The concept of early programming is based on the high plasticity of organisms during their development, allowing them to adapt their phenotype to environmental conditions. In poultry, it has been shown that embryonic thermal programming improves the survival of animals exposed to subsequent heat stress , and it is particularly interesting to note that the best embryonic period to apply the stimulus corresponds to the maturation period of the hypothalamo-hypophysis-thyroid axis, which is involved in thermal regulation . Remarquably, the adapted phenotype may also respond differently to new environmental challenges, such as embryonic thermal manipulation resulting in increased foie gras production in mule ducks at the age of 3 months . Although the mechanisms are not yet fully understood, the timing of the application of the environmental stimulus for programming seems to be very important. In this context, it seems interesting in the field of foie gras production, to study the ontogeny of the metabolic pathways involved in liver fattening, in order to reveal potentially interesting windows of application of the thermal stimulus.
As a first step, the description of gene expression profiles in embryonic duck liver is in itself particularly informative to understand the establishment of hepatic metabolism pathways.
However, since the size of the livers did not allow sampling before E12, it is impossible to conclude on the specifically hepatic expression of developmental genes before this stage. Data on early chicken embryogenesis suggest that hepatic induction of the anterior endoderm via an interaction with the “cardiac” mesoderm  involves many of the pathways depicted in Fig. 1 from the very beginning of ontogeny [15, 16]. Nevertheless, although much of the cell proliferation and hepatic differentiation arise at the earliest stages of liver development , our results suggest that these signaling pathways (supplemental Table 5) are still strongly involved in ducks between E12 and E20, in morphogenetically distinct livers. Consequently, an environmental stimulus occurring during this period could potentially influence the proliferation and differentiation of hepatocytes, thereby causing a modification in the final number of cells in the mature organ, as previously shown for chicken muscle cells [18, 19]. Therefore, even though hyperplasia does not seem to be involved in fatty liver enlargement during overfeeding , it is conceivable that an increase in the number of hepatocytes at birth may enhance the fattening of the liver during force-feeding, since the ability of each cell to expand (hyperphagia) may not be affected. Moreover, recent studies [3, 21] suggest that the histological structure of the liver after overfeeding, particularly the number and size of cells, may play a role in the final quality of the product, mainly indicated by fat loss after cooking. It would therefore be very interesting to determine the precise impact of the embryonic thermal stimulus on the number of hepatic cells at birth and after overfeeding in order to accurately modulate the final yield of fatty liver through a specific programming protocol.
In oviparous animals, the nutrition of the developing embryo depends entirely on the resources from yolk and albumen. Despite the low amount of carbohydrate in the egg [22, 23], glycolysis has been described as an extremely important source of energy during the first third of chicken embryogenesis  and hatching . The present results highlight that expression of carbohydrate-related genes is strongly committed up to E20 in mule duck embryos (Fig. 2), in particular those related to glucose transport (GLUT1 and 2) and glycolysis (GAPDH and HK1), confirming the major role of the liver in systemic glucose homeostasis throughout embryogenesis [26, 27]. Lastly, the drop in carbohydrate-related gene expression observed at D1 might reflect the decline of endogenous resources after hatching, a process involving high energy demand. Since carbohydrate metabolism is a major pathway involved in fattening the liver during overfeeding, the high expression of carbohydrate-related genes around E20 may represent an interesting period for embryonic programming by environmental stimulus. With the exception of ChREBP, the present results suggest that the programming period that may have an impact on carbohydrate metabolism could be centered around E20. Nevertheless, it is still possible that a stimulus applied up to E27 had an impact on the resulting activity of ChREBP. As a major transcription factor playing a key role in carbohydrate and lipid metabolism [28, 29], it cannot be excluded that a programming protocol applied during its peak of expression may make an important contribution to the physiological response after overfeeding. Only programming experiments with different stimulus protocols and an in-depth analysis of the impact on ChREBP mRNA and protein expressions, or activity could provide a definitive answer about its specific role and that of other carbohydrate-related genes.
