Metabolic reprogramming by Zika virus provokes inflammation in human placenta

The recent outbreak of Zika virus (ZIKV) was associated with birth defects and pregnancy loss when maternal infection occurs in early pregnancy, but specific mechanisms driving placental insufficiency and subsequent ZIKV-mediated pathogenesis remain unclear. Here we show, using large scale metabolomics, that ZIKV infection reprograms placental lipidome by impairing the lipogenesis pathways. ZIKV-induced metabolic alterations provide building blocks for lipid droplet biogenesis and intracellular membrane rearrangements to support viral replication. Furthermore, lipidome reprogramming by ZIKV is paralleled by the mitochondrial dysfunction and inflammatory immune imbalance, which contribute to placental damage. In addition, we demonstrate the efficacy of a commercially available inhibitor in limiting ZIKV infection, provides a proof-of-concept for blocking congenital infection by targeting metabolic pathways. Collectively, our study provides mechanistic insights on how ZIKV targets essential hubs of the lipid metabolism that may lead to placental dysfunction and loss of barrier function.


4.
Example chromatograms for lipids measured should be shown in supplementary. 5. Authors should include their criteria for defining a peak e.g. signal/noise and points across a peak required for a lipid to be considered real and not below LOD. 6. For phospholipids/ceramide/sphingomyelins, provide full detail of this extraction method, and the LC/MS method which is missing. Provide information on whether this was an MRM method and if so what transitions were used? MS settings for this method are missing. 7. For bioactive lipid quantitation, provide more detail on SPE method, e.g. volumes, washing and elution solvents, etc. Also, full details on which standards and how this was done need to be included, including quantitation method, and what are the MS conditions, and the transitions used, etc.
8. Figure legends need to be far clearer on how lipids were measured, esp when just called CE or TAG, etc, and it is not clear how this was done.
9. Was smoothing used, and if so, this should either be not used, or a clear justification for why and how it was used needs to be provided, showing that it does not change the data or outcomes measured (generally smoothing is not recommended).
Reviewer #2 (Remarks to the Author): In this paper the authors have described Zika virus infection in placental samples from early human pregnancy. Using large scale metabolomics the authors show that there is extensive lipid metabolic reprogramming, accompanied by membrane rearrangements and mitochondrial dysfunction resulting in secretion of pro and anti-inflammatory mediators which contribute to tissue damage. The data presented in this study are interesting; especially characterisation of the placental cells as a model for Zika infection is certainly very useful. My primary concern is that this study lacks a hypothesis or a working model where the underlying mechanism is being addressed. It appears too descriptive at this point and seems to be a start for the purpose of addressing more in-depth questions in Zika infection. Given the descriptive nature of the data, it is not clear to me if there is a specific immune-metabolic pathway that contributes to Zika virus pathogenesis.
General comments: First, the observations on altered lipid metabolism, membrane rearrangements and mitochondrial dysfunction are not specific to placental cells -this has been observed in other Zika virus susceptible cells, including monocyte-derived cells and hepatocytes. It is not clear from the way the manuscript is written, whether this is a specific phenomenon that results in placental damage and associated foetal complications. Also, if (as per the authors) the primary cause of pathogenesis is virus triggered lipid metabolic alteration, this should be supported by studies with atleast a set of inhibitors targeting some of the obvious lipid metabolic pathways, or genetic depletion/deletion of some of the key genes to demonstrate that this is so. These would be simple experiments where specific depletion of genes for DAG biosynthesis/Kennedy pathway would be tested in infected cells to measure pathogenesis with readouts as production of inflammatory mediators, LD generation etc.
Specific comments: Figure 1b: Since the mock and infected samples are matched, it would be useful if the data points on the two columns were presented in that manner, so that individual increases were evident. Figure 1d-i: Given the significant increase in DAG and downstream products (PC, PE) containing long chain FAs, is it possible to test/comment on whether viral replication compartments are enriched in these specific lipid species? Or are they being utilised for altered placental membrane architecture as the authors suggest? Figure 2: Along the same lines as that of Figure 1, it would be useful to measure whether alterations in the specific enzymes in the Kennedy pathway/CDP-DAG pathway results in altered end products. These could be either enzyme abundance or activity in infected cells. Figure 3: Since most flavivirus infections are accompanied by consumption of lipid droplets, it would be useful to perform LD staining as a function of time in the infected cells to quantitate whether consumption is followed by increased synthesis and storage of LDs or vice versa. What is the authors explanation for increased LDs in uninfected cells? Is it triggered by secreted cytokines or other inflammatory mediators? Figure 5: Given the data on increased abundance and size of LDs in cells treated with conditioned UV-irradiated media, how can we exclude the possibility that the increases are an effect of immune response to infection and mediated through soluble cytokines or inflammatory regulators? It might be useful to treat these cells with a cytokine cocktail (or inflammatory mediators described in figure 8) to test their effect in LD abundance and volume.

