sex²⁷, is necessary when interpreting FBLN3-related observations in the eye, especially those related to pathophysiology of FBLN3-associated ocular diseases, such as retinal degeneration and possibly glaucoma.

MATERIALS AND METHODS

Mice:

C57BL/6J female mice were obtained from Jackson Laboratory (Bar Harbor, ME) at 2, 6, and 18 months of age. Additional male and female C57BL/6 wildtype (2, 6, and 17 months) as well as FBLN3 wildtype and knockout mice²⁶ (15 months, males and females) from separate breeding protocols were also obtained. To ensure consistency, mice were analyzed within each of their respective groups. All mice were maintained in 12 h/12 h light/dark cycle and supplied with standard rodent chow (2016 Teklad Global 16% Protein Rodent Diet) and water ad libitum. All experiments were conducted in accordance with the Association for Research in Vision and Ophthalmology (ARVO) Statement of the Use of Animals in Ophthalmic and Vision Research and the UT Southwestern Medical Center (UTSW) or University of Minnesota (UMN) Institutional Animal Care and Use Committee (IACUC) guidelines. Approved protocol numbers: 2016–101740 (UTSW), 2308–41350A (UMN).

Non-human primate (NHP) eyes (Olive and Olive/Yellow baboon):

Normal NHP eyes (male and females between 1 and 12 years of age) (Sup. Table 1) were obtained from the Southwest National Primate Research Center (San Antonio, TX). Maximum death to preservation time was 4 h. One eye from each pair was preserved in RNALater solution (Invitrogen #AM7020; Carlsbad, CA, USA) at 4°C and used for quantitative PCR (qPCR), while the other was fixed in 10% formalin and used for histology.

Human donor eyes:

Normal adult donor eyes (between 61 and 85 years of age) (Sup. Table 2) with no known history of retinal diseases or corneal transplants were obtained from the Lion’s Eye Institute (Tampa Bay, FL). Maximum death to preservation time was 6 h. One eye from each pair was preserved in RNALater solution and used for qPCR, while the other was fixed in Davidson’s solution (2-parts formalin, 3-parts ethanol, 1-part glacial acetic acid, and 3-parts water) and used for immunofluorescence. Three OCT frozen slides each from two AMD donors were obtained from the Curcio Lab (University of Alabama at Birmingham, Sup. Table 3).

Porcine eyes (Hampshire, Duroc, and Red Wattle hogs):

Adult porcine eyes ages (males and females between 6 months to 1 year of age) were obtained from Huse’s Country Meats Inc. (Malone, TX). Maximum death to preservation time was 4h. One eye from each pair was preserved in RNALater solution at 4°C and used for qPCR, while the other was fixed in 10% formalin and used for histology.

qPCR analysis:

Enucleated eyes were dissected to remove the anterior segment and the posterior eye cup was further dissected to obtain neural retina and RPE/choroid separated from the sclera (except in mice where sclera was included). These tissues were then preserved in RNALater solution. RNA isolation for all tissues were carried out using the Aurum Total RNA isolation kit (BioRad #732–6820; Hercules, CA, USA). Samples were reverse-transcribed to cDNA using qScript cDNA Supermix (Quantabio #101414–106; Beverly, MA, USA). Available TaqMan probes (Applied Biosystems; Waltham, MA, USA) were used for qPCR reactions as follows: Mice: fibulin-1 (Fbln1, Mm00515700_m1), fibulin-2 (Fbln2, Mm00484266_m1), fibulin-3 (Efemp1, Mm00524588_m1), fibulin-4 (Efemp2, Mm00445429_m1), fibulin-5 (Fbln5, Mm00488601_m1), Mmp2 (Mm00439498_m1), Timp3 (Mm00441826_m1), Htra1 (Mm00479887_m1). Humans and NHPs: fibulin-1 (FBLN1, Hs00972609_m1), fibulin-2 (FBLN2, Hs00157482_m1), fibulin-3 (EFEMP1, Hs00244575_m1), fibulin-5 (FBLN5, Hs01056636_m1). Porcine: fibulin-3 (EFEMP1, Ss04326723_m1), fibulin-4 (EFEMP2, Ss06901209_m1), fibulin-5 (FBLN5, Ss04805024_m11) and samples were run on a QuantStudio 6 Real-Time PCR system (Applied Biosystems #4485694) in technical and biological replicates (n=3–5), where n’s indicate separate mice, pigs, NHPs, or human donors. Relative abundance was calculated by comparing expression to β-actin [mice (Bact, Mm02619580_g1), humans and NHPs (BACT, Hs01060665_g1), and pigs (BACT, Ss03376563_uH)] in each respective tissue.

