When the Retina’s Recycling System Falters: Kyoto iPS Cells Reveal a Rare Disease

Across the bottom of a culture dish lies a dark, cobblestone sheet of cells. They are retinal pigment epithelial cells, or RPE: the single layer that supports the light-sensing rods and cones at the back of the eye. RPE cells deliver nutrients, absorb stray light, recycle visual pigments and engulf the worn tips that photoreceptors shed every day. They are quiet custodians of sight, performing an immense cleanup job with little opportunity to replace themselves.

Yumi Inoue, Hanako Ohashi Ikeda and colleagues at Kyoto University turned induced pluripotent stem cells from a patient with Malattia Leventinese into this specialized tissue. The rare inherited macular disease is also known as Doyne honeycomb retinal dystrophy. When the scientists fed the patient-derived RPE a sustained diet of photoreceptor outer segments, whitish structures began to appear beneath the cells after about three weeks. The deposits contained mutant fibulin-3, apolipoprotein E, complement C3, collagen IV and lipid granules—many of the materials implicated in deposits in the human eye.

The study, published July 22 in JCI Insight, places a diminished and malfunctioning lysosomal system in the middle of the disease chain. Lysosomes are acidic cellular compartments that digest and recycle proteins, lipids and damaged components. In the patient-derived RPE, the genetic program supporting this clearance system was weakened, degradation slowed, and incompletely handled material accumulated inside and beneath the cells. When the researchers added the natural disaccharide trehalose, several measures of lysosomal abundance, clearance, deposit formation and cell survival improved.

This was not a treatment in a person, or even in an animal eye. It was an in-vitro experiment using cells from one patient. The trehalose concentration—100 millimolar delivered directly to the culture for four weeks—is not a dietary dose and cannot be translated into advice to consume the sugar. Holding that boundary is essential to understanding both what Kyoto’s team discovered and how much work lies between a rescued cellular phenotype and preserved human vision.

One disease, two names, a century apart

The history begins with one disorder being seen in two places and given two names. In 1899, the English ophthalmologist Robert Doyne described a “peculiar condition of choroiditis” occurring in several members of the same family. Yellow-white spots crowded together with age into a pattern reminiscent of honeycomb, giving rise to the name Doyne honeycomb retinal dystrophy.

In 1925, Swiss ophthalmologist Alfred Vogt described a similar inherited pattern among families in the Leventina Valley of Canton Ticino. The condition became Malattia Leventinese—Italian for “disease of the Leventina.” Small drusen radiating outward from the macula were especially conspicuous in the Swiss families, leaving room for debate over whether the English and Swiss conditions were truly identical.

Genetics settled the question. In 1999, Edwin Stone and colleagues found the same single-nucleotide change in EFEMP1 in every family they studied. The substitution changes the protein’s 345th amino acid from arginine to tryptophan—R345W. Two clinical stories, separated by geography and terminology, converged on one molecular lesion.

Malattia Leventinese is autosomal dominant. In the usual situation in which an affected parent carries one altered copy, each child has a one-in-two chance of inheriting it. Expression is variable even within a family. Drusen may appear in adolescence, yet reduced central detail, distorted straight lines, glare, slow adaptation to darkness and paracentral blind spots may not disrupt life until the fourth or fifth decade. Geographic atrophy can gradually erase central vision. Abnormal choroidal blood vessels can produce a more sudden decline.

The eye’s custodians work a lifetime shift

The retina is often likened to a camera’s film or sensor, but that metaphor leaves out its maintenance system. Rods and cones contain stacks of membrane discs packed with the molecules that convert light into electrical signals. Intense illumination, high oxygen consumption and reactive chemistry damage those membranes. Photoreceptors continually renew them, shedding aged tips toward the RPE.

Each RPE cell supports scores of photoreceptors, and classic estimates suggest roughly a tenth of a photoreceptor’s outer segment is renewed each day. RPE cells engulf the discarded material, deliver it through endosomal compartments to lysosomes, recover useful molecules and dispose of what remains. Unlike a rapidly renewing epithelium, mature RPE is largely postmitotic. The same cells perform this task for decades. A small daily deficit can become a large deposit with time.

