Genome-wide analysis of retinal transcriptome reveals common genetic network underlying perception of contrast and optical defocus detection

Background Refractive eye development is regulated by optical defocus in a process of emmetropization. Excessive exposure to negative optical defocus often leads to the development of myopia. However, it is still largely unknown how optical defocus is detected by the retina. Methods Here, we used genome-wide RNA-sequencing to conduct analysis of the retinal gene expression network underlying contrast perception and refractive eye development. Results We report that the genetic network subserving contrast perception plays an important role in optical defocus detection and emmetropization. Our results demonstrate an interaction between contrast perception, the retinal circadian clock pathway and the signaling pathway underlying optical defocus detection. We also observe that the relative majority of genes causing human myopia are involved in the processing of optical defocus. Conclusions Together, our results support the hypothesis that optical defocus is perceived by the retina using contrast as a proxy and provide new insights into molecular signaling underlying refractive eye development. Supplementary Information The online version contains supplementary material available at 10.1186/s12920-021-01005-x.

Although the role of optical defocus in refractive eye development is a well-established fact, it is still largely unknown how optical defocus is perceived by the retina. Several lines of evidence suggest that the absolute level of visual acuity does not play a significant role in optical defocus detection because a variety of species with different visual acuity ranging from 1.3 cpd in fish [34], 0.6-1.4 cpd in mice [35][36][37], 2.4 cpd in tree shrews [38,39], 2.7 cpd in guinea pigs [40,41], 5 cpd in cats [42,43], 5 cpd in chickens [44,45] and 44 cpd in rhesus monkeys and humans [46][47][48] undergo emmetropization and can compensate for imposed optical defocus [13][14][15][16][17][49][50][51][52][53]. Moreover, it was demonstrated that accommodation and emmetropization are driven by low spatial frequencies even in species with high visual acuity [54,55]; and the peripheral retina, which has much lower visual acuity than the central retina [56][57][58][59][60][61], plays an important role in defocus detection and emmetropization [8,22,24,62]. Conversely, perception of contrast is emerging as an important cue for both defocus-driven accommodation and emmetropization [54,63,64]. The possible role of contrast perception in emmetropization is highlighted by the fact that species with different visual acuity have similar contrast sensitivity at spatial frequencies found to be critical for emmetropization [65]. Optical defocus leads to a proportional degradation of contrast at the luminance edges of the images projected onto the retina, as revealed by the analysis of the effect of defocus on the eye's contrast sensitivity [66][67][68][69]. The retina, as revealed by in vitro recordings from ganglion cells, exhibits the highest sensitivity to contrast at low spatial frequencies, consistent with what is found for the dependency of accommodation and emmetropization on low spatial frequencies [45]. Ganglion cells' firing rate increases proportionally with an increase in both optical focus and contrast, suggesting that contrast may be used by the retina as a proxy to optical defocus [45]. Interestingly, vision across the animal kingdom is tuned to contrast and not to visual acuity [65,70,71]. Several species have been shown to possess high contrast sensitivity and undergo emmetropization despite low visual acuity [65,[72][73][74][75]. Retinae of many species adapted asymmetric photoreceptor distribution which maximizes contrast perception across visual space in expense of other important visual functions such as visual acuity and color perception [70,71,[76][77][78][79][80][81].
Accommodation and emmetropization appear to be driven by both luminance contrast and longitudinal chromatic aberrations, whereby the retina uses the contrast of the luminance edges to determine focal plane and color contrast to identify the sign of defocus [54,63,64,82]. Detection of luminance contrast and color contrast in the retina is organized as ON-center/OFFsurround and OFF-center/ON-surround receptive fields, which divide retinal pathways into ON and OFF channels respectively [83,84]. The importance of these contrast processing retinal channels for refractive eye development was demonstrated by experiments in knockout mice with ON-pathway mutations and in humans [85][86][87][88]. Anatomically, center-surround receptive fields are comprised of photoreceptors, bipolar cells, horizontal cells and amacrine cells, where horizontal and amacrine cells provide lateral inhibition which is critical for the generation of surround and contrast perception [83,84]. Interestingly, it was shown that amacrine cells play a critical role in refractive eye development [89][90][91][92][93][94][95][96][97][98][99] and that loss of amacrine cells negatively affects contrast sensitivity [100].
Here, we systematically analyzed gene expression networks and signaling pathways underlying contrast perception and refractive eye development in a panel of highly genetically diverse mice, which have different baseline refractive errors, different susceptibility to formdeprivation myopia and different levels of contrast sensitivity, and determined the contribution of the genetic network subserving contrast perception to baseline refractive development and optical-defocus-driven eye emmetropization.

Ethics approval and consent to participate
A total of 298 mice comprising collaborative cross (129S1/svlmj, A/J, C57BL/6J, CAST/EiJ, NOD/ShiLtJ, NZO/HlLtJ, PWK/PhJ, and WSB/EiJ mice) were used in this study. Mice were obtained from the Jackson Laboratory (Bar Harbor, ME) and were maintained as an inhouse breeding colony. Food and water were provided ad libitum. All procedures adhered to the Association for Research in Vision and Ophthalmology (ARVO) statement on the use of animals in ophthalmic and vision research and were approved by the Columbia University Institutional Animal Care and Use Committee (IACUC). Animals were anesthetized via intraperitoneal injection of ketamine (90 mg/kg) and xylazine (10 mg/kg) and were euthanized using CO 2 followed by cervical dislocation. The study was carried out in compliance with the ARRIVE 2.0 guidelines (Animal Research: Reporting of In Vivo Experiments), including both the "ARRIVE Essential 10" requirements, which describe information that is the basic minimum to include in a manuscript, and the "ARRIVE Recommended Set" requirements, which add context to the described study [116].

