IN SEARCH OF MOUSE MODELS FOR EXFOLIATION SYNDROME

Abstract

Purpose

Exfoliation syndrome (XFS) is a systemic connective tissue disorder with elusive pathophysiology. We hypothesize that a mouse model with elastic fiber defects caused by lack of lysyl oxidase like 1 (LOXL1 encoded by Loxl1), combined with microfibril deficiency due to Fbn1 mutation (encoding fibrillin-1, Fbn1^C1041G/+) will display ocular and systemic phenotypes of XFS.

Methods

Loxl1^−/− was crossed with Fbn1^C1041G/+ to create double mutant (dbm) mice. Intraocular pressure (IOP), visual acuity (VA), electroretinogram (ERG) and biometry were characterized in 4 genotypes (wt, Fbn1^C1041G/+, Loxl1^−/−, dbm) at 16 weeks old. Optic nerve area was measured by ImageJ and axon counting was achieved by AxonJ. Deep whole-body phenotyping was performed in wt and dbm mice. Two-tailed Student’s t-test was used for statistical analysis.

Results

There was no difference in IOP between the 4 genotypes. VA was significantly reduced only in dbm mice. The majority of biometric parameters showed significant differences in all 3 mutant genotypes compared to wt, and dbm had exacerbated anomalies compared to single mutants. Dbm mice showed reduced retinal function and significantly enlarged ON area when compared with wt. Dbm mice exhibited severe systemic phenotypes related to abnormal elastic fibers, such as pelvic organ prolapse, cardiovascular and pulmonary abnormalities.

Conclusions

Ocular and systemic findings in dbm mice support functional overlap between fibrillin-1 and LOXL1, two prominent components of exfoliation material. Although no elevated IOP or reduction of axon numbers was detected in dbm mice at 16-week-old, their reduced retinal function and enlarged ON area indicate early retinal ganglion cell dysfunction. Dbm mice also provide insight on the link between XFS and systemic diseases in humans.

INTRODUCTION

Exfoliation glaucoma (XFG) is the most common form of secondary glaucoma due to exfoliation syndrome (XFS) which is a systemic disease with defective connective tissue.¹ While all forms of glaucoma may lead to irreversible blindness if not diagnosed and managed properly, XFG is particularly aggressive, often requiring surgical intervention.² For all forms of glaucoma, the initial site of damage occurs at the optic nerve head (OHN), leading to apoptosis of retinal ganglion cells (RGCs).^3–6 A major hypothesis of glaucoma pathogenesis is that alterations in the biomechanical properties and tissue remodeling of the ONH, which is rich in elastic fibers and collagens, cause damage of RGC axons as they pass through lamina cribrosa (LC) of the ONH.⁷ Deformation of the ONH may induce axon transport deficits^8–10 and eventual degeneration and death of RGCs.

A genetic component of XFS is well recognized. An early study conducted by Allingham and colleagues identified 6 Icelandic families with XFS in two generations, in which transmission of the disease to the second generation was through an affected parent, demonstrating that XFS can be genetically inherited.¹¹ Less than a decade later, in 2007, a landmark genome wide association study (GWAS) identified genomic variants of LOXL1 associated with XFS/XFG.¹² Over the ensuring years, the association of LOXL1 with XFS has been confirmed in many studies across multiple ethnic populations.¹³ LOXL1 encodes one of 5 lysyl oxidase enzymes (LOX, LOXL1–4). The major function of lysyl oxidase-like 1 (LOXL1) is to crosslink elastin, which is a pre-requisite for proper formation and stability of elastic fibers.^14,15 Although the most prominent features of XFS manifest in the eyes, XFS has been associated with systemic disorders, many of them having a defective connective tissue component, such as pelvic floor organ prolapse (POP).^16–20 Defective LOXL1 could well explain systemic manifestations of XFS. Indeed, similar systemic phenotypes due to abnormal elastic fibers are also found in loxl1 knockout (Loxl1^−/−) mice.^15,21

In addition to POP, the association of XFS with cardiovascular diseases has been shown by many studies, although conflicting findings have also been reported.^1,22–24 Mechanistically, the higher prevalence of cardiovascular diseases in XFS patients is plausible as the elastic fibers are rich in cardiovascular system and play important roles in maintaining the stability and normal function of cardiovascular system. Mutations in LOX were reported in hereditary aortopathy which is a serious condition that can lead to aortic dissection and sudden death.²⁵ In addition to LOX, mutations in other genes have been discovered causing hereditary aortopathy, one of which is FBN1 which encodes fibrillin-1.²⁶ FBN1 mutations were first identified in Marfan Syndrome (MFS), and subsequently in other MFS related connective tissue diseases.²⁷ MFS is a multisystem disease with aortopathy as the most common feature.

