RNA-binding proteins and post-transcriptional regulation in lens biology and cataract: mediating spatiotemporal expression of key factors in the cell cycle, transcription, cytoskeleton and transparency

Abstract

Development of the ocular lens – a transparent tissue capable of sustaining frequent shape changes for optimal focusing power – pushes the boundaries of what cells can achieve using the molecular toolkit encoded by their genomes. The mammalian lens contains broadly two types of cells, the anteriorly located monolayer of epithelial cells which, at the equatorial region of the lens, initiate differentiation into fiber cells that contribute to the bulk of the tissue. This differentiation program involves massive upregulation of select fiber cell-expressed RNAs and their subsequent translation into high amounts of proteins, such as crystallins. But intriguingly, fiber cells achieve this while also simultaneously undergoing significant morphological changes such as elongation – involving about 1000-fold length-wise increase – and migration, which requires modulation of cytoskeletal and cell adhesion factors. Adding further to the challenges, these molecular and cellular events have to be coordinated as fiber cells progress toward loss of their nuclei and organelles, which irreversibly compromises their potential for harnessing genetically hardwired information. A long-standing question is how processes downstream of signaling and transcription, which may also participate in feedback regulation, contribute toward orchestrating these cellular differentiation events in the lens. It is now becoming clear from findings over the past decade that post-transcriptional gene expression regulatory mechanisms are critical in controlling cellular proteomes and coordinating key processes in lens development and fiber cell differentiation. Indeed, RNA-binding proteins (RBPs) such as Caprin2, Celf1, Rbm24 and Tdrd7 have now been described in mediating post-transcriptional control over key factors (e.g. Actn2, Cdkn1a (p21^Cip1), Cdkn1b (p27^Kip1), various crystallins, Dnase2b, Hspb1, Pax6, Prox1, Sox2) that are variously involved in cell cycle, transcription, cytoskeleton maintenance and differentiation in the lens. Furthermore, deficiencies of these RBPs have been shown to result in various eye and lens defects and/or cataract. Because fiber cell differentiation in the lens occurs throughout life, the underlying regulatory mechanisms operational in development are expected to also be recruited for the maintenance of transparency in aged lenses. Indeed, in support of this, TDRD7 and CAPRIN2 loci have been linked to age-related cataract in humans. Here, I will review the role of key RBPs in the lens and their importance in understanding the pathology of lens defects. I will discuss advances in RBP-based gene expression control, in general, and the important challenges that need to be addressed in the lens to define the mechanisms that determine the epithelial and fiber cell proteome. Finally, I will also discuss in detail several key future directions including the application of bioinformatics approaches such as iSyTE to study RBP-based post-transcriptional gene expression control in the aging lens and in the context of age-related cataract.

Introduction

The lens – a transparent living tissue – refracts and focuses light on the retina to enable high-resolution vision (Anand and Lachke, 2017; Bloemendal, 1977; Cvekl and Zhang, 2017; Lachke and Maas, 2010; Papaconstantinou, 1967; Piatigorsky, 1981). Cataract, defined as loss of lens transparency, causes alteration of refractive properties and elevated light scattering, resulting in sub-optimal vision or blindness (Shiels and Hejtmancik, 2019). Based on early or late onset, cataracts are classified as congenital/pediatric cataract or age-related cataract, respectively (Shiels and Hejtmancik, 2017). Depending on the population, congenital cataracts are detected in the range of 2 to 13 individuals among 10,000 live births and are alone responsible for an estimated one-tenth of childhood blindness cases worldwide (Bermejo and Martínez-Frías, 1998; Deng and Yuan, 2014; Haargaard et al., 2004; Pichi et al., 2016; Sheeladevi et al., 2016; Wu et al., 2016). Congenital cataract, having an underlying genetic basis in an estimated 8.3% to 25% of cases, presents as a clinically and genetically heterogenous defect that is currently associated with alterations in >100 genes (Shiels et al., 2010). On the other hand, while age-related cataract is also a multifactorial disease, it primarily results from environment-based alterations to lens proteins, the susceptibility toward which can be influenced by genomic variants (Choquet et al., 2021; Congdon et al., 2005; Hammond et al., 2001, 2000). Thus, to understand the pathology of congenital and age-related cataract, it is important to first define the cell and molecular biology of the lens, both in development as well as later in life.

The lens harbors in its anterior region, a monolayer of epithelium, cells of which retain the capacity for proliferation as well as for responding to signaling cues to initiate differentiation. Indeed, lens epithelial cells located in the transition zone, in response to fibroblast growth factor (Fgf) signaling, commence cell cycle exit and initiate differentiation into fiber cells (Lovicu and McAvoy, 2005). This differentiation program generates mature fiber cells that form the majority of the lens tissue and render it its characteristic property of transparency (Bassnett et al., 2011). Differentiating fiber cells express high amounts of transcripts encoding crystallins as well as specific membrane proteins. Crystallins provide the lens with refractive properties for transparency, while membrane proteins such as gap junction proteins and aquaporins allow the lens to harbor a water circulation-based physiology, a necessity resulting from the absence of vasculature around the mature lens. Furthermore, as they differentiate, lens fiber cells elongate by 1000X so as to extend across the entire anterior-posterior length of the lens, while also establishing cell-cell contact for optimal alignment (Audette et al., 2017, 2016; Cheng et al., 2016). However, it is intriguing that these molecular and cellular processes are coordinated as fiber cells initiate a program for the degradation of their nuclei and organelles, a process necessary for minimizing diffraction of light as it passes through this remarkable tissue (Morishita et al., 2021; Wride, 2011). Furthermore, the overall process of epithelial cells differentiating into fiber cells continues throughout life in mammals.

In the past three decades, studies on the genetics of inherited cataract in humans, combined with molecular genetics-based analyses of various animal models have led to the identification of the regulatory molecules that are important in lens development and homeostasis, and whose perturbations cause cataract (Cvekl and Zhang, 2017; Lachke and Maas, 2010). Not surprisingly, these data have established that signaling pathways (e.g. Ffg) and transcription factors (e.g. Pax6) play an important role in regulating key downstream targets such as crystallins in lens development and homeostasis (Cvekl and Zhang, 2017; Donner et al., 2006; Padula et al., 2019). Indeed, a detailed gene regulatory network for lens development can now be defined (Cvekl and Zhang, 2017; Lachke and Maas, 2010) and mutations in genes encoding crystallin proteins such as molecular chaperones, gap-junction proteins such as connexins, as well as growth factors, membrane proteins and intermediate filament proteins have been shown to cause cataract (Shiels et al., 2010; Shiels and Hejtmancik, 2019).

More recently, alterations in genes encoding RNA binding proteins (RBPs) – that are involved in post-transcriptional gene expression control – have been associated with lens defects and/or cataract in humans and animal models, which in turn has renewed the interest in this area of lens biology research (Dash et al., 2016; Lachke and Maas, 2011). The importance of post-transcriptional gene expression control was suggested by observations made in several early studies on the lens. For example, Drs. Joram Piatigorsky, David Beebe, and others observed spatiotemporal disconnections between specific mRNAs and their encoded proteins in the lens or lens epithelial cells (Beebe and Piatigorsky, 1981, 1977; Cenedella, 1995; Craig and Piatigorsky, 1973; Piatigorsky, 1981; Wang et al., 2004). Early studies showed that while delta-crystallin mRNA levels increased in cultured chicken lens epithelial cells, a corresponding increase in its encoded protein was not observed (Beebe and Piatigorsky, 1981). Further, these studies also showed that while the relative levels of delta-crytallin mRNAs in epithelial cells were largely similar between day 6 and day 19 lenses, the rate of synthesis of the encoded protein was about three times higher in the former compared to the latter, suggestive of post-transcriptional control mediating on the translational level in these cells (Beebe and Piatigorsky, 1981, 1977; Yoshida and Katoh, 1971). More recently, transcriptome analyses of human and mouse lenses from different embryonic and developmental stages indicate that lens epithelial cells contain mRNAs of genes the encoded proteins of which are predominantly detected in fiber cells (Hawse et al., 2005; Hoang et al., 2014; Nakahara et al., 2007; Zhao et al., 2018), in turn suggesting the involvement of translational control mechanisms in the lens. Dr. Beebe’s laboratory had previously showed that mRNAs encoding proteins with fiber cell “preferred” pattern, namely, crystallins, membrane proteins and transcription factors are present from early stages in lens development, in some cases even prior to formation of fiber cells, and only translated in later stages (Wang et al., 2004), again suggesting that translational control mechanisms are operational in the lens. Besides delta-crystallin, there is data suggesting that mRNAs for other crystallins, such as Beta-crystallin, are translationally controlled as polysomal RNAs were found to produce lesser amount of protein compared to non-polysomal mRNA from 1-day old chicken lenses (Thomson et al., 1978). Because Beta-crystallin levels are high in adult chicken lenses, it was proposed even at the early times, that Beta-crystallin mRNAs may be stored for up-regulation in translation post hatching (Piatigorsky, 1981; Thomson et al., 1978). Interestingly, early studies have also reported the translational inhibition of crystallin mRNAs in chicken optic cup (future neuroretina) cells (Clayton et al., 1979). Furthermore, early studies on the lens also focused on isolation of lens polyribosomes (Bloemendal et al., 1966; Spector and Travis, 1966), characterization of lens protein synthesis by in vitro translation (Strous et al., 1974; Vermorken et al., 1975) and attempts at isolation of lens proteins bound to RNA (Berns and Bloemendal, 1974; Chen et al., 1976). Thus, several studies in chicken or calf lenses showed that crystallin mRNAs were present in ribonucleoprotein particles, in turn supporting the involvement of protein-mRNA interactions in translational control (Chen et al., 1976; Lavers, 1976; Williamson et al., 1972). Additionally, from the time of the early studies on the calf or chicken lenses, it was appreciated that stability of specific mRNAs differed between lens epithelial and fiber cells, suggesting of post-transcriptional control (Papaconstantinou et al., 1966; Reeder and Bell, 1965; Stewart and Papaconstantinou, 1967; Yoshida and Katoh, 1971). In particular, the fiber cell differentiation program poses numerous challenges – translation of extremely high levels of proteins (i.e. crystallins ~450 mg/ml) while also translating sufficient quantities of other key proteins (e.g. TFs such as c-Maf, Mafg, Mafk, Prox1 and Sox1 whose mRNAs are comparatively less abundant as crystallins), and potential movement of molecules across long distances – that can be addressed by post-transcriptional control mechanisms. As another specific example, it can be asked how do transcriptionally and translationally “silent” fiber cells in the “organelle free zone” avoid breakdown of key proteins and ensure proper cytoskeletal structure and cell morphology. Together, these observations support the involvement of additional regulatory mechanisms, besides signaling and transcription, in the spatiotemporal control of the lens and other ocular tissue proteomes. Importantly, advancing the knowledge on post-transcriptional control mechanisms will also extend our understanding of how these pathways are coordinated with signaling and transcription and lead to a more comprehensive definition of the regulatory networks in the lens. Thus far, majority of the studies in the lens on gene regulation, in general, and post-transcriptional control, in particular, have focused on embryonic and early post-natal stages. In this review, I will discuss: (1) the significance of post-transcriptional control in determination of cellular proteome, (2) the role of key RBPs that have been described in the lens, (3) the new challenges in lens RNA biology, and (4) the proposed expansion of the iSyTE tool and other resources to study post-transcriptional control in the aging lens. Through discussion of these topics, this article aspires to point out exciting new directions toward understanding the basic biology of the lens as well as the etiology and pathology of cataract.

