Cis-regulatory basis of sister cell type divergence in the vertebrate retina

Machine Learning Prediction of Non-Coding Variant Impact in Cell-Class-Specific Human Retinal Cis-Regulatory Elements

Non-coding variants in cis-regulatory elements such as promoters and enhancers contribute to inherited retinal diseases (IRDs), however, characterizing the functional impact of most regulatory variants remains challenging. To improve identification of variants of interest, we implemented machine learning using a gapped k-mer support vector machine approach trained on single nucleus ATAC-seq data from specific cell classes of the adult and developing human retina. We developed 18 distinct ML models to predict the impact of non-coding variants on 39,437 cell-class-specific regulatory elements. These models demonstrate accuracy over 90% and a high degree of cell class specificity. Variant Impact Prediction (VIP) scores highlight specific sequences within candidate CREs, including putative transcription factor (TF) binding motifs, that are predicted to alter CRE function if mutated. Correlations to massively parallel reporter assays support the predictive value of VIP scores to model single nucleotide variants and indels in a cell-class-specific manner. These analyses demonstrate the capacity for single nucleus epigenomic data to predict the impact of non-coding sequence variants and allow for rapid prioritization of patient variants for further functional analysis.

Active learning of enhancers and silencers in the developing neural retina

Deep learning is a promising strategy for modeling cis-regulatory elements. However, models trained on genomic sequences often fail to explain why the same transcription factor can activate or repress transcription in different contexts. To address this limitation, we developed an active learning approach to train models that distinguish between enhancers and silencers composed of binding sites for the photoreceptor transcription factor cone-rod homeobox (CRX). After training the model on nearly all bound CRX sites from the genome, we coupled synthetic biology with uncertainty sampling to generate additional rounds of informative training data. This allowed us to iteratively train models on data from multiple rounds of massively parallel reporter assays. The ability of the resulting models to discriminate between CRX sites with identical sequence but opposite functions establishes active learning as an effective strategy to train models of regulatory DNA. A record of this paper's transparent peer review process is included in the supplemental information.

Organ Evolution: Emergence of Multicellular Function

Instances of multicellularity across the tree of life have fostered the evolution of complex organs composed of distinct cell types that cooperate, producing emergent biological functions. How organs originate is a fundamental evolutionary problem that has eluded deep mechanistic and conceptual understanding. Here I propose a cell- to organ-level transitions framework, whereby cooperative division of labor originates and becomes entrenched between cell types through a process of functional niche creation, cell-type subfunctionalization, and irreversible ratcheting of cell interdependencies. Comprehending this transition hinges on explaining how these processes unfold molecularly in evolving populations. Recent single-cell transcriptomic studies and analyses of terminal fate specification indicate that cellular functions are conferred by modular gene expression programs. These discrete components of functional variation may be deployed or combined within cells to introduce new properties into multicellular niches, or partitioned across cells to establish division of labor. Tracing gene expression program evolution at the level of single cells in populations may reveal transitions toward organ complexity.

Pathogenic variants in CRX have distinct cis-regulatory effects on enhancers and silencers in photoreceptors

Dozens of variants in the gene for the homeodomain transcription factor (TF) cone-rod homeobox (CRX) are linked with human blinding diseases that vary in their severity and age of onset. How different variants in this single TF alter its function in ways that lead to a range of phenotypes is unclear. We characterized the effects of human disease-causing variants on CRX cis-regulatory function by deploying massively parallel reporter assays (MPRAs) in mouse retina explants carrying knock-ins of two variants, one in the DNA-binding domain (p.R90W) and the other in the transcriptional effector domain (p.E168d2). The degree of reporter gene dysregulation in these mutant Crx retinas corresponds with their phenotypic severity. The two variants affect similar sets of enhancers, and p.E168d2 has distinct effects on silencers. Cis-regulatory elements (CREs) near cone photoreceptor genes are enriched for silencers that are derepressed in the presence of p.E168d2. Chromatin environments of CRX-bound loci are partially predictive of episomal MPRA activity, and distal elements whose accessibility increases later in retinal development are enriched for CREs with silencer activity. We identified a set of potentially pleiotropic regulatory elements that convert from silencers to enhancers in retinas that lack a functional CRX effector domain. Our findings show that phenotypically distinct variants in different domains of CRX have partially overlapping effects on its cis-regulatory function, leading to misregulation of similar sets of enhancers while having a qualitatively different impact on silencers.

