Worldwide, about 1.3 billion of individuals suffer of excessive adiposity, 940 million are overweight and 400 million are obese 1. This epidemic is driven by a widespread energy imbalance, with energy intake exceeding energy expenditure. Nevertheless, individual differences in the susceptibility to gain an excess of fat mass, in obesogenic environments exist. In addition to genetic factors there is a growing body of evidence, from epidemiological and animal studies, that perinatal nutrition may substantially predispose to the development of obesity. Fetal over and under-nutrition promotes adiposity later in life in different animal models 23 and infants small for gestational age appear to have a higher risk of developing adult obesity 4. Our understanding of the factors and mechanisms, by which perinatal nutrition impacts later susceptibility to obesity, remains limited. Nevertheless, the early nutritional environment may set adipose tissue growth and function towards either fat storage or oxidation, later in life 5. There is some evidence that the composition of dietary fatty acids and especially the essential polyunsaturated fatty acids (PUFA), linoleic acid (LA; C18:2n-6) and α-linolenic acid (ALA; C18:3n-3) and their respective long-chain products, arachidonic acid (AA; C20:4n-6), eicosapentaenoic acid (EPA; C20:5n-3) and docosahexaenoic acid (DHA; C22:6n-3) influence early adipose tissue development. These effects may be mediated by eicosanoids, and more especially by prostaglandins (PG; PGD₂, PGE₂, PGF_2α, PGI₂), involved in adipose tissue growth and metabolism and dependent on the long-chain PUFA availability 6. Animal studies showed that the intake of a high LA/ALA ratio during fetal and postnatal life promotes early adipose tissue growth and obesity 78. Nevertheless, since enhancing the LA/ALA ratio in the diet implies either a substantial increase of LA 8, a decrease of ALA 7 or both 8, it is difficult to determine if the n-6 PUFA promote adipose tissue development or if the n-3 PUFA have anti-adipogenic properties. Because, the adipogenic effects of the n-6 PUFA remain controversial 9, we hypothesized that a low level of ALA rather than an excess of LA leads to an increased adiposity. In order to test part of this hypothesis, 3 groups of newborn guinea pigs (GP) were fed until 21 days of age with milk and pellets containing either a high (10% total FA), medium (2.4% total FA) or low but adequate (0.8% total FA) ALA content corresponding respectively to a LA/ALA ratio of 2:1, 10:1 and 30:1. The short and long term consequences of such early postnatal dietary interventions on adipose tissue development and metabolic markers were determined.

Food intake was not different between groups. The mean cumulative food intake was 118.1 ± 5.8 g (10%-ALA), 124.6 ± 7.0 g (2.4%-ALA) and 125.1 ± 10.7 g (0.8%-ALA group) from d9 to d18 and was 3013.3 ± 228.7 g (10%-ALA), 3138.6 ± 166.2 g (2.4%-ALA) and 3231.4 ± 119.9 g (0.8%-ALA group) from d23 to d129.

Fatty acid profile of plasma lipids was determined at the end of the dietary intervention period (d21). Inclusion of 10% ALA in the diet resulted in its increase in plasma TG and PL (TG: 4.40 ± 0.35%; PL: 0.80 ± 0.26%, figures 1A and 1B), but the increase was not significant with the addition of 2.4% ALA (TG: 2.25 ± 0.23%; PL: 0.20 ± 0.01% and TG: 1.25 ± 0.37%; PL: 0.15 ± 0.07% for 2.4% ALA and 0.8% ALA, respectively). The proportion of total C20-22 n-3 PUFA, including eicosapentaenoic acid (EPA), docosapentaenoic acid (DPA) and docosahexaenoic acid (DHA), in plasma TG (figure 1A) and PL (figure 1B) tended to decrease with decreasing ALA intake, but this effect did not reach statistical significance (p = 0.09 and p = 0.07 for PL and TG, respectively), except for DPA in TG and for EPA in PL. The 0.8%-ALA group (0.01 ± 0.01 mol %) had lower DPA in TG than both other groups (2.4%-ALA: 0.20 ± 0.12, p < 0.05 and 10%-ALA: 0.20 ± 0.01 mol %, p = 0.06). For EPA in PL, there was a significant difference between 10%-ALA group (0.10 ± 0.01 mol %) and 0.8%-ALA group (0.01 ± 0.01 mol %, p = 0.018). As expected, there was no difference in plasma LA between groups and increasing ALA intake did not modify the proportion of n-6 PUFA (figures 1C and 1D).