Post-transcriptional gene expression control: determination of the cellular proteome

The following is a fundamental question in tissue development and homeostasis: how is genomic information precisely interpreted spatiotemporally to coordinate morphogenesis in order to generate distinct cell types, and in turn, control their physiology? So far, eukaryotic gene expression studies have primarily focused on the level of signaling, transcriptional and epigenetic-mediated control of this process. However, it is now becoming clear that RBP-based post-transcriptional control mechanisms – that are downstream of signaling and transcriptional regulation but which may also participate in their control by feedback regulatory mechanisms – are critical in determining a cell’s proteome. It is estimated that the human genome encodes over 1800 RBPs, more in number than transcription factors (TFs), however, compared to the latter, substantially less is known about the former class of proteins in development and disease (Brinegar and Cooper, 2016; Castello et al., 2013, 2012; Chen and Manley, 2009; Gebauer et al., 2021; Gerstberger et al., 2014; Hentze et al., 2018; Manning and Cooper, 2017; Tian and Manley, 2017; Van Nostrand et al., 2020). This glaring knowledge-gap is driven home even more starkly considering that RBPs are involved in every aspect of regulation of the various types of RNA in a cell. For example, with regards to mRNAs, RBPs are involved in regulating it at various different stages, beginning from the initial steps of transcription through their destruction. Indeed, RBPs mediate control over mRNA at key stages such as its capping, splicing, cleavage and polyadenylation, editing, export, localization, stability, translation and decay (Fig. 1). The specificity of RBPs binding to their target RNAs is determined by the specific sequence and structural motifs present in the target RNA. Indeed, mRNAs are “coated” with different RBPs in complexes termed mRNPs. Just as binding of specific miRNAs (micro RNAs; non-coding RNAs of length 21–23 nucleotides) determine translation or stability of their target mRNAs, binding of specific RBPs to their target mRNAs can mediate similar level of control. Further, similar to transcription factors, multiple RBPs can function in a coordinate or combinatorial manner, but in the case of the latter, the outcome is regulation of their target RNAs. Indeed, it has been proposed that RBPs may coordinately regulate multiple functionally related mRNAs, representing “RNA regulons” analogous to operon-mediated control of transcription of multiple genes in a pathway (Keene, 2007). Thus, there are conceptual parallels between TFs and RBPs. However, while TFs dictate the onset of transcription in a cell, it is the function of RBPs along with the translation machinery and other post-transcriptional regulators (e.g. miRNA) that determines how this information is “translated” into the global cellular proteome, and even into “localized proteomes” inside the cell. Therefore, along with TFs, it is critical to identify the key RBPs and characterize their function in specific tissues such as the lens.

Fig. 1.

RNA-binding protein mediated post-transcriptional control of gene expression. RNA-binding proteins (RBPs) play a key role in all aspects of the mRNA life-cycle that determines the cellular proteome. Although RBPs are implicated in all classes of RNA, for convenience, the life-cycle of an mRNA is considered here. As the genome is transcribed into the transcriptome, RBPs are involved in direct function and/or interactions with the RNA ranging from capping, splicing, polyadenylation, quality-control by marking of exon-exon junction and export into the cytoplasm. In the cytoplasm, RBPs interact with the 5’ and/or 3’UTRs as well as other regions in the mRNA to control its localization, stability, decay and translation into protein, thereby determining the proteome of the cell. The schematic is intentionally made simple for clarity and does not include details such as potential use of alternate exons or alternate poly(A) sites, and also does not reflect the dynamics of these processes. However, it should be noted that splicing and other RNA processing events largely occur co-transcriptionally.

iSyTE and the discovery of key RBPs in lens biology and cataract

Within the past decade, the web-resource tool iSyTE (integrated Systems Tool for Eye gene discovery) has advanced our knowledge on the mechanistic basis of lens biology and pathology. Over a decade ago, it was first proposed that the application of “omics-based” approaches to the eye can lead to the definition of the “Oculome” – a comprehensive “systems level” understanding of eye development, homeostasis and pathology (Lachke and Maas, 2010). Soon after, the first version of iSyTE was developed (Lachke et al., 2012b) leading to the identification of TDRD7 as a new gene linked to congenital cataract in human patients and mouse and chicken animal models (Lachke et al., 2011), followed by the identification of another new gene (PVRL3; also known as NECTIN3) linked eye defects (Lachke et al., 2012a). Tdrd7 was the first tudor family protein to be described in the lens, and its discovery implicated the importance of regulators of post-transcriptional gene expression control in lens development (Lachke and Maas, 2011). This initiated the search for other RNA-binding proteins in the lens, leading to the functional characterization of Caprin2 (Dash et al., 2015), Celf1 (Aryal et al., 2020b; Siddam et al., 2018), Rbm24 (Dash et al., 2020; Lachke et al., 2012b) and Elavl1 (Aryal and Lachke, unpublished observations) in the lens. Subsequent studies have established all these post-transcriptional regulators as having a key role in lens biology and have demonstrated their deficiencies to result in eye and lens defects or cataract. I discuss the current knowledge on the function of the well characterized RBPs in the lens, in detail, below.

Tdrd7 in lens development and cataract

TDRD7 (Tudor domain containing protein 7; OMIM: 611258) was first identified in a human brain cDNA library (Nagase et al., 2000) and subsequently in mouse (Yamochi et al., 2001). Early studies involving yeast two-hybrid and co-immunoprecipitation assays demonstrated that TDRD7 interacts with CDK17 (Cyclin dependent kinase 17; also called PCTAIRE 2) (Hirose et al., 2000). In mouse spermatogenesis, Tdrd7 protein was reported to be part of a ribonucleoprotein complex and localize to a cytoplasmic RNA granule structure called “nuage” (also called germinal granule, chromatoid body or intermitochondrial cement) (Hosokawa et al., 2007). These data were not surprising because several Tudor family proteins had previously been described to localize to chromatoid bodies and to be involved in germ cell development across metazoa (Pek et al., 2012).

However, what came as a surprise was iSyTE’s prediction and the subsequent validation that TDRD7 mutations or deficiency caused congenital cataract in human, mouse and chicken (Lachke et al., 2011). Also surprising was the appearance of Tdrd7 protein in mouse lens fiber cell cytoplasm, which was granular and resembled the pattern of RNA granules known as chromatoid bodies found in mouse testes – although the size of the granules differed, with Tdrd7-stained chromatoid bodies in differentiating sperm being generally larger than the Tdrd7-stained granules in the lens. Since the original report, several new mutations in TDRD7 have been identified in inherited congenital cataract in humans (Chen et al., 2017; Fernández-Alcalde et al., 2021; Kandaswamy et al., 2020; Tan et al., 2019) and multiple Tdrd7-deficient mouse models have been generated and shown to develop cataract early in age (Tan et al., 2019; Tanaka et al., 2011). Interestingly, a single nucleotide polymorphism in TDRD7 has been associated with differential susceptibility to human age-related cataract (Zheng et al., 2014) and TDRD7 expression was found to be reduced in human age-related cataract lenses (Hawse et al., 2003), as well as in a rat model of cataract (Ishida et al., 2020). Together, these findings establish TDRD7 as a critical factor in lens development, homeostasis and cataractogenesis.

