henhao
Sign in
GeneticsbioRxiv
Preprint

Host Endoplasmic Reticulum Stress and Interferon Responses Contribute to AAV-Induced Ocular Toxicity

Apolonia GardneriD, Christin M. HongiD, Sophia R. ZhaoiD, Adam J. Daniels, Constance L. CepkoiD

Abstract

Adeno-associated viruses (AAVs) are popular gene therapy vectors, but AAVs can cause toxicity. This is particularly evident following expression of some transgenes, e.g. GFP, in the retinal pigment epithelium (RPE), which leads to loss of RPE cells and photoreceptors. Here, we sought to unravel the toxicity mechanism(s). Several transgenes, self and non-self, were tested for toxicity. Some transgenes were toxic while others were not, with no clear correlation for toxicity with regards to self vs non-self transgenes. RPE RNA-sequencing revealed upregulation of translational processes, cell stress, cytokine release, antiviral responses, and leukocyte infiltration pathways. Toxicity-inducing pathways were explored for causality by injecting toxic AAVs into mice deficient for intrinsic, innate, or adaptive immune pathways. The CHOP KO partially alleviated toxicity for RPE but not photoreceptors, whereas the type I interferon receptor KO partially alleviated toxicity for photoreceptors but not RPE. In situ hybridization of interferon pathway transcripts (IFNB1, IFNAR1) revealed that the RPE and retina can produce and potentially respond to interferon. These data suggest that transgene-induced cell stress responses in the RPE lead to RPE cell death, while interferon signaling contributes to the death of photoreceptors.

PDFDOI
Full-text digest ↓

02 /Specification & Classification

Click any tag to browse matching papers
Citations
0
OpenAlex Topics

In-depth summary

✦ Multi-agent verified full-text synthesis

Claude Sonnet 4.6 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (23 Sept 2026) · CC BY 4.0

In neonatal mice, AAV-driven GFP overexpression in RPE shows ER stress signatures and partly CHOP-dependent RPE damage, while IFNAR loss partly spares photoreceptors.

AAV gene therapy can cause toxicity even in the relatively immune-privileged eye, and the mechanisms are unclear. This work separates RPE and photoreceptor toxicity pathways by comparing a toxic and a benign GFP vector using RNA-seq and a panel of knockout mice.

Key results

  1. ShownIn an otherwise identical vector backbone, RPE-restricted GFP (Best1::GFP) caused RPE damage and progressive photoreceptor loss, whereas rod-restricted GFP (Rho::GFP) was largely benign.

    At 24 weeks, ~10.6 ONL layers (Rho::GFP) vs ~3 (Best1::GFP); week 24 p<0.0001 (n=3-6 eyes). RPE score higher with Best1::GFP at 2 weeks (2e9 gc; p=0.03, n=3-4 eyes). Dose of 2e7 gc was non-toxic. · Assessment of toxicity from AAVs expressing GFP…; Figure legends

  2. ShownRPE RNA-seq showed early translation and ER stress signatures (including CHOP and ATF4) at week 1 and broader cytokine, interferon and myeloid signatures at week 2. Many interferon-stimulated genes (ISGs) were induced.

    n=5 eyes per group. Genes enriched with Best1::GFP: 1688 (week 1) and 1461 (week 2). Interferome ISGs: 245 (week 1) and 281 (week 2); 47 and 59 of these induced by a factor of 2 or more. Visual cycle genes (RPE65, RGR, RDH10) were downregulated. · Analysis of differential gene expression in…; Assessment of type I interferon responses in RPE…; Figure legends

  3. ShownKnockout screening dissociated RPE and photoreceptor toxicity. IFNAR1 KO preserved rods and cones and cGAS KO preserved rods, neither helped RPE; CHOP KO improved RPE but not photoreceptors.

    Mean ONL layers ~9.3 (IFNAR1 KO) and ~8.5 (cGAS KO) vs ~6.8 (B6J). IFNAR1 KO alone had more cones. Mean RPE score 1.1 in CHOP KO (3 eyes) vs 2.2 in B6J, and 1.9 in IFNAR1 KO. · Assessment of the contributions of intrinsic…; Figure legends

  4. SuggestedAdaptive immunity, complement, TLR, MAVS, type II interferon, pyroptosis and chemokine receptor deficiencies, and microglia depletion, gave no clear ONL rescue; cones rose slightly only with PLX5622 at 4e9 gc. TNFR1 KO rescue reflected strain background.

