Impaired yolk sac NAD metabolism disrupts murine embryogenesis with relevance to human birth defects

Reviewer #1 (Public Review):

Summary:

This study investigated the mechanism underlying Congenital NAD Deficiency Disorder (CNDD) using a mouse model with loss of function of the HAAO enzyme which mediates a key step in the NAD de novo synthesis pathway. This study builds on the observation that the kynurenine pathway is required in the conceptus, as HAAO null embryos are sensitive to maternal deficiency of NAD precursors (vitamin B3) and tryptophan, and narrows the window of sensitivity to a 3-day period.

An important finding is that de novo NAD synthesis occurs in an extra-embryonic tissue, the visceral yolk sac, before the liver develops in the embryo. It is suggested that lack of this yolk sac activity leads to impaired NAD supply in the embryo leading to structural abnormalities found later in development.

Strengths:

Previous studies show a requirement for HAAO activity for the normal development of embryos. Abnormalities develop under conditions of maternal vitamin B3 deficiency, indicating a requirement for NAD synthesis in the conceptus. Analysis of scRNA-seq datasets combined with metabolite analysis of yolk sac tissue shows that the NAD synthesis pathway is expressed and functional in the yolk sac from E10.5 onwards (prior to liver development).

HAAO enzyme assay enabled quantification of enzyme activity in relevant tissues including the liver (from E12.5), placenta, and yolk sac (from E11.5).

Comprehensive metabolite analysis of the NAD synthesis pathway supports the predicted effects of Haao knockout and provides analysis of the yolk sac, placenta, and embryo at a series of stages.

The dietary study (with lower vitamin B3 in maternal diet from E7.5-10.5) is an incremental addition to previous studies that imposed similar restrictions from E7.5-12.5.

Nevertheless, this emphasises the importance of the synthesis pathway on the conceptus at stages before the liver activity is prominent.

Weaknesses:

The current dietary study narrows the period when deficiency can cause malformations (analysed at E18.5), and altered metabolite profiles (eg, increased 3HAA, lower NAD) are detected in the yolk sac and embryo at E10.5. However, without analysis of embryos at later stages in this experiment it is not known how long is needed for NAD synthesis to be recovered - and therefore until when the period of exposure to insufficient NAD lasts. This information would inform the understanding of the developmental origin of the observed defects.

More importantly, there is still a question of whether in addition to the yolk sac, there is HAAO activity within the embryo itself prior to E12.5 (when it has first been assayed in the liver - Figure 1C). The prediction is that within the conceptus (embryo, chorioallantoic placenta, and visceral yok sac) the embryo is unlikely to be the site of NAD synthesis prior to liver development. Reanalysis of scRNA-seq (Fig 1B) shows expression of all the enzymes of the kynurenine pathway from E9.5 onwards. However, the expression of another available dataset at E10.5 (Fig S3) suggested that expression is 'negligible'. While the expression in Figure 1B, Figure S1 is weak this creates a lack of clarity about the possible expression of HAAO in the hepatocyte lineage, or especially elsewhere in the embryo prior to E10.5 (corresponding to the period when the authors have demonstrated that de novo NAD synthesis in the conceptus is needed). Given these questions, a direct analysis of RNA and/or protein expression in the embryos at E7.5-10.5 would be helpful.

Reviewer #2 (Public Review):

Summary:

Disruption of the nicotinamide adenine dinucleotide (NAD) de novo Synthesis Pathway, by which L-tryptophan is converted to NAD results in multi-organ malformations which collectively has been termed Congenital NAD Deficiency Disorder (CNDD).

While NAD de novo synthesis is primarily active in the liver postnatally, the site of activity prior to and during organogenesis is unknown. However, mouse embryos are susceptible to CNDD between E7.5-E12.5, before the embryo has developed a functional liver. Therefore, NAD de novo synthesis is likely active in another cell or tissue during this time window of susceptibility.

The body of work presented in this paper continues the corresponding author's lab investigation of the cause and effects of NAD Deficiency and the primary goal was to determine the cell or tissue responsible for NAD de novo synthesis during early embryogenesis.