Minor comments:
Line 90: The choice of first-trimester placenta needs to be better discussed. The sensitivity of placenta to ZIKA infection decreases along with the placental development (Arora, Sadovsky et al. 2017). It would be interesting to discuss the possible differences in lipid metabolic profiles of different placental cells upon ZIKA infection and its implication in ZIKA pathogenesis.
Reviewer #3 (Remarks to the Author): This intriguing manuscript by Chen et al. evaluates the impact of Zika virus (ZIKV) infection of the human placenta on metabolic reprogramming including mitochondrial function and the lipid metabolome. The manuscript is primarily descriptive. Alterations in lipid content might influence disease pathogenesis, although this cannot definitively be known based on the results presented here.
1. Although there are numerous statistically significant differences in lipid abundance resulting from ZIKV infection, it is not clear whether these effects are meaningful in terms of viral replication or disease pathogenesis. Perhaps if there were a way to manipulate lipid levels it would be possible to demonstrate the significance of these phenomena during ZIKV infection. (In other words, even if it is not possible to determine how metabolic reprogramming occurs as a consequence of ZIKV Infection, it seems important to be able to be able to prove whether this directly impacts disease pathogenesis. Perhaps an animal model exhibiting a genetic defects or alteration in a relevant metabolic pathway may be helpful here?) 2. Numerous flaviviruses including ZIKV are essentially pan-tropic and cause cell death. How much is virus-or DAMP-induced cell death contributing to the reported differences in lipid abundance in the placenta? Minor: 1. Although mentioned in the methods, the results section also should include more details about the specific strain, virus dose, etc. The methods also should include passage history. This is critical information for interpreting results. 2. For qRT-PCR experiments, were samples normalized to a standard curve? Were copies per mL determined based on a genome equivalent? 3. Numerous text corrections are needed: Line 38: Flaviviridae should be italicized Line 44: Disagree with the statement that mechanisms of fetal pathogenesis are poorly understood. There is a now a vast literature on the topic. Maybe instead this could be rephrased to say "incompletely understood." Line 91: The text should be specific about the ZIKV strain used. Which Asian strain? What is the name and identifier? Was it deep sequenced to verify absence of other pathogens? How was it propagated? Line 232: "pathlogenesis" (typo) Line 307: "plancental" (typo) Line 437: "health" placenta should be healthy control placenta There are quite a few text errors (e.g. inappropriately capitalized words, spelling, and/or grammatical errors).

Reviewer #1 (Remarks to the Author)
I am commenting on technical aspects of the lipidomics workflow only.The authors need to include more details on their lipidomics workflow in several places in order for the quality of the analysis to be properly evaluated.
1. How was the tissue homogenised for all these measures? Where were standards bought?

Author Response:
We appreciate the reviewer for pointing out that some technical details are not clear or missing. Each placenta sample was cut into several explants of 0.3 cm 2 (~200 mg). At the end of each experiment, tissue samples were snap-frozen then extracted as described in the revised manuscript (page 18, line 425) and supplementary experimental procedures (line 16). The final quantification of lipids is normalized to the protein levels as described in the supplementary experimental procedures (page 4). Information regarding internal standards has been added in supplementary procedures (page 2) "All the internal standards (IS) for neutral lipids were purchased from Sigma Aldrich. IS for phospholipids were purchased from Avanti Polar Lipids (Alabaster, LA, USA). Deuterated IS for bioactive lipids were purchased from Cayman Chemicals (Ann Arbor, MI, USA)".

