A single-cell transcriptome atlas of the adult human retina.

The retina is a specialized neural tissue that senses light and initiates image processing. Although the functional organization of specific retina cells has been well studied, the molecular profile of many cell types remains unclear in humans. To comprehensively profile the human retina, we performed single-cell RNA sequencing on 20,009 cells from three donors and compiled a reference transcriptome atlas. Using unsupervised clustering analysis, we identified 18 transcriptionally distinct cell populations representing all known neural retinal cells: rod photoreceptors, cone photoreceptors, Müller glia, bipolar cells, amacrine cells, retinal ganglion cells, horizontal cells, astrocytes, and microglia. Our data captured molecular profiles for healthy and putative early degenerating rod photoreceptors, and revealed the loss of MALAT1 expression with longer post-mortem time, which potentially suggested a novel role of MALAT1 in rod photoreceptor degeneration. We have demonstrated the use of this retina transcriptome atlas to benchmark pluripotent stem cell-derived cone photoreceptors and an adult Müller glia cell line. This work provides an important reference with unprecedented insights into the transcriptional landscape of human retinal cells, which is fundamental to understanding retinal biology and disease.

Photoreceptor physiology and evolution: cellular and molecular basis of rod and cone phototransduction.

The detection of light in the vertebrate retina utilizes a duplex system of closely related rod and cone photoreceptors: cones respond extremely rapidly, and operate at 'photopic' levels of illumination, from moonlight upwards; rods respond much more slowly, thereby obtaining greater sensitivity, and function effectively only at 'scotopic' levels of moonlight and lower. Rods and cones employ distinct isoforms of many of the proteins in the phototransduction cascade, and they thereby represent a unique evolutionary system, whereby the same process (the detection of light) uses a distinct set of genes in two classes of cell. The molecular mechanisms of phototransduction activation are described, and the classical quantitative predictions for the onset phase of the electrical response to light are developed. Recent work predicting the recovery phase of the rod's response to intense flashes is then presented, that provides an accurate account of the time that the response spends in saturation. Importantly, this also provides a new estimate for the rate at which a single rhodopsin activates molecules of the G-protein, transducin, that is substantially higher than other estimates in the literature. Finally, the evolutionary origin of the phototransduction proteins in rods and cones is examined, and it is shown that most of the rod/cone differences were established at the first of the two rounds of whole-genome duplication more than 500 million years ago.

Visual Opsin Diversity in Sharks and Rays.

The diversity of color vision systems found in extant vertebrates suggests that different evolutionary selection pressures have driven specializations in photoreceptor complement and visual pigment spectral tuning appropriate for an animal's behavior, habitat, and life history. Aquatic vertebrates in particular show high variability in chromatic vision and have become important models for understanding the role of color vision in prey detection, predator avoidance, and social interactions. In this study, we examined the capacity for chromatic vision in elasmobranch fishes, a group that have received relatively little attention to date. We used microspectrophotometry to measure the spectral absorbance of the visual pigments in the outer segments of individual photoreceptors from several ray and shark species, and we sequenced the opsin mRNAs obtained from the retinas of the same species, as well as from additional elasmobranch species. We reveal the phylogenetically widespread occurrence of dichromatic color vision in rays based on two cone opsins, RH2 and LWS. We also confirm that all shark species studied to date appear to be cone monochromats but report that in different species the single cone opsin may be of either the LWS or the RH2 class. From this, we infer that cone monochromacy in sharks has evolved independently on multiple occasions. Together with earlier discoveries in secondarily aquatic marine mammals, this suggests that cone-based color vision may be of little use for large marine predators, such as sharks, pinnipeds, and cetaceans.

Evolution of the vertebrate phototransduction cascade activation steps.

