Poster Session

Retinal degenerations such as age-related macular degeneration and retinitis pigmentosa result in debilitating vision impairment due to irreversible photoreceptor loss. One therapeutic strategy being developed is cell replacement therapy, in which stem cell-derived photoreceptors are transplanted into the retina to replace lost photoreceptors. The success of this approach hinges upon donor cell survival and the formation of appropriate synaptic connections with the residual inner retinal circuitry. Although studies in non-primate retinas have shown some promise for cell replacement therapy, we know very little about how primate retinal circuits respond to photoreceptor loss—particularly the foveal circuits dedicated to high acuity central vision—or whether stem cell-derived photoreceptors can integrate into a primate retina. Here, we used immunohistochemistry and confocal imaging to determine how deafferentation—from a laser ablation model of photoreceptor degeneration—impacts macular bipolar cell (BC) and horizontal cell (HC) survival and synaptic integrity. We also assessed whether transplanted human pluripotent stem cell-derived photoreceptor precursors (hPSC-PRPs) showed signs of synaptic integration into the damaged areas. Focal photoreceptor ablations, between ~1 to 20 degrees from the foveal center, were made with a femtosecond-pulsed laser (730nm, 27-33 J/cm2) with an adaptive optics scanning laser ophthalmoscope. Tissues were processed for histological assessment at time points ranging from < 1 week post-ablation to >3 months post-ablation. Cell type-specific markers and synaptic protein markers were used to determine the impact of photoreceptor loss on the survival and synaptic integrity of BCs and HCs. In deafferented areas, we found the densities of rod BCs, On-BCs, and Off-midget BCs to be unchanged compared to control areas at all time points. However, deafferentation led to the loss of dendritic glutamate receptors on On- and Off-BCs. Moreover, we found a loss of both H1 and H2 HCs in deafferented areas as early as 3 weeks post-ablation. HCs show signs of cell death at 4 and 8 days post ablation—time points when the photoreceptors have been cleared in the lesioned areas—suggesting HC loss occurs secondary to photoreceptor loss. To determine whether BCs can form synaptic connections with donor stem cell-derived photoreceptors, a scaffold seeded with hPSC-PRPs was transplanted into the subretinal space after photoreceptor ablation. Histological assessment showed both On and Off BC dendrites in close proximity to transplanted PRPs with expression of synaptic proteins at putative contact sites. The survival of BCs in deafferented areas and the extension of BC dendrites to transplanted stem cell-derived photoreceptors show promise for the development of cell replacement therapy for photoreceptor degenerations.

Genetic mutations in humans cause improper vascularization of the retina and lead to severe visual impairments manifested in diseases such as retinitis pigmentosa and familial exudative vitreoretinopathy. Feline leukemia virus subgroup C receptor (FLVCR2), also known as MFSD7c, is a member of the major facilitator superfamily (MFS) of membrane transporters. FLVCR2 is a blood brain barrier choline transporter that controls vascularization of the brain. Like the brain, the retina is primarily vascularized through sprouting angiogenesis, raising the hypothesis that FLVCR2 also controls retinal angiogenic growth. Based on mRNA expression and a new FLVCR2-HA (hemagglutinin) reporter in mice, we show that FLVCR2 expression correlates with that of other angiogenesis-related genes. We then performed endothelial-cell specific conditional deletion of FLVCR2 in mice (Cdh5CreER;FLVCR2flox) and demonstrate that FLVCR2 plays a key role in retinal angiogenesis, highlighting the key feature of decreased angiogenic sprouting upon knockout of the gene. Together, our report defines a novel role for FLVCR2 the eye. Our future work will focus on the exact role of FLVCR2, and its transport substrate choline, in developmental angiogenesis and related diseases of the eye.

Macular degeneration (MD) often leads to limitations in physical activities due to poor vision and fear of falling. Here, we present two complementary studies investigating mobility concerns in MD in the clinic and lab.

Study 1 assessed self-reported balance and mobility concerns in 100 patients (?50 years) with binocular maculopathy during their low vision consultation. Patients were asked general balance and mobility questions and a questionnaire on mobility concerns. 53% of patients reported balance complaints, and 39% attributed mobility limitations to vision loss. Severity of vision deficit correlated with higher scores on the questionnaire, particularly in terms of contrast sensitivity (?=-0.58, p<0.001). This awareness of mobility issues led to 32% of patients being referred for mobility rehabilitation.

