The impact of altered nutrient availability on brain development during the perinatal period is widely acknowledged and is corroborated by both observations on humans and animal studies. Fetal malnutrition may be the consequence of an imbalanced maternal nutrition or placenta deficiency [1
], with both possibly resulting in Intra Uterine Growth Restriction (IUGR) and a risk of preterm birth. In addition, very preterm infants often experience poor early postnatal growth, characterized by a deficit of lean mass [3
]. This is often associated with neurological impairments during infancy [3
]. Poor fetal growth is also known to confer a risk of developing metabolic diseases in adulthood according to the thrifty phenotype hypothesis [5
], with the consequence being the impaired control of energy homeostasis. The hypothalamus, because of its central role in the regulation of energy homeostasis and food intake, has been intensely studied using animal models. Malnutrition in the perinatal period is associated with impaired hypothalamus development as well as altered leptin and insulin signaling, leading to defects in the control of food intake [6
]. However, although the impact of perinatal nutrition on postnatal development of the hypothalamus and its functional consequences have been widely described [7
], little is known about alterations that may take place during the morphogenesis of the fetal hypothalamus, probably because of the enormous complexity of the anatomy and functionality characterizing this brain region [11
The development of the hypothalamus starts during early embryonic development with anterior–posterior patterning of the developing neural tube. The numerous nuclei of the hypothalamus are generated between E11 and E17 embryonic days in rodents [11
]. This morphogenesis step is followed after birth, during the first two weeks of life in rodents, by the organization of nuclei that connect to each other and towards other regions of the brain [13
The different cell types found in the hypothalamic nuclei originate from undifferentiated dividing neural progenitor cells (NPCs) residing in the ventricular zone of the brain, a transient embryonic layer of tissue. These NPCs, also named radial glial (RG) cells, are derived from neuroepithelial cells and give rise to committed neuronal and glial progenitors that migrate and differentiate according to a strictly defined program. In rodents, hypothalamic neurogenesis occurs prenatally between E12 and E17 and precedes astrogenesis that takes place during the postnatal period [13
]. Because cell number and neurogenesis are determined during the prenatal period, the fetal environment, including maternal nutrition and metabolic status, may impact these processes, as already demonstrated by several studies on hippocampal and cortical tissues (reviewed in [15
]). For instance, fetal malnutrition was associated with an alteration in the level of neuronal proliferation, maintenance and apoptosis of hippocampic cells [16
]. Maternal protein restriction has been shown to reduce the proliferation of neural stem cells and to influence progenitor cell fate during embryonic cortex development in mice, resulting in increased cortex thickness [18
Neural stem cells show a spectacular plasticity in their capacity to differentiate into a variety of cell types. This confers to the developing brain a great adaptability but also a high sensitivity to external cues. Hypothalamic progenitor cells were shown to respond in vitro to environmental stimuli such as the neurotrophic hormones insulin and leptin [19
] or the endocrine disruptor Bisphenol A [20
However, even if the impact of the perinatal environment on hypothalamus development is now clearly established, the underlying mechanisms remain largely unexplored. The proliferation and differentiation of neural stem cells and progenitors is strongly based on the precise control of the expression of specific genes, including pluripotency genes and lineage-specific genes. The maintenance in an undifferentiated state or the commitment into neuronal or glial differentiation requires a complex interplay between external cues, transcription factors, DNA-binding proteins, epigenetic control of gene expression and possibly other, as yet uncharacterized, mechanisms [21
]. Overall, evidence is growing that gene expression is finely regulated both at the transcriptional and translational level. Given all this complexity, it makes sense that environmental factors can act on many levels.
The objective of this study was to identify early determinants of the impact of maternal protein restriction on hypothalamic development and to characterize, at the molecular level, early indicators of an impaired development. We use a well-characterized rat model of maternal protein restriction during gestation, which was designed to mimic placental defects, often resulting in the altered transfer of amino acids between mother and child [22
]. We have previously shown that protein deficiency during gestation and lactation results in alterations in the development of the hypothalamus, leading to defects in the control of food intake [8
] and metabolic alterations during adulthood [23
Our strategy was to analyze fetal hypothalami at E17, which approximately represents the end of neurogenesis in rats [11
]. Using 3′ digital gene expression sequencing (DGE-seq) for differential gene expression analysis, pathway enrichment analysis and N6-methyladenosine (m6A) RNA methylation assay, we have sought to identify possible molecular targets of fetal undernutrition that underlie alterations of in neurogenesis process in the hypothalamus.
