Defect of TIMP4 Is Associated with High Myopia and Participates in Rat Ocular Development in a Dose-Dependent Manner

Abstract

Thinning of the sclera happens in myopia eyes owing to extracellular matrix (ECM) remodeling, but the initiators of the ECM remodeling in myopia are mainly unknown. The matrix metalloproteinase (MMPs) and tissue inhibitors of matrix metalloproteinase (TIMPs) regulate the homeostasis of the ECM. However, genetic studies of the MMPs and TIMPs in the occurrence of myopia are poor and limited. This study systematically investigated the association between twenty-nine genes of the TIMPs and MMPs families and early-onset high myopia (eoHM) based on whole exome sequencing data. Two TIMP4 heterozygous loss-of-function (LoF) variants, c.528C>A in six patients and c.234_235insAA in one patient, were statistically enriched in 928 eoHM probands compared to that in 5469 non-high myopia control (p = 3.7 × 10⁻⁵) and that in the general population (p = 2.78 × 10⁻⁹). Consequently, the Timp4 gene editing rat was further evaluated to explore the possible role of Timp4 on ocular and myopia development. A series of ocular morphology abnormalities in a dose-dependent manner (Timp4^−/− < Timp4^+/− < Timp4^+/+) were observed in a rat model, including the decline in the retinal thickness, the elongation in the axial length, more vulnerable to the form deprivation model, morphology changes in sclera collagen bundles, and the decrease in collagen contents of the sclera and retina. Electroretinogram revealed that the b-wave amplitudes of Timp4 defect rats were significantly reduced, consistent with the shorter length of the bipolar axons detected by HE and IF staining. Heterozygous LoF variants in the TIMP4 are associated with early onset high myopia, and the Timp4 defect disturbs ocular development by influencing the morphology and function of the ocular tissue.

1. Introduction

Myopia is an epidemic worldwide, with an increasing incidence rate in recent years [1]. High myopia (HM) is a severe form of myopia defined as a refraction error greater than −6.00 diopters (D) or an axial length greater than 26 mm. HM is often accompanied by pathological changes such as scleral thinning, retinal choroidal atrophy, posterior scleral staphyloma, macular degeneration, and choroidal neovascularization, and has become one of the leading causes of blindness [2]. While genetic and environmental factors play essential roles in the occurrence and development of myopia, early-onset high myopia (eoHM) is mainly caused by genetic factors occurring before school age [3,4]. To date, 27 loci (MYP1-MYP28, MYP4 was deprived) and ten genes (OPN1LW, OMIM 300822; SCO2, OMIM 604272; ZNF644, OMIM 614159; CCDC111, OMIM 615421; LRPAP1, OMIM 104225; SLC39A5, OMIM 608730; P4HA2, OMIM 600608; ARR3, OMIM 301770, CPSF1, OMIM 606027; LOXL3, OMIM 607163) are identified to be associated with myopia by linkage analysis combined with whole-exome sequencing (WES) or WES alone, which account for less than 15% patients with eoHM [5,6]. Additionally, variants in RetNet research have been identified in approximately 25% of the eoHM probands based on our previous studies [7,8]. The genetic factors for about 60% of the cause of eoHM are still unknown. Genome-wide association studies on high myopia have reported a package of related single-nucleotide polymorphisms (SNPs), but the role of these SNPs or genes harboring SNPs in the mechanism of myopia is largely unknown [9].

The occurrence and development of myopia are strongly related to scleral matrix remodeling [10] and excessive axial length elongation [11]. The scleral changes are accompanied by reduced synthesis and excessive scleral extracellular matrix (ECM) degradation. Collagen is a primary component of the ECM, and collagen disease is often accompanied by pathological high myopia such as Stickler syndrome and Marfan syndrome [12,13] because of its essential role in maintaining the structural integrity of ocular tissues such as cornea, retina, and sclera [14]. The homeostasis of the ECM is mainly regulated by an enzyme balance system, including the matrix metalloproteinase (MMPs), which degrade the ECM proteins, and the tissue inhibitor of matrix metalloproteinase (TIMPs), which inhibit the MMPs [15]. Twenty-nine genes (including the MMP 23A and MMP 23B) of the MMPs and four TIMPs have been found in vertebrates [16]. Several investigations reported an increase of the TIMPs and MMPs in aqueous humor in high myopia patients [17], where the TIMPs served as a compensatory response to inhibit the MMP-induced degradation of the ECM, thereby constraining myopia’s progression [18]. MMP-2 significantly upregulated in myopia animal models such as tree shrews, chickens, and mice [19,20,21]. The dysregulation of MMP/TIMP balance may contribute to myopia development, but none have been genetically studied in myopia patients.

This study aims to explore the relationship between the thirty-three MMPs/TIMPs genes and eoHM. We further probed the underlying mechanisms by gene mutant rat model.

2. Results

2.1. The TIMP4 Loss-of-Function Variants Clustered in Individuals with High Myopia

We screened variants in twenty-nine TIMP and MMP genes (

Table A1. MMPs and TIMPs families screened in our cohort.

Gene Name	Location	NCBI Gene ID	OMIM
MMP1	11q22.2	4312	120353
MMP2	16q12.2	4313	120360
MMP3	11q22.2	4314	185250
MMP7	11q22.2	4316	178990
MMP8	11q22.2	4317	120355
MMP9	20q13.12	4318	120361
MMP10	11q22.2	4319	185260
MMP11	22q11.23	4320	185261
MMP12	11q22.2	4321	601046
MMP13	11q22.2	4322	600108
MMP14	14q11.2	4323	600754
MMP15	16q21	4324	602261
MMP16	8q21.3	4325	602262
MMP17	12q24.33	4326	602285
MMP19	12q13.2	4327	601807
MMP20	11q22.2	9313	604629
MMP21	10q26.2	118856	608416
MMP23A	1p36.33	8511	603320
MMP23B	1p36.33	8510	603321
MMP24	20q11.22	10893	604871
MMP25	16p13.3	64386	608482
MMP26	11p15.4	56547	605470
MMP27	11q22.2	64066	618101
MMP28	17q12	79148	608417
TIMP1	Xp11.3	7076	305370
TIMP2	17q25.3	7077	188825
TIMP3	22q12.3	7078	188826
TIMP4	3p25.2	7079	601915

) using WES data from 928 eoHM individuals without related syndromic diseases and 5469 non-HM individuals with different eye conditions. Two heterozygous loss-of-function (LoF) variants in the TIMP4 gene were detected in probands from seven unrelated families (

Table 1. Clinical features of high myopia probands carry the TIMP4 LoF variants.

	Nucleotide Change		BCVA		RE		AL (mm)
ID	(NM_003256.4)	Onset Age	OD	OS	OD	OS	OD	OS	Fundus	VEP
P1	c.528C>A	12	1.0	1.0	−8.00	−8.50	/	/	the bilateral temporal lens with cloudy anterior and posterior cortex, large optic disc with conus	/
P2	c.528C>A	2	0.5	1.0	−9.00	−6.00	/	/	large optic papilla, light temporal retina, and peripheral tigroid fundus, fovea reflex present	/
P3	c.528C>A	5	0.9-	0.9−	−6.25	−5.25	25.18	24.89	/	+
P4	c.528C>A	4	0.8-	0.8−	−10.25	−11.25	/	/	tigroid fundus, boundary line on the mid-peripheral retina	/
P5	c.528C>A	3	0.1	0.1	−11.25	−11.00	26.85	26.98	light nipple color, thin arteries, tigroid fundus, arc spots, poor retinal luster, peripheral retina degeneration, boundary line on the peripheral retina	+
P6	c.528C>A	12	1.0	1.0	−6.25	−7.25	26.37	26.83	conus and tigroid fundus	/
P7	c.234_235insAA	4	0.7	0.7	−10.75	−11.50	/	/	tigroid fundus with fovea reflex present	/

BCVA, best corrected visual acuity; RE, refraction error; AL, axial length; OD, right eye; OS, left eye; VEP, visual evoked potential; /, Not available; +, the incubation period is severely delayed, and the amplitude is normal.

