Human Gut Microbiome Response to Short-Term Bifidobacterium-Based Probiotic Treatment

Abstract

Gut microbiota is believed to play a crucial role in modulating obesity in humans, and probiotics affecting gut microbiota can alleviate some of the obesity-related health complications. The study was aimed to investigate changes in the composition of the gut microbiome in obese humans due to short-term (2 weeks) treatment of obese patients with a probiotic preparation containing Bifidobacterium longum. Faecal microbiome diversity was studied using the 16S amplicon sequencing by Illumina MiSeq. Bioinformatic analysis showed distribution across 14 phyla (with Firmicutes and Bacteroidetes dominating), 21 class, 125 genera and 973 OTUs. The probiotic treatment decreased relative abundance of Bacteroidetes (Prevotellaceae and Bacteroidaceae), while increasing that of Actinobacteria (Bifidobacteriaceae and Coriobacteriaceae), and Firmicutes (Negativicutes: Veillonellaceae and Clostridia: Peptostreptococcaceae). The probiotic treatment decreased total blood sugar and increased patients’ assessment of their physical and mental health. Thus even the short-term Bifidobacterium-based probiotic treatment brought significant compositional changes in the 16S rRNA gene diversity in faecal bacterial assemblages by increasing beneficial and decreasing pathogenic or opportunistic bacteria; the related shifts in life quality assessment necessitate further research into the causal relationships involved.

Electronic supplementary material

The online version of this article (10.1007/s12088-020-00888-1) contains supplementary material, which is available to authorized users.

Introduction

By now there is no doubt that microbiome’s functioning is crucial for human health. Currently vast research efforts are channelled into modulating the intestinal microbiota to promote health and prevent diseases [1].

The last decade has seen a renaissance in research focusing on Actinobacteria as their role in gastrointestinal homeostasis and fighting systemic diseases became well acknowledged. Bifidobacteria have been incorporated into many food products, such as yogurt, fermented milk and dietary supplements. For decades actinobacteria have been used as beneficial additives in dairy products in many countries, including the former USSR: for instance, the Bifidobacterium-enriched milk products were distributed via the country-wide network of milk-processing stations according to medical prescriptions [2]. Consequently, a lot of research was carried out on beneficial microorganisms, i.e. probiotics, in the 20th century. However, recent advancement in human physiology, biochemistry and microbiology have shifted pathophysiology paradigms, necessitating once more detailed investigation of probiotic effect on human pathological conditions through modulation of gut microbiota diversity.

The studies on genetics, physiology, biochemistry and other aspects of bifidobacteria have been revived, with Bifidobacterium longum as one of the most popular components of probiotic preparations being one of the foci of such research. Increasing incidence of obesity and related health complications prompted research into novel treatment approaches, such as manipulating gut microbiota by probiotics. In this respect microbiome’s dynamic behaviour over time is extremely important to understand [3].

Alleviating obesity per se is obviously a long process. However, some obesity-related complications might be partially alleviated by a short-term probiotic treatment. The aim of the work was to study the effect of a short-term administration of probiotic formulation based on B. longum subsp. longum strain MC-42 on gut bacterial assemblages, blood biochemistry and health status of obese patients.

Materials and Methods

Probiotic Preparation and Treatment

The commercially available dairy probiotic product Biovestin^® (Bio-Vesta, Russia) contained B. longum subsp. longum strain MC-42 [4]. Patients consumed 50 ml of product (10⁷ CFU/ml) every day during 2 weeks, otherwise following their conventional diet.

Participants and Faecal Sample Collection

Subjects for the trial were selected at random from a pool of obese patients being treated for obesity-related health complications in the Central Clinical Hospital (SB RAS, Novosibirsk, Russia). Thirteen patients were recruited, but the trial was fully completed by five women and three men (between 23 and 57 years old with a BMI of 30.0–37.0 kg/m²). Exclusion criteria were the following: (1) neuroendocrine form of obesity; (2) cancer of any organs; (3) severe somatic pathology at decompensation stage; (4) acute infection within 2 weeks prior to probiotic treatment (PT); (5) exacerbation of a chronic disease; (6) antibiotic therapy within the last 10–14 days prior to enrolment; (7) functional food consumption 1 month prior to enrolment. All patients were duly informed and gave their consent to the study.

