Neuroprotective strategies for retinal disease

ABSTRACT Diseases that affect the eye, including photoreceptor degeneration, diabetic retinopathy, and glaucoma, affect 11.8 million people in the US, resulting in vision loss and blindness. Loss of sight affects patient quality of life and puts an economic burden both on individuals and the greater healthcare system. Despite the urgent need for treatments, few effective options currently exist in the clinic. Here, we review research on promising neuroprotective strategies that promote neuronal survival with the potential to protect against vision loss and retinal cell death. Due to the large number of neuroprotective strategies, we restricted our review to approaches that we had direct experience with in the laboratory. We focus on drugs that target survival pathways, including bile acids like UDCA and TUDCA, steroid hormones like progesterone, therapies that target retinal dopamine, and neurotrophic factors. In addition, we review rehabilitative methods that increase endogenous repair mechanisms, including exercise and electrical stimulation therapies. For each approach, we provide background on the neuroprotective strategy, including history of use in other diseases; describe potential mechanisms of action; review the body of research performed in the retina thus far, both in animals and in humans; and discuss considerations when translating each treatment to the clinic and to the retina, including which therapies show the most promise for each retinal disease. Despite the high incidence of retinal diseases and the complexity of mechanisms involved, several promising neuroprotective treatments provide hope to prevent blindness. We discuss attractive candidates here with the goal of furthering retinal research in critical areas to rapidly translate neuroprotective strategies into the clinic. HIGHLIGHTSNeuroprotective strategies promote survival of retinal neurons.Preserving functional vision supports independence and quality of life.We present six strategies that preserve retinal neurons across multiple diseases.Translation of TUDCA and progesterone can leverage several ongoing clinical trials.Dopamine‐related therapies and exercise are new strategies to prevent vision loss.


Vision loss and neuroprotective strategies
For this review, we are specifically focused on neuroprotective interventions that slow vision loss caused by retinal disease.Neuroprotective strategies encompass a broad range of therapeutics that increase survival of neurons by preserving neuronal structure and function.There is a great need for neuroprotective therapies for retinal disease since nearly all are progressive and lead to vision loss and eventual blindness.Fear of blindness and vision loss is ranked among the top health concerns for Americans, similar to cancer and Alzheimer's disease (National Public Opinion Poll).Reduced visual acuity and blindness is associated with diminished quality of life measures due to limited social interactions and independence, loss of employment, and depression (Coffey et al., 2014;Cumberland and Rahi, 2016;Khorrami-Nejad et al., 2016;Papadopoulos et al., 2013;Senra et al., 2013).This is especially true for retinal diseases that compromise the macular region, thereby affecting the ability to read and recognize faces (Taylor et al., 2016).Thus, a great need exists to treat blinding retinal disease with neuroprotective agents that could slow or halt the progression of vision loss.Additionally, rehabilitative strategies can provide benefit by maintaining or improving visual function.Successful neuroprotective strategies could provide months or years of useful vision that would allow an individual to continue to be gainfully employed and maintain their independence and quality of life.

Primary retinal diseases causing vision loss
The retinal diseases addressed in this review can be classified into three major categories: photoreceptor degenerations, diabetic retinopathy, and retinal ganglion cell disease.These categories are broad, but provide a framework in which the final stages of pathology are often the same-oxidative stress and apoptosis.Importantly, neuroprotective strategies often target these broad final mechanisms of disease, and can be used across multiple retinal diseases.

Photoreceptor degenerations
The first step in converting light energy into a visual signal occurs in the photoreceptors.These cells are essential for detecting light of various wavelengths across a broad range of luminance.Thus, healthy photoreceptors are critical to vision.The loss of photoreceptors quickly leads to visual dysfunction, and the loss of afferent signals to second and third order retinal neurons eventually leads to remodel of the retina in late stage disease (Cuenca et al., 2014;Jones et al., 2003Jones et al., , 2016;;Marc et al., 2003).Two main retinal diseases caused by loss of photoreceptors are retinitis pigmentosa (RP) and age-related macular degeneration (AMD).
RP is due to mutations or other defects in photoreceptor or retinal pigment epithelial cells that lead to photoreceptor death.RP is a family of inherited retinal diseases that have been associated with over 250 different gene mutations (Daiger et al., 2013; http://www.sph.uth.tmc.edu/RetNet/), including some genes located in the choroid (Kennan et al., 2005).RP affects 1 in 4000 people in the US and worldwide (https://nei.nih.gov/health/pigmentosa/pigmentosa_facts).While this level of prevalence would be defined as a rare or orphan disease, RP accounts for the most common cause of complete blindness and is the focus of much research (Parmeggiani, 2011)(Fig.1).The early phenotype of RP is difficultly seeing at night and loss of peripheral vision caused by apoptosis of rod photoreceptors.As the rods progressively die, vision loss becomes more restricted to the central visual field and is eventually lost.Likely due to the multiple mutations that cause RP, the age of onset and rate of degeneration is variable.
Two types of treatments are being pursued for retinal disease.Neuroprotective strategies aim to maintain and preserve the patient's current visual function, with the possibility that vision may decline, but with the ultimate goal of retaining functional vision for many years.In contrast, restoration of vision refers to treatments that replace components of the visual circuit (e.g.mutated genes, degenerated photoreceptors) to repair vision or activate residual vision (Sabel et al., 2011b) with the goal of improved vision.However, neither one of these approaches is capable of returning the retina to normal, healthy visual function.
Mechanisms of disease and treatment for inherited retinal diseases are the focus of a disproportionate large amount of research programs compared to the relatively small number of individuals affected (Parmeggiani, 2011) (Fig. 1).While targeted to a limited number of people, this research is important for understanding inherited retinal diseases and may lead to discovery of treatments for more complex diseases such as AMD.Thus, restorative treatments for this form of vision loss have progressed further along the clinical trials pipeline to translation.Retinal prosthetics directly stimulate the neural retina to restore visual signals and sight.The only FDA approved treatment for RP that is targeted at individuals in advanced stages of disease is the Argus II electronic epiretinal device (Second Sight Medical Products, CA, USA) (Mills et al., 2017).This retinal implant is currently FDA approved as a humanitarian device for individuals with severe RP and bare light perception or no light perception in both eyes (Mills et al., 2017).A similar device, the Alpha-IMS electronic subretinal device (Retina Implant AG, Germany) is currently undergoing multi-center clinical trials in Europe (Cheng et al., 2017;Mills et al., 2017) and is commercially available in some countries.While visual function is rudimentary and not fully restored with these devices, some implanted patients report improvements in quality of life measures but others do not (Mills et al., 2017).Other restorative approaches under development include optogenetic therapy (Yue et al., 2016) and cell transplantation (Luo and Chen, 2018).
Gene therapy could be a restorative or neuroprotective treatment depending on when in the disease course it is applied (before or after pathology) and which gene is being targeted.Gene therapy for the retina has been most successful in correcting RPE65 mutations in Leber's congenital amaurosis (LCA) (Cideciyan, 2010).LCA is an inherited form of retinal degeneration caused by mutations in 14 different genes, including RPE65 (http://www.sph.uth.tmc.edu/RetNet/).LCA is Fig. 1.Prevalence in the US of the four main types of retinal diseases that lead to severe vision loss and blindness (left) compared to the relative number of publications from a Pubmed Central search related to neuroprotective research for each disease (right).The prevalence numbers were obtained from the NIH website for the year 2010 with the number of people affected provided in the graph (https://nei.nih.gov/eyedata).The Pubmed Central search was performed with the key words "neuroprotection" and the disease name with the number of citations highlighted.M = millions; K = thousands.

M.T. Pardue, R.S. Allen
Progress in Retinal and Eye Research 65 (2018)   associated with vision loss in early childhood, nystagmus and hyperopia.Gene therapy for RPE65 has been successful in restoring vision to animal models of LCA and the first clinical trials are underway with positive, albeit temporary, benefits to retinal sensitivity, dark-adapted perimetry, and vision-guided mobility (Kumaran et al., 2017).Currently, retinal prosthetics and gene therapy are targeted to individuals in late stages of inherited retinal disease mainly due to safety concerns.Thus, there is a need for neuroprotective agents that can be applied much earlier after initial detection of vision loss to slow the progression to severe vision loss and blindness.AMD is traditionally described as a disease that affects central vision.During the early stages of AMD, Bruch's membrane thickens due to lipid and protein accumulation, and extracellular deposits called drusen form between the RPE and Bruch's membrane (Bowes Rickman et al., 2013).As the disease progresses, drusen accumulate and geographic atrophy (regional loss of choroid, RPE and retinal neurons) and/or neovascularization develops.Photoreceptor death appears to be preceded by loss of RPE or choroid vasculature (Bhutto and Lutty, 2012), possibly due to mutations in complement factor H (Toomey et al., 2018).There are multiple risk factors for AMD, including genetic factors causing complement dysregulation, age, smoking and sunlight exposure (Armstrong and Mousavi, 2015).AMD has two forms: dry and wet.The distinguishing feature is the involvement of retinal vascular abnormalities in the wet form, including vessel leakage and eventual angiogenesis.AMD is the leading cause of vision loss in the aging population.In the US, 2.07 million people over 65 years of age are affected by AMD (Fig. 1), with an estimated increase to 5.4 million by 2050 due to the aging population (https://nei.nih.gov/eyedata/amd/tables).AMD is the only retinal disease with an FDA-approved neuroprotective agent, Age Related Eye Disease Study (AREDS), which is an oral supplement that contains vitamin C, vitamin E, beta-carotene, zinc oxide, and cupric oxide.AREDS has been reported to decrease the progression to advanced AMD by 25% in patients with wet AMD (Chew et al., 2013).Wet AMD is also treated with anti-VEGF injections and laser therapy to control angiogenesis (Amoaku et al., 2015).Due to the high prevalence of AMD, the projected increase in the aging population, and its devastating effects on central visual function, the development of neuroprotective strategies to treat AMD is greatly needed.

Diabetic retinopathy (DR)
DR is a leading cause of vision loss, affecting over 7 million people in the US (https://www.nei.nih.gov/eyedata/diabetic#5).Since the prevalence of diabetes has increased from 6 million Americans in 1990 to over 23 million Americans in 2015 (https://www.cdc.gov/diabetes/statistics/slides/long_term_trends.pdf), the number of individuals that develop this retinal complication of hyperglycemia is also expected to increase at an alarming rate to 14 million Americans by 2050 (https:// www.nei.nih.gov/eyedata/diabetic#5).DR has been historically described as a microvascular disease that is recognized clinically by the appearance of structural abnormalities in the retinal vasculature.However, over the last 15 years, numerous studies have shown that DR also affects retinal neurons, with reports of abnormalities in photoreceptor, amacrine cells, and retinal ganglion cells (Antonetti et al., 2006;Fletcher et al., 2007).While hyperglycemia underlies both type I and type II diabetes, the exact cause(s) of neuronal and vascular defects in DR are still unknown.Increased oxidative stress potentially due to hyperglycemia, aldose reductase, and/or inflammation (Frank, 2004) appears to lead to neuronal dysfunction (Antonetti et al., 2006), damage to mitochondrial DNA (Kowluru et al., 2015), vascular dysfunction (Garhofer et al., 2004), and eventual vascular leakage and neovascularization.A highly successful public health screening program for retinal disease recommends yearly fundus exams in diabetic patients to detect the first signs of vascular damage.While current data suggests that neuronal changes may occur prior to these structural changes, diabetic patients often don't have visual complaints.By the time structural vascular changes are visible in the fundus, vision loss due to DR is inevitable.The current treatments for DR include laser surgery and anti-VEGF therapies, which provide some control of abnormal blood vessel growth (Frank, 2004), but do not prevent vision loss.Importantly, the current treatments are targeted at mid-to-late stage DR when vision loss is likely irreversible.Combining neuroprotective therapies with new screening methods for early stage DR (Aung et al., 2014) could significantly decrease the number of diabetic patients with vision loss.

