Introduction
Non-communicable diseases, including obesity, diabetes, and cardiovascular disease, are a major health hazard of the modern world. While genetics and suboptimal adult environmental factors can affect an individual’s propensity to develop metabolic abnormalities, early life environmental factors including (maternal) nutrition are increasingly recognized as important contributors influencing health and disease risk in later life.Reference Koletzko, Godfrey and Poston1–Reference Koletzko5 Rodent models help elucidate the effects and mechanisms involved in maternal and early life dietary exposures (e.g., maternal undernutrition, high-fat diet, micronutrient exposure) and their impact on long-term health outcomes.Reference Fernandez-Twinn and Ozanne6–Reference Bianco-Miotto, Craig, Gasser, van Dijk and Ozanne8 We previously established a mouse model of nutritional programming to determine whether dietary manipulations in early life could alter later-life health outcomes. We use this mouse model to test the effects of IMF interventions on offspring growth patterns and susceptibility to diet-induced obesity later in life. This validated model forms the basis for testing our nutritional concepts aimed at promoting healthy growth and development in children. Using this model, we have shown that lipid quality in early life diet affects offspring susceptibility to adult diet-induced obesity.Reference Oosting, Kegler, Boehm, Jansen, van de Heijning and van der Beek9–Reference Baars, Oosting and Engels11
In addition to dietary interventions, rodents in nutritional programming research are exposed to standard rodent diets prior to and/or during the experiment. These diets serve as control or base diets for a nutritional supplementation/intervention. Grain-based and purified diets are two common standard rodent diets in (metabolic) research. Both diets have long been considered nutritionally sufficient to support breeding and long lifespan. However, they are very different in nutritional composition and food matrix.Reference Ricci and Ulman12, Reference Pellizzon and Ricci13 Significant differences include the matrix (multi-nutritional ingredients versus purified, single ingredients); ingredient types (unrefined versus refined); the quantity and source of fibers (diverse range of soluble/insoluble fibers versus mainly insoluble cellulose), carbohydrates (whole grains versus combination of refined corn starch with maltodextrin and simple carbohydrates) and proteins (mainly plant based versus milk casein). Such matrix and compositional differences could lead to different effects on nutrient absorption, gut microbiota, metabolic responses, and subsequent health outcomes. Research has shown varied effects of these diets on gut microbiota and short chain fatty acid (SCFA) profile.Reference Toyoda, Shimonishi, Sato, Usuda, Ohsawa and Nagaoka14–Reference Pellizzon and Ricci18 In addition, rodents exposed to purified diets developed an impaired liver phenotype,Reference Daubioul, Rousseau and Demeure19, Reference Ronda, van de Heijning and de Bruin20 an effect significantly reduced by addition of soluble fiber,Reference Pontifex, Mushtaq and Le Gall21 whilst higher serum cholesterol and triglycerides,Reference Lien, Boyle, Wrenn, Perry, Thompson and Borzelleca22 and slightly lower growth rate and food intake have also been observed.Reference Rutten and de Groot23While the impact of diet is increasingly recognized, the influence of the mother’s background diet on offspring health outcomes (i.e., nutritional programming) may still be overlooked. A recent studyReference Zou, Ngo, Wang, Wang and Gewirtz24 demonstrated that maternal diet lacking soluble fiber during lactation led to changes in offspring microbiota, predisposing them to obesity later in life. Building on these findings, we used our nutritional programming model to determine whether maternal exposure to a grain-based versus purified diet during early lactation could contribute to distinct programming effects on offspring growth, metabolic phenotype (focusing on body weight and composition, metabolic organs, hormones and inflammatory markers), and gut microbiota profile.
Methods
Animal procedures
The C57BL/6J mouse strain was selected as this strain is frequently used in research and is susceptible to diet-induced obesity. Mice used for this study were derived from a larger study with breeding procedures as described in detail elsewhere.Reference Schipper, Tims, Timmer, Lohr, Rakhshandehroo and Harvey25 Briefly, breeding pairs from Charles River Laboratories (Saint-Germain-Nuelles, France) were time-mated and day of birth was recorded as postnatal day 0 (PN0). At PN2, litters were cross fostered and/or culled to six pups/dam. Each litter contained both sexes and 2 to 4 male offspring, depending on birth outcomes. Male offspring were weaned at PN21 and were pair-housed (with same-sex littermate) and followed up into adulthood. Animals were housed in IVC polycarbonate type II cages with bedding, nesting and enrichment materials as previously described.Reference Schipper, Tims, Timmer, Lohr, Rakhshandehroo and Harvey25 All procedures took place during the light phase.
