Retinitis Pigmentosa is comprised of a class of heterogeneous gene mutations and phenotypes (Van Soest et al., 1999). Mutations can be classified as non-syndromic, affecting the retina in isolation, or syndromic in which retinal degeneration coincides with other disease processes that occur outside the eye (Hamel, 2006). The most notable syndromic diseases are Usher syndrome and Bardet-Biedl syndrome, but there are approximately thirty syndromes coincident with RP (Phelan and Bok, 2000).

Non-syndromic retinitis pigmentosa is a class of inherited retinal dystrophies that present without other systematic abnormalities (Pierrottet et al., 2014). Inherited non-syndromic RPs are composed of autosomal dominant (15–20%), autosomal recessive (5–20%), X-linked (5–15%), and a vast majority of unknown genetic patterns (40–50%) (Fahim et al., 1993). Eighty-four gene mutations have been identified so far, affecting a wide range of biological functions, which ultimately result in the death of rod photoreceptors. Principal genes affect the visual transduction cascade, visual cycle, and ciliary function and transport within rod photoreceptors (Verbakel et al., 2018).

Figure 2 displays some genes in the visual transduction cascade and photopigment cycle that when disrupted result in RP. Almost every step in the phototransduction cascade and cycle when disrupted could potentially result in rod death and the development of RP. There are four dominant pathways to apoptosis in rods; mutations that create endoplasmic reticulum (ER) stress, excitotoxic levels of cGMP, excitotoxic levels of intracellular calcium, and continuous phototransduction (Koenekoop, 2011; Newton and Megaw, 2020). RHO mutations are the most common mutations in RP and many affect protein folding. Excess accumulation of misfolded proteins in the ER can result in stress and ultimately the activation of apoptotic pathways via caspases (Comitato et al., 2020). Gene mutations such as those to the phosphodiesterase subunits results in excitotoxic levels of cGMP as they are no longer hydrolyzed to 5’ GMP and increases intracellular calcium (Barabas et al., 2010). Finally, mutations in genes such as RPE65 and GNAT1 result in the continuous activation of the phototransduction cascade (Hao et al., 2002), recruiting similar apoptotic pathways to those observed in light damage.

Ciliopathies, which affect ciliogenesis and ciliary protein trafficking, can have detrimental effects on many bodily organs including photoreceptors which are specialized sensory cilia. Rod-cone atrophy in both Usher Syndrome and Bardet-Biedl Syndrome are thought to be caused by deficits in the trafficking between the rod inner and outer segments due to mutated cilia (Waters and Beales, 2011; Tsang et al., 2018; Forsythe and Beales, 2013). Genes encoding trafficking proteins such as MYO7A (Williams and Lopes, 2011), PCDH15 (Ahmed et al., 2003), and USH2A can have profound adverse effects on photoreceptors when disrupted, such as, impaired intracellular directional transport, impaired outer segment formation (Özgül et al., 2011), mislocalized proteins (Zhang et al., 2015), and misrouting of opsins (Jacobson et al., 2008; Reiter and Leroux, 2017). The pathways to apoptosis of ciliopathies are not fully understood but may also be due to ER stress. In many cases multiple pathways to apoptosis may be recruited or are not completely known.

Cone death is secondary to that of rods and follows rod death. The factors leading to cone degeneration have remained somewhat elusive, but there are four prevailing hypotheses. First, rods typically provide the trophic survival factor, rod-derived cone viability factor (RdCVF), to cones to promote survival (Léveillard et al., 2004). The reduction in RdCVF as rods degenerate could reduce the viability of cones via diminished glucose uptake. The second hypothesis is the ‘Cone Starvation’ hypothesis. Proposed by Punzo et al. (2009), they found the mTOR pathway which is typically involved in mediating anabolic processes of cells is diminished due to low nutrient conditions, and promotes the autophagy of cones. They identified upregulation of GLUT-1 in cones, as a compensatory mechanism for the decrease in glucose. Thus, the first two hypothesis converge on a pathway involving glucose loss and cone starvation. The third hypothesis involves the increase in oxidative stressors called reactive oxygen species (ROS). Cones are especially susceptible to oxidative stress due to their high metabolic load (Shen et al., 2005; Hoang et al., 2002). Within the retina, rods consume 95% of the oxygen delivered by the choroid and as rods degenerate the choroid has no mechanism to autoregulate oxygen levels (Yu et al., 2000). Increasing oxygen results in ROS species, which cause mitochondrial breakdown ultimately leading to apoptosis (Kevany and Palczewski, 2010). Finally, microglia activation coincident with rod death may influence cone apoptosis (Gupta et al., 2003; Zeiss and Johnson, 2004). Microglia recruitment is a normal function of the inflammatory immune response to damage, but in many cases, it has both protective and destructive effects (Block and Hong, 2005; Graeber and Streit, 2010). Under these circumstances microglia release pro-inflammatory cytokines that can bind to cones and activate proapoptotic pathways (Zeng et al., 2005). Evidence suggests that these pathways work in concert to produce cone degeneration.

