Unravelling the gut microbiota of cow’s milk allergic infants, their 1 mothers and grandmothers 2

69 Background: Cow’s Milk Allergy (CMA) is one of the most prevalent food allergies (FA) among infants. 70 Gut microbiota dysbiosis has been related to the development of FA. The primary colonization of the 71 gut microbiota occurs via maternal route. We hypothesized that a longitudinal influence in the 72 composition of the gut microbiota, transmitted from mothers to offspring, could be directly related 73 to CMA development. 74 Methods: 148 faecal samples of 34 CMA and 16 control 0-8 month-old infants and their respective 75 mothers and grandmothers were studied. Gut microbiota was profiled by 16S rRNA gene sequencing 76 using the Illumina MiSeq platform. 16S rRNA sequencing analysis was performed using the DADA2 77 pipeline. Descriptive statistics of the epidemiological variables of the three generations were analysed. 78 Statistical analyses were performed with R 3.4.0 software. 79 Results : Mothers allergy status and smoking habits of mothers and grandmothers were associated to 80 infant CMA. We found that adult gut microbiota is richer and more diverse than that of infants. 81 Relative abundance of the Prevotellaceae family was significantly different between infant groups, 82 and between hydrolysate-fed and formula-fed infants. Finally, the Bray-Curtis distance between 83 members of the same family was independent of the allergy status. Conclusions: Microbiota from allergic children do not differ from non-allergic at the onset of allergy. smoking and were

The human microbiota is a complex ecosystem made up of bacteria, fungi, viruses, archaea and 93 parasites that cohabit on or inside the human body, where bacteria are the most abundant. 1 These 94 microorganisms and their entire set of genes are referred to as the microbiome. Among the microbial 95 ecosystems of the human organism, the most complex and diverse is the one associated with the 96 digestive system, particularly in the gastrointestinal tract (GIT), where the density of microorganisms 97 is the highest of the entire body. The number of bacteria in the GIT, particularly in the section between 98 the colon and the appendix, is around 10 9 per gram of luminal content , and the number of genera 99 swings between 1000 and 3000. 2,3 Faeces are representative of the microbiota composition of the 100 colon segment. This fact and their noninvasive collection method make them the biological sample of 101 choice for the study of the gut microbiota. 4-6 102 Gut microbiota composition changes throughout life. It is thought that 70% of the gut microbiota 103 primary colonization is of maternal origin, 7 and that the first 1,000 days of life, when the body is faced 104 for the first time with external factors, are very important for the development of the intestinal 105 microbiota. 8 Moreover, the development of the gut microbiota in the first years of life correlates with 106 the development and maturation of the intestine and the immune system. From birth, a symbiotic 107 relationship is established between the microbiota and our cells, which evolves over time, adapting 108 to changes. 9 After the first 2-3 years of life, the gut microbiota becomes similar to what it will be for 109 the rest of our life. However, the composition of the gut microbiota is dynamic and dependent on 110 host-associated confounding factors such as age, diet, use of antibiotics, lifestyle, and environmental 111 conditions. 10,11 112 The greatest source of stimulation of the immune system is found in the mucous surfaces of the body, 113 in contact with the external environment. About 70-80% of immune system cells are found in the small 114 and large intestines. 12 The gut microbiota stimulates and modulates the immune system by a DCs-115 mediated immune regulation. Microbes promote the differentiation of regulatory T cells by the 116 activation of DCs present in the mucous surface of the intestine through the Toll-Like Receptor (TLR) 117 pathway. 13 These activated cells produce cytokines that in turn activate naïve T cells or Th0 cells so 118 that they mature towards the corresponding T cell subtype, T helper cells (Th1, Th2, Th17) and 119 regulatory T cells. 13 120 In healthy individuals, all Th cell subpopulations are present in dynamic balance with regulatory T cells. 121 Several studies have shown that the onset of tolerogenic responses to antigens is mediated by the 122 presence of certain bacteria in our GIT. 14-16 Imbalances of microbial communities, known as dysbiosis, 123 have been related to inadequate modulation of the immune system and the development of 124 pathologies not only related to the GIT. Alterations in gut microbiota have been found in people 125 suffering from rhinitis, atopic eczema, asthma or food allergies such as peanut, egg or cow's milk 126 allergy (CMA). 17-21 However, whether the imbalance of the gut microbiota is a cause that triggers the 127 disease or, on the contrary, is a consequence of the disease that alters the bacterial populations and 128 their functionality is still unknown. In this sense, previous work from our group has shown that 129 epithelial barriers are compromised in severe respiratory allergic phenotypes regardless of triggering 130 allergen. 22,23 Moreover, this barrier malfunction is linked to systemic changes. 24 Thus, whether this 131 imbalance in the gut microbiota is a cause of allergy or a consequence of a previously altered state of 132 the mucosa is not yet well defined, and this issue needs to be further addressed. 133 Among food allergies, CMA is one of the most prevalent among children. Tolerance is acquired in 80% 134 of CMA patients by the age of 4; 25,26 despite this, the etiology and pathophysiology of the disease 135 remain unclear. In addition, to what extent alterations of the microbiota of the infant could play a role 136 in this particular allergy is completely unknown. 137 On the other hand, several studies have shown an association between maternal and paternal 138 allergy/asthma and the risk of allergy development in infants. Moreover, most associations are with 139 maternal allergy/asthma. 27-34 One possible explanation to this link could be that the mother's gut 140 microbiota, transmitted to infants, may predispose infants to allergy. However, to our knowledge this 141 has not been studied yet. 142 The aim of this study was to evaluate the intergenerational effect of the gut microbiota in the allergy 143 status of infants over the maternal route within three generations. 144