Retinal ganglion cell disease
Retinal ganglion cells (RGCs) provide the final circuit between retinal processing and higher order visual processing in the midbrain and cortex.Thus, in RGC diseases the photoreceptors may be healthy and able to detect light, but visual information is not transmitted to the midbrain for processing and interpretation.The main diseases that affect the RGCs include glaucoma and anterior ischemic optic neuropathy (AION).Glaucoma is another major cause of vision loss, affecting 2,720,000 people in the US (https://nei.nih.gov/eyedata/glaucoma) (Fig. 1).Glaucoma is a group of ocular disorders united by a clinically characteristic optic neuropathy with associated RGC loss, loss of visual fields, and eventual blindness (Pang and Clark, 2007;Schwartz, 2005).The primary risk factor for glaucoma is high intraocular pressure (IOP), although normal tension glaucoma affects the RGCs without an increase in IOP (Weinreb et al., 2014).Other risk factors include age, family history of glaucoma, black race, and use of systemic or topical corticosteroids have also been associated with glaucoma (Weinreb et al., 2014).
Primary open angle glaucoma is the most common subtype of glaucoma, characterized by RGC axon damage with subsequent loss of RGC cell bodies.There is considerable evidence from both monkey studies (Yang et al., 2009(Yang et al., , 2015) ) and more recent rodent studies (Schaub et al., 2017) that the optic nerve head is the primary site of pathology in glaucoma (Jonas and Budde, 2000).Human pathology reports are limited but also demonstrate axonal injury at the ONH (Vrabec, 1976).The exact mechanisms causing axonal injury at the optic nerve head and precipitating the death of RGCs are unknown (Nickells et al., 2012).A leading theory is that increased IOP may alter the biomechanical aspects of the posterior globe, leading to increased pressure on the RGC axons as they exit the eye in the optic nerve that then leads to RGC apoptosis (Nguyen and Ethier, 2015).However, since glaucoma can develop without high IOP, glaucoma may also be due to vascular dysregulation (Flammer et al., 2013;Mastropasqua et al., 2015;Schwartz, 2005;Yamamoto and Kitazawa, 1998).Other theories include deficits in anterograde and retrograde transport in the RGC axon, metabolic stress, activation of molecular cell death pathways in axons, and neuroinflammation (Nickells et al., 2012).
The leading treatments for glaucoma are eye drops with pharmacological agents (CNTGSG, 1998) or surgery to reduce IOP (Schwartz, 2005) and the risks associated with glaucoma (vision loss, increased risk of retinal detachment, etc.).However, these treatments are not effective for all patients (Heijl et al., 2002) and compliance rates with eye drops are low (McKinnon et al., 2008).Even when these treatment strategies are successful, RGC death and vision loss can persist.Further, in patients with normal tension glaucoma, RGC degeneration progresses even with normal IOP (Acland et al., 2001;Schwartz, 2005).Thus, neuroprotective strategies could be developed to target secondary RGC degeneration alone or used in combination with IOP lowering methods.
A second condition that affects the RGCs is anterior ischemic optic neuropathy (AION), in which a stroke of the optic nerve causes sudden vision loss.AION affects 23 to 102 people per million each year (Johnson and Arnold, 1994).Treatments for AION include aspirin, VEGF inhibitors, and systemic corticosteroids (Bernstein et al., 2011;Foroozan, 2017).However, results are mixed, and other treatment options are needed.

Types of neuroprotective strategies
For this review, we have grouped neuroprotective strategies into two general categories: drugs that target survival pathways and rehabilitative methods that increase endogenous repair mechanisms.Each of these approaches has the potential for use across several different retinal diseases.As detailed in the following sections, these approaches may target a specific pathway or multiple pathways, but often result in pleiotropic effects that lead to increased retinal neuron survival and preserved visual function.Drugs include anti-apoptotic agents like tauroursodeoxycholic acid (TUDCA), steroids, and dopamine-related therapies.Other drug therapies include growth factors or growth factor receptor agonists, such as ciliary nerve trophic factor (CNTF).Rehabilitative methods offer another means to activate endogenous survival pathways.Among them, physical exercise and electrical stimulation are two approaches that are commonly used to rehabilitate motor and cardiovascular systems and brain health.It should be noted that there are other neuroprotective agents reported in the literature demonstrating beneficial effects on the retina, such as red light, nutritional supplements (AREDS, saffron, curcumin, etc.) (Bernstein et al., 2016;Morrone et al., 2015), rod-derived viability factor (Ait-Ali et al., 2015;Byrne et al., 2015) and erythropoietin (Caprara and Grimm, 2012).In this review we have focused on neuroprotective approaches that we have experimented with in our laboratories and we believe have a growing body of evidence suggesting benefit to retinal diseases.The following sections provide detailed examination of these approaches by reviewing the history of the neuroprotective effects, the known mechanisms of action, the protective effects on retinal neurons, and additional points to consider in moving the approach into the clinic, including gaps in knowledge and limitations to the approach.We have emphasized functional and structural outcomes as evidence for translational potential.In order to make comparisons between neuroprotective strategies with different animal models and methods, we calculated a fold difference ratio based on functional and structural data reported in the literature.

Background
Ursodeoxycholic acid (UDCA) and its taurine-conjugated derivative tauroursodeoxycholic acid (TUDCA) are powerful neuroprotective agents with multiple actions.Both are found naturally in bile acid of hibernating bears and their beneficial effects have been noted in ancient homoeopathic Chinese medical texts for detoxifying the liver, dissolving kidney stones and gallstone, stopping convulsions, and improving vision (Cidian, 2004;Williamson and Phipps, 2001).This wide array of actions and benefits to widely varying diseases is mirrored in modern day uses.Synthetic hydrophilic bile acids were first FDA-approved to treat liver disease (Lazaridis et al., 2001), before the discovery of their neuroprotective properties for neurodegenerative diseases, such as Parkinson's and Huntington's diseases in the early 2000s (Duan et al., 2002;Keene et al., 2001Keene et al., , 2002)).In addition to the neuroprotective effects of TUDCA on retinal neurons as detailed below, recent experiments also show beneficial effects of TUDCA on stem cell survival (Yoon et al., 2016), stem cell proliferation and conversion (Soares et al., 2017), cardiovascular disease (Choi et al., 2016;Groenendyk et al., 2016;Qin et al., 2017;Rani et al., 2017;Xie et al., 2016), and diabetes (Chen et al., 2016;Fan et al., 2017;Walsh et al., 2016;Zhang et al., 2016).

Mechanism of action
TUDCA targets the fundamental aspect of neurodegeneration-apoptosis-through multiple actions (Vang et al., 2014).TUDCA prevents apoptosis by stabilizing the mitochondrial membrane (Botla et al., 1995;Rodrigues et al., 1998), suppressing the pro-apoptotic P53 protein (Park et al., 2008), reducing endoplasmic reticulum (ER) stress by improving protein folding capacity (Omura et al., 2013), and decreasing reactive oxygen species (Mantopoulos et al., 2011;Oveson et al., 2011).In addition, TUDCA has anti-inflammatory effects that reduce glial cell activation by inhibiting nuclear factor-kappa beta (NFκβ) activation in glial cells and inducing transforming growth factor β pathway (Noailles et al., 2014;Romero-Ramirez et al., 2017).While not confirmed, one of the benefits of the taurine-conjugated form of UDCA may be the breakdown of TUDCA to release taurine which has been shown to be neuroprotective for both photoreceptors and retinal ganglion cells (Froger et al., 2014).These robust and pleiotropic effects create a powerful inhibitor of apoptotic cell death that promotes neuron survival and function.

Experiments in the retina
In the retina, the pleiotropic action of TUDCA has been shown to provide neuroprotection in several retinal disease models, including photoreceptor degenerations, diabetic retinopathy, and RGC cultures.While the underlying pathophysiology of these retinal diseases differs, TUDCA acts on the common mechanisms of oxidative stress and apoptosis.

Photoreceptor degeneration
TUDCA benefits rodent models of photoreceptor degeneration: light-induced retinal degeneration in BALB/c mice (Boatright et al., 2006;Oveson et al., 2011), Pde6b rd1/rd1 (rd1) mice (Lawson et al., 2016), Pde6b rd10/rd10 (rd10) mice (Boatright et al., 2006;Oveson et al., 2011;Phillips et al., 2008), Bardet-Biedl syndrome mice (Drack et al., 2012), and transgenic P23H rats (Fernandez-Sanchez et al., 2011;Noailles et al., 2014).TUDCA treatments resulted in significant preservation of retinal function and photoreceptor structure in each of these models.Effects on retinal function, as measured by electroretinogram (ERG) ranged from 1.6 to 11x larger dark-adapted a-wave amplitudes and 1.3 to 7x larger dark-adapted b-waves (Fig. 2A and B; Supplemental Table 1).[Note that the largest improvements were from the Boatright study in which ERGs were recorded at 24 h post-light exposure compared to 1 week post-light exposure in other studies].TUDCA significantly preserved the photoreceptor density with reports of 1.2 to 2x greater total photoreceptor numbers (Fig. 2C; although only preserving ∼50% of normal wild-type mice ERG amplitude and photoreceptor numbers (Phillips et al., 2008)).Most notably, TUDCA promoted greater survival (2.4-4.5xgreater) in the cone photoreceptors (Oveson et al., 2011;Phillips et al., 2008) which are responsible for high resolution vision.In regard to dosing, TUDCA was able to provide benefit to retinal function and structure when delivered every three days (Boatright et al., 2006;Phillips et al., 2008) or once weekly (Fernandez-Sanchez et al., 2011) using subcutaneous or intraperitoneal injections, respectively.However, for more aggressive retinal degeneration, like the rd1 mice, daily intraperitoneal injections were necessary to slow photoreceptor degeneration (Lawson et al., 2016), and for some studies, even this dosing was not sufficient to stop photoreceptor cell death (rd1 and rd16 mice) (Drack et al., 2012).
TUDCA has direct action on photoreceptor survival by improving protein folding and trafficking, reducing ER stress, and reducing oxidative stress.In a Leber's congenital amaurosis model, Lrat−/− mice, TUDCA preserved cone photoreceptor density by 3 fold while not altering rod photoreceptor numbers (Zhang et al., 2012a).TUDCA also reduced the unfolded protein response and reduced apoptotic markers (caspase 3) in Lrat−/− mice.Using photoreceptor 661W cell cultures, Duricka et al. showed that TUDCA reduced expression of ER stress markers and improved trafficking of cone photoreceptor cyclic nucleotide-gated channels (Duricka et al., 2012).Additionally, TUDCA provided benefit to the photoreceptors indirectly by enhancing phagocytosis of photoreceptor outer segments by activating MerTK in the retinal pigment epithelial cells (Murase et al., 2015).Furthermore, TUDCA greatly reduced photoreceptor death after retinal detachment, reducing oxidative stress as measured by > 3x reduction in reactive oxygen species which was indistinguishable from controls (Mantopoulos et al., 2011).Interestingly, in this retinal detachment model, TUDCA did not reduce ER stress, contrary to findings above and in other tissues (Duricka et al., 2012;Omura et al., 2013).

Diabetic retinopathy
TUDCA also benefits diabetic retinopathy.TUDCA decreased cell death in retinal cells exposed to elevated glucose, as measured by reduced annexin V and 1.4x less TUNEL labeling (Gaspar et al., 2013).TUDCA also improved translocation of apoptosis inducing factor from the mitochondria.Finally, TUDCA-treated cells exposed to high glucose have fewer reactive oxygen species (1.3x less) and oxidized proteins (1.5x less) indicating reduced oxidative stress.Similar results were reported in explanted retinas from 3 week post-STZ diabetic rats (Oshitari et al., 2014).Apoptotic markers were reduced with TUDCA exposure: TUNEL labeling by 1.6x and phosphorylated c-Jun by 2.6x in the whole retina and 4x in the ganglion cell layer.As further evidence of TUDCA increasing retinal cell survival in diabetic retinopathy, the number of neurites was 2x greater (Oshitari et al., 2014).Diabetic retinopathy is also associated with inflammation and TUDCA may provide benefit on this front too.AGE-exposed rat retinas show decreased NF-κβ after exposure to TUDCA (Bikbova and Oshitari, 2015).Additionally, retinal Muller cells cultured with high glucose and TUDCA show reduced c-Jun N-terminal kinase (JNK) levels (Zhong et al., 2012).Moving to whole animal models, TUDCA treatment was effective in slowing early visual function in STZ-induced diabetic mice.Spatial frequency thresholds were 1.1x greater and contrast sensitivity 1.5x greater in TUDCAtreated diabetic mice compared to vehicle treated at 8 weeks post-STZ and 7 weeks of TUDCA treatment (Fu et al., 2016).When UDCA was given two weeks after STZ-induced diabetes in C57 mice, retinal pericyte loss was attenuated, as evidenced by decreased vascular leakage and preserved retinal capillary numbers at 8 weeks post-STZ through diminution of the unfolded protein response (Chung et al., 2017).Further studies are needed to determine the duration of TUDCA's protective effects in diabetic retinopathy and whether it would alter the vascular pathology associated with late stage disease.Evidence of TUDCA benefiting vasculature comes from a laser-induced choroidal neovascularization rat model in which leakage was reduced by 1.4x and lesion size by 1.4x (Woo et al., 2010).In addition, VEGF levels were also decreased by 3 fold.The authors suggest these results provide evidence of a potential application of TUDCA in the wet form of agerelated macular degeneration.However, neovascularization is common to many retinal diseases and these data suggest the potential use of TUDCA in any retinal disease accompanied by vascular defects.

Retinal ganglion cells
While delivery of TUDCA to a glaucoma model has not yet been reported, some studies show that TUDCA protects RGCs.Intravitreal injections of NMDA into the rat eye induces a model of RGC degeneration.In this study, daily i.p. injections of TUDCA preserved RGC function as measured by 2x larger positive scotopic threshold response, an ERG wave generated by the RGCs (Gomez-Vicente et al., 2015).Furthermore, RGCs had 20% greater survival.In Leber's hereditary optic neuropathy (LHON), RGCs undergo apoptosis caused by a mitochondrial DNA mutation.Chao de la Barca and colleagues demonstrated that ER stress is an integral part of LHON by showing that TUDCA reduced several ER stress markers in fibroblasts harvested from LHON patients (Chao de la Barca et al., 2016).Finally, TUDCA may also provide benefit to healthy RGCs.In whole mount retinas from cats treated with TUDCA, RGC recordings showed reduced luminance thresholds and improved visual response properties (Xia et al., 2015).Thus, it seems TUDCA benefits RGC survival and function and would make a good candidate for testing in glaucoma.