Active learning of enhancer and silencer regulatory grammar in photoreceptors

Cis-regulatory elements (CREs) direct gene expression in health and disease, and models that can accurately predict their activities from DNA sequences are crucial for biomedicine. Deep learning represents one emerging strategy to model the regulatory grammar that relates CRE sequence to function. However, these models require training data on a scale that exceeds the number of CREs in the genome. We address this problem using active machine learning to iteratively train models on multiple rounds of synthetic DNA sequences assayed in live mammalian retinas. During each round of training the model actively selects sequence perturbations to assay, thereby efficiently generating informative training data. We iteratively trained a model that predicts the activities of sequences containing binding motifs for the photoreceptor transcription factor Cone-rod homeobox (CRX) using an order of magnitude less training data than current approaches. The model's internal confidence estimates of its predictions are reliable guides for designing sequences with high activity. The model correctly identified critical sequence differences between active and inactive sequences with nearly identical transcription factor binding sites, and revealed order and spacing preferences for combinations of motifs. Our results establish active learning as an effective method to train accurate deep learning models of cis-regulatory function after exhausting naturally occurring training examples in the genome.

Breaking enhancers to gain insights into developmental defects

Despite ground-breaking genetic studies that have identified thousands of risk variants for developmental diseases, how these variants lead to molecular and cellular phenotypes remains a gap in knowledge. Many of these variants are non-coding and occur at enhancers, which orchestrate key regulatory programs during development. The prevailing paradigm is that non-coding variants alter the activity of enhancers, impacting gene expression programs, and ultimately contributing to disease risk. A key obstacle to progress is the systematic functional characterization of non-coding variants at scale, especially since enhancer activity is highly specific to cell type and developmental stage. Here, we review the foundational studies of enhancers in developmental disease and current genomic approaches to functionally characterize developmental enhancers and their variants at scale. In the coming decade, we anticipate systematic enhancer perturbation studies to link non-coding variants to molecular mechanisms, changes in cell state, and disease phenotypes.

A single-cell massively parallel reporter assay detects cell-type-specific gene regulation

Massively parallel reporter gene assays are key tools in regulatory genomics but cannot be used to identify cell-type-specific regulatory elements without performing assays serially across different cell types. To address this problem, we developed a single-cell massively parallel reporter assay (scMPRA) to measure the activity of libraries of cis-regulatory sequences (CRSs) across multiple cell types simultaneously. We assayed a library of core promoters in a mixture of HEK293 and K562 cells and showed that scMPRA is a reproducible, highly parallel, single-cell reporter gene assay that detects cell-type-specific cis-regulatory activity. We then measured a library of promoter variants across multiple cell types in live mouse retinas and showed that subtle genetic variants can produce cell-type-specific effects on cis-regulatory activity. We anticipate that scMPRA will be widely applicable for studying the role of CRSs across diverse cell types.

Normalization benchmark of ATAC-seq datasets shows the importance of accounting for GC-content effects

The assay for transposase-accessible chromatin using sequencing (ATAC-seq) allows the study of epigenetic regulation of gene expression by assessing chromatin configuration for an entire genome. Despite its popularity, there have been limited studies investigating the analytical challenges related to ATAC-seq data, with most studies leveraging tools developed for bulk transcriptome sequencing. Here, we show that GC-content effects are omnipresent in ATAC-seq datasets. Since the GC-content effects are sample specific, they can bias downstream analyses such as clustering and differential accessibility analysis. We introduce a normalization method based on smooth-quantile normalization within GC-content bins and evaluate it together with 11 different normalization procedures on 8 public ATAC-seq datasets. Accounting for GC-content effects in the normalization is crucial for common downstream ATAC-seq data analyses, improving accuracy and interpretability. Through case studies, we show that exploratory data analysis is essential to guide the choice of an appropriate normalization method for a given dataset.