Analysis of contrast sensitivity
Contrast sensitivity was measured at P40 using a virtual optomotor system (Mouse OptoMotry System, Cerebral Mechanics) (Additional file 1: Tables S3), as previously described [117]. Briefly, the animal to be tested was placed on a platform surrounded by four computer screens displaying a virtual cylinder comprising a vertical sine wave grating in 3D coordinate space. The OptoMotry software controlled the speed of rotation, direction of rotation, the frequency of the grating and its contrast. Contrast sensitivity was measured at six spatial frequencies, i.e. 0.031, 0.064, 0.092, 0.103, 0.192, and 0.272 cycles/degree (cpd) using the staircase procedure. The contrast was systematically decreased from 100% using the staircase procedure until the minimum contrast capable of eliciting a response (contrast sensitivity) was determined. The staircase procedure was such that 3 correct answers in a row advanced it to a lower contrast, while 1 wrong answer returned it to a higher contrast. The contrast sensitivity at each frequency was calculated as a reciprocal of the contrast threshold, which was calculated as a Michelson contrast from the screen luminances ( I max −I min I max +I min ; I white = 208.25 cd/m 2 , I black = 0.21 cd/m 2 )

Analysis of refractive state of the eye
We measured baseline refractive errors in both left and right eyes on alert animals at P40 using an automated eccentric infrared photorefractor as previously described (Additional file 1: Tables S1) [118,119]. The animal to be refracted was immobilized using a restraining platform, and each eye was refracted along the optical axis in dim room light (< 1 lx), 20-30 min after the instillation of 1% tropicamide ophthalmic solution (Alcon Laboratories) to ensure mydriasis and cycloplegia. Refractive error measurements were automatically acquired by the photorefractor every 16 ms. Each successful measurement series (i.e., Purkinje image in the center of the pupil and stable refractive error for at least 5 s) was marked by a green LED flash, which was registered by the photorefractor software. Five independent measurement series were taken for each eye. Sixty individual measurements from each series, immediately preceding the green LED flash, were combined, and a total of 300 measurements (60 measurements × 5 series = 300 measurements) were collected for each eye. Data for the left and right eyes were combined (600 measurements total) to calculate the mean refractive error and standard deviation for each animal.

Analysis of susceptibility to form-deprivation myopia
We measured the extent of myopia induced by diffuserimposed retinal image degradation (visual form deprivation) (Additional file 1: Tables S2). Visual form deprivation was induced in one of the eyes by applying plastic diffusers and refractive development of the treated eye was compared to that of the contralateral eye, which was not treated with a diffuser, as previously described [50,120]. Diffusers represented low-pass optical filters, which degraded the image projected onto the retina by removing high spatial frequency details. Hemispherical plastic diffusers were made from zero power rigid contact lenses manufactured from OP3 plastic (diameter = 7.0 mm, base curve = 7.0 mm; Lens.com) and Bangerter occlusion foils (Precision Vision). Diffusers were inserted into 3D-printed plastic frames (Proto Labs). On the first day of the experiment (P24), animals were anesthetized via intraperitoneal injection of ketamine and xylazine, and frames with diffusers were attached to the skin surrounding the eye with six stitches using size 5-0 ETHILON ™ microsurgical sutures (Ethicon) and reinforced with Vetbond ™ glue (3M Animal Care Products) (the contralateral eye served as a control). Toenails were covered with adhesive tape to prevent mice from removing the diffusers. Animals recovered on a warming pad and were then housed under low-intensity constant light in transparent plastic cages for the duration of the experiment as previously described [50,120]. Following 21 days of visual form deprivation (from P24 through P45), diffusers were removed and the refractive state of both treated and control eyes was assessed using an automated infrared photorefractor as previously described [118,119]. The interocular difference in refractive error between the treated and contralateral control eye was used as a measure of induced myopia.

RNA extraction and RNA-seq
Animals were euthanized following an IACUCapproved protocol. Eyes were enucleated, the retinae were dissected from the enucleated eyes. The retinae were washed in RNAlater (Thermo Fisher Scientific) for 5 min., frozen in liquid nitrogen, and stored at − 80 °C. To isolate RNA, tissue samples were homogenized at 4 °C in a lysis buffer using Bead Ruptor 24 tissue homogenizer (Omni). Total RNA was extracted from each

Post-sequencing RNA-seq data validation and analysis
The FASTQ raw data files generated by the Illumina sequencing system were imported into Partek Flow software package (Partek), libraries were separated based on their barcodes, adapters were trimmed and remaining sequences were subjected to pre-alignment quality control using the Partek Flow pre-alignment QA/QC module. After the assessment of various quality metrics, bases with the quality score < 34 were removed (≤ 5 bases) from each end. Sequencing reads were then mapped to the mouse reference genome Genome Reference Consortium Mouse Build 38 (GRCm38/mm10, NCBI) using the STAR aligner resulting in 95.0 ± 0.4% mapped reads per library, which covered 35.4 ± 1.0% of the genome. Aligned reads were quantified to transcriptome using the Partek E/M annotation model and the NCBI's RefSeq annotation file to determine read counts per gene/genomic region. The generated read counts were normalized by the total read count and subjected to the analysis of variance (ANOVA) to detect genes whose expression correlates with either refractive error, susceptibility to myopia or contrast sensitivity. Differentially expressed transcripts were identified using a p value threshold of 0.05 adjusted for genome-wide statistical significance using Storey's q -value algorithm [121]. To identify sets of genes with coordinate expression, differentially expressed transcripts were clustered using the Partek Flow hierarchical clustering module, using average linkage for the cluster distance metric and Euclidean distance metric to determine the distance between data points. Each RNA-seq sample was analyzed as a biological replicate, thus, resulting in three biological replicates per strain.