Faecal samples were collected prior to the probiotic treatment (PT) and within 5–7 days after it by patients at home into 10 ml sterile faecal specimen containers and stored frozen at approximately − 20 °C. Samples were transferred to the laboratory within 1 week of collection and stored at − 80 °C until used for DNA extraction.

The protocol of the study was approved by the ethics committee of the Institute of Chemical Biology and Fundamental Medicine (SB RAS, Novosibirsk, 630090, Russia). All clinical aspects of the study were supervised by a physician.

Blood Analyses and Questionnaire

Blood samples were collected from the patients prior to PT and within 5−7 days after the end of the treatment. Blood was analysed for total sugar and cholesterol content, and for the content of high density lipoproteins, triglycerides, bilirubin, urea and activity of aspartate and alanine transaminases.

The patients enrolled in the trial also completed SF-36 questionnaire [5] to assess their health status before and 5−7 days after PT.

DNA Extraction and Sequencing

DNA was extracted from 50 to 100 mg of thawed patient faecal samples using MetaHIT protocol. The bead-beating was performed using TissueLyser II (Qiagen, Germany), for 10 min at 30 Hz. No further purification of the DNA was needed. The quality of the DNA was assessed using agarose gel electrophoresis. The 16S DNA region was amplified with the primer pair V3–V4 combined with Illumina adapter sequences. All PCR reactions used 25 ng of faecal DNA as template and were performed in triplicates. A total of 200 ng PCR product from each sample was pooled together and purified through MinElute Gel Extraction Kit (Qiagen, Germany). The obtained libraries were sequenced with 2 × 300 bp paired-ends reagents on MiSeq (Illumina, USA) in SB RAS Genomics Core Facility (ICBFM SB RAS, Novosibirsk, Russia). The read data were deposited in GenBank under the sequence read archive (SRA) accession number SRP194142.

Bioinformatic and Statistical Analyses

Raw sequences were analyzed with UPARSE pipeline [6] using Usearch v.10.0. The UPARSE pipeline included merging of paired reads; read quality filtering; length trimming; merging of identical reads (dereplication); discarding singleton reads; removing chimeras and operational taxonomic unit (OTU) clustering using the UPARSE-OTU algorithm. The OTU sequences were assigned a taxonomy using the SINTAX [7] and 16S RDP training set v.16.

The OTUs datasets were analyzed by individual rarefaction (graphs are not shown) with the help of the PAST software [8]: the number of bacterial OTUs detected, reaching plateau with increasing number of sequences, showed that the sampling effort was close to saturation for all samples, thus being enough to compare diversity. Taxonomic structure of the assemblages, i.e. a collection of different species at one site at one time, was estimated by the ratio of the number of taxon-specific sequence reads to the total number of sequence reads, i.e. by the relative abundance of taxa, expressed as percentage. Comparison of relative abundances of different bacterial taxa in faecal samples of the control and probiotic-treated group was carried out by the Wilcoxon nonparametric test for repeated measures using the Statistica v.13.3 software (Statsoft, USA). The α- and β-biodiversity indices were calculated with the help of PAST 3.19 software [8]. The data are shown as median, minimum and maximum values.

Results

After quality filtering and chimera removal a total of 973 different OTUs were identified at 97% sequence identity level, of which the overwhelming majority (970) was Bacteria, the rest representing the Euryarchaeota phylum of the Archaea domain.

Over all samples 14 bacterial phyla were identified, containing 21 class, 27 orders, 53 families and 125 genera. Most of the bacterial OTUs represented the Firmcutes phylum (697 OTUs, or 72% of the total bacterial OTU number), with Bacteroidetes and Actinobacteria being the second and the third most OTU-rich phyla with 11 and 6%, respectively. Clostridia was the OTU-richest class, accounting for 60% of the total OTU richness, with Bacteroidia and Actinobacteria contributing 11 and 6%, respectively.