Progress in testing UDCA and TUDCA in retinal disease
A number of studies have tested and shown benefit of UDCA and TUDCA to the three major retinal diseases.Comparisons of improvements in retinal function and structure with TUDCA treatment across the different retinal diseases show very similar magnitudes of benefit (Fig. 3).In general, the amount of benefit is approximately 2.2 fold in function and 1.7 fold in structure across photoreceptor degenerations, DR and RGC disease.Since UDCA and TUDCA target apoptosis and ER stress, perhaps these results suggest that these pathophysiological mechanisms affect the retinal neurons similarly, regardless of the origin of the disease or the endstage manifestations.Thus, UDCA and TUDCA provide much promise for a general neuroprotective agent that could provide benefit to multiple retinal diseases.

Progress of UDCA and TUDCA to the clinic
Translation of TUDCA treatments into the clinic has already begun for multiple neurodegenerative diseases.Since UDCA and TUDCA have FDA approval, the needed toxicity screening has already been completed and UDCA/TUDCA is well-tolerated by humans with few side effects (Angelico et al., 1999;Pan et al., 2013).Currently, there are 121 UDCA and 16 TUDCA trials listed in clinicaltrials.gov.Only one trial Modified from (Phillips et al., 2008).
focuses on retina: a study to evaluate UDCA as an adjuvant treatment for rhegmatogenous retinal detachment (NCT02841306; University of Lausanne, France).In this study, one dose of UDCA is given prior to surgery to repair the detachment and then for 4 daily doses after surgery in order to promote photoreceptor survival.This study is still recruiting and results are not yet available.Four trials use UDCA or TUDCA in diabetic patients.With the current pre-clinical data showing TUDCA benefits the retina, there appears to be an opportunity to leverage these current trials to test visual outcomes in diabetic patients that may develop or already have DR.
One of the challenges of translating UDCA/TUDCA to the clinical for eye diseases is delivery of this agent to the eye.The systemic doses used in animal experiments (500 mg/kg) are extremely high for humans.Even manufacturing such large quantities would be a considerable hurdle.As with other neuroprotective strategies, the ideal delivery would target the eye without systemic administration as to provide the optimal dose for the retina with the fewest systemic side effects.Fernandez-Sanchez et al have developed slow-release delivery of TUDCA using microspheres (Fernandez-Sanchez et al., 2017).In their P23H rat model of photoreceptor degeneration, they showed modest functional preservation when microspheres were injected intravitreally once a month compared to their results with the same rat model using weekly systemic injections.Thus, optimization of the dosing is needed to provide sustained delivery at a more efficacious level.

Background
The neuroprotective effects of steroid hormones have been documented in a number of injury models.While this review will focus on progesterone and its metabolites, it is important to note that estrogen has a history of neuroprotection as well, including neuroprotection in the retina (Li et al., 2006;Nonaka et al., 2000;Zhu et al., 2015).Progesterone has been studied in traumatic brain injury for three decades.Endogenous progesterone levels were found to protect female rats against brain injury in 1987 (Attella et al., 1987), with exogenous progesterone administration being tested in males and females soon after (Roof et al., 1992).Since then, progesterone has been shown to have pleiotropic effects in preventing neuronal death and improving behavioral outcomes with results being reported in over 250 articles and 22 + different animal models by 40 different laboratories (Sayeed and Stein, 2009;Schumacher et al., 2007;Stein and Wright, 2010).
Thus, progesterone has generated much interest as a neuroprotective agent with potential to treat a variety of brain-related diseases.

Evidence of progesterone and estrogen, their receptors, and their synthesis in the retina
Estrogen, progesterone, and their respective receptors have been identified at the mRNA and/or protein level in the retina, RPE, optic nerve, and occipital cortex in animals and humans, and in males and females (Allen et al., 2016;Koulen et al., 2008;Lanthier and Patwardhan, 1986;Wickham et al., 2000;Jackson et al., 2016).Synthesis of estrogen and progesterone, as well as intermediate steps including the synthesis of precursors and metabolites, occurs in the retina as well (Cascio et al., 2007;Coca-Prados et al., 2003;Guarneri et al., 1994;Lanthier and Patwardhan, 1988;Sakamoto et al., 2001).In addition to acting through classical receptors (in which receptors form dimers and translocate to the nucleus), progesterone acts at membraneassociated progesterone receptors (which are also found in the retina) (Shanmugam et al., 2016;Swiatek-De Lange et al., 2007), at glucocorticoid receptors as a non-competitive antagonist (Svec et al., 1980(Svec et al., , 1989)), at sigma-1 receptors as an antagonist (Maurice et al., 1998), and at nicotinic acetylcholine receptors as a modulator of activity (Valera et al., 1992).Progesterone, pregnenolone, and allopregnanolone activate the pregnane X receptor (Kliewer et al., 1998;Langmade et al., 2006), which is expressed in the RPE (Zhang et al., 2012b).Allopregnanolone binds to the GABA-A receptor as a positive modulator to increase inhibition (Puia and Belelli, 2001;Reddy et al., 2005).

Research in animal models
The bulk of research on progesterone in the retina has been performed in both light-induced and genetic models of retinal degeneration, with administration of progesterone or synthetic progestins resulting in reduced and/or delayed photoreceptor cell death, improvements in morphology, reduced glial activation, and reduced deficits in retinal function (Fig. 4) (Doonan et al., 2011;Jackson et al., 2016;Roche et al., 2017aRoche et al., , 2016Roche et al., , 2017b;;Ruiz Lopez et al., 2017;Sanchez-Vallejo et al., 2015).Progesterone was shown to protect stressed photoreceptor cells in culture by upregulating fibroblast growth factor, which mediated calcium influx (Wyse- Jackson et al., 2016;Wyse Jackson and Cotter, 2016).Most of these studies focused on the mechanism of underlying progesterone benefit using cell cultures.In vivo studies using orally administrated progesterone to rd1 mice every other day starting at P7 provides modest benefit with the largest preservation (1.4x) of photoreceptor nuclei in the far periphery observed at P15 that was indistinguishable from controls by P17 (Sanchez-Vallejo et al., 2015).Similarly, progesterone-treated rd1 mice have ERG b-waves that were twice as large at P15, but only 1.2 fold difference by P17 (Sanchez-Vallejo et al., 2015).However, the FDA-approved Norgestrel, a synthetic progestin, reduced photoreceptor cell death by 70% in a light damage model and by 75% in the rd10 mice and preserved function (ERG b-wave ∼4-5x higher) (Doonan et al., 2011).
Few studies have evaluated progesterone in diabetic retinopathy.In one study on retinal glial cells, retinal explants from STZ-induced Type 1 diabetes show reduced osmotic swelling, which may promote survival (Neumann et al., 2010).
Ischemic injury can affect the RGCs and other inner retinal neurons.
Progesterone protection has been observed in the retina in middle cerebral artery occlusion (Allen et al., 2015b;Allen et al., 2016).In this model, retinal function was significantly preserved in the contralateral eye by 1.4 fold, and RGC numbers were 1.3 fold higher with progesterone treatment (Allen et al., 2015b).In a high intraocular pressure model of ischemia-reperfusion, progesterone is reported to preserve inner retinal neurons (Lu et al., 2008).Norgestrel in reducing levels of inflammatory cytokines in retinal ischemic injury (Allen et al., 2016) and decreasing proinflammatory activation via microglia activity in retinal degeneration (Roche et al., 2016) may point to the potential of these treatments as anti-inflammatory agents in retinal injury and disease.Negative results with progesterone treatment in eye disease are reported in a few papers (Allen et al., 2015b;Kaldi and Berta, 2004;Nakazawa et al., 2006;O'Steen, 1977).O'Steen did not observe protection with progesterone treatment in light-induced retinal degeneration in rats (O'Steen, 1977) and found that ovariectomized rats exhibited reduced light damage (Olafson and O'Steen, 1976).The rats in this study were treated with 2.5 mg of progesterone or progesterone plus estradiol during 14 days of light exposure.However, the precise treatment regimen and route of administration were not specified (O'Steen, 1977).Negative results could be due to issues with the timing, dose, or tapering of the progesterone treatment, all necessary components for successful progesterone treatment.Additionally, progesterone may not work in every retinal injury model, for example, the anterior ischemic optic neuropathy model (Allen et al., 2015b).

Research in humans
Research on the effects of progesterone on ocular disease in humans is sparse and often anecdotal.A variety of ocular changes have been shown to take place during times of hormonal fluctuation (Avitabile et al., 2007;Errera et al., 2013;Grant and Chung, 2013;Panchami et al., 2013;Yucel et al., 2005;Ziai et al., 1994), and gender differences are observed in rates of DR, glaucoma, AMD, dry eye, and cataracts (Wickham et al., 2000).Progesterone administered exogenously has been shown to reduce intraocular pressure, suggesting potential benefit for glaucoma (Obal, 1950;Posthumus, 1952).
Much of the human research on progesterone and disease incidence is contradictory.Studies on oral contraceptives and hormone replacement therapy often fail to distinguish whether they are using progesterone and estrogen or synthetics.For example, progesterone was claimed to increase glaucoma risk, but the hormone tested in that study was medroxyprogesterone acetate (Thomas et al., 2003), a synthetic progesterone/progestin that has been shown to have progesterone-like behavior in the reproductive system but to increase clotting in blood vessels (Thomas et al., 2003) and exacerbate excitotoxicity in neural tissue (Nilsen et al., 2006).Synthetics differ in metabolism, potency, pharmokinetics, chemical structure, receptors they bind to, and biological effects (Nilsen et al., 2006;Reddy et al., 2005;Thomas et al., 2003Thomas et al., , 2007)), and have even been shown to decrease protection provided by other neurosteroids (Littleton-Kearney et al., 2005;Nilsen and Brinton, 2002).

Applying progesterone treatment to retinal disease
Due to the tremendous overlap between mechanisms modulated by estrogen and progesterone and mechanisms involved in eye disease and injury, steroid hormones make a promising candidate for treatment.Specifically, we think progesterone and estrogen treatments are promising in ocular stroke and trauma (due to success in cerebral stroke and trauma - (Attella et al., 1987) (Roof et al., 1992) (Sayeed and Stein, 2009;Schumacher et al., 2007;Stein and Wright, 2010) (Cutler et al., 2007) (Allen et al., 2016;Ishrat et al., 2010;Yousuf et al., 2013)  2016, 2017b; Ruiz Lopez et al., 2017;Sanchez-Vallejo et al., 2015)), and in glaucoma (due to progesterone's ability to both protect and to reduce intraocular pressure - (Obal, 1950;Posthumus, 1952)).However, due to conflicting reports about whether progesterone increases VEGF and neovascularization or merely protects existing vasculature from dying off (Ishrat et al., 2012;Li et al., 2012;Won et al., 2014;Yousuf et al., 2014), more experiments must be done before testing progesterone in patients with DR or macular degeneration.
The optimal dose, timing, and tapering of dose must also be determined in order to have successful translation from brain to retina.The optimal progesterone dose differs by disease (Cutler et al., 2007;Wali et al., 2014) and would need to be determined for retinal diseases.Timing of dose is also criticalfor example, pretreatment with progesterone can actually be harmful (Murphy et al., 2000), as can abruptly ceasing treatment without tapering the dose (Cutler et al., 2005(Cutler et al., , 2006)).Additionally, progesterone has most often been used in acute brain and nerve injury models, so long term effects in the retina would have to be studied as well.Finally, progesterone was shown to provide greater protection in brain vs. retina in a model that caused a concurrent stroke, which may be explained by an increase in progesterone receptors in the brain and a decrease in progesterone receptors in the retina following ischemia (Allen et al., 2016).It is possible that larger doses of progesterone or treatment with allopregnanolone, a progesterone metabolite that acts independently of the progesterone receptor, may prove useful in retinal disease and injury models.

Using progesterone in the clinic
Progesterone treatment would be relatively easy to translate to the clinic, given that it is safe, FDA-approved, and can be administered systemically through multiple routes (Stein and Wright, 2010).However, while the literature supporting progesterone's protective effects in animal research and Phase II clinical trials (Wright et al., 2007;Xiao et al., 2008) is vast, two Phase III clinical trials testing progesterone treatment in traumatic brain injury failed to show an effect (Skolnick et al., 2014;Wright et al., 2014).This lack of protection could be due to methodological problems that include less than optimal dosing parameters (Howard et al., 2017) and lack of sensitivity in outcome measures (Stein, 2015).Future studies of progesterone treatments in retinal diseases will need careful planning of the experimental design to avoid such potential causes for failure.