Gene ontology analysis and identification of canonical signaling pathways
To identify biological functions (gene ontology categories), which were significantly associated with the genes whose expression correlated with baseline refractive errors, susceptibility to myopia or contrast sensitivity, we used the database for annotation, visualization and integrated discovery (DAVID) version 6.8 [122] and GOplot R package version 1.0.2 [123]. DAVID uses a powerful gene-enrichment algorithm and Gene Concept database to identify biological functions (gene ontology categories) affected by differential genes. GOplot integrates gene ontology information with gene expression information and predicts the effects of gene expression changes on biological processes. DAVID uses a modified Fisher's exact test (EASE score) with a p-value threshold of 0.05 to estimate the statistical significance of enrichment for specific gene ontology categories. The IPA Pathways Activity Analysis module (Ingenuity Pathway Analysis, QIAGEN) was used to identify canonical pathways encoded by the genes involved in baseline refractive eye development, susceptibility to myopia or contrast perception, and to predict the effects of gene expression differences in different mouse strains on the pathways. The activation z-score was employed in the Pathways Activity Analysis module to predict activation or suppression of the canonical pathways. The z-score algorithm is designed to reduce the chance that random data will generate significant predictions. The z-score provides an estimate of statistical quantity of change for each pathway found to be statistically significantly affected by the changes in gene expression. The significance values for the canonical pathways were calculated by the right-tailed Fisher's exact test. The significance indicates the probability of association of molecules from a dataset with the canonical pathway by random chance alone. The Pathways Activity Analysis module determines if canonical pathways, including functional end-points, are activated or suppressed based on the gene expression data in a dataset. Once statistically significant canonical pathways were identified, we subjected the datasets to the Core Functional Analysis in IPA to compare the pathways and identify key similarities and differences in the canonical pathways underlying baseline refractive development, susceptibility to myopia and contrast sensitivity.

Identification of genes linked to human myopia and other human genetic disorders
All mouse genes, which were found to be associated with baseline refractive errors, susceptibility to myopia or contrast sensitivity were analyzed for their association with human genetic disorders. To identify genes linked to human myopia, we compared the genes that we found in mice with a list of genes located within human myopia quantitative trait loci (QTLs). We first compiled a list of all single-nucleotide polymorphisms (SNPs) or markers exhibiting a statistically significant association with myopia in the human linkage or genome-wide association studies (GWAS) using the Online Mendelian Inheritance in Man (OMIM) (McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University) and NHGRI-EBI GWAS Catalog [124] databases. The LDlink's LDmatrix tool (National Cancer Institute) was used to identify SNPs in linkage disequilibrium and identify overlapping chromosomal loci. We then used the UCSC Table  Browser to extract all genes located within critical chromosomal regions identified by the human linkage studies or within 200 kb (± 200 kb) of the SNPs found by GWAS. The list of genes located within human QTLs was compared with the list of genes that we found in mice using Partek Genomics Suite (Partek). To identify genes associated with human genetic disorders unrelated to myopia, we screened mouse genes that we found in this study against the Online Mendelian Inheritance in Man (OMIM) (McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University) database.

Statistical data analysis and data graphing
Statistical analyses of the RNA-seq data were performed using statistical modules integrated into Partek Flow (Partek), DAVID [122] and IPA (QIAGEN) software packages and described in the corresponding sections. Other statistical analyses were performed using the STA-TISTICA software package (StatSoft). The statistical significance of the overlaps between gene datasets was estimated using probabilities associated with the hypergeometric distribution as implemented in the Bioconductor R software package GeneOverlap and associated functions. Data graphing and visualization was performed using Partek Flow (Partek) and IPA (QIAGEN) graphing and visualization modules, as well as Prism 8 for Windows (GraphPad) and GOplot R package [123].

Contrast sensitivity and susceptibility to form-deprivation myopia exhibit strong interdependence in mice
To investigate the role of contrast in defocus perception and refractive eye development, we analyzed the relationship between contrast sensitivity, baseline refractive error, and susceptibility to form-deprivation myopia in eight genetically diverse strains of mice comprising collaborative cross, i.e., 129S1/svlmj, A/J, C57BL/6J, CAST/ EiJ, NOD/ShiLtJ, NZO/HlLtJ, PWK/PhJ, and WSB/EiJ mice, at P40-P45.
Analysis of baseline refractive errors and susceptibility to form-deprivation myopia in these mice (Fig. 1A, B; Additional file 1: Tables S1 and S2) revealed that although both parameters were clearly influenced by the differences in genetic background between the strains (ANOVA refractive error , F(7, 145) = 429.8, p < 0.00001; ANOVA myopia , F(7, 48) = 9.8, p < 0.00001) and both baseline refractive error and susceptibility to form-deprivation myopia were inherited as quantitative traits, correlation between baseline refractive error and susceptibility to myopia was weak (r = 0.2686, p = 0.0305; Additional file 2: Figure S1). Therefore, we then analyzed contrast sensitivity in all eight strains of mice ( Fig. 1C; Additional file 1: Table S3) and found that genetic background strongly influenced contrast sensitivity (ANOVA contrast , F(7, 57) = 1837.7, p < 0.00001). Moreover, contrast sensitivity was also inherited as a quantitative trait. Linear regression analysis showed that the correlation between contrast sensitivity and baseline refractive error was not statistically significant (r = 0.1058, p = 0.4014) (Fig. 1D), whereas contrast sensitivity and form-deprivation myopia exhibited a positive statistically significant correlation (r = 0.5723, p < 0.0001) (Fig. 1E). Mouse strain-specific contrast sensitivity profiles revealed largely consistent differences between the strains at all frequencies ( Fig. 1F; Additional file 1: Table S3), with maximum contrast sensitivity recorded at 0.064 cpd. Therefore, contrast sensitivity at 0.064 cpd was used for all further analyses.
Collectively, these data suggest that processing of contrast by the retina plays an important role in optical defocus detection, visually regulated eye growth and susceptibility to form-deprivation myopia. On the other hand, contrast sensitivity does not seem to play any substantial role in baseline refractive eye development.

