Ophthalmology has emerged as a premier proving ground for gene therapy following the landmark regulatory approval of voretigene neparvovec, yet existing vectors remain limited by static expression profiles that cannot accommodate fluctuating pathologies. Approved by the United States Food and Drug Administration in 2017, voretigene neparvovec utilizes an adeno-associated virus vector to supply a functional RPE65 gene to the retinal pigment epithelium, establishing a pivotal regulatory precedent for inherited retinal dystrophies. This milestone catalyzed a global clinical pipeline that now includes more than sixty active clinical-stage programs. These initiatives address conditions ranging from monogenic photoreceptor degenerations involving PDE6B, RPGR, or CNGA3 to gene-agnostic neuroregenerative strategies delivering neurotrophic factors like ciliary neurotrophic factor and brain-derived neurotrophic factor. Increasingly, gene therapy is expanding toward complex multifactorial disorders such as age-related macular degeneration, diabetic macular edema, and glaucoma, where pathophysiology is driven by dynamic, evolving mechanisms rather than isolated single-gene defects.
While constitutive viral gene delivery has proven effective for monogenic conditions with fixed replacement requirements, this monolithic framework cannot address the changing demands of multifactorial ocular diseases. Virtually every construct currently evaluated in clinical pipelines relies on constitutive promoters that produce non-adjustable protein expression across the entire lifespan of the transduced retinal cell. Once delivered, these vectors lack any mechanism to adjust dosage, sequence distinct biological interventions across successive disease stages, or terminate transgene production if adverse effects occur. Multifactorial retinal diseases exhibit fluctuating stages of metabolic deficiency, cellular inflammation, and structural damage over decades, rendering permanent, fixed-dose expression potentially inadequate or toxic. Uncontrolled lifelong production of potent neurotrophic or anti-angiogenic molecules may induce collateral tissue damage or cellular exhaustion. Consequently, continuing to rely on unmodifiable expression systems designed in the 1990s restricts clinical efficacy, creating a pressing need for adaptable platforms that adjust to evolving chronic pathologies.
To overcome these pharmacological limitations, researchers are advancing molecular toolkits that provide switchable or tunable transgene expression within ocular tissues, establishing distinct mechanisms for binary and graded control. In this terminology, tunable systems allow therapeutic protein levels to be continuously modified in a graded, dose-responsive manner, whereas switchable architectures undergo discrete on-or-off state transitions triggered by external signals. Regulatory control can occur at transcriptional, post-transcriptional, or post-translational stages. Historically studied transcriptional switches, including tetracycline-responsive promoters, rapamycin-dimerization systems, hypoxia-inducible cassettes, and microRNA-response elements, face notable translational constraints in the eye. These inducible genetic systems require complex regulatory machinery to operate flawlessly within non-dividing retinal cells over multi-decade treatment horizons. Additionally, regulating ocular transcription via systemic drug administration frequently results in unpredictable intraocular pharmacokinetics, making precise titrations exceptionally difficult while increasing the probability of systemic drug exposure and unwanted off-target complications across non-ocular tissues.
Post-translational regulation presents a conceptually cleaner strategy by preserving continuous viral transcription while governing therapeutic protein abundance through pharmacological destabilizing domains. Under this design, an adeno-associated virus cassette maintains steady transcription from a strong viral promoter, maximizing cellular transduction efficiency and simplifying genetic architecture. Functional output is regulated downstream at the protein level by fusing the therapeutic transgene to a destabilizing domain derived from Escherichia coli dihydrofolate reductase, encoded by folA, or mutant human FK506-binding protein 12, encoded by FKBP1A. In the absence of a specific small-molecule stabilizer, the destabilizing fusion protein is recognized as misfolded and continuously degraded by the cellular ubiquitin-proteasome system. Delivering an exogenous stabilizer, such as trimethoprim for dihydrofolate reductase or Shield-1 for FK506-binding protein, binds the domain and rescues the fusion protein from proteasomal destruction. This pharmacological stabilization produces a rapid, titratable, dose-dependent, and reversible increase in therapeutic protein concentration within target ocular tissues.