A major criteria for diagnosing MFS is ectopia lentis due to abnormal zonules²⁸ which are predominantly composed of fibrillin-1.²⁹ LOXL1 was also detected in lens zonules.³⁰ We have previously characterized ocular features of mouse models with Fbn1 and Loxl1 mutations and discovered those two models share some common features, such as deepening of anterior chamber suggesting weakening of lens zonules and enlarged optic nerve cross-sectional areas which is an under-recognized early glaucoma phenotype.^31,32 Both prove to be valuable mouse models to study eye diseases, however, the late onset and slower progression limit their utilization. To overcome this, we created a model with both Fbn1 and Loxl1 mutations (Fbn1^C1041G/+; Loxl1^−/−), termed double mutant (dbm). Here we report the exacerbated ocular and systemic features in dbm mice.

METHODS

ANIMALS AND GENOTYPING

All animal studies were performed in accordance with the Association for Research in Vision and Ophthalmology guidelines for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and Use Committee of Vanderbilt University Medical Center. Fbn1^C1041G/+ mice (previously referred to as Fbn1^C1039G/+)³³ on the C57BL/6J background were originally purchased from the Jackson Laboratory (Bar Harbor, ME; Stock No. 012885). Loxl1^−/− mice on a129S1/SvImJ background were obtained from Dr. Tiansen Li¹⁵ and backcrossed for 10 generations onto the C57BL/6J background. To produce experimental mice on the C57BL/6J background, male mice heterozygous for the Fbn1^C1041G allele and heterozygous for the Loxl1⁻ allele were paired with female mice wt for Fbn1 and heterozygous for the Loxl1⁻ allele. Animals were housed in a facility operated by the Vanderbilt University Division of Animal Care, with 12/12 h light/dark cycle and ad libidum access to food and water. All experiments were conducted in 16-week-old mice.

The genotype of each experimental mouse was determined at weaning and confirmed after sacrificing. For Fbn1 genotyping, we utilized a protocol described on the Jackson Laboratory web site (https://www.jax.org/Protocol?stockNumber=012885&protocolID=28863) in which PCR amplification of DNA extracted from ear punch tissue with primers 10958 (5’-CTC ATC ATT TTT GGC CAG TTG-3’) and primer 10959 (5’GCA CTT GAT GCA CAT TCA CA 3’) resulted in bands of 164 bp for wt and 212 bp for C1041G alleles of Fbn1 (Supplemental Figure 1A). For Loxl1 genotyping, PCR amplification of DNA extracted from ear punch tissue using primers S32 (5’-ACA CGT CGG TGC TGG GAT CA-3’); D5 (5’-CTT TCG TAA ACC AGT ATG AGA ACT ACG ATC-3’); and N5 (5’-CGA GAT CAG CCT CTG TTC CAC-3’) (IDT, Coralville, IA) resulted in bands of ~400 bp for wt and ~310 bp for Loxl1⁻ alleles (Supplemental Figure 1B), as previously described.¹⁵ Routine genotyping was performed by a genotyping service (Transnetyx, Memphis, TN) using proprietary assays that were validated using the PCR protocols described above.

INTRAOCULAR PRESSURE (IOP) MEASUREMENTS

Mice were anesthetized by isoflurane inhalation (2.5% in oxygen) delivered at 1.5 L/min (Vet Equip). IOP of the right eyes was measured within 2 min of loss of consciousness to avoid effects of anesthesia on IOP.³⁴ In addition, to avoid IOP diurnal variation,³⁵ all measurements were conducted at the same time of the day (between 3 pm to 5 pm) using TonoLab tonometer (Icare, Finland). IOP was calculated as the average of 3 separate IOP determinations, each consisting of the mean of six error-free readings.

VISUAL ACUITY MEASUREMENTS

Photopic visual acuity (VA) of mice was assessed by the optomotor response OptoDrum (Stria.Tech, Germany). Mice were placed, unrestrained, on an elevated platform surrounded by computer monitors while striped pattern rotates around the animal, triggering the reflex. A camera above the mouse records the behavior, which is automatically detected and analyzed by OptoDrum software. The stimulus pattern is continuously and automatically adjusted during the experiment to find the animal’s visual threshold (cycles/degree).

ANTERIOR SEGMENT EXAMINATION

We performed clinical examination using portable slip lamp (Kowa SL-17, Torrance, CA) by one of the authors (RWK) masked for mouse genotypes. After pupils were dilated with one drop of tropicamide (1%, Bausch & Lomb) and one drop of phenylephrine (2.5%, Paragon Bioteck), anterior segment examination was performed with attention to any corneal abnormalities, ease of pupillary dilation with or without posterior synechia, cataract formation, and exfoliation material on the pupillary margin and anterior lens capsule.

