Integrative Transcriptome and Proteome Analyses Elucidate the Mechanism of Lens-Induced Myopia in Mice

Abstract

Purpose

The purpose of this study was to investigate the underlying molecular mechanism of lens-induced myopia (LIM) through transcriptome and proteome analyses with a modified mouse myopia model.

Methods

Four-week-old C57BL/6J mice were treated with a homemade newly designed –25 diopter (D) lens mounting by a 3D printing pen before right eyes for 4 weeks. Refraction (RE) and axial dimensions were measured every 2 weeks. Retinas were analyzed by RNA-sequencing and data-independent acquisition liquid chromatography tandem mass spectrometry. Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) annotation, and STRING databases were used to identify significantly affected pathways in transcriptomic and proteomic data sets. Western blot was used to detect the expression of specific proteins.

Results

The modified model was accessible and efficient. Mice displayed a significant myopic shift (approximately 8 D) following 4 weeks’ of lens treatment. Through transcriptomics and proteomics analysis, we elucidated 175 differently expressed genes (DEGs) and 646 differentially expressed proteins (DEPs) between binoculus. The transcriptomic and proteomic data showed a low correlation. Going over the mRNA protein matches, insulin like growth factor 2 mRNA binding protein 1 (Igf2bp1) was found to be a convincing biomarker of LIM, which was confirmed by Western blot. RNA-seq and proteome profiling confirmed that these two “omics” data sets complemented one another in KEGG pathways annovation. Among these, metabolic and human diseases pathways were considered to be correlated with the LIM forming process.

Conclusions

The newly constructed LIM model provides a useful tool for future myopia research. Combining transcriptomic and proteomic analysis may potentially brighten the prospects of novel therapeutic targets for patients with myopia.

Myopia is a common ocular disorder in humans, characterized by excessive axial length (AL) elongation, leading to various visual abnormalities.^1,2 Despite its increasing prevalence, the exact mechanism of myopia remains elusive. Ocular refractive development is affected by a complex “retina-to-choroid-to-scleral signaling cascade.”³ However, many questions remain unanswered, such as whether specific elements or pathways play a decisive role in the development of myopia, how factors interact with each other, and whether essential components are yet to be discovered.

To better understand the pathogenesis of myopia, form-deprived myopia (FDM)^4,5 and lens-induced myopia (LIM) are two common modeling methods.^6,7 Both models have been successfully established on multiple animals, including mice,^8,9 chicks,^10,11 guinea pigs,^12,13 and tree shrews.^14,15 Among all, mice were widely used in various fields of biomedical research with multiple advantages and could be the ideal model animals in ophthalmology as well.¹⁶ However, due to the difficulty of conducting lens induction on mice,¹⁶ more studies chose FDM model rather than LIM model.^9,17 Whereas previous studies have used commendable methods to construct an LIM model, such as gluing,¹⁸ stitching,¹⁹ and head-mounted spectacle frames,^20–22 there are still some limitations to be acknowledged. Therefore, the development of a refined LIM model holds promising potential for significant benefits.

Both proteomics and transcriptomics approaches, widely used in previous studies, are useful tools for discovering molecular mechanisms of myopia.²³ Much of our understanding of myopia comes from a single type of omics. However, the limitations of the single type of omics method include one-sidedness and incoherency. To gain a more comprehensive view of potential pathologies, along with recent advances in the detection sensitivity and accuracy in RNA-seq and mass spectrometry-based proteomic techniques, we can incorporate the multi-omics approaches to achieve insights into the gene-protein mechanism of myopia formation. Particularly, due to the difficulty of establishing an effective mouse model of LIM, few studies were published about murine LIM models, let alone multi-omics analysis of LIM. To fill this gap, our present study integrated RNA-seq and proteomics data analysis of the original murine LIM model. Bioinformatics analysis was furthermore conducted to investigate the molecular metabolism mechanisms.

