Comparative transcriptome analysis of human and murine choroidal neovascularization identifies fibroblast growth factor inducible-14 as phylogenetically conserved mediator of neovascular age-related macular degeneration

  1. Julian Wolf
  2. Anja Schlecht
  3. Dennis-Dominik Rosmus
  4. Stefaniya Boneva
  5. Hansjürgen Agostini
  6. Günther Schlunck
  7. Peter Wieghofer
  8. Clemens Lange

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    The authors present here an analysis of the gene expression patterns of lesions as found in patients wet age-related macular degeneration, known as choroidal neovascularization. This gene expression analysis is compared to an experimental mouse model. Using this analysis, the authors have identified a candidate target, FN14, as a potential target for the treatment of choroidal neovascularization. Identifying an alternative treatment modality is important for the treatment of these lesions beyond the current standard of care.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

No abstract available

Article activity feed

  1. eLife

    Evaluation Summary:

    The authors present here an analysis of the gene expression patterns of lesions as found in patients wet age-related macular degeneration, known as choroidal neovascularization. This gene expression analysis is compared to an experimental mouse model. Using this analysis, the authors have identified a candidate target, FN14, as a potential target for the treatment of choroidal neovascularization. Identifying an alternative treatment modality is important for the treatment of these lesions beyond the current standard of care.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)

  2. eLife

    Reviewer #1 (Public Review):

    Wolf et al. utilized transcriptional profiling of choroidal neovascularization (CNV ) membranes from four patients with neovascular age-related macular degeneration or nAMD (re-analyzed from their recent publication; Schlecht et al., 2020, Am J Pathol) and laser-induced murine model of CNV and identified fibroblast growth factor inducible-14 (FN14), as phylogenetically conserved mediator of CNV formation. Furthermore, the authors utilized intravitreal injection of FN14 decoy receptor that blocks interaction between FN14 and its ligand TWEAK to demonstrate reduced CNV size and decreased expression of a specific cytokine (IL6) in the laser induced murine model of CNV. Based on this data, the authors propose FN14 as a novel therapeutic target for nAMD. As highlighted by the authors, anti-VEGF therapy, the current therapeutic intervention for nAMD is not effective in about one third of nAMD patients. Therefore, the identification of novel therapeutics for nAMD has high clinical relevance. Importantly, the data presented in the current study is consistent with prior studies that have shown altered levels of FN14 as well as a role of TWEAK/FN14 pathway in pathological angiogenesis, including retinal neovascularization. Similarly, the claim that FN14 inhibition does not affect VEGF-A levels and displays VEGF-independent anti-angiogenic effect is supported by similar findings reported earlier in a model of ischemia induced retinopathy.

    Overall, although the data show a contribution of FN14 in CNV pathology in laser induced murine model of CNV, the limited sample size of nAMD CNV membranes and lack of mechanistic insights in the rodent model, preclude the definitive link between TWEAK/FN14 pathway in nAMD.

    1. Major challenge for the current study are technical limitations relevant to obtaining a larger sample size for human CNV membranes and lack of a suitable preclinical model to investigate nAMD pathobiology. Specifically to link/implicate FN14 in nAMD would require validating the FN14 expression changes in CNV membranes from a bigger nAMD sample size that is likely not feasible. Similarly, although laser-induced CNV is routinely utilized for nAMD studies, this is a VEGF-dependent process where anti-VEGF is remarkably efficient in targeting CNV and therefore not ideal for investigating mechanisms and drug-targets for patient population not responsive to anti-VEGF therapy.

    2. The mechanistic basis of observed FN14 upregulation and consequence of FN14 upregulation need to be further clarified. FN14 is a highly inducible cytokine receptor that engages multiple signaling pathways, including nuclear factor-kB (NF-kB) and mitogen-activated protein kinase (MAPK). Given that the authors did not observe any significant upregulation of its ligand Tweak in human vitreous and plasma samples as well as in murine retina and RPE samples (Supplementary Fig. 4), makes it likely that FN14 mediated signaling pathway may be activated independent of its ligand TWEAK. Furthermore, it is well-established that FN14 is up-regulated after injury and apart from TWEAK expression of FN14 can be regulated by several cytokines and growth factors, and various other cytokines, including those that were upregulated in the current study (Figure 4). Therefore it would be important to interrogate if any of the upregulated cytokines (Figure 4) increases the expression and acts as a ligand for FN-14. Similarly, the human FN14 promoter region contains several potential transcription factor binding sites, including AP-1 sites and an NF-κB site. The activation of NF-κB involves two major signaling pathways, the canonical and non-canonical (or alternative) pathways. The canonical NF-κB pathway responds to diverse stimuli, including ligands of various cytokine receptors, pattern-recognition receptors (PRRs), TNF receptor (TNFR) superfamily members. Mechanistically, therefore it would be important to understand if FN-14 expression is upstream or downstream of these signaling pathways.