2. Materials and Methods
All experiments were carried out in accordance with current guidelines of the local animal welfare committee and were approved by the Animal Ethics Committee of Pays de La Loire under reference 2016112412253439/APAFIS 7768. Nulliparous female Sprague–Dawley rats were purchased from Janvier Labs (Le Genest Saint Isle, France) and delivered to our facilities at the age of 7/8 weeks. On arrival, rats were housed (two per cage) under controlled conditions (22 °C, 12 h/12 h dark/light cycle) with free access to a standard diet (A04, SAFE-diets, Augy, France). After one week of acclimation, the estrous cycle was determined by vaginal smears and female rats in early estrous were mated overnight with a male. The presence of spermatozoa was verified the next day through vaginal smears and, when positive, this day was considered embryonic day 0 (E0). Pregnant rats were housed individually and randomly assigned to two experimental groups receiving either a control (C) diet containing 20% protein or a protein-restricted (PR) diet containing 8% protein (UPAE, Jouy-en-Josas, France). A detailed composition of the diets was described in [24
]. At embryonic day 17 (E17), dams were anesthetized with 4% isoflurane and fetuses were sampled by caesarian section. Fetuses and placentas were weighed and brains were rapidly removed and dissected under a binocular magnifier to collect the hypothalamus. Some hypothalami were snap frozen in liquid nitrogen and stored at −80 °C for transcriptomic and proteomic analysis, while others were collected in cold PBS containing 2% glucose for cell biology experiments.
2.2. In Vitro Culture of Neurospheres from E17 Hypothalamus
For each litter, we collected six hypothalami to prepare neurospheres. After one wash in 2 mL of sterile PBS containing 2% glucose, hypothalami were mechanically triturated in 1 mL of NeuroCult Basal Medium (STEMCELL Technologies Inc., Vancouver, BC, Canada) using a 1 mL micropipette until a single-cell suspension was obtained. The cell suspension was then filtered on a 40 µm cell strainer (Greiner Bio-one International GmBH, Kremsmünster, Austria) and centrifuged at 500× g for 5 min. Cell pellets were resuspended in NeuroCult Basal Medium and viable cells were counted on a hemocytometer after eosin staining. All cells from one hypothalamus were seeded in a T-12.5 cm2 tissue culture flask containing 5 mL of Complete NeuroCult TM Proliferation Medium and incubated at 37 °C and 5% CO2. The obtained neurospheres were passed after 3 days in vitro. Briefly, cell passages were done by centrifuging neurospheres at 90× g for 5 min and incubating pellets with 200 µL of Accutase (STEMCELL Technologies Inc.) followed by gentle trituration to obtain single-cell suspensions. Cells were washed in NeuroCult Basal Medium, centrifuged at 500× g for 5 min and resuspended in NeuroCult Basal Medium, Complete NeuroCult TM Proliferation Medium or Complete NeuroCultTM Differentiation Medium, depending on the following experiment.