SPECTRAL DOMAIN OPTICAL COHENRENCE TOMOGRAPHY (SD-OCT)

SD-OCT was carried out as previously described.³⁶ Briefly, mice were anesthetized with ketamine (100 mg/kg) and xylazine (7 mg/kg), wrapped in gauze and placed in a holder. Eyes were kept moist using lubricant eye drops (Refresh Optive^®, Allergan, Irvine, CA). All measurements were obtained with “mouse retina” lens using the BioptigenEnvisu R2200 SD-OCT system for rodents (Leica Microsystems, Wetzlar, Germany), after pupils were dilated with1% tropicamide (Bausch & Lomb, Laval, QC, Canada). Mouse position was adjusted until the appearance of Purkinje lines perpendicular to and parallel to the visual axis and centered on the corneal surface. Images were acquired in a rectangular scan pattern consisting of 100 B-scans, each consisting of 1000 A-scans. Image acquisition was completed before lens opacity or corneal damage appeared due to anesthesia.^37,38 Central corneal thickness (CCT) was determined by digital caliper. The anterior chamber depth (ACD) was defined as the distance from the central posterior surface of the central cornea to the central anterior surface of the lens. Axial length (AL) measurements were determined by the acquisition of a series of three images. 1) A posterior image was used to determine the distance from the outer retinal pigment epithelium to the posterior surface of the lens (vitreous + retina); 2) an anterior image was used to determine the distance from the outer corneal surface to the anterior surface of the lens (CCT + ACD); and 3) an image in which the lens was optically folded in half to determine half of the lens axial diameter (1/2 lens). AL was defined as equal to (vitreous + retina) + (CCT + ACD) + 2 × 1/2 lens. Lens thickness was calculated by 1/2 lens diameter multiplied by 2. Upon completion of imaging, the mice were injected with atipamezole (1 mg/kg; Patterson Veterinary, Greeley, CO, USA) to reverse anesthesia and to prevent xylazine-induced corneal damage.³⁸

ELECTRORETINOGRAM (ERG)

Scotopic ERG responses were measured using the Espion system (Diagnosys) as previously reported.³¹. After dark adaptation overnight, mice were prepared for recordings under dim red illumination. Mice were anesthetized with ketamine/xylazine/urethane (28/11.2/800 mg/kg), and their eyes dilated with one drop of tropicamide (1%, Bausch & Lomb) and one drop of phenylephrine (2.5%, Paragon Bioteck). After placing mice under a Ganzfeld dome with a heating pad, gold electrodes were placed on the corneas and ground electrodes placed subcutaneously at the flank. Flash stimuli consisted of flashes of white light of 4-ms duration generated by light emitting diodes. Waveforms were recorded in response to flashes ranging in intensity from −5 to 0 log cd·s/m², in 1 log increments by averaging responses to multiple consecutive flashes at each intensity. Recordings included a 100-ms prestimulus baseline with data collected up to 500 ms after stimulus onset. Raw data were exported into Excel (Microsoft) for analysis. The pSTR, nSTR and a-wave amplitudes were determined by the peak or trough to baseline. The b-wave amplitudes were measured from the a-wave trough to the b-wave peak. Response latency was defined as the time interval between stimulus onset and the corresponding peak or trough.

OPTIC NERVE EVALUATION

Immediately after euthanization, mice were cardiac perfused with phosphate-buffered saline (PBS) followed by 4% paraformaldehyde (PFA) in PBS. Eyes were enucleated and optic nerves cut approximately 1.5 mm from the globe as previously described³¹ and post-fixed in fixative containing 1% glutaraldehyde and 4% PFA in PBS. Optic nerves were transferred to 2% osmium tetroxide in PBS for 1 hour before dehydration and embedding in Epon-812/Araldite resin (Electron Microscopy Sciences, Hatfield, PA, USA), as previously described.³¹ Using an ultramicrotome (Leica EM UC7, Wetzlar, Germany), 1-μm-thick cross-sections of optic nerves were cut, stained with paraphenylenediamine which darkly stains axoplasm of degenerating axons³⁹ and mounted with Permount Mounting Medium (Thermo Fisher Scientific, Waltham, MA, USA). Stained optic nerve cross-sections were imaged with a 100X 1.45 NA oil immersion objective on a Nikon inverted light microscope equipped with an SLR DS-Ri2 camera (Nikon, Melville, NY, USA). Montage images covering the entire nerve cross-section were assembled using NIS-elements software (Nikon). Optic nerve cross sectional area ~1.5 mm from the globe, excluding the pia mater, was determined as previously described³¹ by drawing a polygon around the nerve using ImageJ (https://imagej.net). The total number of axons was determined using AxonJ, an automated counting plugin for Fiji developed by Zarei et al.⁴⁰

DEEP WHOLE-BODY PHENOTYPING

We observed early and sudden death of dbm mice at around 16 weeks of age, therefore 16 weeks old dbm mice and age-matched wt mice were sacrificed for deep phenotyping. Five wt and 6 dbm mice were necropsied for gross and microscopic examination. Collected tissues (heart, aorta, lung, liver, spleen, kidney, brain, spinal column, skin, pancreas, gastrointestinal tract, reproductive tract, skeletal muscle, adrenal glands, and lymph nodes) were fixed en bloc prior to histology processing, paraffin embedding, sectioning, and routine Hematoxylin & eosin (H&E) staining in the Vanderbilt Translational Pathology Share Resource (TPSR). In addition to H&E staining, sections of the ascending aorta were stained with Movat histochemical stain to examine elastic fiber morphology.