Study design 147
We performed an intergenerational and observational case-control study approved by the Regional

Epidemiological and intergenerational risk factors for allergy in infants 192
The characteristics of the 50 infants including variables of their mothers and grandmothers are shown 193 in Table 1. Infant cases and controls had similar age (4.93 vs 5.00 months, respectively) and gender 194 distribution, and no significant differences (p>0.05) were found regarding their delivery mode and the 195 use of antibiotics at birth. Having older siblings or pets was not significantly associated (p>0.05) with 196 an increased risk of allergy in infants. A highly significant association was found between type of 197 feeding at the moment of sample collection of infants (p<0.01). Mothers and grandmothers of cases 198 and controls were of similar ages (p>0.05). In addition, it was observed a trend between allergy status 199 of infants and the counterpart of their mothers (p=0.06), but not with that of their grandmothers 200 (p=1.00). Interestingly, smoking status of both mothers (p=0.019) and grandmothers (p=0.077) might 201 be associated with the allergy status of the infant. 202 We also performed a multivariate analysis including the smoking status of mothers, grandmothers and 210 allergy status of the mothers. In this case, only the association between the smoking status of the 211 mother and the risk of having an allergic child remained significant (OR = 0.10, p-value = 0.04) (Table  212 1S). 213 214 Gut microbiome profile differences over three generations 215 The 16S rRNA gene sequencing was performed in samples from infants between 4 and 6 months which 216 had sufficient quantity of DNA, including 11 AG; 32 NA_G; 18 AM; 27 NA_M; 19 AI and 7 CI (n=114 217 out of 148, 77%). This selection allowed to standardize even more the AI and CI groups. 218 Overall, a total of 19,523,010 sequences were generated, out of which 12,955,391 remained after 219 quality and length filtering and chimera removal, resulting in 9,641 ASVs for the gut microbiome. 220 Relative abundance of the bacterial phyla and families found in individuals of the three generations 221 according to their allergic status are shown in Figure 2A and 2B. 222 The faecal microbiome composition was different between adults and infants (PERMANOVA p=0.017). 223 At the phylum level, the adult microbiota was mainly constituted by Firmicutes (60%), Bacteroidetes 224 (15%), Actinobacteria (7%), Verrucomicrobia (4%) and Proteobacteria (3%), whereas in infants, the 225 relative abundance of both Actinobacteria and Proteobacteria phyla was higher (ANCOMII p=7.4e−06 226 for Actinobacteria and p=9.8e−14 for Proteobacteria), representing 30% and 15% of the infant gut 227 microbiota composition. On the contrary, the relative abundance of Firmicutes was lower in infants 228 were decreased compared to CI. However, these differences were not significant ( Figure 3S_A). On 243 the other hand, AI had decreased within-sample bacterial diversity compared to CI (p<0.05) (Figure  244 2C). Moreover, multivariate analysis using PERMANOVA showed that there were statistical 245 compositional differences at bacterial family level in the gut microbiota between AI and CI (Adonis 246 p=0.025) ( Figure 3B). However, the overall microbiome composition of their mothers did not 247 significantly differ ( Figure 2S). Differences regarding abundant bacterial families and ASVs between 248 infant groups were identified using the ANCOM-II method. Interestingly, the relative abundance of 249 Prevotellaceae and Acidaminococcaceae, two of the families with the lowest relative abundances, was 250 significantly different between AI and CI ( Figure 3C). At ASV level, significant differences between the 251 relative abundances for Veillonella parvula, Veillonella dispar, Streptococcus lutetiensis and 252 Enterococcus casseliflavus were detected between AI and CI ( Figure 3S_B & 3S_C). 253 We identified 5 groups according to the feeding regime of infants: breast milk (BM), breast milk 254 together with hydrolysate (BM_H), breast milk together with formula milk (BM_F), hydrolysate (H) 255 and formula milk (F). In relation to the sequenced samples, AI were fed with H, BM or BM_H while CI 256 were fed with F or BM_F. As can be observed in Figure 4A, the relative abundance of the bacterial 257 families changes according to the feeding regime. The most remarkable compositional changes were 258 observed between infants fed with H (all belonging to the AI group) and those with F feeding (all 259 belonging to the CI group) (PERMANOVA p=0.005), finding an increase of Prevotellaceae bacterial 260 family in F group ( Figure 4B-C). At the ASV level, we identified an increase of the relative abundance 261 of Veillonella parvula and Enterococcus casseliflavus in BM_H and BM_F groups compared to F group, 262 respectively, while Streptococcus lutetiensis was decreased in H group compared to F ( Figure 4S Among all the variables considered in this study (see Table 1) smoking was also found to be associated 303 with risk of allergy in infants. Maternal tobacco consumption was significantly associated with a higher 304 prevalence of infant allergy. This association remained significant even after adjusting for the maternal 305 allergy status and the maternal grandmother smoking status, confirming that maternal smoking is an 306 independent risk factor for allergy status in infants. The loss of significance of the maternal allergy 307 status and maternal grandmother smoking status in the multivariate model maybe due to the limited 308 sample size of our study. 