H&E histology:

All eyes were processed into paraffin-embedded sections (Histo Pathology Core, UT Southwestern). H&E staining was performed, and images of the sections were taken using 10x and 20x objectives on a Leica DMI3000 B manual inverted fluorescence microscope (Leica Microsystems, Buffalo Grove, IL).

Immunohistochemistry:

All fixed eyes were processed into paraffin-embedded sections (serial sections were obtained for each sample), de-paraffinized, and subjected to antigen retrieval (Histo Pathology Core, UT Southwestern). Slides were washed (2 × 5 min) in 1x Tris-buffered saline (TBS) followed by pretreatment in 0.025% Triton X-100/1xTBS with gentle agitation (2 × 5 min). Sections were blocked in blocking solution consisting of 10% normal goat serum (New Zealand origin, #16210064, Gibco, Waltham, MA) with 1% bovine serum albumin (Fisher Bioreagents, #BP1600–100, Waltham, MA) in TBS for 2 hours at room temperature (RT) in a humidified chamber. Slides were drained and incubated in primary antibodies diluted in blocking solution overnight at 4°C. The antibodies used were as follows: anti-fibulin-3 antibody (1:200, Santa Cruz, #sc-365224, Dallas, TX), anti-fibulin-3 antibody (1:100, Chemicon Millipore, #MAB1763, St. Louis, MO), anti-fibulin-3 antibody (1:100, GeneTex, #GTX111657, Irvine, CA), anti-fibulin-3 antibody (rabbit IgG, 1:100, ProSci, #5213, Poway, CA,), anti-rhodopsin antibody (rabbit IgG, 1:500, Cell Signaling, #14825, Danvers, MA), fibulin-3 blocking peptide (5:1 to anti-fibulin-3 antibody, Prosci # 5213P). The next day, slides were rinsed with 0.025% Triton X-100/1xTBS (2 × 5 min) and incubated in goat anti-rabbit AlexaFluor488 secondary antibody (1:1000, Life Technologies, #A27034, Carlsbad, CA) at RT for at least 2 h followed by 4′,6-diamidino-2-phenylindol (DAPI) staining at RT for 20 min. Sections were then rinsed with 1xTBS (3 × 5 min) and mounted with Vectashield (Vector Laboratories, Newark, CA). Images were taken using a 25x or 63x objective on a Leica TCS SP8 confocal microscope.

In situ Hybridization (ISH):

Eyes were harvested, fixed, paraffin-embedded, and sectioned. RNAscope^™ ISH was performed (In-Situ Hybridization Core, Gautron Lab, UT Southwestern) using the 2.5 HD Detection Reagent – RED (Advanced Cell Diagnostics, Hayward, #322360, CA USA) and probes to detect the EFEMP1 expression in mouse (Mm-Efemp1 #429501) and human (Hs-EFEMP1 #433161) ocular cross sections, while the expression of PPIB was used as control (Positive Control Probe- Hs-PPIB # 313901, Positive Control Probe- Mm-Ppib # 313911). Images were taken using a 25x objective on a Leica TCS SP8 confocal microscope.

RESULTS

Mice have patterns of fibulin gene expression in the neural retina and RPE which are distinct from other mammals.

In previous studies, we were surprised to find that FBLN3 expression levels in the neural retina (NR) of C57BL/6 mice were consistently higher than those found in the RPE/choroid²⁶, the primary site of FBLN3-driven retinal pathology^18–20. We reconfirmed these important qPCR data (Fig. 1A) and expanded the same analyses to pigs, NHPs, and humans (Fig. 1B-D), utilizing all the commercially-available TaqMan fibulin probes for each species. Amongst the species analyzed, mice were the only animal where FBLN3 expression was significantly lower (an average of 2-fold, Fig. 1A, *** p < 0.001) in the RPE/choroid compared to the NR, consistent with our past observations. Fold increases in RPE/choroid FBLN3 expression vs. NR expression ranged from 3-fold in pigs, to 2-fold in NHPs, to 10-fold in humans (Fig. 1B-D), all of which were significant (* = p < 0.05, ** = p < 0.01). Interestingly, in tissues with high levels of FBLN3 expression (e.g., RPE/choroid from pigs, NHPs, and humans), little expression of other fibulin family members was observed, indicating that FBLN3 is likely the primary fibulin component in these tissues (Fig. 1B-D). In contrast, in tissues with low FBLN3 expression (e.g., RPE/choroid from mice), high levels of other fibulins (fibulin-1, fibulin-4, and fibulin-5) were found (Fig. 1A), potentially indicating a compensatory mechanism. Upon ratio metric analysis of FBLN3 expression of RPE/choroid over neural retina, we observe a trend of higher RPE/NR in humans, pigs, and NHPs but not in mice (Fig. 1E). Therefore, these FBLN3 expression differences already prognosticate likely FBLN3 protein localization variation amongst the tested species.

