FUT2 gene variants and reported respiratory and gastrointestinal illnesses during infancy

Abstract

Background

FUT2 controls the production on digestive and respiratory epithelia of histo-blood group antigens involved in the attachment of pathogens. The aim of our study was to relate FUT2 variants to reported gastrointestinal and respiratory illnesses in infancy.

Methods

In the UK Southampton Women’s Survey, FUT2 genetic variants (rs601338, rs602662) were genotyped in 1831 infants and related to infant illnesses adjusting for sex, breastfeeding duration and potential confounders.

Results

For FUT2 SNP rs601338, the risk ratios for ≥1 bout of diarrhoea from ages 6-12 and 12-24 months per additional risk (G) allele were 1.23 (95%CI 1.08-1.4, p=0.002) and 1.41 (1.24-1.61 p=1.7x10^-7), respectively; the risk ratio for ≥1 diagnoses of lower respiratory illnesses (pneumonia or bronchiolitis) from 12-24 months per additional G allele was 2.66 (1.64-4.3, p=0.00007). Similar associations were found between rs602662 and gastrointestinal and respiratory illnesses, due to the high linkage disequilibrium with rs601338 (R²=0.92). Longer breastfeeding duration predicted a lower risk of diarrhoea independent of infant FUT2 genotype.

Conclusions

We confirmed that FUT2 G-alleles are associated with a higher risk of infant gastrointestinal illnesses and identified novel associations with respiratory illnesses. FUT2 locus variants need consideration in future studies of infantile gastrointestinal and respiratory illnesses.

Introduction

Diarrhoea and respiratory illnesses are major causes of morbidity and mortality in young children less than five years of age, with over 2 million deaths per year[1]. Preventive measures and improved disease management have reduced mortality, but it remains particularly substantial in low or middle-income countries[1], accounting for 30% of early childhood deaths worldwide[2]. Mortality is especially high in young children with multiple episodes per year[3, 4]. In developed countries, gastroenteritis cases are mainly due to rotavirus (in unvaccinated children), norovirus, adenovirus and Salmonella, with the first two accounting for half and a third of the cases, respectively, and the last two approximately 10%. In young children, rotavirus and norovirus infections are highly prevalent, with a peak age between 3 months and two years[5]. For acute lower respiratory infection in young children, human respiratory syncytial virus (RSV) is the most common causal pathogen[6]. Breastfeeding is proposed as a protective factor against both diarrhoea and respiratory infections, suggesting that early cessation of lactation and the benefits delivered to the infants might contribute to the avertable infections [7, 8].

Host characteristics are known to impact the susceptibility to early childhood infections. Recent studies have reported that genetic polymorphisms affecting the production of histo-blood group antigens (HBGA), which can act as attachment sites for specific pathogens, are associated with incidence of diarrhoea [9–11] . The FUT2 (FUT2 [MIM: +182100]) gene encodes for α1,2-fucosyltransferase 2, which catalyzes the addition of a fucose to the H-type 1 precursor, generating the H-type antigen in saliva and on digestive and respiratory epithelia, producing the secretor phenotype. Inactivating polymorphisms in FUT2 give rise to the non-secretor status, characterized by the absence of H-type antigen in mucosal tissues and secretions[12]. The objective of our study was to assess the genetic risk that FUT2 contributes to infant morbidities in a UK mother-offspring cohort study (the Southampton Women’s Survey (SWS)); given the presence of H-type antigen on respiratory epithelia, our particular focus was on respiratory illnesses which have not previously been related to FUT2 variants in population-based studies.

Methods

Study Population

Offspring of participants in the SWS were studied [13]. Between 1998 and 2007, 3158 infants were born and data were collected on them both during the pregnancy and after the birth. Infant health outcomes (chest wheezing, cough, pneumonia or bronchiolitis, croup, diarrhoea, vomiting and ear infections) were assessed by nurse-led questionnaires 6, 12 and 24 months after birth (Questions asked by research nurses on infant health outcomes are listed in Supplementary Table 2). Detailed information about breastfeeding was also obtained by nurse-led questionnaires administered at 6, 12 and 24 months after the birth of the child. Maternal smoking during pregnancy was assessed by questionnaire in early and late pregnancy. Visits were placed at short intervals (6, 12 and 24 months) and trained research nurses familiar with infant health outcomes administered all questionnaires to minimise recall bias.