), including c.528C>A (p.C176*) in six probands and c.234_235insAA (p.M80Rfs*26) in one proband (Figure 1A). None of the other known causative variants for HM or genetic eye disease were detected in these probands. The frequency of the two heterozygous LoF variants in the TIMP4 was 3.77 × 10⁻³ (7/1856) in our 928 eoHM cohort, 1.83 × 10⁻⁴ (2/10938) in the 5469 non-HM participants, 9.78 × 10⁻⁵ (26/285849) in gnomAD overall population (Figure 1B and Figure A2B) and was 7.06 × 10⁻⁴ (14/19843) in the gnomAD East Asia population. The frequency of the LoF TIMP4 allele in the HM cohort was significantly higher than that in non-HM populations (Fisher’s exact test, p = 3.7 × 10⁻⁵) and in the gnomAD population (Fisher’s exact test, p = 2.78 × 10⁻⁹) as well as in the East Asia population (Fisher’s exact test, p = 1.0 × 10⁻³) (Figure 1B). The LoF variants in the TIMP4 were relevant to eoHM with the mean OR value of 20.70 (CI: 4.30–99.73, p = 3.70 × 10⁻⁵), indicating that individuals with the TIMP4 LoF variants are approximately 20.7 times more likely to develop eoHM compared to those without the LoF variants. Ocular phenotypes of 7 probands with the heterozygous LoF variants were shown in Table 1. The spherical equivalent (SE) fraction was measured to be over −6.00 D in all probands. The typical pathological changes in high myopia were observed, including tigroid fundus, temporal crescent of the optic nerve, and latency delay in the visual evoked potential (VEP) (Table 1, Figure 1C). Co-segregation was observed in family P1, in which the father of proband 1 (P1) carrying heterozygous c.528C>A began to develop myopia at the age of seven and exhibited a SE of −18.00 D in both eyes, along with an ocular axial length of over 34 mm. The mother of P1, who did not harbor this variant, was not diagnosed with high myopia.

The nucleotide change at site 528 encodes amino acid No. 176, which participates in disulfide bond formation and plays a crucial role in protein stability. The frameshift variant c.234_235insAA was predicted to cause premature termination of protein coding, triggering nonsense-mediated mRNA decay (NMD), according to several in silico bioinformatic tools. The two variants are both in the Netrin Module (NTR) domain (Figure 1D) and are highly conserved across vertebrates (Figure A1A). The bioinformatics tool predicted that both variants would dramatically change the corresponding three-dimensional peptide structure of the TIMP4 and might result in an incomplete main structure of the protein, including the NTR domain, and thus pathogenic (Figure 1D).

2.2. Timp4 Gene Editing Rat

Gene-editing rats were constructed to evaluate the potential functional roles of the TIMP4 on eyeball development. The exon 1 and exon 5 of the Timp4 of rats were selected as target sites of CRISPR/Cas9-mediated genome engineering, which resulted in the loss of the whole Timp4 gene in the genome (Figure 2A). Sanger sequencing confirmed the absence of fragment flanking exon 1 to exon 5 of the Timp4 based on genomic DNA extracted from the blood from the tail of the Timp4-gene editing rat (Figure 2B). Three kinds of genotypes, Timp4^+/+, Timp4^+/−, and Timp4^−/−, are present by DNA electrophoresis (Figure 2C). Western blot revealed that the expression of the Timp4 in three-week-old rats’ retinas dramatically decreased after knock-out compared with wild-type (Figure A2A).

2.3. Timp4 Expression in the Retina

To identify the tissue and cell types in which the TIMP-4 protein is located, we further investigated its distribution in the human and rat retina at the protein level by immunohistochemistry (IF). In the human retina, immunostaining was predominantly observed in the outer plexiform layer (OPL) and inner nuclear layer (INL) (Figure 2D). In the rat retina, there was a higher fluorescence density at the OPL with weak staining in INL (Figure 2D). TIMP4 is mainly found in the OPL, which consists of a dense network of synapses from bipolar cells and photoreceptor cells.

2.4. Ocular Morphology and Electroretinography Changes in Timp4^−/− Rats in Normal Development

The CT, ACD, VCD, RT, VCD, and axial length (AL) of Timp4^+/+, Timp4^+/−, and Timp4^−/− rats were measured throughout 3 to 52 weeks (

Table A2. Axial length, retina thickness, vitreous chamber depth, cornea thickness, anterior chamber depth, and lens thickness of three groups were measured by SD-OCT.

	Timp4+/+		Timp4+/−		Timp4−/−
	Mean	SD	Mean	SD	Mean	SD	p Value
Axial length (mm)
3W	5.143	0.025	4.984	0.080	5.246	0.033	1.0 × 10−5
6W	5.611	0.059	5.664	0.089	5.660	0.102	0.401
9W	5.923	0.084	5.891	0.119	5.892	0.094	0.007
10W	5.789	0.040	5.977	0.068	5.928	0.036	0.215
15W	6.084	0.116	5.985	0.070	5.949	0.042	0.786
Retina thickness (mm)
3W	0.2382	0.0121	0.2290	0.0054	0.2212	0.0075	3.6 × 10−5
6W	0.2233	0.0147	0.2171	0.0086	0.2073	0.0055	4.4 × 10−4
9W	0.2148	0.0051	0.2052	0.0062	0.2012	0.0088	0.013
10W	0.2103	0.0061	0.2051	0.0066	0.2009	0.0083	0.022
15W	0.2078	0.0027	0.2015	0.0072	0.2000	0.0094	0.213
20W	0.1984	0.0055	0.1948	0.0086	0.1867	0.0042	0.163
32W	0.1689	0.0026	0.1638	0.0027	0.1602	0.0037	0.055
52W	0.1623	0.0017	0.1591	0.0012	0.1582	0.0041	0.088
Vitreous chamber depth (mm)
3W	1.310	0.031	1.326	0.035	1.325	0.039	0.472
6W	1.310	0.046	1.328	0.050	1.352	0.052	0.079
9W	1.288	0.047	1.318	0.039	1.327	0.054	0.353
10W	1.300	0.033	1.290	0.046	1.295	0.064	0.814
15W	1.246	0.031	1.249	0.029	1.287	0.042	0.113
20W	1.257	0.025	1.232	0.008	1.253	0.021	0.725
32W	1.191	0.012	1.187	0.015	1.213	0.016	0.119
52W	1.180	0.030	1.155	0.033	1.171	0.025	0.165
Cornea thickness (mm)
3W	0.107	0.011	0.093	0.013	0.112	0.019	0.002
6W	0.122	0.016	0.124	0.016	0.118	0.015	0.618
9W	0.121	0.022	0.129	0.018	0.126	0.024	0.819
10W	0.135	0.012	0.137	0.019	0.133	0.014	0.857
15W	0.143	0.013	0.141	0.007	0.130	0.012	0.370
20W	0.141	0.020	0.137	0.003	0.139	0.013	0.161
32W	0.158	0.016	0.161	0.003	0.162	0.010	0.959
52W	0.171	0.010	0.163	0.008	0.162	0.012	0.862
Anterior chamber depth (mm)
3W	0.530	0.034	0.515	0.047	0.524	0.036	0.641
6W	0.621	0.048	0.643	0.042	0.612	0.037	0.103
9W	0.676	0.026	0.674	0.026	0.685	0.041	0.843
10W	0.725	0.041	0.687	0.032	0.628	0.052	0.002
15W	0.737	0.022	0.769	0.036	0.682	0.034	0.075
20W	0.786	0.024	0.787	0.044	0.722	0.032	0.633
32W	0.923	0.068	0.987	0.072	0.927	0.047	0.359
52W	1.087	0.055	1.023	0.067	0.987	0.042	0.163
Lens thickness (mm)
3W	2.932	0.024	2.839	0.037	3.007	0.070	1.3 × 10−5
6W	3.359	0.050	3.359	0.045	3.358	0.079	0.079
9W	3.515	0.021	3.661	0.083	3.589	0.081	0.295
10W	3.728	0.060	3.702	0.015	3.676	0.013	0.751
15W	3.908	0.053	3.830	0.024	3.873	0.039	0.135