2.3. Proliferation Test Using BrdU
Cell proliferation was measured with a BrdU Cell Proliferation colorimetric ELISA Kit (ab126556, Abcam, Cambridge, UK) following the manufacturer’s manual. This experiment was performed on passaged cells following neurosphere culture. Briefly, 20,000 cells resuspended in 100 µL of Complete NeuroCult TM Proliferation Medium were seeded in coated 96-well plates, incubated with 20 µL of 1× BrdU Reagent at 37 °C and 5% CO2 for 24 h and fixed with the provided solution. Fixed cells were incubated with anti-BrdU primary antibody, horseradish peroxidase-conjugated secondary antibody and tetramethybenzidine (TMB) substrate and absorbance at 450 nm was measured using a Varioskan LUX (ThermoFisher Scientific, Waltham, MA, USA)
Immature and mature neurons, undifferentiated cells and proliferative cell proportions were determined by immunochemistry using anti-TUJ1 (1:1000; MMS-435P-100 Eurogentec, Liège, Belgium), anti-MAP2 (1:100; #4542 Cell Signaling Technology, Leiden, The Netherlands), anti-NES (1:250; ab92391, Abcam) and anti-Ki67 (1:250; ab66155, Abcam) antibodies, respectively. Immunocytochemistry was performed on cells obtained after hypothalamus dissociation and on neurosphere cells resuspended in NeuroCult Basal Medium. Cells were fixed on Lab-Tek™ II Chamber Slide™ System (154534, ThermoFisher Scientific) with PBS and 4% paraformaldehyde (PFA). Blocking was done with an incubation step with PBS, 3% Bovine Serum Albumin (BSA) and 0.2% Triton for 1 h at room temperature and primary antibodies were added overnight at 4 °C. After three washings with PBS, secondary antibodies: Alexa 647-conjugated donkey anti-mouse (1:1000; 715-605-150, Jackson, Cambridge, UK), Alexa 647- conjugated donkey anti-rabbit (1:1000; 711-606-152, Jackson, Cambridge, UK) and biotin-conjugated goat anti-rabbit (1:1000; A24541, ThermoFisher Scientific were added and incubated for 1 h. After three washes, streptavidin Alexa 568 (1:1000; s11226, Molecular Probes, Eugene, OR, USA) was added to the wells containing biotinylated secondary antibodies. Cells were then incubated for 5 min at room temperature with DAPI (1/10,000; D3571 Molecular probes) and washed. The chambers were removed and the slides were mounted in VECTASHIELD®
Vibrance™ Antifade Mounting Medium (Vector Laboratories, Burlingame, CA, USA). Pictures of each well were obtained using ×20 magnification on Zeiss Axio Imager.M2m microscope and positive cells for each marker were automatically counted with an ImageJ [25
] script based on object detection using signal intensity.
2.5. Mitochondrial Membrane Potential Determination
For each litter, two hypothalami were collected in cold PBS and 2% glucose was used to determine mitochondrial membrane potential using MitoTracker Red CMXRos (ThermoFisher Scientific) staining. Cell dissociation and counting were performed as previously described for cell culture. About 20,000 viable cells were incubated 30 min with Mitotracker Red CMXRos (500 nM). Cells were then centrifuged at 700× g for 3 min and resuspended in PBS before the fluorescence was read in Varioskan LUX (ThermoFisher Scientific) (Ex 579 nm, Em 599 nm). The values of the fluorescence were normalized to the number of viable cells seeded.
2.6. Western Blotting
Proteins were extracted from E17 fetal hypothalami stored at −80 °C with lysis buffer containing Radio Immunoprecipitation Assay (RIPA) lysis buffer (EMD Millipore Corp, Burlington, MA, USA), protease inhibitor and phosphatase I and II inhibitor (Sigma-Aldrich, Saint-Louis, MO, USA). Lysis buffer was added to each sample then hypothalami were shredded (Precellys® Ozyme, 2 × 15 s at 5000 rpm) and centrifuged at 5590× g for 5 min at 4 °C. Protein concentrations were measured with Pierce™ BCA Protein Assay Kit (ThermoFisher Scientific) and 25 µg of proteins per sample were used for Western Blot. For subsequent labelling with the CSDE1 antibody, proteins were denatured with a heating step at 95 °C for 5 min with Laemmli Sample Buffer (Bio-Rad, Hercules, CA, USA) whereas, for