Contrast sensitivity, baseline refractive eye development, and susceptibility to form-deprivation myopia are controlled by diverse sets of genes and signaling pathways
Our data suggested that contrast sensitivity exhibited strong correlation with susceptibility to form-deprivation myopia and very weak correlation with baseline refractive error. Therefore, we then set out to investigate whether these observations would be reflected at the level of molecular signaling underlying contrast perception, baseline refractive eye development, and the development of form-deprivation myopia. To gain insight into the molecular signaling cascades involved in the regulation of these processes, we performed a genome-wide gene expression profiling in the retina of the eight mouse strains at P28 (age when refractive eye development in mice is progressing towards its stable plateau, but visual input is still influencing eye growth [50,118]) using massively parallel RNA-sequencing (RNA-seq). We found that expression of at least 2050 genes (q-value < 0.01) correlated with baseline refractive error ( Fig. 2A, Additional file 1: Table S4). These genes were organized in two clusters ( Fig. 2A). The expression of 718 genes in the first cluster positively correlated with baseline refractive error (i.e., expression was increased in hyperopic animals and decreased in myopic animals), while the expression of 1332 genes in the second cluster negatively correlated with baseline refractive error (i.e., expression was decreased in hyperopic animals and increased in myopic animals). Analysis of the gene ontology categories associated with biological processes revealed 88 biological processes, which were involved in baseline refractive eye development (Fig. 2B, Additional file 1: Table S7). Among these processes, several were linked to visual perception, synaptic transmission, cell-cell communication, retina development, and DNA methylation, which suggests that these processes play an important role in baseline refractive eye development. Gene ontology data were complemented by the analysis of canonical signaling pathways, which revealed that signaling pathways related to β-adrenergic signaling, protein kinase A signaling, dopamine receptor signaling, HIPPO signaling, mTOR signaling, phototransduction pathway, axonal guidance signaling, synaptic long term potentiation, and oxidative stress response signaling, among others, are involved in baseline refractive eye development (Fig. 2C, Additional file 1: Table S10).
Analysis of the relationship between gene expression and susceptibility to form-deprivation myopia uncovered that expression of at least 1347 genes (q-value < 0.01) correlated with susceptibility to form-deprivation myopia (Fig. 3A, Additional file 1: Table S5), including 455 genes whose expression positively correlated with susceptibility to form-deprivation myopia (i.e., expression was increased in animals with high susceptibility to Fig. 1 Contrast sensitivity strongly correlates with susceptibility to form-deprivation myopia in mice. a Baseline refractive errors range from highly myopic to highly hyperopic in mice depending on the genetic background. Horizontal red lines show mean refractive errors for each strain, while each dot corresponds to mean refractive errors of individual animals. b Susceptibility to form-deprivation myopia in mice is determined by genetic background. Horizontal red lines identify means of induced myopia for each strain, while each dot represents a mean interocular difference between the treated eye and contralateral control eye for individual animals. c Contrast sensitivity in mice is determined by genetic background. Horizontal red lines identify means of contrast sensitivity for each strain, while each dot represents a mean contrast sensitivity for individual animals. form-deprivation myopia and decreased in animals with low susceptibility to form-deprivation myopia) and 892 genes which exhibited a negative correlation with susceptibility to myopia (i.e., expression was decreased in animals with high susceptibility to form-deprivation myopia and increased in animals with low susceptibility to formdeprivation myopia). Gene ontology analysis revealed potential involvement of the biological processes related to axonogenesis, transport, fatty acid oxidation, aging, insulin secretion, cell proliferation, cell-cell adhesion, and cellular response to hypoxia, among others (Fig. 3B, Additional file 1: Table S8). Furthermore, analysis of canonical signaling pathways pointed to the important role of G2/M DNA damage checkpoint regulation, iron homeostasis signaling pathway, tight junction signaling, sirtuin signaling pathway, α-adrenergic and β-adrenergic signaling, HIPPO signaling, mTOR signaling, amyotrophic lateral sclerosis signaling, axonal guidance signaling, and growth hormone signaling, among others (Fig. 3C, Additional file 1: Table S11).
We found that regulation of contrast sensitivity in mice is associated with differential expression of at least 1024 genes in the retina (q-value < 0.01) (Fig. 4A, Additional file 1: Table S6). Expression of 489 of these genes was positively correlated with contrast sensitivity (i.e., expression was increased in animals with high contrast sensitivity and reduced in animals with low contrast sensitivity), whereas expression of 535 of these genes was negatively correlated with contrast sensitivity (i.e., expression was decreased in animals with high contrast sensitivity and increased in animals with low contrast sensitivity). Interestingly, contrast sensitivity was found to be strongly associated with biological processes related to rhythmic process, entrainment of circadian clock by photoperiod, circadian regulation of gene expression, regulation of circadian rhythm, detection of light stimulus involved in visual perception, visual perception, receptor-mediated endocytosis, synapse assembly, cell adhesion, positive regulation of catenin import into nucleus, and RNA splicing (Fig. 4B, Additional file 1: Table S9). At the level of canonical signaling pathways, we found a strong association of contrast sensitivity with CD27 signaling, leucine degradation, senescence pathway, phototransduction pathway, melatonin signaling, synaptic long-term depression, relaxin signaling, PPARα/RXRα activation, HIPPO signaling, and DNA methylation signaling, among other pathways (Fig. 4C, Additional file 1: Table S12).
Taken together, these data highlight the complexity and diversity of biological processes and signaling pathways involved in refractive eye development and suggest that the correlation between contrast perception and susceptibility to form-deprivation myopia may be explained by the common signaling pathways underlying both processes.
The relationship between contrast sensitivity and baseline refractive development appears to be less pronounced.