The authors conclude that visceral yolk sac endoderm is the source of NAD de novo synthesis, which is essential for mouse embryonic development, and furthermore that the dynamics of NAD synthesis are conserved in human equivalent cells and tissues, the perturbation of which results in CNDD.

Strengths:

Overall, the primary findings regarding the source of NAD synthesis, the temporal requirement, and conservation between rodent and human species are quite novel and important for our understanding of NAD synthesis and its function and role in CNDD.

The authors used UHPLC-MS/MS to quantify NAD+ and NAD-related metabolites and showed convincingly that the NAD salvage pathway can compensate for the loss of NAD synthesis in Haao-/- embryos, then determined that Haao activity was present in the yolk sac prior to hepatic development identifying this organ as the site of de novo NAD synthesis. Dietary modulation between E7.5-10.5 was sufficient to induce CNDD phenotypes, narrowing the window of susceptibility, and then re-analysis of RNA-seq datasets suggested the endoderm was the cell source of NAD synthesis.

Weaknesses:

Page 4 and Table S4. The descriptors for malformations of organs such as the kidney and vertebrae are quite vague and uninformative. More specific details are required to convey the type and range of anomalies observed as a consequence of NAD deficiency.

Can the authors define whether the role of the NAD pathway in a couple of tissue or organ systems is the same? By this I mean is the molecular or cellular effect of NAD deficiency is the same in the vertebrae and organs such as the kidney. What unifies the effects on these specific tissues and organs and are all tissues and organs affected? If some are not, can the authors explain why they escape the need for the NAD pathway?

Page 5 and Figure 6C. The expectation and conclusion for whether specific genes are expressed in particular cell types in scRNA-seq datasets depend on the number of cells sequenced, the technology (methodology) used, the depth of sequencing, and also the resolution of the analysis. It is therefore essential to perform secondary validation of the analysis of scRNA-seq data. At a minimum, the authors should perform in situ hybridization or immunostaining for Tdo2, Afmid, Kmo, Kynu, Haao, Qprt, and Nadsyn1 or some combination thereof at multiple time points during early mouse embryogenesis to truly understand the spatiotemporal dynamics of expression and NAD synthesis.

Absolute functional proof of the yolk sac endoderm as being essential and required for NAD synthesis in the context of CNDD might require conditional deletion of Haoo in the yolk sac versus embryo using appropriate Cre driver lines or in the absence of a conditional allele, could be performed by tetraploid embryo-ES cell complementation approaches. But temporal dietary intervention can also approximate the same thing by perturbing NAD synthesis Shen the yolk sac is the primary source versus when the liver becomes the primary source in the embryo.

Author response:

General comments, factual mistakes:

Reviewer 1 - Summary: “This study builds on the observation that the kynurenine pathway is required in the conceptus, as HOO null embryos are sensitive to maternal deficiency of NAD precursors (vitamin B3) and tryptophan, and narrows the window of sensitivity to a 3-day period.”

Correction:

Vitamin B3 should not be in parentheses, because vitamin B3 and tryptophan are both NAD precursors. We also suggest that the second half of this sentence is changed to “…and narrows the window of sensitivity to a 3-day period from embryonic day 7.5 to E10.5.” Currently, it reads as if Haao-null embryos are sensitive to any 3-day period of maternal NAD precursor restriction.

Reviewer 1 – Strengths: “Abnormalities develop under conditions of maternal vitamin B3 deficiency, indicating…”

Correction:

We suggest replacing “vitamin B3 deficiency” with “NAD deficiency”, as this is more accurate.

Reviewer 2 – Strengths: “…and then re-analysis of RNA-seq datasets suggested the endoderm was the cell source of NAD synthesis.”

Correction:

We suggest re-phrasing this sentence to “…and then re-analysis of RNA-seq datasets suggested the yolk sac endoderm cells are the source of NAD de novo synthesis.”

Reviewer 1 (Public Review):

However, without analysis of embryos at later stages in this experiment it is not known how long is needed for NAD synthesis to be recovered - and therefore until when the period of exposure to insufficient NAD lasts. This information would inform the understanding of the developmental origin of the observed defects.