We examine the molecular phylogeny of the proteins underlying the activation steps of vertebrate phototransduction, for both agnathan and jawed vertebrate taxa. We expand the number of taxa analysed and we update the alignment and tree building methodology from a previous analysis. For each of the four primary components (the G-protein transducin alpha subunit, GαT, the cyclic GMP phosphodiesterase, PDE6, and the alpha and beta subunits of the cGMP-gated ion channel, CNGC), the phylogenies appear consistent with expansion from an ancestral proto-vertebrate cascade during two rounds of whole-genome duplication followed by divergence of the agnathan and jawed vertebrate lineages. In each case, we consider possible scenarios for the underlying gene duplications and losses, and we apply relevant constraints to the tree construction. From tests of the topology of the resulting trees, we obtain a scenario for the expansion of each component during 2R that accurately fits the observations. Similar analysis of the visual opsins indicates that the only expansion to have occurred during 2R was the formation of Rh1 and Rh2. Finally, we propose a hypothetical scenario for the conversion of an ancestral chordate cascade into the proto-vertebrate phototransduction cascade, prior to whole-genome duplication. Together, our models provide a plausible account for the origin and expansion of the vertebrate phototransduction cascade.

Evolution of Vertebrate Phototransduction: Cascade Activation.

We applied high-throughput sequencing to eye tissue from several species of basal vertebrates (a hagfish, two species of lamprey, and five species of gnathostome fish), and we analyzed the mRNA sequences for the proteins underlying activation of the phototransduction cascade. The molecular phylogenies that we constructed from these sequences are consistent with the 2R WGD model of two rounds of whole genome duplication. Our analysis suggests that agnathans retain an additional representative (that has been lost in gnathostomes) in each of the gene families we studied; the evidence is strong for the G-protein α subunit (GNAT) and the cGMP phosphodiesterase (PDE6), and indicative for the cyclic nucleotide-gated channels (CNGA and CNGB). Two of the species (the hagfish Eptatretus cirrhatus and the lamprey Mordacia mordax) possess only a single class of photoreceptor, simplifying deductions about the composition of cascade protein isoforms utilized in their photoreceptors. For the other lamprey, Geotria australis, analysis of the ratios of transcript levels in downstream and upstream migrant animals permits tentative conclusions to be drawn about the isoforms used in four of the five spectral classes of photoreceptor. Overall, our results suggest that agnathan rod-like photoreceptors utilize the same GNAT1 as gnathostomes, together with a homodimeric PDE6 that may be agnathan-specific, whereas agnathan cone-like photoreceptors utilize a GNAT that may be agnathan-specific, together with the same PDE6C as gnathostomes. These findings help elucidate the evolution of the vertebrate phototransduction cascade from an ancestral chordate phototransduction cascade that existed prior to the vertebrate radiation.

Quantitative modeling of the molecular steps underlying shut-off of rhodopsin activity in rod phototransduction.

PURPOSE: To examine the predictions of alternative models for the stochastic shut-off of activated rhodopsin (R*) and their implications for the interpretation of experimentally recorded single-photon responses (SPRs) in mammalian rods. THEORY: We analyze the transitions that an activated R* molecule undergoes as a result of successive phosphorylation steps and arrestin binding. We consider certain simplifying cases for the relative magnitudes of the reaction rate constants and derive the probability distributions for the time to arrestin binding. In addition to the conventional model in which R* catalytic activity declines in a graded manner with successive phosphorylations, we analyze two cases in which the activity is assumed to occur not via multiple small steps upon each phosphorylation but via a single large step. We refer to these latter two cases as the binary R* shut-off and three-state R* shut-off models. METHODS: We simulate R*'s stochastic reactions numerically for the three models. In the simplifying cases for the ratio of rate constants in the binary and three-state models, we show that the probability distribution of the time to arrestin binding is accurately predicted. To simulate SPRs, we then integrate the differential equations for the downstream reactions using a standard model of the rod outer segment that includes longitudinal diffusion of cGMP and Ca(2+). RESULTS: Our simulations of SPRs in the conventional model of graded shut-off of R* conform closely to the simulations in a recent study. However, the gain factor required to account for the observed mean SPR amplitude is higher than can be accounted for from biochemical experiments. In addition, a substantial minority of the simulated SPRs exhibit features that have not been reported in published experiments. Our simulations of SPRs using the model of binary R* shut-off appear to conform closely to experimental results for wild type (WT) mouse rods, and the required gain factor conforms to biochemical expectations. However, for the arrestin knockout (Arr(-/-)) phenotype, the predictions deviated from experimental findings and led us to invoke a low-activity state that R* enters before arrestin binding. Our simulations of this three-state R* shut-off model are very similar to those of the binary model in the WT case but are preferred because they appear to accurately predict the mean SPRs for four mutant phenotypes, Arr(+/-), Arr(-/-), GRK1(+/-), and GRK1(-/-), in addition to the WT phenotype. When we additionally treated the formation and shut-off of activated phosphodiesterase (E*) as stochastic, the simulated SPRs appeared even more similar to real SPRs, and there was very little change in the ensemble mean and standard deviation or in the amplitude distribution. CONCLUSIONS: We conclude that the conventional model of graded reduction in R* activity through successive phosphorylation steps appears to be inconsistent with experimental results. Instead, we find that two variants of a model in which R* activity initially remains high and then declines abruptly after several phosphorylation steps appears capable of providing a better description of experimentally measured SPRs.