Comparison of transcriptomes underlying contrast sensitivity and baseline refractive eye development reveals limited contribution of the genetic network regulating contrast perception to baseline refractive eye development
To elucidate the relationship between contrast perception and baseline refractive eye development, we analyzed the overlap between gene expression networks, biological processes, and signaling pathways underlying contrast sensitivity and baseline refractive development. We found that less than 13% of genes involved in the regulation of contrast sensitivity (136 genes) exhibited a correlation with baseline refractive error (OR = 1.9, p = 1.05 × 10 −12 ) (Fig. 5A, Additional file 1: Table S13). These genes were involved in 11 biological processes shown in Fig. 5B (Additional file 1: Table S15), which implicate DNA and histone methylation, as well as cell adhesion and dendrite development. Importantly, both increased contrast sensitivity and hyperopic refractive errors were associated with suppression of many of these processes. Analysis of canonical signaling pathways affected by the genes whose expression correlates with both contrast sensitivity and baseline refractive error revealed that these two processes are controlled by largely distinct pathways (Fig. 5C, Additional file 1: Tables S10 and S12). Nevertheless, several canonical pathways involved in the regulation of contrast perception were also implicated in the regulation of baseline refractive eye development, including protein kinase A signaling, HIPPO signaling, leucine degradation, phototransduction pathway, AMPK signaling, senescence pathway, melatonin signaling, tRNA splicing, synaptic long-term depression, relaxin signaling, PPARα/RXRα activation, xenobiotic metabolism PXR signaling pathway, unfolded protein response, and DNA methylation and transcriptional repression signaling. Thus, although the overlap between the gene expression network underlying contrast perception and that underlying baseline refractive development is statistically significant, the cumulative evidence suggests that the overall impact of the gene expression network subserving contrast perception on baseline refractive eye development appears to be relatively low.

Comparison of transcriptomes underlying contrast sensitivity and the development of form-deprivation myopia reveals significant contribution of the genetic network regulating contrast perception to optical defocus detection and emmetropization
Considering that our data suggested that there was statistically significant correlation between contrast sensitivity and susceptibility to form-deprivation myopia, we analyzed the overlap between genes, biological processes, and canonical signaling pathways associated with contrast perception and the development of form-deprivation myopia. We discovered that more than 30% of genes (315 genes) whose expression correlates with contrast sensitivity are also involved in the regulation of susceptibility to form-deprivation myopia (Fig. 6A, Additional file 1: Table S14). The overlap between these two gene sets was highly significant (OR = 6.6, p = 6.61 × 10 −175 ). These genes were associated with 21 biological processes, implicating regulation of circadian rhythm, positive regulation of focal adhesion assembly, cellular response to hypoxia, axon extension, and DNA methylation (Fig. 6B, Additional file 1: Table S16). Importantly, the processes, which were activated in the animals with high contrast sensitivity, were also activated in the animals with high susceptibility to form-deprivation myopia, while the processes, which were suppressed in the animals with high contrast sensitivity, were also suppressed in the animals with high susceptibility to form-deprivation myopia. Analysis of the canonical signaling pathways encoded by the genes associated with contrast perception and susceptibility to form-deprivation myopia revealed that there was a substantial overlap between these two processes at the level of canonical signaling pathways (Fig. 6C, Additional file 1: Tables S11 and S12). Approximately 75% of pathways linked to form-deprivation myopia development (27 out of 36) were also associated with contrast perception, including tRNA charging, HIPPO signaling, AMPK signaling, NER pathway, IGF-1 signaling, protein kinase A signaling, role of JAK2 in hormonelike cytokine signaling, relaxin signaling, and PPARα/ RXRα activation. In summary, these data suggest that the gene expression network regulating contrast perception significantly contributes to optical defocus detection and visually guided eye emmetropization.

Genes involved in contrast perception are linked to human myopia and several other classes of human genetic disorders
Our data suggested that genes involved in contrast perception, baseline refractive eye development, and susceptibility to form-deprivation myopia encode a variety of signaling pathways regulating a diverse range of biological processes. To obtain an additional layer of information about the spectrum of contrast-perceptionrelated biological processes involved in refractive eye development, we analyzed the association between contrast genes, which were found to be involved in either baseline refractive development or form-deprivation myopia development, and human genetic disorders listed in the Online Mendelian Inheritance in Man (OMIM) database. We also analyzed the overlap between these genes and the genes found to be linked to human myopia by GWAS studies. We found that 26 out of 136 genes underlying both contrast perception and baseline refractive eye development were associated with known human disorders (Fig. 7A, Additional file 1: Table S17). The majority of these genes were linked to metabolic disorders (~ 25.8%), while other genes were associated with myopia (~ 22.6%), developmental disorders (~ 22.6%), connective tissue disorders (~ 9.7%), disorders affecting phototransduction-related signaling (~ 6.5%), neurologic disorders (~ 6.5%), and diseases caused by the breakdown in epigenetic regulation of gene expression (~ 6.5%). Forty-nine out of 315 genes involved in both contrast perception and form-deprivation myopia development were linked to known human diseases (Fig. 7B, Additional file 1: Table S18). The largest number of these genes were associated with myopia (39.7%). Other genes were associated with metabolic disorders (27%), developmental disorders (12.7%), diseases caused by the breakdown in synaptic transmission (11.1%), disorders affecting phototransduction-related signaling (4.8%), connective tissue disorders (1.6%), epigenetic disorders (1.6%), and diseases caused by the breakdown in DNA repair (1.6%). Cumulatively, these data suggest that the contrast-perceptionrelated genetic network contributes to baseline refractive eye development through metabolic and developmental processes, connective tissue restructuring, phototransduction-related signaling, and epigenetic regulation. These data also suggest that the contrast-perceptionrelated genetic network plays a prominent role in optical defocus detection and emmetropization mainly through myopia-related pathways, metabolic and developmental processes, synaptic transmission, and phototransduction-related signaling.