Figure 1. Gene expression profile of fibulins across species.

(A) Expression levels of fibulin 1–5 in the NR and RPE/choroid of mice (9 mo) relative to β-actin (n=3) (FBLN3, p≤0.001). (B) Expression levels of fibulin 3–5 in the NR and RPE of pigs relative to β-actin (n=4) (FBLN3, p≤0.01). (C) Expression levels of fibulin 1–3 and 5 in the NR and RPE/choroid of NHPs relative to β-actin (n=4) (FBLN3, p≤0.05). (D) Expression levels of fibulin 1–3 and 5 in the NR and RPE/choroid of humans relative to β-actin (n=5) (FBLN3, p≤0.01). Unpaired t test. Mean ± SD. (E) Graph representing FBLN3 expression in terms of ratio of RPE/choroid to that of the NR for humans, NHPs, pigs, and humans relative to β-actin.

Retinal distribution and localization of FBLN3 (protein and mRNA) varies among species.

During our efforts to better define FBLN3 protein localization in the eye, we found that many available FBLN3 antibodies produced a significant amount of background staining, even in FBLN3 knockout (KO) mice (Fig. S1F-H). Moreover, in instances where we did not have a FBLN3 KO control (i.e., in pigs, NHPs, and humans), it was difficult to gauge the specificity of staining across species, without the use of a blocking peptide. Thus, throughout our studies, we used an antibody from ProSci (cat# 5213) that showed no staining in FBLN3 KO mice (Fig. S1E) and also was available with a corresponding blocking peptide (cat# 5113P). Moreover, the epitope recognized by this antibody is highly conserved across species (Fig. S1I). Consistent with our observations at the transcriptional level, in mice, FBLN3 is found primarily in the ganglion cells layer (GCL) and inner nuclear layer (INL) with no clear signal originating from the RPE layer (Fig. 2A, Fig. S2A, B), consistent with previous difficulties detecting wild-type FBLN3 in the RPE, even in aged mice²⁸. Porcine ocular cross-sections revealed a unique, retina-wide, almost equal distribution of FBLN3, including in the GCL, inner plexiform layer (IPL), INL, outer plexiform layer (OPL), outer nuclear layer (ONL), photoreceptors (PR), as well as RPE (Fig. 2B, Fig. S2C, D). In contrast, distribution of FBLN3 in the NHP and human ocular cross-sections exhibited patterns of localization particularly focused in the OPL and RPE, although higher levels of expression in the GCL is observed in NHP compared to humans (Fig. 2C, D, Fig. S2E-H). To test whether the type of fixative used affected the specificity of this antibody, we used four different fixation techniques (freeze substitution, 10% normal buffered formalin [NBF], 4% paraformaldehyde [PFA], and Davidson’s solution) to process mouse retina and immunostained them for FBLN3. The prominent observation of FBLN3 staining in the GCL and INL, as depicted in Figure 2A, remained consistent regardless of the fixative used (Fig. S3A-D), while no still staining was detected in the RPE, however we observe positive FBLN3 staining in the posterior eye cup connective tissue of the mouse ocular sections (Fig. S9A). Moreover, FBLN3 staining in porcine retina does not appear to be due to ‘stickiness’ of the tissue or section since other antibodies of the same isotype (rabbit IgG) did not show the same broad staining pattern (Fig. S9B). These findings demonstrate the antibody’s ability to detect FBLN3 across various processing conditions.

Figure 2. FBLN3 distribution and localization in the retina across species.

