Different immunization methods lead to altered gut flora and varied responses to Mycobacterium tuberculosis infection in mice

Introduction: Vaccination is an essential means for prevention of tuberculosis infection, but the effects of various vaccines on the intestinal flora of mice and their response to Mycobacterium tuberculosis (Mtb) infection remain poorly understood. Methodology: In this study, two different vaccinations − ESAT6 and ESAT6 + TLR8 agonists were administered to mice transgenic for human TLR8 to investigate gut microbiota characteristics following vaccination. Gut microbiota was investigated by next generation sequencing in the MiSeq Sequencing System. Adonis analysis was used to evaluate the effect of variables on gut bacterial community stucture. Chao1, Shannon index, and phylogenetic diversity index were used to explore the gut bacterial diversity. Results: The results showed that different vaccines have significant influence on mice intestinal bacteria (adonis analysis, p < 0.01), with gut bacterial diversity within the ESAT6 + TLR8 agonists group being significantly decreased compared to the ESAT6 treatment group (p < 0.01). Following infection with Mtb via tail vein injection, the bacterial community structure within the control versus vaccinated groups altered significantly (adonis analysis, p < 0.01), and the altered changed genera were markedly different between the groups. Following infection, Bifidobacteria differed between the groups, indicated that they play a vital role in the response to infection. Conclusions: Our results indicated that different vaccines might have distinct influences on intestinal flora, and their role should not be ignored.


Introduction
Tuberculosis (TB) remains a significant threat to human health with an increasing number of cases infected with Mycobacterium tuberculosis (Mtb) worldwide. During the invasion of Mtb in the human body, many host-pathogen-recognition receptors can recognize Mtb surface molecules, such as toll-like receptors (TLRs). It has been reported that several TLRs work to recognize Mtb and activate the innate immune responses accompanied by mycobacterial infection [1]. Also, gut microbiota play an important role in the function of host immune system [2]. This microbiome is vital for the development of host immune processes responding to stimuli, which is essential for maintenance of normal homeostasis [3][4][5][6][7]. Thus, the roles of human microbiota within TB infection should not be overlooked. Since the characteristic of sputum microbiota during TB infection was studied by next generation sequencing in 2013 [8], dramatic changes were observed in gut microbiota of tuberculosis patients [9]. The gut microbiota of TB patients was primarily characterized by significantly decreased bacteria related by short-chain fatty acids production [9]. Much evidence indicated the close relationship between the changes of gut and lung microbiota and the Mtb infection [10]. It was also proved in mice [11]. The gut microbiota is also crucial for the response to Mtb in the lung [12]. Gut microbiota rapidly modified following Mtb infection and formed a gut bacterial community structure that was significantly different from uninfected samples [11]. Vaccination is an essential means for prevention of TB. Work is ongoing to improve the vaccine effectiveness in various ways. Evidence indicates that dysbiosis of gut microbiota could affect the function of DCs and the response of T cell against Mtb [12]. Whether different immunization methods could affect gut microbiota as well as how they affect vaccine effectiveness remains poorly understood. Our previous study evaluated the impact of ESAT6 (Abcam)-aluminum hydroxide gel (group C) and ESAT6-aluminum hydroxide gel-TL8-506 (group D) on Mtb vaccine effectiveness. It was demonstrated that group D gained better protection against Mtb infection compared to group C as supported by CFU determination in the lung, bacterial load in lung and liver, lung histology, and percentage of granuloma area within lung lobes ( Figure S1) [13]. In the present study we investigated the fecal bacterial community from our previous research and compared the bacterial community structure of group B and group D within the gut microbiota responses to Mtb infection, which could lead to a better understanding of the shift of gut species within the context of infection.