Study 2 investigated mobility-related cognition and its relationship to movement patterns in 10 individuals with MD and 10 age-matched controls. Participants completed a gait-specific attentional questionnaire focusing on conscious movement processing (CMP) and ruminations. Head and trunk movements were recorded during the Timed Up and Go (TUG) functional mobility assessment. In the MD group, head acceleration and variability while walking increased with vision loss, suggesting poorer control (r=-0.73, p=0.026). Interestingly, higher rumination (?=-0.73, p=0.017) and CMP (?=-0.70, p=0.025) scores correlated with decreased head acceleration variability, suggesting improved control. Individuals with poorer vision also scored higher in CMP (?=0.73, p=0.027).

These findings suggest that greater conscious movement processing and ruminations lead to more careful walking, though this process may be disrupted with greater vision loss. This complex relationship between physical and cognitive mobility changes, and vision loss severity highlights the need to assess and address these issues in the clinic. Though greater mobility concerns are linked to poorer vision, the associations found with cognitive processes while walking suggest protective mechanisms in MD that may be leveraged via training.

Background: The lens epithelial cells (LECs) maintain the overall growth and homeostasis of the eye lens. In prior research, connexin 50 knockout (Cx50KO) mice display microphthalmia and smaller lenses with mild cataracts while mitogen-activated protein kinase kinase1 transgenic (Mek1TG) mice show macrophthalmia and enlarged lenses with severe cataracts, and Cx50KO/Mek1TG double mutant mice develop normal size lenses with severe cataracts.

Purpose: The study aims to investigate the heterogeneity of lens epithelium and to identify the transcriptional changes across different epithelial cell subpopulations using single-cell RNA sequencing (scRNA-seq). Moreover, gene candidates that play important roles in epithelial cell proliferation, differentiation and/or homeostasis are investigated.

Methods: Postnatal day 30 (P30) LECs were collected for scRNA-seq. 10x Genomics pipeline was used to align and preprocess raw scRNA-seq data. Seurat was utilized to process the data and integrate matrices for various genotypes. Monocle3 was employed to predict the pseudotime trajectory of lens epithelium. GO analysis was performed to explore the signaling pathways.

Results: The integrated scRNA-seq datasets reveal seven clusters among wild-type control, Cx50KO, Mek1TG and Cx50KO/Mek1KO, with variable distribution of cells in each cluster. The expression of crystallins, transcription regulatory genes, and metabolic genes varied across LEC subclusters and total cells of different genotypes, indicating different LEC heterogeneity and suggesting candidate genes involved in the regulation of LEC proliferation, differentiation and lens homeostasis.

Conclusions: scRNA-seq and GO analysis provide some molecular insights into the heterogeneity and transcriptomics of lens epithelium in different mutant lenses with different sizes. The scRNA-seq results reveal changes of LECs subpopulations associated with the distinct lens phenotypes of Cx50KO and Mek1TG. Identified specific genes, such as Tob1, Eid1 and Prox1, are likely involved in lens size control, while both Cx50 and Aqp1 are important for water homeostasis in lens epithelial cells.

Purpose: Identifying children at risk of myopia onset and evaluating the efficacy of myopia control requires normative data on ocular growth, particularly refractive error (RE) and axial length (AL), that reflect the US population. This study aims to generate preliminary growth curves for refractive error and ocular biometrics in children aged 3-8 years.

Methods: This cross-sectional study was conducted on site at Head Start centers and public elementary schools near the three participating colleges of optometry: Berkeley (CA), Boston (MA), and Houston (TX). Subjects were provided with comprehensive eye exams. Primary outcomes included cycloplegic RE and AL.

Results: To date, 169 children have participated across the three sites, with 48 from Berkeley (CA). AL and cycloplegic spherical equivalent (CSE) were evaluated using data from the right eyes, as there were no significant differences between the right and left eyes (p = 0.61, p = 0.63). The mean AL was 22.40 ± 0.78 mm (N = 42). Significant positive correlations were found between height and AL (r = 0.43, p = 0.005) and age and AL (r = 0.35, p = 0.02). The mean CSE was +1.09 ± 1.14 D (N = 42). A significant negative correlation was found between CSE and AL (r = -0.52, p < 0.001), with a weak negative correlation between age and CSE (r = -0.12, p = 0.44).