The body weight (BW; 10%-ALA: 869.3 ± 54.2 g; 2.4%-ALA: 860.7 ± 29.6 g and 0.8%-ALA group: 902.3 ± 25.7 g at d136) and BW gain (data not shown) did not differ between groups during the course of the experiment.

Decreased intake of ALA during early life resulted in increased body fat at 136 days; Figure 2 shows the fat mass evolution during the course of the study expressed as g (figure 2A) or % body weight (figure 2B). While no differences were observed at the end of the dietary intervention period (d21), older guinea pigs in the 0.8%-ALA group had or tended to have more fat mass than those in the 10%-ALA group at d79 (figure 2A: +20.3 g, p = 0.08; figure 2B: +2.6%, p < 0.01), at d107 (+29.9 g, p < 0.05; +2.8%, p < 0.01) and at d128 (+22.6 g, p = 0.05; +1.7%, p = 0.08) and than those in the 2.4%-ALA group at d128 (+24.2 g, p < 0.05; +2.2%, p < 0.05).

Likewise, at d 136, epididymal AT was or tended to be heavier in the 0.8%-ALA than in the 10%-ALA (+0.19 g/100 g BW, p < 0.05) and the 2.4%-ALA group (+0.16 g/100 g BW, p = 0.05). The interscapular AT weight also was higher in the 0.8% than in the 2.4%-ALA groups (+0.18 g/100 g BW, p < 0.01). The weights of epididymal, interscapular and retroperitoneal AT relative to BW were not different between groups at d21 (Table S3, Additional file 3). The liver weight as well as the pancreas weight did not differ between groups at d21 and d136 (data not shown).

The fractional proliferation rates (FSR) of cells, measured in the retroperitoneal and epididymal AT, were unchanged between 10%-ALA and 0.8%-ALA groups at both d21 and d136 (Table 4, Additional file 4). Nonetheless, the cell proliferation rate in the subcutaneous AT was higher in the 0.8%-ALA group compared to the 10%-ALA group at d21 (2.1-fold higher; not significant) and d136 (1.8-fold higher; p = 0.021) (Table 4, Additional file 4).

The de novo lipogenesis (DNL) and synthesis rate of TG in the different adipose fat pads did not differ between the 10%- and 0.8%-ALA groups at d21 (Table 5, Additional file 5). A higher hepatic DNL (P < 0.001) was observed at d21 in guinea pigs fed 0.8%-ALA intake compared to 10%-ALA intake (Table 5, Additional file 5). No difference was observed in DNL or synthesis rate of TG in hepatic or adipose tissues between groups at d136 (Table 5, Additional file 5).

The arachidonic acid (AA) and most eicosanoid concentrations determined in retroperitoneal and subcutaneous fat pads were not different between the groups (Table 6, Additional file 6). An exception was observed for PGE2 and PGJ2 that were both significantly higher after medium vs low ALA intake at d136 (Table 6, Additional file 6). In order to evaluate the AT differences, the groups were combined and the eicosanoid profile was compared between the fat depots at both ages. At d21, AA, 6-keto-PGF_1α and PGJ₂ concentrations were 1.6-fold (244.7 ± 24.9 vs 149.6 ± 14.0 ng/50 mg AT, p < 0.01), 3.7-fold (1.95 ± 0.10 vs 0.52 ± 0.09 ng/50 mg AT, p < 0.0001) and significantly (0.015 ± 0.001 vs 0.001 ± 0.001 ng/50 mg AT, p < 0.0001) higher in the retroperitoneal AT when compared to the subcutaneous one, respectively. At d136, the difference of PGJ₂ concentrations was even higher between the retroperitoneal (1.53 ± 0.35 ng/50 mg AT) and subcutaneous (0.01 ± 0.01 ng/50 mg AT, p < 0.0001) AT. In addition, 6-keto-PGF_1α, PGE₂ and PGF_2α were 1.5-fold (p < 0.001), 23-fold (p < 0.0001) and 1.6-fold (p < 0.0001) higher in the retroperitoneal vs subcutaneous AT (data not shown).