Author Response:
We agree with the reviewer that information on quantification lipids should be provided. Quantification of different neutral lipid species was achieved by comparison of the peak area of individual species relative to their class-specific internal standards. Values are then normalized to sample protein levels. The relative abundance of TG represents the sum of the relative abundance of different subspecies in the TG group. This covers TG (16_16_16), TG (16_16_18), TG (16_16_18), TG (18_18_18), TG (18_18_20). The relative abundance of CE represents the sum of relative abundance of different subspecies in CE group including CE (C16) and CE (C18). Similar calculation modes were applied to determine the relative abundance of phospholipid species. All the details on how to determine lipid relative abundance have been added in the revised figure legends ( Fig.1 and Fig.2). The mass values used for each species have been provided in our supplementary tables (Table 2-5).

Author Response
The lipid nomenclature was corrected according to the reviewer's suggestions using the classical nomenclature on the LIPID MAPS. We revised all figures and figure legends throughout the paper as suggested by the reviewer.

Author Response
In the revised version of the manuscript, we included example chromatograms of neutral lipid (Extended Supplementary Figure 1) and phospholipids (Extended Supplementary Figure 2). We also included representative chromatographs of the bioactive mediators PGE2 and 17-HDoHE (Extended Supplementary Figure 3).

Author Response
Since neutral lipids and phospholipids were analyzed using a relative quantification approach, relative values above 10 mV cut-off were considered positive. All the details concerning our criteria for defining the peak definition have been published elsewhere 1,2 . For bioactive lipids, deuterated internal standards were chosen to fit different metabolite families. Absolute quantification was analyzed based on the calibration curves established from internal standards. A signal-to-noise ratio (s/n) over 10 was included in the analysis. Additional details on Machine settings and LOD for different compounds have been published elsewhere 3 . This information is now included in the revised supplementary experimental procedures (page 4).

Author Response
We have added all the details for the extraction of phospholipids/ceramide/sphingomyelins as well as the LC-MS/MS method in supplementary experimental procedures (pages 2-3). For these lipid analyses, we used MRM method. Transition information is now provided in the revised version of the manuscript under supplementary Table 4.

Author Response
No smoothing was used for our data analysis. We do agree with the reviewer that smoothing should be avoided.
Reviewer #2: Summary of comments from the reviewer: In this paper the authors have described Zika virus infection in placental samples from early human pregnancy. Using large scale metabolomics the authors show that there is extensive lipid metabolic reprogramming, accompanied by membrane rearrangements and mitochondrial dysfunction resulting in secretion of pro and anti-inflammatory mediators which contribute to tissue damage. The data presented in this study are interesting; especially characterisation of the placental cells as a model for Zika infection is certainly very useful. My primary concern is that this study lacks a hypothesis or a working model where the underlying mechanism is being addressed. It appears too descriptive at this point and seems to be a start for the purpose of addressing more in-depth questions in Zika infection. Given the descriptive nature of the data, it is not clear to me if there is a specific immune-metabolic pathway that contributes to Zika virus pathogenesis.

Author Response
In healthy pregnancy, the fetal placenta is able to develop in a finely balanced environment between inflammation and immune suppression. Localized inflammation within the maternal-fetal interface is necessary for fetal implantation and tissue remodeling. However, excessive production of inflammatory mediators can cause placental insufficiency. This has been illustrated in many of the great obstetrical syndromes such as preterm birth, growth restriction, gestational diabetes and pre-eclampsia which involve placental insufficiency. Furthermore, lessons learned from infectious agents associated with devastating developmental outcomes suggest that even subtle changes in maternal inflammation can lead to devastating outcomes. Hence, the effect of ZIKV infection on the lipid metabolome and major changes of the pro-and antiinflammatory equilibrium within the placenta may very well result in placental dysfunction. We have now modified the text in the introduction to put forward our working hypothesis and working model (lines 74-76).

General comments:
First, the observations on altered lipid metabolism, membrane rearrangements and mitochondrial dysfunction are not specific to placental cellsthis has been observed in other Zika virus susceptible cells, including monocyte-derived cells and hepatocytes. It is not clear from the way the manuscript is written, whether this is a specific phenomenon that results in placental damage and associated foetal complications. Also, if (as per the authors) the primary cause of pathogenesis is virus triggered lipid metabolic alteration, this should be supported by studies with at least a set of inhibitors targeting some of the obvious lipid metabolic pathways, or genetic depletion/deletion of some of the key genes to demonstrate that this is so. These would be simple experiments where specific depletion of genes for DAG biosynthesis/Kennedy pathway would be tested in infected cells to measure pathogenesis with readouts as production of inflammatory mediators, LD generation etc.