Rhodopsin, light-sensor of vision.

The light sensor of vertebrate scotopic (low-light) vision, rhodopsin, is a G-protein-coupled receptor comprising a polypeptide chain with bound chromophore, 11-cis-retinal, that exhibits remarkable physicochemical properties. This photopigment is extremely stable in the dark, yet its chromophore isomerises upon photon absorption with 70% efficiency, enabling the activation of its G-protein, transducin, with high efficiency. Rhodopsin's photochemical and biochemical activities occur over very different time-scales: the energy of retinaldehyde's excited state is stored in <1 ps in retinal-protein interactions, but it takes milliseconds for the catalytically active state to form, and many tens of minutes for the resting state to be restored. In this review, we describe the properties of rhodopsin and its role in rod phototransduction. We first introduce rhodopsin's gross structural features, its evolution, and the basic mechanisms of its activation. We then discuss light absorption and spectral sensitivity, photoreceptor electrical responses that result from the activity of individual rhodopsin molecules, and recovery of rhodopsin and the visual system from intense bleaching exposures. We then provide a detailed examination of rhodopsin's molecular structure and function, first in its dark state, and then in the active Meta states that govern its interactions with transducin, rhodopsin kinase and arrestin. While it is clear that rhodopsin's molecular properties are exquisitely honed for phototransduction, from starlight to dawn/dusk intensity levels, our understanding of how its molecular interactions determine the properties of scotopic vision remains incomplete. We describe potential future directions of research, and outline several major problems that remain to be solved.

Evolution of the vertebrate eye: opsins, photoreceptors, retina and eye cup.

Charles Darwin appreciated the conceptual difficulty in accepting that an organ as wonderful as the vertebrate eye could have evolved through natural selection. He reasoned that if appropriate gradations could be found that were useful to the animal and were inherited, then the apparent difficulty would be overcome. Here, we review a wide range of findings that capture glimpses of the gradations that appear to have occurred during eye evolution, and provide a scenario for the unseen steps that have led to the emergence of the vertebrate eye.

Slowed recovery of human photopic ERG a-wave amplitude following intense bleaches: a slowing of cone pigment regeneration?

We have used the post-bleach recovery of the ERG a-wave to estimate the time-course of regeneration of cone pigment, following bleaching exposures far stronger than in a previous study. We recorded the photopic electroretinogram a-wave from two subjects, in response to dim red flashes delivered following 1-min exposures to intensities ranging from 1.1 × 10(4) to 1.3 × 10(5) photopic cd m(-2). The measured response amplitudes were "linearized" to derive estimates of pigment level. These estimated pigment levels were found to increase at an initially linear rate, consistent with a "rate-limited" model of photopigment regeneration. The extracted time-course was similar to that previously reported in densitometric studies of cone pigment regeneration after similarly intense exposures. On the other hand, the rate of regeneration was slower than measured in the same subjects following less intense bleaches. These results are consistent with the notion that cone pigment regeneration is slowed following very strong bleaching exposures, possibly as a result of depletion of a pool of retinoid.

Modelling the initial phase of the human rod photoreceptor response to the onset of steady illumination.