Discussion
Several lines of evidence suggest that the process of eye emmetropization is regulated by optical defocus, which is perceived by the retina using luminance contrast and longitudinal chromatic aberrations [8,64]. The eye is the most sensitive to optical defocus during a critical period in postnatal development, which continues in mice from the eye opening (P12-P14) to approximately P60 [50,118]. Our data support this hypothesis and provide experimental data, which suggest that the genetic network and signaling pathways subserving perception of contrast by the retina play an important role in emmetropization. Although our data indicate that the signaling pathways underlying perception of contrast contribute to both baseline refractive eye development and the optical-defocus-driven emmetropization process, contrast perception plays a much more prominent role in optical defocus detection and emmetropization than in baseline refractive development. We did not find a significant correlation between contrast sensitivity and baseline refractive errors, whereas sensitivity to contrast strongly correlated with susceptibility to form-deprivation myopia. The current study used a much larger dataset compared to our earlier study, which found that the genetic networks subserving baseline refractive eye development and susceptibility to myopia are largely distinct (more than 28,800 versus ~ 18,000 unique transcripts) [113]. Our current results mostly replicate our previous findings regarding the genes and signaling pathways involved in baseline refractive eye development and optical-defocus-driven eye emmetropization, [113]. Similar to the aforementioned study, we found that baseline refractive eye development was strongly associated with DNA and histone methylation, dopamine signaling, β-adrenergic signaling, protein kinase A signaling, HIPPO signaling, mTOR signaling, phototransduction, and oxidative stress response. We also found that visually guided eye emmetropization was strongly associated with circadian rhythms, response to hypoxia, α-adrenergic and β-adrenergic signaling, and growth hormone signaling, in addition to previously identified signaling pathways related to rRNA processing, protein transport, metabolism, autophagy, iron homeostasis signaling, tight junction signaling, methylmalonyl pathway, HIPPO signaling, mTOR signaling, axonal guidance, and amyloid processing.
We found that contrast perception is strongly dependent on the signaling pathways involved in axonogenesis, synaptic signaling, cell-cell communication, and regulation of circadian rhythms. However, further analysis of the biological functions and signaling pathways subserving contrast perception revealed that the contribution of contrast-related pathways to baseline refractive eye development and optical-defocus-regulated eye emmetropization is different. While contrast-related pathways involved in baseline refractive development were primarily related to DNA methylation, histone methylation and phototransduction, contrast-related pathways underlying optical defocus detection and emmetropization were primarily involved in circadian rhythms, response to hypoxia, metabolism and synaptic transmission (see, for example, how suppression of the pathway for synaptic long-term depression underlies increased contrast sensitivity and increased susceptibility to myopia, Additional file 3: Figure S2).
DNA methylation and phototransduction were previously implicated in refractive eye development. For example, genome-wide methylation status was linked to myopia development in humans [125,126]. In-utero epigenetic factors were found to be associated with refractive status in young children, and grandmothers' smoking causing epigenetic modifications of the genome was shown to be linked to less myopic refractive errors in children [127][128][129][130][131]. Light-induced signaling and the phototransduction pathway were also implicated in refractive eye development [12,113,114,132].
The finding that contrast perception is dependent on synaptic transmission is supported by previous reports that detection of contrast in the retina is organized as ON/OFF receptive fields, with a prominent role played by horizontal and amacrine cells which provide lateral inhibition via synaptic contacts [83,84]. Signaling pathways underlying response to hypoxia and metabolism were also implicated in myopia development [112,114]. However, we found that genes associated with the signaling pathways involved in the regulation of circadian rhythms Fig. 6 Genetic network underlying contrast perception has significant contribution to the regulation of susceptibility to form-deprivation myopia. a Venn diagram showing overlap between genes underlying contrast sensitivity and genes regulating susceptibility to form-deprivation myopia. b Top 11 biological processes affected by the genes associated with both contrast perception (top panel) and susceptibility to form-deprivation myopia in mice (bottom panel). Outer circle shows gene ontology IDs for the biological processes; middle circle shows up-or down-regulated genes in animals with high contrast sensitivity versus animals with low contrast sensitivity (top panel), or in mice with high susceptibility to form-deprivation myopia versus mice with low susceptibility to form-deprivation myopia (bottom panel); inner circle shows activation or suppression of the corresponding biological processes, while the size of the sector corresponds to statistical significance (larger sectors correspond to smaller p-values). c Comparison of canonical signaling pathways involved in contrast perception and susceptibility to form-deprivation myopia. Vertical red line indicates p = 0.05. Vertical grey line indicates p = 0.1. Colors identify pathways associated with either contrast perception or susceptibility to form-deprivation myopia and correspond to the colors in the Venn diagrams (a) (See figure on next page.) Tkatchenko and Tkatchenko BMC Med Genomics (2021) 14:153 were particularly overrepresented within the genetic network that controls contrast perception underlying defocus detection and emmetropization. This finding is especially intriguing because several studies found a link between circadian rhythms and refractive eye development [8,133]. Nickla et al. [134,135] discovered that the impact of optical defocus on refractive eye development was strongly dependent on the time of day and was associated with the endogenous rhythms in choroidal thickness and eye growth. It was also found that the effect of anti-myopia drugs quinpirole and pirenzepine on myopia development depends on the time of day [136]. Moreover, it was observed that contrast sensitivity strongly depends on the level of oxygen and glucose in the retina, both of which are under circadian control [137]. In line with this evidence, it was reported that optical defocus alters the expression of several genes encoding endogenous eye clock [138] and that targeted retina-specific disruption of the clock gene Bmal1 induces myopia-like phenotype in mice [139]. This evidence and our results suggest that the link between circadian rhythms and sensitivity of the eye to optical defocus, hence susceptibility to optical-defocus-induced myopia, may be explained by the strong dependence of contrast perception on the retinal circadian clock signaling. Analysis of the specific genes encoding components of the contrast-related signaling pathways led to several important findings. We found that 44% of contrast-related genes involved in the development of form-deprivation myopia in mice were linked to human myopia, while only 27% of contrast-related genes involved in baseline refractive eye development were associated with human myopia (Additional file 1: Tables S17 and S18). This suggests that the relative majority of genes causing human myopia are associated with pathways responsible for the processing of optical defocus.