Animal and cell line-based studies have provided insights into Tdrd7 function. Tdrd7 encodes a large protein (the longest isoform is 1098 and 1119 amino acids in human and mouse, respectively) that harbors three OST-HTH (Oskar-Tdrd7-Helix turn helix)/LOTUS domains and three tudor domains. The OST-HTH/LOTUS domains – spanning about 80 amino acids – were identified in two independent reports and predicted to bind RNA (Anantharaman et al., 2010; Callebaut and Mornon, 2010). Subsequent structural studies have demonstrated that OST-HTH domains can interact with both protein and RNA. Studies on the oskar protein in Drosophila have shown that OST-HTH interacts with a dead-box helicase (Jeske et al., 2017, 2015). However, recent studies have shown that OST-HTH domains can also bind to guanine (G)-rich RNA sequences, including a unique RNA tertiary structure called RNA G-quadruplex that is found in coding and non-coding RNAs and can impact their regulation on the post-transcriptional level (Ding et al., 2020). Tudor domains, on the other hand, function as a scaffold by facilitating binding to methylated arginine or lysine in proteins, thereby allowing protein complex formation (Chen et al., 2011; Gan et al., 2019; Pek et al., 2012). Thus, Tdrd7 is expected to participate in specific protein-RNA and protein-protein interactions that may be important for its function in the lens.

Characterization of the lens in multiple Tdrd7-deficient mouse models have provided insights into the morphological, cellular and molecular nature of lens defects and cataract pathobiology (Anand et al., 2021; Barnum et al., 2020; Lachke et al., 2011; Tan et al., 2019; Tanaka et al., 2011). The lens defects in the Tdrd7-deficient mouse models morphologically resemble the defects observed in human cataract. This work showed that while the lens defects are fully penetrant in all the different mouse models (Lachke et al., 2011; Tan et al., 2019; Tanaka et al., 2011), Tdrd7-targeted knockout mice (Tdrd7−/−) on the C57/Bl6 background exhibit visible cataract early in life, at postnatal (P) day 22 (Barnum et al., 2020). Interestingly, just four days earlier at stage P18, Tdrd7−/− lenses do not exhibit defect as determined by light microscopy or histological analysis, but they exhibit cellular defects that are discernable by scanning electron microscopy (SEM). These data suggested that the cellular defects – and the underlying molecular changes – are already present in Tdrd7−/− lenses and contribute to the morphological defects detectable by light microscopy later. This also highlights the importance of using high-resolution imaging such as SEM for characterization of cellular phenotypes that may be missed by other approaches.

Further insight on the cellular nature of these lens defects was obtained through staining for phalloidin and what germ agglutinin, which showed that the morphology of fiber cells – specifically in the organelle degradation zone and deeper inside – was severely abnormal in Tdrd7−/− lenses. This suggested that Tdrd7 was necessary for proper morphology in fiber cells after initiation of organelle degradation (Barnum et al., 2020). Use of microarrays, RNA-sequencing and two-dimensional difference in-gel electrophoresis (2D-DIGE) led to the identification of several differentially expressed candidates in Tdrd7−/− lenses (Anand et al., 2021; Barnum et al., 2020; Lachke et al., 2011). Downstream analysis of these data using different parameters, including iSyTE-based information of gene expression during normal lens development, have led to prioritization of candidates that help explain specific aspects of the lens defects. For example, the heat shock protein Hspb1 (also known as Hsp27) is found to be reduced – both RNA and protein levels – in lens cells upon Tdrd7 deficiency (Barnum et al., 2020; Lachke et al., 2011). Further, RNA-immunoprecipitation as well as single-molecular fluorescence in situ hybridization assays indicate that Hspb1 mRNA colocalizes with Tdrd7 protein in specific regions in the lens.

These data have led to a working model wherein Tdrd7 protein is necessary for optimal build-up of Hspb1 mRNA and protein in the lens, especially in fiber cells prior to nuclear degradation (Fig. 2). Tdrd7-based control of Hspb1 may reflect a mechanism to ensure that key mRNAs with lower expression compared to crystallins are also translated at optimal levels necessary for their function. Tdrd7-based regulation of Hspb1 may be achieved through a mechanism wherein Tdrd7 protein’s direct interaction with Hspb1 mRNA may result in the latter’s increased stability, perhaps by reducing its decay. It is also possible that Tdrd7 may positively impact translation of Hspb1 protein, and the above two scenarios need not be mutually exclusive. Optimal Hspb1 levels, in turn, are necessary for development and/or maintenance of the characteristic cellular morphology of fiber cells after organelle degradation – a state that may resemble a “stress” condition. Indeed, Hspb1 – a protein that shares sequence conservation with crystallin alpha A and B – is shown to be needed for F-actin stability in other cells under conditions of stress (Arrigo, 2017; Clarke and Mearow, 2013; Doshi et al., 2010). It is possible that Hspb1 provides molecular chaperone activity that is necessary in the differentiating fiber cells because of the unusually high translation events and abundance of proteins. This model provides a potential explanation as to how Tdrd7-mediated post-transcriptional control over a key factor (Hspb1) in fiber cells allows these cells to address challenges arising from organelle degradation. Importantly, these data provided the first evidence that there are different mechanisms in the lens that regulate morphology of cells in the stages prior to organelle degradation and after organelle degradation (Barnum et al., 2020). Additionally, Tdrd7 deficiency results in misexpression of several miRNAs in the lens (Anand et al., 2021). Correlation of these miRNA to their potential mRNA targets provide insights into the complexity of Tdrd7-controlled downstream regulatory events in the lens. Further, Tdrd7 has recently been implicated in control of autophagy based on its direct binding to mRNA encoding Tbc1d20, a protein that regulates autophagosome maturation (Tu et al., 2021). These data show that several distinct pathways can be controlled by Tdrd7-based regulation of specific mRNAs in the lens, and alterations of these events upon Tdrd7 deficiency/mutation contributes to the pathology of cataracts.

Fig. 2.

Regulatory network controlled by RNA-binding proteins in the lens. Several conserved RNA-binding proteins (RBPs) – Caprin2, Celf1, eIF3h, Rbm24, Tdrd7 – have been characterized with regards to their function in the lens. The experimental-validated relationships (“edges”) between RBPs and their downstream targets (“nodes”) are outlined. Where there is evidence of RBPs to control translation, this is indicated by depicting both the mRNA and the resulting protein for a specific gene. Edges denoted by “broken lines” are indicative of RBP-based indirect control of a target, or where evidence of RBP-based direct control is not yet conclusively described. Direct relationships are defined as such when there is experimental evidence either of direct interaction (e.g. by RNA-immunoprecipitation, cross-linked immunoprecipitation, electrophoretic mobility shift assay) or function (e.g. by reporter analyses, RNA stability assay). Unless specifically described (e.g. Rbm24 positives controls crystallin mRNA poly(A)), RBP-based positive control of its target (denoted by arrows) generally reflects its positive impact on mRNA abundance, likely through elevation of mRNA stability.

Caprin2 and lens and ocular defects

Interestingly, a second gene encoding an RNA granule-associated protein, Caprin2 (Cytoplasmic activation- and proliferation-associated protein 2; also known as RNA granule protein RNG140; OMIM: 610375), was found by iSyTE to be a candidate of interest in the lens based on its highly enriched lens-expression (Dash et al., 2015). Independently, Caprin2 was identified as a gene with high expression in fiber cells compared to epithelial cells in stage E12.5 mouse lens (Dr. David Beebe, unpublished observations) and was found to be induced by fibroblast growth factor (Fgf) signaling – specifically FGF8 – in chicken lens fiber cells (Lorén et al., 2009). Lens-specific Caprin2 conditional knockout mice (Caprin2 cKO) exhibit two types of ocular defects. More commonly observed is a relatively mild defect which is characterized by the abnormal reduction of the central “nucleus” region of the lens in Caprin2 cKO animals (Dash et al., 2015). This is relevant to age-related cataract because aging in humans can cause a similar reduction in lens nucleus region termed nuclear compaction, which is linked to accommodation defects and cataract (Al-Ghoul et al., 2001; Augusteyn, 2010; Dubbelman et al., 2003). Indeed, Caprin2 has been identified in a locus associated with age-related cataract in a large, multiethnic genome-wide association study (GWAS) (Choquet et al., 2021). Less commonly observed – in about 8% of Caprin2 cKO mouse embryos – is a striking phenotype characterized by the presence of an abnormal lens-cornea “stalk”, which results from the faulty separation of these tissues during early eye development, and is associated in an ocular birth defect called as Peters anomaly in humans (Dash et al., 2015). Recently, the reduction of lens nucleus region defect has been independently demonstrated in a CRISPR/Cas9-based Caprin2 germline knockout (Caprin2 KO) mice (Nakazawa et al., 2020).

In mouse, Caprin2 encodes a 1031-long amino acid protein with two each of basic helix (coiled-coil) and RGG box domains (Shiina and Tokunaga, 2010) suggesting its function may involve binding to RNA and mediating post-transcriptional control. Interestingly, Caprin2 contains a complement component C1q domain in its C-terminal region and a nuclear localization signal upstream near the basic helix domains. Caprin2 has been shown to associate with RNA granules (Shiina and Tokunaga, 2010). Ectopic expression of Caprin2-GFP localizes to cytoplasmic aggregates of mitochondria in Chinese hamster ovary cells and results in elevated apoptosis. Further, stably expressed high levels of Caprin2 correlate with reduced cell growth. Interestingly, Caprin2 expression has been found to sharply elevate with onset of nuclear condensation and reduced cell proliferation in erythroid cell differentiation (Aerbajinai et al., 2004) – drawing parallels to lens fiber cell differentiation. Caprin2 expression is highly elevated in the cytoplasm of lens fiber cells from early stages of differentiation (e.g. E11.5 in mouse). However, interestingly, Caprin2 protein is detected in a granule-like pattern in mouse lens at E10.5. At this stage, Caprin2 granules are predominantly detected in the apical region of cells residing in the anterior-most “collar” part of the lens pit, suggesting a potential role in separation of the lens tissue from the surface ectoderm and consistent with the Peters anomaly-like abnormal lenti-corneal stalk that is present in a minority of Caprin2 cKO mice. Interestingly, the Caprin2 ortholog in Drosophila is termed Capr, which is described in RNA granules involved in translation inhibition and is linked with regulation of eye size in flies (Baumgartner et al., 2013; Papoulas et al., 2010; Shiina and Tokunaga, 2010).