Background
Dopaminergic amacrine cells in the retina play a role in light adaptation, growth cues, and signal modulation (Witkovsky, 2004).In diseases like DR, dopamine loss and disruption cause early visual dysfunction that appears even before a clinically recognizable vascular phenotype (Aung et al., 2014), and treatment with L-DOPA (the precursor to dopamine) or dopamine agonists reduces these deficits (Aung et al., 2014).Additionally, dopamine has been shown to be an antiangiogenic in research outside of the eye (Sarkar et al., 2015).While L-DOPA and dopamine agonists have long been used in the treatment of Parkinson's disease (Katzenschlager and Lees, 2002;Oertel, 2017), treatments to increase retinal dopamine levels as a neuroprotective strategy represent a novel and exciting avenue for retinal disease research.
There is accumulating evidence of dopamine loss in retinal disease.A reduction of dopamine and its metabolites has been observed in both diabetic rat retinas (Aung et al., 2014) and retinas from Parkinson's patients (Harnois and Di Paolo, 1990) and animal models (Cuenca et al., 2005;Esteve-Rudd et al., 2011).Dopaminergic amacrine cell loss has been observed in diabetic retinas as well (Gastinger et al., 2006).The visual changes accompanying dopamine deficiency in these diseases include reduced visual acuity (Aung et al., 2014;Holroyd et al., 2001;Matsui et al., 2006) and contrast sensitivity (Adam et al., 2013;Aung et al., 2014;Bulens et al., 1988;Hutton et al., 1993) and deficits in flash and pattern ERGs, visual evoked potentials, and electrooculograms (Deak et al., 2016;Esteve-Rudd et al., 2011;Ikeda et al., 1994;Schneck et al., 2008).Retinal thinning occurs in the retinas of Parkinson's patients (Adam et al., 2013;Bodis-Wollner, 2009) with specific loss of AII amacrine cells reported in a monkey model of Parkinson's disease (Cuenca et al., 2005).In addition, reduced dendritic branching and synaptic contacts with AII amacrine cells has been observed in dopaminergic amacrine cells in several models of retinal degeneration, indicating that dopamine may play a role in functional loss in these diseases as well (Ivanova et al., 2016).

Dopamine as a neuroprotective treatment: mechanisms of action
Evidence is lacking that dopamine acts directly to reduce apoptosis, inflammation, or oxidative stress in a fashion similar to other neuroprotective agents reviewed here.However, reduced dopamine appears to have a negative effect on neuronal survival in the brain, such as in Parkinson's disease (Lotharius and Brundin, 2002).Additionally, neuronal degeneration and pathology can also lead to secondary effects on dopaminergic amacrine cell morphology, as is the case in retinal degeneration mouse models that show decreased soma size, dendrite length, and disrupted innervation (Ivanova et al., 2016).Thus, L-DOPA and dopamine agonists may simply act to restore visual and neuronal function in these diseases by restoring the dopamine that has been lost as a result of disease (Aung et al., 2014;Harnois and Di Paolo, 1990;Onofrj et al., 1986;Peppe et al., 1995).However, there is evidence that L-DOPA may stimulate G protein coupled receptors (GPCR) like GPR143 in RPE that control secretion of pigment epithelium-derived factor (PEDF) which may benefit diseases like AMD (Lopez et al., 2008;McKay and Schwartz, 2016).RPE cells produce L-DOPA as an intermediate of the melatonin pathway (Schraermeyer et al., 2006) and are capable of synthesizing L-DOPA into dopamine (Ming et al., 2009).It is possible that the reduced pigmentation that occurs with aging (Sarna et al., 2003) causes a reduction in L-DOPA in the RPE, which makes older adults more susceptible to AMD (Brilliant et al., 2016).Additionally, dopamine acts as an anti-angiogenic in models of acute lung injury (Vohra et al., 2012), irritable bowel syndrome (Tolstanova et al., 2015), and a number of types of cancer (Basu et al., 2001;Sarkar et al., 2015), and with fewer side effects than the usual anti-angiogenic drugs (Sarkar et al., 2015).Specifically, dopamine inhibits angiogenesis by causing the endocytosis of VEGF receptor 2, thus disrupting the VEGF pathway (Basu et al., 2001).

Research in animal models
Dopamine agonists were shown to ameliorate functional loss and cell death in a light-damage model of retinal degeneration, maintaining ERG function and photoreceptor cell counts similar to healthy control mice (Shibagaki et al., 2015).Dopamine deficiency has been identified as a mechanism causing early visual dysfunction in Type I diabetes, and treatments that increase retinal dopamine reduce visual and retinal function deficits (Aung et al., 2014).L-DOPA treatment reduced retinal deficits in STZ rats (Aung et al., 2014).L-DOPA treatment delayed decreased spatial frequency and contrast sensitivities thresholds that occur as early as 3 and 4 weeks post-STZ in diabetic rats by 3 weeks and completely preserved retinal function as measured by ERG in diabetic mice at 5 weeks post-STZ compared to non-diabetic control levels (Aung et al., 2014) (Fig. 5).Additionally, treating with dopamine agonists reduced visual function deficits selectively -D4 receptor agonists partially rescued contrast sensitivity deficits (1.6x) while D1 receptor agonists partially rescued spatial frequency deficits (1.1x) (Aung et al., 2014).It is unclear whether dopamine therapy is acting solely by neutralizing dopamine deficiencies, or if, in addition, dopamine is directly neuroprotective.Future studies could examine whether dopamine deficiencies are present in all diseases where dopamine protection is observed and whether protection observed with dopamine therapies is due to restoration or neuroprotection or both.

Research in humans
In a retrospective study comparing AMD incidence in patients taking L-DOPA, L-DOPA was found to be protective, delaying AMD onset by 8 years (Brilliant et al., 2016), potentially through a GPR143 mechanism (McKay and Schwartz, 2016).L-DOPA treatment was also shown to increase dopamine levels in the retinas of Parkinson's patients (Harnois and Di Paolo, 1990), and to reduce deficits in contrast sensitivity (Hutton et al., 1993), visual evoked potentials (Onofrj et al., 1986), and pattern ERG (Peppe et al., 1995).Furthermore, L-DOPA administration may protect against visual function loss in patients with ischemic and traumatic optic neuropathies (Lyttle et al., 2016;Razeghinejad et al., 2010Razeghinejad et al., , 2016)).In addition to its potential neuroprotective role, the dopaminergic system may be involved in regulating intraocular pressure, with D1 receptor agonists increasing aqueous production (and thus IOP), and D2 and D3 agonists decreasing intraocular pressure (Pescosolido et al., 2013).In fact, D1 agonists can be used as an IOP challenge to identify impaired outflow (and thus, a potential predisposition to glaucoma) in non-glaucomatous children of parents who have glaucoma (Virno et al., 2013).

Applying dopamine treatments to retinal disease
Research on dopamine treatments for retinal disease are promising, but currently very limited.More experiments need to be done to determine the potential of dopamine-related treatments for retinal disease.Retinal degenerations and DR both show decreased dopamine levels and evidence of dopaminergic amacrine cell loss with progression.In addition, dopaminergic system may also modulate glaucoma through dopamine receptors in the anterior segment that affect IOP (Pescosolido et al., 2013;Platania et al., 2013).Glaucoma may also affect retinal dopaminergic cells although the evidence is currently sparse with only one report of decreased dopaminergic cells and dopamine rhythms in the retinas of a glaucoma-like disorder in quails (Dkhissi et al., 1996).
Current evidence suggests that dopamine focused treatments would be beneficial in DR because of the potential to reduce early visual dysfunction and protect against later stage vascular pathology, and AMD because L-DOPA can target RPE cells.The optimal dose and the optimal type of dopamine-targeted treatment (L-DOPA, dopamine agonists, dopamine transporter inhibitor) must be determined for each retinal disease.Additionally, options for local delivery to the eye and/ or retina should be investigated as this could help with the issue of side effects.One study showed that apomorphine (a dopamine receptor agonist) could be delivered via eye drops as a treatment for myopia with minimal, if any, side effects (Iuvone et al., 1991).Further studies could investigate delivery of dopamine-targeted treatments to the retina, focusing on sustained release, microneedles, and molecules that are specifically tagged for retinal delivery.

Bringing dopamine treatments to the clinic
Treatments targeting dopamine are ripe for clinical translation, given that L-DOPA or levodopa is already approved for Parkinson's and other diseases (Foster and Hoffer, 2004;Katzenschlager and Lees, 2002) and the dopamine agonist, bromocriptine, is already FDA approved for treatment of Type II diabetes.Because dopamine is a neurotransmitter utilized throughout the brain and the rest of the body in a number of functions, it is important to consider the side effects of systemic dopamine treatment.While it is early to consider treating the retina with dopamine, these side effects need to be considered and potentially addressed before dopamine therapies could be used for retinal disease.For example, it is recommended that all individuals receive a cardiac evaluation before beginning treatment with a stimulant (The American Heart Association) (Thomas et al., 2011).In patients with Parkinson's disease, chronic L-DOPA treatment has been shown to cause motor side effects, like dyskinesia, dystonia, and wearing off effects at the end of dose (Li et al., 2017) (Katzenschlager and Lees, 2002), and systemic side effects, like insomnia, gastrointestinal symptoms, and hallucinations (Foster and Hoffer, 2004).These sides effects are greatly reduced when treating with dopamine agonists instead of L-DOPA, though dyskinesia and other side effects can still occur (Li et al., 2017) (Katzenschlager and Lees, 2002).Similar side effects are observed in children given L-DOPA for myopia and in children given drugs that increase extracellular dopamine in the brain (Adderall, Ritalin, etc.) for attention deficit hyperactivity disorder (ADHD) (Jeffers et al., 2013;Repka et al., 2010;Salardini et al., 2016).Using carbidopa in conjunction with L-DOPA treatment results in fewer side effects because its actions as a peripheral dopa decarboxylase inhibitor inhibit peripheral dopamine metabolism and improve delivery to the central nervous system.Side effects are also greatly reduced when L-DOPA treatment is combined with an antioxidant diet (antioxidants help prevent L-DOPA breakdown into toxic metabolites) (Foster and Hoffer, 2004).Additionally, in Parkinson's disease, clinicians continue to increase the dose because of homeostatic reductions in the dopamine receptors with increased levodopa which contributes to the negative side effects.It is not clear if L-DOPA treatments to the eye would require such large dosing.Finally, bromocriptine treatment in Type II diabetes has already been shown to improve glucose control (Chamarthi and Cincotta, 2017), increase fat loss (Meier et al., 1992), and prevent progression of kidney disease (Mejia-Rodriguez et al., 2013), suggested that systemic disease might benefit with dopamine agonist treatment, in addition to potential retinal benefits.

Background
Growth factors are endogenously produced substances that bind to receptors to promote cell proliferation, survival or regeneration (Kolomeyer and Zarbin, 2014).Neurotrophic factors are a family of growth factors that are neuron-specific.Two classic families of neurotrophic factors include neurotrophins and glial-derived neurotrophic factor (GDNF).In mammalian species, the neurotrophin family consist of nerve growth factor (NGF), brain-derived growth factor (BDNF), neurotrophin 3 (NT-3), and neurotrophin 4 (NT-4) (Dawbarn and Allen, 2003).The GDNF family consists of GDNF, neurturin, artemin, and persephin.Other classes of trophic factors include fibroblast growth factors (FGF), gp130-binding growth factors (such as ciliary nerve trophic factor (CNTF)), insulin-like growth factors, and transforming growth factors (TGF) (Kolomeyer and Zarbin, 2014).Neurotrophic factors are a target of several neuroprotective interventions which aim to upregulate endogenous levels of these survival factors (see Sections 6 and 7).However, exogenous neurotrophic factors can also be delivered to neuronal tissue to promote neuronal survival, regeneration and plasticity (Hodgetts and Harvey, 2017;Onger et al., 2017).
Growth factors are released by glial cells, including microglia and macroglia, like the retinal astrocytes and Muller cells.Muller cells release VEGF, pigment epithelium-derived factor (PEDF), transforming growth factor-β (TGFβ), BDNF, and NGF (Whitmire et al., 2011).In many retinal diseases, the release of these growth factors declines, promoting the hypothesis that restoration of these factors to normal levels will enhance neuronal survival.Some trophic factors such as BDNF are critical for synaptic plasticity, i.e. the strengthening of synaptic transmission following activation of neuronal firing (long-term potentiation) (Leal et al., 2015;Sasi et al., 2017).Thus, BDNF may be able to enhance synaptic strength in addition to promoting survival, both of which would be useful for neuronal fucntioning.
While neurotrophic factors have considerable potential to act as a powerful neuroprotective agent, there are significant challenges.Neurotrophic factors are hydrophilic and typically basic, monomeric or dimeric proteins (Thorne and Frey, 2001).Their ability to target the central nervous system is limited by their inability to easily pass through the blood-brain, blood-retinal or blood cerebrospinal barrier.In addition, they have short in vivo half-lives and poor pharmacokinetics (Thorne and Frey, 2001).Thus, local delivery of neurotrophic factors to the eye avoids some of these limitations.