    PLX5622 reduced Iba1+ cells by ~91% (4e8 gc) and ~98% (4e9 gc) with no change in ONL layers. RAG1, IL2RG, GSDMD and Cx3cr1/Ccr2 KOs showed no rescue. TNFR1 KO on Sv129 rescued; on C57BL/6 it did not. · Assessment of the contributions of intrinsic…; Assessment of the contributions of microglia…

  5. ShownToxicity depended on the transgene but did not track self versus foreign status. Of five transgenes, only GFP and B2M were toxic; RPE65, Ctsd, CasMini and a non-expressing GFP control were not.

    At 4-5 weeks: ~6.8 ONL layers with Best1::GFP (~41% reduction) and ~8.4 with Best1::B2M (~27% reduction), vs ~11.5 in uninjected controls (n=4-17 eyes per group). · Assessment of Transgene Dependence on…; Figure legends

Methods, data and reproducibility

Toxic Best1::GFP and benign Rho::GFP AAV2/8 vectors were compared after neonatal subretinal injection in mice, using bulk RPE RNA-seq (DESeq2, GO enrichment), HCR-FISH, and a knockout-strain screen scored by RPE morphology, cone counts and ONL layer counts.

  • Bulk RPE RNA-seq of Best1::GFP vs Rho::GFP (4e8 gc, contralateral eyes) at 1 and 2 weeks, plus uninjected controls · Mus musculus (C57BL/6J) · n = n=5 eyes per group; 20 samples analyzed together
  • KO strain screen (ten KO strains, 4e8 gc Best1::GFP, neonatal); ONL layers, cone counts, RPE scores at 4-5 weeks · mouse · n = n=4-17 eyes per group (ONL layers)
  • Best1::GFP dose response (2e7, 2e8, 4e8, 2e9 gc) and Rho::GFP vs Best1::GFP time course to 24 weeks · mouse (C57BL/6J) · n = n=3-6 eyes per group (time course)
  • Transgene panel (GFP, B2M, Ctsd, RPE65, CasMini, 6xSTOP-mutGFP control) under Best1 promoter · mouse (C57BL/6J) · n = n=4-17 eyes per group
  • PLX5622 microglia depletion experiments at 4e8 and 4e9 gc Best1::GFP; Cx3cr1-GFP reporter depletion checks · mouse · n = n≥4 eyes per group (RPE and cones)
  • HCR-FISH for IFNAR1 and IFNB1 on RPE and retinal flatmounts at P14 · mouse (C57BL/6J)
  • Adult subretinal injections (2e10 gc Best1::GFP; 2e9 gc CMV::GFP) · mouse (C57BL/6J) · n = n=4-16 eyes per group (RPE scores)
Methods & tools
DESeq2, STAR 2-pass, featureCounts, clusterProfiler, ashr, Interferome database, HCR RNA-FISH, PLX5622 (CSF1R inhibitor), ImageJ/Fiji, GraphPad Prism, Ensembl GRCm39 release 112, Illumina NovaSeq 6000
Reagents & environment
Reads aligned to Ensembl Mus musculus GRCm39 release 112 with STAR 2-pass; DESeq2 median-of-ratios normalization; ashr log2 fold-change shrinkage; statistics in GraphPad Prism.
Reproducibility gaps
Raw data available only upon request; no public accession or code repository reported for the RNA-seq data or image quantification.
Limitations (6)
  • Chow-based microglia depletion was ineffective in pups before weaning (only slight, non-significant depletion at P19), so depletion covered only about the last week; a microglial role cannot be excluded. · stated by the authors
  • The Sv129-background TNFR1 KO gave a spurious rescue that vanished on a C57BL/6 background, showing strain background can confound KO comparisons. · stated by the authors
  • Loss-of-function tests of adaptive immunity were not performed in adult injections; neonatal mice may have an immature adaptive system, and IFNAR KO did not help adults injected with CMV::GFP. · stated by the authors
  • A survey of only five transgenes gave no strong predictors of toxicity, so findings from GFP may not generalize to clinically relevant transgenes. · stated by the authors
  • ER stress is inferred from GO terms, CHOP and ATF4 transcript levels and CHOP KO rescue (3 eyes in the RPE score panel); no direct protein-level ER stress measurement is reported. · noted in this summary
  • IFNB1 and IFNAR1 HCR-FISH is described qualitatively without quantification, and KO rescues were partial, so the causal weight of interferon signaling remains uncertain. · noted in this summary

Who should read:Read this if you develop ocular AAV gene therapies and want knockout-mouse evidence on how ER stress and type I interferon signaling contribute to RPE and photoreceptor toxicity.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISSourd '26Ail '21VENDOMELE '22Kinane '26Pereira '26Friese '25
67% semantic matchImmunology
Célia Sourd, Joel Quinn, Molly C. John+10 more·Molecular Therapy Advances·Mar 31, 2026