Periodic Acid-Schiff (PAS) Staining

PAS staining has been shown capable of recognizing exfoliation material in a previous study,⁴¹ therefore, we chose to use this method to investigate the presence of exfoliation material on the anterior lens capsule. A PAS kit (Sigma, catalog# 395B) was used following the manufacturer’s instruction. Briefly, 7 μm thick sections were deparaffinized in xylene and re-hydrated in deionized water. Slides were immersed in Periodic Acid Solution for 5 min at room temperature, rinsed with distilled water, then immersed in Schiff’s Reagent for 15 min at room temperature and washed in running tap water for 5 min. Sections were counterstained with Gill No. 3 Hematoxylin Solution for 90 sec and rinsed in running tap water. Finally, sections were dehydrated in gradient ethanol and mounted using Permount mounting media. Brightfield images were acquired using a microscope equipped with 20X objective (Nikon).

STATISTICAL ANALYSIS

All experiments were carried out by masked observers. All data are presented as mean ± standard deviation (SD). Graphs were made and statistical analyses were performed using GraphPad Prism version 9.0 (GraphPad Software, San Diego, CA). Data were analyzed using Student’s two-tailed t-test as indicated in the figure legends. We defined statistical significance as p ≤ .05. Number of measurements and specific p-values are indicated in results or figure legends.

RESULTS

DBM MICE DO NOT HAVE ELEVATED IOP

We previously reported 1-year-old Loxl1^−/− non-anesthetized mice did not have elevated IOP.³² Similarly, we did not observe elevated IOP in Fbn1^C1041G/+ mice when compared with wt mice at 1 year of age (data not shown). Not surprisingly, at 16 weeks of age, as shown in Figure 1A, no elevated IOP was detected in Fbn1^C1041G/+, Loxl1^−/− or dbm mice compared with age-matched wt mice.

Figure 1.

Normal IOP but reduced VA in dbm mice. At 16 weeks old, there were no differences in IOP compared to wt (A). No difference of VA was detected in either Fbn1^C1041G/+ or Loxl1^−/− mice at 16 weeks old compared to wt mice. However, the VA in dbm was reduced compared to wt mice (B). Numbers of mice in each group were indicated above x axis. The p values from Student’s 2-tailed t-tests for comparisons to wt are indicated above the brackets. Numbers of mice in each group are indicated above x axis.

DBM MICE HAVE REDUCED VISUAL ACUITY

As shown in our previous publication, 1-year-old Loxl1^−/− mice showed decreased VA compared with age-matched wt mice.³² We also reported decreased VA in mice carrying the Tsk mutation of Fbn1 (Fbn1^Tsk/+) at the advanced age of 16 months, although not at 6 months of age.³¹ At 16 weeks old, we did not detect VA changes in either Fbn1^C1041G/+or Loxl1^−/− mice as expected. However, as shown in Figure 1B, we observed significantly reduced VA in 16-week-old dbm when compared with age-matched wt mice.

DBM MICE DO NOT HAVE DETECTABLE EXFOLIATION MATERIAL OR CATARACT FORMATION

Slit lamp examination was performed to check for the presence of exfoliation material and to explore the etiology of reduced VA in dbm mice at 16 weeks of age. We did not observe any exfoliation material either on the pupillary margin or anterior lens capsule. All mice experienced equal dilation of the pupils without any evidence of posterior synechia formation. We carefully examined lens and detected no visible cataract formation in any of the mice, regardless of their genotypes. We did not observe the presence of exfoliation material after staining the anterior lens capsule with PAS.

EXACERBATED OCULAR BIOMETRIC CHANGES INDUCED BY FBN1 AND LOXL1 MUTATIONS

Thin cornea is a common ocular feature of Marfan patients.⁴⁰ In addition, thin cornea was observed in embryos of Marfan mice.⁴¹ We previously reported reduced CCT of Loxl1^−/− mice at 1 year old.³² We also reported thinning of CCT of Fbn1^Tsk/+mice, which was detectable as early as 3 months old and persistent up to 9 months old.³¹ Taken together, it was not surprising that we observed by SD-OCT significant thinning of CCT in both Fbn1^C1041G/+and Loxl1^−/− mice at 16 weeks old when compared with wt mice. More significantly, dbm mice showed even thinner CCT compared with wt and each single mutant mice (Figure 2A and 2B).

Figure 2.

