Prenatal stress linked to alterations in the gut microbiota of infants

Mother’s chronic prenatal psychological distress and elevated hair cortisol concentrations are associated with gut microbiota composition of the infant, according to a new publication from the FinnBrain research project of the University of Turku, Finland.

The results help to better understand how prenatal stress can be connected to infant growth and development. The study has been published in the esteemed Psychoneuroendocrinology journal.

Prenatal stress can be associated with infant growth and development. However, the mechanisms underlying this association are not yet fully understood.

“We were able to show that maternal chronic psychological distress and elevated hair cortisol concentrations during pregnancy are associated with infant gut microbiota composition but not diversity,” says Doctoral Candidate, Doctor Anna Aatsinki.

The study used hair cortisol analysis which enabled measuring the concentration averages of stress hormone cortisol over several months.

In addition, the symptoms of the mother were assessed three times during pregnancy. The infant gut microbiota was analysed early at the age of 2.5 months with next generation sequencing.

Previously, similar studies have focused on animals and two have been smaller human studies making this data consisting of 399 mothers and their infants the largest in the world so far. The received results provide significant new information on the phenomenon. In addition, this study was able to confirm previously made observations.

Studying the Role of Microbes as Mediators of Stress

Both Proteobacteria and Lactobacillus are common infant gut microbes.

“We discovered, for instance, that mother’s chronic prenatal psychological distress was linked to increased abundances of Proteobacteria genera in infant microbiota. In addition, chronic psychological symptoms were connected to decreased abundances of Akkermansia genera which is considered to promote health at least in adults,” summarises Aatsinki.

According to Aatsinki, it was also interesting that low cortisol concentrations were associated with increased abundances of Lactobacillus in infant gut microbiota. Lactobacillus bacteria are considered to promote health.

However, Proteobacteria also contain species that are able to cause inflammation in the body. Proteobacteria can also be associated with the child’s disease risk later in life. Therefore, researchers consider it important to study how the observed changes are connected to later child development.

“Our study does not explain the cause-effect relationship, or whether prenatal psychological stress is linked to differences in microbial metabolic products or e.g. in immune system function. In other words, important questions still need to be answered,” notes Aatsinki.

The study is part of the FinnBrain research project and its gut-brain axis sub-project. The sub-project led by Docent, Child and Adolescent Psychiatrist Linnea Karlsson studies how prenatal stress affects infant microbiota development and how infant gut microbes affect later brain development.

The FinnBrain research project of the University of Turku studies the combined influence of environmental and genetic factors on the development of children. Over 4,000 families participate in the research project and they are followed from infancy long into adulthood.

---

Although the underlying mechanisms remain unclear, an increasing number of studies link maternal prenatal stress to infant physical development and health, and psychological functioning and behavior. Stress during pregnancy predisposes to premature birth and low birth weight (Mulder et al., 2002; Beydoun and Saftlas, 2008), eczema (Sausenthaler et al., 2009), asthma (Cookson et al., 2009), and respiratory, general and skin illnesses (Beijers et al., 2010).

Regarding psychological functioning and behavior, children of prenatally stressed mothers often show more impulsivity, anxiety problems, ADHD symptoms, and worse cognitive and psychomotor development (Beydoun and Saftlas, 2008).

Recently, the development of the infant gut microbiota has been put forward as a possible factor underlying the links between maternal prenatal stress and infant development (Beijers et al., 2014).

Rhesus monkey infants whose mothers had experienced stress during late pregnancy, in the form of repeated exposure to an acoustic startle, had lower levels of Bifidobacteria and Lactobacilli and more diarrheic symptoms than the infants of non-stressed mothers (Bailey et al., 2004).

Also, in adult mice a social stressor (i.e., social disruption) provoked a decrease in the relative abundance of the genus Bacteroides together with an increase in the relative abundance of the genus Clostridium (Bailey et al., 2011). The goal of the present study is to investigate the relation between maternal prenatal stress (i.e., reported stress and cortisol concentrations) and the development of infant intestinal microbiota and health in the first 110 days of life in humans.

The intestinal microbiota are known to play an impor- tant role in the maturation of an infant’s gastrointestinal tract, immunity, metabolism, as well as the hypothalamic- pituitary-adrenal system (Sudo et al., 2004; Dimmitt et al., 2010; Bäckhed, 2011).

An aberrant acquisition of intestinal bacteria or a reduced complexity of the microbiota may delay immune maturation or alter the development of the immune system and stress responses (Sudo et al., 2004; Adlerberth and Wold, 2009; Sekirov et al., 2010). Bacterial colonization of the infant gut is thought to begin in utero (Gosalbes et al., 2013), and to accelerate dramatically during and after delivery, and during the first months of life (Palmer et al., 2007; Fallani et al., 2010).