Dams were fed a grain-based diet throughout gestation. After birth, dams and litters were randomly allocated to a grain reference group [Grain-Ref] that remained on grain-based diet throughout the study, or one of four experimental groups that experienced a shift to purified diet at either PN2 [MatAIN] or PN16 [MatGrain] (Fig. 1 and Supplementary Table 1). Offspring in MatAIN and MatGrain groups were exposed to a standard AIN-93G based infant milk formula (IMF) diet containing soluble galacto-oligosaccharides and fructo-oligosaccarides (GOS/FOS) between PN16 and PN42, followed by the semisynthetic control AIN-93 M [Con] or Western-Style Diet [WSD] until PN126, following a previously described nutritional programming model.Reference Oosting, Kegler, Boehm, Jansen, van de Heijning and van der Beek9–Reference Baars, Oosting and Engels11 The Grain-Ref group was included in the study as a reference to health outcomes of mice kept in same conditions as the experimental groups but without any diet interventions.
Figure 1. Experimental design. From two weeks before mating and throughout gestation dams were subjected to grain-based growth diet (Teklad 2920X-irradiated). Dams and litters in the grain reference (Grain-ref) group remained on the grain-based diet until PN42 and were switched to grain-based maintenance diet (Teklad 2916C) from PN42 to PN126. In the other four groups, from PN2 to PN16 (early lactation), dams were exposed to either the grain-based diet or purified AIN-93 growth (AIN-93-G) diet which resulted in two groups based on maternal diet type abbreviated as Mat; MatGrain or MatAIN accordingly. Between P16 and P42, dams and litters were exposed to standard infant milk formula (IMF) diet which was AIN-93G based and between P42 and P126 (adulthood), male offspring received a purified control (AIN-93-M) or Western-style diet (WSD, consisting of 20% w/w fat −17% w/w lard, 3% w/w soy, 0% w/w cholesterol). Body composition was measured by echo-MRI on PN28, PN42, PN98 and PN126. Fecal samples were collected on PN28, PN42 and PN126. The experimental groups are represented in the figure. 1) Grain-Ref (n = 12); 2) MatGrain-Con (n = 12); 3) MatGrain-WSD (n = 12); 4) MatAIN-Con (n = 12); and 5) MatAIN-WSD (n = 12). One mouse in the MatAIN-Con presented malocclusion, resulting in low body weight gain after PN42; data from this animal were excluded from analyses.
Body weight, energy intake and body composition
Body weight of dams and litters were recorded weekly and, after weaning, offspring body weight was monitored twice weekly. Body composition was measured at PN28, PN42, PN98 and PN126 by magnetic resonance imaging (EchoMRI-100™ analyzer, EchoMRI Medical Systems, Houston, TX) as previously reported.Reference Schipper, Tims, Timmer, Lohr, Rakhshandehroo and Harvey25 Food intake was roughly monitored per cage by weighing the food on rack twice a week between PN42 and PN126.
Tissue collection
Fecal samples were collected at PN28, PN42, and PN126 and were processed for fecal DNA extraction and sequencing as previously reported.Reference Schipper, Tims, Timmer, Lohr, Rakhshandehroo and Harvey25 At PN126 animals were deeply anesthetized (isoflurane) and sacrificed as previously reported.Reference Schipper, Tims, Timmer, Lohr, Rakhshandehroo and Harvey25 Subcutaneous and visceral (perirenal, retroperitoneal and epidydimal) white adipose tissue depots, intrascapular brown adipose tissue depots, adrenal glands, (tibialis anterior) muscle and liver were dissected and weighed. Cecum content was collected and processed for analysis of SCFAs.
Liver histology and triglycerides
For histological analysis, liver (left lobe) samples were placed in 10% formalin for approximately 48 hours followed by storage in 70% ethanol until paraffin embedded. Paraffin sections were stained with hematoxylin and eosin (H&E) for routine histological analysis.Reference Feldman and Wolfe26 Liver sections were cut to 5 μm thickness. Sections were air dried for 30 min, followed by fixation in 4% formaldehyde for 10 min. Hematoxylin nuclei staining was subsequently carried out for 5 min followed by several rinses with distilled H2O. Sections were mounted in aqueous mounting media (Imsol, Preston, UK). The H&E slides from the liver specimens were blindly evaluated by using an adapted version of the nonalcoholic steatosis scoring system for nonalcoholic fatty liver diseaseReference Kleiner, Brunt and Van Natta27 and reviewed by two certified veterinary pathologists. This scoring system considers the presence or absence of steatosis in hepatocytes examined at low magnification, the presence of ballooning cells, and incidence of lobular inflammation.
For triglyceride analysis, part of the left lobe was snap frozen and stored at −80°C. Liver triglycerides were determined in liver homogenates prepared in buffer containing 250 mM sucrose, 1 mM EDTA, and 10 mM Tris-HCl at pH 7.5 using a commercially available kit (Instruchemie, Delfzijl, The Netherlands) according to the manufacturer’s instructions.