Chromatin Accessibility and Transcriptional Differences in Human Stem Cell-Derived Early-Stage Retinal Organoids

Retinogenesis involves the specification of retinal cell types during early vertebrate development. While model organisms have been critical for determining the role of dynamic chromatin and cell-type specific transcriptional networks during this process, an enhanced understanding of the developing human retina has been more elusive due to the requirement for human fetal tissue. Pluripotent stem cell (PSC) derived retinal organoids offer an experimentally accessible solution for investigating the developing human retina. To investigate cellular and molecular changes in developing early retinal organoids, we developed SIX6-GFP and VSX2-tdTomato (or VSX2-h2b-mRuby3) dual fluorescent reporters. When differentiated as 3D organoids these expressed GFP at day 15 and tdTomato (or mRuby3) at day 25, respectively. This enabled us to explore transcriptional and chromatin related changes using RNA-seq and ATAC-seq from pluripotency through early retina specification. Pathway analysis of developing organoids revealed a stepwise loss of pluripotency, while optic vesicle and retina pathways became progressively more prevalent. Correlating gene transcription with chromatin accessibility in early eye field development showed that retinal cells underwent a clear change in chromatin landscape, as well as gene expression profiles. While each dataset alone provided valuable information, considering both in parallel provided an informative glimpse into the molecular nature eye development.

A mechanism for red coloration in vertebrates

Red coloration is a salient feature of the natural world. Many vertebrates produce red color by converting dietary yellow carotenoids into red ketocarotenoids via an unknown mechanism. Here, we show that two enzymes, cytochrome P450 2J19 (CYP2J19) and 3-hydroxybutyrate dehydrogenase 1-like (BDH1L), are sufficient to catalyze this conversion. In birds, both enzymes are expressed at the sites of ketocarotenoid biosynthesis (feather follicles and red cone photoreceptors), and genetic evidence implicates these enzymes in yellow/red color variation in feathers. In fish, the homologs of CYP2J19 and BDH1L are required for ketocarotenoid production, and we show that these enzymes are sufficient to produce ketocarotenoids in cell culture and when ectopically expressed in fish skin. Finally, we demonstrate that the red-cone-enriched tetratricopeptide repeat protein 39B (TTC39B) enhances ketocarotenoid production when co-expressed with CYP2J19 and BDH1L. The discovery of this mechanism of ketocarotenoid biosynthesis has major implications for understanding the evolution of color diversity in vertebrates.

Single-cell transcriptomics identifies conserved regulators of neuroglandular lineages

Communication in bilaterian nervous systems is mediated by electrical and secreted signals; however, the evolutionary origin and relation of neurons to other secretory cell types has not been elucidated. Here, we use developmental single-cell RNA sequencing in the cnidarian Nematostella vectensis, representing an early evolutionary lineage with a simple nervous system. Validated by transgenics, we demonstrate that neurons, stinging cells, and gland cells arise from a common multipotent progenitor population. We identify the conserved transcription factor gene SoxC as a key upstream regulator of all neuroglandular lineages and demonstrate that SoxC knockdown eliminates both neuronal and secretory cell types. While in vertebrates and many other bilaterians neurogenesis is largely restricted to early developmental stages, we show that in the sea anemone, differentiation of neuroglandular cells is maintained throughout all life stages, and follows the same molecular trajectories from embryo to adulthood, ensuring lifelong homeostasis of neuroglandular cell lineages.

Functional analysis of the Vsx2 super-enhancer uncovers distinct cis-regulatory circuits controlling Vsx2 expression during retinogenesis

Vsx2 is a transcription factor essential for retinal proliferation and bipolar cell differentiation, but the molecular mechanisms underlying its developmental roles are unclear. Here, we have profiled VSX2 genomic occupancy during mouse retinogenesis, revealing extensive retinal genetic programs associated with VSX2 during development. VSX2 binds and transactivates its enhancer in association with the transcription factor PAX6. Mice harboring deletions in the Vsx2 regulatory landscape exhibit specific abnormalities in retinal proliferation and in bipolar cell differentiation. In one of those deletions, a complete loss of bipolar cells is associated with a bias towards photoreceptor production. VSX2 occupies cis-regulatory elements nearby genes associated with photoreceptor differentiation and homeostasis in the adult mouse and human retina, including a conserved region nearby Prdm1, a factor implicated in the specification of rod photoreceptors and suppression of bipolar cell fate. VSX2 interacts with the transcription factor OTX2 and can act to suppress OTX2-dependent enhancer transactivation of the Prdm1 enhancer. Taken together, our analyses indicate that Vsx2 expression can be temporally and spatially uncoupled at the enhancer level, and they illuminate important mechanistic insights into how VSX2 is engaged with gene regulatory networks that are essential for retinal proliferation and cell fate acquisition.