the OXPHOS antibody, proteins were not denatured as recommended by the manufacturer. Proteins from the extracts were separated on a 4%–15% precast polyacrylamide gel (Bio-Rad) then transferred onto nitrocellulose membrane with the Trans-Blot Turbo Transfer System (Bio-Rad). For the membranes that were subsequently labelled with the OXPHOS antibody, total proteins were stained using the RevertTM 700 Total Protein Stain (LI-COR Biosciences, Lincoln, NE, USA) following the manufacturer recommendations. The total amount of proteins per sample was quantified on the Odyssey (LI-COR Biosciences) using Image Studio Ver 5.2 (LI-COR Biosciences) and the EMPIRIA Studio software Ver 1.2 (LI-COR Biosciences) and used for normalization. The Revert stain was removed from the membrane using the Revert Reversal Solution before the incubation with the antibody. Membranes were then blocked in TBST containing 5% dried fat-free milk and incubated overnight with primary antibodies anti-CSDE1 (1:1000; ab 201688,), anti-β-ACTIN (1:7,500; A5441, Sigma-Aldrich) or anti-OXPHOS cocktail (1:250; ab 110413, Abcam) then one hour with secondary antibodies goat anti-rabbit IgG DyLight 800 (1:10,000, SA5-10036, ThermoFisher Scientific), goat anti-mouse IgG DyLight 680 (1:10,000, 35519, ThermoFisher Scientific) and goat anti-mouse IgG Dylight 800 (1:10,000, SA5610176, ThermoFisher Scientific). Immunolabelling was then revealed on the Odyssey (LI-COR Biosciences) using Image Studio Lite Ver 5.2 (LI-COR Biosciences). For the experiment with the anti-CSDE1 antibody, normalization of the signal was performed using the anti-β-ACTIN antibody.
2.7. m6A RNA Methylation Assay
The total amount of m6A in total RNA was measured using the m6A RNA Methylation Assay Kit (Fluorometric) (ab 233491, Abcam), following the manufacturer manual. For each sample, 200 ng of total RNA from E17 hypothalamus preparation were used.
2.8. 3′DGE Library Preparation, Differential Gene Expression Analysis and Enrichment Analysis
Total RNA and DNA were extracted simultaneously from hypothalami stored at −80 °C using NucleoSpin®
RNA columns and RNA/DNA buffer set (Macherey–Nagel, Hoerdt, France) following the manufacturer’s manual. Transcriptomic analysis was performed using 3′DGE (Digital Gene Expression)-sequencing in accordance with [26
]. Briefly, mRNA libraries were prepared from 10 ng of total RNA from 96 individuals (eight males and eight females from each experimental group collected at three different ages (E17, D0 and D130)). Poly(A) mRNA tails were tagged using universal primers, sample-specific barcodes and a unique molecular identifier (UMI) and cDNA synthesis was performed using template-switching reverse transcriptase. Samples were then pooled, amplified and fragmented using a transposon fragmentation method that enriches for 3′ ends of cDNA. Fragments ranging from 350–800 bp were selected and sequenced on an Illumina Hiseq 2500. Paired-end sequencing was performed using a Hiseq Rapid SBS kit v2 50 cycles (FC-402-4022) and a Hiseq Rapid PE Cluster kit v2 (PE-402-4022). The first read of 16 bp corresponds to the sample-specific barcode and the second read of 57 bp to the mRNA in the 5′ 3′ direction. Alignments were done on RefSeq rat mRNA sequences (Rn6) by using BWA (version 0.7.15-0). The number of unique UMI associated with each RefSeq gene was counted. Only genes with three or more reads per sample in at least four samples were kept in the count table. Normalization and differential gene expression analyses were processed using DESeq2 (version 1.24.0) [27
] with a correction for sex effect. Data from the 96 samples were used for normalization and differential expression analysis was performed separately for each age. Functional enrichment was done using FGSEA [28
] from Gene Ontology (GO) [30
], Kyoto Encyclopedia of Genes and Genomes (KEGG) [31
] and Reactome [32
] databases. The datasets generated for this study can be found at the European Nucleotide Archive (ENA) under the accession PRJEB35794.