Author Response
We appreciate the reviewer's comment. During pregnancy, the placenta exhibits high metabolic activity. Lipid metabolites are not only participated in the placenta itself development but also can be transported to the fetus circulation. Metabolic alterations have been clearly associated with placental dysfunction. All these metabolism alterations can affect the placental membrane fluidity and metabolites transport to the fetus and impair fetus development. To our knowledge, our study is the first report that ZIKV infection induces global metabolic to the placental. In addition, our study provides the topological evidence that the ZIKV infection induces lipid droplet biogenesis and viral replication complex formation which fosters viral replication. The mitochondria damage and inflammatory mediators' production can induce server tissue damage. Impairment of these cellular hubs function has been clearly associated with pregnancy disorders.
According to the reviewer's suggestion, we performed additional analysis to identify cellular pathways impaired during ZIKV infection, and which can lead to placental metabolic alterations. We analyzed the expression of key enzymes involved in fatty acid transport/synthesis and LD biogenesis. Our data revealed that ZIKV induces a significant increase in the expression levels of all three enzymes. Since we are dealing with primary placenta cells, we used chemical inhibition rather than genetic modulation. Our analysis demonstrates that pharmacological inhibition of DGAT1 enzymatic activity suppresses LD biogenesis and reduces ZIKV replication, further providing a proof-of-concept approach for new opportunity to block placental infection (Fig. 6).

Author Response
We agree with the reviewer that it is clearer to present the data points per matched placenta sample. Therefore, graph in Fig. 1b was changed to depicted clearly matched placenta samples from mock-or ZIKV-infected conditions.

Author Response
It is likely that increased in DG as well as the downstream phospholipids PC and PE with long acyl chain participate in membrane biogenesis supporting building of replication factories. ZIKV, similar to other (+) RNA viruses, has to create new membrane webs with unique lipid composition to bring together cellular and viral factors that are necessary for viral replication. Many (+) RNA viruses appear to be dependent on PE-enriched membranes for replication, and depletion of PE can significantly inhibit viral RNA replication. Due to the cone-shaped lipid of PE, their enrichment during viral infection will induce negative membrane curvature and promote the formation of viral replication complexes. Enhanced PC levels were also shown to accumulate at HCV and poliovirus replication sites, revealing a common feature among the group of (+) RNA viruses. On the other hand, human placenta preferentially take-up long acyl chain fatty acids and regulate their delivery to the developing fetus for CNS development. Long acyl chain fatty acids are fundamental to membrane biogenesis during placental development and function. Furthermore, many physiological and biochemical adaptations that occur during pregnancy ensure availability of these fatty acids. Our observed alteration to specific phospholipids with long-chain fatty acid tails would impair the function of placenta and the channeling of fatty acids to the developing conceptus. Whist, the mechanisms of fatty acid transfer to the fetus are not clear, impaired placental delivery of fatty acids may constitute potential underpinning key to developmental outcomes with life-long health problems in congenitally afflicted infants. Therefore, by usurping essential fatty acids the virus can hamper also the placenta membranes.
In the future, it would be interesting to find out which lipids are specifically enriched in the replication sites and whether this would alter the membrane function of the placenta and/or deprive the fetus of essential lipids. Figure 2: Along the same lines as that of Figure 1, it would be useful to measure whether alterations in the specific enzymes in the Kennedy pathway/CDP-DAG pathway results in altered end products. These could be either enzyme abundance or activity in infected cells.

Author Response
We thank the reviewer for the suggestion and reiterate our response to the general comments above. We performed additional analysis to identify cellular pathways impaired during ZIKV infection, and which can lead to placental metabolic alterations. We analyzed the expression of FASN, FAT/CD36 and DGAT1 key enzymes involved in the fatty acid synthesis, transport, and lipid droplet biogenesis. Our analysis revealed that ZIKV induces a significant increase in the expression these three enzymes. Our findings suggest increased synthesis/ transport of fatty acids can serve as building blocks for biogenesis of LD and virus replication organelles. These new data are shown in the revised Figure 2 (Fi. 2h).