Patient Consent and Ethical Approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The study was approved by the Southampton and South West Hampshire Research Ethics Committee approved all procedures (276/97, 307/97). Informed consent was obtained from all individual participants included in the study.

Research Reporting Checklist

The STROBE guideline checklist (http://www.strobe-statement.org/index.php?id=strobe-home) was used in the preparation of this manuscript.

DNA extraction

A 5-10 cm segment was cut from the mid portion of each cord, immediately following delivery, flushed with saline to remove fetal blood, flash-frozen in liquid nitrogen and stored at −80°C until required for DNA isolation. Genomic DNA was prepared from umbilical cord and cord blood by a standard high salt method [14], and from adipose tissue using the QIAamp DNA mini kit (Qiagen, Germany).

Genotyping of FUT2 variants

1831 offspring and their mothers from the SWS cohort with infant health outcomes and available DNA were genotyped for 2 polymorphisms, rs601338 and rs602662, in the FUT2 gene located on chromosome 19. The genomic region containing rs601338 and rs602662 was amplified by PCR following manufacturer’s guidelines (HotStarTaq Plus DNA Polymerase, Qiagen, 203605) (Supplementary Table 1). SNP genotyping was carried out on a Pyromark MD.

Statistical Analysis

Statistical procedures were performed in Stata version 13 (StataCorp, Texas, USA) and SPSS version 21 (IBM, Armonk, New York). The small number of offspring (n=83) whose mothers were recorded as having non-white ethnicity were excluded from analysis to ensure genetic homogeneity in the study population. Genotypes were coded according to the additive model (0, 1 or 2 copies of the G allele). Infant health outcomes were coded as dichotomous variables indicating whether the child had suffered one or more episodes of the health outcome or not

Binary regressions were completed with each genotype individually used as a predictor variable for each infant health outcome, adjusting for child’s sex, parity, mode of delivery (vaginal or Caesarean), maternal SES and length of breastfeeding split into 3 groups (never, <= 3 months, > 3 months). Analyses of infant diarrhoea, from 6 to 12 and 12 to 24 months (which do not commonly result in a doctor’s prescription of antibiotics) were additionally adjusted for infant antibiotic use to reduce the likelihood of any observed associations resulting from antibiotic use rather than an infective episode. Infant antibiotic use from 6 to 12 months was also examined to see if it was associated with episodes of diarrhoea from 12 to 24 months. Regressions showing association between offspring FUT2 SNPs and health outcomes were additionally adjusted for matching maternal genotype where available, and for preterm birth (<37 weeks). Results from binary regression are expressed as risk ratios per unit increase in number of risk alleles. A Benjamini-Hochberg procedure with a False Discovery Rate (FDR) of 10% was applied to account for multiple testing.

Results

Supplementary Table 3 shows genotype distributions for white and other ethnicities in the SWS cohort. Chi squared tests show that there were associations between rs602662 and ethnicity (p=0.004) and between rs601338 and ethnicity (p=0.025). Expected values show that whites had fewer g/g genotypes and more a/a genotypes than other ethnicities. Therefore analysis was restricted to white infants only due to insufficient numbers of other ethnicities in the SWS cohort.

Both SNPs, rs601338 and rs602662 in FUT2, were in Hardy-Weinberg equilibrium in the infants with white ethnicity (chi squared p-value>0.05). rs601338 had a minor allele frequency of 48.7% (A) and rs602662 had a minor allele frequency of 48.6% (G). rs602662 is in high Linkage Disequilibrium (LD) with rs601338 in Caucasian populations (R²=0.92; Ensembl).

Analysis of infant health outcomes was based on 1831 SWS singleton births with available phenotype data from birth to 24 months and genotype for at least one SNP. Table 1 shows summary statistics for the study sample from the SWS cohort (n=1831). A comparison of infants included (n=1831) and not included (n=1327) in the study is shown in Supplementary Table 4.

Table 1. Summary statistics for the study sample.