Extensive preclinical investigations have validated the efficacy of the dihydrofolate reductase and trimethoprim regulatory axis across multiple retinal disease models. In vivo studies confirm that proteasomal activity remains fully functional and robustly capable of maintaining basal fusion protein degradation in aged mice, during light-induced retinal degeneration, and in genetic photoreceptor degeneration models including rd2 and Abca4-knockout strains. Santiago and colleagues provided direct proof of concept for tunable ocular therapy by fusing a dihydrofolate reductase domain to RPF2, an engineered neuroprotective cytokine derivative of leukemia inhibitory factor. By altering trimethoprim dosage, the researchers titrated RPF2 abundance in vivo to protect photoreceptors against degenerative cellular death. Furthermore, non-antibiotic small-molecule stabilizers have been developed to eliminate potential complications associated with long-term antibiotic administration. Complementing these post-translational tools, small-molecule-responsive riboswitches operating at the RNA level have demonstrated a fifteen-fold induction of adeno-associated virus transgene expression in mice with verified reversibility upon repeat dosing schedules.
A compelling case study demonstrating the necessity of drug-tunable destabilizing domains involves outer retinal neovascularization, where continuous vascular endothelial growth factor suppression risks serious collateral damage. While constant anti-angiogenic suppression is therapeutically appealing, eliminating physiological vascular endothelial growth factor endangers the healthy choroidal vasculature and impairs retinal neurons dependent on sustained trophic signaling. To address this risk, Liu and colleagues engineered a destabilizing domain fusion linking Escherichia coli dihydrofolate reductase to Flt23k, an intracellular vascular endothelial growth factor trap. In the absence of trimethoprim, the intracellular trap is rapidly degraded, but trimethoprim administration rescues protein expression in a dose-dependent manner to inhibit both intracellular and extracellular growth factor targets. Tested in a rat model of oxygen-induced retinopathy, this platform significantly reduced retinal vascular endothelial growth factor levels and pathological neovascular area exclusively when trimethoprim was provided, verifying on-demand anti-angiogenic control that can be engaged during active disease flares and withdrawn during quiescence.
The limitations of permanent transgene expression become particularly stark when examining glaucoma, the leading global cause of irreversible blindness. Glaucoma is a chronic, heterogeneous neurodegenerative condition characterized by the progressive degeneration of retinal ganglion cells and their axonal projections over thirty to forty years. Disease pathogenesis involves overlapping metabolic, inflammatory, and mechanical insults whose relative contributions shift as the disease advances and vary significantly between individual patients. A patient diagnosed around age fifty requires therapeutic protection for several decades, during which clinical needs evolve from early metabolic rescue to late-stage anti-inflammatory and regenerative interventions. Throughout this prolonged timeframe, required dosages fluctuate alongside changes in intraocular pressure control and comorbid pathologies. Delivering a single therapeutic payload via a permanently active viral vector cannot accommodate this biological complexity, highlighting why modular architectures capable of sequential staging represent an essential evolutionary step forward in ocular gene therapy design and translational clinical application.
Biological insights derived from studying retinal nicotinamide adenine dinucleotide metabolism provide a rigorous foundation for developing tunable gene therapies in glaucoma. Experimental findings reveal that age-dependent reductions in retinal nicotinamide adenine dinucleotide render retinal ganglion cells highly vulnerable to intraocular pressure-dependent stress. Boosting cellular dinucleotide levels through dietary supplementation with nicotinamide, or vitamin B3, or via adeno-associated virus overexpression of NMNAT1 or NMNAT2 robustly prevents glaucomatous neurodegeneration across animal models. Furthermore, combining oral nicotinamide with the Wallerian degeneration slow allele, which increases tissue dinucleotide concentrations through an adenylyltransferase-dependent mechanism, protected ninety-four percent of treated eyes from detectable glaucoma. This dual intervention achieved substantially greater preservation than either treatment alone, demonstrating additive, multi-compartment neuroprotection spanning ganglion cell somas, synaptic connections, and axons. These experimental results emphasize that achieving robust clinical neuroprotection in glaucoma requires engaging multiple distinct protective pathways rather than depending exclusively on single-target interventions.
