Consumption of artificially sweetened beverages during pregnancy impacts infant gut microbiota and body mass index

Artificial sweetener consumption by pregnant women has been associated with an increased risk of infant obesity, but the underlying mechanisms are unknown. We aimed to determine if maternal consumption of artificially sweetened beverages (ASB) during pregnancy is associated with modifications of infant gut bacterial community composition during the first year of life, and whether these alterations are linked with infant body mass index (BMI) at one year of age. This research included 100 infants from the prospective Canadian CHILD Cohort Study, selected based on maternal ASB consumption during pregnancy (50 non-consumers and 50 daily consumers). We identified four microbiome clusters, of which two recapitulated the maturation trajectory of the infant gut bacterial communities from immature to mature and two deviated from this trajectory. Maternal ASB consumption was associated with the depletion of several Bacteroides sp. and higher infant BMI. As we face an unprecedented rise in childhood obesity, future studies should evaluate the causal role of gut microbiota in the association between maternal ASB consumption, infant development and metabolism, and body composition.


INTRODUCTION 38
Childhood obesity in the United States increased from 5 to 18.5 percent between 1978 and 39 2016 1 , magnifying the risk of cardiometabolic disease and mental health disorders later in life 2 . 40 Recent work from the CHILD Cohort Study showed that maternal consumption of artificially 41 sweetened beverages (ASB) during pregnancy is associated with higher infant body mass index 42 (BMI) at one year of age 3 . Importantly, this association was independent of key obesity risk 43 factors, such as maternal BMI, smoking, poor diet, diabetes, short breastfeeding duration, and 44 earlier introduction of solid food 3 . Similar associations have been reported in several other 45 prospective birth cohorts 4 , but the underlying mechanism has not been studied. 46 The gastrointestinal tract, a key site for host metabolic regulation 5,6 , is colonized by a vast 47 community of microbes including bacteria, viruses, and micro-eukaryotes 7 . The gut microbiome 48 is highly heterogeneous during infancy, characterized by colonization patterns 8-10 that are 49 influenced by the maternal microbiome 11,12 , method of birth 13-15 , infant nutrition (breast milk or 50 formula) [16][17][18] , and antibiotic treatment 14,19 . Simultaneously, important aspects of metabolic 51 development occur during this period of life, many of which rely on interactions between 52 microbes and host cells 20 . Recent studies in mice show that artificial sweetener consumption 53 during pregnancy predisposes offspring to increased weight gain through behavioral (i.e. 54 preference for sweet foods, appetite increase) and physiological mechanisms (i.e. stimulation of 55 intestinal sugar absorption, increased postnatal weight gain, altered lipid profiles, 56 downregulation of hepatic detoxification, and increased insulin resistance) [21][22][23][24] demonstrated that artificial sweetener consumption in adult mice directly impacts gut 58 microbiome composition and function, leading to an increase in host glucose intolerance. More 59 recently, Stichelen et al. 24 addressed gestational exposure to artificial sweeteners, finding 60 characterize the gut microbiome, stool samples were acquired at three and 12 months of age for a 84 total of 200 samples. This study was approved by the University of Calgary Conjoint Health 85 Research Ethics Board (CHREB) and ethics committees at the Hospital for Sick Children, and 86 the Universities of Manitoba, Alberta, and British Columbia. Written informed consent was 87 obtained from mothers during enrollment to the CHILD Study. 88 89