Recently, Caprin2’s involvement in translation control over specific mRNAs has been examined in mammalian lens development. These new data indicate that Caprin2 suppresses eIF3 (eukaryotic initiation factor 3)-based initiation of translation of mRNA into protein (Nakazawa et al., 2020). Studies involving ribosome profiling, which were carried out on Chinese hamster ovary cells, showed that Caprin2 overexpression resulted in reduced translation of long-length mRNAs. These findings are complemented by studies performed on Caprin2 KO mice, which suggest that Caprin2-based suppression of long-length mRNAs also occurs in eye tissue. Interestingly, in non-lens cells – eIF3, a large translation initiation multi-protein complex generally required for translation initiation – is shown to be involved in translational regulation of specific mRNAs (Choudhuri et al., 2013; Hu et al., 2014; Lee et al., 2015). While it remains unclear how Caprin2 targets long-length mRNAs, based on present findings, a model is proposed to explain its function in the lens. In this model, Caprin2 suppresses relatively long-length mRNA such as those involved in cell cycle proliferation, thus indirectly allowing the translation machinery to “preferably” translate relatively short-length mRNA such as crystallins. Taken together, the studies described above provide new understanding into the function of Caprin2-based post-transcriptional control in lens development and fiber cell differentiation (Fig. 2).

Celf1’s conserved role in lens development and cataract

iSyTE identified yet another conserved RBP called Celf1 (CUGBP, Elav-like family member 1; also known as Cugbp1, Brunol2, Eden-bp; OMIM: 601074) to be linked to lens development and cataract (Aryal et al., 2020b; Siddam et al., 2018). Celf1 protein contains three RRM (RNA recognition motif) domains and is known to bind to specific RNAs and regulate their alternative splicing, translation into protein, or decay (Barreau et al., 2006; Beisang et al., 2012; Vlasova-St Louis et al., 2013; Vlasova-St Louis and Bohjanen, 2011). Interestingly, Celf1 has also been shown to associate with RNA granules (Fujimura et al., 2008). Celf1 expression in the lens is found to be conserved among different vertebrate model organisms, including fish, frog, chicken and mouse (Blech-Hermoni et al., 2013; Day and Beck, 2011; Farnsworth et al., 2021; Gautier-Courteille et al., 2004; Siddam et al., 2018; Suzuki et al., 2000). Celf1 germline knockout (Celf1 KO) or lens conditional knockout (Celf1 cKO) in mouse results in early-onset lens defects (Siddam et al., 2018). These lens defects are detectable by morphological, histological, and marker analyses starting from early stages (E12.5) with newborn mice exhibiting cataract. Moreover, celf1 knockdown in fish (Zebrafish) and frog (Xenopus) animal models results in lens and/or eye defects. These data suggest that Celf1 is essential for proper development of the lens in various vertebrate models.

Examination of the lens defects in Celf1 cKO mice showed that Celf1 is required for degradation of nuclei in fiber cell differentiation as well as for proper fiber cell morphology (Siddam et al., 2018). Subtle abnormalities in morphology of the lens epithelium were also noted. Molecular characterization using microarrays, RT-qPCR, immunostaining, Western blotting, RNA-immunoprecipitation (RIP), RNA crosslinked immunoprecipitation (CLIP), and reporter assays demonstrated that Celf1 regulated many mRNAs involved in development of the lens development, differentiation of fiber cells and epithelial cell biology (Aryal et al., 2020b; Siddam et al., 2018). Most notably, these studies showed that Celf1 is involved in the translational suppression of the cyclin-dependent kinase inhibitor, p27^Kip1 (Cdkn1b), in mid stages of fiber cell differentiation, which contributes to Cdk1 activation and its downstream phosphorylation of nuclear laminin proteins, a critical step in the breakdown of fiber cell nuclear envelope. Interestingly, Celf1 depletion results in high expression of transcripts of another cyclin-dependent kinase inhibitor, p21^Cip1 (Cdkn1a), indicating that it normally functions in suppression of these important cell cycle regulatory factors, albeit by distinct mechanisms, in the lens. Further, Celf1 also positively regulates, likely by impacting mRNA stability, the nuclease Dnase2b, which is necessary for DNA hydrolysis of fiber cell nuclei. Consequently, Celf1 regulates both the enzyme (Dnase2b) and its access to nuclear DNA, thereby facilitating degradation of nuclei in fiber cell differentiation, which is necessary for optimal passage of light through the lens (Siddam et al., 2018).

In addition to controlling fiber cell nuclear degradation, Celf1 is involved in controlling key transcription factors in the lens. In Celf1 cKO lenses, Pax6 protein is found to be elevated to abnormal levels in lens fiber cells, while Prox1 protein is abnormally elevated in both epithelial cells and fiber cells (Aryal et al., 2020b). However, a corresponding abnormal elevation in the mRNAs of these TFs is not detected in Celf1 cKO lenses, indicating that Celf1 post-transcriptionally controls the levels of Pax6 and Prox1 proteins in lens tissue. This is further supported by RIP assays that suggest Celf1 protein directly binds Pax6 and Prox1 mRNAs, and reporter assays that indicate that Celf1-based control is likely mediated via their 3’ UTRs. Finally, Celf1 cKO lenses exhibit altered levels of distinct splice isoforms (splice “isoforms” refers to the alternatively spliced forms of mRNAs, transcribed from the same transcription unit, but distinct in the sense that they differ in their composition of exons) of Sptb (spectrin, beta (erythrocytic); also termed Spnb1), suggesting that Celf1 may play a role in alternative splicing of specific mRNAs (Siddam et al., 2018). Together, these studies present a model for Celf1-based post-transcriptional gene expression regulation in lens development and fiber cell differentiation, which defines its control over key factors that determines epithelial cell and fiber cell proteome and orchestrates nuclear degradation in fiber cell maturation (Fig. 2).

Rbm24’s conserved role in eye and lens development and their associated defects

Rbm24 (RNA binding motif protein 24; also known as Seb4 in Xenopus; OMIM: 617603) was predicted by iSyTE to be a high-priority gene in the lens based on its high lens-enriched expression (Lachke et al., 2012b). Multiple independent investigations support this finding, demonstrating that Rbm24 and its orthologs are expressed in the lens in fish, frog, chicken and mouse (Dash et al., 2020; Grifone et al., 2018, 2014; Lachke et al., 2012b; Li et al., 2010; Maragh et al., 2014; Oberleitner, 2008; Poon et al., 2012; Shao et al., 2020). In mouse eye development, in early stages at E9.5, Rbm24 expression is detected in the optic cup and the overlying surface ectoderm, including the presumptive lens ectoderm (Dash et al., 2020). In later stages, Rbm24 exhibits high expression in lens fiber cells in mouse as well as in other vertebrate models such as zebrafish (Dash et al., 2020; Shao et al., 2020). Besides the lens, Rbm24 shows robust expression in other tissues such as skeletal muscle, cardiac muscle and inner ear, suggesting its role in these tissues (Grifone et al., 2018, 2014; Jin et al., 2010; Li et al., 2010; Poon et al., 2012; Yang et al., 2014). Rbm24 and its orthologs have been examined in several animal models, which have informed on its role in development of different organs and their associated defects.

Rbm24 germline knockout (KO) in mice results in features of anophthalmia, microphthalmia and lens defects (Dash et al., 2020). Conditional deletion of Rbm24 in the optic cup results in severe ocular defects, including anophthalmia and microphthalmia, suggesting its autonomous and specific role in mouse early eye development (Lachke, unpublished observations). Knockdown or targeted deletion/mutation of rbm24a by CRISPR/Cas9 or TALENs (transcription activator-like effector nucleases) in zebrafish results in microphthalmia and lens defects (Brastrom et al., 2019; Dash et al., 2020; Shao et al., 2020). Further, in zebrafish, rbm24a mutation is shown to result in lens fiber cell differentiation defects (Shao et al., 2020). These data suggest that Rbm24 controls distinct aspects of eye development in vertebrates.

Molecular analyses have led to novel insights into Rbm24 function in eye tissues. Rbm24 protein contains a single RRM domain that allows it to directly bind to its target RNAs and control their alternative splicing, cytoplasmic polyadenylation and stability (Grifone et al., 2020). In mouse, Rbm24 KO results in reduced expression of key eye expressed TF-genes, namely Lhx2 in the optic cup and Sox2 in the optic cup and the surface ectoderm, including the presumptive lens ectoderm (Dash et al., 2020). The regulatory relationship between Rbm24 and Sox2 seems to be conserved in vertebrates as Rbm24 deficiency results in reduction of Sox2 in both mouse and zebrafish (Brastrom et al., 2021; Dash et al., 2020). On the other hand, the expression of the key eye TF, Pax6, is reduced in the lens but not in the retina, in severely affected Rbm24 KO mouse eye tissue. The canonical Notch signaling pathway ligand, Jag1, and its downstream target Birc2, an inhibitor of apoptosis, are also found to be reduced in the lens in Rbm24 KO mice, which may explain the increased apoptosis in these animals. In Rbm24 KO mouse embryos exhibiting severely defective ocular tissues, lens markers such as crystallin gamma are profoundly reduced or absent. Electrophoretic shift and RIP assays show that Rbm24 protein directly binds to Sox2 mRNA via AU-rich elements (AREs) in its 3’UTR. Reporter assays combined with site-directed mutagenesis showed that Rbm24 directly and positively controls stability of Sox2 mRNA, requiring the presence of intact AREs, which ensures optimal levels of Sox2 (Dash et al., 2020). Because miRNAs are known to bind to Sox2 mRNA and regulate its downstream fate in other cell types (Otsubo et al., 2011), and because some of these miRNAs are known to be expressed in the lens (e.g. miR-126) (Kubo et al., 2013), it will be interesting to examine how miRNAs and RBPs such as Rbm24 mediate combinatorial control over Sox2 in lens cells. In zebrafish, loss of rbm24a resulted in reduced stability of cryaa mRNA (Shao et al., 2020). Interestingly, rbm24a loss also resulted in reduction in polyadenylated mRNAs of several crystallin genes, which results in the reduction of their encoded proteins, contributing to the lens fiber cells defects observed in zebrafish (Shao et al., 2020). While reduction in crystallin proteins has been attributed to decreased translational efficiency, the precise underlying mechanism remains to be determined. Some important insights on rbm24a’s impact on polyadenylation in zebrafish have been gained from the demonstration that it interacts with Cpeb1b (cytoplasmic polyadenylation element binding protein 1b) and Pabpc1l (poly(A) binding protein cytoplasmic 1 like), which are involved in promoting poly(A) tail elongation and translation stimulation as well as poly(A) protection, respectively (Shao et al., 2020). Together, these studies uncover some of the key mechanistic aspects of Rbm24’s role in the eye and lens (Fig. 2).