The number of dominant OTUs, i.e. OTUs contributing ≥ 1% into the total sequence number in a sample, was 91, i.e. less than 10% of the total number of OTUs. They represented 5 phyla, 9 classes, 11 orders, 18 families and 32 genera, i.e. much less taxonomic richness as compared to the total list of identified OTUs. The number of dominant OTUs per sample varied from 9 to 29. Averaged over the samples, collected before the PT, 41 of 73 dominant OTUs (56%) represented Firmicutes phylum, 24 OTUs (33%) represented Bacteroidetes phylum, and 4 OTUs (5%) belonged to Actinobacteria phylum. After the PT 48 OTUs of 77 dominant ones (62%) were found to represent the Firmicutes phylum, while 16 (21%) and 7 (9%) OTUs represented Bacteroidetes and Actinobacteria phyla, respectively. Thus more than half of the of the dominant OTUs’ list were ascribed to the Firmicutes phylum.

The number of dominant OTUs, ascribed to Clostridia class, comprised 45% of the total number of OTUs, identified in samples before the PT, and 51% after it, whereas the number of OTUs, representing Bacteroidia class decreased from 33 to 21%. Thus these classes drastically prevailed on the lists of dominant OTUs both before and after the PT.

The PT did not affect the relative abundance of the Firmicutes phylum representatives, while decreasing the Bacteroidetes percentage almost 2 times and increasing the Actinobacteria phylum percentage by 6.4 times (Table 1).

Table 1.

Relative abundance (%) of phylum-specific 16S rRNA gene sequences in faecal samples collected before and after the probiotic treatment (median and fluctuation range)

Identified phylum	Before		After		p- value
	Median	Min–max	Median	Min–max
Firmicutes	38.9	29.7–76.6	53.3	23.2–83..4	0.67
Bacteroidetes	58.5	6.5–69.5	19.9	3.7–58.6	0.03
Actinobacteria	2.3	0.1–10.1	7.4	3.3–67.8	0.01
Proteobacteria	0.7	0.1–3.7	1.2	0.2–10.9	0.40
Verrucomicrobia	0.002	0.00–0.87	0.010	0.0–3.4	0.69
unclassified Bacteria	0.04	0.00–2.41	0.4	0.0–2.3	0.21
Others	0.006	0.00–0.40	0.030	0.00–0.21	0.78

At the class level all the representatives of the latter two phyla belonged to Bacteroidia and Actinobacteria classes, respectively (Fig. 1), but the Firmicutes phylum was represented by four identified classes, of which Negativicutes class increased twofold.

Fig. 1.

Relative abundance (%) of class-specific 16S rRNA gene amplicon sequences in human faecal samples collected before and after the probiotic treatment (median; box shows 75–25% quartile range, while the lines indicate the fluctuation range, and the violin shows the probability density of the data). Classes and differential p-values based on Wilcoxon test: 1—Clostridia (p = 0.67), 2—Bacteroidia (p = 0.04), 3—Actinobacteria p = 0.01), 4—Negativicutes (p = 0.04), 5—Others (p = 0.21), 6—Gammaproteobacteria (p = 0.26), 7—Bacilli (p = 0.57), 8—Verrucomicrobia (p = 0.69)

At the order level Bacteroidia class was represented by Bacteroidales, while Actinobacteria was represented by Bifidobacteriales and Coriobacteriales, both responding to PT by increasing relative abundance (Fig. 2). The Selenomonadales order was the sole one representing the Negativicutes class.

Fig. 2.

