Introduction

Insulin resistance (IR) has been suggested to have a fetal programming [4]. The fat mass and obesity (FTO)-associated gene is located on 16q12.2. The rs9939609 FTO (T ≥ A) has been associated with body mass index (BMI) increasing weight of A-homozygous adults about 3 kg. FTO association with increased risk of childhood obesity was also demonstrated by Frayling et al. [11], although the effects were observed at age seven. The FTO expression is especially high in the hypothalamus, suggesting its involvement in the energetic balance [42]. The risk alleles have been associated with poorer eating behaviors [9, 11, 39]; higher food, energy, and fat intakes; and specific preference for energy-dense foods with high-fat content [20, 36]. In vitro experiments suggest that FTO exert gene regulation at RNA levels in humans [13, 23, 24]. Its role in energy homeostasis is supported by some studies showing that FTO mRNA expression is regulated by food intake [12, 13, 41, 42], glycemia [35], weight status [43, 44], and energy expenditure [3, 10, 41].

Harbron et al. [19] investigated the association of FTO polymorphisms (rs1421085 and rs15817449) and food intake, eating behavior, and changes in BMI in adults with BMI ≥27 kg/m2. The risk alleles were associated with higher hunger and emotional disinhibition scores and with higher intakes of high-fat foods and refined starch. Ortega-Azorín et al. [34] reported that the association of the FTO rs9939609 polymorphism with type 2 diabetes (T2DM) is highly modulated by diet in subjects presenting low adherence to the Mediterranean diet (MDA) pattern.

High-quality diets during pregnancy have been recommended to assure a correct embryo and fetal growth and functionality [38]. Thus, diets with high score for healthy eating index (HEI) or MDA followed during pregnancy decrease the risk of both high insulin and homeostatic model assessment insulin resistance (HOMA-IR) at birth [15]. In addition, neonates from mothers that followed high-score MDA diets during pregnancy showed less cholesterol transported by LDL (LDLc), apolipoprotein (Apo) B, and homocysteine (tHcys) but higher ApoA1/ApoB ratio than neonates whose mothers followed low MDA score diets [17].

Higher IR and preference for hypercaloric diets have been described in the A carriers [36]. Information about the effects of FTO polymorphism on glucose and insulin levels in term neonates is scarce [29]. To the best of our knowledge, this is the first study associating neonatal and maternal FTO polymorphisms and lipid profile and tHcys at birth. Moreover, the effect of this polymorphism on glucose and insulin levels at birth in appropriate-for-gestational-age (AGA) term neonates has never been tested.

Present paper hypothesizes that maternal and neonatal FTO polymorphisms affect neonatal parameters related to metabolic syndrome (MS). The aims of this paper were (a) to find out the effects of neonatal and maternal FTO rs9939609 polymorphism on different parameters related to IR, tHcys, lipid, and lipoprotein levels at birth; (b) the possible associations between the maternal and neonatal FTO polymorphism on those parameters; and (c) the possible modulating effect of the maternal diet quality during pregnancy on these parameters.

Methods and materials

Subjects

It is a cross-sectional study performed in the Mérida Hospital (Spain), in accordance with the Helsinki Declaration and approved by the Management and Ethical Committee of the Mérida Hospital. Permission for DNA genotyping was obtained from 65 participating mothers. All selected neonates were Caucasian, singleton, full-term (born between weeks 37 and <42), normoweight (between 2.5 and <4.0 kg), AGA (weight ≥ percentile 10 for its gestational age), and born without fetal distress, by eutocic delivery with cephalic presentation from mothers without gestational diabetes (GDM). Thus, 53 mother-neonate couples were finally selected according to the inclusion criteria. The participation rate was about 24 % of participating women in the Mérida cohort study [16]. Nutritional information during pregnancy was available in 30 mothers.