Future challenges in RBP-based control in the lens

It is evident from the knowledge gathered thus far that RBP-based post-transcriptional gene expression control is necessary for spatiotemporal control of cellular proteomes in the lens. Some of the molecular and cellular events where RBP-based control has been shown to function in the lens are outlined below – these events represent excellent candidates for identifying new regulatory relationships between established as well as new RBPs and their downstream targets (Fig. 3). These are: (1) repression of translation of fiber-cell enriched mRNAs that are also expressed in the anterior epithelium of the lens (e.g. Celf1-based negative control over Prox1 (Aryal et al., 2020b)), (2) decay of epithelial cell-enriched mRNAs in transition zone cells and early differentiating fiber cells (P-bodies, which are sites of mRNA decay, are described in lens epithelium and fiber cells (Lachke et al., 2011; Terrell et al., 2015) and mRNA encoding proteins (e.g. Foxe3) that need to be reduced in fiber cell differentiation may represent a good candidate for evaluating RBP-based decay in early fiber cells – in future studies), (3) elevated stability of mRNAs expressed in the lens (e.g. Celf1-based control of Dnase2b (Siddam et al., 2018); also likely is Rbm24-based control of Sox2 (Dash et al., 2020) and Tdrd7-based control of Hspb1 (Barnum et al., 2020; Lachke et al., 2011)), (4) decay and/or translational repression of factors associated with early differentiaon in late fiber cells (e.g. Celf1-based negative control over p27^Kip1 (Siddam et al., 2018)), and (5) preferential translation of proteins in fiber cells for lens transparency (e.g. eIF3h-based translation of gamma-crystallins in fish (Choudhuri et al., 2013); Rbm24-based control of translation of crystallins in fish (Shao et al., 2020), Caprin2-based inhibition of translation of long-length mRNAs, thereby indirectly facilitating translation of short-length mRNAs such as crystallins in lens fiber cells (Nakazawa et al., 2020)). Because fiber cell differentiation and maturation occurs throughout life, characterization of these factors will be significant to our understanding of the aging lens. The specific directions and challenges for advancing knowledge in RBP-based post-transcriptional gene expression regulation in the lens are outlined below (Fig. 3).

Fig. 3.

Challenges in lens biology likely to involve post-transcriptional control-based solutions. The key challenges that are posed due their cellular characteristics or other properties in the anterior epithelium of the lens (AEL), the transition zone (TZ) and fiber cells (FCs), and how these can be addressed by RNA-binding protein (RBP)-based post-transcriptional control are outlined. Observations, challenges and post-transcriptional “solutions” are described. In specific cases where “observations” or “solutions” are supported by the published data, the related key publications are cited.

Identifying new RBPs in the lens.

For addressing the goals outlined above, it would be necessary to identify new candidate RBPs for future studies in the lens. This can be can done by expanding and mining databases such as iSyTE (please see separate section below on expansion of iSyTE for these goals) and analyzing the rich transcriptome data and proteome data generated for whole lenses (Agrawal et al., 2015; Anand et al., 2018, 2015; Aryal et al., 2020a; Audette et al., 2016; Barnum et al., 2020; Cavalheiro et al., 2017; Kakrana et al., 2018; Khan et al., 2018, 2015; Lachke et al., 2012b; Siddam et al., 2018; Wolf et al., 2013) as well as isolated lens epithelial and fiber cells (Hoang et al., 2014; Nakahara et al., 2007; Zhao et al., 2019, 2018). Additionally, mining of the literature and expression atlases (e.g. GenePaint; https://gp3.mpg.de/) (Diez-Roux et al., 2011; Visel et al., 2004) can also be used to identify new RBPs in the lens. Indeed, in situ hybridization atlases for RBPs expressed in the brain can also be examined for these purposes these also contain images of the eye (McKee et al., 2005). Based on initial analysis of the literature, and while far from being comprehensive, we can already start to assemble a list of several RBPs, wherein their expression in the lens is independently validated by RT-qPCR, in situ hybridization or immunostaining and, in some cases, a function in the lens has also been experimentally demonstrated (Table 1). This list include known developmental defects-linked RBPs, such as Fxr1, Msi1, Stau1, Stau2 and Tia1, as well as candidates from the polypyrimidine tract-binding protein family that have been shown to be important in development of specific cell types, e.g. B cell development (Monzón-Casanova et al., 2020). Several of these RBPs have gene-perturbation animal models that have been generated but have not yet been examined for defects in the lens, thus making these goals a priority (Table 1). For example, Stau1 protein is detected in the lens, having a granule-like pattern in fiber cells, appearing to colocalize with Tdrd7 more so at the apical ends (Lachke et al., 2011). A related protein, Stau2, is also expressed in lens fiber cells and its knockdown in chicken results in a small eye defect (Cockburn et al., 2012; Lachke et al., 2011). Because both these RBPs are involved in mRNA localization in other cell types such as neurons and oocytes, it will be intriguing to examine if they play a role in potentially transporting select mRNAs, perhaps for localized translation in fiber cells, which have an elongated morphology. Interestingly, Abi2 and Cdk5 proteins are found to be located at the apical and basal regions of lens fiber cells, while myosin IIB is located at the posterior region of fiber cells, suggesting that some proteins have a localized pattern in fiber cells (Grove et al., 2004; Maddala et al., 2007; Negash et al., 2002). Further, Stau2 associates with translational repressors such as Pum2 in RNA granules in radial glial precursor cells (mouse) and potentially binds mRNAs that encode Prox1 and beta-actin – proteins that are also important for lens fiber cell differentiation and characteristic morphology (Vessey et al., 2012). Thus, Stau1 and Stau2 function in the lens can be examined in the future. Besides these RBPs, other candidates can be prioritized for studies in the lens by analysis of high-throughput RNA-sequencing data using web-based tools for identification of RBP-binding sites in the transcripts (Hwang et al., 2020). Initial efforts using such approaches have led to yet another newly identified RBP in the lens, Elavl1 (Embryonic lethal abnormal vision like 1; also known as HuR; OMIM: 603466), that plays an important role in eye and lens development (Lachke, unpublished observations). Interestingly, recent evidence suggests that the well-established transcription regulator and Rbm24 target, Sox2, may itself be involved in binding RNA in mouse embryonic stem cells (Holmes et al., 2020). Conversely, ectoderm-specific conditional deletion of an RBP called Quaking (QKI) in mouse results in cataract, but interestingly, its function in the lens is considered to involve binding to DNA and regulating Srebp2-based transcription of genes in the cholesterol biosynthesis pathway (Shin et al., 2021). Thus, in addition to proteins containing known RNA-binding domains/motifs, other “non-traditional” roles of known RBPs or TFs should also be investigated in the lens.

Table 1.