Relative abundance (%) of order-specific 16S rRNA gene amplicon sequences in human faecal samples collected before and after probiotic treatment (median; box shows 75–25% quartile range, while the lines indicate the fluctuation range and the violin shows the probability density of the data). Classes and differential p-values based on Wilcoxon test: 1—Clostridiales (p = 0.67), 2—Bacteroidales (p = 0.03), 3—Bifidobacteriales (p = 0.03), 4—Selenomonadales (p = 0.04), 5—Coriobacteriales (p = 0.01), 6—Others (p = 0.16), 7—Erysipelotrichales (p = 0.58), 8—Aeromanadales (p = 0.58), 9—unclassified Bacteria, 10—Lactobacillales (p = 0.58), 11—Verrucomicrobiales (p = 0.69)

More interesting pattern emerged while analyzing the effect of probiotic on the relative abundance of family-specific sequences (Table 2) as one family of Firmicutes (Clostridiads, Clostridiales), namely Peptostreptococcaceae, was found to increase its percentage in the sequence assemblage almost three times due to PT. Among Bacteroidales the Prevotellaceae family decreased almost two times, while the decreased abundance of Bacteroidaceae was not statistically significant.

Table 2.

Relative abundance (%) of family-specific 16S rRNA gene amplicon sequences in human faecal samples collected before and after probiotic treatment

Identified families	Before		After		p value
	Median	Min–max	Median	Min–max
Lachnospiraceae	D007	10.4–27.0	11.0	4.7–38.1	0.40
Ruminococcaceae	11.9	11.0–41.9	24.7	9.1–39.1	0.78
Prevotellaceae	54.4	1.4–68.8	16.3	1.0–58.0	0.05
Bacteroidaceae	1.6	0.1–21.3	1.9	0.1–7.7	0.21
Bifidobacteriaceae	0.8	0.0–8.9	4.6	0.0–65.7	0.03
Coriobacteriaceae	1.1	0.1–2.5	3.6	1.5–13.5	0.01
Veillonellaceae	2.1	0.0–7.4	3.8	0.1–14.8	0.04
unclassified Clostridiales	1.8	0.5–6.3	2.0	0.2–8.3	0.33
Eubacteriaceae	1.0	0.3–2.3	1.2	0.4–10.9	0.21
Streptococcaceae	0.1	0.0–3.6	0.4	0.0–1.2	0.67
Erysipelotrichaceae	0.5	0.1–2.2	0.7	0.1–1.3	0.58
Enterobacteriaceae	0.2	0.0–2.1	0.1	0.0–1.9	0.89
Porphyromonadaceae	0.3	0.0–2.6	0.8	0.1–1.2	0.78
Acidaminococcaceae	0.1	0.0–2.7	0.3	0.0–5.5	0.50
Peptostreptococcaceae	0.1	0.0–1.6	1.1	0.0–3.6	0.02
Lactobacillaceae	1 × 10−2	0.0–0.1	2 × 10−2	0.0–0.6	0.87
Clostridiaceae_1	8 × 10−2	0.1–0.8	0.1	0.0–3.7	0.58
Rikenellaceae	9 × 10−2	0.0–3.6	0.4	0.0–1.9	0.78
Verrucomicrobiaceae	2 × 10−3	0.0–0.8	1 × 10−2	0.0–3.4	0.68
Others	0.9	0.3–6.4	2.7	0.7–6.3	0.21

At the OTU level seven OTUs increased their relative abundance due to the PT (Table 3), with three of them belonging to Actinobacteria and the other four to Firmicutes/Clostridia. Three OTUs were found to decrease their relative abundance (in total about tenfold) due to the PT (Table 3); all of them represented Firmicutes/Clostridia, being rare or minor components of the assemblage.

Table 3.

Changes (at p ≤ 0.05 level) in relative abundance (%) of OTUs in human faecal samples collected before and after the probiotic treatment