Protocol

Data related to the delivery, mothers’ age, and neonatal characteristics were obtained from hospital records. Anthropometrical measurements were performed by trained personnel. After delivery, in order to avoid any contamination with maternal DNA (blood and efflux), cord blood was obtained by arterial puncture and collected in BD Vacutainer® SST II tubes containing separation gel (Becton Dickinson, Plymouth, UK). Serum was obtained by blood centrifugation (3500 rpm, 5 min), and aliquots were stored at −18 °C until processing.

Between weeks 24 and 28 of pregnancy, future mothers were screened for GDM using the O’Sullivan glucose tolerance test [33]. Shortly, pregnant women received 50 g of glucose and after 1 h were tested for glycemia. They were considered to have impaired glucose tolerance when glycemia was over 140 mg/dL (≥7.78 mmol/L). When mothers were diagnosed of impaired glucose tolerance, a second test was done. They received 100 g of glucose and were tested after 1, 2, and 3 h for GDM.

Assays

Genomic DNA was isolated from whole blood. The rs9939609 FTO single-nucleotide polymorphism (SNP) was determined using a 7900HT sequence detection system (Applied Biosystems by Life Technologies, Foster City, CA, USA) and a fluorescent allelic discrimination TaqMan™ assay by standard procedures. For quality control purposes, 10 % of randomly selected samples were genotyped a second time, and there were no discrepancies. The calling rate was 95 %. Genotype frequencies did not deviate from Hardy-Weinberg equilibrium expectation (P > 0.1). Similarly to other Mediterranean populations, about two third of participating mothers and neonates were AA + AT [5].

Four genotype neonatal × maternal allele associations were tested: (a) neonates homozygous for the FTO rs9939609 T allele × mothers homozygous for the FTO rs9939609 T allele (nTT × mTT), (b) neonates homozygous for the FTO rs9939609 T allele × mothers carrying the FTO rs9939609 A allele (nTT × mAA + AT), (c) neonates carrying the FTO rs9939609 A allele × mothers homozygous for the FTO rs9939609 T allele (nAA + AT × mTT), and (d) neonates carrying the FTO rs9939609 A allele × mothers carrying the FTO rs9939609 A allele (nAA + AT × mAA + AT).

Serum glucose concentrations were measured by the glucose hexokinase method (Gluco-quant®), total cholesterol (TC) by the colorimetric enzymatic method (CHOD-PAP), cholesterol transported by high-density lipoproteins (HDLc) by the homogeneous enzymatic method (HDL plus second generation), triglycerides (TG) by the colorimetric enzymatic method (GPO-PAP), and ApoA1 and ApoB by the immunoturbidimetric method (Tina-quant®). All of them were supplied by Roche Diagnostics (Basel, Switzerland) and processed in a Roche/Hitachi Modular P (Roche Diagnostics, Basel, Switzerland) analyzer. LDLc was calculated using the formula of Friedewald et al. as validated in neonates by Glueck et al. [18]. Insulin concentrations were determined by electrochemiluminescence immunoassay (ECLIA) supplied by Roche Diagnostics in a Roche/Hitachi Modular Analytics E 170 analyzer (Roche Diagnostics, Basel, Switzerland). tHcys was measured by fluorescence polarization immunoassay (FPIA) supplied by Abbott in an IMX® System analyzer (Abbott Diagnostics, IL, USA). Arylesterase (AE) activity was measured according to Nus et al. method [32] at 37 °C. One unit of AE was defined as the millimole of phenol formed from phenyl acetate per minute monitorized using a thermostated T80+ spectrophotometer (PG Instruments® Ltd., Wibtoft, Leics, UK). Oxidized LDL (oxLDL) was determined by an ELISA test kit from Mercodia Laboratories (Upsala, Sweden) using an ELx808 BioTek® spectrophotometer (BioTek Instruments, Winoosky, VT, USA).