RNA-binding proteins with validated expression in vertebrate lens

RNA-binding protein	Expression in the lens and other eye tissues (by RT-PCR, in situ or Immunostaining)	Animal model examined or available	Linked to eye phenotype	Protein detected in specific RNA granules in any cell type	References
Caprin1 (RNG105)	Lens, Retina (Mouse)	Knockout mouse available	Eye examination not reported	SG	(Dash et al., 2015; Jain et al., 2016; Markmiller et al., 2018; Solomon et al., 2007; Youn et al., 2018)
Caprin2 (RNG140)	Lens (Chicken, Mouse)	Conditional knockout mouse and CRISPR/Cas9 deletion mouse examined	Compaction of lens central fiber region, Lenti-corneal stalk feature of Peter’s anomaly	RG*	(Lorén et al., 2009; Dash et al., 2015; Nakazawa et al., 2020; Shiina and Tokunaga, 2010)
Celf1 (Cugbp1)	Lens (Fish, Frog, Chicken, Mouse)	Germline knockout mouse examined, Conditional knockout mouse examined	Lens defects, cataract	PB, SG	(Suzuki et al., 2000; Gautier-Courteille et al., 2004; Day and Beck, 2011; Blech-Hermoni et al., 2013; Siddam et al., 2018; Aryal et al., 2020b; Fujimura et al., 2008; Cui et al., 2012; Jain et al., 2016; Youn et al., 2018; Markmiller et al., 2018)
Ddx6	Lens (Mouse)	-	-	PB, SG	(Hubstenberger et al., 2017; Jain et al., 2016; Lachke et al., 2011; Markmiller et al., 2018; Youn et al., 2018)
Elavl1 (HuR)	Cornea (Human), Lens (Mouse)	Conditional knockout mouse available	Down-regulation in human patients with Keratoconus	NG, PB, SG	(Bitel et al., 2010; Gallouzi et al., 2000; Guo et al., 2010; Jain et al., 2016; Joseph et al., 2012; Lachke et al., 2011; Markmiller et al., 2018; Tsai et al., 2009)
Elavl2 (HuB, Elrb)	Lens (Mouse), GCL (Frog)	-	-	PB, SG	(Amato et al., 2005; Bitel et al., 2010; Hubstenberger et al., 2017; Jain et al., 2016; Markmiller et al., 2018)
Elavl3 (HuC, Elrc)	Lens (Mouse), GCL, INL (Frog)	-	-	SG	(Amato et al., 2005; Bitel et al., 2010; Markmiller et al., 2018)
Elavl4 (HuD, Elrd)	Lens (Mouse), GCL, INL (Frog)	Overexpression in transgenic mouse	HuD-target splice forms in transgenic mouse lens	NG, PB, SG	(Amato et al., 2005; Bitel et al., 2010; Burry and Smith, 2006; De Santis et al., 2019; Markmiller et al., 2018)
Fxr1 (Xfxr1p)	Lens, Retina (Frog)	-	-	SG	(Huot et al., 2005; Jain et al., 2016; Markmiller et al., 2018; Mazroui et al., 2002; Youn et al., 2018)
Msi1 (Musashi-1)	Lens (Mouse), RPE (Newt, Frog, Mouse), Photoreceptors (Newt, Frog, Mouse), Regenerating retina (Newt), CMZ (Newt), Iris (Human)	Knockout mouse available, Conditional knockout mouse available	Photoreceptor degeneration, RPE65 protein reduced in RPE	PB, SG	(Raji et al., 2007; Susaki et al., 2009; Nickerson et al., 2011; Dhamodaran et al., 2014; Kaneko and Chiba, 2009; Sundar et al., 2021; ErLin et al., 2015; Kawahara et al., 2008; Markmiller et al., 2018; Hubstenberger et al., 2017)
Pabp (pabpc1.S)	Lens, retina (Frog)	-	-	NG, PB, SG	(Hoyle et al., 2007; Jain et al., 2016; Kedersha et al., 2000, 1999; Markmiller et al., 2018; Youn et al., 2018; Zelus et al., 1989)
Ptbp1	Lens (Mouse)	Knockout mouse available	Early embryonic lethality in mouse	SG	(Bitel et al., 2010; Markmiller et al., 2018; Shibayama et al., 2009)
Ptbp2	Lens (Mouse)	Knockout mouse available	Eye examination not reported	-	(Bitel et al., 2010; Li et al., 2014; Licatalosi et al., 2012)
Rbfox1	Lens (Mouse), GCL (Mouse)	-	-	PB, SG	(Bitel et al., 2011; Jain et al., 2016; Kucherenko and Shcherbata, 2018; McKee et al., 2005; Park et al., 2017)
Rbm24	Lens (Fish, Frog, Chicken, Mouse)	Knockdown in Zebrafish, Knockdown in Xenopus, Germline knockout mouse examined, Conditional knockout mouse available	Microphthalmia (Fish, Mouse), Anophthalmia (Mouse), Cataract (Fish), Lens alphacrystallin downregulation (Frog),	-	(Lachke et al., 2012b; Oberleitner, 2008; Li et al., 2010; Poon et al., 2012; Maragh et al., 2014; Grifone et al., 2014, 2018; Shao et al., 2020; Dash et al., 2020)
Stau1 (Staufen 1)	Lens (Mouse)	Knockout mouse available	Eye examination not reported	NG, SG	(Jain et al., 2016; Köhrmann et al., 1999; Lachke et al., 2011; Markmiller et al., 2018; Thomas et al., 2005)
Stau2 (Staufen 2)	Lens (Chicken, Mouse), neuroretina, retinal pigment epithelium (Chicken)	Overexpression in Chicken, Knockdown in Chicken	Increased eye size upon overexpression, Microphthalmia upon knockdown	PB, RG**, SG	(Cockburn et al., 2012; Hubstenberger et al., 2017; Jain et al., 2016; Lachke et al., 2011; Markmiller et al., 2018; Vessey et al., 2012; Youn et al., 2018)
Tdrd7	Lens (Chicken, Mouse), RPE (Mouse)	Knockout, ENU-mutant, CRISPR/Cas9 mutant mouse models examined (4 independent lines available), Knockdown in Chicken	Cataract (Human, Mouse, Chicken); Glaucoma (Human, Mouse)	CB, RG**, SG	(Anand et al., 2021; Barnum et al., 2020; Chen et al., 2017; Fernández-Alcalde et al., 2021; Hosokawa et al., 2007; Kandaswamy et al., 2020; Lachke et al., 2011; Tan et al., 2019; Tanaka et al., 2011; Youn et al., 2018; Zheng et al., 2014)
Tia1	Lens (Mouse)	Knockout mouse available	Eye examination not reported	PB, SG	(Jain et al., 2016; Kedersha et al., 1999; Lachke et al., 2011; Markmiller et al., 2018)
Tial1 (Tiar)	Lens (Mouse)	Knockout mouse available	Eye examination not reported	PB, SG	(Jain et al., 2016; Kedersha et al., 1999; Lachke et al., 2011; Markmiller et al., 2018; Youn et al., 2018)

CB, chromatoid body; NG, neuronal granule; PB, processing body; SG, stress granule; RG*, Caprin2 RNA granules that co-localize with FMRP and Poly(dT); RG**, Stau2 RNA granules are described in mammalian embryonic radial glial precursor cells; RG***, Tdrd7 RNA granules exhibit partial co-localization with Stau1, PB in lens fiber cells and with chromatoid body in sperm; RBPs in bold have been functionally characterized in the lens.

Defining RBP function in lens epithelium and fiber cells.

It will be important to define the roles different RBPs may play that are specific to lens epithelium and fiber cells. It will also be important to examine how the abundance of specific RNAs in lens epithelium and fiber cells and the resulting cell-specific differences in RBP-mRNA stoichiometry impact the overall outcome of RBP-based control (Fig. 4). For example, expression of Caprin2, Celf1, Rbm24 and Tdrd7 is significantly higher in fiber cells compared to epithelial cells in early lens development. In addition to addressing their function in these cells, the spatiotemporal expression pattern of these RBPs (as well as others) should be characterized in detail, which may point to different aspects of their function in epithelial or fiber cells, and in later stages of life. Further, just as there are differences in transcriptional machineries – in terms of combinatorial control mediated by transcription factors – it can examined whether there are differences between specific aspects of the post-transcriptional regulatory machinery and/or translational machinery in lens epithelial and fiber cells (Fig. 4). Models based on hypotheses resulting from these possibilities (that are not necessarily mutually exclusive) are proposed (Fig. 4). Indeed, it has been shown that specific translation initiation factors such as eIF3h are necessary for the optimal translation of gamma crystallins in zebrafish lens (Fig. 2) (Choudhuri et al., 2013). Toward this goal, it will be important to identify the RNA-bound proteome in lens epithelial and fiber cells. Indeed, early studies on the lens include attempts to identify lens proteins associated with RNA (Berns and Bloemendal, 1974). Present efforts to identify RNA-bound proteomes have been initiated on global scales for RBPs that bind to mRNA, non-coding RNA and newly transcribed RNA and new protocols have been optimized (Bao et al., 2018; Conrad et al., 2016; Huang et al., 2018; Perez-Perri et al., 2021; Wessels et al., 2016). Similarly, in order to characterize ribosomes spatiotemporally in the aging lens, it will be important to identify ribosomal components that may differ between lens epithelial and fiber cells. Perhaps such studies will uncover “translational” bursts involved in the build-up of specific proteins in lens fiber cells – analogous to the transcriptional bursts described in these cells (Cvekl and Eliscovich, 2021; Limi et al., 2018). Finally, the application of ribosomal profiling (also termed ribosome footprinting or ribo-seq), an approach that allows identification of which mRNAs are actively translated (Ingolia, 2014; Wang et al., 2021), to lens epithelium and fiber cells will uncover the “translatome” of these cells. Comparative analysis of RNA-sequencing data and ribosomal profiling data will allow identification of transcripts that are targets of cell-specific post-transcriptional regulation in the lens. Indeed, protocols for isolation of polyribosomes from calf lens epithelium and analyses of lens protein synthesis were described over 50 years ago (Bloemendal et al., 1966; Spector and Travis, 1966). Interestingly, new ribosome profiling studies in other systems have shown that upstream open reading frames (uORFs), which are regions within 5’UTRs (untranslated regions) of mRNAs, are capable of initiation and termination of translation upstream of the start codons of the protein coding region (Calvo et al., 2009). Such uORFs can impact translation of protein-coding regions and thereby control overall abundance of the protein. Indeed, changes in translated uORFs can contribute to the pathology in different diseases in humans (Lee et al., 2021). It will be interesting to examine the impact of the uORFs on the lens proteome.

Fig. 4.