Materials and Methods

Mouse Breeding and Treatments

Mice were housed and treated following the guidelines of the ARVO Statement for the use of animals in ophthalmic and vision research, and approved by the animal care and ethics committee at Zhongshan Hospital, Fudan University, Shanghai, China (#2021-068). Before different treatments, mice with abnormal status were excluded, including severe anisometropia, cataracts, eye closure, abnormal mental state, et al. Fifteen 4-week-old C57BL/6J mice (referred to as p4w, p6w, and p8w denoted the corresponding time after birth, respectively) were treated with –25 diopters (D) minus the lens mounts¹⁹ in front of their right eyes for 4 weeks. We checked to see if the lenses have fallen out three times a day. We would attach the eyeglasses onto the mice right the first time. Mice with spectacle frame loss twice or more would be excluded.

Myopia Induction

As depicted in Figure 1, a modification was made to construct the LIM model, which has been patented by China National Intellectual Property Administration (ZL 2021 1 0156065. 5). Initially, we produced open collars and frames with diameters of approximately 1.5 cm using 3D printing technology and polylactic acid (SL-300, SUNLU, Guangdong, China). Then, the collars were placed around the anesthetic mice's necks (pentobarbital solution, 70 mg/kg i.p.) and the 3D materials were slightly warm and adjustable in size at this stage. Thereafter, they were sealed to ensure a secure fit, preventing any movement or displacement once they were in place. Accordingly, connecting lines were drawn between the collar and the spectacle frame. Finally, we shaped the eyewear to position the lenses correctly before the connecting lines cooled down and solidified. Throughout the experiment, the apparatus was regularly adjusted to adapt to the changing sizes and growth of the mice's bodies.

Figure 1.

Establishment of the modified LIM model and induction of myopia changes. (A) Equipment preparation included -25 D lenses, a 3D printing pen, and polylactic acid filament. A PVC plastic sheet was used for fur and skin protection. (B) A diagram illustrated the connection of components in the modified LIM model. (C) Frame and collar fabrication. (D) Linkage production. (E) An example of a complete lens mounting for mice. (F) Mice with 2 or 4 weeks of minus lens treatment displayed a myopic change tendency. (G) After 4 weeks of lens treatment, ALs of the right eyes were longer than those of the control eyes. (H) At p8w, the myopic eyes had deeper VCD than the control ones. (I) At p8w, the myopic eyes had less dopamine than the control; n = 10/group in F, G. and H; n = 4/group in I. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001.

Ocular Biometric Measurements

Refraction (RE) and axial dimensions, including vitreous chamber depth (VCD) and axial length (AL), were measured every 2 weeks. A custom-made eccentric infrared photorefractor²⁴ and a spectral-domain optical coherence tomography (SD-OCT) system (TEL210C1; Thorlabs Inc.) were used to do the measurements as reported.²⁵ Using the heated tip of the 3D printing pen, we could engage the junction point, leading to its disconnection and enabling us to remove the spectacle. Before SD-OCT measurement, the mice were anesthetized with pentobarbital solution (70 mg/kg i.p.). We adjusted the position of the mice to keep the measured eye aimed squarely at the optical scanning probe and scanned along the entire AL. The experiments were performed at least three times to get the average value. After completing all the measurements, we seamlessly used the 3D printing pen to reconnect the separated ends for reassembly.

High-Performance Liquid Chromatography-Electrochemical Detection

Animals were euthanized from 8:00 AM to 10:00 AM. High-performance liquid chromatography (HPLC) analysis was done as previously described.²⁶ Briefly, homogenate was added to the retinas (n = 4/group, 0.1 g tissue to 1 mL 0.2N perchloric acid). An Agilent 1200 HPLC apparatus (Agilent Technologies, Santa Clara, CA, USA) equipped with an electrochemical detector (Antec, Zoeterwoude, The Netherlands) and the Chem Station software (Agilent Technologies) were used to determine the dopamine level.

Western Blot

As described previously,²⁵ total protein extracts from retinas were prepared in RIPA lysis buffer (P0013B; Beyotime, China). Then, the BCA Protein Assay Kit (23235; ThermoFisher, USA) was used to measure the concentration. After being boiled with loading buffer (9173; Takara, Japan), the samples were separated on 10% Acrylamide/Bis-Tris gels and electroblotted onto polyvinylidene difluoride (PVDF) membranes. We used primary antibodies directly against IGF2BP1 (1:1000, abs135956; Absin, China) and GAPDH (1:5000, HRP-60004, Chicago, IL, USA). Protein bands were detected with a chemiluminescence kit (34094; ThermoFisher, USA). We quantified the protein bands with Image J software (Wayne Rasband, National Institutes of Health, USA).