Many identified genes causing RP affect the visual transduction cascade in rods, where mutations in many steps in phototransduction can result in photoreceptor death (Phelan and Bok, 2000). As rods are the first to degenerate, patients first experience night blindness, before peripheral vision loss, and the transition from partial to complete blindness results from secondary loss of cones following rod degeneration. The clinical presentation of RP typically takes decades to evolve, and for much of the early stages, patients may not notice progressive visual changes. The diagnostic criterion for RP encompasses night blindness, peripheral field deficits, lesions within the fundus, and hypovolted electroretinogram (ERG) (Michael and Fletcher, 2004; Hamel, 2006; Kaplan et al., 1990; Baumgartner, 2000; Li et al., 1995). In the early stages of RP, changes in the ERG may be the first indicator that a patient is developing RP, as the fundus can appear clear of pigmentary deposits and the patient may not report any perceptual changes to their vision.

As the disease progresses to intermediate stages, patients begin to notice worsening night vision, culminating in the inability to drive at night or navigate in dim light. In addition, patients also experience visual changes under photopic light conditions. Scotomas can appear, creating blind spots in the peripheral visual field (Grover et al., 1998). Patients may also become photophobic, especially under diffuse lighting conditions, such as bright cloudy weather (Otsuka et al., 2020). Color vision can be disrupted leading to dyschromatopsia specifically in blue/yellow color perception (Jeffrey et al., 2014). Often patients do not seek help until their central visual acuity is compromised due to extensive cone loss, and they are no longer able to read. In this stage, the scotopic ERG is nonexistent and the photopic ERG is significantly reduced (Nagy et al., 2008). Fundoscopy images show bone-spicule shaped deposits in the periphery and narrowing of retinal vasculature (Li et al., 1995).

In the late stages of RP, patients are left with very little, if any, foveal vision. The fundus shows dramatic bone spicule pigmentation, thin vasculature, and optic disk discoloration. The choroid also begins to atrophy in the periphery and macular regions. With little to no vision remaining, this is the stage in which retinal prosthesis have traditionally been implemented to restore vision, given the low risk to further vision loss. Prior work shows that patients with late stage RP have a harder time using assistive technologies limiting their ability to pursue higher education and maintain jobs (Chaumet-Riffaud et al., 2017). As such, vision restoration could have a large impact on patients’ quality of life as they regain the ability to independently navigate. However, extensive disease mediated plasticity in the inner retina may also limit the potential quality of restored vision. These considerations demonstrate the importance for gauging the best disease stage to implement vision restoration strategies for both RP and AMD from a cost/benefit perspective.

Given the relatively simple genetic origins of most RP, gene therapies have garnered much excitement for the potential to “cure” RP by delivering the required gene to correct rod dysfunction before degeneration begins. Nevertheless, only one gene therapy has been approved by the FDA (Luxturna, Spark Therapeutics), which is approved for use in patients with Leber Congenital Amaurosis. Luxturna targets the RPE65 gene and showed improvements in the majority of patients at follow-up. While Luxturna paved the way for gene therapies it only targets 0.3–1% of RP cases (Wu et al., 2023), and other gene therapies remain in the clinical trial stage. As others have reviewed the current state of gene therapies for inherited retinal diseases, we direct interested readers to Amato et al. (2021), for a more in depth discussion.

Other therapies are under development to slow the progression of RP such as injections of Rod Cone viability factor (Yang et al., 2009), glucose addition (Kaplan et al., 2021), or the anti-inflammatory drug dexamethasone (Guadagni et al., 2019). Interestingly, these trophic drugs may also be useful for AMD as rod loss precedes cone loss and metabolic stress is a feature of both diseases. However, any benefit is only afforded after the loss of rods, and may only slow, not prevent, the loss of cones.

The etiology and pathophysiology of AMD are more complex and less well understood than RP, but several factors significantly increase the incidence of developing AMD. While no one gene or factor will in isolation result in AMD, the largest factor contributing to disease is age. Other environmental factors include smoking (Klein et al., 1993a,b; Smith et al., 1996), diet (Chapman et al., 2019), hypertension (Hogg et al., 2008; Katsi et al., 2015), oxidative stress (Beatty et al., 2000; Shaw et al., 2016; Jarrett and Boulton, 2012), alcohol intake (Chapman et al., 2019), BMI, and sunlight exposure (Age-Related Eye Disease Study Research Group, 2000; Coleman et al., 2008), while traits not directly associated with disease progression include race (Friedman et al., 1999; Klein et al., 1999), iris color (Cruickshanks et al., 1993; Taylor et al., 1992), and gender (Mousavi and Armstrong, 2013). The degree to which genes result in AMD and the severity of progression is not completely understood. However, a risk factor scale has been developed using additive odds ratios for each individual risk factor. Risk categories are low (1.0–7.90), average (8.0–28.9), moderate (29.0–100), high (101–184), severe (185–2,600) (Zanke et al., 2010). With increasing risk score, patients increase the prevalence of progressing to the late-stage AMD outcomes, geographic atrophy and choroidal neovascularization. Primary genes include CFH, ARMS2, C3, CFB, and C2, with risk scores of 7.2, 25.4, 3.6, 0.3, and 0.3, respectively, (Seddon et al., 2009).