Biometric changes in mutant mice. The majority of biometric parameters showed significant differences in all 3 mutant genotypes compared to wt, with exacerbated anomalies found in dbm. Representative OCT images of CCT and ACD measurements in the 4 genotypes, each genotype labeled in lower left corner (A). Thinning of CCT and deepening of ACD in the 3 mutant lines can be appreciated here using wt as a reference. The significant differences of CCT are shown in B. Significant deepening of ACD is shown in C. No difference of lens thickness was seen among the 4 genotypes (D), and mild elongation of axial length was seen in Loxl1^−/− and dmb when compared with wt mice (E). The p-values are shown above or below the brackets indicating comparisons (B, C, D and E). Numbers of mice in each group are indicated above x axis (B, C, D and E).

Similar to CCT, we also observed by SD-OCT other significant biometric changes in dbm mice. The most notable finding was the significant deepening of ACD in dbm mice compared to wt as well as single mutant mice (Figure 2A and 2C). We also noticed significantly deeper ACD in Loxl1^−/− mice compared to wt, but the deepening of ACD in Fbn1^C1041G/+ mice did not reach statistical significance, which could be explained by the mild nature of MFS due to the C1041G mutation and young age. However, the ACD of dbm was markedly deepened when compared to wt as well as single mutant mice. This is not due to the lens thickness as we did not observe any significant differences among all 4 genotypes as shown in Figure 2D. No significant elongation of the eye was observed in mild Marfan Fbn1^C1041G/+ mice, but longer AL was seen in both Loxl1^−/− and dbm when compared to wt mice, although the difference between Loxl1^−/− and dbm was not significant (Figure 2E).

RETINAL FUNCTION IS REDUCED IN DBM MICE

Retinal function measured by ERG, particularly pSTR component has been widely utilized to assess RGC function in rodent glaucoma models as shown previously by us and others.^31,42,43 As shown in Figure 3F, we detected significantly prolonged latency of scotopic threshold response in dbm when compared with wt mice, whereas no difference was detected between Fbn1^C1041G/+ and wt or between Loxl1^−/− and wt mice at 16 weeks of age. Similarly, as indicated by the average waveform of each genotype, reduced peak of scotopic threshold response in dbm was observed as shown in Figure 3A to 3D, although when compared with wt, such reduction was only trending significant. No reduction of response peak was observed in either Fbn1^C1041G/+ or Loxl1^−/− mice.

Figure 3.

Prolonged latency of scotopic threshold response of electroretinography in dbm mice. Raw waveforms of each individual eye of mice with 4 genotypes are indicated in gray and colored waveform represents average of all eyes of mice with 4 genotypes (A-D). There is trend of reduced peak response in dbm mice, but not statistically significant when compared with wt (E). There is prolonged latency in dbm when compared with wt (F). The p values are indicated above the brackets and Numbers of eyes in each group are indicated above x axis.

DBM MICE HAVE ENLARGED OPTIC NERVE WITHOUT AXONAL LOSS

We and others have observed enlarged optic nerve cross sectional area preceding reduced RGC function and axon numbers in rodent glaucoma models.^44–46 We reported enlarged optic nerve cross sectional area in 1-year-old Loxl1^−/− mice.³² We also previously observed same phenotype in Fbn1^Tsk/+ mice at 6 months old and persisted to 16 months old.³¹ At 16-week-old, as shown in figure 4, there was significant enlargement of optic nerve cross sectional area in both Loxl1^−/− and dbm mice when compared to wt. We were not surprised that there was no significant difference between wt and Fbn1^C1041G/+ mice as these Fbn1^C1041G/+ mice have overall mild phenotypes. We also compared Loxl1^−/− and dbm mice and did not observe any significant difference between those two groups, indicating that the enlarged optic nerve phenotype in dbm is driven by the Loxl1^−/− allele. We also performed axon counting using AxonJ in wt and dbm mice. There was no difference between wt and dbm at 16-week-old (data not shown).

Figure 4.

Enlarged optic nerve in dbm mice. When compared with wt, dbm mice showed significantly increased optic nerve area (2-tailed Student’s t-test, P=.012). Similar phenotype was observed in Loxl1^−/− mice when compared with wt, whereas no significant difference between Fbn1^C1041G/+ and wt mice. Comparison between Loxl1^−/− and wt mice showed no significant difference. The p values are indicated above and below the brackets and numbers of eyes in each group are indicated above x axis.

LIFE SPAN IS REDUCED IN DBM

Both Fbn1^C1041G/+ and Loxl1^−/− mice have relatively normal life span, up to 2 years old. However, dbm mice died much younger, starting at about 12 weeks of age. The survival rate dropped to approximately 80% at 16 weeks. Based on these observations, we chose 16 weeks of age for all experiments reported here.

PELVIC ORGAN PROLAPSE (POP) WAS FREQUENTLY OBSERVED IN DBM MICE

One of the phenotypes of Loxl1^−/− mice is POP, as reported in the original publication.¹⁵ We observed 44% dbm mice experienced POP within their life span as shown in Table 1. The percentage of POP in female mice was higher than that in male mice (54% vs 33%). We did not observe any wt mice with POP at this age.