Models for RNA-binding protein-based control in the lens. Two different models are proposed to explain how RNA-binding proteins (RBPs) that are expressed in anterior epithelium of the lens (AEL), transition zone (TZ) and fiber cells (FCs) can have different outcomes on the expression of downstream targets. In the model based on “differential RBP-mRNA stoichiometry”, the relative abundance of mRNAs, mRNA 1 (blue) and mRNA 2 (red) compared to that of the RBP (RBP1) that controls them, determines the overall abundance of the translated proteins. In these examples, for simplicity, the default outcome of binding of RBP1 on its target mRNAs is considered to be inhibition of translation. In this model, in the AEL, mRNA 1 is highly abundant compared to overall RBP1 protein levels and thus although RBP1 binds to a subset of mRNA 1 molecules and inhibits their translation, there is sufficient translation from other non-RBP1-bound mRNA 1 molecules to accumulate “Protein 1” (blue). A similar example is shown for FCs with regards to RBP1 and mRNA 2 that ensures sufficient levels of “Protein 2” (red). For simplicity, in the TZ where the transcriptome is considered to be a composite as gene expression transits from being AEL-like to more FC-like, a balance of the above scenarios is described. In the model based on “RBP-RBP combinational control”, presence of another RBP, RBP2 (magenta), can alter the outcome of the relatively straightforward “inhibition of translation” scenarios described above. In this model, binding of RBP1 and RBP2 can have multiple outcomes. Outcome 1 is that RBP1 and RBP2 binding together to its target mRNA (e.g. mRNA 1, blue) in AEL prevents translation inhibition that would normally occur by binding of RBP1 alone. Further, binding of RBP1 to RBP2 may prevent binding of both proteins to mRNA 1, or alternatively, RBP1 could competitively inhibit RBP2 binding to mRNA 1, the resulting outcome of both scenarios would be preventing translation inhibition. Thus, in these scenarios, mRNA 1 is translated into Protein 1 (blue). Conversely, RBP1 and RBP2 binding together to mRNA 2 (red) in AEL results in translation inhibition, and lower abundance of “Protein 2” (red). In the above events, the nature of the combinatorial control by different RBPs (i.e. RBP1 and RBP2) and the nature of the target (i.e. cis-binding elements in mRNA 1 or mRNA 2) determine the outcome of whether the mRNA will undergo translation. An analogous model is described for FCs which results in translation of “Protein 2” but not “Protein 1”. Key to the various molecules is provided in the grey box.

Uncovering RBP-based combinatorial control in the lens.

It is known that distinct transcription factors mediate combinatorial control to spatiotemporally express cohorts of genes in a cell or tissue-specific manner (Lambert et al., 2018). Similarly, it should be determined whether RBPs mediate a coordinate or combinatorial control over mRNAs in the lens, for example, those enriched in lens fiber cells and encoding crystallins and key membrane proteins. As discussed in the above sections, new RBPs expressed in the lens epithelium and fiber cells can be identified using various databases and their function in the lens can be examined using animal models. The overlap between the function of such newly identified RBPs and those already characterized (e.g. Caprin2, Celf1, Rbm24, Tdrd7) in the lens should be defined. Toward these goals, the interplay between Celf1 and a newly identified RBP in the lens, Elavl1, and the overlap between their downstream-regulatory networks, can be focused on, in future studies.

Understanding the significance of RNA granules in lens biology.

The significance of RNA-protein complexes, such as RNA granules – defined as membrane-less ribonucleoprotein (RNP)-based cellular “compartments” – in the lens, remains to be adequately addressed. Interestingly, 90% (18 out of 20) of the RBPs with independently validated expression in the lens are also recognized to be associated with RNA granules in other cell types, and some have even been associated with RNA granules in the lens (Table 1). RNA granules are classified into distinct classes, such as Processing bodies (P-bodies) and Stress granules (SGs) – among others, which are formed in the cytoplasm by localized high concentrations of key proteins and RNAs that travel back and forth between these subcellular compartments and the cytoplasm (Ivanov et al., 2019; Riggs et al., 2020; Tauber et al., 2020). SGs form in response to various conditions of stress (e.g. oxidative stress, heat shock, hypoxia, starvation, viral infection) and are made of excess mRNPs that are formed by protein-protein, protein-RNA and RNA-RNA interactions as a result of stalled translation initiation. P-bodies, on the other hand, are present in the cytoplasm under normal conditions (i.e. not necessarily only under stress conditions, although their numbers are known to be elevated by stress) and contain RNA. P-bodies are associated with mRNA decay and silencing and may represent centers for RNA storage. Because both SGs and PBs contain mRNAs, they are linked to cellular RNA metabolism.

Thus, it needs to be determined if assembly of RBP-RNA in the form of large membrane-less structures such as RNA granules, perhaps by offering localized environment within the cell, is a necessary aspect of the function of at least some of the RBPs in the lens. This would add additional layers of control, involving space and compartmentalization, necessary for achieving the characteristic proteomes of lens cells. Phase separation can allow the cytoplasm to be organized in membrane-less compartments such as RNA granules – which in turn allow spatiotemporal regulation over molecular interactions limited by constraints rendered by diffusion kinetics. It is proposed that aberrant phase separation may be linked to neurodegenerative diseases and aging of the cell (Alberti and Hyman, 2016). While challenging, it will be important to examine the significance of specific RNA granules to aging-related defects in the lens – similar to other aging-associated diseases (Alberti et al., 2017; Cao et al., 2020). In particular, it will be important to examine whether RNA granules and/or other RBP-RNA complexes ensure optimal RNA levels and spatiotemporal translation of specific RNAs such as crystallins or TFs in the lens, and whether their function differs between epithelial and fibers cells.

Lens epithelium-derived cell lines have been shown to express the relevant proteins and support the formation of P-bodies or SGs, and therefore can be used as a model for these studies (Dash et al., 2016; Lachke et al., 2011; Terrell et al., 2015; Weatherbee et al., 2019). For example, it can be examined whether lens epithelial cells or lens explant cultures form SGs when subjected to stress conditions such as ultraviolet radiation or low-dose ionizing radiation (Giblin et al., 2002; Uwineza et al., 2019), or under conditions of disease or developmental defects such as mutations in crystallin genes or alteration of crystallin proteins that cause protein-protein aggregation in cataract. Indeed, RBPs have been described in addressing UV-induced stress in other cells (Yang and Carrier, 2001) and preliminary findings suggest that lens epithelial cells form SGs in response to exposure to ultraviolet radiation (Lachke, unpublished observations). Further, cell lines derived from the lens epithelium of human patients with Myotonic Dystrophy Type 1 (DM1) – a disorder in which cataracts are prevalent (Smith and Gutmann, 2016) – exhibit defects in localization of the RBPs CELF1 (CUGBP1) and MBNL1 (Muscle-bind) to P-bodies and SGs, likely impacting the structure and function of these RNA granules (Gulyurtlu et al., 2021). P-bodies have been reported in embryonic and early postnatal lens, but have not been studied in the aging lens (Dash et al., 2016; Lachke et al., 2011; Terrell et al., 2015). It will be interesting to see if P-bodies are elevated in number – or SGs are detected – regardless of external conditions of stress, in the aging lens.

Further, misfolded proteins have been shown to induce aberrant SGs, which can be prevented by mechanisms involving chaperones (Mateju et al., 2017). Because autophagy defects are linked to lens pathology and cataract (Costello et al., 2013), it will be interesting to examine if pathways of autophagy are also involved in disassembly of SGs upon removal of stress conditions in the lens (in the event that SGs are assembled upon stress). Proteins containing intrinsically disordered regions (IDRs) are considered to function in stabilizing interactions in RNA granules (Mittag and Parker, 2018; Protter et al., 2018). Indeed, several RBPs that contain prion-like domains (PrLDs) – a subclass of IDRs prone to aggregation – are found in SGs, and interestingly, mutations in these proteins are associated with aggregation of proteins and degenerative diseases (Baradaran-Heravi et al., 2020; Ramaswami et al., 2013). Thus, to potentially identify new RNA granule components in the lens, iSyTE could be used to identify proteins that contain such domains and exhibit significant expression in development or aging. Finally, liquid-liquid phase separation (LLPS) – resulting from separation of dense liquid droplets from less-dense droplets based on differences in concentration of macromolecules, in turn determined by numerous weak multivalent interactions – is considered to contribute RNA granule formation (Mittag and Parker, 2018). LLPS may allow rapid exchange of molecules. Whether LLPS plays a role in RNA granule formation in the lens needs to be examined. Indeed, proteins that facilitate phase separation and mediate formation of RNA granules such as SGs are found to be expressed in the lens as per iSyTE (Guillén-Boixet et al., 2020; Sanders et al., 2020; Yang et al., 2020). Finally, it will be important to study RNA granules in the lens using live cell imaging as it will inform on their potential dynamic interactions.

Defining the significance of chemical modification of RBPs and their target RNAs in the lens.

Similar to DNA and proteins, RNA can be chemically modified after it is made. The field of RNA epigenetics has been opened due to the recognition of the “epitranscriptome” - defined as all the chemical modifications of RNA in a cell (Wiener and Schwartz, 2021). The fate of mRNA can be regulated by chemical modifications and over 170 distinct types of chemical modifications of RNA have been recognized thus far. The commonly described mRNA modifications are N⁶-methyladenosine (m⁶A), N¹-methyladenosine (m¹A), N⁷-methylguanosine (m⁷G), 5-methylcytidine (m⁵C), N⁴-acetylcytidine (ac⁴C), pseudouridine (Ψ), adenosine-to-inosine (A-to-I) and ribose methylations (N_m) (Walkley and Li, 2017; Wiener and Schwartz, 2021). Modification of RNA can impact its stability and function (Boo and Kim, 2020), with recent studies suggesting that small RNAs present on cell surface can be modified by glycosylation (Flynn et al., 2021). Finally, in addition to chemical modifications of RBP and their target RNAs, it will be interesting to examine if specific metabolites can also impact RBP function in the lens, based on previous findings that have shown UDP-glucose to impact interaction of specific RBPs (e.g. Elavl1/HuR) with its target mRNA (Wang et al., 2019).

Detailed characterization of known lens RBPs.