Transcriptome Analysis

Bilateral retinas were harvested from mice after 4 weeks of myopia induction (n = 3/group). As previously described,^27,28 RNA extraction, transcriptome sequencing, data process, quality control, and expression analysis were done. Briefly, RNA was purified and concentrated, and the RNA Integrity Number (RIN) was determined by an Agilent Bioanalyzer 4150 system (Agilent Technologies, CA, USA). The ABclonal mRNA-seq Lib Prep Kit (ABclonal, China) was used for transcriptome sequencing. The library was sequenced on an Illumina Novaseq 6000 (or MGISEQ-T7), generating 150 bp paired-end reads. The data were obtained from the Illumina (or BGI) platform and used for bioinformatics analysis. Clean data were obtained from raw data of the fastq format using in-house perl scripts that removed unqualified reads, including reads containing the adapter sequences, poly-N reads (concealing base information), and low-quality reads (containing > 50% bases with a Phred quality score < 20). RNA-seq data were aligned to the Ensembl Mus musculus reference transcriptome (GRCh38, Release 102) by HiSat2 (version 2.1.0). FeatureCounts (http://subread.sourceforge.net/) was used to count the reads mapped to each gene. Gene expression levels were quantified using fragments per kilobase of mappable length and million counts (FPKM) for paired terminal RNA-seq. Differential expression analysis was performed using DESeq2 (http://bioconductor.org/packages/release/bioc/html/DESeq2.html) to acquire the volcano plot, which indicates the number and distribution of DEGs. Genes with |log₂FC| > 1 and adjusted P values < 0.05 were considered to have considerably differing expression levels.²⁹

Quantitative Real-Time PCR

As previously reported,²⁵ total RNA was extracted and reverse-transcribed into cDNA. The SYBR Green quantitative real-time PCR (qRT-PCR) kit (AG11701; Accurate Biotechnology [Hunan] Co., Ltd, China) and QuantStudio 5 real-tme PCR system (ThermoFisher, USA) were used to quantify the PCR-amplification products. The specific primers used for quantitative real‐time PCR are shown in Supplementary Table S1 and Actb was used as an internal control. The 2^− ΔΔCT method was applied to calculate the relative expressions of the genes.

Quantitative Proteomic Analysis by Liquid Chromatography Tandem Mass Spectrometry

The retinas of both eyes were harvested from mice after 4 weeks (n = 5/group). Protein extraction, protein purification, peptide fragment separation, and liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis, were performed following our previously published paper.³⁰ Normalized enrichment score (NES) was used to quantify the magnitude of enrichment, whereas the false discovery rate (FDR) was used to determine statistical significance. Protein data with an overall missing value exceeding 50% were removed, and the remaining blank values were refilled with a random number between 0 and the minimum area. Significant discrepancies were detected using Benjamin-Hochberg's false discovery testing. Differentially expressed proteins (DEPs) were identified based on the t-test (P values < 0.05) and a fold change (FC) > 2 or < 0.5. To verify DEPs, high-scoring peptides were selected for analysis using the PRM-PASEF experiment, which was based on the TIMS-TOF Pro mass spectrometry method described previously.³⁰ The peptide areas were manually checked for accuracy.

Statistical Analysis

We used the clusterProfiler R software package for Gene Ontology (GO) function enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis to interpret functional enrichment and create clusters that clarify the diversity between different functional genes/proteins. Protein-protein interaction (PPI) analysis was performed to demonstrate the interactions between proteins translated from target genes. The STRING database (https://www.string-db.org/) was used to recognize and predict PPIs.

IBM SPSS Statistics 20 was conducted to analyze data, and GraphPad Prism 8 (GraphPad Software Inc., San Diego, CA, USA) was used to create figures. Changes in refractive status between the lens-treated eyes and the control eyes were analyzed by repeated-measures 2-way analysis of variance (ANOVA), followed by Holm-Sidak post hoc tests for statistical significance. Independent t-tests were used to compare expression levels of dopamine or different mRNAs or insulin like growth factor 2 mRNA binding protein 1 (Igf2bp1) protein between groups. Differences were considered significant when P < 0.05. Data were presented as mean ± standard error (SEM).