Incidence of POP in dbm mice

Number of mice	25
Number and percentage (%) of mice with POP	11 (44%)
Number of male mice	12
Number and percentage (%) of male mice with POP	4 (33%)
Number of female mice	13
Number and percentage (%) of female mice with POP	7 (54%)

DBM MICE HAVE DISORDERS IN CARDIOVASCULAR AND PULMONARY SYSTEMS

The early death in dbm mice prompted us to perform necropsy. As aortic dissection is a common cause of sudden death in patients with MFS,²⁶ we expected to see hemothorax in dbm mice. Interestingly, we instead observed cardiomegaly without obvious hemothorax. This further prompted us to perform thorough gross and histological deep phenotyping of dbm mice. As shown in Figure 5C, we observed markedly dilated atria and ascending aorta in dbm mice. Although we did not observe gross body size abnormalities, we measured tibia length, which revealed no difference between wt and dbm mice (data not shown). Using tibia length for normalization, the heart weight in dbm was significantly greater than wt as shown in Figure 5A. In addition, the width of ascending aorta of dbm was significantly increased compared to wt (Figure 5B). Movat Pentachrome histochemical staining of the ascending aorta showed thinning and fragmented elastic fibers in dbm mice (Figure 5D).

Figure 5.

Enlarged heart and aorta phenotype of dbm compared to wt mice. Significant cardiomegaly was observed in dbm compared with wt mice (A). The ascending aorta width of dbm was also significantly dilated compared to wt mice (B). Gross comparison of hearts from wt and dbm hearts showing both right and left atria (chevrons) were markedly dilated, and the ascending aorta (arrowheads) was visibly wider in dbm hearts (scale bar = 5mm). Movat Pentachrome histochemical staining of the ascending aorta showed linear and continuous elastic fibers in wt. In contrast the elastic fibers are thin, fragmented and discontinuous in dbm mice (D) (scale bar = 50 μm).

Although the histological findings of aortic wall are typical of MFS with aneurism, we suspected other causes of sudden death in dbm mice. Hematoxylin and eosin staining of the heart revealed myxomatous valvular degeneration of both atrioventricular (AV) valves in dbm mice as shown in Figure 6. Both mitral and tricuspid valve leaflets are elongated with irregular and segmental thickening of the spongiosa in dbm mice. The valvulopathy and associated cardiac chamber enlargement likely contributes to early death in dbm mice.

Figure 6.

Myxomatous valvular degeneration of both atrioventricular (AV) valves in dbm mouse hearts. Hematoxylin & eosin staining of wt mouse heart showed consistently smooth and thin tricuspid and mitral valves (A). Both AV valve leaflets are elongated with irregular and segmental thickening of the spongiosa (B). (tricuspid valve: black arrowheads; mitral valve: white arrowheads). Scale bar = 500 μm.

In addition, as shown in Figure 7, severe pulmonary emphysema and bronchiestasis were detected in dbm mice with abundant evidence of diffusely dilated alveoli and bronchioles. We also observed multifocal leukocytoclastic arteritis in the abdominal aorta, peripheral muscular arteries and aortic root as shown in Supplemental Figure 2.

Figure 7.

Pulmonary emphysema and bronchiectasis in dbm mouse lungs. Lung section from wt mice showed normal bronchioles (Br) and alveoli (Alv) (A), whereas dbm mouse lungs had diffusely dilated alveoli and bronchioles. (B) (scale bar = 100 μm).

DISCUSSION

The discovery of genomic variants of LOXL1 associated with XFS by GWAS¹² has been replicated by many studies, including the largest GWAS using geographically comprehensive samples with diverse populations.¹³ However, the LOXL1 allele flip phenomenon was also well recognized, indicating abnormal LOXL1 is necessary for XFS development, but not sufficient, suggesting that additional factors are needed.⁴⁷ We hypothesize that one of the additional factors could be fibrillin-1, which is a key component of microfibril scaffold for elastin deposition and formation of stable elastic fibers.^48,49 The interaction between fibrillin-1 and LOXL1 has been shown outside the eye. In a study by Busnadiego et al., when Fbn1^C1041G/+ mice were challenged with β-aminopropionitrile (BAPN) which inhibits all lysyl oxidases, mice with mild MFS developed accelerated aortopathy, including ascending aorta dilation with increased aortic wall elastic fiber fragmentation resulting in early death.⁵⁰ Within the eye, we utilized the same protocol by Busnadiego et al. to treat Fbn1^C1041G/+ mice with BAPN to test if double hits with both fibrillin-1 and LOX defects will worsen the ocular phenotypes. We discovered that inhibition of entire LOX family proteins indeed enhanced optic nerve expansion in Fbn1^C1041G/+ mice.⁵¹ Although in that study, the entire LOX family (LOX, LOXL1-4) was inhibited by BAPN, it lends support for overlapping and possibly, interacting roles of fibrillin-1 and LOXL1. In terms of their roles within the eye, specifically related to XFS, LOXL1 and fibrillin-1 are two major components of exfoliation material.⁵² In addition, using human samples, Schlotzer-Schrehardt demonstrated reduced expression of LOXL1 and fibrillin-1 in lamina cribrosa of XFS and XFG patients.⁵³ To focus on interaction specifically between LOXL1 and fibrillin-1, we created our unique dbm mouse model with haploinsufficiency C1041G mutation of Fbn1 and knockout of Loxl1 allele to investigate ocular and systemic manifestations by deep phenotyping and in-depth ocular structural and functional analyses.