OTU	Identified OTUs	Before		After		p value
		Median	Min–max	Median	Min–max
1	Bifidobacterium adolescentis	0.53	0.00–4.7	1.71	0.00–50.9	0.036
4	Collinsella aerofaciens	0.51	0.1–1.0	2.07	0.7–11.0	0.012
15	Romboutsia sedimentorum	0.05	0.0–0.7	1.14	0.0–3.2	0.012
24	un. Clostridiales	0.06	0.0–0.5	0.75	0.0–1.9	0.046
95	Slackia isoflavoniconver-tens	0.00	0.0–0.2	0.15	0.0–0.4	0.043
154	Oscillibacter sp.	0.003	0.00–0.02	0.012	0.00–0.13	0.028
805	un. Ruminococcaceae	0.02	0.00–0.11	0.12	0.00–0.22	0.017
	Increased relative abundance (7 OTUs)	1.14		5.94
8	Ruminococcus bromii	0.07	0.0–7.0	0.02	0.0–0.9	0.046
263	un. Lachnospiraceae	0.15	0.0–0.5	0.00	0.0–0.1	0.018
721	Roseburia sp.	0.22	0.1–1.0	0.09	0.0–0.3	0.036
	Decreased relative abundance (3 OTUs)	0.44		0.11

un. unclassified

If we increase the statistical significance threshold to 0.05 ≤ p ≤ 0.10, we get additional list of bacteria with their relative abundance changed due to PT (Supplementary Table 2). Eight of these OTUs increased summarily their percentage two times, the main contribution into the increase done by a Bifidobacterium sp., Dialister succinatiphilus and Eubacterium desmolans. Bacterial α-diversity indices did not change due to the PT (Supplementary Table 2). The Jaccard index of similarity between faecal bacterial assemblages of a patient before and after the PT was calculated for each patient; it averaged 0.58, ranging 0.52−0.64.

After the PT, the total sugar level was slightly decreased, but no differences observed in the other blood parameters (Supplementary Table 3). After the PT the patients’ assessment of their health status improved both on physical and mental health scales (Supplementary Table 4).

Discussion

The predominance of Firmicutes and Bacteroidetes in faecal microbiomes of the patients agrees with the general view of the structure of human gut microbiome. These phyla were the core ones both before and after the Bifidobacterium-based PT, despite their drastically changed abundance. The fact that even the short duration (just 2 weeks) of the PT was found to result in the structural changes at the phylum level, as Bacteroidetes decreased and Actinobacteria increased their abundance almost threefold, seems quite promising. Nearly the entire decrease in Bacteroidetes phylum was due to the Prevotellaceae family.

The PT changed the structure of faecal microbiome at OTUs level as well, though changed OTUs were minor, rare or extremely rare. This finding complies with the idea that small adjustments to shifts in environments are achieved via low-abundance OTUs. Such OTUs in the human gut microbiome may have systemic interactions, mostly synergistic ones, with potentially important consequences for the microbiome performance within a host organism.

Mostly beneficial OTUs were found to increase their abundance after the PT: for instance, Bifidobacterium adolescentis can produce short-chain fatty acids and vitamins, has been used as a beneficial microorganism to promote health and is also known for degrading resistant starch in human colon. Another actinobacterium with increased abundance, namely Collinsella aerofaciens, is a common member of healthy human colon microbiota, producing lactate and short-chain fatty acids, albeit opposing patterns of its association with various disorders have been reported [9]. Other actinobacteria with PT-related increase in the faecal microbiome, i.e. Slackia isoflavoniconvertens, can convert dietary flavonoids, thus exerting health-promoting effects in a human host [10]. Some clostridia were found to become more abundant, for instance, Oscillibacter sp. The Oscillibacter genus can comprise small percentage of the human gastrointestinal microbiome and seems to be able to induce remission in ulcerative colitis after faecal microbiota transplantation [11]; a fat-rich diet was shown to decrease Oscillibacter abundance and correlate with insulin resistance [12]. The type strain of this genus (Oscillibacter valericigenes) has valeric acid as its main metabolic end product, which structurally resembles γ-aminobutiric acid and was shown to be negatively correlated with depression [13]. Thus it seems that increased abundance of Oscillibacter can be beneficial for human health. Another clostridium with PT-increased presence in the faecal microbiome, i.e. Romboutsia sedimentorum, was found to ferment glucose to acetic acid, ethanol, iso-butanoic acid and iso-valeric acid [14], thus benefiting human health, but information about its physiology and biochemistry is scarce. Some unclassified below the family level (Ruminococcaceae) OTUs, that increased due to the PT, are problematic for interpreting in terms of whether they are beneficial for human health.