Experiments in the retina
Various neurotrophic factors are known to be present in the eye and have been shown to have a protective or rescue effect with mechanical injury, environmental toxicity, or hereditary disease.A simple piercing wound through the eye ball causes the upregulation of basic fibroblast growth factor (FGF-2), brain-derived neurotrophic factor (BDNF) and ciliary neurotrophic factor (CNTF) in a localized region which will protect the retina from degeneration due to light damage (Wen et al., 1995) or hereditary retinal disease (Faktorovich et al., 1990).In addition, neurotrophic factors such as CNTF, BDNF, FGF-2, and interleukin 1α have been injected into the eye directly (Faktorovich et al., 1990(Faktorovich et al., , 1992;;LaVail et al., 1992) or via viral vector (Cayouette and Gravel, 1997;Lau et al., 2000), or encapsulate cell technology (Tao et al., 2002) delivery systems as a treatment for degenerative retinal diseases.

Ciliary nerve trophic factor (CNTF)
Of the neurotrophic factors that have been tested, CNTF has progressed the furthest to clinical trials.Early experiments showed CNTF to have a neuroprotective effect on rod photoreceptors, followed by a series of studies that demonstrated similar effects in mice, rats, cats, and dogs (see review (Wen et al., 2012)).These studies promoted a clinical trial using encapsulated cell technology with CNTF for human retinal degeneration and demonstrated the safety of this delivery approach and the targeted levels of CNTF output (Sieving et al., 2006).However, long-term studies in RP patients have not reported functional benefits with this treatment (Birch et al., 2016).Early findings in RP mice and rat models showed preserved photoreceptor structure, but decreased ERG amplitudes after AAV-mediated CNTF gene delivery (Liang et al., 2001).Further research revealed that CNTF can reduce the levels of rhodopsin and transducin proteins, and decrease photoreceptor outer segment length while increasing rod arrestin (Wen et al., 2006).These results emphasize the importance of functional testing in vetting neuroprotective approaches for retinal disease as retinal function testing provides essential information about whether the intervention will provide functional, and therefore, meaningful improvements to the patient.While caution is warranted in using CNTF as a neuroprotective agent in the retina, the transient deconstruction of photoreceptors by CNTF has been used to increase delivery efficiency of gene therapy in an achromatopsia model (Komaromy et al., 2013).

Brain derived neurotrophic factor (BDNF)
In glaucoma, BDNF levels decrease in the retina (Kimura et al., 2016).Since BDNF does not cross the blood-brain barrier, potential therapies have targeted activation of the BDNF receptor, TrkB.TrkB agonists are showing positive results in promoting RGC survival in acute and chronic models of glaucoma (Bai et al., 2010;Hu et al., 2010).Activation of BDNF-TrkB signaling through genetic means in animal models or through gene therapy demonstrate RGC survival (Kimura et al., 2016).Furthermore, BDNF release and therefore synaptic strength can be promoted through repeated neuronal activation, as might be generated with visual stimulation (Mui et al., 2018).Future studies are needed to determine the feasibility of such approaches for their ability to enhance neuronal function and structure with progression of disease.

Other growth factors
DR could benefit from a number of different growth factors.Anti-VEGF therapies are clinically used to control neovascularization in late stages of disease.However, use of growth factors in earlier stages, prior to vascular changes may promote long-term neuronal and vascular function.PEDF was first discovered in the RPE where it is secreted to function as a neurotrophic, anti-angiogenic, and anti-inflammatory agent (Zhu and Zou, 2012).Many studies are underway to evaluate PEDF as a potential treatment for DR (see review (Whitmire et al., 2011)).Additional growth factors of interest include insulin-like growth factor (IGF), NGF, and BDNF (Whitmire et al., 2011).More research is needed with these neurotropic factors to determine their potential efficacy to slow the early stages of neuronal dysfunction and prevent vision loss.

Treating retinal disease with neurotrophic factors
While the clinical efficacy of CNTF for photoreceptor degenerations may be questionable (Birch et al., 2016), the number of publications on growth factors and the retina has increased at a rapid rate for the last 20 years from 160 in 1995 to over 850 publications in 2015 (PubMed search October 2017).Thus, interest in use of growth factors as neuroprotective interventions to the retina continues to grow.The greatest challenges for exogenous delivery of growth factors are to determine the proper dosing and delivery of each neurotrophic factor to provide the optimal preservation.It is not clear that the same dose will be ideal for all retinal cell types.The proper timing of delivery has also not been established.Perhaps there are specific periods during the progression of retinal disease when growth factors would be most effective.For instance, if BNDF was used to target neuronal synaptic connections, perhaps delivery prior to peak periods of neuronal remodeling would be most beneficial and we cannot be certain if the best site of delivery is the cell bodies or the axon terminals.Thus, more research is needed on the pathophysiology of each retinal disease to better understand when and at what locations interventions would provide the most benefit.Additionally, as suggested in the early work by LaVail et al. (1992), combinations of neurotrophic factors may provide the most benefit, an area that has been understudied.

Bringing neurotrophic factors to the clinic
As detailed above for CNTF, delivery of neurotrophic factors is a major hurdle for translation.Due to short half-lives and the inability to easily cross the blood-brain or blood-retina barrier, delivering neurotrophic factors to the retina in the optimal levels remains a major challenge.Potential solutions include the use of cell delivery methods (like CNTF release from encapsulated cell implants), receptor agonists that cross the blood retina barrier (Bai et al., 2010;Hu et al., 2010), or repeated microinjections (Patel et al., 2012).In addition to the studies described above to determine the most optimal dose, timing and tissue target, drug delivery experiments need to be performed in parallel to have delivery options available to translate these neuroprotective agents into the clinic.

Background
Exercise is a rehabilitation approach that has been shown to be protective and neuroregenerative in a number of injury models and diseases (Speisman et al., 2013;Svensson et al., 2015).In animal models, exercise enhances performance on tests of learning and memory (Anderson et al., 2000;Radak et al., 2001;van Praag et al., 1999), promotes hippocampal neurogenesis (van Praag et al., 1999), and reduces neuronal death while increasing axon regeneration after peripheral nerve transection (Wood et al., 2012).The protective effects of exercise have also been observed in the clinic, with functional improvements in patients with stroke (Mang et al., 2013), Alzheimer's disease (Groot et al., 2015;Vreugdenhil et al., 2012), amyotrophic lateral sclerosis patients (McCrate and Kaspar, 2008), Parkinson's disease (Amano et al., 2013), and diabetes (Anderson-Hanley et al., 2012;Billinger et al., 2017;Cooper et al., 2016;Nuhu and Maharaj, 2017;Sjoberg et al., 2017), as well as cognitive function gains in healthy adults (Chaddock et al., 2010;Erickson et al., 2011;Griffin et al., 2011;Ploughman, 2008;Voss et al., 2013b).More recently, exercise has been shown to be protective in animal models of retinal disease, including retinal degeneration (Hanif et al., 2015;Lawson et al., 2014), glaucoma (Chrysostomou et al., 2014), and diabetic retinopathy (Allen, 2015a;Kim et al., 2013).Visual exercises or training (Matteo et al., 2016;Sabel et al., 2011b) and environmental enrichment (Barone et al., 2014) represent an additional field of neuroprotection, with studies showing cortical plasticity and improvements in vision.For the purposes of this article, we have focused on general physical exercise as opposed to modality specific training and rehabilitation.

Mechanisms of action
While exercise likely acts via multiple mechanisms, a number of studies identify the BDNF/TrkB signal transduction pathway as a mediator of exercise protection (Chen et al., 2017;Chrysostomou et al., 2016;Griffin et al., 2011;Hanif et al., 2015;Lawson et al., 2014;Leckie et al., 2014;Marosi and Mattson, 2014;Ploughman et al., 2007;Voss et al., 2013a).BDNF has been reported to increase both locally and systemically after exercise (Griffin et al., 2011;Ploughman et al., 2007;Tonoli et al., 2015).Rodents given exercise treatment after light-induced retinal degeneration showed higher levels of BDNF in the retina and hippocampus, and in serum (Lawson et al., 2014).In an increased ocular pressure injury model, a post-injury decrease in BDNF was not observed in exercised animals (Chrysostomou et al., 2016).
Both BDNF and TrkB receptor expression have been observed in the retina, with BDNF localized to cone outer segments, amacrine cells, retinal ganglion cells, and Müller cells (Avwenagha et al., 2006;Di Polo et al., 2000;Fujieda and Sasaki, 2008;Seki et al., 2003), and TrkB to the RPE, outer plexiform layer, inner nuclear layer, inner plexiform layer, retinal ganglion cell layer, and Müller cells (Asai et al., 2007;Saito et al., 2009).To confirm TrkB pathway involvement, exercise experiments have utilized ANA-12, a TrkB specific inhibitor that blocks BDNF from binding and can be given systemically as it passes through the blood-brain barrier (Cazorla et al., 2011a(Cazorla et al., , 2011b)).Blocking TrkB pathway activation using ANA-12 has been shown to abolish the protective effects of exercise in both hereditary and light-induced retinal degeneration models (Fig. 6) (Hanif et al., 2015;Lawson et al., 2014), to block exercise's protective effects on both retinal and cognitive function in a Type I diabetes model (Allen, 2015a), and to prevent exercise-mediated protection of retinal ganglion cell function in a glaucoma model (Chrysostomou et al., 2016).ANA-12, as well as another TrkB inhibitor (K252a), have also been shown to overturn the protective effects of exercise in stress-induced dendritic spine loss (Chen et al., 2017), in a rat model of Parkinson's disease (Real et al., 2013), and in healthy prenatal rat pups (M Akhavan et al., 2013).Additionally, exercise's protective effects after peripheral nerve injury did not occur in BDNF knockout mice (Krakowiak et al., 2015), lending further support to BDNF's involvement.
Other, non-TrkB-mediated mechanisms could be involved in exercise-mediated protection as well, such as inflammation (Svensson et al., 2015) and oxidative stress (Kim et al., 2015).Exercise's actions on the whole body system to increase blood flow and modulate VEGF and its receptors may also play a role in its protective effects (Stevenson et al., 2018).Exercise has been shown to elevate dopamine receptor binding potential (Fisher et al., 2013) and reduce clinical symptoms in Parkinson's patients (Amano et al., 2013), and to improve performance on a discrimination task through effects on the striatal dopamine system (Eddy et al., 2014), so it is also possible that the protective effects of exercise on retinal disease is acting through a dopamine-mediated mechanism.Further, neurotrophins 3 and 4 also bind to and activate the TrkB receptor and could play a role in TrkB-mediated exercise protection (Reichardt, 2006).Future studies are needed to identify  (Hanif et al., 2015).
exactly where BDNF is acting and how TrkB pathway activations protects against retinal disease and injury.

Research in animal models
The protective effect of exercise in retinal disease and injury models is a relatively recent finding.Exercise reduces retinal dysfunction and photoreceptor loss in a light-induced retinal degeneration model (forced treadmill exercise) (Chrenek et al., 2016;Lawson et al., 2014).Mice exercised on a treadmill for 5 days/week for 2 weeks prior to light damage and one week after showed 2.3x greater ERG b-wave and 2x more photoreceptor nuclei.In order to establish that the forced nature of the exercise wasn't the cause of protection due to stress hormones, the rd10 hereditary model of retinal degeneration were exposed to voluntary wheel running from birth (dam had wheel and then weaned into cage with own wheel) (Hanif et al., 2015).Exercised rd10 mice showed 2.3x greater spatial frequency thresholds and 2.1x more photoreceptor nuclei (Fig. 6).
In an STZ-induced Type I model of diabetes, treadmill exercise reduced VEGF expression (M Akhavan et al., 2013) and apoptosis in the retina (Kim et al., 2013;M Akhavan et al., 2013).Treadmill exercise was also shown to reduce spatial frequency and contrast sensitivity deficits by 1.2-1.5x,reduce delays in ERG implicit times by 1.1x, and reduce cognitive deficits measure by a Y-maze task (Allen, 2015a).In these experiments, rats were confirmed to be diabetic before beginning the exercise program to be more clinically relevant to the human condition regarding the timing of the exercise intervention.Interestingly, the effects of exercise on diabetic retinopathy seem to be independent of an exercise effect on blood glucose, as the diabetic active and inactive groups exhibited similar blood glucose levels (Allen, 2015a).
In a model of increased ocular pressure injury with aged animals, swimming exercise was shown to reduce retinal ganglion cell death and dysfunction (Chrysostomou et al., 2014(Chrysostomou et al., , 2016)), increase synapse preservation (Chrysostomou et al., 2016), and reduce gliosis and macrophage activation (Chrysostomou et al., 2014).Exercise also reduced oxidative stress in the retinas of naturally aging mice (Kim et al., 2015).

Research in humans
There is little research on exercise as a treatment in clinical retinal disease, and the papers that exist are retrospective studies that correlate exercise history with disease prevalence.Specifically, a history of an active lifestyle was associated with a reduction in AMD in a group of men and women followed for 15 years (Knudtson et al., 2006), and in women doing physical activity (McGuinness et al., 2016).Additionally, minimal physical activity is associated with increased macular drusen (a sign of early AMD) (Munch et al., 2013) and increased the risk of DR (Loprinzi, 2016).Evidence that exercise benefits glaucoma is demonstrated by reduced incidence of glaucoma in men running long-distances (Williams, 2009) and protective effects of physical activity for open-angle glaucoma (Yip et al., 2011)[see (Zhu et al., 2018) for recent review].Controlled prospective studies that measure quantifiable outcomes in retinal disease patients in response to exercise treatment are desperately needed.