Drusen are yellow-white accumulations between the RPE and Bruch’s membrane. They contain lipids, ApoE, complement proteins and extracellular-matrix components. They are also a signature of early age-related macular degeneration. Malattia Leventinese resembles an accelerated, genetically focused version of part of that biology, which is why it has long attracted researchers studying AMD. The diseases are not interchangeable, however: the EFEMP1 R345W variant that causes MAL is not a common cause of AMD.

A protein that cannot leave cleanly

EFEMP1 encodes fibulin-3, a secreted glycoprotein associated with the extracellular matrix. Rich in disulfide bonds, it must fold correctly inside the endoplasmic reticulum before it can be exported. The R345W substitution disrupts that process. Mutant fibulin-3 is retained inside the cell, activating the unfolded-protein response, and it accumulates abnormally on the basement-membrane side of the RPE.

Human tissue studies in 2002 documented aberrant EFEMP1 accumulation. In 2007, knock-in mice carrying R345W developed sub-RPE deposits, complement activation and progressive RPE and choroidal abnormalities. Engineered ARPE-19 cells later showed that mutant extracellular matrix could drive basal deposits and chronic activation of the alternative complement pathway. A 2021 patient-derived iPS-RPE study implicated impaired cholesterol efflux and disordered lipid homeostasis.

Those findings supplied important pieces: misfolded protein, extracellular matrix, complement and lipid. What remained incomplete was the machinery connecting them to cellular damage and deposit formation. Kyoto’s group turned to the organelle that sits at the end of photoreceptor waste processing and at the center of lipid turnover: the lysosome.

From a hidden enzyme to the cell’s command center

Christian de Duve discovered lysosomes in 1955 while studying acid phosphatase. The enzyme appeared strangely inactive in freshly isolated cell fractions, then became more active after storage or membrane-disrupting treatment. De Duve inferred that digestive enzymes were being sequestered behind a membrane. Electron microscopy supplied the visual evidence. De Duve, Albert Claude and George Palade shared the 1974 Nobel Prize in Physiology or Medicine for discoveries concerning the structural and functional organization of the cell.

A lysosome is more than a rubbish bag. Its acidic interior activates dozens of enzymes that break proteins, lipids, carbohydrates and damaged organelles into reusable material. It is also a signaling hub where cells sense nutrients and decide whether to grow or dismantle and recycle their contents.

In 2009, Marco Sardiello and colleagues identified transcription factor EB, TFEB, as a coordinator of a broad lysosomal gene network. When activated and moved into the nucleus, TFEB turns on genes that build lysosomes and strengthen clearance. That history makes the markers in Kyoto’s study intelligible: TFEB is part of the control program, LAMP2 helps define the organelle and its trafficking, and cathepsin D is one of its major digestive proteases.

Turning a skin cell into a window on the retina

A living sample of a patient’s central retina cannot be removed simply to study it. Induced pluripotent stem cells provide a route around that barrier. In 2006, Shinya Yamanaka and Kazutoshi Takahashi showed that four transcription factors could reprogram mature mouse cells to pluripotency. In 2007, Yamanaka’s Kyoto University team generated human iPS cells from adult skin fibroblasts. Yamanaka and John Gurdon shared the 2012 Nobel Prize for discovering that mature cells can be reprogrammed to a pluripotent state.

In regenerative medicine, the aim is to transplant cells made from iPS cells. In disease modeling, the purpose is different: preserve a patient’s genome, direct the reprogrammed cells into the tissue that disease attacks, and replay part of the pathology in a dish. Candidate drugs can then be tested against the cellular defect. The eye has been a symbolic arena for both approaches; in 2014, Kobe hosted the first transplant in a clinical study using an iPS-derived RPE sheet.