Results

Inducement and Evaluation of a Murine Model of Lens Induced Myopia

The mice wearing the homemade spectacle frames appeared to have normal activity (see Supplementary Videos S1 and S2). These two situations, a spectacle frame “lost” and an entire apparatus “lost,” are crucial in the assessment of device stability.^21,31 Throughout the 4-week induction period, there were no instances of complete equipment loss. Partial unit loss (spectacle frame only) was found in two cases (2/15) during the first week and one (1/15) during the second week , which were retightened in minutes once spotted. The mounts remained complete in the rest of the experiment period. Notably, all “lost” frames were from different mice, ensuring that they were not eliminated from the study.

Myopia was successfully induced in mice using the method described in Figure 1. Lens treatment induced myopic refractive errors in the right eyes compared to the left eyes (2-way repeated-measures ANOVA main effect of treatment, F [1, 18] = 37.23, P < 0.001). After 2 weeks’ treatment, the right eyes displayed myopic refractive errors compared to the left eyes (LIM versus control: 2.24 ± 1.05 D vs. 8.43 ± 1.00 D, Holm-Sidak multiple comparisons, P < 0.01). By the end of the 4-week experiment period, the right eyes demonstrated significantly more myopic refractive error, along with longer AL and deeper VCD (Holm-Sidak multiple comparisons, LIM versus control: RE = 0.36 ± 0.80 D vs. 8.26 ± 0.42 D, P < 0.001; AL = 3.22 ± 0.01 mm vs. 3.19 ± 0.01 mm, P < 0.05 and VCD = 0.64 ± 0.01 mm vs. 0.59 ± 0.02 mm, P < 0.001).We compared the dopamine level in the two groups.³² It showed that the dopamine level was significantly decreased after 4-week’ lens-wearing (t-test, LIM versus control: 752.0 ± 37.43 pg/uL vs. 954.7 ± 69.03 pg/uL, P < 0.05).

Global Gene Expression in the Mouse Model of LIM

The flow chart of the transcriptomic and proteomic analysis is shown in Figure 2A. First, we investigated single-gene expression alterations to elucidate the transcriptome mechanisms underlying the response generated by minus lenses in the myopic mouse model. The entire retina was chosen for analysis, as it plays a critical role in converting optical defocus information into molecular signals. Reads mapping and quantification of transcript expression levels were performed. Hierarchical clustering of differentially expressed genes (DEGs; see Fig. 2B) revealed a clear separation between healthy control and LIM eyes. Based on the volcano plot, we identified 175 significant DEGs (see Fig. 2C, Supplementary Table S2), of which 152 genes showed increased expression and 23 had decreased expression. To validate the results of transcriptome sequencing, four genes (Alox12e, Cidec, Fgfbp3, and Prss56) were randomly selected from the transcriptome results for qRT-PCR validation (see Figs. 2D-G). Figure 2H and Supplementary Table S3 provides information on the results of GO annotation analysis. The most enriched biological process (BP) terms of the DEGs were “ERK1 and ERK2 cascade” and “positive regulation of leukocyte-mediated immunity.” Regarding molecular function (MF), the most enriched activities were related to serine.

Figure 2.

Identification of global DEGs in the LIM model. (A) Flowchart of the transcriptomic and proteomic analysis of the myopic retina. (B) Heatmap of DEGs. (C) Volcano plot of the DEGs. Genes significantly dysregulated in LIM compared to the control group (|log₂FC| > 2) were represented in green (23 downregulated) and red (152 upregulated). (D, F, G) qPCR results of Alox12e, Cidec, Fgfbp3, and Prss56 genes. (H) GO annotation analysis of the DEGs (n = 3/group).