We first investigated biometric features of dbm mice and compared them with wt, Fbn1^C1041G/+ and Loxl1^−/− mice at 16 weeks old. We detected thinning of CCT of dbm mice. We previously reported ocular findings of mice carrying a different Fbn1 mutation (Fbn1^Tsk/+).⁴⁴ The systemic phenotypes of Fbn1^Tsk/+ mice include thickened skin and visceral fibrosis.⁵⁴ In the eyes, we discovered that they have thin CCT, which could be detected as early as 3 months of age. This phenotype likely occurred earlier, as we observed persistent but non-progressive nature of thin CCT in Fbn1^Tsk/+ mice up to 9 months of age. Corneal thinning is a common ocular phenotype of Marfan syndrome in human patients.⁵⁵ When comparing adult with pediatric MFS patients, Suwal et al. did not detect significant difference between them, indicating non-progressive nature of thin CCT.⁵⁶ Research led by Meek using different Marfan mouse model revealed thinning of CCT detectable from embryonic stage to adulthood of 3 months old.^41,57 Although the Fbn1^C1041G/+ Marfan mouse model reported here is considered a mild Marfan model because of its mild and late onset of cardiovascular phenotypes,³³ we were not surprised to see the thinning of CCT of Fbn1^C1041G/+ mice at 16 weeks of age, and we speculate this was congenital. It is worth noting that FBN1 has been associated with central corneal thickness, a known risk factor for primary open angle glaucoma in a cross-ancestry GWAS study⁵⁸ as well as in a large Australia population study⁵⁹. Similarly, we reported thinning of CCT in Loxl1^−/− mice at 1 year of age.³² In this study, in line with previous findings, we observed thinning of CCT in both Fbn1^C1041G/+ and Loxl1^−/− mice at 16 weeks old. Fibrillin-1 and all 5 LOX family members are present in cornea, and LOX mutations have been found in keratoconus patients,^60,61 although the role of LOXL1 in keratoconus is less clear. Intriguingly, we observed exacerbated thinning of CCT in dbm mice when compared to each single mutant genotype (Figure 2B), indicating interaction of fibrillin-1 and LOXL1 within the eye.

In addition to thin corneas, one of the key findings of Marfan syndrome is progressive ectopia lentis due to weakened zonular fibers which are largely composed of fibrillin-1 protein.⁵⁵ Because it is challenging to detect ectopia lentis in mice by slit lamp examination, even at advanced stage, we utilized SD-OCT to measure the ACD. As shown in Supplemental Figure 3, we detected extreme deepening of the ACD in Fbn1^Tsk/+ mice at advanced age likely due to posterior movement of lens caused by abnormal zonular fibers. This lends support for utilizing SD-OCT measurement of ACD as a surrogate of zonule stability. We did not observe deepening of ACD in Fbn1^C1041G/+ mice at 16 weeks old, which was consistent with mild nature of the C1041G mutation of Fbn1 in terms of MFS phenotypes and the progressive nature of ectopia lentis in MFS. We did observe deepening of the ACD in Loxl1^−/− mice which recapitulate some XFS phenotypes, as dislocated lens due to zonular fiber defects is a common phenotype of XFS.¹ Using the same Loxl1^−/− mice but at older age (1 year old), Wiggs et al., reported a number of ocular features resembling XFS in human patients, such as cataract formation and blood-aqueous barrier breakdown.⁶² Although the biometric phenotypes observed by us here as well as ocular phenotypes reported by Wiggs et al, are mild, the deepening of ACD in Loxl1^−/− mice at 16 weeks of age is consistent with an XFS phenotype. More significantly, similar to CCT, we observed exacerbated deepening of ACD in dbm mice, further supporting the interactions between fibrillin-1 and LOXL1. It is interesting to note that the key function of LOXL1 is cross-linking tropoelastin to elastin for stable elastic fiber formation.⁶³ Lens zonules are composed of elastin-free microfibrils,²⁹ yet LOXL1 protein was detected in human and bovine zonules,³⁰ raising the possibility of LOXL1 acting on fibrillin-1, in addition to crosslinking elastin.