For the purpose of catching a broader range of specific bacteria, that changed their presence in the faecal microbiome due to the PT in our study, we did not use any fdr correction; instead we also used the 0.10 statistical threshold, which resulted in additional set of five OTUs with increased abundance due to the PT. Among the latter, Eubacterium desmolans is a butyrate-producing bacterium, contributing to steroid and bile acid metabolism and hence generally beneficial for human health, although its association with some neurological disorders was also reported [15]. In a similarly, i.e. apparently controversial way, Dialister succinatiphilus was positively correlated with spondyloarthritis [16]; and oppositely, its abundance was found to be negatively correlated with many other disorders as well [17]. Dialister genus was decreased in the faecal microbiomes of autistic subjects [18], and in long-living subjects [19], and increased in patients with irritable bowel syndrome [20]; so the unequivocal conclusion about the genus’s effect on human health is varying. As for Asaccharobacter celatus, another PT-increased actinobacterium in the faecal microbiome, it can participate in the transformation of dietary flavonoids in the human gut and equol production [21], so can be beneficial for the host even at the very low abundance.

Several minor OTUs became much less abundant or practically disappear from the faecal microbiome after the PT. For some of them it is difficult to infer any host benefits because of the controversial information reported so far. For instance, Ruminococcus bromii, that decreased its abundance due to the PT, can degrade resistant starch in human colon, which is generally regarded as beneficial for human health; it is also known to have decreased abundance in people with some gastrointestinal disorders, cystic fibrosis, acquired immunodeficiency syndrome; however, it’s relative abundance was increased in patients with Parkinson’s disease and ectopic eczema [9].

At the same statistical significance threshold (0.05 ≤ p ≤ 0.10) Prevotella stercorea reduced its abundance twofold. As its increased abundance can be associated with disease, e.g. untreated HIV infection [22], the change may be expected to be beneficial.

Despite some PT-related changes in faecal 16S rRNA gene assemblage structure, the biodiversity indices did not change over the short-term (2 weeks) treatment. The Shannon biodiversity index (2.2–2.3) seems low, yet falling within the range reported earlier [23].

By now there is little doubt that gut microbiota is a key factor impacting our emotional and behavioural health. Notably, our study revealed that the assessment of their physical and mental performance by patients themselves showed improvement already after the short-term PT. The data about the effect of PT on physical performance so far have been controversial: recently some studies concluded that athletes may benefit from probiotics [24], while others [25] concluded that probiotic supplementation did not provide ergogenic effect. The effect may be strain-, duration and disorder-related.

Conclusions

Based on our results, we concluded that short-term Bifidobacterium-based probiotic treatment caused significant compositional changes in the 16S rRNA gene sequence diversity in faecal bacterial assemblages of obese patients by increasing beneficial and decreasing pathogenic or opportunistic bacteria. These changes were accompanied by shifts in the quality of life assessment, thus necessitating further research targeting specific bacterial taxa that changed due to the probiotic treatment to get into details of the mechanisms involved.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author Contributions

Conceptualization, G.S. and A.K.; methodology, T.A.; software, M.K.; validation, N.N.; formal analysis, M.K.; investigation, T.A. and G.S.; resources, V.V.; data curation, A.T. and N.N.; writing

—

original draft preparation, N.N.; writing−review and editing, V.V.; supervision, V.V.; project administration, M.K.; and funding acquisition, A.K. and V.V.

Funding

This research was funded by the Russian Ministry of Science and Higher Education, the projects (2018−2020) Nos. 0309-2018-0004 and AAAA-A17-117020210021-7.

Availability of Data and Material

The read data were deposited in GenBank under the sequence read archive (SRA) accession No. SRP194142.

Compliance with Ethical Standards

Conflict of interest

Author A.K. is an employee of Bio-Vesta LLC. Any opinions or scientific interpretations expressed in this manuscript are those of the author and do not necessarily reflect the position or policy of Bio-Vesta LLC. Otherwise the authors declare that they have no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Consent to participate

All patients were duly informed and gave their consent to the study.