AMD can be clinically classified into five stages. Stage 1 manifests as small drusen deposits that are less than 63 μm in diameter and is typically diagnosed at routine optometry visits. In Stage 2, drusen deposits become larger than 63 μm, but smaller than 125 μm. Stage 3 is classified as ‘intermediate AMD’, where drusen deposits become over 125 μm and the retinal pigment epithelium (RPE) begins to show abnormalities. Patients with early to intermediate AMD typically perform well on high contrast, high luminance best-corrected visual acuity tasks (Pondorfer et al., 2019). Interestingly, low luminance tasks show the most pronounced effects of visual impairment in early AMD (Chandramohan et al., 2016; Owsley et al., 2016; Nigalye et al., 2022). This shows the importance of parafoveal rod vision, which can precede inner macular cone loss, resulting in scotopic visual acuity deficits (Nebbioso et al., 2014). Stage 4, or advanced AMD shows RPE damage and geographic atrophy of photoreceptors. The final stage is the transition from non-neovascularized (dry) to neovascularized (wet) AMD (Ferris et al., 2013). Patients with severe AMD can eventually experience bilateral macular vision loss. Complete loss of macular vision affects 15–20 degrees of the visual field (Kolb, n.d.). From the age of 65 to 80 the prevalence of developing intermediate, geographic, and neovascular AMD increase from 5.4 to 23.6%, 0.3 to 6.9%, and 0.4 to 8.2%, respectively (Friedman et al., 2004; Yonekawa and Kim, 2015).

Significant visual impairment occurs with progression to either the geographic atrophy or neovascularized forms of AMD (Figure 1, right middle & bottom). Vision loss following diagnosis of neovascular AMD is rapid (Munk et al., 2012) leading to a rapid decline in quality of life. In one study, patients with vision limited to light perception stated they would be willing to trade 60% of their remaining lifespan to regain perfect vision (Brown et al., 2000). Clearly, the AMD population stands to greatly benefit from vision restoration technologies, yet most vision restoration research has been focused on retinitis pigmentosa (see Table 1).

In the absence of vision restoration therapies, a large effort has been focused on slowing the progression of AMD. Given the apparent role of inflammation and immune response in AMD, multiple classes of anti-inflammatory drugs have been tested on AMD, namely, corticosteroids, nonsteroidal anti-inflammatory drugs, immunosuppressants, and biologics (Wang et al., 2011), however, there has been little success in slowing the progression of dry AMD and to date there is no standard treatment. Currently, anti-VEGF medications such as, ranibizumab and bevacizumab are the only available treatments for patients with neovascular AMD, and these have good efficacy at slowing the progression of CNV and prolonging visual acuity (Pieramici and Rabena, 2008). However, in six clinical trials for anti-VEGF medication 30–98% of patients eyes still developed macular atrophy at follow-up (Foss et al., 2022). Because these drugs do not fully prevent the disease from progressing to vision loss, there exists a strong need for vision restoration technologies in AMD, despite the success of disease slowing therapies.

Structural plasticity occurs throughout the nervous system; especially after injury and when presynaptic inputs are lost (Butz et al., 2009). In both RP and AMD structural plasticity begins before complete degeneration of photoreceptors (Jones et al., 1995; Fariss et al., 2000; Phillips et al., 2010; Jones et al., 2003). As the retinal environment becomes unhealthy, reactive gliosis occurs to prevent further retinal damage and promote repair. However, the primary structural glial cells of the retina, Müller glia, also undergo reactive gliosis resulting in hypertrophy and glial scarring (Graca et al., 2018). Glial scarring can be a significant hurdle to subretinal prosthetic vision restoration as it creates an additional barrier between stimulating electrodes and target neurons (Graca et al., 2018). In addition, as photoreceptors degenerate and second order neurons lose their connections, dendrite retraction occurs which is followed by neurite expansion in search of new presynaptic targets. Generally, RP and AMD both exhibit similar structural changes after degeneration of photoreceptors, despite their difference in causes (Jones et al., 1995). This may indicate that loss of input and not disease origin is a major driver of structural plasticity, however this question is far from settled.