Blood and plasma measurements
Blood glucose levels were determined immediately after sacrifice using a commercial blood glucose meter and test strips (Accu-Chek Performa, Roche Diabetes Care, Inc.). Blood was collected in EDTA-coated tubes (Sarstedt, Etten-Leur), centrifuged (13,000 rpm, 15 min, 4°C), and plasma was removed and stored at –80°C until analysis. Interleukin-6 (IL-6), insulin, leptin and resistin were measured using the Mouse Metabolic Hormone Expanded Panel multiplex assay (MILLIPLEX® MAP), monocyte chemoattractant protein-1 (MCP-1) was quantified with the Mouse MCP-1 SimpleStep ELISA® Kit (Abcam) and lipopolysaccharide binding protein (LBP) was measured with the mouse LBP ELISA kit (Hycult®Biotech). All the assays were performed according to the manufacturer’s instructions. Plasma analyses were performed in duplicate, and samples were excluded when duplicate measurements had coefficient of variation (CV)>20%.
SCFA analyses
Cecum content was weighed, and samples were diluted 1:10 according to weight in pre-cooled phosphate-buffered saline. Samples were vortexed 3 times for 30 s and centrifuged at 4°C for 5 min at 15 000 × g. The supernatant was collected and 200 µL was used for SCFA analysis. The following SCFAs –acetic, propionic, n-butyric, iso-butyric, n-valeric, and isovaleric acids – were quantified on a Shimadzu-GC2025 gas chromatograph with a flame ionization detector and hydrogen as mobile phase. Quantification was performed by using 2-ethylbutyric acid as an internal standard and generating a calibration curve from the peak area after which the concentration in the samples was calculated.
Analysis of sequencing results
Analysis of fecal sequencing results was performed as extensively described previously.Reference Schipper, Tims, Timmer, Lohr, Rakhshandehroo and Harvey25 Rarefaction was applied to the taxa by phyloseqReference McMurdie and Holmes28 and vegan packagesReference Jari Oksanen, Friendly and Kindt29 in R v3.5.1 for α-diversity calculations using the Chao1 and Shannon index metrics. The β-diversity was computed using the Bray-Curtis distance over all samples with functions vegdist and betadisper from the vegan package in R v3.5.1. Statistical significance of differences in α-diversity were assessed with pairwise_wilcox_test function from the rstatix package in R v4.0.2Reference Rstatix30 followed by Benjamini-Hochberg p-value adjustment per timepoint. Statistical significance of differences in β-diversity were assessed using the permutation ANOVA function adonis2 from the package vegan in R. Using Spearman, the phenotypic metadata was correlated to genera with a minimum mean relative abundance of 0.5% across all samples and tested for significance using cor.test function with default settings from the R stats package. Differential abundance was performed with generalized linear models with mixed effects on the sequencing counts using the glmmTMB package v 1.1.2.3 in R v4.0.2Reference Brooks31 followed by Anova.glmmTMB applying the Chi Squared test for significant differences. After adjustment, a p-value < 0.05 was considered significant for all statistical tests applied to the sequencing data.
Statistical analysis
Phenotypic data were analyzed using SPSS 20.0 (IBM software) and GraphPad Prism 8 (GraphPad software, GraphPad Holdings, LLC, La Jolla, CA, USA). Data from the Grain-Ref group were not included in the statistical analyses, but data are added to figures as a visual reference. Due to the color and texture difference between grain-based diet and purified diet, researchers were not blinded to diet type. However, ex vivo analyses and data processing was performed by researchers blinded to the groups.
Data were analyzed using linear mixed models. Effect of maternal diet type on changes in body weight and body composition over 3 weeks (PN21 – PN42) in the MatGrain (group 2 and 3 combined) versus MatAIN (group 4 and 5 combined) was analyzed by one-way repeated measures ANOVA using maternal diet type as fixed factor and time as repeated measure, excluding data at missing timepoints. Post hoc analyses were performed using Bonferroni’s test. Effect of adult diet type on changes in body weight and body composition over 12 weeks (PN42-PN126) in the groups 2 – 5 was analyzed by two-way repeated measures ANOVA using maternal/adult diet types as fixed factors and time as repeated measure. Effect of diet type on organ and plasma parameters at PN126 was analyzed by two-way ANOVA using maternal and adult diet types as fixed factors. Individual animals were considered as statistical units, however, as the study included multiple batches of mice and mice were always housed two animals per cage throughout the study, all analyses included batch and cage as random factors. The relation between diet type and liver phenotype as indicated by %responder was analyzed using Chi-square test.
All data are expressed as mean ± standard error of the mean (SEM). Data were considered statistically significant when p < 0.05. Statistical trends were reported in case of a p-value between 0.05 and 0.06. Three-way interactions were considered statistically significant when p < 0.1 as a common practice in more complex models. Power calculations were based on published data from previous experiments with comparable design and based on fat accumulation in response to WSD in male adult offspring.Reference Oosting, Kegler and Wopereis10 Using an error-probability of 5% and power of 80%, sample size was calculated as 12 animals per group. There was one animal in the MatAIN-Con that presented malocclusion, resulting in low body weight gain after PN42; data from this animal were excluded from all analyses.