Post hoc comparison results: 3W retina thickness: F = 13.256, η² = 0.393, p < 0.001, Timp4^+/+ vs. Timp4^+/− p = 0.125, Timp4^+/+ vs. Timp4^−/− p = 0.005 < 0.01; 6W retina thickness: F = 8.708, η² = 0.209, p < 0.001, Timp4^+/+ vs. Timp4^+/− p = 0.117, Timp4^+/+ vs. Timp4^−/− p < 0.001; 9W retina thickness: F = 6.041, η² = 0.463, p = 0.013 < 0.05, Timp4^+/+ vs. Timp4^+/− p = 0.040 < 0.05, Timp4^+/+ vs. Timp4^−/− p = 0.026 < 0.05; 10W retina thickness: F= 4.289, η² = 0.211, p = 0.022 < 0.05, Timp4^+/+ vs. Timp4^+/− p = 0.032 < 0.05, Timp4^+/+ vs. Timp4^−/− p = 0.025 < 0.05; 3W axial length: F = 25.574, η² = 0.761, p < 0.001, Timp4^+/+ vs. Timp4^+/− p < 0.001, Timp4^+/+ vs. Timp4^−/− p = 0.045 < 0.05; 9W axial length: F = 11.035, η² = 0.759, p = 0.007 < 0.01, Timp4^+/+ vs. Timp4^+/− p = 0.002 < 0.01, Timp4^+/+ vs. Timp4^−/− p = 0.015 < 0.05. The number of rats in each group at different time points was provided in Figure 2.

). The RT of Timp4^−/− rats was thinner than those of Timp4^+/+ rats and decreased in a Timp4 dose-dependent manner (Timp4^−/− < Timp4^+/− < Timp4^+/+). These significant differences of RT were observed at four timepoints including 3W (Timp4^+/+ = 0.2382 ± 0.0121 mm, Timp4^+/− = 0.2290 ± 0.0054 mm, Timp4^−/− = 0.2212 ± 0.0075 mm, p = 3.6 × 10⁻⁵), 6W (Timp4^+/+ = 0.2233 ± 0.0147 mm, Timp4^+/− = 0.2171 ± 0.0086 mm, Timp4^−/− = 0.2073 ± 0.0055 mm, p = 4.4 × 10⁻⁴ < 0.001), 9W (Timp4^+/+ = 0.2148 ± 0.0051 mm, Timp4^+/− = 0.2052 ± 0.0062 mm, Timp4^−/− = 0.2012 ± 0.0088 mm, p = 0.013 < 0.05), 10W (Timp4^+/+ = 0.2103 ± 0.0061 mm, Timp4^+/− = 0.2051 ± 0.0066 mm, Timp4^−/− = 0.2009 ± 0.0083 mm, p = 0.022 < 0.05). The thinner retina was presented in a Timp4 dose-dependent manner, similar to other myopia gene-editing animal models [22]. The axial length of Timp4^−/− rat was longer than those of Timp4^+/+ and Timp4^+/− rats in 3 and 9 weeks (3W: Timp4^+/+ = 5.143 ± 0.025 mm, Timp4^+/− = 4.984 ± 0.080 mm, Timp4^−/− = 5.246 ± 0.033 mm, p = 1.04 × 10⁻⁵ < 0.001; 9W: Timp4^+/+ = 5.923 ± 0.084 mm, Timp4^+/− = 5.891 ± 0.119 mm, Timp4^−/− = 5.892 ± 0.094 mm, p = 0.007 < 0.01) (Figure 2E). Also, a visual tendency of VCD was observed in the different groups from 3 weeks to 52 weeks (Timp4^−/− > Timp4^+/− > Timp4^+/+), but no significant difference was found yet. No significant differences or tendencies were found among the three groups at most time points of the other ocular parameters, including CT, ACD, and LT (Figure A3).

ERG changes in the three groups were assessed at three weeks and six weeks of age, corresponding to the time points of the remarkable eye tissue parameter differences observed in the Timp4 gene editing rats. Analysis of scotopic ERG (dark-adapted ERG) in the Timp4 gene editing rats revealed that the b-wave amplitude was reduced in both the Timp4-gene editing rats compared to wild-type rats at 3 and 6 weeks of age under different intensities. A significant difference was observed at six weeks of age (One-way ANOVA, Timp4^−/− < Timp4^+/− < Timp4^+/+, 0.01 log scotopic cd.s/m²: Timp4^+/+ = 150.17 ± 31.32 μV, Timp4^+/− = 106.31 ± 29.94 μV, Timp4^−/− = 118.70 ± 31.21 μV, p = 0.037 < 0.05; 3.0 log scotopic cd.s/m²: Timp4^+/+ = 234.33 ± 37.58 μV, Timp4^+/− = 193.30 ± 21.72 μV, Timp4^−/− = 165.63 ± 6.41 μV, p = 0.007 < 0.01; 10.0 log scotopic cd.s/m²: Timp4^+/+ = 202.67 ± 29.14 μV, Timp4^+/− = 175.60 ± 20.67 μV, Timp4^−/− = 155.11 ± 9.77 μV, p = 0.029 < 0.05). No significant difference was observed in the a-wave amplitude among the groups (Figure 2F).

2.5. Ocular Morphology Changes of Timp4 Defect Rats under form Deprivation Pressure

Since the axial length elongation was observed in the Timp4 gene editing rats discontinuously, we burdened the environment pressure of form deprivation (FD) model on gene-editing rats to enlarge the possible influence of the Timp4 knockdown on ocular development (Figure A2C). Fifteen rats from three genotypes were used (n = 5, respectively). After wearing a translucent diffuser for 72 h (Figure 3A), ocular change was recorded by comparing the value before FD and after FD in both experimental eyes (OD) and the control eyes (OS) as previously reported [23] (ocular change = value after FD minus value before FD). Changes of different ocular parts, including ACD, VCD, RT, and AL, were not significant in wild-type rats but were significant in Timp4-deficient rats (Paired T-test, change in ACD: Timp4^+/+, p = 0.412; Timp4^+/−, p = 0.644; Timp4^−/−, p = 0.007 < 0.01; change in VCD: Timp4^+/+, p = 0.310; Timp4^+/−, p = 0.654; Timp4^−/−, p = 0.008 < 0.01; change in RT: Timp4^+/+, p = 0.901; Timp4^+/−, p = 0.194; Timp4^−/−, p = 0.027 < 0.05; change in AL: Timp4^+/+, p = 0.680; Timp4^+/−, p = 5.29 × 10⁻⁵ < 0.001; Timp4^−/−, p = 0.521) (Figure 3B). After FD modeling, FD eyes of gene defect rats showed a significant reduction of bipolar axon length in the central and mid-peripheral retina compared to non-FD eyes, while similar differences were not observed in wild-type rats (Paired T-test, central: Timp4^+/+ p = 0.1129, Timp4^+/− p = 0.0398 < 0.05, Timp4^−/− p = 0.392; mid-peripheral: Timp4^+/+ p = 0.1674, Timp4^+/− p = 0.0087 < 0.01, Timp4^−/− p = 0.0071 < 0.01; peripheral: Timp4^+/+ p = 0.1065, Timp4^+/− p = 0.6741, Timp4^−/− p = 0.1219) (Figure 3C,D). Detailed images are provided in Figure A4.

2.6. Content and Morphology Change in the Collagen of Timp4 Defect Rats

Since collagen is involved in myopia progression and is the target factor of TIMP4, we further evaluated the ocular tissues’ collagen to explore whether TIMP4 is related to a reduction in collagen content. In the Timp4-knockdown rats, the collagen contents of the sclera, cornea, and retina were significantly lower than those in wild-type rats (sclera: Timp4^+/+ = 2719.9 ± 535.7 μg/mL, Timp4^+/− = 2112.9 ± 167.4 μg/mL, Timp4^−/− = 1941.2 ± 190.0 μg/mL, p = 0.047 < 0.05; cornea: Timp4^+/+ = 854.2 ± 156.4 μg/mL, Timp4^+/− = 676.3 ± 148.5 μg/mL, Timp4^−/− = 522.4 ± 101.4 μg/mL, p = 0.031 < 0.05; retina: Timp4^+/+ = 62.8 ± 21.5 μg/mL, Timp4^+/− = 33.5 ± 5.5 μg/mL, Timp4^−/− = 32.6 ± 5.8 μg/mL, p = 0.025 < 0.05) (Figure 4A,

Table A3. Total collagen concentration and ocular tissue weight of Timp4-deficient and wild-type rats at 6 weeks old.