Several of these genes deserve special attention. One of the genes, APH1B, encodes a critical component of the gamma-secretase complex, which is known to play an important role in the processing of amyloid beta (A4) precursor protein (APP) [140][141][142]. APP interacts with its homologs APLP1 and APLP2 to form a presynaptic complex in neuronal axons, which plays a critical role in synaptic transmission [143][144][145][146][147][148]. Importantly, APLP2 was found to play an important role in geneenvironment interaction underlying the development of childhood myopia [7]. Another gene, CACNA2D1, which encodes a subunit of calcium voltage-gated channels mediating the influx of calcium ions into the cell upon membrane polarization, also plays an important role in synaptic transmission in the retina [149][150][151][152]. Two genes encoding components of ubiquitin-protein-ligase complex, beta-transducin repeat containing E3 ubiquitin protein ligase (BTRC ) and ubiquitin recognition factor NPL4 homolog (NPLOC4), confirm the importance of the protein ubiquitination pathway in myopia development identified by several studies [12,[112][113][114][153][154][155][156][157]. Very little is known about the function of the tight junction protein 2 encoded by the TJP2 gene, which is primarily expressed in the inner nuclear layer of the retina [158]; however, our findings and recent studies point to a potentially important role of TJP2 in refractive eye development [113,159,160]. Another interesting gene, which we found to be linked to both contrast perception and susceptibility to formdeprivation myopia, is PER1. The PER1 gene encodes period circadian regulator 1 protein, which plays a critical role in the regulation of circadian rhythms [161][162][163]. PER1 is expressed in the inner nuclear layer of the retina harboring amacrine cells [162], which were implicated in optical defocus detection and myopia development [89][90][91][92][93][94][95][96][97][98][99]. Interestingly PER1 expression is regulated by the EGR1 transcription factor (also known as ZENK) and vasoactive intestinal polypeptide (VIP) [164,165]. The EGR1 gene is expressed in the amacrine cell of the retina and was shown to respond to optical defocus in a sign-of-defocus sensitive manner [92], while VIP is the principal neurotransmitter of the VIPergic amacrine cells of the retina, which were shown to be involved in myopia development [89,166,167]. Moreover, PER1 expression is controlled by the hypoxia signaling pathway [168], which was shown to play a critical role in the signaling cascade underlying the eye's response to optical defocus [112,114]. Thus, our data suggest an intriguing interaction between contrast perception, the retinal circadian clock pathway and the signaling cascade underlying optical defocus detection.
We also found a large number of contrast-related genes, which were not previously implicated in refractive eye development or myopia, but were found to be linked to a multitude of other human genetic (See figure on next page.) Fig. 7 Genes underlying contrast perception are associated with diverse group of human genetic disorders. a Chord diagram showing genes (left semicircle) and human genetic disorders (right semicircle) associated with contrast perception and baseline refractive eye development in mice. b Chord diagram showing genes (left semicircle) and human genetic disorders (right semicircle) associated with contrast perception and susceptibility to form-deprivation myopia in mice. Colored bars underneath gene names show up-or down-regulation of the corresponding genes in mice with high contrast sensitivity versus mice with low contrast sensitivity disorders. Analysis of these genes also provides additional insights into biological processes involved in contrast perception and refractive eye development. For example, several such genes associated with both contrast perception and baseline refractive development point to the involvement of several seemingly unrelated biological processes in baseline refractive eye development. The causal gene for a congenital form of cone-rod dystrophy POC1B and the gene causing Oguchi disease GRK1 are involved in the functioning of photoreceptor synapses and light-dependent deactivation of rhodopsin respectively [169][170][171], which suggests that photoreceptor-related signaling is involved in baseline refractive eye development. The involvement of TSHR gene influencing expression and patterning of retinal cone opsins points to the important role of the thyroid-stimulating hormone signaling pathway in baseline refractive eye development [172]. A causal gene for Parkinson disease GPR37 implicates dopamine signaling in baseline refractive eye development [173]. Finally, it was shown that a histone methyltransferase encoded by the causal gene for Weaver syndrome EZH2 interacts the polycomb repressive complex 2 (PRC2) and directly controls DNA methylation, implicating epigenetic regulation of gene expression in baseline refractive eye development [174,175].
On the contrary, analysis of the genes associated with both contrast perception and susceptibility to form-deprivation myopia implicates synaptic transmission and retinal ON/OFF signaling pathways in optical defocus detection and emmetropization. For example, the causal gene for juvenile amyotrophic lateral sclerosis and Kjellin syndrome SPG11, GABRA4 causing autism, DMXL2 linked to an autosomal dominant form of deafness and early infantile epileptic encephalopathy, as well as NTNG2 associated with a neurodevelopmental disorder are all involved in synapse function and synaptic transmission [176][177][178][179][180][181][182][183][184][185][186]. The causal gene for Chudley-McCullough syndrome GPSM2, POC1B linked to a congenital form of cone-rod dystrophy, GRK1 associated with a congenital stationary night blindness are involved in photoreceptor functioning [169-171, 187, 188], while NYX causing a congenital form of stationary night blindness and LHX4 linked to congenital pituitary hormone deficiency are involved in the communication between photoreceptors and cone bipolar cells [87,[189][190][191][192][193][194][195][196]. Considering that communication between photoreceptors and bipolar cells plays an important role in the organization of ON/OFF receptive fields [83,84], these two groups of genes point to the important role of ON/OFF signaling pathways in contrast perception and optical defocus detection.