Results
Maternal exposure to purified diet (AIN-93G) during early lactation (PN2-PN16) impacted the pattern of maternal and litter weight gain
During lactation, dams and litters in the MatGrain and MatAIN group showed a different pattern of body weight accumulation (diet*time, dams: F (3,12) = 8.51, p < 0.01; litters: F (3,13) = 4.53, p = 0.02) with animals in MatAIN showing lower bodyweight compared to animals in MatGrain at PN14 and PN21 (Fig. 2).
Figure 2. Dam body weight (A) and litter weight (B) in the period PN2-PN21 in the MatGrain, MatAIN and Grain-ref groups. Effect of diet type on body weight was analyzed by one-way repeated measures ANOVA using maternal diet type as fixed factor and time as repeated measure. ainteraction effect between maternal diet and time. *MatGrain and MatAIN groups differed at depicted time points by post hoc analysis using bonferroni testing, p < 0.05. n = 5–8 (A). Grain-ref (n = 5 litters); MatGrain (n = 8 litters); and MatAIN (n = 7 litters). Each litter contained 6 pups in total (2–4 of which were males, depending on birth outcomes) (B). Values are given as mean ± SEM.
Maternal exposure to purified diet during early lactation increased offspring growth velocity after weaning and decreased offspring response to WSD challenge in adulthood
In the offspring, there was a significant increase in body weight over time in both MatGrain and MatAIN groups (time: F (2,69) = 5586.97, p < 0.001) during the post-weaning period (PN21 – PN42), as well as an interaction between maternal diet and time on offspring body weight (maternal diet*time: F (2,69) = 9.26, p < 0.001) during the same period. Post hoc testing indicated that offspring from MatAIN mice had lower body weight compared to offspring from MatGrain mice at PN21 (p < 0.001) and PN28 (p < 0.001) and had similar body weight at PN42 (p = 0.97) (Fig. 3A). There was no difference in fat mass and lean body mass at PN28 and PN42 (Fig. 3).
Figure 3. Longitudinal body weight (BW) in the post-weaning period PN21-PN42 (A), in the groups MatGrain, MatAIN and Grain-Ref. Average fat mass (% BW) at PN28 (B) and lean mass (% BW) at PN28 (C), longitudinal BW (D), fat mass (% BW) (E) and lean mass (% BW) (F) in the groups MatGrain-Con, MatGrain-WSD, MatAIN-Con, MatAIN-WSD and Grain-Ref. Maternal diet (Grain versus AIN-93G), adult diet (WSD versus AIN-93M), time, and diet-by-time interaction effects were determined by repeated measures one-way ANOVA for the period (PN21-PN42) and repeated measures two-way ANOVA for the period (PN42-PN126). a interaction effect between maternal diet and time (A) and interaction between maternal diet, adult diet and time (E), b interaction effect between adult diet and time (D–F), p < 0.05. *MatAIN and MatGrain groups in panel a and MatGrain-WSD and MatAIN-WSD groups in panel E differed at depicted time points by post hoc analysis using Bonferroni testing, p < 0.05. n = 11**–12. **MatAIN-Con group in panel A-D. Values are given as mean ± SEM.
During the adult phase (PN42 – PN126) there was an interaction effect between adult diet and time on offspring body weight (F(2,79) = 81.10, p < 0.001) and relative lean body mass (F (2,88) = 40.56, p < 0.001)(Fig. 3D and 3F). Post hoc analysis indicated that body weight was significantly higher and relative lean body mass was significantly lower at PN98 and PN126 due to adult WSD exposure.
There was an interaction between maternal diet, adult diet and time on offspring fat mass (F (2,86) = 3.95, p = 0.02). Post hoc analysis indicated that relative fat mass was significantly lower in the offspring from MatAIN compared to MatGrain at PN98 (p < 0.01) and PN126 (p = 0.02) only following WSD challenge in adulthood (Fig. 3E). There was no statistically significant effect, though visually there seemed to be an interaction between maternal diet, adult diet and time on offspring relative lean body mass when exposed to WSD challenge; lean body mass seemed to be higher in the offspring from MatAIN compared to MatGrain at PN98 and PN126 when exposed to WSD (Fig. 3F). Maternal diet had no effect on offspring energy intake from PN42 to PN126 whereas, WSD exposure increased caloric intake (F (1,20) = 66.72, p < 0.001) (Supplementary Fig. 1).
The Grain-Ref group showed similar patterns of weight gain to both MatGrain and MatAIN groups in the post-weaning period (PN21-PN42) (Fig. 3A). There was a difference between the Grain-Ref, MatGrain-Con and MatAIN-Con groups in terms of body weight and body composition development in adulthood period (PN42-PN126); numerically the Grain-Ref group had higher body weight and relative fat mass and lower relative lean mass compared to the other groups (Fig. 3D and 3E and 3F).