	Timp4+/+		Timp4+/−		Timp4−/−
	Total Collagen Concentration (μg/mL)	Tissue Weight (g)	Total Collagen Concentration (μg/mL)	Tissue Weight (g)	Total Collagen Concentration (μg/mL)	Tissue Weight (g)
cornea	854.2 ± 156.4	0.005 ± 0.001	676.3 ± 148.5	0.006 ± 0.002	522.4 ± 101.4	0.005 ± 0.001
lens	38.0 ± 5.4	0.022 ± 0.005	39.8 ± 3.1	0.023 ± 0.004	32.1 ± 3.2	0.019 ± 0.003
vitreous	80.4 ± 27.1	0.004 ± 0.001	68.5 ± 26.3	0.004 ± 0.001	51.0 ± 7.8	0.005 ± 0.002
retina	62.8 ± 21.5	0.013 ± 0.003	33.5 ± 5.5	0.010 ± 0.002	32.6 ± 5.8	0.010 ± 0.001
sclera	2719.9 ± 535.7	0.013 ± 0.002	2112.9 ± 167.4	0.015 ± 0.004	1941.2 ± 190.9	0.013 ± 0.002

Post hoc comparison: sclera: F = 4.899, η² = 0.583, p = 0.047 < 0.05, Timp4^+/+ vs. Timp4^+/− p = 0.044 < 0.05, Timp4^+/+ vs. Timp4^−/− p = 0.021 < 0.05;cornea: F = 5.219, η² = 0.537, p = 0.031 < 0.05, Timp4^+/+ vs. Timp4^+/− p = 0.097, Timp4^+/+ vs. Timp4^−/− p = 0.011 < 0.05;retina: F = 5.746, η² = 0.561, p = 0.025 < 0.05, Timp4^+/+ vs. Timp4^+/− p = 0.017 < 0.05, Timp4^+/+ vs. Timp4^−/− p = 0.022 < 0.05. Number of animals: Timp4^+/+ = 5, Timp4^+/− = 4, Timp4^−/− = 3. The results are presented as mean ± SD. Total collagen content was quantified based on hydroxyproline residues obtained from acid hydrolysis. ug/mL is the concentration unit of every 50ug dry-weight tissue dissolved in 200 μL hydrochloric acid.

). No significant differences were found in the lens and vitreous body. The ultrastructural morphology of collagen fiber of the sclera area surrounding the optic nerve was observed by using electron microscopy, and the results showed that in wild-type rats, the collagen fiber bundles had smooth edges, relatively uniform thickness, and an average diameter of approximately 114.5 μm, with smaller gaps and a tight arrangement between adjacent fibers. The sclera of Timp4-deficient rats exhibited collagen fibers with diameters of about 82.8 μm, varying sizes, unsmooth edges, and more significant gaps between adjacent collagen fibers. The statistical analysis revealed that the sclera collagen fiber diameter was significantly reduced in the Timp4-gene editing rats compared to wild-type rats (Timp4^+/+ = 114.5 ± 29.22, Timp4^+/− = 103.4 ± 20.21, Timp4^−/− = 82.77 ± 26.68, p < 0.0001) (Figure 4B,

Table A4. Collagen bundle diameter (nm) of the sclera of Timp4 deficient and wild-type rats at 6 weeks old.

	Timp4+/+	Timp4+/−	Timp4−/−
Average	114.5 ± 29.22	103.4 ± 20.21	82.77 ± 26.68
First quartile	94.75	90.52	63.45
Median	113.1	103.7	84.7
Third quartile	135.0	117.3	102.3

Post hoc comparison: F = 230, η² = 0.1959, p < 0.0001, Timp4^+/+ vs. Timp4^+/− p < 0.0001, Timp4^+/+ vs. Timp4^−/− p < 0.0001. Number of animals: Timp4^+/+ = 4, Timp4^+/− = 4, Timp4^−/− = 4. The results are presented as mean ± SD.

). The alterations in collagen fiber structure in the Timp4-deficient rats may have implications for the biomechanical properties of the sclera, such as its tensile strength, elasticity, and resistance to deformation, which is closely related to myopia formation.

2.7. Morphology Changes in the Retina of Timp4^−/− Rats

The result of the HE staining revealed that the thickness of the retina decreased with increasing age across different genotypes (Figure 5A). The thickness of the central retina of Timp4^+/− and Timp4^−/− rats was significantly decreased compared to that of the Timp4^+/+ rats from 3 w to 9 w (Timp4^−/− < Timp4^+/− < Timp4^+/+) (Figure 5B,

Table A5. Retina thickness (μm) of Timp4 deficient and wild-type rats from HE staining.

		Central		Mid-Peripheral		Peripheral
		Mean	SD	Mean	SD	Mean	SD
3W	Timp4+/+	208.22	3.76	184.83	15.67	141.55	10.2
	Timp4+/−	187.44	7.97	171.84	12.25	135.22	7.2
	Timp4−/−	175.87	11.31	172.65	28.49	138.5	10.37
	p	3.41 × 10−6		0.349		0.359
6W	Timp4+/+	166.67	12.29	143.53	17.65	111.8	16.18
	Timp4+/−	153.95	13.73	128.06	20.86	103.61	15.73
	Timp4−/−	137.65	13.53	133.2	15.42	101.85	13.86
	p	0.006		0.296		0.481
9W	Timp4+/+	149.47	8.8	122.23	2.4	100.33	7.46
	Timp4+/−	133.23	2.73	138.03	4.91	103.95	16.01
	Timp4−/−	129.07	6.1	114.13	5.75	97.1	3.03
	p	4.27 × 10−4		4.22 × 10−5		0.591

Post Hoc comparison results: 3W central retina thickness: F = 24.324, η² = 0.698, p < 0.001, Timp4^+/+ vs. Timp4^+/− p < 0.001, Timp4^+/+ vs. Timp4^−/− p = 0.002 < 0.01; 6W central retina thickness: F = 7.196, η² = 0.489, p = 0.006 < 0.01, Timp4^+/+ vs. Timp4^+/− p = 0.197, Timp4^+/+ vs. Timp4^−/− p = 0.005 < 0.001; 9W central retina thickness: F = 18.596, η² = 0.788, p < 0.001, Timp4^+/+ vs. Timp4^+/− p = 0.220, Timp4^+/+ vs. Timp4^−/− p = 0.02 < 0.05; 9W mid-peripheral retina thickness: F = 18.596, η² = 0.867, p < 0.001, Timp4^+/+ vs. Timp4^+/− p = 0.002 < 0.01, Timp4^+/+ vs. Timp4^−/− p < 0.001. Number of animals: 3W, Timp4^+/+ = 3, Timp4^+/− = 6, Timp4^−/− = 4; 6W, Timp4^+/+ = 9, Timp4^+/− = 10, Timp4^−/− = 6; 9W, Timp4^+/+ = 3, Timp4^{+/ −}= 3, Timp4^−/− = 3; SD, standard deviation.