Protein Expression in the Mouse Model of LIM

To gain insight into cellular and molecular dysfunction in this myopic model, we used an MS-based proteomics approach to analyze global changes in protein abundance. DEPs readily distinguished between groups, and we demonstrated the hierarchical clustering of top100 DEPs (ranked by |log₂FC|) in Figure 3A. We identified 646 significant protein changes out of 8004 total quantified proteins (Fig. 3B, Supplementary Table S4). Of these 646 DEPs, 249 had increased fold changes in myopic mice, whereas 397 had decreased fold changes. We also performed GO annotation analysis on the proteome data (Fig. 3C, Supplementary Table S5). The most enriched BPs were “chromatin organization,” “regulation of RNA splicing,” and “mRNA processing.” DEPs were predominantly located in the cytoplasm, cytosol, and nucleus. Furthermore, proteins with the most enriched molecular functions were those that bind to “protein,” “metal ion,” and “nucleotide.”

Figure 3.

Identification of global DEPs in the LIM model. (A) Heatmap of the top100 differentially expressed proteins in LIM versus control. (B) Volcano plot of the DEPs displaying downregulated proteins (397, green) and upregulated proteins (249, red), respectively. (C) GO annotation analysis of the DEPs (n = 3/group).

Transcriptome-Proteome Expression Comparison

We compared the proteome to the RNA-seq data set and observed a low correlation between the transcriptome and proteomic data we analyzed (Pearson's R = 0.01, P value = 0.5). Through matched transcriptomics and proteomics analysis, we compared the log₂ FCs in the transcriptome and proteome data sets between the LIM and control groups (Fig. 4A, Supplementary Table S6). Transcripts/proteins were annotated in quadrants 1 to 9 as 0, 458, 11, 15, 6264, 121, 1, 662, and 17, respectively. Of the 7549 proteins with both mRNA and protein values, 17.0% (1285 proteins) exhibited at least 2-fold differences at the mRNA and/or protein levels. We found 12 homodirectional changes (quadrant 3 and quadrant 7), and 17 opposite changes (quadrant 9). The detailed information on the 12 homodirectional changes of the mRNA/proteins was demonstrated in Table 1. Among the mRNA-protein matches, Igf2bp1 showed a P value < 0.05 for both mRNA and protein fold change. We validated the protein level of Igf2bp1 through Western blot (see Fig. 4B) and mapped a cluster relating Igf2bp1 to other DEPs based on the PPI network by STRING database (Fig. 4C). In quadrant 6, the majority (51.5% or 662 proteins) showed distinct fold changes only in protein content but not in mRNA, implying that the changes in most proteins were independent of synthesis.

Figure 4.

Transcriptome-proteome expression comparison. (A) The 9-quadrant associate analyses of mRNA and proteins. Numbers 1 to 9, quadrant NO. The number of points in each quadrant is shown in parentheses. (B) The protein level of IGF2BP1. (C) Constructed protein-protein interaction (PPI) network. Minimum required interaction score = 0.4. The larger size represented the greater degree (n = 3/group).

Table 1.

Information of the 12 Homodirectional Changes of the mRNA/Proteins

Gene ID	Gene Fold Change	P Value	P Adj	Protein Fold Change	P Value
Actbl2	−1.3756	0.043195	0.51822	−7.754325236	>0.99999999
Drc7	3.9835	0.051155	0.5626	4.912400134	0.000000494
Igf2bp1	1.58	0.046106	0.53582	4.154427252	0.000271165
Gipc3	1.104	0.079936	0.67781	4.860984717	0.000351621
Fsip2	3.6202	0.09541	0.73092	1.20385343	0.001470854
Optc	1.1471	0.035533	0.46972	1.52824293	0.048879176
Saa4	1.5475	0.56625	0.95862	1.50619561	0.051855061
Pbp2	1.1331	0.78124	0.95862	1.387571582	0.111511383
Rab44	3.3751	0.0042686	0.13186	3.952375447	>0.99999999
Galnt2	1.0013	0.49394	0.95862	2.593819633	>0.99999999
Tnnc2	1.3271	0.53095	0.95862	1.896054823	>0.99999999
Ctcfl	4.5732	0.010844	0.24108	2.564572367	>0.99999999

First column was the gene id for the DEG/DEP identified in this study. The gene fold change (second column), p-value (thrid column) and adjust p-value (forth column) in transcriptome were listed. The protein fold change (fifth column) and p-value (sixth column) in proteome were listed.