Maternal diet in pregnancy 90
Maternal dietary assessment in pregnancy has previously been described 3 . Briefly, a food 91 frequency questionnaire (FFQ) was completed during the second or third trimester and ASB 92 consumption was evaluated using reports of "diet soft drinks or pop" (i.e. soda) 93 (serving = 12 oz / one can) and "artificial sweetener added to tea or coffee" (serving = 1 packet). 94 Other dietary variables included: sugar-sweetened beverages, Healthy Eating Index (HEI) total 95 score (see eMethods), added sugar and total energy intake. 96 97 Infant BMI 98 BMI was measured by CHILD staff to the nearest 0.1 kg around one year of age (mean = 12.0 99 months ± 0.8 [sd]) and height to the nearest 0.1 cm. Age-and sex-specific BMI-for-age z-scores 100 were calculated following the World Health Organization reference 29 . 101 102 Other variables 103 The following variables were considered in univariable analyses (see eMethods): (1) infant's sex, 104 age at sample collection, breastfeeding duration (BF duration; months), breastfeeding status at 105 three months (BF at 3M; yes or no), diet at three and six months (Diet at 3M and Diet at 6M; 106 both defined in 8 categories allocated based on the presence in the infant's diet of breastfeeding, 107 formula, and solids), solids at three and six months (Solids at 3M and Solids at 6M), formula 108 feeding at three months (FF at 3M), number of antibiotic treatments received from six to twelve 109 months (Child 6-12 abx), and secretor status (determined from the single nucleotide 110 polymorphism rs601338 in the FUT2 gene); (2)  in four unique clusters (Figure 1-2 and eFigure 1). 133 The distribution of variables as well as the variation in bacterial richness (Chao 1), alpha-134 diversity (Shannon index), and community evenness (Shannon index / log n (species richness)) 135 across the DMM clusters were examined by non-parametric Kruskal-Wallis tests followed by 136 post-hoc Dunn tests or generalized linear models (glm) with a binomial/logistic distribution. To 137 explore the changes in taxonomical community structure at a fine scale, we tested for significant 138 differences in the relative abundance of the 10 most dominant bacterial genera across clusters 139 using non-parametric Kruskal-Wallis tests followed by post-hoc Dunn tests with Benjamin-140 Holmes False Discovery Rate (FDR) correction. To account for potential heteroskedasticity in 141 bacterial community dispersion between groups and avoid the loss of information through 142 rarefaction 35 , we performed a variance stabilizing transformation 35,36 prior to any statistical tests 143 on beta-diversity. To select variables that could be drivers of infant gut bacterial community 144 structure, we tested for correlations between our variables and community scores on the Principal 145 Component Analysis (PCoA) ordination axes in univariable models (envfit function of vegan 37 ). 146 The relative influence of the significant drivers of gut bacterial community structure was then 147 assessed statistically in multivariate models using a Permutational Multivariate Analysis Of 148 Variance (PERMANOVA; adonis function of vegan 37 ) with 999 permutations and visualized 149 using PCoAs based on Bray-Curtis dissimilarities. We used DESeq2 to test for differentially 150 abundant bacterial taxa according to maternal ASB consumption on the 100 most relatively 151 abundant bacterial taxa to limit spurious significance driven by very rare ASVs. Finally, we used 152 linear models on the three-and twelve-months-old samples to test for the influence of maternal 153 ASB consumption and microbial ordination axes (PCoA1 and PCoA2) on infant BMI z-score. 154 The full model's formula was the following: 155 [ Infant BMI ~ ASB + PCoA1 + PCoA2 ] 156 All analyses and graphs were computed in R version 3.6.1 (R Development Core Team; 157 http://www.R-project.org). Compared to the other three clusters, cluster 1 showed a higher proportion of exclusive 171 breastfeeding. Cluster 3 included a higher proportion of mothers receiving antibiotics, infants 172 born by C-section and formula feeding ( Figure 1). However, there was no difference in maternal 173 ASB consumption between clusters, suggesting that this exposure did not influence the 174 compositional differences that drove cluster classification ( Figure 1F). In addition, the clusters 175 did not differ in terms of maternal sugar intake, gestational diabetes, age, parity, ethnicity, 176 education, antibiotics, study site, infant antibiotics, or infant or mother secretor status. 177