Applying exercise therapy to the retina
Research on exercise treatment in the retina is promising, but there are still questions to answer as we move forward.Specifically, optimal exercise regimens would need to be determined, with consideration for speed, duration, and number of times per week, as well as the type of exercise (aerobic exercise vs. interval training).It is possible that exercise may show greater protection in some retinal diseases over others.Specifically, exercise may show an increased benefit in diseases involving an intersection of retinal damage and systemic risk factors, for example, DR in which exercise could benefit insulin resistance and systemic hyperglycemia, or glaucoma in which exercise could have beneficial effects on high blood pressure.Research on exercise effects in the retina in humans is very limited, and clinical trials in this area are needed.

Bringing exercise therapy to the clinic
Exercise is an appealing treatment option because anyone can do it without a prescription (although it is recommended to have a doctor sign off on any new exercise program).Exercise does not require patients to take a drug or have a surgery and it does not need to be approved by the FDA.Despite the health benefits, getting people to exercise is no easy feat.However, it is likely that the duration and intensity of exercise needed to achieve functional differences is less than the general public might think.For example, in our experiments with diabetic rats, we implemented an exercise paradigm of 30 min of treadmill exercise, 5 times a week, at a speed that could be classified as "purposeful walking."Additionally, others have found that 2.1 to 2.5 sessions a week is enough to confer a measurable difference in fracture risk (Kemmler et al., 2017), and that high intensity training can be used both to shorten workout duration and increase exercise enjoyment (Heinrich et al., 2014).Patients might be more motivated to incorporate exercise as part of a treatment plan if informed about the increases in health and protection that can be achieved by adding even small amounts of exercise into their daily life.
Furthermore, implementing an exercise program may be more difficult for patients who already have diminished visual function and have progressed to later stages of disease.Finding exercise equipment that is accessible for low vision patients, and also for patients with disabilities, is an important consideration when translating exercise to the clinic.
Another important consideration is the type of exercise paradigm.Evidence from animal models shows that voluntary exercise is better than forced treadmill exercise (which proved harmful) in a mouse model of colitis (Cook et al., 2013).However, in a focal ischemia model, intense treadmill running resulted in greater increases in BDNF than voluntary wheel running (Ploughman et al., 2007).Additionally, gender differences exist in the efficacy of different exercise regimens: high intensity interval training seems to be more protective in females, while continuous walking seems to be more protective in males in a mouse model of peripheral nerve injury (Wood et al., 2012).
Finally, in a study comparing physical activity in patients with RP, patients with higher levels of physical activity not only reported better visual function, but also greater quality of life (Levinson et al., 2017), indicating that implementing exercise treatment in visual disorders can have far reaching benefits outside the realm of vision.

Background
Electrical stimulation therapies are becoming increasingly popular for rehabilitation, from neuromuscular stimulation to strengthen muscles and synaptic connections (Ethier et al., 2015;Pette and Vrbova, 2017), to neurostimulation devices that target different central or peripheral nervous systems to modulate neuronal activity and provide therapeutic effects (Edwards et al., 2017).Neurostimulation for neurological disorders are varied with deep brain stimulation applied for Parkinson's disease, essential tremor, dystonia, and obsessive-compulsive disorder; responsive neurostimulation for epilepsy; vagus nerve stimulation for epilepsy and depression; and spinal cord stimulation for chronic pain and functional restoration (Edwards et al., 2017).
For vision, two types of neurostimulation are being used: prosthetics to activate visual circuits by stimulating retinal neurons directly or by stimulating the visual cortex, thereby restoring visual function (Bosking et al., 2017;Cheng et al., 2017), and low level electrical stimulation that provides neuroprotection (Pardue et al., 2014;Sehic et al., 2016).The first report of electrical stimulation having neuroprotective effects in the eye appears to be a study by Morimoto et al. showing benefits to transected RGCs (Morimoto et al., 2002).This was followed by evidence of electrical stimulation having neuroprotective effects on photoreceptors when Chow et al. reported that subretinal microphotodiode arrays implanted into patients with RP for restoration had visual improvements in retinal regions far from the site of the electrical device (Chow et al., 2004).Current approaches for low level electrical stimulation use mainly minimally invasive techniques in which an electrode(s) is placed near or on the surface of the eye with the reference electrode located on the retina, or in another location like the mouth or head which spreads the current flow from the active electrode (Sehic et al., 2016).The current levels for low level electrical stimulation therapy are in the range of 100-1000 μA and are unlikely to activate retinal neurons at the same level as light stimulation (Pardue et al., 2014).
Additionally, electrical stimulation therapy promotes survival of neurons by downregulating IL-1β, TNFα, and Bax, and upregulating Bcl-2 (Ni et al., 2009).It is possible that different growth factors may be released depending on the location of the electrodes [subretinal electrical stimulation increased FGF2 (Ciavatta et al., 2013) while whole eye electrical stimulation increased BDNF and FGF2 (Hanif et al., 2016)] or whether the retina is healthy or diseased (Willmann et al., 2011).Some evidence suggests that transcorneal electrical stimulation may alter vascular flow which could produce activation of retinal neurons (Morimoto et al., 2014).

Research in animal models
Current evidence of the neuroprotective effects of electrical stimulation on retinal disease has used photoreceptor degeneration and glaucoma models.In the Royal College of Surgeons (RCS) rat, a common model of photoreceptor degeneration, subretinal or transcorneal electrical stimulation preserved retinal function from 1.6 to 4.3x and increased ONL thickness from 2.3-3.4x(Ciavatta et al., 2013;Morimoto et al., 2007;Pardue et al., 2005).Transcorneal electrical stimulation benefited the P347L rabbit with ERG amplitudes 1.4x larger and ONL thickness 1.8x thicker (Morimoto et al., 2012).Using whole eye electrical stimulation, the P23H rat shows ERG amplitude benefits from 1.6-2.2xgreater and a-wave sensitivity increased 11.6x (Rahmani et al., 2013).Finally, in a light-induced retinal degeneration model, whole eye stimulation preservation of ERG amplitudes ranged from 11.5-13x greater with repeated stimulation (Ni et al., 2009).The frequency of the stimulation varied in these studies from once a week to every three days.With the evidence that gene expression quickly decreases after stimulation (increased at 1hr and normal at 1 day) (Hanif et al., 2016), the dosing of electrical stimulation therapy may be critical.Additionally, repeated dosing greatly increased the effect as a  (Sehic et al., 2016).
single stimulation for one hour after light induced retinal degeneration had minimal effects (ERG amplitude of 1.1x and ONL thickness 1.3x greater) (Schatz et al., 2012) compared to two weeks of stimulation in the same model (ERG amplitude 11.5-13x greater) (Ni et al., 2009).
Electrical stimulation also protects RGCs during mechanical injury or elevated IOP.Using rats with optic nerve transection or crush, transcorneal electrical stimulation increased RGC density by 1.3x (Morimoto et al., 2005) and increased VEP amplitude by 2.4x (Miyake et al., 2007).RGC death due to ischemia from elevated IOP is also delayed using transcorneal electrical stimulation with 1.2x greater RGC density and ERG amplitudes 2.2x greater (Wang et al., 2011).Furthermore, treating with repetitive transorbital alternating current stimulation following optic nerve crush in rats induced dendritic pruning in surviving neurons, potentially protecting them from excitotoxicity, although they did not produce functional signals (Henrich-Noack et al., 2017).In a model of nonarteritic ischemic optic neuropathy, electrical stimulation sustained RGC function measured by scotopic threshold response with 1.9x greater amplitudes than sham and 1.5x greater RGC density (Osako et al., 2013).However, electrical stimulation sessions with long intervals (> 10 days) provided only short-term, temporary benefit to RGCs after optic nerve crush (Henrich-Noack et al., 2013b).Thus, low level electrical stimulation benefits RGCs, with repeated stimulations creating more optimal preservation (Morimoto et al., 2010).

Research in humans
The benefits of electrical stimulation to retinal function have already been translated to humans.Aside from the first study showing neuroprotective effects in visual function from implantation of a subretinal implant (Chow et al., 2004), most studies have applied minimally invasive methods of stimulation, such as transcorneal or whole eye for neuroprotection of retinal function and structure.
Applying transcorneal electrical stimulation to patients with RP provided an increase of 1.3x in b-wave amplitude and a 1.1x increase in visual field area (Schatz et al., 2017).Stimulation was provided weekly for one year at two different levels of stimulation (based on the percentage of current to elicit a phosphine response).Results for AMD have reported mixed results using transcutaneous stimulation.In patients with dry AMD, transpalpebral stimulation was applied twice daily for 5 days (Anastassiou et al., 2013).While the majority of patients receiving stimulation had improved visual acuity, only contrast sensitivity was significantly different from the sham group at 4 weeks after stimulation.Similarly, transpalpebral stimulation (150 μA) applied weekly for 3 months maintained visual acuity in patients with dry AMD, but did not benefit patients with wet AMD (Chaikin et al., 2015).However, in another study with higher current (800 μA) applied transcutaneously 4 times daily, the benefits to dry and wet AMD patients did not reach statistical significance (Shinoda et al., 2008).
In regards to RGCs, one study reported improvements with transcorneal electrical stimulation in patients with nonarteritic ischemic optic neuropathy and traumatic optic neuropathy (although the results were only compared to baseline and not sham controls) (Fujikado et al., 2006).Furthermore, transorbital stimulation in patients with optic nerve lesions showed improved visual field size and visual acuity (Fedorov et al., 2011;Sabel et al., 2011a) or increase detection ability within the visual field (Gall et al., 2011).Similarly, a recent multicenter, randomized, double-blind, sham-controlled trial of transorbital alternating current stimulation in patients with optic neuropathy showed improved visual fields and increased thresholds in static perimetry (Gall et al., 2016).Other studies have reported benefits of electrical stimulation on the retinal vascular that could promote neuronal survival, including increased chorioretinal blood flow with electrical stimulation (Kurimoto et al., 2010) or improved function with retinal artery occlusion (Inomata et al., 2007;Naycheva et al., 2013;Oono et al., 2011).The animal studies performed with electrical stimulation provide solid evidence that this neuroprotective strategy can provide benefit to retinal neurons and shows potential mechanisms through neurotrophic factors and other survival pathways (Fig. 7).However, the optimal stimulation paradigms (current levels and frequency) has not yet been determined for each cell type and/or retinal disease.Furthermore, there is a lack of research on potential combination therapies in which electrical stimulation could be paired with other neuroprotective strategies.For instance, Osaka et al. report on the benefits of progesterone or transcorneal electrical stimulation in a rat model of nonarteritic ischemic optic neuropathy and determined that the two treatments may be complimentary: progesterone reduces edema and electrical stimulation preserving RGC function and structure (Osako et al., 2013); however, the two treatments were not tested in combination.Another exciting use for electrical stimulation is to improve survival of cells prior to or just after retinal transplantation (Manthey et al., 2017).

Bringing electrical stimulation therapy to the clinic
The lack of consensus for the optimal stimulation parameters and even placement of electrodes on the orbit and/or eye is reflected in the various human studies.It is currently difficult to directly compare the results from these studies and determine if multiple labs are getting the same results because each research group is testing a slightly different stimulation paradigm.
A major challenge of applying electrical stimulation therapy is the proper treatment window.Some studies have used patients with longterm vision loss from retinal artery occlusion and found some benefit (Inomata et al., 2007;Oono et al., 2011).Whether these results could be improved by treating earlier, or optimizing the stimulation paradigms is not known.Additionally, some of these benefits could be produced by activation of brain regions (Bola et al., 2014;Henrich-Noack et al., 2013a, 2017;Sergeeva et al., 2015).
Overall, the safety of electrical stimulation therapy is good.The only side effect reported in the current studies has been contact dermatitis (Shinoda et al., 2008) or corneal irritation with the transcorneal electrode (Inomata et al., 2007).These issues are easily addressable, although not to be ignored if repeated stimulation is needed for optimal results.
One issue that is not addressed in most manuscripts of electrical stimulation is the stability of the device that delivers electrical stimulation.The current levels (< 1 mV) are very low to induce neuroprotection and most commercial instruments do not provide much technical accuracy at this minimal current level.Since some studies report benefit with 150 μA (Chaikin et al., 2015), while others report no benefit with 800 μA (Shinoda et al., 2008), the exact level of current may be critical to induce protection.Related to this issue is the fact that electrical stimulation is minimally invasive and even applied transorbitally on the skin in some studies.This provides the opportunity for devices to be easily marketed for neuroprotection prior to rigorous testing.Thus, the consumer needs to be extra careful when examining such products to be sure that the device has been scientifically tested.To date, the FDA has not approved electrical stimulation devices specifically for neuroprotection of the eye, although several are being used off-label (Bikson et al., 2013).