Our laboratory participates in the Spanish Clinical Chemistry Society (SEQC) External Quality Evaluation Program, which follows UNE-EN-ISO 9001:2000 standards and is certified by AENOR. All the assays were properly calibrated and performed under internal and external quality control provided by the manufacturers and SEQC, respectively. Intra-assay and inter-assay variation coefficients were 1 and 1.7 % for glucose, 1.5 and 4.9 % for insulin, 0.8 and 1.7 % for TC, 1.5 and 1.8 % for TG, 0.9 and 1.85 % for HDLc, 1 and 2.4 % for ApoA1, 1.5 and 2.5 % for ApoB, 2.3 and 2.8 % for tHcys, 8 and 8.9 % for AE, and 4.5 and 5.0 % for oxLDL, respectively.

The indices used to test IR or insulin sensitivity were the quantitative insulin sensitivity check index (QUICKI), HOMA-IR, homeostatic model assessment insulin sensitivity (HOMA-IS), and the glucose/insulin ratio [15].

P25 and P75 described in the Mérida study reference population for some components of the MS [14, 16] were used in this study as cutoff point to set the percentage of neonates from AA + AT and TT mothers with low and high values, respectively.

Dietary data collection

Details of dietary collection and managements have been already published [17]. Briefly, participants, guided by a trained dietician, completed a food frequency questionnaire that included 169 items classified according to food groups and based on questionnaires used and validated previously. Photographs of sample portions were used to estimate the serving size and volumes consumed. The dietician reviewed, together with the volunteer, the usual consumption frequency of each food (per day/week/month), together with the normal food helping size. Information about different food group consumption (oils and fats, vegetables, fruits, legumes, cereals, fish, meat and eggs, sweets and bakery, nuts, poultry, and dairy products), and dietary habits was also obtained during pregnancy.

HEI and MDA, used as global dietary quality indices, were evaluated in 30 mothers following criteria stated in a previous paper, where cutoff points for HEI and MDA were set at 70 and 7, respectively [15]. The test for evaluating MDA was an adaptation to 13-point scale of the one used in the PREDIMED study (acronym of “Efectos de la dieta mediterránea en la prevención primaria de la enfermedad cardiovascular”), after considering that alcohol should not be consumed during pregnancy as indicated in [15]. From the 30 mothers evaluated, 10 were TT while 20 were AA + AT. Nine of their neonates were TT while 21 were AA + AT. Attending to the neonatal × maternal FTO association, 5 were nTT × mTT, 4 nTT × mAA + AT, 5 nAA + AT × mTT, and 16 nAA + AT × mAA + AT.

Statistics

The Z Kolmogorov-Smirnov test was used for testing the normal data distribution. IR markers and TG were normalized by natural log transformation. To avoid any potential effects of maternal glycemia on neonatal parameters, glycemia at O’Sullivan test was used as covariate for data adjusting. Neonatal insulin and TC were considered major outcomes. Taken into account allele prevalence of rs9939609 FTO in Mediterranean populations [34], this study showed a power at least of 85 % (nominal alpha = 0.05) to detect mean differences between A allele carrier and TT neonate groups of 12 mg/dL for TC and 3 mIU/L for insulin. The interaction between maternal and neonatal FTO polymorphisms was tested in the following four groups: nTT × mTT, nTT × mAA + AT, nAA + AT × mTT, and nAA + AT × mAA + AT by two factorial (mother and neonatal allele effects) univariate ANOVA. Differences between the nTT × mTT and nTT × mAA + AT, nAA + AT × mTT and nAA + AT × mAA + AT, nTT × mTT and nAA + AT × mTT, and nTT × mAA + AT and nAA + AT × mAA + AT associations were tested by the unpaired t test. Statistical significance was set at P < 0.05. Statistical analyses were performed using the SPSS (version 22.0) and SAS (version 12.3) statistical software packages.

Results

Anthropometrical and biochemical data of selected neonates and their respective mothers did not significantly differ from AGA-neonate sample of the Mérida study (data not shown). Eighteen mothers were T homozygous and 35 were AA + AT, while 17 neonates were T homozygous and 36 were AA + AT.