Pathway Analysis in Transcriptome and Proteome Implicated Similar Metabolic and Human Diseases Pathways

To better understand the underlying biology of dysregulated genes and proteins, we used KEGG annotation to interpret the RNA and protein data sets. The significant pathways in the transcriptome and proteome data sets are listed in Figure 5A and Figure 5B, respectively. We identified 5 pathways significantly enriched in the DEG data set and 16 pathways significantly enriched in the DEP data set (P value < 0.05; Supplementary Table S7). Additionally, we highlighted functional categories in Figure 5C that shared between the transcriptome and proteome data sets. Both the DEG and DEP data sets showed enrichment in pathways associated with metabolism and human diseases, including metabolism of cofactors and vitamins, energy metabolism, and neurodegenerative disease. Among these, “Metabolic pathways (mmu01100)” and “Huntington disease (mmu05016)” were notably observed because they were supported by both the DEG and DEP data sets. The transcriptome analysis indicated dysregulation of metabolism associated with energy and vitamins, whereas the DEP analysis implicated dysregulation of carbohydrate metabolism. In the “human disease” category, both DEGs and DEPs participated in pathways related to neurodegenerative diseases. Moreover, DEPs could participate in pathways related to substance dependence, a symptom that might be caused by a neurological disorder.

Figure 5.

KEGG Pathway analysis in transcriptome and proteome. The lists of the significant pathways in the individual transcriptome (A) and proteome data (B). (C) KEGG pathways in transcriptome (T) and proteome (P) datasets were grouped by broad categories associated with “metabolism” and “human diseases.” Respective pathways are plotted by -log10 P value (n = 3/group).

Comparison With Previous Published Omics-Based Data on Myopia

To assess the consistency of our data with previously published myopia research, we compared our findings to results from formal transcriptomic and proteomic studies in animal models based on both FDM and LIM.^33–40 In our transcriptome data set, we identified 17 out of the 175 DEGs as previously known myopic DEGs, which are listed in Supplementary Table S8. Among them, PRSS56 was identified in both a marmoset model of LIM³³ and mouse models of FDM, whereas the other 16 genes were only identified in previous studies on mouse models of FDM.^34–36 We applied the same criteria to identify hits in the results of our proteomic analysis (Supplementary Table S9), where we found that 155 out of the 646 DEPs were previously reported targets in myopia studies. The majority of these DEPs were identified in previous transcriptome studies of mouse models of FDM.^34–36

We also compared our transcriptome-proteome expression data with other LIM models in different species, including marmosets, guinea pigs, and chickens.^33,37–40 Table 2 shows the significant DEGs and DEPs that overlapped with previously identified LIM markers. Prss56 was the only DEG that overlapped with the results of a previous transcriptome study on marmosets. In that study, prss56 was found to increase in marmosets exposed to lens-defocus for 10 days and 5 weeks, whereas our study showed that prss56 decreased in mice after 4 weeks of lens defocus. Additionally, 24 DEPs overlapped with previous studies, of which 14 had the same direction of expression change as in the previous studies.

Table 2.