We are currently seeking funds to investigate the developmental origin of the observed defects. This study includes assessing how the timing of maternal NAD precursor restriction corresponds to the timing of NAD deficiency in the embryo.

More importantly, there is still a question of whether in addition to the yolk sac, there is HAAO activity within the embryo itself prior to E12.5 (when it has first been assayed in the liver - Figure 1C).

We have additional data showing that at E11.5 the embryo has no HAAO activity. We also tested E14.5 embryos with their livers removed, and these also do not have HAAO activity. We are planning to include these data sets in the revised version of this manuscript.

Reviewer 2 (Public Review):

Page 4 and Table S4. The descriptors for malformations of organs such as the kidney and vertebrae are quite vague and uninformative. More specific details are required to convey the type and range of anomalies observed as a consequence of NAD deficiency.

Kidney defects were classified as described in Cuny et al. 2020 PNAS (PMID:32015132). In brief, kidneys with a length (tip to tip) of ≤ 1.5 mm in length were counted as hypoplastic, because the average length of a normal kidney at E18.5 is 2.98 mm (2.75-3.375 mm). The one dysmorphic kidney we observed in our dataset had a cyst. We plan to include this information plus more details of the observed vertebral defects in the revised version of this manuscript.

Can the authors define whether the role of the NAD pathway in a couple of tissue or organ systems is the same? By this I mean is the molecular or cellular effect of NAD deficiency is the same in the vertebrae and organs such as the kidney. What unifies the effects on these specific tissues and organs and are all tissues and organs affected? If some are not, can the authors explain why they escape the need for the NAD pathway?

We agree that this is a very important question, but consider it beyond the scope of this manuscript. To elucidate the underlying cellular and molecular mechanisms in individual organs will require a multiomic approach because NAD is involved in hundreds of molecular and cellular processes affecting gene expression, protein levels, metabolism, etc. For details of NAD functions that have relevance to embryogenesis see Dunwoodie et al 2023 https://doi.org/10.1089/ars.2023.0349. Furthermore, organs develop at different times during embryogenesis with both distinct, but in some cases shared, molecular and cellular processes. Relating these to specific NAD functions is the challenge. We are currently seeking funds to investigate how NAD deficiency disrupts organogenesis.

Page 5 and Figure 6C. The expectation and conclusion for whether specific genes are expressed in particular cell types in scRNA-seq datasets depend on the number of cells sequenced, the technology (methodology) used, the depth of sequencing, and also the resolution of the analysis. It is therefore essential to perform secondary validation of the analysis of scRNA-seq data. At a minimum, the authors should perform in situ hybridization or immunostaining for Tdo2, Amid, Kmo, Kanu, Haao, Qprt, and Nadsyn1 or some combination thereof at multiple time points during early mouse embryogenesis to truly understand the spatiotemporal dynamics of expression and NAD synthesis.

We have tested antibodies against HAAO, KYNU, and QPRT in adult mouse liver samples (the main site of NAD de novo synthesis) which produced non-specific bands with western blotting. Therefore, in situ immunostaining studies on embryonic tissues are not feasible. We will investigate the possibility of effectively localizing transcripts of NAD de novo synthesis enzymes using in situ hybridization.

Absolute functional proof of the yolk sac endoderm as being essential and required for NAD synthesis in the context of CNDD might require conditional deletion of Haoo in the yolk sac versus embryo using appropriate Cre driver lines or in the absence of a conditional allele, could be performed by tetraploid embryo-ES cell complementation approaches. But temporal dietary intervention can also approximate the same thing by perturbing NAD synthesis Shen the yolk sac is the primary source versus when the liver becomes the primary source in the embryo.

Reviewer 1 has a related comment. We have additional data showing that at E11.5 the embryo has no HAAO activity, like the placenta. Similarly, E14.5 embryos with their livers removed, do not have HAAO activity either. We believe this provides sufficient proof that the yolk sac endoderm is the only site of NAD de novo activity in the conceptus until the liver has formed and takes over this function.