An enhancer located in a Pde6c intron drives transient expression in the cone photoreceptors of developing mouse and human retinas

How cone photoreceptors are formed during retinal development is only partially known. This is in part because we do not fully understand the gene regulatory network responsible for cone genesis. We reasoned that cis-regulatory elements (enhancers) active in nascent cones would be regulated by the same upstream network that controls cone formation. To dissect this network, we searched for enhancers active in developing cones. By electroporating enhancer-driven fluorescent reporter plasmids, we observed that a sequence within an intron of the cone-specific Pde6c gene acted as an enhancer in developing mouse cones. Similar fluorescent reporter plasmids were used to generate stable transgenic human induced pluripotent stem cells that were then grown into three-dimensional human retinal organoids. These organoids contained fluorescently labeled cones, demonstrating that the Pde6c enhancer was also active in human cones. We observed that enhancer activity was transient and labeled a minor population of developing rod photoreceptors in both mouse and human systems. This cone-enriched pattern argues that the Pde6c enhancer is activated in cells poised between rod and cone fates. Additionally, it suggests that the Pde6c enhancer is activated by the same regulatory network that selects or stabilizes cone fate choice. To further understand this regulatory network, we identified essential enhancer sequence regions through a series of mutagenesis experiments. This suggested that the Pde6c enhancer was regulated by transcription factor binding at five or more locations. Binding site predictions implicated transcription factor families known to control photoreceptor formation and families not previously associated with cone development. These results provide a framework for deciphering the gene regulatory network that controls cone genesis in both human and mouse systems. Our new transgenic human stem cell lines provide a tool for determining which cone developmental mechanisms are shared and distinct between mice and humans.

Building a Mammalian Retina: An Eye on Chromatin Structure

Regulation of gene expression by chromatin structure has been under intensive investigation, establishing nuclear organization and genome architecture as a potent and effective means of regulating developmental processes. The substantial growth in our knowledge of the molecular mechanisms underlying retinogenesis has been powered by several genome-wide based tools that mapped chromatin organization at multiple cellular and biochemical levels. Studies profiling the retinal epigenome and transcriptome have allowed the systematic annotation of putative cis-regulatory elements associated with transcriptional programs that drive retinal neural differentiation, laying the groundwork to understand spatiotemporal retinal gene regulation at a mechanistic level. In this review, we outline recent advances in our understanding of the chromatin architecture in the mammalian retina during development and disease. We focus on the emerging roles of non-coding regulatory elements in controlling retinal cell-type specific transcriptional programs, and discuss potential implications in untangling the etiology of eye-related disorders.

Information content differentiates enhancers from silencers in mouse photoreceptors

Enhancers and silencers often depend on the same transcription factors (TFs) and are conflated in genomic assays of TF binding or chromatin state. To identify sequence features that distinguish enhancers and silencers, we assayed massively parallel reporter libraries of genomic sequences targeted by the photoreceptor TF cone-rod homeobox (CRX) in mouse retinas. Both enhancers and silencers contain more TF motifs than inactive sequences, but relative to silencers, enhancers contain motifs from a more diverse collection of TFs. We developed a measure of information content that describes the number and diversity of motifs in a sequence and found that, while both enhancers and silencers depend on CRX motifs, enhancers have higher information content. The ability of information content to distinguish enhancers and silencers targeted by the same TF illustrates how motif context determines the activity of cis-regulatory sequences.

Different cell types are established by activating and repressing the activity of specific sets of genes, a process controlled by proteins called transcription factors. Transcription factors work by recognizing and binding short stretches of DNA in parts of the genome called cis-regulatory sequences. A cis-regulatory sequence that increases the activity of a gene when bound by transcription factors is called an enhancer, while a sequence that causes a decrease in gene activity is called a silencer.

To establish a cell type, a particular transcription factor will act on both enhancers and silencers that control the activity of different genes. For example, the transcription factor cone-rod homeobox (CRX) is critical for specifying different types of cells in the retina, and it acts on both enhancers and silencers. In rod photoreceptors, CRX activates rod genes by binding their enhancers, while repressing cone photoreceptor genes by binding their silencers. However, CRX always recognizes and binds to the same DNA sequence, known as its binding site, making it unclear why some cis-regulatory sequences bound to CRX act as silencers, while others act as enhancers.

