Nuclear NAD⁺-biosynthetic enzyme NMNAT1 facilitates development and early survival of retinal neurons

Decision letter after peer review:

Thank you for submitting your article "Nuclear NAD+-biosynthetic enzyme NMNAT1 facilitates survival of developing retinal neurons" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Marianne Bronner as the Senior Editor. The reviewers have opted to remain anonymous.

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

Essential revisions:

1) The authors need to experimentally address the potential role of NMNAT1 in retinal progenitor differentiation, with especial consideration of amacrine, photoreceptor and bipolar progenitors. The reviewers share the opinion that the data presented in the manuscript could be the result of defective specification/differentiation and not only linked to degeneration as the authors claim. Thus, staining for progenitor markers and proliferation antigens at embryonic and postnatal stages is required. Any other approach aimed at addressing this issue will be equally valid.

2) Analysis of cone, amacrine and Müller cell specification/differentiation should also be included as morphological and RNA-seq data suggests possible decrease of their specification/differentiation.

3) A better description of the knockout efficiency with the Cre-lox system used and its comparison with those used in the other studies to investigate NMNAT1 requirement in the retina is needed.

4) Statistical analysis of the data should be revised in Figure 1 and added in Figure 5.

5) A number of different studies have used similar approaches (retina-specific NMNAT1 KO or KD, transcriptomics, and metabolomics) to shed light on NMNAT1 function in the retina. The authors need to clarify what are the knowledge gaps they are trying to fill in and should discuss better their findings in relation to the existing data (i.e. in relation to those of Eblimit et al., 2018 and Kuribayashi et al., 2018).

6) The authors may also consider rescuing the phenotype of the CKO retina by the administration of NAD or NMN, if experimentally feasible.

Reviewer #1:

Mutations in the gene nicotinamide mononucleotide adenylyltransferase-1 (NMNAT1) causes a specific type of severe and early onset retinal degeneration known as Leber congenital amaurosis type 9. By conditional inactivating the NMNAT1 gene in the mouse eye bud, this study extends on previous reports showing that NMNAT1 is required for postnatal survival of photoreceptors and specific retinal interneurons. Using transcriptomic and metabolomic approaches, the authors provide novel evidence that retinal degeneration is linked to the disruption of carbon, purine and amino acid metabolism and to the activation of distinct cell death pathways, including apoptosis, pyroptosis and necroptosis.

Although well conducted the study does not consider the possible role of NMNAT1 in retinal progenitor differentiation. This is an important point because the phenotypic characterization of the retina at postnatal development suggest that possibility that amacrine, photoreceptor and bipolar cells might be generated in lower number besides undergoing later degeneration. This point needs to be addressed. Furthermore, the authors fell short of discussing in the details their data in comparison with those of Eblimit et al., 2018, who have use different Cre line to delete NMNAT1 at slightly later stages of development or only in photoreceptors.

Figure 2. When comparing the number of amacrine and bipolar cells in the control groups, it is clear that both populations undergo an increase in size between P4 and P10. The same comparison in the mutants indeed suggest that bipolar cells degenerate and are lost. However, the number of amacrine cells at P10 is similar to that at P4. The authors conclude that there is also degeneration of amacrine cells but an alternative possibility is that cells are no longer generated and thus the population does not expand as in controls.

The HandE staining in Figure 1F seems to indicate the presence of cells in the ONL, presumably photoreceptor nuclei that however are strongly decreases at P30. The data in Figure 3 and the related supplementary information indicate a strong decrease of photoreceptor specific-gene expression. Again, the author conclusion is that NMNAT1 is required for postnatal survival of photoreceptors. Although it is clear that part of the phenotype is linked to the activation of cell death pathways, the authors should also consider the possibility that NMNAT1 may be also required for the generation of a proper number of photoreceptors. This needs to be determined. Markers such as Otx2 or Crx could be used to determine if the same number of photoreceptors are generated and then die perhaps because they cannot terminally differentiate or they are simply not generated in the same amount, as it may also happen for amacrine and even bipolar cells. This is particularly important because the Six3-Cre is an early driver and thus NMNAT1 absence could affect progenitor cells. Furthremore, the authors state "In conclusion, this study presents the most comprehensive evaluation of NMNAT1-associated retinal dysfunction to date and suggests crucial roles for nuclear NAD+ in the proper development and early survival of the mammalian retina". This conclusion at the moment is an overstatement as the development function of NMNAT1 is not addressed.