The initial time course of the change in photoreceptor outer segment membrane conductance in response to light flashes has been modelled using biochemical analysis of phototransduction, and the model has been successfully applied to a range of in vitro recordings and has also been shown to provide a good fit to the leading edge of the electroretinogram a-wave recorded in vivo. We investigated whether a simple modification of the model's equation would predict responses to the onset of steady illumination and tested this against electroretinogram recordings. Scotopic electroretinograms were recorded from three normal human subjects, using conductive fibre electrodes, in response to light flashes (0.30-740 scotopic cd m(-2) s) and to the onset of steady light (11-1,900 scotopic cd m(-2)). Subjects' pupils were dilated pharmacologically. The standard form of the model was applied to flash responses, as in previous studies, to obtain values for the three parameters: maximal response amplitude r (max), sensitivity S and effective delay time t (eff). A new "step response" equation was derived, and this equation provided a good fit to rod responses to steps of light using the same parameter values as for the flash responses. The results support the applicability of the model to the leading edge of electroretinogram responses: in each subject, the model could be used to fit responses both to flashes of light and to the onset of backgrounds with a single set of parameter values.

Human scotopic dark adaptation: Comparison of recoveries of psychophysical threshold and ERG b-wave sensitivity.

We have compared the time course of dark adaptation of the human scotopic visual system, measured psychophysically and from the b-wave of the electroretinogram (ERG), for bleaches ranging from a few percent to near total. We also measured light adaptation, in order to apply a "Crawford transformation" to convert the raw measurements of dark adaptation into equivalent background intensities. For both the "psychophysical threshold equivalent" intensity and the "ERG b-wave sensitivity equivalent" intensity, the equivalent background declined over much of its range with an "S2" component, though with somewhat different slopes of -0.36 (psychophysical) and -0.22 (ERG) log(10) unit min(-1), respectively. In addition, the magnitude of the equivalent background was approximately 1 log(10) unit lower in the psychophysical experiments than in the ERG experiments. Despite these differences, the two approaches extract a common time course for the decline in level of free opsin following moderately large bleaches. We conclude that the recovery of psychophysical scotopic visual threshold over the S2 region reflects events that are present by the stage of the first synapse of rod vision, stemming ultimately from the presence of unregenerated opsin in the rod outer segments.

Analysis of Paralogons, Origin of the Vertebrate Karyotype, and Ancient Chromosomes Retained in Extant Species.

A manually curated set of ohnolog families has been assembled, for seven species of bony vertebrates, that includes 255 four-member families and 631 three-member families, encompassing over 2,900 ohnologs. Across species, the patterns of chromosomes upon which the ohnologs reside fall into 17 distinct categories. These 17 paralogons reflect the 17 ancestral chromosomes that existed in our chordate ancestor immediately prior to the two rounds of whole-genome duplication (2R-WGD) that occurred around 600 Ma. Within each paralogon, it has now been possible to assign those pairs of ohnologs that diverged from each other at the first round of duplication, through analysis of the molecular phylogeny of four-member families. Comparison with another recent analysis has identified four apparently incorrect assignments of pairings following 2R, along with several omissions, in that study. By comparison of the patterns between paralogons, it has also been possible to identify nine chromosomal fusions that occurred between 1R and 2R, and three chromosomal fusions that occurred after 2R, that generated an ancestral bony-vertebrate karyotype comprising 47 chromosomes. At least 27 of those ancestral bony-vertebrate chromosomes can, in some extant species, be shown not to have undergone any fusion or fission events. Such chromosomes are here termed "archeochromosomes," and have each survived essentially unchanged in their content of genes for some 400 Myr. Their utility lies in their potential for tracking the various fusion and fission events that have occurred in different lineages throughout the expansion of bony vertebrates.

Evolution of the genes mediating phototransduction in rod and cone photoreceptors.

This paper reviews current knowledge of the evolution of the multiple genes encoding proteins that mediate the process of phototransduction in rod and cone photoreceptors of vertebrates. The approach primarily involves molecular phylogenetic analysis of phototransduction protein sequences, combined with analysis of the syntenic arrangement of the genes. At least 35 of these phototransduction genes appear to reside on no more than five paralogons - paralogous regions that each arose from a common ancestral region. Furthermore, it appears that such paralogs arose through quadruplication during the two rounds of genome duplication (2R WGD) that occurred in a chordate ancestor prior to the vertebrate radiation, probably around 600 millions years ago. For several components of the phototransduction cascade, it is shown that distinct isoforms already existed prior to WGD, with the likely implication that separate classes of scotopic and photopic photoreceptor cells had already evolved by that stage. The subsequent quadruplication of the entire genome then permitted the refinement of multiple distinct protein isoforms in rods and cones. A unified picture of the likely pattern and approximate timing of all the important gene duplications is synthesised, and the implications for our understanding of the evolution of rod and cone phototransduction are presented.