We also detected reduced VA in dbm mice. Although the reduction of VA was mild in dbm, this was not observed in Fbn1^C1041G/+ or Loxl1^−/− mice. While the reduced VA could simply be due to refractive error from biometric changes in dbm mice, it could also be due to reduced retinal function as shown in Figure 3. The enlarged optic nerve phenotype of dbm mice as shown in Figure 5 further supports this possibility. Although reduced VA generally occurs much later in human glaucoma patients, we cannot rule out the possibility of reduced VA in dbm mice related to glaucoma. Interestingly, the dbm mice did not exhibit elevated IOP as shown in Figure 1A. It is well established that CCT affects IOP measurement in humans,⁶⁴ however, we do not believe normal IOP in all three mutant genotypes was influenced by thin CCT of those mice, as we previously demonstrated.⁴⁴ This promotes the notion that the reduced retinal function and enlarged optic nerve phenotypes are independent of IOP. It has been recognized that XFG occurs in some patients without elevated IOP.^65,66 The observation of reduced fibrillin-1 and LOXL1 expression in the lamina cribrosa of patients with both XFS and XFG reported by Schlotzer-Schrehardt et al. supports the significant contribution of optic nerve head structure in XFG pathogenesis.⁵³

We observed a series of systemic findings through deep phenotyping of dbm mice. The dbm mice sustained early mortality ranging from 12 to 16 weeks of age. This is consistent with increased mortality in BAPN-treated Fbn1^C1041G/+ mice attributed to exacerbated dilated aorta.⁵⁰ The early death of dbm mice prompted us to perform necropsy initially, which revealed cardiomegaly with severe bilateral atrial enlargement. This observation led us to deep phenotyping of dbm mice at 16 weeks old which revealed cardiomegaly, aortic dilation and valvulopathy in dbm mice. These findings suggest the early death of the dbm may be largely attributed to these cardiac changes and potentially leading to congestive heart failure or fatal cardiac arrythmia. Further investigations likely will shed more light on the observation of XFS associated with atrial fibrillation in humans.²⁰

The diagnosis of XFS is through the readily detectable exfoliation material in the anterior segment of the eye. The discovery of exfoliation material elsewhere in the body confirms the systemic manifestations of XFS,⁶⁷ although the existing data in the literature on the association with systemic diseases, especially cardiovascular disorders, remain controversial. Using a large electronic medical records database of Maccabi Health Services, Zehavi-Dorin et al. found that among individuals with XFS, the risk of cardiovascular diseases, including hypertension, myocardial infarction and congestive heart failure, was significantly higher than the control group.⁶⁸ The Blue Mountain Eye Study showed that XFS was associated with a history of angina or hypertension or a combined history of angina, acute myocardial infarction, or stroke.⁶⁹ In another study, in addition to confirming an increased risk of respiratory, cardiovascular, and urogenital comorbidities, Scharfenberg et al. found an increased risk of cardiac valve disorders among XFS patients.⁷⁰ However, in a Russian population, no significant association between XFS and history of cardiovascular disease was found.⁷¹ It is interesting to note that even in highly controlled homogenous genetic background and other variables, there is a large heterogeneity of cardiomegaly in dbm mice (Figure 5A). It is also intriguing that the enlarged optic nerve phenotype in dbm mice also exhibits large heterogeneity. Although we did not observe cataract formation or exfoliation material in dbm mice, which could be attributed by the young age of those mice, because of other findings we believe dbm mice should be further explored as an excellent model to understand both ocular and cardiovascular disorders in XFS patients.

The dbm mice also exhibit other systemic findings commonly observed in XFS patients. Using the Utah database, Wirostko et al. discovered increased risk of XFS in women with POP.¹⁶ As shown in Table 1, we observed 44% rate of POP in dbm mice, females greater than males, and none of the female dbm mice had history of pregnancy. The average age of detected POP was 12 weeks. One female mouse had POP at 3 weeks of age, and interestingly, that individual died earlier than average. The POP phenotype was well documented in the original report of Loxl1^−/− mice, and it is correlated with abnormal elastic fibers of pelvic floor.¹⁵ The Loxl1^−/− mice have since been extensively studied as a model of POP, especially pregnancy induced POP.^21,72 The earlier onset of POP unrelated to pregnancy in our dbm mice indicate that dbm may be a better model to understand POP in human patients with XFS. In addition to POP, the Utah database also revealed increased risk of chronic obstructive pulmonary disease in patients with XFS.¹⁸ We observed severe pulmonary emphysema and bronchiectasis in dbm mice, which further supports the utility of this model for human XFS. Lastly, the association between XFS and retinal vein occlusion has been frequently observed, although the mechanisms remain unknown.⁷³ We observed striking multifocal leukocytoclastic arteritis in dbm mice. It remains speculative that the arterial wall changes observed in the dbm mice may be present in XFS patients which may contribute to the higher incidence of retinal vein occlusion.

There are limitations of the dbm model because exfoliation material deposition, elevated IOP and RGC loss, features commonly observed in XFS/XFG patients were not observed. We speculate this is largely attributed by the young age of those mice as XFS/XFG is an age-related condition in humans. The unique mouse model with microfibril and elastic fiber defects present many other ocular and systemic disorders that are similarly observed in patients with XFS. Although mice differ from humans, the model offers opportunities to investigate XFS pathogenesis.