The idea that there might be an impact on progenitors is possibly supported by comparing the present results with those obtained by Eblimit et al., 2018 as well as those by Kuribayashi et al., 2018. In both cases knock down of NMNAT1 occurred with a different timing that correlated with less severe phenotypes.

The authors should also consider discussing their results with those of Eblimit et al., 2018 in more details, given that in the latter the reported use of photoreceptor-specific drivers suggests that reduction of the INL might be secondary to the loss of photoreceptors.

Reviewer #2:

NMNAT1 is known to be causal to Laber congenital amaurosis (LCA) which is congenital retinal dystrophies that appears at an early age. In this paper, the authors aimed to reveal the roles of Nmnat1 in retinal development by using retina specific conditional knockout mice. Interestingly, the loss of Nmnat1 by Six3-cre resulted in severe perturbation of the retinal development. The authors claimed the degeneration of the photoreceptors, but the results suggested lack of rod differentiation rather than degeneration since no evidence showing rod differentiation in the mice was included. In contrast, interneurons were once differentiated, then disappeared in the postnatal period. Since the special expression pattern of Cre or NMNAT1 in the retina was not available, it made difficult to relate the complex and cell type specific phenotype and Nmnat1. There are several other papers examining the role of Nmnat1 in retinal development in in vitro and ex vivo, precise comparison of the methods and data among the works may give interesting insight.

This paper describes examination of retinal phenotype of Nmnat1 by using retina specific knockout mice of Nmnat1. The paper contains the results of variety of analyses i.e. histological analysis, transcriptome analysis, and metabolic pathway examination. However, the whole part of this manuscript is descriptive, and did not give mechanisms how NMNAT1 regulates retinal differentiation. The authors claim that the severe photoreceptor degeneration was observed in this mice; however, I think that the phenotype showed lack of rod photoreceptor differentiation but not degeneration.

1. The Six3-cre mouse was used in this work, but this mouse line is known to show chimeric expression pattern of cre in the retina. i.e. "More detailed analysis of the recombination pattern of Six3-cre in the mature retina suggested it may incompletely excise target genes of interest. JNS 2007, 27, 12707-.". The authors observed strong phenotype in the central retina but not in the peripheral. Does it correlate with the expression pattern of the Cre protein? They should examine spatial expression of Cre or Nmnat1 in the CKO in retinas of several different developmental stages.

2. They did not perform any detailed examination of the retinal progenitor cells. Staining of progenitor markers and proliferation antigens of embryonic as well as P0 to P4 retinas may give useful information to understand the phenotype.

3. The authors claims that they observed photoreceptor degeneration, and NMNAT1 is crucial for the early survival of photoreceptors. If the authors observe the transition of rod photoreceptor that is once differentiation and then disappearance of them, this statement is appropriate. However, since neither Rhodopsin nor Recoverin expressed in the P4 retina, I think that failure of rod differentiation is suitable description to account for this phenotype.

4. RNA-seq data showed decreased cone related genes. The authors should confirm differentiation of cone in embryonic as well as postnatal retina of the CKO mice.

5. The retina at P4 showed no OPL, and previous literature indicated that the OPL is established by the interaction of photoreceptor and horizontal cells. Is horizontal cell differentiate in the CKO retina? In addition, how about Mueller glia differentiation in the CKO retina?

6. The authors observed lack of Synaptophysin, and I wondered postsynaptic formation of horizontal and bipolar cells.

7. Figure 1D lacks the scale bar, and that made us difficult to examine the panels.

6. Is it possible to rescue the phenotype of the CKO retina by administration of NAD or NMN locally or systemically?

Reviewer #3:

The authors aimed to disentangle the peculiar and complex aspects of NMNAT1-dependent retinal degeneration, with a particular focus on the underlying metabolic and transcriptional changes, used to hypothesize a mechanistic scenario explaining the pathophysiological picture associated with mutations at this locus. Overall, the paper is very well written and logical and provides a very comprehensive analysis spanning from detailed phenotypic analysis to metabolomics and transcriptomics approaches. This is definitely the major strength of the paper: experimental comprehensiveness and multilevel analysis. Using this approach, in general, the authors managed to provide interesting insights on the putative mechanistic scenario on the basis of the retinal degeneration caused by NAD+ and NMNAT1 deficiencies. Moreover, the cell-specific analysis provided also brings value and specificity to the study. Although the aim was obviously ambitious, the results provided well support most of their conclusions. In spite of this, I do have a major concern about the paper submitted. In general, the authors cite several papers with overlapping data/approaches to those proposed in the submitted manuscript (i.e. Lin et al., 2016, Greenwald et al., 2021, Wang et al., 2017, Lundt et al., 2021; Kuribayashi et al., 2018, Sasaki et al., 2020b). I understand that this is a hot topic in the field, that converging data is understandable, and that independent studies providing similar results are an extraordinary opportunity to validate the strength of the findings. However, I would encourage the authors to better state/define what differentiates some of their findings from those reported in the mentioned papers which used both retina or photoreceptor-specific NMNAT1 ablation, as well as metabolomic and transcriptional profiling of such transgenic animals. Although the authors use a very logical and comprehensive approach (and did a great job in presenting their findings with clarity), in several parts I kind of struggled a bit in recognizing the novelty of what they are bringing, and which kind of knowledge gaps they aim to fill, beyond the rhetoric of scientific writing.