Mother	N	Percent
Maternal smoking during pregnancy status	1798
Smoking during pregnancy	300	16.7
Non-smoker during pregnancy	1498	83.3
Parity	1829
First born child	901	49.3
Second or higher birth	928	50.7
Mode of delivery	1811
Vaginal delivery	1395	77
Caesarean Section	416	23
Child	N	Percent
Male/Female	951/ 880	51.9 /48.1
Breastfeeding status	1759
Never tried	322	18.3
<=3 months	739	42
>3 months	698	39.7
Wheezing: birth to 6 months	1814	24.4
Nocturnal cough: birth to 6 months	1816	15.3
Pneumonia or bronchiolitis: birth to 6 months	1817	4.3
Croup: birth to 6 months	1817	3.7
Vomiting: birth to 6 months	1817	9.7
Diarrhoea: birth to 6 months	1817	16.8
Ear infection: birth to 6 months	1817	5.7
Wheezing: 6 to 12 months	1743	30.2
Nocturnal cough: 6 to 12 months	1742	21.9
Pneumonia or bronchiolitis: 6 to 12 months	1743	4
Croup: 6 to 12 months	1743	6.7
Vomiting: 6 to 12 months	1740	22.4
Diarrhoea: 6 to 12 months	1742	36
Ear infection: 6 to 12 months	1742	19.3
Atopic eczema 6 to 12 months	1743	10.7
Atopy 6 to 12 months	1466	11.1
Wheezing: 12 to 24 months	1681	26.4
Nocturnal cough: 12 to 24 months	1680	22.5
Pneumonia or bronchiolitis: 12 to 24 months	1682	2.9
Croup: 12 to 24 months	1682	11.3
Vomiting: 12 to 24 months	1682	24.5
Diarrhoea: 12 to 24 months	1681	34.9
Ear infection: 12 to 24 months	1681	26

(N: Number)

Figure 1 shows risk ratios and 95% Confidence Intervals for each copy of the risk allele (G) for rs601338 and all infant health outcomes tested.

Fig 1.

Risk Ratios (RR) and 95% Confidence Intervals (CI) for each copy of the risk allele (G) for rs601338 and all health outcomes tested. The dashed line shows the risk ratio of null effect; confidence intervals crossing this line shows that the outcome is not associated with FUT2 SNP rs601338.

For FUT2 SNP rs601338, infants who possessed the G allele were more likely to suffer infant vomiting, diarrhoea, pneumonia/bronchiolitis and nocturnal cough than infants possessing A alleles. Thus, FUT2 SNP rs601338 showed a significant association with one or more episodes of diarrhoea from birth to 6 months (risk ratio 1.20 (95%CI 1.03-1.4), p=0.017), 6 to 12 months (risk ratio 1.15 (1.04-1.26), p=0.004) and 12 to 24 months (risk ratio 1.29 (1.17-1.43), p=3.27x10^-7). rs601338 was associated with one or more episodes of vomiting from 6 to 12 months (risk ratio 1.23 (1.08-1.4), p=0.002) and from 12 to 24 months (risk ratio 1.41 (1.24-1.61 p=1.7x10^-7). rs601338 was also associated with one or more diagnoses of lower respiratory illness (pneumonia or bronchiolitis) between 12 to 24 months (risk ratio 2.66 (1.64-4.3), p=0.00007) and with nocturnal cough between 12 to 24 months (risk ratio 1.16 (1.02-1.33), p=0.029). No significant associations were observed between rs601338 and croup, wheezing, ear infections, atopic eczema or atopy from birth to 24 months (Figure 1). Similar results were observed for rs602662 as the two SNPs are in high Linkage Disequilibrium (R²=0.92) (Supplementary Figure 1).

Antibiotic use from 6 to 12 months was associated with increased episodes of diarrhoea from 12 to 24 months (p=0.006).

Length of breastfeeding (>3 months compared to never breastfed) was also a significant predictor of infant diarrhoea and pneumonia/ bronchiolitis from 12 to 24 months (p=0.042). Longer breastfeeding duration lowered the risk of diarrhoea and pneumonia/bronchiolitis independent of infant FUT2 genotype. Maternal FUT2 genotype was not found to be a significant predictor of infant health outcomes when infant FUT2 genotype was included in the model. Preterm birth was not a significant predictor of infant health outcomes with a Benjamini and Hochberg FDR of 10%.

Discussion

Our study showed significant associations of FUT2 SNPs rs601338 and rs602662 with the risks of reported respiratory and diarrhoeal illnesses, as well as vomiting and nocturnal cough from birth to 24 months in infants in a population based study. The risk of diarrhoeal infections was significantly higher in the infants possessing the G allele compared to those with the A alleles in FUT2 SNPs. The result was consistent from birth to 6 months, to 12 months and up to 24 months. Prevalence rates for one or more reported bouts of diarrhoeal lasting more than 2 days were highest between 6 and 12 months of age, being reported for 36% of infants. Reported pneumonia or bronchiolitis was less frequent in this population, ranging from 2.9% between birth and age 6 months to 4.3% between ages 12 and 24 months. G alleles on both FUT2 SNPs tested conferred higher risk of pneumonia or bronchiolitis in this population. Other comorbidities potentially linked to infectious diseases, like vomiting lasting more than 2 days and nocturnal cough, were also associated with these FUT2 SNPs at 6 to 12 months and at 24 months respectively. The direction of the association was the same, with a lower risk of illness in the carriers of G alleles.