), consistent with the SD-OCT results. At 3W and 6W, the central retina thickness is significantly thinner in the Timp4-deficient rats than that in wild-type rats (3W central retina thickness: Timp4^+/+ = 208.22 ± 3.76 μm, Timp4^+/− = 187.44 ± 7.97 μm, Timp4^−/− = 175.87 ± 11.31 μm, p = 3.41 × 10⁻⁶; 6W central retina thickness: Timp4^+/+ = 166.67 ± 12.29 μm, Timp4^+/− = 153.95 ± 13.73 μm, Timp4^−/− = 137.65 ± 13.53 μm, p = 0.006 < 0.01). At 9W, the central and mid-peripheral retina is thinner in Timp4-deficient rats than that in the wild-type (9W central retina thickness: Timp4^+/+ = 149.47 ± 8.8 μm, Timp4^+/− = 133.23 ± 2.73 μm, Timp4^−/− = 129.07 ± 6.1 μm, p = 4.27 × 10⁻⁴; 9W mid-peripheral retina thickness: Timp4^+/+ = 122.23 ± 2.4 μm, Timp4^+/− = 138.03 ± 4.91 μm, Timp4^−/− = 114.13±5.75 μm, p = 4.22 × 10⁻⁵). The differences in retinal thickness between different groups were mainly observed in the bipolar cells and photoreceptor cell layer of the retina, specifically in the IPL to INL (

Table A6. The thickness of central retina layers (μm) of Timp4 deficient and wild-type rats from HE staining.

		NFL + GCL		IPL + INL		OPL + ONL + IS/OS		RPE
		Average	SD	Average	SD	Average	SD	Average	SD
3W	Timp4+/+	36.42	7.26	85.00	7.81	89.67	4.55	6.67	0.81
	Timp4+/−	28.66	8.69	71.74	8.12	75.48	8.72	5.58	0.72
	Timp4−/−	27.58	4.14	76.93	3.61	73.72	7.31	6.68	1.28
	p	0.0920		0.005		0.002		0.026
6W	Timp4+/+	26.32	2.54	69.00	4.63	66.48	8.35	4.87	0.67
	Timp4+/−	25.27	4.26	51.93	16.09	50.77	16.51	3.96	1.32
	Timp4−/−	25.52	1.28	53.00	2.91	52.90	3.95	5.03	0.98
	p	0.8220		0.021		0.059		0.138
9W	Timp4+/+	37.53	2.32	57.20	2.79	52.27	1.39	3.73	1.00
	Timp4+/−	35.90	7.55	50.58	4.80	44.17	2.55	4.80	0.69
	Timp4−/−	26.13	0.67	44.57	2.10	40.63	0.51	3.50	0.46
	p	0.0668		0.011		0.0002		0.057

Post Hoc comparison results: 3W IPL+INL: F = 6.765, η² = 0.392, p = 0.005 < 0.01, Timp4^+/+ vs. Timp4^+/− p = 0.004 < 0.01, Timp4^+/+ vs. Timp4^−/− p = 0.2; 3W OPL+ONL+IS/OS: F = 8.643, η² = 0.452, p = 0.002 < 0.01, Timp4^+/+ vs. Timp4^+/− p = 0.004 < 0.01, Timp4^+/+ vs. Timp4^−/− p = 0.005 < 0.01; 3W RPE: F = 4.386, η² = 0.295, p = 0.026 < 0.05, Timp4^+/+ vs. Timp4^+/− p = 0.026 < 0.05, Timp4^+/+ vs. Timp4^−/− p = 0.975; 6W IPL+INL: F = 4.826, η² = 0.349, p = 0.021 < 0.05, Timp4^+/+ vs. Timp4^+/− p = 0.04 < 0.05, Timp4^+/+ vs. Timp4^−/− p < 0.001; 9W IPL+INL: F = 7.738, η² = 0.632, p = 0.011 < 0.05, Timp4^+/+ vs. Timp4^+/− p = 0.124, Timp4^+/+ vs. Timp4^−/− p = 0.01 < 0.05; 9W OPL + ONL + IS/OS: F = 26.601, η² = 0.855, p < 0.001, Timp4^+/+ vs. Timp4^+/− p = 0.002 < 0.01, Timp4^+/+ vs. Timp4^−/− p = 0.006 < 0.01. Number of animals: 3W, Timp4^+/+ = 3, Timp4^+/− = 6, Timp4^−/− = 4; 6W, Timp4^+/+ = 9, Timp4^+/− = 10, Timp4^−/− = 6; 9W, Timp4^+/+ = 3, Timp4^+/− = 3, Timp4^−/− = 3. Abbreviation: SD, standard deviation; NFL, nerve fiber layer; GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nucleus layer; OPL, outer plexiform layer; ONL, outer nucleus layer; IS/OS, inner segment/outer segment; RPE, retina pigment epithelial. One-way ANOVA was used for statistical analysis.

). Thus, we further observed the changes in ON bipolar cells whose axon is in IPL-INL. IF analysis showed that PKC-α area was shorter in the Timp4 defect rats in the central and mid-peripheral retina (central: Timp4^+/+ = 64.91 ± 4.58 μm, Timp4^+/− = 54.06 ± 8.62 μm, Timp4^−/− = 49.06 ± 2.26 μm, p = 0.0155 < 0.05; mid-peripheral: Timp4^+/+ = 76.01 ± 5.62 μm, Timp4^+/− = 60.25 ± 0.91 μm, Timp4^−/− = 56.90 ± 11.48 μm, p = 0.041 < 0.05, peripheral: Timp4^+/+ = 50.34 ± 3.81 μm, Timp4^+/− = 50.41 ± 1.12 μm, Timp4^−/− = 49.09 ± 7.91 μm, p = 0.94), which indicated that the gene defect rats had shorter bipolar cells axons in central and mid-peripheral regions of the retina compared to wild-type rats (Figure 6). The decrease of b-wave revealed by ERG was further corroborated by structural changes observed in the IF and HE staining.

3. Discussion

This study identified two TIMP4 LoF variants of the TIMP4 in seven of 928 probands with early-onset high myopia, and the defect of the TIMP4 was significantly enriched in eoHM compared to 5469 non-myopia controls (p = 3.7 × 10⁻⁵) and the general population (p = 2.78 × 10⁻⁹). We further confirmed the possible role of the TIMP4 by gene defect rat model and found that several changes in ocular structure and function were observed in the Timp4 defect rats in a dose-dependent manner, including the thinner retina, reduced b-wave amplitude, shortened axon length of bipolar cells, smaller sclera collagen fibers, and decreased collagen content in the sclera, retina, and cornea. Extra pressure induced by FD modeling would amplify these structure and function changes. These findings suggested that a defect of the TIMP4 is associated with high myopia and participates in rat ocular development in a dose-dependent manner.

The TIMP-4 belongs to the TIMP family, a group of structurally similar proteins that inhibit the activity of metalloproteinases. Unlike other family members (TIMP1, TIMP2, TIMP3), the role of TIMP-4 in ocular tissue remains limited [24]. Previous studies revealed that the protein structure of the TIMP-4 can be inferred from the TIMP-2, sharing 51% homology at the amino acid level, including the location of 12 cysteine residues and a leader sequence comprising 29 amino acids, which may be the target sequence cleaved to generate the mature protein [16,25]. The TIMP-4 efficiently inhibits the activity of MMP-2 via its N-terminus [25,26], and MMP2 upregulation was accompanied by significant myopia [18]. Both the TIMP4 LoF variants detected in our study are in the NTR domain, which is implicated in many protein functions such as axon guidance, regulation of the Wnt signaling pathway, and, notably, metalloproteinase activity [27,28]. Our findings linked these variants to high myopia based on the following reasons: (1) the TIMP4 LoF variants are enriched in the high myopia cohort in our study; (2) the TIMP4 LoF variants were predicted to result in truncation changes, disrupting domain with important biological functions; (3) pathological changes resembling those in HM individuals were observed in the Timp4-knockdown rat model in a dose-dependent manner; (4) form deprivation could exacerbate the pathological changes in gene defect rats. Previous studies of extracellular matrix-related genes (the MMP and TIMP family) were likely driven by the known biological roles in tissue remodeling, possibly overlooking their molecular genetic analysis on patients with high myopia. While the TIMPs has been studied through WES in other contexts, such as schizophrenia [29], no systematical studies of ECM-associated high myopia were performed based on WES data before. This might explain why the LoF variants of the TIMP4 were not previously associated with high myopia in large cohorts. Currently, there is no entry of human disease caused by TIMP4 mutation listed in the OMIM database. To our knowledge, this study is the first to discover the role of TIMP4 in high myopia, providing fresh insights into the etiology of myopia.