2.9. Mitochondrial DNA Quantification Using qPCR
Mitochondrial DNA was quantified by quantitative PCR using primers designed against the mitochondrial CytB gene (Forward 5′-TTCCGCCCAATCACCCAAATC-3′, Reverse 5′-GCTGATGGAGGCTAGTTGGCC-3′) and normalized against the geometric mean of the amplification signal from two nuclear genes: Gapdh (Forward 5′-TTCAACGGCACAGTCAAGG-3′, Reverse (5′-CTCAGCACCAGCATCACC-3′) and Zfx-ZFy (5′-AAGCATATGAAGACCCACAG-3′, Reverse 5′-CTTCGGAATCCTTTCTTGCAG-3′). Ten ng of total DNA were amplified in a total volume of 15 µL using the iTaq™ Universal SYBR®Green Supermix (Biorad) and 0.25 µM of each primer following the manufacturer’s instructions, in a CFX Connect™ Real Time PCR Detection System (Biorad). A relative amount of mitochondrial DNA was quantified using the 2−ΔΔCt method.
Either a Mann–Whitney or t-test was used to evaluate differences between groups and a two-way analysis of variance (ANOVA) was used to additionally assess sex effects. For each experiment, the tests are indicated in Figure and Table legends. The statistical analyses were performed using R software (version 3.6).
In this study, we used a well-characterized model of maternal protein restriction during gestation to identify gene families and physiological pathways that were altered in the fetal hypothalamus in response to the maternal PR diet.
The impact of an imbalanced maternal diet on the proliferation and differentiation capacities of neural stem cells during embryo and fetal development is now clearly established [18
]. Our observations on the cells sampled on E17 fetuses and grown in vitro as neurospheres confirmed these alterations. The total number of cells after dissociation of the fetal hypothalami was lower in the PR group, which may reflect reduced proliferation in an earlier period in the PR group. In addition, we observed, after three days of proliferation in vitro, a higher proportion of TUBB3+ cells in the PR group, which may reflect that E17 PR fetuses had initially a higher number of committed neuronal progenitors that proliferated in culture. Similarly, Gould et al. [18
] showed that low protein diet throughout gestation was associated with an increase in the number of late neural progenitors in mice brain but tempered by increased apoptosis. Further investigation would be required in order to establish whether this was also the case in our model.
Interestingly, the proportion of cells expressing NES and MAP2 were not different between groups whereas the genes encoding these proteins were under-expressed in the PR group. This may be related to the fact that the level of expression of these genes varies throughout the differentiation process, from early to late progenitors until differentiated neurons. Therefore, the difference in expression level might reflect an alteration in the timing of differentiation which cannot be seen in the immunochemistry experiments that do not distinguish between cells that have variable levels of gene expression. Additionally, we cannot exclude post-transcriptional control of expression.
E17 corresponds approximately to the time when neurogenesis is complete and residual NPCs start to differentiate into astrocytes [13
]. Indeed, no cell was GFAP-positive at E17 and the Gfap
gene was not expressed. The switch between neurogenesis and astrocytogenesis is based upon a complex interaction between external signals and a cell-intrinsic program via a strict control of gene expression. This interaction first requires nutrient sensing and detection of metabolic and hormonal signals coming from the mother and the placenta and then the activation of regulatory pathways that control cell differentiation. By using a large scale transcriptomic approach on whole fetal hypothalamus, we highlighted several metabolic pathways and molecular regulation systems that were impacted by maternal PR and led us to propose some mechanistic hypotheses in order to explain alterations of various neurodevelopment processes.