Microbes from the mother and, to a lesser extent, of the environment are thought to be the first colonizers of the infant’s gut (Tannock et al., 1990; Gosalbes et al., 2013). After the initial establishment of the intestinal microbiota during the first year of life, the microbiota begins to stabilize to a unique individual composition, continuing to develop gradually throughout childhood and adolescence.

To what extent the early colonization dictates later development and finally the stable adult composition, is currently unknown. Due to the intimate interaction between the developing intestinal

microbiota and the immune system, the early-life devel- opment of the intestinal microbiota may have long-lasting consequences (Bäckhed, 2011).
Distortions in the intestinal microbiota are associated with a wide range of diseases, including the risk of diarrheal illness, food allergy, inflammatory diseases (atopic diseases and inflammatory bowel disease), irritable bowel syndrome, obesity, and diabetes (Sekirov et al., 2010).

Furthermore, as is the case with irritable bowel syndrome, gut-related diseases can develop or worsen during stressful periods (O’Mahony et al., 2009; De Palma et al., 2014). This may be due to the bidirectional communication between the central nervous system (CNS) and the gut (brain-gut axis; Dinan and Cryan, 2012), where both the autonomic nervous system (ANS) and the hypothalamic pituitary adrenal (HPA) axis play important roles (Rhee et al., 2009).

When the HPA axis is activated in reaction to stress, cortisol is produced as an end product. In rats, experimentally increased cor- tisone levels in pregnant females resulted in lower levels of total bacteria and gram negatives in the intestine of the pups (Schiffrin et al., 1993).

This suggests that cortisone may influence the maternal microbiota, and thereby the transmission of bacteria to offspring. In humans it is as yet unknown if maternal prenatal psychological stress and corti- sol concentrations are related to the infant gut microbiota.

The goal of the present human study is to prospectively investigate the relation between maternal prenatal stress and the development of infant intestinal microbiota and health in the first 110 days of life. A limitation of the Rhesus monkey study of Bailey et al. (2004) is that the intestinal microbiota analyses were carried out with traditional culturing approaches and were not able to show the more complex microbiota signatures.

The present study avoids this limitation by using a high-throughput phylogenetic microarray
(Rajilic-Stojanovic et al., 2009).

Results

Correlations between prenatal stress and anxiety variables

None of the reported stress and anxiety variables was signifi- cantly correlated with the cortisol variables, indicating that cortisol and reported stress represent independent meas- ures of stress.

The questionnaire-based stress variables were weakly to moderately correlated (r ranging from 0.22 to 0.74), and the cortisol concentrations measured at differ- ent times of day were weakly to strongly correlated with each other (r ranging from 0.15 to 0.87).

Comparison of the different stress indicators

Based on the number of significantly associated bacteria, the sum of the stress questionnaire scores and the cortisol con- centration measured at noon were most strongly associated with the infant microbiota (Fig. 1).

Of the individual questionnaire scores, the fear of a handicapped child (PRAQ2) had the strongest association with the infant microbiota, but not as strong as the sum of the questionnaire scores.

We therefore selected the sum of questionnaire scores and the 12:00 cortisol concentration as measures of reported stress and cortisol concentration for further analyses. There was a weak positive correlation between the reported stress and the cortisol concentration (r = 0.25, p < 0.001).

Both indicators were significantly associated with the relative abundances of over 60% of the bacterial genus-level groups at one or more time points during the first four months of the infants’ life. A major part (78%) of the microbial groups were associated with either reported stress or cortisol concentration.

The magnitude of the associations between the total infant microbiota and the prenatal stress (sum of questionnaires) and noon cortisol concentration were similar (Fig. 2).

The effects of both prenatal stress indicators appeared comparable to or higher than those of breastfeeding, and were usually higher than the effects of postnatal maternal stress. The effects of prenatal stress on the infant microbiota were modest over the first month, peaked at 80 days, and were still clearly evident at 110 days.

As both stress indicators had similar associations with the microbiota, we combined the reported stress and the cortisol measure, forming a ‘prenatal cumulative stress index’: low reported stress + low cortisol concentration = low cumulative stress; low reported stress + high cortisol concentration, or high reported stress + low cortisol concentration = moderate cumulative stress; high reported stress + high cortisol concentration = high cumulative stress.

See Table 1 for the descriptive statistics of these three groups.
Based on the temporal dynamics and the associations with the two chosen stress indicators (i.e., sum of the questionnaire scores and noon cortisol), we grouped the stress-associated bacterial genera into clusters within which the bacteria behaved uniformly (Table 2).

Prenatal cumulative stress is a major driver of inter-individual variation in infant microbiota

The major source of inter-individual variation in the infant microbiota was the ratio between a group of pro- teobacteria (Escherichia, Enterobacter, Serratia; referred to as bacterial group PRO1), and a group of lactic acid bacteria (Lactobacillus, Lactococcus, Aerococcus; group LAB) and Actinobacteria (Bifidobacterium, Collinsella, Eggerthella; group ACT1) (Fig. 3).