With regard to the lipid metabolism, the significant overall change occurring on the 4th day after birth suggests that unlike the genes involved in carbohydrate metabolism, the expression of lipid-related genes could be strongly affected by first meals. Indeed, ducklings sampled on D1 were slaughtered before the first meal, while the ducklings sampled on D4 were all fed ad libitum since day 2. The use of yolk lipids during the development of avian embryos has been well described in a previous review . These lipids are the main source of energy during the last week of embryogenesis, when the embryos exhibit an exponential growth [24, 31]. Therefore, the starting diet, mainly composed of wheat and corn, can be interpreted as a nutritional transition since the ducklings move from an energy source consisting primarily of lipids from egg yolk to an exogenous diet with high carbohydrate content . This crucial transition phase is also accompanied by a major change in the metabolism of the liver that acquires the ability to synthetize its own lipids . The present results, like previous studies on chickens [34, 35], illustrate this modification of hepatic lipid metabolism by highlighting the sharp increase in the expression of lipogenic genes such as SCD1 (Fig. 3.2.b) and FASN (Fig. 3.2.a) at D4 in mule ducklings. These genes are involved in the de novo lipogenesis pathway [36, 37] which reflects the ability to store carbohydrate sources as lipids . In a context of nutritional change with a sudden high intake of carbohydrates, it is consistent to stimulate their storage by increasing the expression of genes involved in lipid synthesis, the liver being the predominant site of lipogenesis in birds [39, 40].
However, we observe that the pathway of lipid catabolism is also still engaged at D4, with high expression of ACOX1, ACAD11, CPT1A, ACAA2, suggesting that energy metabolism depends on the use of both carbohydrates and lipid at this stage in mule ducks. Therefore, environmental programming during this critical period could be particularly interesting to study in the context of the response to overfeeding and the production of foie gras. Finally, several genes mainly involved in lipid catabolism (PPARA, CPT1A, ACAA2, ACAT1) also showed high expression at the beginning of the kinetics, between E12 to E20. Indeed, beta-oxidation of fatty acids provides a large part of the energy demand during embryogenesis . Consequently, the application of an environmental stimulus during this period could potentially program a different response to force-feeding and thus improve the phenotype.
However, the negative correlation measured between the expression of several carbohydrate and lipid-related genes during embryogenesis suggest that these two pathways, which seem to work in mirror mode during development [24, 30], could be affected differently by early-life programming. Targeting both with a thermal stimulus around E20, where most carbohydrate-related genes and some of the genes related to lipid catabolism are strongly expressed, seems to be the most appropriate choice. Nevertheless, these results also open a new programming window, around the first meals and specific to lipid-related genes, which could be interesting to explore in the context of the production of foie gras.
The overall increase in stress-related gene expressions occurred after the transfer of ducklings from the hatchery to the breeding facility, resulting in a significant temperature change from 37.3 °C to 26–28 °C. It is interesting to note that a change in the ambient temperature induced a significant increase in the hepatic expression of heat-sensitive genes involved in protein folding [41,42,43] (supplemental Table 8). If the thermal stimulus applied during embryogenesis induced a direct modification of their expression, it might be of interest to use them as positive markers of stimulation. Since the products of these genes are involved in the folding of different types of proteins, a change in their expression profiles could have an impact on several enzymatic activities, even those involved in metabolic processes. To answer this question, an upcoming study will focus on the immediate impact of the thermal change during embryogenesis on the expression level of these genes.
The hatching process represents a major challenge in terms of nutritional regulation, control of body temperature, but also of transition from chorioallantoic to pulmonary respiration [30, 35]. This abrupt metabolic change with the sudden onset of elevated oxygen levels may result in an increase in oxidative stress that must be controlled to maintain overall cellular homeostasis. The enhanced expression of several genes involved in cellular detoxification such as GSTK1 , GSTT1  or CYP2E1  in the liver of newborn ducklings may be a reflection of this control system. Finally, these expression patterns confirm that hatching is certainly the most brutal challenge a bird faces throughout its life and suggest that embryonic thermal stimulus could be specifically traced by some of these stress-related biological markers.