Early detection of retinal disease and monitoring progression
Bringing neuroprotective treatments to the clinic requires showing that the intervention has protective effects in rigorous clinical testing.There are three major hurdles to achieving this goal.The first is that retinal diseases progress very slowly, over many months and years.Neuroprotective agents are expected to slow or halt the progression of the retinal disease.Thus, the study design needs to follow the subjects for a sufficient length of time such that control vehicle/sham groups show measureable declines in visual function/structure compared to treatment groups.For most retinal diseases, this can take more than 5 years, longer than typical grant cycles.Thus, clinical trials need to be carefully developed with consideration for the patient population being studied and potential diversity in the disease state (subtypes, stages, etc) that may increase variability and dilute positive effects.
The second major challenge is the need to start neuroprotective treatments at the earliest stage of disease when they will be most effective in delaying progression.In experiments with animal models, it is convenient to start a treatment with the initiation of the retinal insult (e.g.light-induced retinal degeneration).However, patients typically are not seen in the clinic until they are experiencing visual loss and this may be months to years after the initiation of the retinal disease.When clinical trials for neuroprotective agents have to begin at these mid-to late stages of the retinal disease, the chances of significantly halting vision loss is minimal, as suggested in Fig. 8.In the first stage of retinal disease, adaptive changes like oxidative stress and neuronal dysfunction occur.During this stage, various endogenous defense mechanisms become active to maintain homeostasis in the tissue.As the disease progresses, the normal defense mechanisms become overwhelmed, oxidative stress levels increase and both neuronal and vascular dysfunction are present.In this second stage, it is possible to deliver neuroprotective agents to reverse the pathology and prevent vision loss.If no therapeutic intervention is present, retinal diseases progress to late stage pathology with neuronal loss, vascular defects like vascular leakage and neovascularization, and eventual blindness.In this late stage of the disease, the changes are irreversible.Thus, there is an urgent need for early detection of all retinal diseases.The time spent at each of these stages can vary depending on the mechanisms underlying each retinal disease.The inability to detect retinal disease at the earliest signs of neuronal and vascular dysfunction stops patients from getting neuroprotective agents that could delay or prevent late stage vision loss.Furthermore, the lack of sensitive screening methods to detect early stage disease has likely contributed greatly to the sparsity of neuroprotective agents that have shown positive effects in clinical trials.
The third major challenge to translation of neuroprotective strategies to the clinic is the need for sensitive tools for retinal screening and monitoring of disease.Current technology showing promise in this area includes imaging and functional testing.Imaging technologies like adaptive optics scanning laser ophthalmoscopy (AO-SLO), optical coherence tomography (OCT) and optical coherence tomography angiograms (OCT-A) are providing unprecedented opportunities to non-invasively track retinal structure and blood flow in vivo.AO-SLO is capable of imaging single photoreceptors in vivo (Roorda and Williams, 1999) and other retinal layers/structures (Arichika et al., 2014;Takayama et al., 2013).Several research groups are using this technology to characterize photoreceptor changes during retinal diseases like RP and AMD (Ratnam et al., 2013;Sun et al., 2016;Zayit-Soudry et al., 2013, 2015), as well as erythrocyte aggregates in retinal capillaries in DR (Arichika et al., 2014) and nerve fiber bundles in glaucoma (Takayama et al., 2013).To date, AO-SLO instruments are not readily available for all vision researchers, but are limited to experts in adaptive optics which hinders their use as outcome measures in intervention studies.As the technology becomes more accessible to all and further progress is made to image other layers of the retina, including vascular layers, AO-SLO has the potential to greatly impact how disease is detected and monitored in the clinic.
OCT provides in vivo imaging of the eye and retina in cross-section, allowing for the individual retinal layers to be visualized (Huang et al., 1991).OCT has quickly become a standard of care in clinics around the world as patient retinal structure can be non-invasively imaged and monitored.OCT is routinely used to detect macular edema, retinal detachment, ocular tumors, and to monitor retinal thickness.With the sensitivity and resolution of OCT continually increasing over the last several years, OCT will continue to be a mainstay of the clinical exam.However, structural changes in the neural retina are only going to be observed after cell death.Thus, using OCT for detection of a disease may miss the earliest functional changes.Several innovative improvements to OCT are providing novel assessments of the retina without dyes or other contrast agents, providing non-invasive imaging of retinal vessels.For instance, OCT-A is a modified version of OCT that provides information on retinal and choroidal blood flow (Hagag et al., 2017).Retinal blood flow changes have not been fully characterized for most retinal diseases due to the lack of in vivo assessments.However, the development of OCT-A has already provided important insights into blood flow changes in AMD and DR.As characterization of the retinal and choroidal blood flow for retinal disease becomes well-established, OCT-A will be well-poised as an in vivo technique to detect and monitor retinal disease.Functional tests that do not require imaging may be less expensive and also offer the potential to probe neuronal dysfunction in the earliest stages of retinal disease.Visual acuity charts (Snellen, Early Treatment of Diabetic Retinopathy Study), are often the go-to functional test for clinical trials along with near-threshold and super-threshold visual field tests.However, visual acuity assessed in this manner is not very sensitive to changes in visual function and does not offer the earliest detection of retinal dysfunction.Thus, better, more sensitive methods for assessing functional vision are needed.The electroretinogram (ERG) is available clinically and is often used in pre-clinical studies to measure retinal function.Full-field ERG recordings produce small amplitudes that have minimal sensitivity to detect small differences between timepoint or subjects groups.For instance, in the docosahexaenoic acid (DHA) and X-linked RP clinical trial (NCT00100230), the primary and secondary outcome measures were cone and rod ERGs, which did not show any differences.The third outcome measure was loss of peripheral visual fields, and this showed high significance (p < 0.001).While ERG amplitude is not always useful for detection of differences, ERG implicit time is much less variable and may offer more sensitivity in diseases like DR, in which oscillatory potentials delays are observed with dim flash stimuli (Aung et al., 2013(Aung et al., , 2014)).Additionally, the development of hand-held ERG systems avoids the cumbersome table-top Ganzfeld dome used in most clinics to present the flash stimuli, making the hand-held ERG more feasible as a screening tool (Maa et al., 2016).Another functional test showing promise for early detection of AMD is dark adaptometry, which measures the speed of rod and cone pathway adaptation to darkness using a shortened protocol that optimizes the bleaching intensity (Jackson et al., 2014).More research is needed to determine if dark adaptometry should be considered for future clinical trials as an outcome measure with potential sensitivity to detect small changes in retinal function.

Delivery of neuroprotective agents to the retina
Another challenge for treatment of disease is delivery of the intervention to the target tissue.The eye has many advantages for drug delivery since it is easily accessible (not encased in the body like the heart or liver), has its own blood-retinal barrier (providing potential isolation of the eye from the rest of the body), and is easily monitored through imaging and other outcomes (intraocular pressure, visual function, pupillary reflex, etc.).For non-drug strategies, the eye and visual system also have advantages.Electrical stimulation to the eye is easily applied via transcorneal or transorbital stimulation (Sehic et al., 2016).While we have focused on the neuroprotective effects of whole body physical exercise, the visual system can be "exercised" directly with visual training, as discussed below.
In this review, we have focused on several drugs with neuroprotective properties that could be delivered systemically (injection or orally) or locally (injection or sustained release).Modes of local intraocular delivery include intravitreal injections, sub-tenon's injections, suprachorical microneedle injections, or topical drops (Edelhauser et al., 2010).Neuroprotective strategies will likely need to be taken frequently and chronically (particularly if started early in disease).Thus, other delivery methods that have better safety profiles, can potentially be self-administered and have minimal costs should be considered.Intravitreal injections are routinely done in clinics around the world.Bioavailability of many drugs to the retina is excellent; however, this method does have risks for vitreous hemorrhage, retinal detachment, or infection (Edelhauser et al., 2010).With sub-tenon's injections, the needle is placed between the sclera and the more external tenon's capsule.Thus, bioavailability of the drug is decreased due to the need for absorption through the sclera.Microneedle injections into the suprachoroidal space offer a promising new delivery method (Patel et al., 2012).In this approach, the needle is only long enough to penetrate the sclera and enter the suprachoriodal space located between the choroid and sclera, decreasing many of the risks of intravitreal injections.A microneedle injection could be performed with less risk than an intravitreal injection, perhaps even allowing for self-administration which would greatly increase the availability of this approach for drug delivery.Finally, while topical eye drops are an appealing delivery method for localized delivery to the eye because of ease of use, the absorption of most drugs and molecules through the anterior layers of the eye to the retina is limited (Edelhauser et al., 2010).However, several innovative advances in topical delivery improve drug solubility and cell entry, such as polyguanidilyated translocators, nanoparticles that can quickly penetrate the corneal epithelium layer (Durairaj et al., 2010), or hydrogels for more sustained delivery (Yang et al., 2012).Sustained release and nanoparticles can increase bioavailability of drugs delivered through intravitreal injections, as demonstrated by several studies using microspheres (Fernandez-Sanchez et al., 2017) and cell encapsulating technologies (Tao et al., 2002).
Physical exercise is the only neuroprotective strategy discussed here that cannot be localized to just the eye.However, the pluripotent nature of exercise and its benefit to multiple systems of the body is part of its appeal as a neuroprotective strategy.Furthermore, exercise is particularly easy to deliver since exercise does not need FDA approval and can be done anywhere with minimal equipment (even walking may provide benefit).In addition, this neuroprotective strategy provides a self-controlled and self-managed approach for the patient that provides some sense of control while functional vision is being lost.That said, visionspecific exercise such as increased visual stimulation or training exercises that increase the visual field has not been fully explored as a potential neuroprotective intervention.Increased visual stimulation during the critical period increases spatial frequency thresholds to above normal (Prusky et al., 2008), however, more research is needed on whether this type of visual stimulation provides benefit during retinal disease when it occurs in adulthood.A report using visual stimulation in several rat models of photoreceptor degeneration showed that neuroprotective effects are not uniform across models; visual stimulation slowed the rate of degeneration in some models, but increased it in others (McGill et al., 2012).However, vision training provided some benefit to adults as demonstrated by improved visual fields in patients with glaucoma after 3 months of daily 1 h vision restoration training (Sabel and Gudlin, 2014).Since these experiments originated in humans, it is not known if an increase in neurotrophic factors or synaptic connections underlie results.

Pharmacokinetics of neuroprotective strategies
Another challenge for applying neuroprotective treatments to the clinic is determining the proper dosage.While many of the strategies discussed here have very good safety profiles (TUDCA, progesterone, low level electrical stimulation, exercise), the exact dosage to achieve optimal efficacy is not known.This is particularly challenging when most testing starts in pre-clinical studies in rodent models of retinal disease.Rodents have different metabolic rates and body mass that makes pharmacokinetics comparisons to humans difficult.Allometric equations exist to estimate a reasonable dose conversion between rodents and humans, but the accuracy is questionable (Lucas et al., 2016).Thus, calculating optimal doses is another area that hinders the success of neuroprotective strategies in human trials.Funding for clinical trials is highly competitive and only the most promising treatments are going to be pushed forward.The misfortune of having negative results from a clinical trial could result in doubt of an otherwise promising treatment strategy, such as that in the progesterone Phase III clinical trial (Section 3.4.2) (Howard et al., 2017).Furthermore, translating exercise from rodents to humans has similar challenges.Many studies are performed on the protective effects of exercise on cognitive health.Yet, a systematic study to determine optimal dosing in humans has yet to be performed given the substantial effort and cost of such a study.We are missing potential opportunities to slow retinal and systemic disease if we don't invest in studies that elucidate optimal dosing before moving forward into Phase II and III clinical trials.

Lack of implementation science for successful translation
Another issue that we feel is impeding the progress of neuroprotective strategies for retinal disease into the clinic is the lack of studies that include clinicians and patients on the research team.Implementation science is "the study of methods to promote the adoption and integration of evidence-based practices, interventions and policies into routine health care and public health settings" (https:// www.fic.nih.gov/researchtopics/pages/implementationscience.aspx).This field promotes partnerships with key stakeholder groups (patients, providers, organizations, systems and/or communities) to more quickly and effectively adopt treatments into the clinic.Having these novel perspectives from the start can properly inform the research goals so that the treatment is implemented into the clinic in a timely fashion.The traditional research training model provides excellent training for basic researchers to do benchtop research in their labs.This includes excellent pre-clinical research to test neuroprotective strategies in animal models of disease.However, moving these treatment strategies to clinical trials requires knowledge of IRB regulations and the proper framework to test human subjects (including expertise in human vision testing) as well as business and finance partnerships to support the project.Many basic researchers likely feel that highly significant results in an animal model will be the incentive to engage a clinical collaborator who could run the clinical trial.While this may be possible, in the case of neuroprotective treatments, the effects are rather modest in the short-term--with the longer term goals of preserved vision being the most valuable to the patient with retinal disease.Thus, it may be difficult to entice an already busy clinician to take the lead on such a clinical study.This leaves many pre-clinical studies in an intermediate stage of development without a "champion" to bring it to the clinical for the benefit of patients.Thus, we highly recommend considering the clinical trials pipeline early in the research process.This pipeline moves studies from pre-clinical research in animals to clinical studies to community based or implementation studies that move the treatment into routine health care and public health settings.To move treatments to this last stage, the ultimate goal in order to get our treatments to the clinic, a team approach of basic scientists, clinicians, and patients (as well as regulatory specialists and financial backers) is needed in order to design research studies that provide the most meaningful results for rapid translation.These studies may include dosage, toxicity, the proper timing of the treatment for the retinal disease, etc.They may also include whether the treatment is acceptable to the patient.For instance, while intravitreal injections may be acceptable to an ophthalmologist in the clinic, a patient may prefer a self-administered microneedle injection into the suprachoroidal space (if it was available).Likewise, developing exercise interventions for people with visual impairments fails if the patients are not consulted to determine how they are going to exercise and if standard exercise equipment is accessible to visually impaired individuals.