Table 1 shows some anthropometrical, analytical, and dietary characteristics of the studied mothers. Weight gain during pregnancy was significantly higher in mAA + AT (P = 0.038). mTT showed significantly lower MDA score diet (P = 0.006) but non-significantly higher HEI scores. Mothers of nAA + AT and nTT showed equivalent age, pregnancy weight gain, and diet quality scores.

Table 1 Maternal age, pregnancy weight gain, serum glucose and cholesterol, health eating index (HEI), and Mediterranean diet adherence (MDA) score of the studied population depending on the maternal and neonatal FTO rs9939609 polymorphism
Full size table

The anthropometrical and clinical characteristics of the group of 30 mTT and mAA + AT (and their respective children) whose nutrition data were available did not significantly differ from those of the whole group of 53 mother-neonate couples studied in the present paper.

Table 2 shows the effects of maternal and neonatal FTO polymorphisms on the neonatal anthropometry and IR markers. Offspring of mAA + AT presented lower glucose (P = 0.015), insulin (P = 0.049), and HOMA-IR (P = 0.015) but higher HOMA-IS (P = 0.015) at birth than that of mTT. However, nAA + AT had higher insulin (P = 0.048) and HOMA-IR (P = 0.035) than their T homozygous counterparts.

Table 2 Neonatal anthropometrical parameters, glucose, and insulin resistance/sensitivity markers of the studied population at birth depending on the maternal and neonatal FTO rs9939609 polymorphism
Full size table

Table 3 shows the effect of maternal and neonatal FTO polymorphisms on neonatal tHcys, lipids, and lipoproteins. Only tHcys (P = 0.029) was affected by the maternal FTO polymorphism. Neonates born from mAA + AT showed higher tHcys levels than those from mTT. Considering the neonatal FTO polymorphism, nAA + AT showed marginally significant lower (P = 0.058) TG levels than their TT counterparts.

Table 3 Lipids, lipoproteins, apolipoproteins, homocysteine, and arylesterase at birth depending on the maternal and neonatal FTO rs9939609 polymorphism
Full size table

Figure 1 shows the percentage of neonates from mAA + AT and mTT presenting values that lay below P25 or above P75 described in the Mérida study reference population for some components of the MS [14, 16]. Lower levels of MS markers were more prevalent in neonates from mAA + AT.

Fig. 1
figure 1

Percentage of neonates whose mothers were genotyped for FTO rs9939609 as AA + AT or TT presenting high or low metabolic syndrome (MS) risk components. The percentile 25 for low levels and the percentile 75 for high levels stated by Gesteiro et al. [14, 16] were considered

Full size image

Table 4 shows the effect of the FTO neonate × mother allele interaction on the different parameters tested. Only significant results are shown. None of the parameters tested were significantly affected by the neonate × mother allele factor interaction. Maternal allele significantly affected pregnancy weight gain, neonatal glycemia, insulinemia, HOMA-IR, HOMA-IS, the glucose/insulin ratio, and homocysteinemia (at least P < 0.05), while neonatal allele affected AE activity. Lower levels of glucose, insulin, AE, and HOMA-IR (all P < 0.05) but higher (P < 0.05) pregnancy weight gain and the HOMA-IS were found in nAA + AT × mAA + AT vs. nAA + AT × mTT. Higher AE activity (P < 0.05) was found in nAA + AT × nTT vs. nTT × mTT.

Table 4 Pregnancy weight gain, neonatal glucose and insulin resistance/sensitivity markers, homocysteine, and arylesterase of the studied population depending on the neonatal and maternal FTO rs9939609 polymorphism interaction
Full size table

Figure 2 shows that mothers’ diet in the nAA + AT × mTT interaction group had significantly lower MDA score (P = 0.05) than those in the nAA + AT × mAA + AT group. The anthropometrical and clinical characteristics of the 30 mothers (and that of their respective children) distributed in the four association groups where nutrition data were available did not significantly differ from those of the 53 neonate-mother couples distributed in the four groups and tested for all other parameters (data not shown).