    3. With regard to the data on FN14 decoy receptor, given that a decoy receptor binds to a receptor and does not signal or activate the intended receptor complex thus acting as inhibitor, it would be important to show the effect of the decoy receptor on the known downstream molecules of FN14 such as NF-κB or MAPK. This is particularly important as only Il-6 expression were significantly down regulated in experiments utilizing FN14 decoy receptor (Figure 4).

  3. eLife

    Reviewer #2 (Public Review):

    Wolf and colleagues compare transcriptomic changes in human CNV membranes to-matched control RPE/choroid as well as in the mouse laser-induced CNV model. While it is conceptually a great study, this reviewer has concerns regarding the methodology. First, the comparison of isolated CNV membranes to normal RPE/choroid complex does not seem appropriate unless the CNV membrane contains RPE and choroid, which does not appear to be the case based on the histology shown. Since the RPE are believed to play a key role in CNV formation, analyzing the tissue without their contribution tells us little about this pathological process. Rather, analysis of the CNV membrane alone gives more information regarding CNV growth and scar formation.

    A related second concern is the timepoint of analysis for the mice. At 7 days post laser, the CNV is still forming in the mouse model. Therefore, it is not surprising that comparing this to advanced human CNV generated different results. Perhaps a later timepoint would have been more informative and comparable to human CNV membranes. Still, this timepoint yields interesting information regarding changes during CNV formation. Another question regarding the mouse laser model is how many laser CNV spots were included in each section for sequencing analysis. Also, it is unclear if CNV was somehow isolated or if entire sections from laser-induced CNV mice were used for RNA isolation. If CNV areas were not isolated, the normal tissue in these sections could reduce the transcriptional changes detected.

    Despite questions regarding the methodology, the conservation of the increased fibroblast growth factor inducible-14 between human and mouse is intriguing. Moreover, the inhibition of FN-14-TWEAK pathway further supports the role of FN-14 in CNV formation. The authors successfully demonstrate the potential role of FN-14 in CNV formation and, in doing so, present a potentially novel therapeutic target for neovascular AMD.

  4. eLife

    Reviewer #3 (Public Review):

    The authors here present a transcriptome analysis of choroidal neovascular (CNV) membranes and compare these data to a mouse model of laser-induced CNV. Through this analysis they found several genes that were differentially expressed within these human CNV lesions and the mouse CNV lesions, and found common orthologous genes that were differentially expressed. Via these analyses the authors identified conserved mediators of CNV formation in human and mouse including FN14, LGALS3, AIF1, UNC5B, ADAM15, MCAM, CYBB, APLNR, KCNN4, UNC93B1. The authors focused on FN14 and its ligand TWEAK as it has been previously shown to be involved in angiogenesis, and suggests the FN14 / TWEAK pathway to be a potential target for the treatment of CNV.

    They provide data that supports their conclusions, which have some strengths and weaknesses as follows.

    1. The study was able to perform transcriptome analysis of difficult to obtain patient-derived CNV samples. They used 4 sections from CNV samples that were surgically removed. This material is difficult to obtain, and is directly relevant for human disease. However, I am concerned about the age of these samples and the fact that these samples were paraffin embedded, and how this could have affected the transcriptome data and analysis.

    2. The authors used the commonly used mouse laser-induced CNV model, which is a surrogate model of human CNV. This model is more consistent with a wound healing model, and there is a combination of inflammatory, angiogenic, and fibrotic processes that occur throughout the model. Thus the rationale to compare these two different lesions may be limited, and makes it somewhat difficult to compare directly through a transcriptome analysis.

    3. Their transcriptome analysis is consistent with genes that are known to be unregulated in CNV lesions. FN14 itself has previously been shown by Ameri et al. to mediate angiogenesis in retinal neovascularization. The authors do an admirable job of attempting to validate this result in the CNV lesions

    4. The authors validated their study by immunofluorescence staining of FN14 within the the human and mouse CNV lesions. With respect to their suggestion that it may be used as a therapeutic, it is concerning that there is staining of the choroidal and retinal vessels in mice suggesting a normal physiologic role of FN14 within the retinal and choroidal vasculature as seen in Figure 3E, A. Though in humans there seems to be relatively low expression in the human retina.

    5. They observed the presence of the FN14 ligand TWEAK in human vitreous samples, again providing relevance for human pathology. However, there was no difference in TWEAK levels within the vitreous and plasma of control and patients with wet AMD, suggesting its does not play a significant role in pathology.

    6. Using an FN14 decoy they show decreased CNV lesion size and decreased IL-6 expression in the mouse CNV lesion, again providing supporting evidence of its role in treating CNV. However, these findings may only be relevant in the setting of laser-induced CNV, where there is known to be increased mediators of inflammation.

  5. Version published to 10.1016/j.bbadis.2022.166340
  6. Version published to 10.1101/2021.05.10.443381v1 on bioRxiv