Conclusion: The positive correlations between height, age, and AL support findings that AL elongates as children grow. The negative correlation between CSE and AL supports the understanding that myopia involves eyeball elongation. The weak negative correlation between age and CSE aligns with research indicating a gradual decrease in refractive error from hyperopia toward emmetropia during childhood. Further longitudinal research with a larger sample size and additional sites is needed to establish the dynamics of refractive error development throughout childhood, underscoring the importance of normative ocular growth curves for early myopia

Amblyopia is a neurodevelopmental disorder characterized by reduced visual acuity (VA) in one or both eyes. Amblyopic observers perceive stimuli distorted (e.g., thinner strokes, blurred, indefinite edges). Additionally, stronger and more extensive crowding (i.e., impaired target identification due to neighboring stimuli) has been found in the central visual field of observers with amblyopia. Here, we investigated the appearance of isolated and crowded (by black bars) high contrast letters in amblyopia. Observers (16 controls, 14 persons with amblyopia) were presented with 5 letters monocularly (either to the dominant/fellow eye (DE/FE) or non-dominant/amblyopic eye (NDE/AE)). Letters were presented at 1.0 and 1.5 x observers’ VA threshold to the fovea for 500 ms. Observers were instructed to recreate target appearance by selecting squares on a 9x9 square-grid interface used to create the letter targets. Results showed that observers' responses were most similar to the target for stimuli 1.5 x the VA threshold (vs. 1.0 x threshold size), for isolated (vs. flanked) targets and for controls (vs. observers with amblyopia). No crowding differences were found between groups. The responses’ similarity to the target between the FE and AE of observers with amblyopia showed no differences. Surprisingly, response-target similarity was lower for controls’ DE, compared to the NDE. The captured appearance of the targets showed distortions in shape (e.g., straight line elements depicted as curves), fusion of elements (e.g., connecting the flanking bars), as well as truncation and extension of elements. The current results reveal characteristics of target appearance and highlight appearance differences between normal and amblyopic visual perception when controlling for VA differences.

Humans interact with the same objects in different ways based on the task at hand, and correspondingly, brain responses to the same objects can be modulated by task demands. Flexible perception is thought to rely on a network of areas termed the Multiple Demand Network (MDN), spanning parts of the parietal, frontal, and lateral occipital cortex. Here, we investigate the degree of functional specificity vs domain generality among regions in the MDN. To this end, we constructed an fMRI experiment in which human participants made judgments using different features of the same object. For each trial, participants either passively fixated or indicated whether a) medial axis structure or b) set of local shapes comprising the object or c) object texture was different from that of the previous object. Performing multiple tasks while presenting the same stimuli allows the study of task-specific responses independent of the stimuli. We created an encoding model with indicator variables for task conditions to derive weights associated with each task for every voxel. This model explained unique variance in a withheld portion of the dataset over and above stimulus-based models in regions of the frontal, parietal, and lateral occipital cortex. Principal Component Analysis of task weights across task-selective voxels revealed that less than half the variance was explained by a component indicative of activation across all tasks. Furthermore, contrast analysis of weights pointed to a significant number of voxels more tuned to specific tasks as opposed to uniform activation across all three tasks. Among these regions, we found that posterior LOTC ventral to hMT+ and dorsal to OFA responded more in the local shape task. A region just anterior to this responded more in the medial axis task. The texture task engaged more posterior regions. We also found spatially distinct activations in the IPS for different tasks. Although these varied across participants, some regions including lateral IPS were reliably activated by the medial axis task. These results point towards functional specificity of regions within the MDN, with different regions representing relatively more global and local aspects of object shape.

Echolocation, a strategy used by animals like bats and dolphins, is also employed by a growing number of blind and visually impaired (BVI) individuals, who use tongue clicks and listen to the resulting echoes to create mental representations of their surroundings. Despite its potential, the mechanisms underlying human echoacoustic perception are not well understood, limiting its broader application in the BVI community. To address this, our lab has developed "Robin," a wearable ultrasonic echolocation device that emits user-initiated ultrasonic signals, records the resulting echoes, and processes them into human-audible frequencies for playback. Robin interfaces with a web app that offers a wide range of customizable parameters, allowing users to tailor signals to their personal needs while providing researchers with a large, customizable parameter space for investigating the perceptual mechanisms of human echolocation. To assess the efficacy of this platform, we tested the perceptual discriminability of Robin-processed echoes from common furniture items. Preliminary results from sighted participants (no echolocation expertise) on a 2-AFC match-to-sample task showed above-chance performance in discriminating Robin-processed echoes reflected by four different objects (tested against every other object). These results indicate that the artificially processed echoacoustic signals produced by Robin encode perceptually relevant information that enables non-experts to successfully distinguish between different objects, based solely on their reflected echoes. Furthermore, visualizing the perceptual and stimulus spaces for this task using MDS revealed that perceptual discriminability reflected acoustic dissimilarity in the echo signatures of different objects, suggesting that observers were able to integrate increasing acoustic differences along multiple dimensions to improve performance. Ongoing work aims to leverage Robin’s customizable signal parameter space to identify specific acoustic cues best suited to different tasks and environmental features, paving the way for designing future echolocation-based aids for BVI individuals.