The four RBPs whose roles are described in the lens – Caprin2, Celf1, Rbm24 and Tdrd7 – should be studied further to identify their post-transcriptional regulatory networks in lens development, homeostasis and aging. Presently, these RBPs have be shown to regulate key processes in fiber cell differentiation in different model organisms in early developmental stages. For example, Celf1 mediates control over p27^Kip1 and the pathway for fiber cell nuclear degradation (in mouse), Tdrd7 mediates control over Hspb1 and maintenance of the cytoskeleton in fiber cells that have undergone post-organelle degradation (in mouse), Caprin2 function is considered to indirectly result in preferential translation of crystallin transcripts in fiber cells (in mouse) and Rbm24 is necessary for normal poly(A) tail of key lens fiber proteins (in fish). Thus, a reasonable hypothesis/question can be addressed, i.e. are these RBPs retained in the control of these processes in aging lenses? Some future experiments addressing these and other challenges are proposed below. For example, Caprin2 cKO should be analyzed in the context of Pax6 or Foxe3 deficiency, representing a sensitized background, to determine if Peters anomaly-related defects are worsened in these compound animals. Further, Caprin2 protein appears as granules in epithelial cells that contribute to the future lens and separate from the surrounding surface ectoderm, which contribute to the future cornea. Interestingly, the presumptive lens epithelial cells have a sharp reduction of specific proteins that are highly expressed in the presumptive corneal cells (e.g. p63) (Kuracha et al., 2011). Whether Caprin2 is involved in this sharp reduction of p63 in lens epithelial cells, perhaps by translational inhibition, should be examined. Because Caprin1 is also expressed in the lens and may contribute to redundancy, Caprin2 and Caprin1 double knockouts can be analyzed to determine if the combined deficiency of these proteins worsen the lens defects observed in Caprin2 cKO mice.

While it is known that key TFs, such as Pax6, are expressed as multiple isoforms in the lens (Epstein et al., 1994), alternative splicing events and the RBPs that controlling these have not been examined in the lens. Because Celf1 and Rbm24 are known to play a role in regulating mRNA alternative splicing in other cells, and indeed, Celf1 cKO lenses exhibit altered levels of Sptb splice isoforms (Siddam et al., 2018). Thus, the potential role of these RBPs in controlling splicing events should be investigated in lens cells. With regards to Tdrd7, the downstream-regulatory targets of this tudor family protein should be comprehensively identified and the nature of its control over them should be further investigated. Recent studies indicate Tdrd7’s role in controlling cytoskeleton of mature fiber cells (Barnum et al., 2020) and Celf1 deficient lenses also exhibit cytoskeletal defect (Siddam et al., 2018). Tdrd7 granules are often detected in a linear manner, antero-posteriorly, and appear to correspond to the length of fiber cells. It is not known whether this expression impacts Tdrd7’s role in ensuring optimal levels of mRNAs and/or their encoded proteins that in turn regulate the cytoskeleton. Thus, the role of these RBPs in controlling cytoskeleton has to be examined in more detail. Toward this goal, a proteomic screen to comprehensively identify RBPs associated with the cytoskeleton should be performed in the lens, similar to screens carried out in other cell types (Doroshenk et al., 2009). Further, whether these RBPs are involved in auto-regulatory feedback should be addressed (Müller-McNicoll et al., 2019). Moreover, it will be important to define the cross-talk of RBPs with regulators of signaling and transcription in the lens. Indeed, some data already offers interesting insights on this topic. For example, FGF8 induces Caprin2 expression in chicken lenses (Lorén et al., 2009), and deletion of the transcription factor Prox1 or expression of a mutant form of Ubiquitin results in reduction of Caprin2 and Tdrd7 expression in the lens (Audette et al., 2016; Liu et al., 2015). Finally, the mode of growth of the lens differs between different species and it will be interesting to examine the function of these RBPs (as well as other RBPs) that may be involved in these processes. For example, in humans, the lens switches from a phase of rapid growth in prenatal development to a liner phase of growth postnatally and throughout life (Augusteyn, 2010, 2008, 2007). It will be important to understand how the function of RBPs in the lens may differ between early development and later in life. These questions can be addressed specifically in primate or human lens development by the use of induced pluripotent stem cell-based lentoid model systems (Anchan et al., 2014; Fu et al., 2017; Murphy et al., 2018; Ooto et al., 2003; Yang et al., 2010). Together, these proposed studies will lead to a detailed understanding of the post-transcriptional regulatory network in the lens.

Expanding iSyTE to study the aging lens

The web-based bioinformatics tool iSyTE has been effective in identifying genes linked to pediatric cataract (Anand et al., 2018; Anand and Lachke, 2017; Aryal et al., 2020a; Kakrana et al., 2018; Lachke et al., 2012b, 2011; Patel et al., 2017) and recently has been helpful in prioritization of genes linked to age-related cataract as well (Choquet et al., 2021). iSyTE has also assisted in the characterization of numerous regulatory pathways in the lens (Audette et al., 2016; Cavalheiro et al., 2017; Kasaikina et al., 2011; Krall et al., 2018; Manthey et al., 2014; Padula et al., 2019; Wang et al., 2017; Wolf et al., 2013; Zhang et al., 2016). However, the transcriptome data – microarrays as well as RNA-seq – in the present version of iSyTE predominantly covers embryonic and early postnatal development stages, with the most “aged” stage being 2-month old mouse lenses (Anand et al., 2018; Kakrana et al., 2018). The inclusion of proteome data in iSyTE is limited to a single embryonic E14.5 lens (Aryal et al., 2020a). Further, recent findings point to sex-specific association of specific genomic loci with age-related cataract (Choquet et al., 2021). Dr. Melinda Duncan’s laboratory has generated new sex-specific transcriptome data on aged mouse lens epithelial and fiber cells (age: 24 months) (Faranda et al., 2021a, 2021b). In future, iSyTE can be expanded to include these data, as well as other new sex-specific lens cell-specific transcriptomics data on aged lenses. Importantly, analysis of these omics data with whole-embryo body (WB) in silico subtraction will potentially identify sex-independent and sex-specific genes enriched in aged lens cells. This approach can be used for generating sex-specific aged mouse lens cellular proteomes. Together, the expansion of iSyTE through inclusion of these datasets, and their presentation in the user-friendly iSyTE web-portal, will potentially lead to identifying new candidates associated with age-related cataract.

Applying iSyTE-based integrated approach to the aging lens

To prioritize new candidates in control of lens homeostasis and aging, integration of data from various public sources can be analyzed in the context of lens-specific information in iSyTE. This can be illustrated by the following example for Tdrd7, a gene associated with both early and late onset cataract. Thus far, TDRD7 is found to participate in protein-protein interactions with 32 other proteins, namely, ATXN2, BUB1, CCND1, CDK17, CDKN2A, CPEB4, CTNNA1, DCC, DDX19B, DLC1, FRAT1, GABPB1, HSPD1, IGF2BP2, INS, KIF1B, MAGEA8, MEX3B, MKRN2, MLH1, MLH3, MSH2, MUTYH, NRAS, ODC1, PABPC1, PAIP1, PIMREG, STK11, TACC1, TDRD3, and ZC3HAV1 (Source: NCBI, human TDRD7, Interactions). iSyTE analyses shows that majority of these potential Tdrd7 interactors are expressed in normal mouse lens development, several exhibiting lens-enriched expression, and some being mis-expressed in other gene-perturbation animal models of cataract. Furthermore, DAVID analysis of these TDRD7-interacting proteins identifies “poly(A) RNA binding” among the top gene ontology terms using the Functional Annotation Chart feature. Combining these various datasets can be used to develop informed hypotheses on the significance of the gene regulatory network associated with Tdrd7. Extending such analyses in the context of transcriptome/proteome information from aged lenses will point to new pathways in lens homeostasis and cataract pathology.

Conclusion

Studies as early as from 50 years ago suggested that post-transcriptional gene expression control may have an important role in lens development and homeostasis. Since the 2011 discovery of Tdrd7 as a cataract-linked gene with a critical function in the lens, the past decade has witnessed the identification of a growing number of RNA-binding proteins (e.g. Caprin2, Celf1, Rbm24) that are involved in post-transcriptional gene expression control in the lens. Majority of these genes have been shown to be important in the lens in multiple vertebrate species. These studies have showed that RBPs control key factors in distinct aspects of lens development and/or homeostasis, and that perturbations in RNA metabolism of lens cells results in tissue morphology defects and/or cataract. Now the mechanistic basis of how these RBPs control RNA stability and optimal protein translation in the lens has to be determined in detail. Further, because RBPs – similar to TFs – can function in combination with other factors, the complex RBP-RBP interaction networks in the lens need to be uncovered. Importantly, while significant insights have been gained into the importance of RBPs in the lens, for now these studies have focused on lens development or early postnatal lens biology. While this knowledge can be extended to later stages – because epithelial to fiber differentiation occurs throughout life – future studies should focus on understanding the role of RBPs in regulating the gene regulatory network and cellular proteome in aged lenses and cataract. Indeed, there is already some genetic evidence that RBPs such as CAPRIN2 and TDRD7 may be relevant to aged related cataract. Thus, future advancement of database resource-tools such as iSyTE to include transcriptomics and proteomics data on aged lens will set the stage for identifying the key RBPs – from around 1800 RBPs encoded in the human genome – that are relevant to lens homeostasis and cataract pathology.

Highlights.

RNA-binding proteins control lens development and homeostasis

Mutations or deficiency of RNA-binding proteins results in lens defects and cataract

RNA-binding proteins mediate post-transcriptional gene expression control in the lens

Tudor protein Tdrd7 positively controls heat shock protein Hspb1 and cell morphology

Celf1 (Cugbp1) negatively controls CDK-inhibitors p27^Kip1 and p21^Cip1 in lens cells