Fig. 2
figure 2

Mediterranean diet adherence (MDA) in the four neonate groups (nTT × mTT, N = 5; nTT × mAA + AT, N = 4; nAA + AT × mTT, N = 5; and nAA + AT × mAA + AT, N = 16; for abbreviations, see text). MDA scores were significantly higher (P = 0.05) in the interaction of nAA + AT × mAA + AT than in the nAA + AT × mTT one

Full size image

Discussion

Present most relevant results show that newborns from mAA + AT have lower insulin and HOMA-IR than their TT counterparts, while opposite results were found in nAA + AT. A higher percentage of neonates under the P25 level assessed by Gesteiro et al. [14] for some MS components were delivered by mAA + AT.

In agreement with previous studies [11, 25, 26, 34], no significant associations between the maternal or neonatal FTO rs9939609 polymorphism and birthweight, height, BMI, or ponderal index (PI) at birth were found. Several research groups tried to find when the FTO effect on BMI begins. Frayling et al. [11] reported that rs9939609 polymorphism was associated with changes in BMI and obesity at age 7. Da Silva et al. [7] observed significant differences in BMI and energy intake from lipids at 4 years old. Rzehak et al. [37] found no association of FTO variants with birthweight but could show that significant association developed over time in their population at age 4. López Bermejo et al. [29] assessed that at age of 2 weeks, some differences in PI, weight, and body fat composition associated to the FTO polymorphism were observed.

mAA + AT had 6.9 % higher pre-pregnancy weight and gained 18.8 % more weight during pregnancy than their T-homozygous counterparts. Similar results on pregnant A carriers have been observed by Lawlor et al. [26]. FTO polymorphisms have been associated with eating behavior and type of food consumed [9, 11, 19, 28, 42]. MDA and HEI scores tended to be lower and higher, respectively, in mTT than in mAA + AT. The former display a tendency for a higher consumption of animal foods and lower for some vegetables (data not shown), explaining the differences found in the quality of diets. As dietary information was not available in all the genotyped mothers, values have to be taken just as an approach. Nonetheless, according to a previous study [15], despite differences in mothers’ MDA or HEI scores, neonates displayed similar bodyweight but their IR markers were highly affected by their mother’s diet quality during pregnancy.

In adults, several studies have reported significant higher risk of T2DM in FTO A carriers [1, 6, 22]. Others have defined the association of FTO and obesity or MS in different populations [2, 30].The association between the FTO rs9939609 polymorphism and T2DM in adults appears modulated by diet [34]. Thus, in A carriers, there was a high risk of T2DM when MDA was low even after adjustment for BMI, which did not appear when MDA was high [5, 34].

All participant mothers had equivalent fasting glycemia in pregnancy. The distribution of mothers being positive for the O’Sullivan test [33] was not significantly different in the nTT vs. nAA + AA groups or in the mTT vs. mAA + AT groups.

Mothers of both nTT and nAA + AT display similar HEI but significantly different MDA scores. Thus, MDA maternal diet quality seems to be determinant for differences in analytical data. As expected, neonates carrying the obesogenic A allele showed higher insulinemia and HOMA-IR than their TT counterparts [6, 22].

However, it seems controversial that newborns from mAA + AT had lower MS level components than their TT counterparts, as A carriers are more prone to develop obesity and T2DM in adulthood, thus to a higher-glucose homeostasis and lipid marker modification [6, 22]. Moreover, in 30–40 % of the neonates from mTT glucose, insulin and HOMA-IR were higher than the P75 reference value, while less than 10 % were lower than the P25 described by Gesteiro et al. [14]. In the case of AA + AT’s offspring, between 20 and 40 % of neonates had values of these glucose homeostasis markers under the P25 of Gesteiro et al. (Fig. 1) [14]. In addition, mTT delivered more frequently neonates with high (>P75) and less frequently neonates with low (<P25) levels of MS markers than their AA + AT counterparts (Fig. 1).

