Probiotic Supplementation During Human Pregnancy Affects the Gut Microbiota and Immune Status

Abstract

The consumption of probiotics and fermented foods has been very popular in recent decades. The primary aim of our study was to evaluate the effect of probiotics on the gut microbiota and the changes in inflammatory cytokines after an average of 6.7 weeks of probiotic administration among normal pregnant women. Thirty-two healthy pregnant women at 32 weeks of gestation were recruited and divided into two groups. The probiotic group ingested combined probiotics until after birth. The base characteristics of the probiotics and control groups showed no significant differences. The structure of the fecal microbiota at the genus level varied during the third trimester, and administration of probiotics had no influence on the composition of the fecal microbiota however, many highly abundant taxa and core microbiota at the genus level changed in the probiotic group when compared to the control group. The analysis of cytokines showed that IL-5, IL-6, TNF-α, and GM-CSF had equal levels between the baseline and control groups but were significantly increased after probiotic administration (baseline = control < probiotics). Additionally, levels of IL-1β, IL-2, IL-12, and IFN-γ significantly increased among the three groups (baseline < control < probiotics). This result demonstrated that probiotics helped to shift the anti-inflammatory state to a pro-inflammatory state. The correlation analysis outcome suggested that the relationship between the microbiota and the cytokines was not strain-dependent. The gut microbiota varied during the third trimester. The probiotics demonstrated immunomodulation effects that helped to switch over to a pro-inflammatory immune state in the third trimester, which was important for labor.

Keywords: immunomodulation; interaction network; machine learning; pregnancy; probiotics.

Figures

Figure 1

Flow diagram.

Figure 2

Microbial variations of fecal samples from the third trimester. (A) Species accumulation and rarefaction curves calculated for each group. (B) Relative abundance of bacterial phyla for each group. (C) Alpha-diversity estimated by Simpson indices showed no difference in groups. (D) Principal coordinates analysis (PCoA) based on Manhattan distances with ellipses representing 68% confidence intervals. Baseline vs. control: p < 0.001; baseline vs. probiotics: p < 0.001; control vs. probiotics: p = 0.464. Statistical significance was determined by the Adonis test (Vegan, R script package).

Figure 3

OTUs differentially abundant during pregnancy. Only the OTUs that were assigned to the Greengenes 16S rRNA gene reference database at the genus level and had significant differences were labeled. (A–C) Volcano plot of the OTU abundance demonstrates intragroup differences. The y-axis represents the log2-fold change of relative abundance calculated by EdgeR, and the x-axis is the p-value adjusted by the Benjamini-Hochberg method. CPM, counts per million; prevalence indicates the percentage of participants in which a given OTU is present.

Figure 4

The corresponding bacteria of ingested probiotics varied in each group at the genus level. The data were processed for standardization using the Z-score method, and the y-axis represents the rate of change in relative abundance. lac, Lactobacillus; str, Streptococcus thermophilus; bif, Bifidobacterium.

Figure 5

The core taxa selected by the random forest algorithm. (A–B) The feature importance scores of 12 taxa at the genus level, which can explain the variance among groups (baseline vs. control: 74.5%; baseline vs. probiotics: 80%). (A) Baseline vs. control; (B) baseline vs. probiotics. (C–D) The relative abundance of relevant taxa changes between groups. *p < 0.05. The data were processed for standardization using the Z-score method, and the y-axis represents the rate of change in relative abundance.

Figure 6

Cytokine variations and their interactions in pregnant women in the third trimester. (A) Twelve cytokines vary in blood samples. The first row showed no significant differences in the three groups. In the second row, the cytokine levels of the probiotic group increased, while the cytokine levels in the other two groups remained the same (baseline = control < probiotics). All cytokines increased, but the cytokines in the probiotic group increased more than those in the other groups (baseline < control < probiotics). *p < 0.05; **p < 0.01. The data were processed for standardization using the Z-score method, and the y-axis represents the rate of change in concentration. (B) Interaction network of cytokines in the three groups. The correlation value calculated by Spearman's method was >0.5, and p < 0.05 (adjusted by the Benjamini-Hochberg method) was retained. Edge width represents the correlation value. (C) The eigenvector index of the network demonstrates the variation tendency.

Figure 7

(A–C) High abundance taxa at the genus level correlated with cytokines (calculated by Spearman's method and p-values were adjusted by the Benjamini-Hochberg method. *p < 0.05; **p < 0.01). The average prevalence of groups: baseline: 97%, probiotics: 97%, control: 93%.

Figure 8

The anti-inflammatory/pro-inflammatory stage balance changed after probiotic administration. ↑↑, significantly increased; ↑, mildly increased; —, no change.