For the new work, skin cells were obtained from one male patient carrying EFEMP1 R345W and from two healthy people. Episomal vectors created integration-free iPS cells. The team tested two independently established lines from the patient, MAL1 and MAL2, and one line from each healthy donor, NOR1 and NOR2. Directed differentiation produced pigmented, polygonal RPE with expected markers of identity, junctions and apical polarity. The patient cells could become recognizable RPE; disease emerged as they performed an RPE cell’s demanding work.

Making decades of disease appear in weeks

After allowing the iPS-RPE to mature, the researchers added bovine photoreceptor outer segments for as long as four weeks. The treatment imposed the recurring phagocytic and metabolic burden faced by RPE in the eye. Initial uptake was not clearly defective. The trouble emerged in what happened after engulfment and under sustained stress.

Mutant fibulin-3 was distributed abnormally. HSPA5, a marker of endoplasmic-reticulum stress, increased. Caspase-3 measurements and TUNEL staining showed greater apoptosis. Outside the cells, matrix metalloproteinase 2 became more active. Because MMP2 degrades collagen IV, its activation offers a possible link from intracellular stress to weakening of the basement membrane and an environment permissive to atrophy or neovascularization.

The whitish material beneath the patient-derived cells was examined with confocal microscopy, transmission electron microscopy and focused-ion-beam scanning electron microscopy. Thick elongated fibers, lipid granules, stacked degenerating cells and a broad extracellular matrix appeared on the basal side. Fibulin-3, ApoE, complement C3, scattered C5b-9 and collagen IV localized to the structures. The model did not recreate a patient’s macula in full, but it created deposits with enough of drusen’s architecture and ingredients to merit the careful term “drusen-like.”

A lipid map points toward lysosomal gridlock

Untargeted lipidomics found significantly more hexosylceramide and bis(monoacylglycero)phosphate, or BMP, in the MAL cells. LBPA, a specific configuration of BMP associated with late endosomes and lysosomes, formed aggregates. The pattern suggested congestion in a pathway responsible for breaking down and redistributing lipids. Free cholesterol, however, was not significantly increased, a difference from an earlier model.

RNA sequencing showed lower expression of lysosomal genes including LAMP2, cathepsin D and prosaposin. Immunostaining revealed fewer LAMP2-positive puncta, and the mature double-chain form of cathepsin D was reduced. TFEB transcript levels were lower, while TP53 transcripts were higher. The team proposes that retained mutant fibulin-3 creates chronic stress that dysregulates lysosomal gene programs. It remains a working model: static TFEB protein measurements did not explain the entire sequence, and the direction and timing of every causal link have not been established.

The investigators also tested function rather than relying on marker abundance. Activated epidermal growth factor receptor is normally internalized and degraded in lysosomes. That process was slower in MAL RPE. After four weeks of photoreceptor-outer-segment exposure, phosphorylated EGFR had not declined significantly even 60 minutes after the stimulus was removed. The recycling machinery was not merely smaller by staining; it was slower at doing its job.

What trehalose restored—and what it did not prove

The team tested trehalose, a sugar found in fungi, yeast and insects that has been reported to modulate autophagy and lysosomal programs. Mature RPE cells received 100 mM trehalose alongside photoreceptor outer segments for four weeks.

The fraction of TFEB localized to the nucleus increased. LAMP2 and mature cathepsin D rose, as did punctate staining for the mannose-6-phosphate receptor involved in late-endosome-to-lysosome trafficking. The treated MAL cells cleared phosphorylated EGFR significantly within 60 minutes. This was evidence of restored degradative activity, not merely a more lysosome-like appearance.

Pathology improved in parallel. Trehalose preserved cellular morphology and reduced TUNEL-positive apoptosis toward control levels. Dense fibulin-3 structures, intracellular ApoE aggregates, complement C3 and fibrotic collagen IV accumulation declined. MMP2 expression fell. Total fibulin-3 and total ApoE did not simply disappear; the more defensible interpretation is that handling and distribution shifted toward a healthier state.