A second aspect I would like to point out concerns the conclusions drawn by the authors concerning the different sensitivity of the distinct cell populations to the altered NAD+ homeostasis/NMNAT1 knockdown. It is well possible that amacrine, photoreceptor, bipolar cells, and RGCs respond differently to the tested conditions. However, it is also possible that photoreceptor cell degeneration (which occurs relatively early) represents also a cause of the later impairments in the other cell populations, affecting in turn their survival. Definitely, photoreceptors are, among the mentioned cells, those where altered REDOX chemistry (where NAD+ plays a crucial role) can have a bigger impact. I think that in the model proposed, it is not very clear why, the other cell types are affected as well, although later (even at P30 for RGCs). In other words, can we be sure that the degeneration observed in the different cell populations is actually cell-autonomous and not the result of a malfunctioning of one of the cell types? Are the effects on later cell types specific or just a consequence of the earlier impairment in photoreceptors? Maybe the authors could comment on this.

– Line 60: I found it interesting that retinas are particularly sensitive to NAD+ metabolism perturbations. Although the authors provide several examples in the literature, I would ask them to further comment (also in the paper) on why this tissue should be particularly reliant on NAD+ metabolism compared to the others (that to my knowledge should require a fine-tuned NAD+ homeostasis as well), also considering what they state in lines 71-75.

– Line 76: The authors define the other NMNAT as isoforms, but they are actually different genes, so I would define them as "other NMNAT genes", or, in case something is known about their phylogeny, use the more appropriate term paralogs (as I assume they are the results of vertebrate genome duplications).

– Line 123: what is the general range of knockout efficiency with the Cre-lox system, and what was the efficiency in the other papers that used a similar strategy to investigate NMNAT1 (see above). Can this explain some inconsistencies with some of the other studies?

– In Figure 1 legend it is not clear how many animals have been used as biological replicates (Figure 1C-J).

– Figure 1B: t-test is often accepted with 3 biological replicates, yet, theoretically, to assume a normal distribution you need more than 3 datapoints, so I would suggest a Mann-Whitney test (same applies to Figure 2, Figure 4E, J and supplementary figures using t-test, like Figure S3C, Figure S4E).

– Figure 1H-J: It is not possible in this case to use a t-test, as you would need anyway to correct for multiple comparisons even if you are interested in the KO vs wt comparison at a single specific distance. In this case, the authors should use an ANOVA or, given the consideration above (t-test vs Mann-Whitney test), with a non-parametric equivalent, like Kruskal-Wallis (in both cases corrected for multiple comparisons, as the more comparisons you are interested in, the more likely is to find something significant).

– Line 156: In this section, some % are given, yet it is not clear how many animals have been used to calculate such numbers. Did the authors use only one animal per timepoint? Even without statistical constraints (this is not the goal here), I would suggest at least 2-3 specimens.

– Line 332: not necroptosis as well? (See lines 218-221)

– As far as I see, no statistical analysis has been performed for Figure 5C, E, H. Is there a valuable and specific reason? If not, I would include a statistical significance assessment.

– Lines 445-452: I am not fully convinced by the explanation concerning the contrasting evidence between the manuscript and Kuribayashi et al., 2018. The current study reports defects in amacrine and bipolar cells in relatively later stages, so I am not sure whether an earlier knockdown can explain these discrepancies.

– Line 606: although I recognize this study as very well done and comprehensive, I do not think that the authors themselves should define their manuscript as the most… I suggest they change "the most… to date" with "a". I also would like to remember the authors that yes, their study might be the most comprehensive, but they largely base their study on a plethora of studies using very similar approaches and providing similar advancements in the field.

– Especially for scientists from different fields, I would spend few words explaining the relationships between the different cell types mentioned (how they are contributing to retinal development and vision, for instance).