The plasma FFA were not different between groups at d21 (data not shown) or at d136 (10%-ALA, 17.6 ± 0.7; 2.4%-ALA, 18.0 ± 0.6 and 0.8%-ALA group, 18.0 ± 1.1 mg.dL^-1). The plasma TG concentrations were unchanged at d21 (data not shown) but were 1.3-fold lower in the 0.8%-ALA vs 10%-ALA (1.31 ± 0.05 vs 1.78 ± 0.21 mmol.L^-1, p < 0.05) and vs 2.4%-ALA group (1.61 ± 0.15 mmol.L^-1, not significant), at d136. There were no differences in plasma insulin concentrations between groups at the end of the suckling/weaning period (figure 3A). At d136 (figure 3B) the 0.8%-ALA group (6.8 ± 1.5 ng.mL^-1) showed about 1.5-fold higher concentrations of plasma insulin when compared with other groups (2.4%-ALA, 4.3 ± 1.3 and 10%-ALA group, 5.1 ± 1.7 ng.mL^-1, not significant). Similar results were observed regarding the pancreatic insulin content. The 0.8%-ALA group showed about 1.6-fold increase in the pancreatic insulin content (3.5 ± 0.8 ng.μg^-1 of protein) when compared to the 10%-ALA group (2.2 ± 0.3 ng.μg^-1 of protein, p = 0.06, not significant) at d136 (figure 3B). There were no significant differences in plasma glucose between groups at d21 (data not shown) and neither at d136 (10%-ALA, 112.0 ± 3.3; 2.4%-ALA, 110.0 ± 6.7 and 0.8%-ALA group, 104.5 ± 4.6 mg.dL^-1).

To our knowledge this is the first study showing that dietary ALA intake during the suckling/weaning period influences fat mass development later in life; the lowest dietary ALA intake programming the highest adiposity in the adult guinea pig.

In our study, the adipose tissue development was not impacted during the period of dietary intervention with different ALA levels. Indeed, the fat mass and adipose tissue weight at d21 were not different between the high, moderate and low ALA fed guinea pigs. Our results are in agreement with Korotkova et al. who showed that offspring of rat dams, fed a 6.2% ALA (total FA) enriched diet during the gestation and suckling period, did not differ in AT weight at 3 weeks of age compared to a low ALA diet with the same amount of dietary lipids 8. On the contrary, the dietary intake of chia seed (rich in ALA) decreased adiposity in adult rats fed an obesegenic diet 26. Interestingly this effect was associated with a large increase in plasma lipid n-3 LC-PUFAs (EPA and DHA), known to have anti-adipogenic effects 27, probably through inhibition of the proliferation of adipose tissue cells 27. While no one can exclude that ALA has its own specific physiological effects on AT development, it is tempting to speculate that these effects are more dependent on its long-chain products. Unlike other rodents and similarly to humans, the efficiency of ALA conversion into its long-chain products is limited in guinea pigs 28. As previously shown in AT and other tissues of the guinea pigs 29, we observed in our study a higher ALA proportion in plasma TG and phospholipids with an increasing ALA intake but with a limited increase of >C20 n-3 PUFA. Indeed, EPA, DPA or DHA represents less than 0.8% of total FA in our study, even in the group with the highest ALA intake, and could explain the absence of impact on the adipose tissue development of the 21-day old guinea pigs. Previous studies have indicated that n-3 LC-PUFAs enriched diets diminish the hepatic enzymatic activities of endogenous fatty acids synthesis 30 associated with a decrease in plasma TG 3132. In the present study, the high ALA intake induced a 2-fold reduction of fractional DNL in the liver at d21 with no impact on total plasma TG concentration. A higher >20 n-3 PUFA plasma concentration during the high ALA intake could modulate transcription factors in the liver. It is possible that that >20 n-3 PUFAs reduce DNL by down-regulating SREBP-1 (sterol regulatory element-binding protein-1) and activating PPAR alpha (peroxisome proliferator-activated receptor alpha) as shown by previous studies 33. To explain no difference on total plasma TG, we considered that fatty acids for TG synthesis come from the DNL, the dietary source or from the adipose tissue lipolysis. Based on the calculation of the FA contribution from DNL to TG synthesis (ratio of DNL to TG FSR), we estimated that the contributions from the dietary source plus the adipose tissues were 72% and 87% for the low and high ALA groups, respectively 10. Since dietary FA contributed equally between groups, plasma TG containing FA from the adipose tissue may have been greater in the high ALA group compared with low ALA group at d21. It is possible that more plasma TG containing FA from the adipose tissue was utilized by the liver or muscle (e.g. for oxidation). Further analysis of VLDL-TG (very large density lipoprotein-TG) secretion and removal rates in and from the circulation could clarify the unchanged plasma TG. Ultimately the fat mass was unchanged between the groups at the end of the suckling/weaning diet treatment.