Conclusions
In conclusion, we used a new larger gene expression dataset to answer one of the most fundamental questions of eye biology, i.e., how optical defocus is perceived by the retina. Our current results largely replicate our previous findings regarding the genes and signaling pathways involved in baseline refractive eye development and susceptibility to myopia [113]. They also reveal for the first time the importance of circadian rhythms, response to hypoxia, α-adrenergic and β-adrenergic signaling, and growth hormone signaling in visually guided eye emmetropization. In addition, our data provide evidence that the genetic network subserving contrast perception plays an important role in optical defocus detection and emmetropization. Our results reveal that contrast-related pathways involved in baseline refractive eye development are primarily related to DNA methylation, histone methylation, phototransduction, photoreceptor-bipolar cell signaling, thyroid-stimulating hormone signaling, dopamine signaling and epigenetic regulation of gene expression. Contrast-related pathways underlying optical defocus detection and emmetropization are primarily involved in retinal ON/OFF signaling, synaptic transmission, metabolism, response to hypoxia, and circadian rhythms. Our results suggest that the interaction between contrast perception, the retinal circadian clock pathway and the signaling cascade underlying optical defocus detection plays a key role in visually guided eye emmetropization. We note that the link between circadian rhythms and sensitivity of the eye to optical defocus may be explained by the strong dependence of contrast perception on the retinal circadian clock signaling. This study also suggests that the relative majority of genes causing common human myopia are involved in the processing of optical defocus, i.e., gene-environment interaction.
Additional file 1: Table S1. Refractive errors in mice comprising collaborative cross (P40, diopters). Table S2. Form-deprivation myopia in mice comprising collaborative cross (deprived eye versus control eye, diopters). Table S3. Contrast sensitivity in mice comprising collaborative cross. Table S4. Genes whose expression correlates with refractive error in mice comprising collaborative cross (FC = fold change, hyperopia versus myopia). Table S5. Genes whose expression correlates with susceptibility to myopia in mice comprising collaborative cross (FC = fold change, high versus low myopia). Table S6. Genes whose expression correlates with contrast sensitivity in mice comprising collaborative cross (FC = fold change, high versus low contrast sensitivity). Table S7. Gene ontology categories associated with genes whose expression correlates with refractive error in mice comprising collaborative cross (BP, biological process; CC, cellular component; MF, molecular function). Table S8. Gene ontology categories associated with genes whose expression correlates with susceptibility to myopia in mice comprising collaborative cross (BP, biological process; CC, cellular component; MF, molecular function). Table S9. Gene ontology categories associated with genes whose expression correlates with contrast sensitivity in mice comprising collaborative cross (BP, biological process; CC, cellular component; MF, molecular function). Table S10.
Canonical pathways associated with genes whose expression correlates with refractive error in mice comprising collaborative cross. Table S11.
Canonical pathways associated with genes whose expression correlates with susceptibility to myopia in mice comprising collaborative cross. Table S12. Canonical pathways associated with genes whose expression correlates with contrast sensitivity in mice comprising collaborative cross. Table S13. Genes whose expression correlates with both contrast sensitivity and refractive error in mice comprising collaborative cross (FC = fold change, contrast sensitivity = high versus low contrast sensitivity, RE = hyperopia versus myopia). Table S14. Genes whose expression correlates with both contrast sensitivity and susceptibility to myopia in mice comprising collaborative cross (FC = fold change, contrast sensitivity = high versus low contrast sensitivity, myopia = high versus low myopia). Table S15. Gene ontology categories associated with genes whose expression correlates with both contrast sensitivity and refractive error in mice comprising collaborative cross (BP, biological process; CC, cellular component; MF, molecular function). Table S16. Gene ontology categories associated with genes whose expression correlates with both contrast sensitivity and susceptibility to myopia in mice comprising collaborative cross (BP, biological process; CC, cellular component; MF, molecular function). Table S17. Genes whose expression correlates with both contrast sensitivity and refractive error in mice comprising collaborative cross and linked to human diseases (FC = fold change, high versus low contrast sensitivity). Table S18. Genes whose expression correlates with both contrast sensitivity and susceptibility to myopia in mice comprising collaborative cross and linked to human diseases (FC = fold change, high versus low contrast sensitivity). (XLS 898 KB) Additional file 2: Figure S1. Baseline refractive error correlates weakly with susceptibility to myopia. Linear regression showing weak correlation between baseline refractive error and susceptibility to myopia. r, Pearson's correlation coefficient; P, Pearson's correlation significance. (TIF 729 KB) Additional file 3: Figure S2. Suppression of the pathway for synaptic long-term depression leads to increased contrast sensitivity and increased susceptibility to form-deprivation myopia in mice. The diagram shows genes associated with the modulation of synaptic long-term depression in the retina and linked to both contrast perception and optical defocus detection in mice. (TIF 3156 KB)