Maternal brief exposure to purified diet during early lactation did not have a significant effect on organ weights
At PN126, WSD resulted in a decrease in relative tibialis anterior muscle mass and an increase in relative total fat, visceral fat, subcutaneous fat and brown fat mass. Relative tibialis anterior muscle mass was higher in offspring from MatAIN compared to MatGrain (p = 0.05). In line with effect of maternal diet type on changes in body composition observed during adulthood, the weight of the adipose tissue depots in the groups exposed to WSD appeared to be lower in MatAIN compared to MatGrain at dissection, however this effect was not significant in the statistical model used (Table 1). Moreover, neither maternal nor adult diet affected cecum content weight and adrenal gland weights. At PN126, cecum short chain fatty acid profile was analyzed which indicated no effect of maternal diet nor an interaction between maternal and adult diet on cecum content weight, total amount of SCFAs and the relative levels of individual SCFAs (data not shown).
Table 1. Average weight of fat depots and organs at PN126
Values are mean ± SEM, n = 11–12. Statistical analyses were performed using two-way ANOVA, no interaction effects.
Histological analysis indicated presence of liver steatosis and inflammation in all the experimental groups that switched to purified diet at either PN2 or PN16
Liver sections obtained at PN126 were stained and scored for anomalies/pathologies. While there was no fat accumulation in the liver of the Grain-Ref mice, a marked heterogeneity in fat accumulation and histology was observed in the four experimental groups. A few animals in all experimental groups developed liver steatosis (Fig. 4A) or inflammation (Fig. 4B) while there was no evidence of nonalcoholic fatty liver disease (Supplementary Table 2). We measured the response rate to purified diet based on the presence of steatosis and/or inflammation, which indicated, surprisingly, that 30%–50% mice per experimental group were responders yet there was no significant correlation between experimental diet groups and response rate (Fig. 4C). Liver weight was not modulated by maternal nor adult diet. Quantitation of hepatic triglycerides confirmed liver fat accumulation in all the experimental groups. Hepatic triglycerides seemed to be lower following WSD challenge in the offspring from MatAIN compared to MatGrain group, however, this effect was not statistically significant (Fig. 4E).
Figure 4. Hematoxylin and eosin (H&E) staining of representative liver sections of the mice scored positive for steatosis (A) and inflammation (B) in the study groups with a switch to AIN-93 diet. % responder rate (defined by the outcome of H&E staining and based on the presence of steatosis and/or inflammation) (C), liver mass (% BW) (D), liver triglyceride (TG) content (mg/g protein) (E). The relation between experimental diet group and liver phenotype as indicated by %responder was analyzed using chi-square test. Values are given as mean ± SEM. n = 11*-12 mice per group (C–E). *MatAIN-Con. central vein (CV), portal tract (PT).
Maternal brief exposure to purified diet during early lactation seemed to decrease plasma leptin levels following WSD challenge
The WSD challenge resulted in an overall increase in blood glucose and plasma insulin levels at PN126 (glucose: adult diet, F (1,21) = 20.79, p < 0.001; insulin: adult diet, F (1,15) = 3.18, p = 0.09), however, maternal diet type had no effect (Fig. 5A and 5B). The WSD challenge also increased plasma leptin, MCP-1 and resistin, supporting a WSD induced obesogenic phenotype, but did not modulate IL-6 and LBP levels (leptin: F (1,31) = 22,12, p < 0.001; MCP-1: F (1,19) = 7.60, p = 0.01; resistin: F (1,37) = 5.51, p = 0.02, Fig. 5C–G). Plasma leptin levels seemed to be lower in MatAIN versus MatGrain group following WSD challenge, although the interaction effect did not reach significance (F (1,31) = 3.55, p = 0.07, Fig. 5C). The Grain-Ref group had a visually higher leptin level compared to MatGrain-Con and MatAIN-Con groups.
Figure 5. Plasma glucose is expressed as mmol/L (n = 11–12). Plasma markers (leptin (n = 7–11), monocyte protein-1 (MCP-1, n = 6–11), resistin (n = 9–11) and interleukin-6 (IL-6, n = 3–9)) and insulin (n = 5–10) are expressed as pg/ml and lipopolysaccharide binding protein (LBP) as ng/ml (n = 7–11). Volume of plasma collected was not insufficient for all analyses resulting in lower n/group. Effect of diet type on plasma measures was analyzed by two-way ANOVA using maternal and adult diet types as fixed factors. Data presented as mean ± SEM. b significant effect of adult diet, p < 0.05.
Maternal diet type during early lactation affected the offspring gut microbiota diversity and composition in the post-weaning period
Analysis of alpha diversity showed lower species richness, measured by Chao1 index, in the offspring from MatAIN compared to MatGrain group at early time points PN28 (Fig. 6A) and PN42 (Fig. 6B) which was not present at later time point PN126 (Fig. 6C). Shannon index indicated no differences at PN28 and PN42 and a significantly increased diversity by WSD challenge at PN126 (Fig. 6D–F). Grain-Ref group visually had a higher alpha diversity by both indices at all the time points (Fig. 6).