The long-term adipogenic effect of a low ALA intake during early postnatal life seems to be more pronounced on subcutaneous than visceral AT. Indeed, the 2 to 3% differences of total fat mass found between the low and high ALA group at d107 and d128 may not be explained by the observed difference of epididymal fat pad weights, knowing that the visceral fat depots (calculated by summing the weight of epididymal, retroperitoneal and mesenteric AT) represent only about 18% of total fat mass in the adult GP. The subcutaneous adipose tissue which represents the major fat depot in adult (about 80%) revealed a significant increase in cell proliferation in adults after the 0.8% ALA versus 10% ALA postnatal diet treatment. No hyperplasia was noticed in other fat depots after 0.8% ALA intake. This supports a specific role of the subcutaneous adipose tissue. Previous studies indicated a direct effect of n-3 fatty acid enriched diet on decreasing adipocyte differentiation and increasing apoptosis in vitro 34 and in high-fat fed rats 35. In the present study, we could not measure the deuterium enrichment decay in DNA to determine the apoptosis rate that could have been lowered in subcutaneous AT of the adult GP of low ALA group. On the contrary, we observed a clear imprinted effect of low ALA early intake on increasing the cell proliferation rate by 75% versus high ALA early intake in the subcutaneous AT of adult GP. In addition, we showed a transient increase of hepatic DNL in the low ALA group at d21 that could be involved in the imprinting mechanism. Further investigations are needed to clarify the mechanism.

Interestingly, the 0.8% ALA group had higher insulin at the end of the experiment, though the differences between groups did not reach statistical significance due to a high variability. This tendency for higher insulinemia is probably linked to an increased adiposity, as known for other animal species and human. We found a positive correlation between insulin (expressed as log) and fat mass in the adult guinea pigs (Pearson r = 0.77, p < 0.0001). Although all in a normal range 36, plasma TG were significantly higher in the high versus medium and low ALA groups at d136. This was not associated with an increased adipose tissue FSR of total TG in the high ALA group. Previous studies showed the capabilities of n-3 LC-PUFAs to modulate lipolysis and consequently alter TG level 2737. Although we described a possible imprinted effect in the present study, the increased plasma TG could be due to an increased lipolysis to support a lower fat mass in the adult GP of high ALA group.

The mechanism by which a low ALA intake programs an excess of adipose tissue is unclear. It may imply cell proliferation as indicated in the present study by the increase of fractional proliferation rate of cells in the subcutaneous AT. Because eicosanoids are modulated by LC-PUFAs and are involved in the differentiation and proliferation of adipose tissue cells we expected to see differences between groups 38. Beside the large differences between the types of fat depots (visceral vs subcutaneous) on eicosanoid levels, we found only minor changes between groups and none in subcutaneous AT. Further studies are definitely needed to understand the mechanism by which early dietary ALA impacts adipogenesis and lipogenesis later in life. The Guinea Pig represents a good model to the human term newborn 39 but shows some experimental limitations due to the lack of commercially available immunological kits and epigenetic markers. This may change with the recent first release of the genome sequencing and assembly of the Guinea Pig 40. Future investigations using the Guinea Pig model will help understanding the mechanisms by which early nutrition may program adiposity later in life. It would be also interesting to evaluate whether the programming effect of a low ALA intake may have been more pronounced with dietary interventions including gestation and/or with more obesogenic post-weaning diets (e.g., high fat/high sucrose diet).

Our main finding shows that a suboptimal ALA intake during suckling/weaning promotes the susceptibility to an increased adiposity later in life. It is important to note that the ALA diet (0.8% total FA; 0.16 g ALA/100 g diet), used for the low intake group, does not correspond to a deficient diet (0.01 g/100 g diet) as previously described 41. Interestingly, smoking mothers have lower ALA content in breast milk than non-smoking ones 42 and maternal smoking is associated with an increased risk of late onset overweight 43. Consequently, if relevant to the human newborn, our findings suggest the need of recommending n-3 PUFA supplementation in the pregnant/lactating mother with a low ALA intake or at risk of suboptimal breast milk content.

EP, OA, ECG, JM, GP and KM authors work for Nestle Company.

EP wrote the manuscript and designed the use of the deuterium approach. OA designed and conducted the study. CG and CC developed the specific insulin analysis for GP. DR and CPA contributed with the eicosanoid analyses. ECG interpreted the data and drafted the manuscript. JM and GP performed the statistics. CB and ST developed the deuterium approach and did the analyses. KM directed the study and wrote the manuscript. All authors read and approved the final manuscript.