As a comparison, 2-hydroxypropyl-beta-cyclodextrin did not restore lysosomal enzymes or prevent deposits, matrix damage and apoptosis. Because that compound can extract cholesterol and had helped a different iPS-RPE disease model, its failure here strengthened the argument that MAL’s lysosomal dysfunction was not being driven by cholesterol accumulation alone.

Readout	MAL iPS-RPE	After trehalose
Lysosomal system	Lower LAMP2, mature CTSD and related transcripts; delayed pEGFR clearance.	More LAMP2, mature CTSD and M6PR; improved pEGFR degradation.
Lipids and proteins	Abnormal hexosylceramide, BMP/LBPA, ApoE and α-synuclein accumulation.	Less prominent ApoE and α-synuclein aggregation and mislocalization.
Drusen-like deposits	Fibulin-3, C3, C5b-9, collagen IV and lipid granules beneath RPE.	Reduced dense deposits, C3 and collagen IV accumulation.
Cell and matrix injury	ER stress, apoptosis and MMP2 activation.	Reduced cell death and MMP2, with better-preserved morphology.
Evidence level	One-patient, in-vitro disease model.	Rescue in the same model; no animal or patient treatment evidence.

A natural sugar is not automatically a medicine

Trehalose is used in foods and as a stabilizing excipient in some medicines. Familiarity can create a dangerous shortcut in interpreting laboratory science. The 100 mM concentration in this study bathed cells directly; it is not convertible into a useful dietary dose. Mammalian intestines express trehalase, an enzyme that splits ingested trehalose into glucose. Even material that survives digestion must enter the circulation, cross ocular barriers and reach RPE lysosomes at an effective concentration.

The mechanism is not settled either. Prior work suggests trehalose can create low-grade lysosomal stress that moves TFEB into the nucleus, but mTOR, Akt, calcineurin and transporter-dependent pathways have all been discussed. In Kyoto’s experiment, total TFEB and phosphorylated TFEB were not elevated at the final time point even though nuclear enrichment increased. Dynamic TFEB regulation and TFEB-independent effects may both be involved.

A drug program would need a route of delivery, retinal exposure data, dose-response curves, duration, repeat-dose toxicology and evidence that normal RPE is not harmed. The eventual medicine might be formulated trehalose, a trehalase-resistant analogue, a locally delivered agent or an entirely different molecule that activates the same clearance program more selectively. This study is not evidence for self-treatment with trehalose foods or supplements.

The power—and narrowness—of one patient’s cells

The model’s strength is that it follows a human patient’s genotype through RPE differentiation, physiological stress, deposit formation and a pharmacological rescue in one experimental system. Two independently derived iPS lines showed consistent phenotypes. Whole-exome sequencing did not uncover another known pathogenic inherited-retinal-degeneration variant. Transcriptomics, lipidomics, microscopy and a functional degradation assay converged on the same pathway.

But two lines are not two patients. Both came from one man. The healthy controls were unrelated people, so background genetic differences may contribute to some contrasts. A gene-corrected isogenic control—identical except for repair of EFEMP1 R345W—would provide a stricter test of causality, but the study did not include one. Replication in women, additional families and diverse genetic backgrounds remains necessary.

The culture also lacks choroidal blood vessels, immune cells, an aged Bruch’s membrane, ocular fluid dynamics and systemic metabolism. Four weeks of bovine outer-segment feeding is an ingenious accelerated stress test, not a literal copy of changes that unfold over decades in a human macula. “Drusen-like” is scientifically important language.

Next steps include EFEMP1-corrected lines, cells from more patients, long-term co-culture or retinal-choroidal organoid systems, and dosing and safety studies in animals. Time-resolved experiments could establish whether mutant-protein retention, p53 activity, TFEB dysregulation, lysosomal decline and MMP2 activation occur in the proposed order—or whether feedback loops make the biology less linear.