Neonates from mAA + AT had higher tHcys levels than their TT counterparts, potentially related with the dysfunctional methylation suggested by Selhub [40]. It is known that FTO works as part of a complex process involving methylation and demethylation of nucleic acids [8, 23, 24]. These mAA + AT neonates had also significantly lower glucose, insulin, and HOMA-IR and higher HOMA-IS than their mTT counterparts. This was unexpected taking into account results in adults, showing that T homozygous vs. A carriers are less affected by BMI changes, obesity [11], and T2DM [21] and are less prone to develop cardiovascular disease (CVD) [27].

In contrast to the influence of maternal FTO on neonatal IR profile already discussed, nAA + AT showed increased insulin and HOMA-IR, suggesting that IR was already present in normoweights at birth. These findings concur with data for adults where A allele was associated to higher MS risk [2, 6, 45]. However, López Bermejo et al. [29] did not find any significant difference in glucose or insulin levels between the different FTO genotypes at birth. Although our study presents limitations due to the small sample size, it seems relevant the fact that belonging to mAA + AT offspring does not necessary imply suffering the negative effects attributed to the A allele. In fact, lower neonatal levels of glucose, insulin, AE, and HOMA-IR were found in the nAA + AT × mAA + AT group with respect to its nAA + AT × mTT counterpart (Table 4). Although those results seem difficult to be explained, they could be related to the higher MDA score of maternal diet found in the nAA + AT × mAA + AT group with respect to that of the nAA + AT × mTT one, suggesting that the benefits of the high MDA already indicated as mothers with higher MDA pregnancy diet deliver neonates with lower glucose, insulin, and HOMA-IR levels [17]. Western diets are rich in saturated fatty acids and poor in carbohydrates, dietary factors related with the IR development [15, 31, 38]. Adequate weight gain during pregnancy plays a central role on glucose homeostasis at birth and should be adapted to pre-conception weights [38]. Mother pregnancy weight gain of all mothers scored in the range considered appropriate [38] (8.5 to 15 kg in our study). However, those mothers belonging to the nAA + AT × mAA + AT group gain the average weight gain and about 2.3 kg more weight than their nAA + AT × mTT counterparts, suggesting that higher MDA diet also contributes to weight gain and in turn to a better glucose homeostasis marker levels [17, 38]. Nonetheless, we are far from knowing major mechanisms involved in this neonatal × maternal genotype association. Moreover, we also do not know whether the association effects found at birth would be also present later in life in case of consuming diets similar to those followed during gestation by their respective mothers and should be checked in future follow-up study.

Although this study shows some limitations, (a) the low number of neonate × mother couples tested, (b) only SNP rs9939609 was studied, and (c) diet pattern was considered in 57 % of mothers, it has several strengths counterbalancing the mentioned limitations, (a) neonates were all AGA, normoweight, and strictly selected and their mothers did not suffer from GDM and had an adequate weight gain during pregnancy; (b) results were adjusted for maternal glycemia at the O’Sullivan test as confounder factor; and (c) it is the first study associating neonatal and maternal FTO polymorphisms with glucose homeostasis, lipid profile, and tHcys at birth.

Conclusion

Maternal FTO rs9939609 polymorphism affects in an opposite manner than neonatal FTO the IR and CVD markers in term, normoweight, AGA neonates at birth. Maternal FTO alleles affected much more than neonatal FTO alleles several parameters linked to the glucose homeostasis. The potential negative effects on glucose homeostasis observed in nAA + AT were counterbalanced in neonates delivered by mAA + AT but not by mTT, probably due to the higher MDA of their diets followed during pregnancy. Further investigations, replicating present associations in a larger population, are desirable to understand this interesting neonatal-maternal interaction.