Figure 6. Alpha diversity assessed by Chao1 index at PN28 (A), PN42 (B) and PN126 (C) and Shannon index at PN28 (D), PN42 (E) and PN126 (F). Statistical significance of differences in alpha diversity were assessed with pairwise_wilcox_test followed by Benjamini-Hochberg p-value adjustment per timepoint. Data presented as median ± interquartile range. * p < 0.05, ** p < 0.01, *** p < 0.001 n = 23*-24 mice per group (panel A, B, D, E) * MatAIN. n = 11*-12 mice per group (panel C and F) * MatAIN-con.
Analysis of beta diversity, quantifying (dis-)similarities in microbiota composition between samples, showed differences in microbiota composition due to maternal diet at PN28, PN42 and PN126 as well as main effect of adult diet type at PN126 (Fig. 7). The Grain-Ref group was clearly separated from all the other groups at early and later time points (Supplementary Fig. 2). Analysis of microbial taxa relative abundance at phylum level both at PN28 and at PN42 showed a slightly higher relative abundance of Verrucomicrobiota and Actinobacteria and a lower relative abundance of Firmicutes and Bacteroidota in MatAIN compared to MatGrain group. Analysis at PN126 showed an increase in the relative abundance of the phylum Firmicutes, particularly in the WSD-challenged groups, compared to the levels observed at PN42. No differences between MatGrain and MatAIN groups were observed at P126. In the Grain-Ref group, the microbiota profile was dominated by the phyla Firmicutes and Bacteroidota (Supplementary Fig. 3).
Figure 7. Beta diversity computed with functions vegdist and betadisper from the vegan package in R v3.5.1. Statistical significance of differences in the beta diversity were assessed using the permutation ANOVA function adonis2 from the package vegan in R. PN28; MatDiet: F = 3.62; p = 0.00. PN42; MatDiet: F = 2.54; p = 0.00. PN126; MatDiet: F = 2.64; p = 0.04; AdultDiet: F = 17.85; p = 0.00; MatDiet:AdultDiet: F = 1.16; p = 0.26.
Analysis at genus level showed that at PN28, the offspring from MatAIN compared to MatGrain group, showed a significantly higher relative abundance of Bacteroides and lower relative abundances of Faecalibaculum, an unknown genus of Muribaculaceae and Parasutterella (Fig. 8A). At PN42 only the abundance of the genera Alistipes and the Lachnospiraceae NK4A136 group were lower in the MatAIN compared to the MatGrain group (Fig. 8B). At PN126, offspring exposed to WSD compared to AIN control diets showed significant difference in the relative abundance of many bacterial genera, among which a few belonging to the Firmicutes phylum, such as Colidextribacter and Lactobacillus, were higher and Akkermansia and Parasutterella were lower in the groups exposed to WSD (Supplementary Fig. 4). There was no significant maternal diet effect nor an interaction effect between maternal and adult diet on relative abundance of microbial taxa at PN126.
Figure 8. Relative abundance at genus level for PN28 (A) and PN42(B) performed with generalized linear models. Cross-sectional correlations between bacterial taxa at PN28 (groups 2, 3, 4 and 5 combined) and body weight, fat mass (% body weight) and lean mass (% body weight) using Spearman correlation analysis (C). Statistical significance of the relative abundance data was assessed using Chi Squared test. The resulting p-values were corrected using Benjamini-Hochberg. Data presented as median ± interquartile range. * p < 0.05, ** p < 0.01, *** p < 0.00. n = 23*-24 mice per group. * MatAIN.
Next, we examined the cross-sectional correlations between relative abundance of bacterial groups at genus level and measured metabolic outcomes, i.e., body weight, relative fat mass and relative lean body mass, at PN28 and PN42. At PN28, a few bacterial taxa from Bacteroidetes phylum correlated (ρ > 0.5 or ρ < −0.5) with body weight. There was a negative correlation between body weight and the Bacteroides genus, and positive correlations between body weight and the following genera: Alistipes, an unknown genus of Muribaulaceae, Rikenellaceae RC9 gut group, and an unknown genus of Tannerellaceae. At PN42, relative lean body mass correlated positively with the Akkermansia genus (ρ = 0.48, data not shown).
Discussion
We have previously shown effects of early life nutrition on adult (metabolic) health outcomes using a nutritional programming model.Reference Oosting, Kegler and Wopereis10,Reference Baars, Oosting and Engels11,Reference Oosting, van Vlies and Kegler32–Reference Kodde, van der Beek, Phielix, Engels, Schipper and Oosting34 In this study, we describe the persistent programming effects of maternal exposure to standard purified diet versus grain-based diet during early lactation on offspring’s response to WSD in adulthood. Offspring of dams exposed to a purified compared to grain-based diet exhibited reduced body weight at weaning, increased growth velocity in the post-weaning period and a lower fat accumulation (% total weight) in response to adult WSD challenge. These effects were in parallel with an adolescent microbiota profile characterized by reduced alpha diversity and a distinct composition depending on maternal diet type.