309 Maternal smoking during pregnancy is a modifiable environmental risk factor for many diseases like 310 atopic eczema, dermatitis syndrome and bronchial asthma, and has intergenerational and organ-311 specific effects for the lungs as well as inducing epigenetic changes in the foetal allergen-specific 312 immune responses. 47 In addition, recent studies have shown that smoking can alter the vaginal 313 microbiota. 48 314 In our infant multi-factor models, maternal smoking and allergy were identified as significant factors 315 associated with infant allergy, suggesting that tobacco smoke in general is an important 316 determinant. 49 317 In this work, we have identified that Prevotellaceae family was significantly increased in CI and in F fed 318 infants compared to AI and H fed infants, respectively. However, at genus level there was no 319 significant increase within this family. The significant association between the feeding regime and 320 allergy status in infants may be explained by the fact that for most of the infants diagnosed with CMA, 321 the feeding was changed to H either alone or combined with BM (~70%) as a regular clinical practice. 322 In contrast, most of the controls were fed with F. Therefore, each infant's diet was conditioned by 323 their allergy status and it is highly difficult to separate both variables as they are usually linked. To 324 minimize this bias as much as possible, infants who ingested H for more than two weeks were excluded 325 from the study. Likewise, since samples were collected only once, we cannot state that those changes 326 are a consequence of either the allergy status of the infant or their feeding regime. Prevotella together 327 with Bacteroides are the most prevalent genera of the Bacteroidetes phylum within the gut. Prevotella 328 is a large genus with high species diversity and high levels of genome diversity between strains. 329 Previous studies have shown that a decrease of this genus in the lung microbiota is associated to 330 asthma and chronic obstructive pulmonary disease (COPD). 50 In other studies, the increased 331 abundance of Prevotella has been associated to rheumatoid arthritis 51 and inflammatory bowel 332 disease. 52 On the other hand, members of the Prevotella genus have also been associated with 333 beneficial effects such as an improved glucose metabolism 53 and correlation with a plant-rich diet 54 , 334 so the role of this genus is still poorly understood and is probably highly dependent on the specific 335 strains involved. 336 Other potential risk factors, including delivery mode (caesarean or vaginal delivery) and antibiotics 337 consumption were analysed although no significant results were obtained as most of the participants 338 recruited in the study had a vaginal delivery and very few participants received antibiotics at birth. In 339 addition, no significant association was observed between CMA risk and having pets or older siblings, 340 probably due to the limited sample size of our study. 341 Several authors have reported the association between age and the human microbiome. 10,55 However, 342 a longitudinal study including three generations has rarely been performed before. 56 The structure of 343 the gut microbiome community showed differences in diversity and richness when age groups were 344 compared according to their allergic status, showing a statistically significant decrease in infants 345 compared to adults, understood as both mothers and grandmothers (Figure 2). The Actinobacteria 346 and Proteobacteria phyla were highly abundant in infants compared to adults. The Actinobacteria 347 phylum is mainly represented by the Bifidobacteriaceae family and in the case of infants it is 25-30% 348 of the total gut microbiota. 57 349 In addition, we found that microbiota Bray-Curtis distances between infants and their respective 350 should be conducted to confirm them as independent risk factors. Regarding gut microbiota 371 composition, we have confirmed that the microbiota of infants is less diverse than that of adults. 372 Finally, it was not possible to assign the differences we have found to diet or allergy. Taking this into 373 account, prospective studies with longitudinal follow-up in close intervals during the first six months 374 will be optimal to shed light into the causal effect of microbiome differences between infant cases and 375 controls. 376 377 Footnote. Epidemiological characteristics of the participants in the study. "Ever", within the variable smoking p-value). C, Bacterial families with a significantly different abundance between groups using ANCOMII 402 test (q<0.1). * p<0.05, ** p<0.01, ***p< 0.001. 403 404 Figure 5. Beta diversity analysis based on Bray-Curtis distance between members of the same family. 405 T-test was performed for statistical analyses using GraphPad Prism. 406

407
Ethics approval and consent to participate 408 The study was approved by the Regional Ethics Committee for Clinical Research of Hospital 409 Universitario Infantil Niño Jesús in Madrid (R-0004/17) according to the ethical guidelines outlined in 410 the Declaration of Helsinki and its amendments. All participants provided informed consent. 411

Consent for publication 412
Not applicable 413

Availability of data and materials 414
Data sharing not applicable to this article as no datasets were generated or analysed during the 415 current study. 416