Our data suggested an alteration of the mitochondrial respiratory chain activity in the PR group, as evidenced both by the over-expression of genes encoding the complexes of the respiratory chain and several enzymes from the pyruvate and citric acid metabolism as well as the increased mitochondrial membrane potential of the E17 hypothalamic cells. Mitochondrial DNA copy number was not different between control and PR fetal hypothalamus, suggesting that the difference in the respiratory activity was not a consequence of a major shift in the number of mitochondria, but possibly a difference in their metabolic activity. One interesting observation that would require to be extended to other proteins from the respiratory chain complexes was the fact that the protein level of four of these proteins was increased mostly in females. Although brain mitochondrial metabolism is known to differ between adult males and females both in human and rodents [39
], there is, to our knowledge, no data in the literature regarding sex effect on mitochondria dynamics and metabolism in the developing fetal brain. Only the testosterone surge occurring around birth in male mice was shown to impact the synthesis of the mitochondrial-specific phospholipid cardiolipin [41
Mitochondria dynamics is closely associated with cell fate and differentiation process during brain development. Mitochondria structure and metabolism change throughout the differentiation process [42
]. At the metabolic level, while energy production relies mostly on glycolysis in undifferentiated cells, it progressively switches to oxidative phosphorylation throughout neural differentiation in order to meet the higher energy requirements of the differentiated neurons [42
]. It has also been illustrated that mitochondria dynamics and metabolic shift precede and functionally regulate neuronal differentiation [43
Several evidences have already established a link between PR during early life and alterations in the mitochondrial metabolism at a later stage of life. Maternal PR was shown to be associated with (1) impaired mitochondrial metabolism in the brain of adult rat offspring [44
] and (2) alteration in the expression level of several proteins from the mitochondrial respiratory chain complexes in the hypothalamus of pre-weaned rat [45
]. In addition, in human, mitochondrial metabolism is altered in the placenta of neonates suffering from Intra Uterine Growth Restriction that is often the result of a reduced provision of nutrients to the fetus [46
]. Oxidative stress that may results from impaired mitochondrial function is indeed evoked as a major programming mechanism in the increased risk of chronic degenerative diseases induced by neonatal protein restriction [47
]. However, although the consequences of neonatal PR on mitochondrial metabolism and oxidative stress on several tissues after birth are widely acknowledged, the link between PR, mitochondrial metabolism in fetal brain and an impaired neurodevelopment is, to our knowledge, not documented. Are mitochondria of neural stem/progenitor cells able to function as nutritional sensors and integrate very early on signals from their environment? In a mice model of maternal protein restriction, Eckert et al. [48
] demonstrated that, as early as E3.5, the blastocyst was able to sense maternal metabolic alterations, including deficiency in essential amino acids, within uterine fluid and they showed evidence of the implication of the mammalian Target of Rapamycin Complex 1 (mTORC1) signaling pathway in this process. Mitochondrial activity reflects the energetic status of the cells and mitochondria architecture was recently suggested to play an important role in bioenergetics adaptation to metabolic demands [49
]. Therefore, mitochondria dynamics constitute a way for the cell to adapt to nutrient shortage or excess. For instance, nutrient shortage was shown to result in the fusion of mitochondria associated with increased oxidative phosphorylation, which is for the cell the most efficient way to produce ATP [50
Our data are not sufficient to conclude that mitochondrial respiratory chain activity was definitely increased in the fetal hypothalamus of the PR group and additional experiments are certainly required to confirm this hypothesis. For instance, it could be interesting to measure mitochondrial mass and ROS production. However, the combination of our results regarding this point, together with what is already known about the link between mitochondrial activity and neuronal differentiation strongly supports the impact of maternal PR on these processes.
How the mitochondrial adaptive mechanism, linked to nutrient shortage, interacts with the program of differentiation of neural cells remains to be clarified.
The ubiquitin gene and several genes encoding the subunits of the proteasome complex were also upregulated in the PR fetal hypothalamus. The UPS (Ubiquitin Proteasome System) is closely associated with the mitochondrial metabolism. Mitochondria need the UPS for the removal of proteins that are damaged by ROS (Reactive Oxygen Species) and the proteasome function requires ATP [51
]. These two functions are especially important in neuronal function and differentiation [52
] as well as adult neurogenesis [55
]. Although the precise mechanisms of the role of the amino acid sensing pathway mTORC1 in the activation of the proteasome activity in situation of nutrient shortage are still under debate [56
], these two major pathways are obviously interconnected [57
], suggesting that they may interact in the response of neural cells to amino acid shortage. Cellular response to nutrient shortage may also involve autophagy, another major stress-response system that is closely linked to the mTORC1 detection system and which was shown to be important for neuronal development and axon growth [58
]. We did not find evidence of alterations in the expression of genes involved in the autophagy in our model but we did notice enrichment in upregulated genes from the lysosome pathway (Table S2
Epigenetic control of chromatin conformation [59
], DNA-binding proteins [61
] and transcription factors [62
] are among the best known molecular actors of the highly complex process of neuronal differentiation. Their action is itself modulated by factors related to the metabolic status of the cell via the remodeling of chromatin [65
]. Recently, post-transcriptional control was highlighted as a new layer of regulation in the determinism of cell fate and differentiation, particularly in the brain [66
]. Post transcriptional regulation include (1) chemical modifications of mRNA such as epitranscriptomic marks and (2) RNA-binding proteins that are involved in mRNA stability, turn-over, trafficking, degradation and translation.