Our main findings are in line with existing literature on diarrhoeal infections and host genetic susceptibility. FUT2 genetic variations have been associated with susceptibility to diarrheal infections in several studies [9, 10, 15, 16]. SNP rs601338 (W154X) was identified as the possible causal variant in recent meta-analysis of birth cohorts studies consisting of >5,000 young children [9]. The strongest association was found for at “at least one episode of diarrhoea” reported around age 1 year but strong results were also reported for other time-points. Similarly, in our study, a reduced risk of diarrheal episodes with rs601338 SNP was consistently reported from 6 to 24 months but with a stronger association for diarrheal episodes from 12 to 24 months. A community-based birth study from Ecuador assessed the associations of norovirus gastroenteritis with secretor status at first 3 years of life and identified that secretor children were more susceptible to norovirus GII.4 genotype infections, whereas non-secretors were susceptible to non-GII.4 infections. [17]. Another study in China of children with acute diarrhoea was in line in these findings, although the detection of cases of norovirus GII.4 and GII.3- associated diarrheal infections in non-secretors showed that these children are not fully protected [18]. Our data did not include information on the specific diarrheal aetiology or identification of pathogens; it is however known that norovirus and rotavirus infections are the commonest among children and they are influenced by FUT2 genotype [10, 16, 19]. Previous studies suggest that most probably a direct association between secretor phenotype or FUT2 genotypes and incidence of norovirus or rotavirus-identified infection outbreaks exists, with non-secretor or carriers of FUT2 non-functional mutations having protection against these infections[10, 16, 19] (Figure 2). FUT2 genetic variations or secretor status, as parts of the innate host genetic factors have been also shown to interact with intestinal microbiota [20–22]. In light of this, a recent study aimed to investigate the interactions of among FUT2 genotypes, gut microbiota and viral infections typically linked to norovirus and rotavirus [23]. The study however could not confirm significant differences in intestinal microbial composition between FUT2 genotype groups, suggesting that these associations may be specific to the disease context or environmental exposures [24, 25]. Whether there is a direct link of association between risk of diarrheal infections and FUT2 genotypes, or if intestinal microbiota are interacting with host genetics to modulate risk of infections, is not well understood.

Fig 2.

The proposed mechanism of interaction between H epitope defined by FUT2 and pathogens. FUT2 inactivating mutations (rs601338 (W154X) and rs1047781 (I140F)) leading to non-secretor phenotype are associated with reduced susceptibility to the risk of infections.

Our novel findings on reported respiratory illnesses are in the same direction as in the recent study by Taylor et al., 2017, in which secretor patients with non-cystic fibrosis bronchiectasis had an increased susceptibility to Pseudomonas aeruginosa-dominated airway infection compared to non-secretors among other disease outcomes [26]. It would be important to confirm these observations in future studies of children with known aetiology of the respiratory infections and other comorbidities.

FUT2 is highly polymorphic with different inactivating mutations in particular ethnic groups, but with this genetic heterogeneity giving rise to similar proportions of the non-secretor phenotype; for example, approximately 20% of Europeans and 22% of East Asians are non-secretors [27]. American populations of Hispanic ancestry are less likely to be non-secretors compared to non-Hispanic populations, highlighting the low prevalence of the FUT2 inactivating mutations in Amerindian populations [15]. Non-secretors are characterized by an absence of H antigen in salivary secretions and on mucosal surfaces of the body. In European and African populations, the most common inactivating variant is rs601338 (W154X), whereas in Asian populations the missense SNP rs1047781 (I140F) leads to a truncated protein with weak bioactivity [28]. G risk allele in our study on rs601338 encodes the secretor phenotype, whereas the A allele the non-secretor. The second FUT2 variant included in our analysis, rs602662, is a missense SNP (G258S) in high LD with rs601338, as shown in our results. This SNP was included in our study, as it has been shown to lead independently of rs601338 to the expression of a FUT2 enzyme with very low activity and thus might influence the expression of H antigen as well[29]. HBGA groups like the H antigen may mediate the attachment of pathogens, leading to infection. In the case of rotavirus, the cell attachment viral spike protein VP8* can recognize A-type HBGA, perhaps leading to increased susceptibility of secretors compared to non-secretors to specific rotavirus strains[30, 31].