A quantitative account of mammalian rod phototransduction with PDE6 dimeric activation: responses to bright flashes.

We develop an improved quantitative model of mammalian rod phototransduction, and we apply it to the prediction of responses to bright flashes of light. We take account of the recently characterized dimeric nature of PDE6 activation, where the configuration of primary importance has two transducin molecules bound. We simulate the stochastic nature of the activation and shut-off reactions to generate the predicted kinetics of the active molecular species on the disc membrane surfaces, and then we integrate the differential equations for the downstream cytoplasmic reactions to obtain the predicted electrical responses. The simulated responses recover the qualitative form of bright-flash response families recorded from mammalian rod photoreceptors. Furthermore, they provide an accurate description of the relationship between the time spent in saturation and flash intensity, predicting the transition between first and second 'dominant time constants' to occur at an intensity around 5000 isomerizations per flash, when the rate of transducin activation is taken to be 1250 transducins s-1 per activated rhodopsin. This rate is consistent with estimates from light-scattering experiments, but is around fourfold higher than has typically been assumed in other studies. We conclude that our model and parameters provide a compelling description of rod photoreceptor bright-flash responses.

Functional Imaging of the Outer Retinal Complex using High Fidelity Imaging Retinal Densitometry.

We describe a new technique, high fidelity Imaging Retinal Densitometry (IRD), which probes the functional integrity of the outer retinal complex. We demonstrate the ability of the technique to map visual pigment optical density and synthesis rates in eyes with and without macular disease. A multispectral retinal imaging device obtained precise measurements of retinal reflectance over space and time. Data obtained from healthy controls and 5 patients with intermediate AMD, before and after photopigment bleaching, were used to quantify visual pigment metrics. Heat maps were plotted to summarise the topography of rod and cone pigment kinetics and descriptive statistics conducted to highlight differences between those with and without AMD. Rod and cone visual pigment synthesis rates in those with AMD (v = 0.043 SD 0.019 min-1 and v = 0.119 SD 0.046 min-1, respectively) were approximately half those observed in healthy controls (v = 0.079 SD 0.024 min-1 for rods and v = 0.206 SD 0.069 min-1 for cones). By mapping visual pigment kinetics across the central retina, high fidelity IRD provides a unique insight into outer retinal complex function. This new technique will improve the phenotypic characterisation, diagnosis and treatment monitoring of various ocular pathologies, including AMD.

Evolution of the calcium feedback steps of vertebrate phototransduction.

We examined the genes encoding the proteins that mediate the Ca-feedback regulatory system in vertebrate rod and cone phototransduction. These proteins comprise four families: recoverin/visinin, the guanylyl cyclase activating proteins (GCAPs), the guanylyl cyclases (GCs) and the sodium/calcium-potassium exchangers (NCKXs). We identified a paralogon containing at least 36 phototransduction genes from at least fourteen families, including all four of the families involved in the Ca-feedback loop (recoverin/visinin, GCAPs, GCs and NCKXs). By combining analyses of gene synteny with analyses of the molecular phylogeny for each of these four families of genes for Ca-feedback regulation, we have established the likely pattern of gene duplications and losses underlying the expansion of isoforms, both before and during the two rounds of whole-genome duplication (2R WGD) that occurred in early vertebrate evolution. Furthermore, by combining our results with earlier evidence on the timing of duplication of the visual G-protein receptor kinase genes, we propose that specialization of proto-vertebrate photoreceptor cells for operation at high and low light intensities preceded the emergence of rhodopsin, which occurred during 2R WGD.

Implications of dimeric activation of PDE6 for rod phototransduction.