178
Relative influence of ASB on microbial community structure 179 Envfit analysis (univariable models) identified thirteen variables as significant drivers of gut 180 bacterial beta-diversity from which we selected eight non-redundant variables to build our 181 models: infant age, maternal intrapartum antibiotics, maternal ethnicity, birth mode, 182 breastfeeding status at three months, presence of older siblings, infant secretor status, and 183 maternal ASB consumption ( Figure 3A and eFigure 2). Considering the complete dataset, the 184 significant predictors were infant age, maternal ethnicity, intrapartum antibiotics, and birth 185 mode. The same four variables, plus breastfeeding status at 3 months, were tested in a 186 PERMANOVA (multivariable model), altogether explaining 14.2% of community variance 187 (Table 1). Maternal ASB consumption was a significant predictor of infant gut bacterial 188 composition only in the multivariable model (R 2 = 0.7%; Table 1). Birth mode (vaginal vs. C-189 section) had also a significant influence on community composition (R 2 =0.8%), but to a lesser 190 extent than infant age (R 2 = 7.3%) and mother's ethnicity (R 2 = 2.5%; Table 1). 191 Next, we repeated the beta-diversity analyses separately within each of the 4 clusters. Envfit 192 univariable models identified distinct drivers for each cluster ( Figure 3A). Interestingly, the 193 drivers of beta-diversity in cluster 1 (only 3M samples) were mainly maternal factors (i.e. birth 194 mode, mother's ethnicity, intrapartum antibiotics) whereas the drivers of cluster 4 (mostly 12M) 195 were infant factors (infant's secretor status, breastfeeding at three months, and infant age ( Figure  196 3A). Cluster 2 was the only cluster in which maternal ASB consumption was associated with 197 beta-diversity (R 2 = 3.2%), and this association was confirmed by the univariable ( Figure 3A, 198 eFigure 2) and multivariable (Table 1)

Association of ASB and the microbiome with infant BMI at one-year-old 210
Finally, using a multivariable linear model on the complete dataset, we tested the association of 211 maternal ASB consumption and microbial community composition with infant BMI z-score at 212 one year of age. Our results confirmed that daily maternal ASB consumption is associated with 213 higher infant BMI (ß-estimate = 0.42, 95%CI 0.03:0.80, P = 0.037; Table 2), and showed that 214 BMI was associated with the microbiome composition at 12 months (PCoA1 axis; ß-estimate = -215 0.71, 95%CI -1.40:-0.01, P = 0.048; Table 2) but not at three months (not shown). These results 216 suggest that features of PCoA1 (i.e. lower relative abundance of Bacteroidetes and 217 Faecalibacterium, and higher relative abundance of Escherichia, Klebsiella,Bifidobacterium,218 Haemophilus, Clostridium, and Veillonella; eFigure 3) are inversely associated with infant BMI. 219