Representative images (n=3–5) of cross-sections showing differential interface contrast (DIC), FBLN3 expression (green), and nuclear stain (DAPI, blue) for (A) mus, (B) pig, (C) NHP, and healthy (D) human eyes. (Scale bar = 100 μm). GCL = ganglion cell layer, INL = inner nuclear layer, ONL = outer nuclear layer, RPE = retinal pigmented epithelium.

To further validate the localization of FBLN3 expression within the retina, we conducted in situ hybridization on mouse and human retinal sections using RNAscope¹⁵. In both mice and humans, the positive control PPIB expression was observed throughout the retina, whereas the localization of FBLN3 mRNA expression was distinct in both species, thereby supporting the findings of protein localization by immunohistochemistry (Fig. S4).

Since FBLN3 is also expressed more broadly throughout the eye than just the retina and RPE²⁶, we also tested FBLN3 staining in the cornea (Fig. S5A-D) and ciliary body/iris (Fig. S6A-D) in mice, pigs, NPHs and humans. Strong FBLN3 staining was observed in the corneal epithelium of mice along with relatively weaker staining of the corneal endothelium (Fig. S5A). Similar degrees of epithelium staining were detected in human corneal epithelium (Fig. S5D). Pig and NHP demonstrated negligible FBLN3 staining in the epithelium and stroma (Fig. S5B, C), however, it is important to emphasize here that these samples were more challenging to preserve during the fixation process. Very strong FBLN3 expression was evident in the ciliary body and iris of mice (Fig. S6A), consistent with previous RNAScope analyses¹⁵. Similar, but weaker staining was observed in human ciliary body/iris (Fig. S6D), and very low, if any FBLN3 protein in pig and NHP ciliary body/iris tissue (Fig. S6B, C). Importantly, blocking peptide eliminated FBLN3 staining in all ocular tissues tested (Fig. S7A-D), which, when combined with validation of the antibody in FBLN3 KO mice (Fig. S1A-H), and our parallel mRNA analysis (Fig1A-D), support the specificity of our staining for FBLN3 across species.

FBLN3 localization varies within the human retina and is associated with RPE deposits in AMD tissue.

Until this point, we have used bulk RNA from specific ocular tissue to evaluate FBLN3 expression, or relied on retinal FBLN3 staining patterns within a single location, typically in the central retina adjacent to the optic nerve. However, retinal anatomy and physiology varies greatly depending on location. Thus, we decided to evaluate FBLN3 staining throughout the human retina, ora serrata to ora serrata (Fig. 3A). In doing so, we detected surprising differences in FBLN3 localization in human tissue. Peripheral retinal FBLN3 staining (close to the ora serrata) was most intense in the RPE, followed by lower degrees of staining in the INL and ONL (Fig. 3B). As we moved inward towards the central retina (~3 mm from the optic nerve head), FBLN3 staining began to shift from primarily RPE staining to predominantly OPL staining, with faint RPE and GCL/nerve fiber layer staining (Fig. 3C). Finally, within the foveal pit of the macula, FBLN3 staining was predominantly in the OPL with faint staining in the INL and surprisingly no observable FBLN3 staining in the RPE (Fig. 3D). No such differences in FBLN3 localization in the far-periphery, mid-periphery, or central retina were observed in mice or pigs (Figs. S8). We speculate that these differences might be present in the NHP retina, however, due to limited availability of ideal sections (entire retina with macula) we were unable to demonstrate this. Nevertheless, our results delineate another key difference observed in expression profile of FBLN3 between mice and humans which is important for understanding basic FBLN3 biology and eventually pathophysiology.

Figure 3. FBLN3 localization varies within a human retina.

(A) Schematic of the region of interest imaged in the human retina. Illustration from BioRender. Representative images (n=3) of cross-sections showing DIC, FBLN3 expression (green), nuclear stain (DAPI, blue) for (B) peripheral (at ora serrata), (C) central (~3 mm from optic nerve head), and (D) fovea/macula region of a human retina. (Scale bar = 100 μm). GCL = ganglion cell layer, INL = inner nuclear layer, ONL = outer nuclear layer, RPE = retinal pigmented epithelium.

Previous reports have suggested that FBLN3 surrounds drusen in AMD patients^{28, 29}. To confirm these results with a FBLN3 KO-validated antibody, we stained ocular sections from two AMD patients (Table S3, Fig. 4A, B). In fields of view that contained small drusen surrounded by sloughed off RPE cells (Fig. 4A, B), we confirmed positive FBLN3 staining surrounding some, but not all small druse, consistent with previous reports^{28, 29}. It is difficult to conclude whether the observed FBLN3 staining is due to FBLN3 expression inside the RPE, or if it is deposited, i.e., extracellular FBLN3. Staining of FBLN3 in the OPL within these sections suggests that that this field of view originates from the central retina.

