Table 1.

Summary of experimental models relevant for pre-clinical studies of Stargardt disease therapies. A2E = N-retinylidene-N-retinylethanolamine; ABCA4 = ATP-binding cassette transporter protein 4; ELOVL4 = elongation of very-long-chain fatty acids 4; ERG = electroretinogram; KO = knockout; ONL = outer nuclear layer; PROM1 = prominin 1; Rd19 = retinal degeneration model 19; RPE = retinal pigment epithelium.

Model Type	Details	Structural Features	Functional Features	Strengths/Limitations
Mouse
STGD1	Abca4 KO [25,26]	Absence of Abca4 expression. On an albino background, loss of outer nuclear layer (ONL) structure at 11 months. Pigmented mice show no loss in structure. Lipofuscin granule accumulation in the RPE.	Abca4 KO models exhibit increased autofluorescence compared to age-matched wild-type mice that correlates to accumulation of bisretinoids/A2E/lipofuscin.	Easy detection of ABCA4 protein following gene supplementation. Assessment of pharmaceutical, dietary and gene therapy efficacy achievable by reduction in autofluorescence and associated build-up of bisretinoids/A2E/lipofuscin. However, the KO genotype and absence of Abca4 does not reflect typical human disease.
STGD1	Leu451Pro and Ala1038Val (PV/PV) [29] Asn965Ser [30]	Reduced expression of Abca4 with mislocalisation within the photoreceptor cells.	Models exhibit increased autofluorescence compared to age-matched wild-type mice that correlates to accumulation of bisretinoids/A2E/lipofuscin.	Efficacy evident in these models would be more relevant to human disease and achieved by rescue of bisretinoid/A2E/lipofuscin build-up and the associated autofluorescence phenotype.
STGD3	Elovl4 KO [31,32]	Normal retinal structure.	Normal retinal function.	The KO is of limited value as it can only be reared as a heterozygous model and offers no clear features of retinal disease.
STGD3	ELOVL4 5-bp deletion knock-in [33,34,35,36]	Accumulation of ELOVL4 at 4 months with progressive loss of ONL and, in particular, cones at 6–18 months.	Abnormal ERG and accumulation of lipofuscin.	Rescue of retinal structure and function. Transgenic models are more representative of human disease both in genotype and phenotype.
STGD4	Rd19	Progressive loss of ONL beginning at 2 months of age.	Normal cone ERG but abnormal rod a-wave responses.	This naturally occurring model has yet to be used in pre-clinical studies.
STGD4	Prom1 KO [37]	Extensive loss of ONL beginning at 2 weeks of age.	Abnormal ERG.	Loss of retinal structure and function begins early; therefore, treatment intervention may not be provided in time to observe efficacy. Rearing in the dark could be applied to slow the rate of degeneration.
STGD4	PROM1 Arg373Cys knock-in [20]	Mislocalisation of PROM1with abnormal outer segment morphology and degeneration.	Abnormal rod and cone ERG by 3 months of age.	The knock-in better reflects the human state and offers an opportunity to assess treatment efficacy through correction of structural and functional changes.
In Vitro
Immortalised cell lines	Wild-type	Lack of native retinal gene expression and absence of specialised retinal structures.	Enables expression and localisation assessments plus downstream isolation and functional assays.	Exogenous delivery of retinal genes of interest is required but basic assessments of vectors and downstream functional assays are achievable.
Induced pluripotent stem cells (iPSCs)	Patient-specific genotype	Cells can be differentiated to better reflect photoreceptor cell morphology and gene expression profiles.	Functional outputs could be achieved by expression profile analysis and downstream protein isolation and functional assays.	These will be particularly useful for future gene-editing techniques in assessing mutation-specific therapies. Editing efficiencies and protein outputs could be compared to cells from control donors.
Fibroblasts	Patient-specific genotype	Some retinal gene expression may be evident, as for ABCA4 [41].	Functional outputs could be achieved by expression profile analysis and downstream protein isolation and functional assays.	The use of these will likely be supplementary to preliminary pre-clinical assessments of new therapies as expression of retinal genes will be limited. However, being patient-derived, they will have the added benefit of being useful for gene-editing strategies.
Hair follicles	Patient-specific genotype	Some retinal gene expression may be evident, as for ABCA4 [41].	As for fibroblast samples, functional outputs could be achieved by expression profile analysis and downstream protein isolation and functional assays.	As for fibroblast samples, the use of these will likely be supplementary to other preliminary pre-clinical assessments but being patient-derived they will have the added benefit of being useful for gene-editing therapies.
Retinal organoids	Patient-specific genotype	Structural differences may be evident and include protein mislocalisation [42,43,44,45,46,47,48,49].	As for other patient-derived samples, functional outputs could be achieved by expression profile analysis and downstream protein isolation and functional assays.	Changes in expression profiles and protein localisation plus cell morphology could be assessed following treatment application. Retinal organoid will provide an ideal model for mutation-specific treatments.