Considering the nutritional differences and the less favorable attributes of a purified diet on health outcomes in metabolic research,Reference Toyoda, Shimonishi, Sato, Usuda, Ohsawa and Nagaoka14–Reference Rutten and de Groot23 one might speculate that early life exposure to such a diet, compared to a grain-based diet, could increase susceptibility to adult diet-induced obesity. We observed decreased fat mass accumulation in response to WSD challenge due to maternal exposure to a purified diet during early lactation. This effect was seen in WSD-challenged offspring but not in non-WSD groups, suggesting it is not due to a general alteration in body fat accumulation. The response to a high-fat diet varies across studies and even among mice within the same study group,Reference Duval, Thissen and Keshtkar35 often attributed to gene-environment-microbiome interactionsReference Siersbaek, Ditzel and Hejbol36–Reference Yang, Smith, Keating, Allison and Nagy38 and sexual dimorphism.Reference Casimiro, Stull, Tersey and Mirmira39 However, early life nutrition is an often-overlooked determinant of this variability.
Emerging evidence suggests that early nutritional experiences significantly influence metabolic responses to dietary challenges later in life. While we cannot conclusively determine whether a maternal grain-based diet promotes fat mass accumulation in response to WSD or if a purified diet impedes it, we can assert that fat mass accumulation triggered by WSD is substantially influenced by the maternal diet during early lactation. Contrary to our findings, a recent studyReference Zou, Ngo, Wang, Wang and Gewirtz24 observed that feeding dams a standard purified diet during lactation resulted in offspring with higher body weight and adiposity at weaning, and increased sensitivity to diet-induced obesity later in life. These discrepancies may be due to differences in study design, particularly the nutritional environments between PN16 and PN21. In the previous study,Reference Zou, Ngo, Wang, Wang and Gewirtz24 pups were exposed to a fiber-rich grain-based diet starting at PN21, whereas in our study, exposure to a diet devoid of soluble fiber was limited to early lactation and ended at PN16. At PN16, pups were transitioned to an AIN-93G-based IMF diet with GOS/FOS as a source of soluble fiber. This critical phase for organ development and programming, including adipose tissue development,Reference Kodde, Engels, Oosting, van Limpt, van der Beek and Keijer40 may account for the different outcomes observed.
Given the critical role of gut microbiota in nutrient digestion, energy harvest and production of bioactive metabolites,Reference Wopereis, Oozeer, Knipping, Belzer and Knol41–Reference Blanton, Charbonneau and Salih44 we assessed gut microbiota profiles. We identified an (adolescent) microbiota profile characterized by reduced alpha diversity and a distinct composition at PN28 and PN42 in MatAIN versus MatGrain groups. This distinct composition was further characterized by a decrease in the Bacteroidota phylum and an increase in the genera Bacteroides in the offspring from dams exposed to a purified compared to a grain-based diet specifically at PN28. Bacteroides species are well-known for their ability to utilize various carbohydrate structures. The higher abundance of Bacteroides in MatAIN versus MatGrain may imply that more carbohydrates are reaching and/or being released in the colon when exposed to purified versus grain-based diets and that, in addition to (soluble) fiber, digestible carbohydrate composition of these diets could contribute to the observed effects.
The Bacteroidota phylum has been associated with the modulation of body weight, and we also found moderate, but significant, correlations between the PN28 levels of bacterial genera belonging to the phylum Bacteroidota and body weight. We acknowledge that these observed changes might function more as markers than direct causative factors. For instance, cultured isolates from the Alistipes genus have demonstrated bile resistance.Reference Parker, Wearsch, Veloo and Rodriguez-Palacios45 Therefore, variations in Alistipes abundance could potentially serve as a marker of alterations in the host’s fat metabolism rather than a direct cause. However, it is particularly intriguing that these correlations, along with significant changes in the relative abundance of certain genera, are evident at the time point when differences in body weight are observed, specifically at PN28.
There were two interesting additional observations in the microbiota data. The MatGrain group, despite the exposure to (AIN-based) IMF diet containing soluble fibers (GOS/FOS), had a very distinct microbiota composition compared with the Grain-Ref group at PN42 (Supplementary Fig. 2). It is noteworthy to mention that the (AIN-based) IMF diet contains less fiber (AIN-based IMF: 3% GOS/FOS and 3% cellulose) than a grain-based diet (15%–25% mostly soluble fiber) which could have played a role. In addition, after the WSD challenge at PN126, we observed a notable stimulatory impact of the WSD on gut microbiota diversity and certain bacterial genera, consistent with previous findings.Reference Malesza, Malesza and Walkowiak46–Reference Jo, Seo and Park48 Although we detected a statistically significant effect of the maternal diet type on Beta diversity at PN126, this effect was not attributable to consistent taxonomic changes and was not as pronounced as the impact of WSD at this timepoint.