Figure 4. FBLN3 accumulates around drusen in human eyes with AMD.

(A-B) Representative images of cross-sections showing DIC, FBLN3 expression (green), nuclear stain (DAPI, blue) for retinas of 2 AMD patients focused on RPE pathology. Asterix (*) denotes FBLN3 staining in small druse and surrounding FBLN3 aggregation. (C) Zoomed-in image of druse and FBLN3 accumulation from inset in (B) (Scale bar = 100 μm). GCL = ganglion cell layer, INL = inner nuclear layer, ONL = outer nuclear layer, RPE = retinal pigmented epithelium.

Aging alters the expression of FBLN3 and ECM proteases involved in FBLN3 turnover.

The amount of FBLN3 protein that we observe in any given tissue sample is a result of two primary factors, i) extent of FBLN3 expression, and ii) FBLN3 protein turnover. Recent studies have suggested that FBLN3 expression increases with age³⁰ and that FBLN3 protein stability (in the form of amyloid) may also be increased with age³¹. In fact, FBLN3 was identified as one of the top 5 genes overexpressed during aging across all analyzed tissues³⁰. To determine whether FBLN3 expression and protein levels also increases with age in the retina and RPE, we performed gene expression analysis in mice at 2–18 months of age (Fig. 5A) along with corresponding FBLN3 immunostaining at similar ages of 4–19 months (Fig. 5B-D). As a control tissue for the expression experiments, we included RNA isolated from mouse liver. Surprisingly, we found that over the course of advance aging, FBLN3 expression did not change in any tissue between 2 mo (young) and 6 mo (adult) of age in mice (Fig. 5A), but significantly increased between 6 mo and 18 mo (geriatric) mice in the NR (average of 10-fold) and liver (average of 25-fold) (p < 0.0001, Fig. 5A). FBLN3 also increased in the RPE, but not significantly (Fig. 5A). Immunostaining for FBLN3 confirmed the NR observations of increased protein levels in geriatric mice compared to young or adult mice (Fig. 5B-D).

Figure 5. Age-related changes in FBLN3 expression in mice.

(A) Graph representing expression levels of FBLN3 in the liver, NR and RPE/choroid of mice at 2 mo, 6 mo, and 18 mo relative to β-actin (n=4, ****p<0.0001). Analyses performed by 2-way ANOVA and represented as mean ± SD. (B) Graph interpreting graph A in terms of ratio of RPE/choroid to that of the NR at 2 mo, 6 mo, and 18 mo relative to β-actin. (C-E) Representative images (n=4) of ocular cross-sections showing DIC, FBLN3 expression (green), nuclear stain (DAPI, blue) of mice at 4 mo, 8 mo, and 19 mo. (Scale bar = 100 μm). GCL = ganglion cell layer, INL = inner nuclear layer, ONL = outer nuclear layer, RPE = retinal pigmented epithelium.

While little is truly known about the totality of factors that regulate FBLN3 protein turnover, two enzymes present in the eye that have been demonstrated to promote the degradation of FBLN3 are matrix metalloproteinase 2 (MMP2)³² and high-temperature requirement A serine peptidase 1 (HTRA1)³³. We next asked if the buildup of FBLN3 that we observe with advanced age may be due in part to declining levels of these ECM homeostasis enzymes. qPCR was performed on two major ocular MMPs, Mmp2³⁴ and Mmp9, and their cognate inhibitor, tissue inhibitor of matrix metalloproteinase 3 (Timp3)^35–37, which has also been identified as a FBLN3 binding partner³⁷. With increasing age, we found a corresponding significant increase in Timp3 expression (p <0.0001, Fig. 6A) which was paralleled by a decrease in Mmp2 and Mmp9 expression (p < 0.001, p < 0.0001, Fig. 6B, C). Additionally, Htra1, a member of the secreted serine protease family linked which is linked to FBLN3 turnover³³ and AMD pathology³⁸, decreases significantly in the NR as the mice age (p < 0.0001, Fig. 6D) (note: due to difficulties in Htra1 amplification in liver, this tissue was excluded from this analysis). Our data strongly suggest that age causes an increase in FBLN3 expression while also downregulating enzymes responsible for FBLN3 extracellular degradation, the culmination of which promotes increased FBLN3 production and deposition in the retina with advanced age, possibly promoting retinal dysfunction and age-related pathologies.

