Merging organoid and organ-on-a-chip technology to generate complex multi-layer tissue models in a human retina-on-a-chip platform

Thank you for submitting your article "Merging organoid and organ-on-a-chip technology to generate complex multi-layer tissue models in a human Retina-on-a-Chip platform" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Milica Radisic as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by a Reviewing Editor and Marianne Bronner as the Senior Editor.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This paper describes a development of a retina organ-on-a-chip model. The paper is at a very forefront of the organ on the chip field. Retina is especially complex to model due to the large number of different cell types involved and complex structural organization. This paper describes integrating more than seven different cell types derived from human induced pluripotent stem cells in a vasculature like perfusion.

Importantly, the authors combined organoids with organ-on-a-chip engineering. This represents a high degree of novelty as it is not commonly done. They combined hiPSC derived retinal organoids with hiPSC-derived retinal pigment epithelial cells in a PDMS-based retina-on-a-chip device (RoC). Each device had four identical micro-tissues connected via a microchannel combining two transparent and biocompatible polymer layers.

The authors observed enhanced inner and outer segment formation and preservation, and they demonstrated the utility of this platform for drug testing. They demonstrated that retinal organoids expressed markers of major retinal cells such as: ganglion cells, bipolar cells, horizontal cells, amacrine cells, Müller glia and photoreceptors.

Overall this paper is strong in terms of the biological assessment as well as the engineering aspects related to device fabrication.

Essential revisions:

The reviewers agree that this manuscript is interesting and novel enough to warrant further consideration in eLife. The reviewers also agree that the authors should show the importance of perfusion in this system and demonstrate how it is superior to a transwell, which would form an important control. The reviewers agree this issue should be addressed experimentally. It is also necessary to perform a statistical review with an experienced statistician. The remaining small comments could be answered through appropriate explanation with references. Please see more detailed explanations of these comments in the reviews below.

Specific points:

Clarification on how long the entire structure is stable for and is there further remodelling of the structure with prolonged time in culture?

The phagocytic activity described in the paper is an inherent property of RPEs, and commercial assays exist to measure this function in mono-culture of RPEs using isolated, fluorescently labeled POS. Reproducing this function in the presented model system does not fully support the claim that RPEs interact with the retina organoids.

The acellular perfusion channel described in this model is not at all representative of the vasculature in the retina that is structurally and functionally integrated with other cell/tissue types in a complex 3D environment to maintain homeostasis. The configuration of the system may be used to simulate choroidal vessels but the channel used in this study is significantly divergent from choriocapillaris in vivo.

One potential benefit of perfusion is to maintain paracrine gradient between the retinal organoids and the epithelium. But without the transwell control, it's not clear how significant is this benefit in these experiments. For example, in Figure 5H, where would the condition with the co-culture of organoids and epithelium in a transwell setting be?

In the fluorescent intensity section, the authors should write out the formulas in the proper format.

The fourth sentence in the Abstract might need to be reworded.

Reviewer #1:

This paper describes a development of a retina organ-on-a-chip model. The paper is at a very forefront of the organ on the chip field. Retina is especially complex to model due to the large number of different cell types involved and complex structural organization. This paper describes integrating more than seven different cell types derived from human induced pluripotent stem cells in a vasculature like perfusion.

Importantly, the authors combined organoids with organ-on-a-chip engineering. This represents a high degree of novelty as it is not commonly done. They combined hiPSC derived retinal organoids with hiPSC-derived retinal pigment epithelial cells in a PDMS-based retina-on-a-chip device (RoC). Each device had four identical micro-tissues connected via a microchannel combining two transparent and biocompatible polymer layers.

The authors observed enhanced inner and outer segment formation and preservation, and they demonstrated the utility of this platform for drug testing. They demonstrated that retinal organoids expressed markers of major retinal cells such as: ganglion cells, bipolar cells, horizontal cells, amacrine cells, Müller glia and photoreceptors.