Previously, it has been noted that mice fed AIN-93G diet have higher TG accumulation in the liverReference Ronda, van de Heijning and de Bruin20, Reference Aguiar, Moura and Ballard49 and C57BL/6J inbred mice showed heterogeneity in liver response to WSD challenge.Reference Duval, Thissen and Keshtkar35, Reference Koza, Nikonova and Hogan50, Reference Burcelin, Crivelli, Dacosta, Roy-Tirelli and Thorens51 Our findings indicate a clear development of liver steatosis and/or inflammation in 30%–50% of the offspring across all groups that transitioned to a purified diet, a phenomenon not observed in the Grain-Ref group. Importantly, the liver phenotype was not influenced by the maternal diet type. Mechanistically, the process by which a purified diet induces the development of liver steatosis is not well understood. However, the lower quality and quantity of fiber in a purified diet compared to a grain-based dietReference Daubioul, Rousseau and Demeure19–Reference Pontifex, Mushtaq and Le Gall21, Reference Aguiar, Moura and Ballard49 suggests a significant role for microbial involvement. Notably, the observation of liver steatosis and/or inflammation across all study groups after transitioning to a purified diet (compared to the Grain-Ref group), even in the absence of a WSD challenge later in life, raises legitimate concerns about the long-term effect of purified AIN diet on liver health.
This study has some limitations. First, we studied the effects of maternal dietary exposure on offspring health outcomes in male mice only. While exclusion of female offspring reduced the total number of animals needed for this study, we acknowledge that this choice contributes to the sex bias prevailing in preclinical research.Reference Karp and Reavey52 Next, we indicated an accelerated growth rate in the MatAIN group compared to the MatGrain group during the post-weaning period. Unfortunately, individual-level caloric intake during in this period could not be determined due to pair housing. While we didn’t expect variations in food intake, we cannot eliminate the possibility. Moreover, we investigated the weight of both dams and litters during the pre-weaning period (PN2-PN21). Notably, differences in weight accumulation were already evident by PN7, as indicated by lower weight in dams and litters exposed to the purified diet. However, the underlying mechanisms responsible for this apparent disparity in offspring weight – whether related to altered energy transfer from mother to pup (such as variations in dam milk availability or composition) or other contributing factors – remain to be elucidated. In addition, we identified a different microbiota profile in MatAIN compared to MatGrain group at PN28 and PN42. Whilst we acknowledge the role of SCFAs in host energy metabolismReference LeBlanc, Chain, Martin, Bermudez-Humaran, Courau and Langella53, Reference Portincasa, Bonfrate and Vacca54 we could not measure ceacal SCFAs at these earlier time points. Finally, we compared different diets, rather than focusing on single nutrient variations, this approach restricted our ability to draw definitive conclusions regarding the underlying nutrients behind the observed effects. While we elaborated on potential effects of soluble fiber exposure, the differences in fat, protein and carbohydrate profiles between purified and grain-based diets could have also contributed to the observed phenotype in our study.Reference EMvdBaA55–Reference Bouwman, Fernandez-Calleja, van der Stelt, Oosting, Keijer and van Schothorst58
Our findings not only reconfirm the role of maternal diet on offspring growth, development, and programming response to an obesogenic environment in later life, but also strongly highlight the critical impact of standard background diet choice in any study with (early life) nutritional interventions. In line with this, others have previously reported that the prevalent practice of using inappropriate control diets, like employing grain-based diets as controls for refined high-fat diets,Reference Pellizzon and Ricci18, Reference Pellizzon and Ricci59 introduces considerable challenges in isolating the effects that are solely attributable to the dietary intervention from those created by the background diet/s. Further research is crucial to safeguard the quality of preclinical animal research by unraveling the mechanisms that drive the impact of maternal purified diets during early lactation on offspring growth velocity and responsivity to a high-fat diet in adulthood.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S2040174424000436
Acknowledgments
The authors would like to acknowledge the many people who contributed to this study either in the design phase, the execution or in the analysis: Cleo Arkenaar, Martin Balvers, Eline van der Beek, Martijn Breeuwsma, Nicole Buurman, Francina Dijk, Miriam van Dijk, Jessica Freesse, Johanneke van der Harst, Andrea Kodde, Stephan Pouw, Hanil Quirindongo, Noela Schaap, Sudarshan Shetty, Heleen de Weerd, Tjalling Wehkamp, Rachel Thomas and Simon De Neck.
Financial support
This work was supported by Danone Research & Innovation, Utrecht, The Netherlands.
Competing interests
Rakhshandehroo M, Harvey L, Lohr J, Tims S, Schipper L are employees of Danone Research & Innovation, Utrecht, The Netherlands. De Bruin A and Timmer E declare no conflict of interest.
Ethical standard
This study was conducted under an ethical license of the national competent authority (CCD, Centrale Commissie Dierproeven) following a positive advice from an external, independent Animal Ethics Committee (St. DEC consult, Soest, the Netherlands), and all animal procedures were captured in a detailed protocol approved by the Animal Welfare Body – by this process securing full compliance the European Directive 2010/63/EU for the use of animals for scientific purposes.
Open access