Here I describe a new binocular stereopsis phenomenon with wavy water: This phenomenon may be of considerable interest to those who are interested in binocular vision as it may be used as a new experimental paradigm to explore motion-elicited stereoscopic depth perception. If one happens to stand by a swimming pool with a horizontal-and-vertical grid (HVG) painted at its bottom, when there are smooth water waves in the pool and with binocular viewing, one will see an illusory stereopsis phenomenon: Either the horizontal (H) lines or the vertical (V) ones, or both of them, of the HVG may appear to “float” to the surface of the water; the most dramatic case is that the H/V lines are seen at the water surface while the V/H ones are seen at the bottom–i.e., they are perceived in two depth planes. This phenomenon can be easily reproduced with a small water container: Put a HVG pattern at its bottom, make smooth water waves, and the illusory depth effect can be readily observed. It can be observed with black lines over white background as well as with colored repetitive patterns–e.g., patterns of red/green lines over white background as well as white lines over red/green backgrounds can be used to induce this illusory depth effect. Furthermore, there is also a hysteresis (memory) effect: After all the water waves cease, the illusory depth may linger on for a few seconds. Finally, I suggest that this depth effect belongs to the same class of motion-elicited stereoscopic effects first reported by D.N. Lee (1971, Binocular stereopsis without spatial disparity) and K. Prazdny (1984, Stereopsis from kinetic and flicker edges)–with a salient difference that these researchers all used random-dot stereograms as visual stimuli whereas the phenomenon described here consists of natural and continuous motions. In motion-elicited stereoscopic depth effects, the determining factor is the difference between the two eyes’ motion signals instead of the spatial disparity of their monocular images.

Amblyopia is a neurodevelopmental disorder characterized by reduced visual acuity due to different visual input from the two eyes early in life. Previous research has shown that observers with amblyopia have stronger foveal crowding (i.e., worse performance with flanked targets) than neurotypical controls and perceive stimuli as distorted (e.g., straight line elements are perceived as jagged) and lower in contrast. In a previous study (Gomes Tomaz et al., 2023), observers were presented with isolated and flanked letter stimuli to the fovea of the dominant /fellow, or non-dominant/amblyopic eye. Observers recreated stimulus appearance by selecting squares on a 9x9 square-grid interface. Here, we trained a convolutional neural network (CNN) with the dataset of observers’ depictions to find the best models to predict whether a depiction was made by an observer with or without amblyopia, and to an isolated or flanked stimulus. The CNN architecture consisted of 3 convolutional, activation, and maximum pooling layers, followed by fully connected layers to classify for group (control, amblyopia) and crowding (isolated, flanked) variables. The model architecture was trained 10 times. The validation dataset was randomized, but balanced between all variables present in the dataset to ensure a representative sample of the appearance space. The most accurate models for classifying between groups (control, amblyopia) and crowding condition (isolated, flanked) have training accuracies of 74.4% and 97.2% and validation accuracies of 68.2% and 93.8%, respectively. The current results indicate that CNNs are useful in classifying images of target depictions. Lower accuracy for the group prediction model is potentially due to high variability in the appearance space that characterizes each group. In the future, accurate models that classify and predict stimulus appearance in amblyopia will be lesioned to further study amblyopic visual perception.

How much of our internal goals get expressed in the pattern of eye movements that we make? We created a dataset preparing for public sharing to answer this question. Participants (n=15) wore wearable eye-tracking glasses (Neon, Pupil Labs, Germany) and performed tasks in the same sensory environments with different goals. In a pair of tasks, after walking the same path, participants were either asked about the surroundings (walking-no memory task) or tested on a list of 20 words they had memorized before the walking (walking-memory task; the two tasks were counterbalanced across participants). In another pair, participants constructed structures (LEGO-building task; 5 min), and deconstructed the structures and sorted the LEGO blocks by color (LEGO-sorting task; until completed). We validated head movement (i.e., IMU data) in a separate session and calibrated eye positions three times during the experiment, and validated the data using the saccade peak velocity and amplitude. As a first step of our analysis, we looked into saccadic eye movement profiles. We classified saccades using the ReModNav algorithm (Dar, Wagner, & Hanke, 2021). The two walking tasks showed no difference in the behaviors (p’s > 0.05). Compared to the LEGO-sorting task, the LEGO-building task showed smaller saccade amplitude (M = 11.204°, STE: 0.399°; LEGO-sorting: M = 12.585°, STE: 0.422°; t = -2.3, p = 0.03) and longer intersaccadic intervals (M = 0.713 sec, STE = 0.036 sec, LEGO-sorting: M = 0.516 sec, STE = 0.014 sec; t = 4.08 p < 0.01) compared to the sorting task. Our result suggests that internal goals can be distinguished from the patterns of eye movement even in the same sensory environments. It also opens up the possibility that computational modeling techniques may be used to determine internal goals participants had from the eye movement patterns.