DISCUSSION 220
In defining links between maternal ASB consumption and infant BMI, we provide new evidence 221 suggesting that maternal consumption of ASB during pregnancy (1) influences the establishment 222 of the infant gut microbiome, particularly in infants diverging from what has previously been 223 described as the typical microbiome maturation trajectory (Table 1, Figure 3A); and (2) is 224 associated with an increase in infant BMI at one-year-old (Table 2). To our knowledge, this is 225 the first human study to report the impact of maternal consumption of ASB on the infant gut 226 microbiome, and its potential influence on infant BMI. In light of recent data showing that ASB 227 can drive dysregulation of energy metabolism in mice through changes in the gut 228 microbiome 24,25,38,39 , our study suggests that infants exposed to ASB through their mothers may 229 be at higher risk of shifts in microbial community structure related to early-life predisposition to 230 metabolic diseases 40,41 . 231 In our study, broad shifts in bacterial community structure were significantly associated 232 with infant BMI at one-year-old. We also identified 9 bacterial taxa from Bacteroides sp. that 233 were enriched (3 ASVs) or depleted (6 ASVs) at high levels of maternal ASB consumption, 234 suggesting a mechanism of influence on infant weight gain involving specific taxa of the gut 235 microbiome. The taxa Akkermansia municiphila and genus Bacteroides have previously been 236 identified by various studies to be respectively decreased and enriched as a consequence of ASB 237 consumption 25,38,39,42 . Our results differ from previous findings for A. municiphila and suggest 238 that Bacteroides patterns of enrichment or depletion might be species-or strain-specific, 239 warranting further research with deeper resolution. 240 As reported by Bian et al. 38,39 in two studies with adult mice, and by Nettleton et al. 43  neotame altered the alpha-and beta-diversity of mice gut microbiome, and led to a decrease in 249 butyrate synthetic genes and changes to the fecal short chain fatty acids cluster. Overall, 250 accumulating evidence suggests that the alterations of host gut bacterial community structure 251 through the consumption of ASB is reflected in bacterial and host metabolic gene clusters, which 252 might explain the increase in weight gain. Based on this evidence and our current results, we 253 hypothesize that gestational exposure to ASB impacts infant gut bacterial communities either 254 indirectly through disruption of vertical transmission of the maternal microbiome, or directly 255 through lactation during breastfeeding. However, our study is underpowered to definitively 256 assess whether gut microbiome mediate the relationship between maternal ASB and infant BMI. 257 Additional work including functional evidence from metagenomics and metabolomics will 258 determine if the bacterial taxa and compositional changes associated with high maternal ASB 259 consumption in our study are causally implicated in energy metabolism dysregulation and infant 260 body composition. 261 Overall, our study validates previous findings 3 that maternal consumption of artificial 262 sweeteners is associated with a higher BMI at one-year-old, and provides unique and timely 263 evidence that the infant gut microbiome could play a role in this effect, especially for susceptible 264 infants displaying a disrupted maturation trajectory (reduced alpha-diversity and species 265 richness) of their gut microbiome and a high relative abundance of Bacteroides. Our study also 266 confirms recent descriptions of infant microbiome development and confirms the influence of 267 several known determinants of the gut microbiome during the first year of life 11-14,16,17,19 268 including maternal antibiotics, breastfeeding, birth mode and ethnicity. 269 The major strength of our study is the combination of state-of-the-art community typing Although we were unable to explore these mechanisms in our study, they will be addressed by 283 future work in the CHILD cohort involving metagenomics of infant stool and metabolomics of 284 infant stool, urine and serum. 285

CONCLUSION 286
In this study, we characterized the infant gut microbiome of 100 infants and found evidence that 287 maternal ASB consumption during pregnancy might have unforeseen effects on infant gut 288 microbiome development and body mass index during the first year of life. As we face an 289 unprecedented rise in childhood obesity and related metabolic diseases, further research is 290 warranted to understand the impact of artificial sweeteners on gut microbiome and weight gain, 291 especially during critical periods of early development. 292  Minuscule letters indicate statistical differences between clusters from post-hoc generalized 463 linear model (glm) with a binomial/logistic distribution. "BF at 3M" stands for "breastfeeding at 464 three months" and "FF at 3M" for "formula feeding at three months". Aside from maternal ASB 465 consumption (F), only the variables that showed a statistical difference in distribution between 466 clusters are presented. No differences were found for maternal age, ethnicity, education, 467 diabetes; study site, household pets, siblings, or introduction of solid foods at 3 or 6 months.

NS NS
Total R 2 (%) 15.1 13.6 9.1 14.0 10.3 NS P > 0.05, * P < 0.05, ** P < 0.01, *** P < 0.001  Relative abundance (%)  community structure across all data and each cluster subset. Horizontal bars show the amount of 5 variance (R 2 ) explained by each covariate in the model as determined by envfit. Asterisk denotes 6 the significant covariates in each data subset (P<0.05). All 32 variables considered in this study 7 are shown in Figure S2. In this figure, ASB represents artificially sweetened beverages and BF at 8 3M represents infant's breastfeeding status at three months (see methodology). (B) 14 bacterial 9 taxa identified as significant features associated with maternal consumption of ASB by DESeq2. 10