Conclusions: neuroprotective therapies with greatest promise in retinal disease
In comparing the neuroprotective strategies reviewed here, it is tempting to try to determine the most promising candidates for neuroprotection of retinal diseases.This question is not easily answered as there are few strategies that have been tested in the same animal models using the same outcome measures.Even in our own lab, we have not always used the same animal models for photoreceptor degeneration and thus, it is difficult to draw direct comparisons.In the STZ rodent models, we have tested L-DOPA (Aung et al., 2014), TUDCA (Fu et al., 2016) and exercise (Allen, 2015a), although L-DOPA and exercise were tested in STZ-treated rats and TUDCA in STZ-treated mice.However, even with these caveats, it is surprising how similar the fold improvements are in functional measurements: flicker implicit time improved 1.1-1.2x,spatial frequency 1.1x, and contrast sensitivity 1.5x.These values are very similar to the fold improvements reported for TUDCA treatments across several different models for both function (1.1-4x) and structure (1.1-4.5x;Fig. 3), and for progesterone functional preservation.However, it is possible to generate greater levels of neuroprotection in the retina.Exercise produced improvements of 2-2.3x greater function for RP models and electrical stimulation varied from 1.6-4.3xgreater function and 2.3-3.4xgreater structure, to even 12x greater in one study.Thus, the potential for significant preservation is possible with multiple neuroprotective strategies, but it is hard to conclude which strategy is best.The ultimate goal of a neuroprotective strategy is to maintain vision at normal levels which is only possible if treatment is started at the earliest signs of disease.The data in Fig. 5D clearly shows that promise for L-DOPA treatments in DR, as ERG flicker amplitude and implicit time was maintained at normal levels out to 5 weeks post-STZ.Since the exact conversion from weeks of progression in a rat to months or years of progression in a human is not known, we can only speculate on the potential long-term benefit to humans.
Table 1 shows a summary of the research that has been done for each of the three major retinal disease categories addressed here.Each drug shows benefit to multiple diseases due to the pluripotent nature of the neuroprotective strategies and the common final mechanisms leading to cell death.The majority of studies have established potential benefit to retinal diseases, but only L-DOPA, neurotrophic agents and electrical stimulation have been used in prospective human studies of retinal disease, in which the treatment is begun after deficits have been observed.Few animal studies test neuroprotective agents at a timepoint when functional deficits or other disease biomarkers are measurable (Aung et al., 2014;Fu et al., 2016).In fact, some studies have preconditioned the retina by providing the treatment prior to the start of the disease (Boatright et al., 2006;Lawson et al., 2014;Oveson et al., 2011;Woo et al., 2010).The clinical translation of testing treatments given prior to the start of the disease is questionable.While this many provide proof that a particular neuroprotective agent may be suitable to explore further, experiments in which administration of the treatment is after clinically recognizable deficits provides a better understanding of the translational potential.One promising, yet fairly new neuroprotective agent for both photoreceptor and RGC degeneration, is L-DOPA.Dopamine deficiencies may play a larger role in retinal disease than previously understood.Since levodopa is already FDA approved for Parkinson's disease and various dopamine receptor agonists are also on the market, dopaminerelated treatments are an underexplored treatment for retinal neuroprotection.
Furthermore, attention to drugs that are already undergoing clinical trials for other diseases need to be considered for the retina.Large numbers of clinical studies have already been performed or are currently ongoing with TUDCA and progesterone.These treatments hold promise for easy translation-the clinical trials pipeline has already been worked out in regards to safety and efficacy for other neuronal diseases.Thus, more effort is needed to leverage existing studies and determine if visual outcomes could be assessed (particularly in TUDCA treatment of diabetes).
The bulk of neuroprotection studies have been performed in photoreceptor degeneration models, reflecting the numbers shown in Fig. 1.The good news is that photoreceptor survival can be enhanced using all of neuroprotective strategies presented here, although some neurotrophic factors (like CNTF) do not benefit function, and electrical stimulation therapy shows only modest effects, potentially due to the lack of optimization.Reflecting the larger number of pre-clinical studies, there are also more neuroprotective agents that have been tested in humans for retinal degenerations, including TUDCA, neurotrophic agents (CNTF), and electrical stimulation.
Given the large number of people with DR (Fig. 1), it is disappointing that more studies have not been performed to test the efficacy of neuroprotective agents to slow DR progression.One explanation for this sparsity may be the relatively recent discovery that DR also involves neuronal dysfunction at early stages (Antonetti et al., 2006;Aung et al., 2013;Fletcher et al., 2007).Thus, as the neuronal aspects of DR become characterized, the opportunity to test potential neuroprotective agents becomes more feasible.However, as discussed above, a critical component of testing neuroprotective agents for DR is the detection of early neuronal defects using functional tests.
Several studies have evaluated neuroprotective treatments for RGC degeneration.Most studies have been done with electrical stimulation.Currently, only L-DOPA and electrical stimulation have moved to human trials.It is notable that most of these studies have not used the standard glaucoma models, like the Morrison or microbead models that demonstrate high IOP, loss of RGCs, and optic nerve cupping.Instead, the majority of studies have used RGCs in culture or acute models of RGC damage or have simply examined RGC loss in other models.Research in glaucoma animal models may provide information about mechanisms of neuroprotection for RGCs as well as optimization of treatments.

Future directions
The number of people with retinal disease, including AMD, DR, and glaucoma is expected to more than double by 2050 (http://www.nei.nih.gov/eyedata).These statistics predict a healthcare crisis that affects not only the individual with vision loss but also the caregivers and the overall healthcare system.Thus, there is urgency to develop new neuroprotective approaches for retinal disease.We conclude this review with specific recommendations that we feel are imperative to move neuroprotective strategies into the clinic more quickly, thus, preventing vision loss and blindness in patients.
1. Screening tools for early detection of retinal disease need to be developed, tested, and adopted.Not only will these screening techniques open a critical treatment window by which neuroprotective treatments can be applied, they will also identify sensitive assessment tools that can be used for monitoring disease progression and treatment efficacy.We have identified several imaging tools, such as AO-SLO and OCT-A, that hold great promise for screening and detection.However, these instruments need to become commercially available and thus, widely accessible in order to be adopted.Less expensive and simpler screening tools also need to be pursued and tested.2. Neuroprotective strategies need to be tested in animal models using clinically relevant starting points, such as measurable delays in ERG waves or deficits in spatial frequency/contrast sensitivity thresholds.Using such experimental design will aid in determining the clinical potential for translation.3.More research is needed on combination therapies to slow degeneration or function loss, such as TUDCA and electrical stimulation.
Many of these approaches have potentially complimentary mechanisms that could produce synergistic effects, thereby improving the overall efficacy of the approach (although antagonistic effects are possible and need to be tested).In addition, neuroprotective agents can be used in combination with other ocular procedures, like retinal detachment repair (Mantopoulos et al., 2011) and retinal cell transplant (Manthey et al., 2017), to increase the success of such procedures in improving survival of the retinal neurons.4. For photoreceptor degenerations, early detection is imperative for success, as a number of agents already show promise.For diseases like AMD and RP, detection is performed by the patient when visual disturbances are noticed and clinical care sought.The visual system is superbly effective at compensating for small visual defects.Thus, simpler, more accessible self-screening is needed, for example, apps that patients could access on a device that would alert them of a problem in visual function and advise them to get a clinical exam. 5.For DR, more neuroprotective agents need to be tested.Multiple animal models of type I and type II diabetes exist that can be used to develop early detection screening and test the agents reviewed here or others.The development of OCT-A and other functional measurements are predicted to be particularly advantageous to DR detection and treatment.6.For RGC degeneration, glaucoma models, such as the Morrison and microbead models, need to be adopted in order to determine if the neuroprotective agents will provide benefit to a chronic model of RGC loss with elevated IOP that mimics human glaucoma.7.For neuroprotective strategies to reach the clinic in a rapid and efficient matter, it is imperative that teams of basic scientists, clinicians, and patients work together to develop, test, and implement these approaches into the clinic.
Finally, the current review was focused on the three major classes of disease that cause vision loss.However, given the pleiotropic effects of the neuroprotective strategies examined here and their generalized benefit across disease models, other types of ocular disease may also benefit from these approaches.For instance, traumatic injury to the eye and retinal and optic nerve stroke/ischemia share many of the same mechanisms covered here, such as cell death and oxidative stress, and could potentially benefit from the same neuroprotective strategies.Neuroprotective strategies for retinal diseases offer hope that functional vision can be sustained until the end of life, supporting independence and quality of life.

Fig. 2 .
Fig. 2. Functional and structural benefits of TUDCA in rd10 mice at 30 days of age.A) Dark-adapted ERG responses show significantly larger a-and b-wave amplitudes in TUDCA-treated mice, indicating preservation of rod and mixed rod-cone retinal pathways.B) Light-adapted ERG b-wave amplitudes are significantly larger in TUDCA treated mice, indicating preservation of isolated cone function.C) Total photoreceptor cell counts at different locations across the retina are significantly preserved in the TUDCA treated mice.Asterisks indicate significance from Holm-Sidak post-hoc comparisons after finding significant interaction effects with two-way repeated ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001 Mean ± SEM.Modified from(Phillips et al., 2008).

Fig. 3 .
Fig. 3. Relative comparisons for functional and structural improvements in studies using UDCA or TUDCA to treat retinal disease.Functional values are ERG amplitude differences compared to vehicle treated groups.Structural values are retinal neuron counts or layer thickness.LIRD: Light-induced retinal degeneration; Other RD: retinal detachment, Lrat−/− mice and Bardet-Biedl mice; NMDA injection into the eye causes RGC death.Note that the values in Boatright et al. (2006) were collected much earlier after retinal injury than other studies (24 h vs 1 week).

Fig. 4 .
Fig. 4. A rodent model of retinal degeneration (rd1 mice) treated with oral progesterone showed reductions in gliosis and glutamate and increases in glutathione, leading to delays in photoreceptor cell death and a partial preservation of photoreceptor function (ERG response).Used with permission from (Sanchez-Vallejo et al., 2015).

Fig. 5 .
Fig. 5. Daily L-DOPA treatments in diabetic rats (A, B) and mice (C, D) provide significant preservation to visual function (A, B) and retinal function (C, D).C and D are from data collected at 5 weeks post-STZ.Asterisks indicate post-hoc comparisons *p < 0.05, **p < 0.01, ***p < 0.001.The color of the asterisk indicates the treatment groups in which significance was reached.Mean ± SEM.Modified from (Aung et al., 2014).

Fig. 6 .
Fig. 6.Running wheel treatment in rd10 mice protected against visual function loss (A) and photoreceptor cell death (B: rd10 inactive, C: rd10 active at 44 days of age), including cones (arrowheads) and rods (arrows).Treatment with the TrkB pathway inhibitor, ANA-12, blocked the protective effects of exercise on both visual function (D) and cell death (E) at 41 days of age.Asterisks indicate post-hoc comparisons *p < 0.05, **p < 0.01, ***p < 0.001.The color of the asterisk indicates the treatment groups in which significance was reached.Mean ± SEM.Modified from(Hanif et al., 2015).

Fig. 7 .
Fig. 7. Summary of possible neuroprotective mechanisms of transcorneal electrical stimulation.Electrical stimulation preserves both photoreceptors and RGCs while also benefiting other retinal neurons through the release of neurotrophic factors and increasing other survival factors while decreasing apoptotic signals.PG: primary microglia, MCs: Muller cells, GS: glutamine synthetase.See text for more detail.Used with permission from(Sehic et al., 2016).

Fig. 8 .
Fig. 8. Stages of retinal disease that reflect when neuroprotective strategies should be started to provide the most benefit.The black line shows the hypothetical progression of retinal disease through adaptive, early and late pathology.During early pathology, retinal disease is detectable and potentially reversible while irreversible damage occurs in late stage pathology.Starting neuroprotective treatments at the first signs of retinal disease would provide the most benefit in preserving vision.

Table 1
Comparisons of neuroprotective strategies tested in photoreceptor degeneration, diabetic retinopathy, and retinal ganglion cell disease with an emphasis on pre-clinical studies in animal models and human studies in bold text.