Author Response
Kinetics of viral replication revealed that ZIKV RNA can be detected 24h post-infection reaching a plateau at 5 dpi. Although the infection stayed high for several days, we decided to perform our lipid analysis on 5 dpi for optimal infection and to avoid artifacts from cytopathic viral effects or tissue decay. Nonetheless, we did perform analysis both at 3 days and five on isolated placental cells ( Fig. 4 and Supplementary Fig. 3). Our IF analysis of LD revealed increased the LD size and number at both time points with even higher increases at 5 dpi. Furthermore, inhibition of the DGAT1-mediated LD formation further supports the essential role of these hubs in viral replication. Altogether, our data suggest that LDs can act as a scaffold for efficient production of infectious progeny virions. Previous studies have shown that the dengue virus is able to deplete LD stores through an autophagydependent mechanism in Huh-7.5 cell line. Therefore, we consider the differences in virus-induced LD dynamics can be inherent the metabolic status of the type of cells. However, we cannot exclude that part of the LD stock is consumed to provide the necessary energy for the replication process.
The observed bystandard effect was further characterized in freshly isolated placenta cells. This effect can occur in cell-to-cell exchange of components or paracrine pathways via soluble factors. Our findings that UV-irradiated conditioned media can reproduce similar effects on non-infected cells favors a paracrine effect mediated through secreted key inflammatory mediators (lipid mediators, cytokine/chemokines resulting from placenta innate immune response and even extracellular vesicles needed for the crosstalk between other and fetus). However, we can exclude the effect of viral proteins and/or RNA in the absence of any replication. Future studies should identify what are the factors involved in ZIKV-induced bystandard effect. We agree with the reviewer that some of the pathways induced in our placental cells might be induced LD dynamic and expansion. Some of these inflammatory factors have been shown to induce lipogenesis. For instance, PGE2 from the COX pathway can induce the expression of a specific enzyme of lipogenesis and trigger lipid accumulation hepatocytes. At this point, we did not test the direct effect of specific lipid mediators or cytokine/chemokine on first trimester placenta samples During early pregnancy, the level of prostaglandins is tightly regulated to establish a balanced e inflammatory microenvironment needed for the implantation of the embryo. Their upregulation can hamper implantation and placental development. Future targeted studies would establish whether a given bioactive mediator and/or cytokines/chemokine can induce LD dynamics in placental cells knowing that the balance of immune mediators is finely tuned in this tissue.

Minor comments:
Line 90: The choice of first-trimester placenta needs to be better discussed. The sensitivity of placenta to ZIKA infection decreases along with the placental development (Arora, Sadovsky et al. 2017). It would be interesting to discuss the possible differences in lipid metabolic profiles of different placentalcells upon ZIKA infection and its implication in ZIKA pathogenesis.

Author Response
We agree with the reviewer and we amended the text and provided a justification for our choice to work with first trimester samples (lines 90-94). We also included the reference by Arora Sadovsky et al. 2017 (ref 37 of the revised manuscript).

Author Response
The typo mistake has been corrected in the revised manuscript ( In our paper, we have shown that ZIKV infection reprograms placental cell metabolism. To illuminate cellular pathways responsible for ZIKV-induced alteration, we analyzed the expression of key enzymes involved in fatty acid synthesis, transport, and LD biogenesis. Our analysis revealed that ZIKV induces significant increase in the expression of FASN, FAT/CD36 and DGAT1 enzymes. These findings suggest that increased synthesis/transport of fatty acids can serve as building blocks for biogenesis of LD and virus replication organelles. More importantly, our new data demonstrating the efficacy of chemical inhibitors of DGAT1 enzymatic activity in limiting ZIKV infection provide proof that LD can be used a scaffold for viral replication. Previous studies have provided evidence that alterations and overload of specific lipids in the fetal placenta are associated with placental insufficiency and severe pregnancy complications such as preeclampsia, preterm birth and fetal growth restriction. Our data suggest that a safe manipulation of the placental lipids might be an alternative therapy to limit congenital infection and/or the occurrence of adverse pregnancy disorders.
Mouse models have proven useful to study several aspects of the ZIKV infection, including maternalfetal transmission, tissue tropism, and mechanisms of virus replication as well as vaccine development. Human pregnancy differs from mouse pregnancy because of the unique fetal trophoblast invasion into the maternal decidua basalis as well as an increasing number of factors that are specific to human placentation. Nonhuman primates are phylogenetically closely related to humans. However, the use of nonhuman primates, particularly during gestation, raises major ethical issues. Despite the fact that many aspects of human placentation can only be addressed using human cells and tissues, it is very unlikely for us to access such a model.