There was a negative correlation between the relative abundances of PRO1 and LAB (r = 0.23, p < 0.001), which corresponded to the first principal coordinate (PC 1), and between PRO1 and ACT1 (r = 0.40, p < 0.0001), which corresponded to the second PC (PC2) (Fig. 3).

The infants in the low cumulative stress group were characterized by high relative abundance of LAB (69%, 377%, 305%, and 39% higher in the low cumulative stress, as compared to the high cumulative stress group at ages 7, 14, 28, and 110 days, respectively) and ACT1 (23%, 29%, 66%, 53% higher in the low cumulative stress group at ages 7, 14, 80, and 110 days).

The infants in the high cumulative stress group tended to localize in the Escherichia—Enterobacter- end of the microbiota gradient (Fig. 3), with 853%, 256%, 1244%, and 699% higher abundances in the high cumulative stress group at ages 14, 28, 80, and 110 days, respectively.

The sum of the first two principal coordinates (indicating a high relative abundance of PRO1 and low relative abundance of LAB and ACT1) was strongly associated with prenatal cumulative stress (Fig. 3).

Infants in the high cumulative stress group, with the combined effect of high maternal prenatal reported stress and cortisol concentration, had thehighest summed PC scores, and those in the low cumulative stress group had the lowest scores. Infants with only one prenatal cumulative stress factor, either high cortisol concentration or high reported stress, had an intermediate microbiota, suggesting a relatively linear association between cumulative prenatal stress and infant microbiota.

Temporal dynamics of the infant microbiota

We selected the low and the high cumulative stress groups and the bacterial groups most strongly associated with prenatal stress to illustrate the differences in the temporal dynamics (Fig. 4).

In the low prenatal cumulative stress group, the overall diversity of the infants’ microbiota decreased during the first four months of life, and was associated with the gradual establishment of dominant bacteria (mainly Bifidobacterium) (Fig. 4, blue lines; Fig. 5). How- ever, the diversities within Actinobacteria, Proteobacteria, and Clostridia, the most abundant groups in the infant microbiota, increased with time (Fig. 4, blue lines).

During the first month, the intestinal microbiota of the infants with low prenatal cumulative stress was characterized by a high abundance of streptococci and PRO1, which were grad- ually replaced by ACT1, LAB, Clostridium spp. (CLO), and Proteobacteria Group 2 (PRO2) (including low-abundance Proteobacteria, such as Sutterella; see Table 2). Similar patterns were evident at the phylum level (Fig. 5A).

In the high cumulative stress group, the temporal devel- opment was different (Fig. 4, red lines). The overall diversity was consistently higher, due to more even relative abun- dances of different bacterial groups (Fig. 5), but the diversities within Actinobacteria and Proteobacteria were lower than in the low stress group.

Furthermore, the rel- ative abundance of PRO1 (and the total abundance of Proteobacteria, Fig. 5) was higher, and correspondingly the abundances of ACT1, ACT2, LAB, CLO and PRO2 were lower.

The abundance of Akkermansia also differed significantly between the groups, as it declined dramatically in the high cumulative stress group after the first month and remained low thereafter (Fig. 4, red lines). The total abundance of Bacilli remained relatively high throughout the study period, whereas it declined considerably in the low cumulative stress group (Fig. 5).

Although we did not have sufficient power to properly investigate the potential influence of breastfeeding interacting with prenatal stress, we checked the results also separately for the breastfed and non-breastfed infants (data not shown). Compared to the observed association of prenatal stress and the infant microbiota, breastfeeding had a minor influence on the microbiota.

Also, the microbiota association with stress was comparable between breastfed and non-breastfed infants. Therefore, we conclude that in this cohort, breastfeeding did not confound the interpreta- tion of the results.

Prenatal cumulative stress predisposes to gastrointestinal and allergic symptoms

Gastrointestinal symptoms (e.g., diarrhea, gastroenteritis, presumed infection and constipation) were more prevalent during the first three months of life in the high cumula- tive stress group (38%), as compared to the low cumulative stress group (22%).

By the age of three months, 43% of the infants in the high cumulative stress group, and none in the low cumulative stress group, had shown allergic reactions. Note that due to the small sample sizes, these differences in symptom frequency were non-significant based on x²– test (p > 0.05).

The health differences between the groups appeared attributable to the differences in the intestinal microbiota. The infants with gastrointestinal symptoms had lower (albeit non-significantly) relative abundances of LAB (on average 0.5% of total microbiota) and Akkermansia (0.1%) than the infants without gastrointestinal symptoms (2% and 0.5%, respectively).

Infants with allergic reactions by the age of three months had consistently lower abun dances of LAB (0.5% vs. 2%, non-significant difference) and ACT1 (15% vs. 60%, p < 0.001), lower abundance of Akker- mansia (only during the first month of life: 0.7% vs. 2%, p < 0.05), and higher abundances of PRO1 (30—50% vs. <10%, p < 0.001) than the infants without allergic reactions.

---

Source:
University of Turku