Figure 6. Age-related changes in the expression of ECM remodeling enzymes in mice.

(A) Graph representing expression levels of Timp3 in the liver, NR and RPE/choroid of mice at 2 mo, 6 mo, and 18 mo relative to β-actin (n=4, ****p<0.0001). (B) Graph representing expression levels of Mmmp9 in the liver, NR and RPE of mice at 2 mo, 6 mo, and 18 mo relative to β-actin (n=4, ****p<0.0001). (C) Graph representing expression levels of Mmp2 in the liver, NR and RPE of mice at 2 mo, 6 mo, and 18 mo relative to β-actin (n=4, ****p<0.0001, ***p=0.0003). (D) Graph representing expression levels of Htra1 in the liver, NR and RPE of mice at 2 mo, 6 mo, and 18 mo relative to β-actin (n=4, ****p<0.0001, **p=0.003). All analyses performed by 2-way ANOVA and data are presented as mean ± SD.

DISCUSSION

FBLN3 is emerging as a unique and physiologically important member of the fibulin family involved in several ocular pathologies in both the anterior and posterior chamber. Our results demonstrate the importance of incorporating anatomical, age, and species related considerations when evaluating FBLN3 expression and protein levels/localization in eye tissue. In addition to these factors, other studies have indicated that subject sex or sex hormones likely influence FBLN3 expression^{39, 40} or BLamD formation²⁷, although our studies were unable to confirm or refute those observations. The culmination of these data suggest that FBLN3-based experiments must be appropriately thoroughly controlled and that observations made in mice should be validated in separate, complementary model systems to enhance experimental rigor.

Our study has unveiled several intriguing insights into the behavior of FBLN3 across different species, shedding light on aspects that have often been overlooked. Notably, the expression patterns of FBLN3 in mice, pigs, NHPs, and humans display interesting variations, which hold potential implications for our understanding of FBLN3-related ocular diseases. The observed lower level of RPE expression in mice compared to other species emphasizes the importance of thoughtfully selecting model systems. In particular when investigating FBLN3-related ocular diseases given the prevalent use of mice as the primary laboratory model for retinal diseases like AMD. It’s noteworthy that R345W FBLN3 knock-in mice do exhibit extensive and continuous BLamDs; however, the development of this phenotype spans 12–18 months and is typically concentrated in specific quadrants of the eye²⁷. These findings underscore the need for nuanced consideration in model selection for a comprehensive understanding of FBLN3-related ocular conditions.

It is interesting to speculate that identifying transgenic approaches aimed at elevating FBLN3 RPE expression to levels observed in humans could be an approach to generate a more robust (or earlier onset) AMD-like mouse model. For example, using RPE-specific promoters (BEST1⁴¹ or RPE65⁴²) could drive higher FBLN3 levels in the RPE which may more closely match humanized FBLN3 expression patterns. In addition, generating knock-in mice harboring glaucoma-associated mutations in FBLN3¹⁴⁻¹⁶ might provide an excellent glaucoma model, especially due to the high levels of FBLN3 expression in the trabecular meshwork ring and ciliary body/iris²⁶.

Secondly, in the healthy human eye, we observed drastic changes in FBLN3 protein localization in the retina/RPE depending on anatomical region – essentially favoring RPE accumulation in the peripheral retina, and OPL accumulation as one approaches the central retina and macula/fovea. One aspect that is surprising about these findings is the lack of apparent FBLN3 RPE localization in areas that are susceptible to drusen formation (like the macula)⁴³, especially in the case of DHRD/ML¹⁸. While the reasons for these distinct FBLN3 localization changes is not clear, it is possible that there are FBLN3 expression differences in retinal regions, or alterations in proteins that turnover FBLN3 (e.g., MMP2 and/or HTRA1) in these particular regions that change with age and/or disease. Indeed, even in ocular cells that were once thought to be rather uniform, such as the RPE, recent evidence suggests that there are physical and genetic differences that distinguish cell populations in the periphery vs. central retina vs. the macula⁴⁴, and these results likely extend to additional retinal cell types as well.