Significant DEGs and DEPs Overlapping With Known LIM Markers

Data	Gene Name	log2FC	P Value	Species	Time	Direction	Reference
T	PRSS56	−1.0953	8.86E-11	marmosets	10 d/5 wk	Increase	33
P	ADSS2	2.009405	0.018587	guinea pigs	5 d	Increase*	37
P	AMOTL2	−1.04468	0.012301	marmosets	5 wk	Decrease*	33
P	DBP	−1.79853	0.007655	marmosets	5 wk	Decrease*	33
P	MEAF6	-1.27143	0.049282	marmosets	5 wk	Decrease*	33
P	RFC2	1.683214	2.15E-07	marmosets	10 d	Increase*	33
P	UTP18	1.194589	0.026089	marmosets	10 d	Increase*	33
P	TOMM6	1.894216	0.02549	chicken	48 h	Increase*	40
P	ENPP2	−2.36184	0.008387	marmosets	10 d	Increase	33
P	CTNNA2	1.082536	0.006486	chicken	48 h	Increase*	39
P	ENO1	1.247267	1.65E-20	chicken	3 d	Increase*	38
P	GABRA2	1.284586	0.035835	chicken	48 h	Increase*	39
P	ACBD7	1.590129	0.018252	chicken	24 h/48 h	Decrease	39
P	CELF2	2.480019	8.76E-06	chicken	24 h/48 h	Decrease	39
P	COX17	1.487404	7.14E-06	chicken	48 h	Decrease	39
P	NCKAP1	−3.33823	0.00041	chicken	48 h	Increase	39
P	SCARB1	−1.12175	0.037411	chicken	48 h	Increase	39
P	CERS6	1.283034	0.014169	marmosets	5 wk	Increase*	33
P	NETO2	−2.9058	0.008961	marmosets	5 wk	Decrease*	33
P	PDIK1L	3.998712	0.000786	marmosets	5 wk	Increase*	33
P	ATP1A2	1.288902	0.044579	chicken	24 h	Decrease	39
P	NASP	−1.32101	0.005673	marmosets	5 wk	Increase	33
P	TMEM109	−1.07732	0.02248	marmosets	5 wk	Increase	33
P	Vip	−1.43829	0.011422	chicken	24 h/48 h	Decrease*	40
P	Tmsb4x	1.173933	8.58E-17	chicken	24 h	Decrease	39

The first column represents whether the data was from transcriptome (T) or proteome (P). The gene id (second column) for the DEG/DEP identified in this study, was listed along with the Log₂FC expression (third column) and p-value (fourth column) of the DEG/DEPs in our data set. We also listed the species (fifth column) and the experiment duration (sixth column) used for the LIM models in previous studies. The alternation direction of the DEG/DEPs expression was also provided in the seventh column.

^*Represented that the direction change of our result was in consist with the previous studies.

Discussion

Myopia is a leading cause of visual impairment globally, and understanding its molecular mechanisms can help establish appropriate treatment strategies for patients with myopia. Various animal models of experimental myopia have been developed due to the scarcity of human eye specimens. Breakthroughs in outstanding molecular signaling of myopia briefly come from the FDM and LIM animal models, in which emmetropization processes are extremely active and susceptible to the disruption of the visual environment.⁴¹ However, attaching diffusers or lenses to mice poses a challenge due to their small eyes and active motor behaviors. Barathi et al.¹⁸ initially described an LIM model by attaching lenses to the hair, and, subsequently, Tkatchenko et al.¹⁹ further improved stability by incorporating suturing procedures. Nevertheless, the lenses remained prone to detachment. Pardue et al.^20,31 introduced an exemplary noncontact head-mounted goggle for myopia research. Later, Jiang et al.²¹ and Kurihara et al.²² sequentially introduced refined head-mounted spectacle frames. These approaches all had favorable position stability and successful induction of myopia. However, the assembly comprised multiple components, and the complex surgical procedure necessitates exposure of the cranial surface, which may result in potential postoperative complications. In light of these challenges, we streamlined the surgical procedure, and facilitated easy adjustments to ensure accurate lens positioning.” This method proved to be effective, which could induce a relative myopic shift of about –8 D after 4 weeks, making it a valuable model for myopia research.

The correlation between certain mRNAs and their corresponding proteins showed poor consistency, as commonly observed in previously studies.^42,43 Our proteomic analysis revealed significant dysregulation of RNA metabolism, indicating a likely contribution of the active post-transcriptional regulation to the discrepancies between the two data sets. Complex post-translational events might also result in these disparities, but our approach lacked post-translational data like protein degradation rates, phosphorylation, and cellular localization. Additionally, differences in stability and lifetime of the two types of molecules could also contribute to the low correlation. Thus, whereas correlations between transcriptome and proteome data might be weak at times, a direct comparison provided a more comprehensive understanding of gene and protein expression regulation.