Overall this paper is strong in terms of the biological assessment as well as the engineering aspects related to device fabrication.

I would appreciate clarification on how long the entire structure is stable for and is there further remodelling of the structure with prolonged time in culture?

Reviewer #2:

This study is interesting and significant in that it is the first to demonstrate the incorporation of stem cell-derived organoids into RPE culture to model the retina, especially the photoreceptor cells which have proven challenging to recapitulate in vitro. Despite this positive aspect, the bulk of data presented in the paper are very preliminary and fail to substantiate the main claims made by the authors. First off, one major concern is that this system is presented as a vascularized retina model, which is completely misleading. The acellular perfusion channel described in this model is not at all representative of the vasculature in the retina that is structurally and functionally integrated with other cell/tissue types in a complex 3D environment to maintain homeostasis. The configuration of the system may be used to simulate choroidal vessels but the channel used in this study is significantly divergent from choriocapillaris in vivo.

The absence of data demonstrating the functionality and benefit of the perfusion channel is another major issue. In the introduction, the lack of vascularization in existing model systems is highlighted as one of the major problems that have motivated this study, but the presented results do not show whether and how the channel structure in the bottom layer serves to make this model superior and more physiological. Control data are nowhere to be found in the manuscript that compare the outcome of cell culture in the presence or absence of flow. On a related note, another critical question that remains unanswered is, is a chip-based system really necessary for this study? As discussed by the authors, MPS and organ-chip models offer numerous novel capabilities (e.g., precise control over the soluble and ECM microenvironment, incorporation and spatial patterning of different cell types, controlled delivery and application of external factors, etc.) not achievable in traditional in vitro models. It is highly questionable whether this study leverages any of these capabilities and whether this system is advantageous over conventional Transwell-based in vitro models. It would be readily possible to establish a similar co-culture model in Transwell inserts and conduct the same assays.

The claim that this model is the first to recapitulate the interaction between photoreceptor cells and RPEs also raises significant concerns. The phagocytic activity described in the paper is an inherent property of RPEs, and commercial assays exist to measure this function in mono-culture of RPEs using isolated, fluorescently labeled POS. Reproducing this function in the presented model system does not fully support the claim that RPEs interact with the retina organoids. The method of analysis used in this study is also problematic as poor resolution of imaging makes it difficult to interpret the data. It is understandable that the configuration of the device, especially the thickness of the bottom layer, presents challenges to high-resolution imaging analysis. If this really was the case, the poor imaging data seem to suggest the limitation of this chip-based model.

Reviewer #3:

The manuscript by Loskill et al., presents a Retinal-on-a-Chip that combines hiPSC-derived retinal organoids composed of all known major retinal subtypes and a monolayer of retinal pigment epithelial in a membrane-based microfluidic device. The paper provided through characterization on both retinal organoids and the retinal epithelium. When integrated, the study clearly demonstrated the functional indigestion of segment structures by the retinal pigment epithelium, which is a functional hallmark of the visual cycle. Furthermore, the Authors demonstrated pharmaceutical testing with an anti-malaria drug and an antibiotic and indicated that the co-culture condition could provide a protective role in the pharmaceutical response. Overall, I think the work is impressive, novel, and would be of great interests to the community. The data are well-presented and the discussions are very extensive and through. My only concern is regarding the utility of perfusion in the existing model. Although the experiments were done on a microfluidic platform, it seems all the experiment could also be performed in a standard transwell to achieve the same result. One potential benefit of perfusion is to maintain paracrine gradient between the retinal organoids and the epithelium. But without the transwell control, it's not clear how significant is this benefit in these experiments. For example, in Figure 5H, where would the condition with the co-culture of organoids and epithelium in a transwell setting be? I certainly see the potential utility of this model, especially by incorporating vasculature and perfusion in the future, which the authors have mentioned. Considering the potential of this platform and the biological finding presented, I think this paper can be accepted for publication.