The primary staining of FBLN3 within the fovea was restricted to the OPL, a retinal cell layer typically packed with neuronal synapses originating from horizonal cells, bipolar cells, and photoreceptors. According to single cell RNAseq (scRNAseq) data, human FBLN3 is primarily produced in fibroblasts > RPE > Müller glia⁴⁵. Thus, it is likely that the FBLN3 we observe in the OPL originated from Müller glia, which span the height of the retina from the inner limiting membrane to the external limiting membrane. Due to their expansive nature, alterations in FBLN3 protein patterns within the retina could reflect Müller glia FBLN3 expression, but differential deposition due to altered FBLN3 secretion or turnover patterns. Indeed, it is quite possible that the strong FBLN3 staining we observe in the OPL may correlate to Müller glia endfeet⁴⁶, or cone pedicles/spherules. Nonetheless, while we believe that FBLN3 likely acts proximal to the cells that it is produced in, since it is a secreted protein, we cannot rule out that it may be produced in one location, but accumulates in locales distant from its source.

Third, we demonstrate that FBLN3 expression and protein levels in the mouse retina significantly increase with advanced age. The mechanisms that regulate FBLN3 expression and turnover within the eye are virtually unstudied. Thus, it is difficult to even speculate what transcriptional regulatory pathways are involved in this age-dependent observation. Regardless, concomitant with the increase in FBLN3 expression levels, we also noted significant reductions in the expression of proteins that have been demonstrated to proteolyze FBLN3 in biochemical assays (Mmp2 and Htra1). These two findings highlight the likelihood of increased FBLN3 deposition and dysfunction with age. Moreover, when these phenomena occur in the presence of an FBLN3 mutation that causes misfolding^{17, 47–49}, retinal disease could be further accelerated as an individual ages. However, given the vast differences we have observed in FBLN3-based behavior between species, it is important that these results be reproduced in another species, ideally from human donor tissue spanning a wide range of ages.

The link between FBLN3 and ocular diseases is intricate. While decreased FBLN3 expression or function is associated with conditions like inguinal hernias, advanced bone age, and Marfan-like syndromes, increased FBLN3 expression, or copy number has been linked to age-related macular degeneration (AMD) and other vision disorders. The paradoxical role of FBLN3 in ocular pathologies underscores the intricate nature of ECM regulation in the eye. These findings prompt us to consider that the development of ocular pathologies related to FBLN3 might not be solely due to its expression level but rather its protein fold, localization, interaction with other ECM components, or perhaps post-translational modifications.

Cross-species comparisons of protein localization patterns can be a challenging endeavor that could be influenced by species-specific sources of cross-reactivity (creating a false signal) and differences in antibody affinity in different species (affecting how well an antibody could bind to proteins from different species). While our FBLN3 ocular localization studies are corroborated by the use of knockout mice, utilization of an antibody blocking peptide, mRNA expression/in situ hybridization correlations, and ≥98% shared homology of the antibody epitope across species, we must acknowledge that more studies should be performed that are specifically directed at differential cross-species immunoreactivity and antibody affinity for FBLN3 originating from different species (by surface plasmon resonance [SPR], for example). Inclusion of these additional studies will be important for advancing our existing observations of FBLN3 ocular dynamics in a more definitive manner.

Nevertheless, our findings offer valuable information by revealing important differences in FBLN3 expression and localization across a range of relevant species. These observations provide a critical baseline for comparing and contextualizing future FBLN3-related studies. Moreover, our research lays the groundwork for distinguishing “normal” versus “diseased” characteristics of FBLN3, shedding light on the multifaceted role of FBLN3 in ocular biology and pathology, which is pivotal in developing interventions aimed at modulating the function of this intriguing protein. Ultimately, this research will facilitate the development of novel treatment approaches targeted at modulating FBLN3 function, preventing FBLN3 accumulation, and ameliorating vision disorders associated with it.

Supplementary Material

1

Sup. Figure 1. Comparison of commercially available FBLN3 antibodies for specificity. Representative images of cross-sections showing merged (DIC/FBLN3(green)/DAPI (blue) images for FBLN3 wild-type (WT) and knockout (KO) mus (15 mo) using antibodies against FBLN3 from ProSci (A&E), GeneTex (B&F), Chemicon (C&G), and Santa Cruz (D&H). Arrows indicate non-specific staining. (Scale bar = 100 μm). (I) Clustal alignment of FBLN3 c-terminal amino acid sequence recognized by the ProSci antibody in humans, pigs, NHPs, and mice.