Our study showed an association of FUT2 SNPs with reported pneumonia or bronchiolitis and nocturnal cough from age 12 to 24 months. A single previous study reported that secretors had increased susceptibility to respiratory infections of influenza viruses A and B, rhinoviruses, respiratory syncytial viruses and echoviruses, however, further evidence for the role of FUT2 genetic polymorphisms in respiratory illnesses is scarce, with none in population-based studies [32]. We acknowledge with caution the findings on reported pneumonia or bronchiolitis, as the number of episodes in our population within the specific time periods were low (Table 1) but a clear association was nonetheless observed.

Breastfeeding was identified as an independent protective exposure in our results, raising the possibility that some of the factors in human milk may reduce the risk of infections independently. The FUT2 enzyme also controls the production of specific human milk oligosaccharides by adding a fucose molecule to lactose in an α1,2-fucosylation activity to generate more complex oligosaccharides such as 2’-fucolsyllactose (2’FL) and lacto-N-fucosylpentaose I (LNFPI). These human milk oligosaccharides are present in the milk of all secretors mothers and are absent in the milk of non-secretor mothers. Previous studies have demonstrated a specific protective effect of these oligosaccharides against diarrheal infections in a cohort of Mexican mothers and infants[33]. In that study, where infants were predominantly breastfed, it was postulated that the presence of these indigested oligosaccharides could compete with the attachment of pathogens to the intestinal epithelial surfaces by inhibiting their colonization and subsequent infection. Using this paradigm, a hypothesis could be proposed that presence of these α1,2-fucosylated human milk oligosaccharides provided by a secretor lactating mother to her secretor infant might attenuate the increased risk that these infants inherently have to specific strains of viruses (like rotavirus or norovirus), considering that the glycan structures present in both human milk oligosaccharides and the epithelial epitopes are similar. Our study examined the association of maternal secretor status in relation with the risk of infections but no independent effect was found. However, many of the population of mothers and infants we studied were not exclusively breastfed and samples were not available to analyze the presence of oligosaccharides in the mother’s breast milk. We might also expect differences in the causal pathogens between our study and the previous Mexican study[33]. Future studies will need to assess this hypothesis by determining both infant secretor status and sequencing for FUT2 variants, as well as analyzing the presence of α1,2-fucosylated human milk oligosaccharides in a predominantly breast fed population.

The main strength of our study was the relatively large population, providing statistical power to examine a number of associations with FUT2 genotypes. The main limitations are first, that information was not available on the aetiology of the illnesses, and available data on both reported illnesses and genotyping information were available only for a part and not for the entire SWS cohort[13]. Despite the large size of our population, we also did not have the power to analyse subpopulations with different ethnicity. Infant outcomes were self-reported by the mothers through questionnaires administered by research nurses, and could possibly be affected by recall bias; this was however minimised by short intervals between visits and by trained research nurses administering the questionnaires. Despite this lack of phenotypic specificity we identified significant associations, consistent through time and in the same direction. Second, we could not accurately measure the level of exposure to pathogens, because our study could assess neither the level of exposure to these pathogens in the population, nor variations in hygiene between the FUT2 subgroups analysed.

Conclusion

Our study has highlighted the consistent association of FUT2 variants with risk of diarrhoea in infants, as well as describing novel associations with respiratory illnesses related and comorbidities in a large population of infants. Host genetic susceptibility to infections needs to be accounted in future epidemiological and interventional studies investigating these outcomes, alongside studying the protective role of early life factors such as breastfeeding.

Abbreviation

FDR

False Discovery Rate

2’FL

2’-fucosyllactose

FUT2

α1,2-Fucosyltransferase 2

HBGA

Histo-blood group antigens

HMO

Human Milk Oligosaccharide

HWE

Hardy-Weinberg Equilibrium

LD

Linkage Disequilibrium

LNFPI

Lacto-N-fucosylpentaose I

MAF

Minor Allele Frequency

RSV

Respiratory syncytial virus

SNP

Single nucleotide polymorphism

SWS

Southampton Women’s Survey