We have shown, in our large-scale transcriptomic study, and confirmed (by the quantification of m6A) that the expression of several genes from the m6A epitranscriptomic machinery was significantly disturbed in the PR fetuses and that this was linked to a decrease in the global level of m6A. The m6A epitranscriptomic mark is the major mRNA modification identified so far and is associated with the control of various aspects of mRNA functions including stability, degradation, trafficking, splicing and translation [67
]. The brain is the tissue where this mark is the highest and it is especially present in mRNAs implicated in transcriptional regulation, cell adhesion, axon guidance and synaptogenesis [68
]. In addition, Yoon et al. [69
] recently demonstrated that depleting the writing of the m6A mark by inactivation of the Mettl14
gene in the embryonic brain of mice prolongs neurogenesis postnatally. This was associated with a decrease in the turnover of mRNAs encoding proteins involved in cell cycle, neurogenesis, and neuronal differentiation. The action of the m6A mark is mediated through interaction with different RNA-binding proteins that specifically recognize methylated or unmethylated mRNAs and will subsequently promote transcript fate [67
]. Interestingly, the m6A mark was demonstrated to be involved in the action of the Fragile X Mental Retardation Protein (FMRP) which is an RNA-binding protein with a major role in synapse function [68
] and that was shown to be over-expressed in the cortex of mice that suffered from fetal protein restriction [18
]. Although the expression of the Fmr1
gene was not altered in our model, the Fxr1
gene was downregulated in the PR group. Since this gene also encodes an RNA-binding protein that was recently shown to control the translation of the mitochondrial Cox2
] and since Cox 2
was one of the most expressed and significantly upregulated genes in the PR group, a possible link could exist between the m6A mark, RNA-binding proteins and mitochondrial function. On the other hand, FTO protein, which acts as a m6A demethylase, was shown to have a role in the cellular sensing of amino acids via the mTORC1 pathway and this activity was associated with its demethylase function [71
]. In addition, the mTORC1 pathway was demonstrated to mediate the link between nutrient shortage and the control of protein synthesis at the post transcriptional level [66
]. Interestingly, the mRNAs that are regulated by this system are enriched for the consensus motif of the m6A epitranscriptomic mark [66
]. All these elements converge to propose the m6A mark as a major actor in the impact of amino acid deficiency on hypothalamus development.
The role of post-transcriptional regulation in the consequences of PR on the timing of neurogenesis was also suggested by the over-expression at both the transcription and translation levels of CSDE1 in the PR group. The CSDE1 protein was recently shown to be a master regulator of neuronal differentiation, by regulating, at the translational level, the expression of a large number of genes [35
]. The expression of CSDE1 decreases throughout differentiation and modulates the transcriptional landscape by controlling the expression of key regulators of cell fate and neuronal differentiation. Therefore, the overexpression of both the gene and the protein that we observed in the PR group may reflect a delay in the neuronal differentiation process that may be consistent with a higher number of neuronal progenitors.
The DGE-seq approach was rather helpful for the detection of the pathways impacted by maternal diet. Of course, transcriptomic data do not always reflect the amount of proteins, but the DGE-seq approach is rather straightforward and more sensitive that a proteomics approach for the detection of mild effects. The proof is that we have indeed been able to identify key players in post-transcriptional regulation who are certainly involved in the impact of fetal nutrition in the precise control of neuronal differentiation. We made the choice to focus here on metabolic pathways and gene families that were, in our opinion, relevant regarding the physiological and cellular alterations observed in our model, but the transcriptomic approach also identified many other genes that certainly may require further investigation. The magnitude of gene expression change between PR and control fetuses was rather modest. However, major cellular pathways related to energy metabolism and neuronal differentiation have been impacted, so we believe that even mild disturbances can have repercussions on a process as precise and finely regulated as neuronal differentiation.