By analyzing the overlapped genes/proteins between control and LIM eyes, we identified 12 mRNAs whose expression levels were highly associated with their protein levels. Igf2bp1/IGF2BP1 was a significantly changed mRNA-protein match. The Igf2bp1 protein contains four K homology domains and two RNA recognition motifs and functions by binding to the mRNAs of certain genes including insulin-like growth factor 2 (Igf-2), β-actin, and β-transducin repeat-containing protein. Although the involvement of Igf2bp1 in the development of myopia has not yet been documented in the literature, recent research has recognized the importance of Igf-2 in myopia.^44,45 Our analysis found an underlying association between Igf2bp1’s physiological functions and retinal protein alterations, further strengthening the hypothesis linking myopia to insulin metabolism.³⁸ We mapped all the DEPs through the STRING database, which revealed that Igf2bp1 interacts with downregulated Ptbp1 and upregulated Srsf3. Ptbp1 was previously found to influence the homeostasis of choroidal vascular function, whereas Srsf3 is a member of the serine/arginine-rich family of pre-mRNA splicing factors that function as nucleic acid and RNA-binding proteins. Ptbp1 and Srsf3, along with Igf2bp1, were vital RNA-binding proteins previously implicated in the intron mutation process of genes related to lactic acidosis development.⁴⁶ The potential RNA regulation, including splicing and methylation,⁴⁷ suggested a crucial role for these proteins in the LIM model, warranting further investigation.

Furthermore, by leveraging the KEGG database for both proteomics and transcriptomics data sets, we gained a better understanding of the biochemical and cellular pathways involved in myopia formation. Among all pathways noted, “Vitamin B6 metabolism,” “Oxidative phosphorylation,” and “One carbon pool by folate” pathways only appeared in the transcriptome data set, whereas the proteome data indicated other disrupted pathways, including the “MAPK signaling pathway,” “Nicotine addiction,” “Glutamatergic synapse,” and “Retrograde endocannabinoid signaling.” From the two omics platforms, we summarized and highlighted the pathways that belong to similar categories. These aberrantly regulated metabolites were also supported by previous findings, including energy,^38,48,49 vitamins,⁵⁰ and carbohydrate metabolism.³⁸ Furthermore, our findings may have ramifications for the existing knowledge of the role of dopamine in myopia, as both “neurodegenerative diseases (Huntington's disease and amyotrophic lateral sclerosis)” and “substance dependence (nicotine addiction)” are closely correlated with the dopamine system.

We compared our myopia model to published models by identifying 17 DEGs and 155 DEPs that overlapped with previous research.^33–36 Among these, 48 DEPs were exclusively found in the LIM model, indicating the distinct mechanisms underlying myopia development between FDM and LIM models. Furthermore, most of the overlap was observed in studies involving FDM murine models. On the one hand, it might result from differences in research subjects, with LIM models primarily focusing on non-mouse animal species. Most of the overlap was observed in studies involving FDM murine models from Tkatchenko's laboratory.^34–36 On the other hand, one DEG (prss56) and 32 DEPs were shared between FDM and LIM models across species, suggesting the potential common pathways in the development of myopia. For instance, prss56 was implicated in both FDM and LIM models, which was known to regulate ocular axial growth^51,52 and play a crucial role in human myopia.^53–55 The detailed comparison between FDM and LIM could provide fundamental information for future myopia research, especially those using murine models for myopia research. Moreover, we compared our transcriptome-proteome expression data to other studies of LIM models using different modeling animals, including marmosets, guinea pigs, and chickens. Besides prss56, we found a total of 24 DEPs overlapped with former studies, which could provide a clue to the regulation of LIM. For instance, vasoactive intestinal peptide (Vip), belonging to the glucagon family, was a key candidate gene/protein for myopia. The dysfunctional retinal VIP-VIPR2 signaling pathway axis leads to myopia in both humans and mice.⁹

Conclusions

We initiated an extensive comparison of transcriptome and proteome data sets, using the refined LIM model, yielding a more comprehensive view of the biological dysfunction of LIM. By integrating these unbiased multi-omics approaches, we have identified potential novel genes, proteins, and pathways associated with myopia formation, with Igf2bp1 being a candidate biomarker in the process of myopia. The complementarity of transcriptomic and proteomic profiling unveiled altered metabolism and human diseases pathways, highlighting the close association between retina dysfunction and myopia development. The detailed comparison between FDM and LIM could provide fundamental information for future myopia research, particularly in the context of mice as an animal model for such investigations.