During myopia development, axial length elongation is regulated by multiple mechanisms across the retina with a retina-to-sclera direction [30,31]. The ECM of the sclera was remodeled as the consequence of a growth-regulating cascade initialed from the retina. Although 14 high myopia genes were reported (

Table A7. Gene defect animal models of known causative genes of high myopia.

Gene	MYP	OMIM	Inheritance Patterns	Animal Models	Phenotypes Supporting High Myopia			Reference (PMID)
					Axial Length	Refractive Error	Others
SCO2	6	604272	AD	mouse	/	/	/	[69]
ZNF644	21	614159	AD	zebrafish	/	/	/	[70]
LRPAP1	23	104225	AR	mouse	/	/	/	[23,71]
SLC39A5	24	608730	AD	mouse	/	/	/	[72]
P4HA2	25	600608	AD	mouse	/	/	/	[73]
ARR3	26	301770	XL	mouse	/	/	/	[74,75]
CPSF1	27	606027	AD	zebrafish	/	/	small eye size	[49]
LOXL3	28	607163	AR	mouse	/	/	/	[54]
CTSH	/	116820	AR	mouse	/	/	abnormally gross eyeball	[22]
NDUFAF7	/	615898	AD	mouse	/	/	/	[76]
BSG	/	109480	AD	mouse	AL elongation	/	/	[32]
OPN1LW	/	300822	XL	mouse	/	/	/	[77]
XYLT	/	608125	AD	mouse	/	/	/	[78]
DZIP1	/	608671	AD	mouse	/	/	/	[79]

AD, autosomal dominant inheritance; AR, recessive dominant inheritance; XL, X-linked dominant inheritance; AL, axial length; /, no direct evidence for high myopia (axial elongation, refractive error, or other related ocular phenotypes).

), only CTSH and BSG, reappeared with longer axial length in gene deficiency animals [32] without further mechanism explorations. In this study, we observed that the quality and quantity of collagen in the sclera of Timp4^−/− rats were decreased simultaneously, which might address the elongation of vitreous and axial length. Our finding supports the view that ECM remodeling contributes to myopia development by changing the biomechanical properties of the sclera. Besides remodeling the sclera, Timp4 also participated in the abnormality of bipolar cells in the retina. TIMP4 protein expressed from OPL to IPL of both human and rat where synapses of the bipolar cell were involved and the primary defect in the retina of TIMP4^−/− rat were also focused on bipolar cell including reduced amplitude of the b-wave as well as shorter bipolar cells axons in eyeball development and under FD. Total axon length (from the axon hillock under the soma to the axon terminal tip) and IPL axon length (measured by the distance from the INL-IPL border to the axon terminal) [33] were often used to explore the functional impact of the neuron in literature [34,35]. Retinal bipolar cells act as the “projection neurons” in the visual circuit of vertebrates and convey essential visual information via synaptic from photoreceptor to ON bipolar cells mediating by the mGluR6 receptor [36]. In our study, we measured PKCα area to find that Timp4 defect rats have shorter axon length and reduced b-wave amplitude compared to wild-type rats, which indicates both functional and structural damage of bipolar cells in Timp4 defect rats. The abnormalities of bipolar cells or neural channels in which they revolved have qualitatively different effects on ocular growth and myopia development [37,38,39,40]. ECM plays an essential role in retinal development by regulating cell shape, proliferation, differentiation, migration, and morphogenesis [41,42] and has been used to facilitate the differentiation of human neural stem cells into neurons and to increase axon length [43]. In Figure 4B, we present the frequency distribution of collagen fiber diameters, where a notable error bar in Timp4^−/− rats suggests heterogeneity and dispersed distribution of fiber sizes, which aligns with the visual trends depicted in Figure 4A. Our study suggests a model wherein Timp4 deficiency may lead to a change of ECM in the retina, which changes the function and structure of bipolar cells, resulting in high myopia.

Although rats are not a classical model for myopia study, we chose rats for this research considering the following reasons: (1) gene-editing rats are easy to obtain; (2) rats have eyes twice the size of mice and are convenient for performing FD modeling and collect ocular parameters; (3) rat as an animal model for FDM was successfully established in previous literature [44,45]. In this study, the purpose of FD was to discover whether or not the ocular structure and function changes caused by the Timp4 defect in rats would be influenced by myopia-related pressure. After FD modeling, retina thickness and more extended vitreous chamber depth were observed, which was not exhibited in untreated gene deficiency rats. The model exhibited phenotypic features such as thinner retina thickness, shorter bipolar cell length, deeper VCD, and longer AL compared to the control eyes. The significant differences between FD and control eyes were not observed in wild-type rats. Timp4-defect animal models have mainly been studied in the cardiovascular system (

Table A8. Timp4 deficient animal model in literatures.

Organ/Tissue	Phenotype	Reference (PMID)
Heart	Timp4−/− mice were born and developed healthy, but their mortality increased significantly by TIMP2 expression increasing after cardiac pressure overload-induced by aortic banding, which could be rescued by exogenous MMPs inhibitors	[46]
Heart	after ischemia-reperfusion in vivo, more severe systolic and diastolic dysfunction, myocardial hypertrophy, and fibrosis appeared in Timp4−/− mice than in wild-type	[80]
Heart	left ventricular hypertrophy appeared, ventricular fibrosis increased, and long-term survival decreased in Timp4−/− mice under left ventricular pressure overload	[81]
Vascular	no difference in blood pressure hypertension induced by angiotensin II between Timp4−/− mice and wild-type mice	[82]
Fat	Timp4−/− mice showed higher food intake than the wild-type mice under a high-fat diet but less fat gain	[83]

). Although TIMP4 is known to participate in physiological and pathological processes such as cell proliferation, differentiation, apoptosis, and extracellular matrix degradation, no studies have reported ocular changes associated with this gene in animal models or human diseases.

Interestingly, researchers discovered that in Timp4-defect animal models, its loss moderately compromised cardiac functions with aging, and gene-defect animals exhibited a higher susceptibility to mortality in induced myocardial infarction [46]. This pattern was also observed in our rat FDM models, as evidenced by retinal thinning during development and changes after FD modeling. To some extent, such a “coincidence” reveals this gene’s interactive nature with external environmental stress.

The association of TIMP4 deficiency and myopia was robustly supported by findings based on myopia patients and the gene editing rat model, but our study had several limitations. First, although the TIMP4 LoF variant enriched the HM population in our study, it was also detected in non-high myopia individuals and GenomAD. It remains to be seen whether incomplete penetrance happened in the pathogenic mechanisms of TIMP4 on eoHM based on more individuals with TIMP4 LoF. Secondly, statistically significant differences in retinal thickness and axial length were concentrated in the adolescent period of Timp4 deficiency rats. It is unknown whether the differences are the causal link to the special role of Timp4 in the eye development stage or to the limited animal number at different time points after 15 weeks in this study (Figure 2E). Thirdly, both functional and structural damage was observed in bipolar cells from Timp4 deficiency rats. However, it is unknown which subtypes of ON and OFF bipolar cells are involved, and the mechanism of how ECM affects the function of bipolar cells remains unknown. Further studies are necessary to answer these questions.

4. Materials and Methods

4.1. Patients and Clinical Information

This study was approved by the institutional review board of the Zhongshan Ophthalmic Centre. Participants with various ocular conditions, including early-onset high myopia, were recruited from the Pediatric and Genetic Clinic, Zhongshan Ophthalmic Centre, Guangzhou, China. Clinical information was gathered according to the established standard operation protocols [47]. The individuals with eoHM were selected based on criteria of high myopia that occurs before school age and is considered to be caused mainly by genetic variations with minimal environmental involvement [32]. The control groups of subjects with other ocular disorders and healthy controls were recruited as previous studies [48]. Before collecting clinical data and peripheral blood samples, written informed consent was obtained from the participants, their guardians, or relatives. The informed consent was obtained adhering to the tenets of the Declaration of Helsinki. Human retina sections were obtained from Zhongshan Ophthalmic Center eye bank under the Medical Ethics Committee of Zhongshan Ophthalmic Center, Sun Yat-Sen University (Approval number: 2011KYNL012).