We are developing smartphone eye-tracking for precision brain health and early detection of neurodegenerative diseases. Our goal is to minimize user burden and to optimize engagement for remote monitoring. The performance of eye-tracking on smartphones has progressed to the point where it rivals specialized hardware. We will verify smartphone eye-tracking performance and benchmark with specialized eye-trackers. With our MVP eye-tracking app we have collected eye-tracking data from 115 participants and correlated it with a clinically validated, self-reported fatigue survey, the brief fatigue inventory (BFI). We will complete a proof-of-concept for clinical validation using smartphone measurements of eye-tracking speed, accuracy and precision to enable longitudinal monitoring of cognitive health and demonstrate utility for objective remote monitoring of progression, recovery and effectiveness for potential treatments.

Amblyopia is associated with impaired visual acuity, abnormal processes in the primary visual cortex (V1), and reduced coordination of activity in binocular neurons. Recent studies have demonstrated that these abnormalities extend beyond V1 with potential impacts on higher order processing and in oculomotor areas that affect visually-guided behaviors. While large fixational instabilities found in amblyopic eyes are highly correlated with worsening acuity in these patients, no such correlation was found in a neurotypical population (Raveendran et al., 2019). It is unclear whether the poor acuity causes fixational instability or vice versa or if the two are correlated as a byproduct of abnormal processes in V1. Therefore, we investigated whether visual acuity is truly a limiting factor in oculomotor and visually-guided behavior. We simulated reduced visual acuity, as observed in the amblyopic eye, in the non-dominant eyes (NDEs) of corrected-to-normal neurotypical observers by applying convex lenses to induce retinal-defocus while the observers viewed naturalistic stimuli – a filtered version of “Where’s Waldo” – while performing a visual search task. We tested a set of logMAR acuity differences (0.2, 0.4, 0.6, and 0.8) as observed in amblyopic patients and compared the oculomotor behavior of the normal eye with lens-induced acuity loss in the following conditions: monocular NDE and binocular viewing. We observed lens induced acuity loss had a significant impact on the observer’s drift magnitude and fixational instabilities compared to the no-blur condition in the NDE; however, further reductions in visual acuity did not have any significant impact. Binocular viewing significantly improved fixational instability and reduced reaction time when compared with monocular viewing in both amblyopes and neurotypicals; this effect persisted as a function of retinal-defocus induced reduction in visual acuity. Our results demonstrate that the reduced visual acuity of patients with amblyopia cannot be fully accounted for by abnormal fixational eye movements.

Retinal degeneration is often driven by mutations in the outer retinal layer, primarily affecting the retinal pigment epithelium (RPE) and photoreceptor cells. Although Adeno-Associated Virus (AAV) vectors show promise as a therapeutic approach, the conventional subretinal delivery method is invasive and requires multiple reinjections over time, posing significant challenges for widespread application. This study aims to identify AAV variants capable of efficiently transducing RPE cells through the less invasive and more sustainable intravitreal injection method. We engineered two AAV Serotype 2 libraries, 7mer and LoopSwap453, designed to modify the capsid's outer surface structure, which influences its binding capabilities to different retinal components. These libraries consist of 1.28 × 10? and 4.10 × 10¹? variants, respectively. To date, AAV2.7mer has been intravitreally injected into three species, followed by posterior eye cup harvesting. DNA extraction and Next-Generation Sequencing (NGS) were used to quantify the prevalence of each variant capable of transducing the outer retinal layers. Through a streamlined selection process, we identified top-performing AAV variants, which were subsequently repackaged with a ubiquitous promoter and GFP transgene to assess transduction efficiency. Our data indicate that upon repackaging with GFP, the top third variant from the initial AAV2.7mer screening, which we called Variant 3 (Var3) , proved to effectively transduce RPE cells. Currently, AAV2.LS453 trials have identified top-performing variants in mice and are awaiting repackaging. We anticipate similar results with AAV2.LS453 variants optimized for selective and efficient RPE transduction.