Numerous flaviviruses including ZIKV are essentially pan-tropic and cause cell death. How much is virus-or DAMP-induced cell death contributing to the reported differences in lipid abundance in the placenta?
Author Response: It is clear that oxidized lipids and DAMPs released by dead cells can change the inflammatory microenvironment further amplifying the viral effect. Our analysis of freshly isolated placental cells was performed 3dpi and 5dpi with a low multiplicity of virus. We did not observe any cytotoxic effect at these time points, yet we have a significant effect on LD morphology and growth dynamics. Although we cannot exclude that early release of mediators can act as DAMPs or PAMPs, we believe that the effect we are seeing is virus induce. In addition, Ceramide is well known for its role as second messengers inducing cell apoptosis. But, we did not observe significant changes in the ceramide level under ZIKV infection.

Minor point
1. Although mentioned in the methods, the results section also should include more details about the specific strain, virus dose, etc. The methods also should include passage history. This is critical information for interpreting results.

Response
We agree with the reviewer that information on the virus could be given in the result section. Therefore, we did summarize to provide minimal information about the specific strain and the amount of virus in the result section. These changes are highlighted in (lines 92-94) of the revised result section "Placentas were challenged with the Asian strain (KU886298) of ZIKV at 6x10 10 RNA copies/mL (equivalent to MOI of 1) obtained from second passage in Vero cells [4][5][6][7][8] .We also provided extensive information about the strain, propagation mode and included additional references in the revised methods (lines 436-443) (see detailed response bellow).
2. For qRT-PCR experiments, were samples normalized to a standard curve? Were copies per mL determined based on a genome equivalent?

Author Response
ZIKV RNA was detected by amplification of an NS5 fragment using RealStar Zika virus RT-PCR kit 1.0 (Altona Diagnostic GmbH, Germany). Quantification of viral production was determined as RNA copy number per mL, calculated based on a standard curve. All the details are given in the revised methods (lines 460-466)

Numerous text corrections are needed:
Line 38: Flaviviridae should be italicized

Author Response
Flaviviridae in Line 38 was changed to Flaviviridae (line 39) Line 44: Disagree with the statement that mechanisms of fetal pathogenesis are poorly understood. There is a now a vast literature on the topic. Maybe instead this could be rephrased to say "incompletely understood."

Author Response
We agree with the review that huge amount of work has been provided by several groups. Thus, the term 'poorly understood' has been replaced according to reviewer suggestion by the term 'not fully understood' (line 44) Line 91: The text should be specific about the ZIKV strain used. Which Asian strain? What is the name and identifier? Was it deep sequenced to verify absence of other pathogens? How was it propagated?

Author Response
We reiterate our response to point 1. ZIKV was isolated from sperm sample of a 32-year-old immunocompetent Caucasian patient returning from Brasil (KU886298) [4][5][6][7][8] The patient was tested HIV negative. Potential infection with dengue virus or chikungunya virus was ruled out by ELISA Diapro (Diagnostic Bioprobes Srl, https://www.diapro.it) and RealStar Dengue and Chikungunya qRT-PCR (Altona Diagnostics, https://www.altona-diagnostics.com). The partial sequence of NS5 protein has been submitted to GenBank (KU886298). Sequence alignment and phylogenetic clustering attributed the isolated strain to the Asian lineage of ZIKV. Although the description of viral strain was given in the original submission under the method section, we, now provide additional information about the viral strain identification and virus propagation under the revised methods section. High titer stocks were obtained from early passaging in Vero cells as previously described 4 . Viral titers were determined by plaque assays on Vero cells. High titer viral stocks were stored at -80°C in singleuse aliquots.