4.2. Candidate Gene Detection by WES and Multistep Bioanalysis

Potential pathogenic variants (PPVs) in TIMPs and MMPs were selected from an in-house whole exome sequencing (WES) dataset comprising 928 probands with high myopia and 5469 probands with other forms of genetic eye diseases as our previous process [49]. The WES procedures and multistep bioinformatic analysis of detected variants were detailed in our previous studies [47]. The frequency of PPV in the HM cohort was calculated and compared with that of the non-HM genetic eye disease cohort and the general population database (gnomAD v2.1.1) to confirm the candidate gene where LoF variants were rare and intolerant. As described in our previous study, all PPVs were confirmed by Sanger sequencing in probands and available family members [47]. The PPVs were described according to the nomenclature for sequence variations (Human Genome Variation Society nomenclature v20.05, http://varnomen.hgvs.org/, accessed on 1 May 2022) and confirmed by Mutalyzer [50] (https://www.mutalyzer.nl/, accessed on 10 March 2023). Pathological classifications were performed according to the guidelines of the American College of Medical Genetics and Genomics (ACMG) [51]. Prediction of variants was conducted using ENTPRISE-X (http://cssb2.biology.gatech.edu/entprise-x/, accessed on 1 May 2022) and SIFT (https://sift.bii.a-star.edu.sg/www/SIFT_indels2.html, accessed on 1 May 2022). In silico prediction of protein structure was conducted using AlphaFold (https://alphafold.ebi.ac.uk/, accessed on 1 May 2022), DUET (http://biosig.unimelb.edu.au/duet/stability, accessed on 1 May 2022) and MaestroWeb (https://pbwww.che.sbg.ac.at/maestro/web, accessed on 1 May 2022).

4.3. Knockdown of Timp4 Gene in Rats

The experiments were approved by the Animal Experimental Ethical Inspection to Ethics Committee of Zhongshan Ophthalmic Center, Sun Yat-Sen University (Approval number: 2020-030, 2020-036, 2020-172). All procedures were strictly carried out per the ARVO statement for the Use of Animals in Ophthalmic and Vision Research and the guidance of the IACUC (Institutional Animal Care and Use Committee). The Timp4 gene (GenBank accession number: NM_001109393.1; Ensembl: ENSRNOG00000007955) knockdown Sprague-Dawley rat (SD rat) model was commercially constructed by Cyagen Bioscience Inc. (Guangzhou, China) using CRISPR/Cas-mediated genome engineering. The construction and validation details are shown in Figure 2A–C, Cas9 mRNA and gRNA generated by in vitro transcription were injected into fertilized eggs for knock-out rat productions. The sequences of gRNA were as follows: gRNA1: 5′-TAGGGCCAATCCCGCCCCCATGG-3′; gRNA2: 5′-CTTCCATAGCGCACATCGCAAGG-3′. The primer sequences of genotyping were as follows: Rat Timp4-F: 5′-CCCTGAGTAGCTTGTCTCATTCCCAG-3′, Rat Timp4-R: 5′-AGAAACCTAACGGAACTGAAGAGAAGTCT-3′, Sequencing primer: 5′-GGTCCGTCCTCCTCTGTCACTGT-3′, and the reaction method of polymerase chain reaction were listed in

Table A9. PCR system applied in our study.

	Material	Volume/Well (μL)	Final Concentration
Mix1	2x PCR buffer for KOD FX	25	1×
	2 mM dNTPs	10	0.4 mM each
	Primer-F	1.5	/
	Primer-R	1.5	/
	Template DNA	2	200 ng/50 μL
	KOD FX (1.0 U/μL)	1	1.0 U/50 μL
	Ultra-pure water	9	/
Mix2	2x PCR buffer for KOD FX	25	1×
	2mM dNTPs	10	0.4 mM each
	Primer-F	0.75	/
	Primer-R	1.5	/
	Primer-WT-F	0.75	/
	Template DNA	2	200 ng/50 μL
	KOD FX (1.0 U/μL)	1	1.0 U/50 μL
	Ultra-pure water	9	/

The reagents for the PCR system were obtained from TOYOBO (No. KFX-101, Japan).

and

Table A10. PCR amplification program.

Temperature	Time	Cycle
94 °C	2 min
98 °C	10 s	35x
60 °C	30 s
68 °C	1 min
68 °C	7 min

. The positive founders (F0) genotyped by PCR and DNA sequencing analysis were then bred to the next generation (F1). We used littermates as controls to ensure genetic and environmental consistency in our experimental design. Three genotypes were included in this study: wild-type (Timp4^+/+), heterozygous (Timp4^+/−), and homozygous (Timp4^−/−). All rats were transported under the standard procedure of the Ophthalmic Animal Laboratory in special containments equipped with filtration systems to maintain specific pathogen-free conditions. Upon arrival at the laboratory, the rats were subjected to a seven-day acclimatization period before being delivered to the animal facility. All animals were housed in the Ophthalmic Animal Laboratory, Zhongshan Ophthalmic Center, Sun Yat-sen University. Controlled photoperiods (12 h of light and 12 h of darkness) and sufficient water and food were provided in a temperature-controlled room in a specific pathogen-free environment. To ensure consistency, three-week-old rats of the same sex (male) and the same body weight (50 ± 10 g) were utilized. Abnormally growing rats were excluded from the study (body weight below or over -2 standard deviations of the group,

Table A11. Body weight of experimental rats in this study.

	Age (Weeks)	3	5	6	9	10	12	15	18	20	22	52
Timp4+/+	Mean	49.3	124.0	167.0	252.2	271.5	366.5	370.3	421.5	423.5	440.0	604.7
	SD	8.1	13.7	9.9	10.1	16.3	14.8	8.7	9.2	29.0	5.7	13.7
Timp4+/−	Mean	52.5	129.7	165.6	259.3	275.4	365.8	382.3	435.5	447.0	441.5	623.0
	SD	6.4	10.8	10.4	20.8	12.6	13.3	16.7	16.7	15.3	24.3	28.3
Timp4−/−	Mean	53.2	122.0	160.7	247.4	303.3	385.0	399.6	443.3	463.5	444.3	608.3
	SD	7.9	25.5	26.1	18.7	10.6	14.3	7.7	13.3	7.1	17.5	11.6

). Western blot was conducted as previously reported [49]. The primary antibodies used in the western blot were TIMP-4 (PA5-92959, 1:1000, Invitrogen, Waltham, MA, USA) and Beta-actin (AC004, 1:1000, Abclonal, Wuhan, China). Peroxidase-conjugated secondary antibody antibodies were anti-rabbit IgG (4413, 1:2000, Cell Signaling Technology, Danvers, MA, USA) and anti-mouse IgG antibodies (4408, 1:2000, Cell Signaling Technology, USA) diluted by 1:2000.

4.4. Determination of Ocular Parameters

Axial length (AL), corneal thickness (CT), anterior chamber depth (ACD), vitreous chamber distance (VCD), and retinal thickness (RT) from three weeks to 52 weeks of age were measured by spectral-domain optical coherence tomography (SD-OCT, Envisu R4310, Leica, Germany) as previously described [52,53] (Figure A2B). Rats were administered 0.5% tropicamide and 0.5% phenylephrine eye drops (Santen, Japan) to induce mydriasis. After an intraperitoneal injection of Zoletil 50 (Virbac, Carros, France) at 75 mg/kg dosage for anesthesia. Rats were immobilized on a platform for scanning. The configuration detail was as follows: a scanning circle with a radius of 2.0 cm (axial resolution: 2.6 μm, central wavelength: 840 nm, 24 frames × 5 B scans, each B scan averaged five times) was centered on the macula. Accurate shots were obtained when a cross flash (signal targeting optic disc) appeared on the screen to ensure that the longest eye axis was captured each time. The images obtained from the scans were analyzed using InVivoVue software 2.4 or DIVER 3.4, and data around the optic papilla at 360° was measured to calculate the average value of CT, ACD, VCD, and RT. The lens thickness (LT) measurement was limited to rats aged between three and 15 weeks, as the distance from the cornea to the lens of older rats would be beyond the maximum range of R4310 (5.2 mm). The AL was calculated by summing the values of CT, ACD, LT, VCD, and RT.