2

Sup. Figure 2. Side-by-side comparison of FBLN3 staining with corresponding H&E histology. (A, C, E, G) Representative images (n=3–5) of cross-sections showing DIC, FBLN3 expression (green), nuclear stain (DAPI, blue) for mus, pig, NHP, and healthy human eyes with their corresponding H&E (B, D, F, H) histology (GCL-ganglion cell layer, IPL-inner plexiform layer, INL-inner nuclear layer, OPL-outer plexiform layer, PR-photoreceptor, RPE-retinal pigment epithelium). (Scale bar = 100 μm).

3

Sup. Figure 3. FBLN3 staining in mouse retina under different fixation parameter. Representative images of cross-sections showing DIC, FBLN3 expression (green), nuclear stain (DAPI, blue) for mouse eyes fixed using Freeze substitution, 10% Normal buffered formalin [NBF], 4% paraformaldehyde [PFA], and Davidson’s solution. (Scale bar = 100 μm).

4

Sup. Figure 5. FBLN3 distribution and localization in the cornea across species. Representative images (n=3–5) of cross-sections showing DIC, FBLN3 expression (green), nuclear stain (DAPI, blue) for (A) mus, (B) pig, (C) NHP, and (D) human corneas. EP-epithelium, EN-endothelium. (Scale bar = 100 μm [mus], 500 μm [pig, NHP, human]).

5

Sup. Figure 6. FBLN3 distribution and localization in the ciliary body and iris across species. Representative images (n=3–5) of cross-sections showing DIC, FBLN3 expression (green), nuclear stain (DAPI, blue) for (A) mus, (B) pig, (C) NHP, and (D) human ciliary body (CB) and iris (IR). (Scale bar = 100 μm [mus] 500 μm [pig, NHP, human]).

6

Sup. Figure 7. Absence of positive FBLN3 staining in the presence of FBLN3 blocking peptide demonstrates specificity of FBLN3 antibody across tissues and species. Representative images (n=3-5) of cross-sections showing DIC, FBLN3 expression (green), nuclear stain (DAPI, blue) for (A) mus, (B) pig, (C) NHP, and (D) human- cornea, ciliary body, iris, and retina. Scale bar [cornea, ciliary body, and iris] = 100 μm, [mus] 500 μm, [pig, NHP, human]. (Scale bar [retina]= 100 μm).

7

Sup. Figure 8. Uniform FBLN3 staining observed in different regions of mouse and pig retina. Representative images (n=3–4) of cross-sections showing merged [DIC/FBLN3 (green)/DAPI (blue)] for mus, pig retina. (Scale bar = 100 μm).

9

Sup. Figure 4. In situ hybridization of FBLN3 transcripts in mouse and human retinas. Representative images (n=3) of cross-sections showing DIC and strongest localization of transcripts FBLN3 (red punctate staining) or PPIB (red punctate staining) in mouse and human retinas. Black arrows indicate presence of FBLN in the ganglion cell layer. White arrows indicate absence of FBLN3 transcripts in mouse RPE and yellow arrows indicate the presence of FBLN3 transcripts in human RPE. (Scale bar = 100 μm).

10

Sup. Figure 9. (A) FBLN3 staining in the posterior eye cup of mice. Representative images of cross-sections showing DIC, FBLN3(green), DAPI (blue), and merged images for FBLN3 staining in the neural retina as well as the posterior eyecup connective tissue. (Scale bar = 100 μm). (B) Uniform rhodopsin staining of the photoreceptors in porcine retina. Representative images of cross-sections showing DIC, RHO (red), DAPI (blue), and merged images for rhodopsin staining in the photoreceptors of porcine retina using a rabbit IgG anti-rhodopsin antibody (#14825 Cell Signaling). This antibody serves as an isotype control for the ProSci FBLN3 antibody that we use throughout this manuscript and demonstrates that FBLN3 staining in the pig is not necessarily due to ‘stickiness’ of the preserved section. (Scale bar = 100 μm).

Highlights.

• Contrasting FBLN3 expression in the RPE of humans/NHP vs. mice

• FBLN3 protein localization varies with retinal anatomical region in humans

• FBLN3 expression and protein levels increase with age in mice

Data availability statement:

Data supporting the findings of this study are available within the article and its supplementary materials. Upon request, the authors are willing to share additional data.