Friedman et al. sought to understand how enhancers and silencers, both bound by CRX, can have different effects on the genes they control. Since both enhancers and silencers contain CRX binding sites, the difference between the two must lie in the sequence of the DNA surrounding these binding sites.

Using retinas that have been explanted from mice and kept alive in the laboratory, Friedman et al. tested the activity of thousands of CRX-binding sequences from the mouse genome. This showed that both enhancers and silencers have more copies of CRX-binding sites than sequences of the genome that are inactive. Additionally, the results revealed that enhancers have a diverse collection of binding sites for other transcription factors, while silencers do not. Friedman et al. developed a new metric they called information content, which captures the diverse combinations of different transcription binding sites that cis-regulatory sequences can have. Using this metric, Friedman et al. showed that it is possible to distinguish enhancers from silencers based on their information content.

It is critical to understand how the DNA sequences of cis-regulatory regions determine their activity, because mutations in these regions of the genome can cause disease. However, since every person has thousands of benign mutations in cis-regulatory sequences, it is a challenge to identify specific disease-causing mutations, which are relatively rare. One long-term goal of models of enhancers and silencers, such as Friedman et al.’s information content model, is to understand how mutations can affect cis-regulatory sequences, and, in some cases, lead to disease.

Partitioning of gene expression among zebrafish photoreceptor subtypes

Vertebrate photoreceptors are categorized into two broad classes, rods and cones, responsible for dim- and bright-light vision, respectively. While many molecular features that distinguish rods and cones are known, gene expression differences among cone subtypes remain poorly understood. Teleost fishes are renowned for the diversity of their photoreceptor systems. Here, we used single-cell RNA-seq to profile adult photoreceptors in zebrafish, a teleost. We found that in addition to the four canonical zebrafish cone types, there exist subpopulations of green and red cones (previously shown to be located in the ventral retina) that express red-shifted opsin paralogs (opn1mw4 or opn1lw1) as well as a unique combination of cone phototransduction genes. Furthermore, the expression of many paralogous phototransduction genes is partitioned among cone subtypes, analogous to the partitioning of the phototransduction paralogs between rods and cones seen across vertebrates. The partitioned cone-gene pairs arose via the teleost-specific whole-genome duplication or later clade-specific gene duplications. We also discovered that cone subtypes express distinct transcriptional regulators, including many factors not previously implicated in photoreceptor development or differentiation. Overall, our work suggests that partitioning of paralogous gene expression via the action of differentially expressed transcriptional regulators enables diversification of cone subtypes in teleosts.

A combinatorial cis-regulatory logic restricts color-sensing Rhodopsins to specific photoreceptor subsets in Drosophila

Color vision in Drosophila melanogaster is based on the expression of five different color-sensing Rhodopsin proteins in distinct subtypes of photoreceptor neurons. Promoter regions of less than 300 base pairs are sufficient to reproduce the unique, photoreceptor subtype-specific rhodopsin expression patterns. The underlying cis-regulatory logic remains poorly understood, but it has been proposed that the rhodopsin promoters have a bipartite structure: the distal promoter region directs the highly restricted expression in a specific photoreceptor subtype, while the proximal core promoter region provides general activation in all photoreceptors. Here, we investigate whether the rhodopsin promoters exhibit a strict specialization of their distal (subtype specificity) and proximal (general activation) promoter regions, or if both promoter regions contribute to generating the photoreceptor subtype-specific expression pattern. To distinguish between these two models, we analyze the expression patterns of a set of hybrid promoters that combine the distal promoter region of one rhodopsin with the proximal core promoter region of another rhodopsin. We find that the function of the proximal core promoter regions extends beyond providing general activation: these regions play a previously underappreciated role in generating the non-overlapping expression patterns of the different rhodopsins. Therefore, cis-regulatory motifs in both the distal and the proximal core promoter regions recruit transcription factors that generate the unique rhodopsin patterns in a combinatorial manner. We compare this combinatorial regulatory logic to the regulatory logic of olfactory receptor genes and discuss potential implications for the evolution of rhodopsins.