From a rare dystrophy to AMD: a bridge, not an equivalence

Malattia Leventinese matters beyond the number of people who carry it. MAL and age-related macular degeneration share sub-RPE drusen, complement, ApoE, lipid, extracellular-matrix remodeling, RPE atrophy and neovascular complications. If those features can be connected mechanistically in a single-gene disorder, a tractable vulnerability may emerge within the much more complex biology of AMD.

Yet the leap from “trehalose helped MAL cells” to “trehalose treats AMD” is not justified. AMD reflects interacting effects of age, smoking, complement genetics, lipid handling, oxidative stress and other influences. Similar-looking drusen may arise through more than one route. Kyoto’s result supplies a reason to test lysosomal enhancement in MAL and other deposit-associated disorders; it does not establish an AMD therapy.

There is currently no approved treatment that stops the underlying deposits and atrophy in Doyne honeycomb dystrophy. Anti-VEGF injections have been used when choroidal neovascularization develops, but evidence in this ultra-rare condition comes from small reports. Low-vision rehabilitation, visual aids, specialist monitoring and genetic counseling remain central. A sudden new distortion or loss of vision warrants prompt ophthalmic evaluation because a treatable vascular complication may be present.

After 127 years, the dots begin to form a pathway

Doyne saw a honeycomb in 1899. Vogt documented familial drusen in an Alpine valley in 1925. Genetics joined the names through R345W in 1999. Knock-in mice showed in 2007 that the variant could produce deposits. The iPS revolution that began in 2006 made it possible, two decades later, to turn one patient’s skin into the retinal tissue the disease injures.

In that dish, two other histories converged: the organelle de Duve discovered in 1955 and the TFEB network defined in 2009. Mutant fibulin-3 fails to fold and traffic normally. Chronic stress develops. Lysosomal capacity declines. Lipids and proteins move through the system poorly. ApoE, complement and collagen-rich material accumulates beneath the RPE. Matrix remodeling accelerates and cells die. Parts of that chain remain provisional, but observations once scattered across pathology, mouse models and molecular biology now fit into a testable mechanism.

Trehalose loosened several links in the chain in cultured cells. The next act is not a larger claim; it is replication in gene-corrected controls, additional patients, living eyes, realistic delivery systems and rigorous toxicology. One person’s cells have supplied a new map of a rare disease. Whether it becomes a route to treatment will depend on how carefully researchers travel from here.

Sources and references

This article draws on the Kyoto-led paper in JCI Insight, the research announcement, and primary or authoritative sources on Malattia Leventinese, EFEMP1, lysosomes, TFEB and induced pluripotent stem cells. Trehalose findings are preclinical results in patient-derived cultured cells; they do not establish safety or efficacy in patients.

• Inoue et al., Restoration of impaired lysosomal function mitigates drusen-like deposit formation and cell death in Malattia Leventinese, JCI Insight (2026)
• Kyoto research announcement: Lysosomal abnormalities identified in a rare eye disease (2026, Japanese)
• Stone et al., A single EFEMP1 mutation associated with both Malattia Leventinese and Doyne honeycomb retinal dystrophy, Nature Genetics (1999)
• Sheyanth et al., Clinical and historical review of Doyne honeycomb retinal dystrophy / Malattia Leventinese (2021)
• Fu et al., R345W mutation in EFEMP1 is pathogenic and causes AMD-like deposits in mice (2007)
• Marmorstein et al., Formation and progression of sub-RPE deposits in Efemp1 knock-in mice (2007)
• Nobel Prize: Christian de Duve and the discovery of lysosomes
• Sardiello et al., A gene network regulating lysosomal biogenesis and function, Science (2009)
• Nobel Prize: Mature cells can be reprogrammed to become pluripotent
• Takahashi et al., Induction of pluripotent stem cells from adult human fibroblasts, Cell (2007)
• RIKEN: First transplant of an iPS-derived RPE sheet in a clinical study (2014)
• Macular Society: Doyne honeycomb dystrophy symptoms and treatment
• Jeong et al., Trehalose causes low-grade lysosomal stress to activate TFEB (2021)