Human tissue mapping confirms that the nicotinamide adenine dinucleotide salvage pathway is enriched in retinal ganglion cells, where expression of the cytoplasmic synthesizing enzyme NMNAT2 declines during glaucoma pathogenesis. Downregulation of NMNAT2 precedes overt neurodegeneration, identifying it as a critical early biomarker and therapeutic target. Two independent research groups demonstrated that adeno-associated virus overexpression of full-length or truncated NMNAT2 isoforms confers robust neuroprotection in glaucoma models. However, while NMNAT2 provides neuron-specific survival benefits, it does not confer anti-inflammatory or pro-regenerative activities required to reverse advanced tissue damage. Nicotinamide has already completed three successful Phase II clinical trials and is undergoing active Phase III investigation, making dinucleotide replenishment a premier candidate for establishing baseline ganglion cell resilience. Nevertheless, comprehensive long-term protection demands combining metabolic support with additional downstream pathways, reinforcing the rationale for modular vector systems capable of staging axon-protective, anti-inflammatory, and regenerative payloads across changing disease stages.
Based on these metabolic insights, researchers propose a modular, drug-tunable architecture delivering two therapeutic modules from a single adeno-associated virus cassette. In this conceptual design, Module 1 encodes NMNAT2 fused to a dihydrofolate reductase destabilizing domain for neuroprotection, while Module 2 incorporates an anti-inflammatory or pro-regenerative factor fused to an FK506-binding protein destabilizing domain. Module 2 remains deliberately undefined because no clinically validated regenerative factor for glaucoma currently exists, making payload identification an external prerequisite rather than an inherent feature of the architecture. Each module would be independently regulated by distinct oral or topical small-molecule ligands, allowing clinicians to activate neuroprotection upon diagnosis and subsequently initiate regeneration decades later following a single intravitreal injection. However, engineering true ligand independence presents substantial challenges, requiring separate intraocular pharmacokinetics, non-overlapping target specificity, and manageable systemic exposure over decades. Furthermore, packaging constraints near the vector limit of approximately 4.7 kilobases necessitate compact isoforms, self-cleaving peptides, or dual-vector strategies, while CMC hurdles demand novel potency assays and stopping rules to mitigate basal leakiness, incomplete shutoff, and chronic ligand toxicity.
Subject of Research: Drug-tunable modular gene therapy architectures for chronic ocular neurodegenerative diseases
Article Title: A new translational architecture for protective and regenerative ocular gene therapy
Article References: Liu, G.-S., Jovancevic, D., & Williams, P. A. (2026). A new translational architecture for protective and regenerative ocular gene therapy. Molecular Neurodegeneration Advances, 2(1), Article 52. https://doi.org/10.1186/s44477-026-00061-w
Image Credits: AI Generated
DOI: 10.1186/s44477-026-00061-w
Keywords: ocular gene therapy, adeno-associated virus, destabilizing domains, NMNAT2, glaucoma neuroprotection, trimethoprim, retinal ganglion cells, nicotinamide adenine dinucleotide, FKBP12, ecDHFR, VEGF-trap, Flt23k, voretigene neparvovec, retinal pigment epithelium, ubiquitin-proteasome system, WldS allele, riboswitches, AAV packaging capacity
News Source: Juliet Wilcox. (November 16, 2026). Modular Gene Therapy Architecture Enables Tunable Treatment for Complex Retinal Diseases. Scienmag.