Each type of sensory receptor neuron in our body expresses a specific sensory receptor protein, which allows us to detect and discriminate a variety of environmental stimuli. The regulatory logic that controls this spatially precise and highly restricted expression of sensory receptor proteins remains poorly understood. As a model system, we study the mechanisms that control the expression of different color-sensing Rhodopsin proteins in distinct subtypes of Drosophila photoreceptors, which is the basis for color vision. Compact promoter regions of less than 300 base pairs are sufficient to reproduce the non-overlapping rhodopsin patterns. However, the regulatory logic that underlies the combination (sometimes called ‘grammar’) of the cis-regulatory motifs (sometimes called ‘vocabulary’) within the rhodopsin promoters remains poorly understood. Here, we find that specific combinations of cis-regulatory motifs in the distal and the proximal core promoter regions of each rhodopsin direct its unique expression pattern.

Ubiquitous Chromatin Modifiers in Congenital Retinal Diseases: Implications for Disease Modeling and Regenerative Medicine

Retinal congenital malformations known as microphthalmia, anophthalmia, and coloboma (MAC) are associated with alterations in genes encoding epigenetic proteins that modify chromatin. We review newly discovered functions of such chromatin modifiers in retinal development and discuss the role of epigenetics in MAC in humans and animal models. Further, we highlight how advances in epigenomic technologies provide foundational and regenerative medicine-related insights into blinding disorders. Combining knowledge of epigenetics and pluripotent stem cells (PSCs) is a promising avenue because epigenetic factors cooperate with eye field transcription factors (EFTFs) to direct PSC fate - a foundation for congenital retinal disease modeling and cell therapy.

Cell type- and stage-specific expression of Otx2 is regulated by multiple transcription factors and cis-regulatory modules in the retina

Transcription factors (TFs) are often used repeatedly during development and homeostasis to control distinct processes in the same and/or different cellular contexts. Considering the limited number of TFs in the genome and the tremendous number of events that need to be regulated, re-use of TFs is necessary. We analyzed how the expression of the homeobox TF, orthodenticle homeobox 2 (Otx2), is regulated in a cell type- and stage-specific manner during development in the mouse retina. We identified seven Otx2 cis-regulatory modules (CRMs), among which the O5, O7 and O9 CRMs mark three distinct cellular contexts of Otx2 expression. We discovered that Otx2, Crx and Sox2, which are well-known TFs regulating retinal development, bind to and activate the O5, O7 or O9 CRMs, respectively. The chromatin status of these three CRMs was found to be distinct in vivo in different retinal cell types and at different stages. We conclude that retinal cells use a cohort of TFs with different expression patterns and multiple CRMs with different chromatin configurations to regulate the expression of Otx2 precisely.

Mapping the cis-regulatory architecture of the human retina reveals noncoding genetic variation in disease

The interplay of transcription factors and cis-regulatory elements (CREs) orchestrates the dynamic and diverse genetic programs that assemble the human central nervous system (CNS) during development and maintain its function throughout life. Genetic variation within CREs plays a central role in phenotypic variation in complex traits including the risk of developing disease. We took advantage of the retina, a well-characterized region of the CNS known to be affected by pathogenic variants in CREs, to establish a roadmap for characterizing regulatory variation in the human CNS. This comprehensive analysis of tissue-specific regulatory elements, transcription factor binding, and gene expression programs in three regions of the human visual system (retina, macula, and retinal pigment epithelium/choroid) reveals features of regulatory element evolution that shape tissue-specific gene expression programs and defines regulatory elements with the potential to contribute to Mendelian and complex disorders of human vision.

Cis-regulatory basis of sister cell type divergence in the vertebrate retina

Multicellular organisms evolved via repeated functional divergence of transcriptionally related sister cell types, but the mechanisms underlying sister cell type divergence are not well understood. Here, we study a canonical pair of sister cell types, retinal photoreceptors and bipolar cells, to identify the key cis-regulatory features that distinguish them. By comparing open chromatin maps and transcriptomic profiles, we found that while photoreceptor and bipolar cells have divergent transcriptomes, they share remarkably similar cis-regulatory grammars, marked by enrichment of K50 homeodomain binding sites. However, cell class-specific enhancers are distinguished by enrichment of E-box motifs in bipolar cells, and Q50 homeodomain motifs in photoreceptors. We show that converting K50 motifs to Q50 motifs represses reporter expression in bipolar cells, while photoreceptor expression is maintained. These findings suggest that partitioning of Q50 motifs within cell type-specific cis-regulatory elements was a critical step in the evolutionary divergence of the bipolar transcriptome from that of photoreceptors.