Jones et al. (1995) have provided an extensive categorization of the structural changes in RP between human and animal model retinas. Primary hallmarks of structural remodeling after degeneration include second order neuron dendrite retraction, bipolar cell death, second order neurite expansion, and gliosis. While the severity of remodeling depends on each RP variant, the basics of remodeling remain the same. As rods degenerate, rod bipolar cells retract their dendrites and form new contacts with remaining cones, prolonging signaling (Peng et al., 2000). As cones begin to degenerate, all bipolar cells retract their dendrites and send ectopic processes into the inner plexiform layer (Phillips et al., 2010). Cone death initiates Müller glia hypertrophy and the formation of a glial scar. Müller glial cell bodies also migrate into the inner nuclear layer. Both GABAergic and glycinergic amacrine cells have also been found to ectopically migrate out of the inner nuclear layer. Finally, once cells within the inner nuclear layer lose all inputs from photoreceptors, they begin sending neurite projections both into the inner plexiform layer and outer nuclear layer (Jones et al., 2016). These projections seem to have no specificity in their targets and the functional impact of these aberrant connections on signal encoding has not yet been explored.

In addition, the primary glutamate receptor on rod bipolar and ON-cone bipolar cells (mGluR6) also significantly decreases with degeneration in a mouse model of RP (Puthussery et al., 2009). Receptor changes could be deleterious to vision restoration strategies. However, most vision restoration strategies target cells in the inner nuclear or ganglion cell layer and loss of mGlur6 on ON-cone dendrites will have little effect on encoding of prosthetic stimuli.

Comparative studies have shown that RP and AMD human retinas exhibit similar gross structural remodeling after degeneration. AMD retinas exhibit gliosis of Müller cells, translocation of glycinergic amacrine cells, and ectopic processes of GABAergic amacrine cells (Jones et al., 2016; Sullivan et al., 2003). This remodeling is especially severe directly over drusen deposits, but there is extensive remodeling in areas where no drusen deposits are found, suggesting the degree of remodeling is not restricted to the bounds of macular drusen formation. The progression of remodeling is also noticeably faster than in RP and occurs even when rod photoreceptors are still present (Jones et al., 2016). This may indicate separate mechanisms by which structural plasticity is initiated in RP and AMD retinas and indicates that factors beyond loss of inputs contribute to changes. We return to this point of quicker remodeling and its possible impact on functional responses in the sections below.

Changes in circuit structure will certainly cause changes in function, but it is also possible to have functional changes in a network beyond what can be seen structurally. Here we define structural plasticity as discrete changes to cell location, dendrite morphology, or receptor number and functional plasticity as changes in the electrical, or biochemical activity within and between cells. There is still a large gap in our understanding of how the retina functionally changes after degeneration (Puthussery and Taylor, 2010). For example, although animal models of RP have been used extensively to characterize physiological responses to prosthetic electrical stimulation or optogenetic stimulation, we still do not fully understand the complexities of how these responses are generated by spared retinal circuitry. In particular, inhibition is a key circuit element shaping retinal response during natural vision, however, few studies have examined if or how inhibitory function is altered following degeneration (Ivanova et al., 2016; Margolis and Detwiler, 2011; Ye and Goo, 2007; Haq et al., 2014), although recent work suggests that it is disrupted (Carleton and Oesch, 2022; Carleton and Oesch, 2024). In this section, we cover the known functional changes of the retina in response to prosthetic stimuli in mouse models of RP and AMD.

Animal models are a powerful tool used to assess circuit function. There are over 10 different mouse models of retinitis pigmentosa; the rd1 and rd10 mouse lines being the two most commonly used. Both the rd1 and rd10 models of retinitis pigmentosa are valuable models of autosomal recessive retinitis pigmentosa in humans (Chang et al., 2002). Both of these models have mutations in the gene PDEb6, which disrupt the function of the beta subunit of phosphodiesterase resulting in excitotoxic cGMP (see Figure 2), similar to common human forms of RP.

Although there are many animal models trying to recapitulate the AMD disease state, due to the diffuse and incompletely understood causes, no one model fully captures all the factors leading to atrophy of photoreceptors (Pennesi et al., 2012). The models that can recapitulate both an ‘AMD like state’ and geographic atrophy are not consistent across all animals, requiring large cohorts and continued genetic modification. Due to these challenges, there has been limited work on the functional changes in the retina in AMD. One recent model that has been used to examine retinal function is laser induced geographic atrophy. While this model does mimic some features of AMD, such as central photoreceptor death and recruitment of the complement pathway, it does not include the chronic inflammatory or pro-angiogenic phenotypes (Khan et al., 2023), which may drive both functional and structural plasticity beyond what is caused by the loss of photoreceptors. This gap in knowledge about retinal function in the AMD disease state presents a significant challenge for the development and implementation of vision restoration strategies in AMD.