4.5. Determination of Ocular Components under Extra Pressure from Form Deprivation

Several genes were reported to be associated with HM, but classical myopia characteristics such as refractive error change and prolonged axial length were not observed in gene mutant animal models [22,54]. The form deprivation myopia (FDM) was a classical approach to induce a high myopia model and it was successfully established in different animal species, including monkeys, mice, chickens, tree shrews, guinea pigs, and rats [44,55,56,57,58,59,60]. To amplify the potential impact of Timp4 knockdown on ocular development, we established a form deprivation model on the three-week-old rats of three genotypes, referring to the previously reported form deprivation models of chicken and rats [45,61] (Figure A2B). Male rats were randomly grouped and anesthetized as described above. After ocular parameters measurement, a form deprivation (FD) translucent diffuser made of soft polyvinyl chloride (outside diameter: 20 mm; inside diameter: 10 mm) was fitted to the right eye (OD) of each rat by matching Velcro rings. The left eyes (OS) were untreated and served as control eyes. The diffusers utilized in this study were fabricated from a soft polyvinyl chloride material, designed to induce a blurring effect without causing harm to the ocular or facial tissues. Following anesthesia, the diffusers were gently affixed on the surface of the orbital skin by super glue (Pattex, Henkel, Düsseldorf, Germany). These lenses diffused light, resulting in the retina’s inability to form clear images, a condition hypothesized to contribute to the development of myopia. The diffusers were cleaned twice daily, and the rats whose diffusers fell off during the FDM modeling were excluded from the experimental group. After 72 h of the FDM modeling, the diffusers were removed, and the ocular components of both eyes were measured immediately using the SD-OCT.

4.6. Determination of Retina Function by Electroretinography

The retina’s function was evaluated using electroretinography (ERG) in experimental animals, as previously described [62]. Animals were divided into different groups as described above and tested at 3 and 6 weeks. Before the experiment, the rats were acclimated to complete darkness for at least eight hours, with access to sufficient food, water, and exercise space. The anesthetized rats were measured with a Roland electrophysiology instrument (RETI-port/scan 21, Roland Consult, Brandenburg an der Havel, Germany) at different scotopic levels (0.1, 3.0, and 10.0 log scotopic cd.s/m²) as well as oscillatory potentials. Each group consists of three to five animals.

4.7. Hematoxylin-Eosin (HE) Staining and Immunofluorescence (IF)

We conducted the HE and IF staining to validate the results from the SD-OCT and ERG and investigate changes occurring beyond the detection range. Tissues were harvested after execution by cardiac perfusion, and the eyes were enucleated and fixed with a fixative solution (FAS, G1109, Servicebio, Wuhan, China) for 24 h before preparing paraffin slices. The IF and HE staining were performed on sections prepared from paraffin-embedded tissue. During the embedding process, we orient the optic nerve parallel to the plane of sectioning to ensure maximal representation of the ocular axis on the sections. Tissues were sectioned by 4 μm. Sections containing the optic nerve were selected for subsequent staining procedures. Standard cardiac perfusion, HE, and IF staining procedures were conducted as previously described [49,63]. Due to the decreased b-wave amplitude observed in the functional experiment, we assessed potential structural changes of bipolar cells using antibodies against PKCα for the IF staining [64]. The axon length of a bipolar cell is measured from under the soma to the axon terminal tip [33]. The information on primary antibodies is provided in

Table A12. Information of antibody or fluorescent dye used in this study.

Antibody	Target	Immunogen	Host Species	Supplier	RRID/SKU	Concentration
PCKα	Protein Kinase C-α	amino acids residue 652–672	Mouse	Santa Cruz, Dallas, USA	AB_628142	1:1000
TIMP4	Tissue inhibitor of matrix metalloproteinase4	Amino acids residue 30–224	Rabbit	Invitrogen, USA	PA5-92959	1:100
DAPI	Nucleic acid	/	/	Roche, Basel, Swiss	10236276001	1:3000

Abbreviation: RRID, Research Resource Identifier; SKU: Stock Keeping Unit.

. The immune reaction complexes were detected using secondary antibodies (goat anti-mouse IgG and goat anti-rabbit IgG) Alexa Fluor 488 (1:1000) diluted in goat serum. The sections were visualized using a confocal microscope (Zeiss LSM 980 with Airyscan 2, Zeiss, Oberkochen, Germany) following standard procedure. Three parts (central, mid-peripheral, and peripheral) of the retina of each section were collected according to previous literature [65]. The average thickness of the whole retina, as well as the length of bipolar cells, were quantified, as illustrated in Figure 5. Three to five cross-sectional slices were analyzed for each group.

4.8. Determination of Collagen Levels

Tissues including the cornea, lens, vitreous, retina, and sclera, of six-week-old rats of three different genotypes were carefully dissected and collected under a microscope after execution. The muscle and connective tissue attached to the cornea, lens, vitreous, retina, and sclera were separated with fine forceps as much as possible. The samples were homogenized in 6M hydrochloric acid and incubated at 95 °C for 20 h. After centrifugation (13,000× g, 10 min, 20 °C), hydroxyproline content in the supernatants was quantified using the Total Collagen Assay Kit (QZBtotcol1, QuickZyme, Biosciences, Rotterdam, The Netherlands) strictly following the standard protocol provided in the kit. For comparability between groups, the collagen content measured in each sample was normalized by the samples’ dry weight (collagen concentration/dry weight). The ocular tissue weight was provided in Table A3, which provides a standardized metric that facilitates accurate comparison of collagen content across different samples. Based on the consistency in processing methods for different groups, we adhere to the standard protocols of commercial assay kits and literature procedures [66] to quantify the total collagen content within tissues from rats of varying genotypes. Four animals were used in each group.

4.9. Determination of Collagen Ultrastructure in the Sclera

To investigate the morphology of collagen fibers, samples were obtained by cutting the sclera into a four-leaf clover shape, and approximately one mm³ was taken from each leaf surrounding the optic nerve within 1–3 min after execution and were fixed in fixative solution (G1102-100ML, Servicebio, Wuhan, China). The tissues were treated as previously described [67]. At least 200 collagen fibers were measured for diameter in each electron microscope image (Figure A2D). According to previous literature, the observation was performed using a transmission electron microscope (HITACHI, HT7800/HT7700) [68]. Three to five animals were used in each group.

4.10. Statistical Analysis

The allele number of the LoF TIMP4 variants in individuals with high myopia and the population database were compared by Fisher’s exact test. The association between the LoF TIMP4 variants and high myopia was calculated by odds ratio (OR), representing the effect size for Fisher’s exact test. The ocular parameters, amplitude of the ERG, thickness of the retina by HE, and height of bipolar cells were statistically analyzed using one-way ANOVA among three different genotypes of rats (Timp4^−/−, Timp4^+/−, and Timp4^+/+). Before performing statistics analyses, the normal distribution (Shapiro-Wilk Statistic, p > 0.05) and homogenous variance (Levene test) were confirmed. When conducting post hoc analyses on the results of the one-way ANOVA, the Bonferroni correction method is employed to adjust for multiple comparisons. In instances of heteroscedasticity, Tamhane’s T2 correction is utilized to account for unequal variances. η² represents the effect size for the one-way ANOVA. A paired T-test was used to compare the differences between the same rat before and after form deprivation. The detailed results of the analysis are provided in corresponding Appendix Figure legends and Tables. All the statistical results were calculated with SPSS 26.0 and plotted with GraphPad Prism 9.5.1. p < 0.05 is used as the basic statistical significance.

5. Conclusions

In summary, this study indicated that heterozygous LoF mutations in TIMP4 are associated with high myopia, and Timp4 deficiency disturbs rat ocular development through ECM remodeling by changing the biomechanical properties of the sclera and disturbing the structure and function of bipolar cells in the retina. Further research is warranted to confirm and expand upon these findings, deepening our understanding of the underlying myopia mechanism.