We examine the implications of a recent report providing evidence that two transducins must bind to the rod phosphodiesterase to elicit significant hydrolytic activity. To predict the rod photoreceptor's electrical response, we use numerical simulation of the two-dimensional diffusional contact of interacting molecules at the surface of the disc membrane, and then we use the simulated PDE activity as the driving function for the downstream reaction cascade. The results account for a number of aspects of rod phototransduction that have previously been puzzling. For example, they explain the existence of a greater initial delay in rods than in cones. Furthermore, our analysis suggests that the 'continuous' noise recorded in rods in darkness is likely to arise from spontaneous activation of individual molecules of PDE at a rate of a few tens per second per rod, probably as a consequence of spontaneous activation of transducins at a rate of thousands per second per rod. Hence, the dimeric activation of PDE in rods provides immunity against spontaneous transducin activation, thereby reducing the continuous noise. Our analysis also provides a coherent quantitative explanation of the amplification underlying the single photon response. Overall, numerical analysis of the dimeric activation of PDE places rod phototransduction in a new light.

Evolution of the shut-off steps of vertebrate phototransduction.

Different isoforms of the genes involved in phototransduction are expressed in vertebrate rod and cone photoreceptors, providing a unique example of parallel evolution via gene duplication. In this study, we determine the molecular phylogeny of the proteins underlying the shut-off steps of phototransduction in the agnathan and jawed vertebrate lineages. For the G-protein receptor kinases (GRKs), the GRK1 and GRK7 divisions arose prior to the divergence of tunicates, with further expansion during the two rounds of whole-genome duplication (2R); subsequently, jawed and agnathan vertebrates retained different subsets of three isoforms of GRK. For the arrestins, gene expansion occurred during 2R. Importantly, both for GRKs and arrestins, the respective rod isoforms did not emerge until the second round of 2R, just prior to the separation of jawed and agnathan vertebrates. For the triplet of proteins mediating shut-off of the G-protein transducin, RGS9 diverged from RGS11, probably at the second round of 2R, whereas Gß5 and R9AP appear not to have undergone 2R expansion. Overall, our analysis provides a description of the duplications and losses of phototransduction shut-off genes that occurred during the transition from a chordate with only cone-like photoreceptors to an ancestral vertebrate with both cone- and rod-like photoreceptors.

Topographic Rod Recovery Profiles after a Prolonged Dark Adaptation in Subjects with Reticular Pseudodrusen.

PURPOSE: Although rod function is known to be severely impaired in eyes with reticular pseudodrusen (RPD), it remains unknown whether this impairment is associated with a total loss of rod function or merely a delay in rod recovery. The purpose of the study was to determine rod functional recovery profiles after prolonged dark adaptation (DA) in eyes with age-related macular degeneration (AMD) and RPD. DESIGN: A cross-sectional, case-series study. PARTICIPANTS: Subjects with AMD and RPD. METHODS: Retinal sensitivity was assessed simultaneously at 14 retinal locations within the central 12° in the study eye of each subject after the eye received approximately 20% bleach. Recovery of retinal sensitivity was monitored at regular intervals up to 30 minutes after bleach. If retinal sensitivity of all test points had not recovered to the rod criterion level (-3.0 log units of stimulus intensity) after 30 minutes of DA, monitoring recovery of retinal sensitivity was extended up to 24 hours of DA. MAIN OUTCOME MEASURES: Rod functional recovery profile at each test point. RESULTS: Six AMD cases with RPD were included, aged 69 to 79 years, and visual acuity ranged from 20/20 to 20/25. All cases had a delay in rod functional recovery at many retinal locations, with test points within the central 6° most affected. The recovery rate was variable between retinal loci and between subjects, although RPD were present at all test locations. In 5 cases with stage 3 RPD, rod function recovered at all tested locations, but many locations took hours to do so. The case with stage 4 RPD had locations that failed to recover even after 24 hours of DA. CONCLUSIONS: Eyes with AMD and RPD are associated with severe rod dysfunction throughout the macula; however, rod function does recover in most cases after an extended DA time. These findings suggest that the delay in rod recovery in eyes with RPD is, in most cases, associated with the impairment rather than the total loss of rod photoreceptor function. Stage 4 RPD may represent a point at which some rod photoreceptors are nonfunctional.