Experimental process
In brief, 12 six-week-old female human TLR8 transgenic mice were used for experiments. Mice were divided into three equal groups, which were intramuscularly injected with either PBS (group B), ESAT6 (Abcam)-aluminum hydroxide gel (group C), or ESAT6-aluminum hydroxide gel-TL8-506 (group D). A second immunization with the same agent was performed four weeks later. Feces were collected four weeks following the second vaccination. Then, mice were challenged with 10 5 CFU of Mtb H37Rv in 200 µL of PBS, and feces were subsequently collected on days 18 and 90 post-Mtb infection ( Figure 1). Fecal samples were immediately subjected to DNA extraction. All the animal experiments were permitted by the Institutional Animal Care and Use Committee of the Institute of Laboratory Animal Sciences, CAMS & PUMC. Detailed information could refer to Tang's paper [13].

Next-generation sequencing
Feces DNA was extracted according to the manufacturer's recommendations using the PowerSoil ® DNA Isolation Kit (MO BIO Laboratories, Inc., Carlsbad, CA, USA). Firstly, the DNA was diluted to 1ng/uL and stored at −20°C for further process. Secondly, PCR of V3-V4 variable regions was performed using Nossa's (2010) primers [14] with the diluted template. The same amplification system was employed using sterile deionized water as the template for the negative control. Thirdly, Amplicon quality was detected using gel electrophoresis, after which amplicons were purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA), amplified with another round of PCR, and purified again using AMPure XP beads. The batch containing no visible band at 400 bp for the negative control was treated using the following steps. Equal amounts of the purified amplicons were pooled for subsequent sequencing using the MiSeq Sequencing System (Illumina, Inc., San Diego, CA, USA).

Sequence processing
Raw sequencing data were processed according to the protocol followed by Chen Lin [15] to remove the low quality sequences. Further denoizing of the sequences involved first abandoning reads containing ambiguous, homologous sequences or that were < 200 bp, retaining reads with 75% of bases above Q20, and finally detecting and removing any reads containing chimeras using the QIIME software (version 1.8.0) [16]. Clean reads were subjected to primer sequence removal and clustering to generate operational taxonomic units (OTUs) using the UPARSE software with a 97% similarity cutoff [17]. Representative reads for each OTU were then selected using the QIIME package. All representative reads were annotated and blasted against the Silva database Version 123 (16s rDNA) using the Ribosomal Database Project Classifier (confidence threshold = 70%) [18]. Each sample was rarefied to 16,973 sequences. The Chao1 richness estimator, Shannon index, and phylogenetic diversity (PD) [19] index were calculated using the mother software. We used the Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) to predict the KEGG category and functional enzyme [20].

Statistical analysis
The α-diversity [21] of the bacterial community was analyzed on the basis of Chao1, Shannon-Wiener, and Faith's phylogenetic diversity (PD) [22] index using the OTU matrix file. Phylogenetic tree was built using approximately maximum likelihood based on the evolutionary distance calculated by Jukes and Cantor model with QIIME. Phylogenetic tree was used for PD index. Principal coordinates analysis (PcoA) and a permutational multivariate analysis of variance were carried out using distance matrices (adonis) with Bray-Curtis distance in the package "vegan" in the R environment. All other statistical analyses were performed in the package "state" in the R environment. p values below 0.05 indicated statistically significant differences.

Effects of different immunization modes on intestinal flora within mice
PcoA based on the Bray-Curtis distance showed that the samples were clustered into three groups consistently within the method of immunization, and the gut microbiota structure differed significantly among the groups ( Figure 2A). Furthermore, it was found the prevalence of classes such as Actinobacteria, Bacilli, Bacteroidia, Coriobacteriia, Erysipelotrichia were significantly different among the groups ( Figure   2B). Of note, the abundances of Bacilli, Bacteroidia, and Erysipelotrichia displayed considerable changes ( Figure 2B). Analysis of bacterial diversity showed that the ESAT6-aluminum hydroxide gel-TL8-506 had a highly significant effect on Chao1, Shannon diversity, and PD diversity ( Figure 2C). However, the ESAT6 (Abcam)-aluminum hydroxide gel inoculation had no effect on bacterial diversity ( Figure 2C).

Changes of intestinal flora within immunized mice following Mtb challenge
Bacterial community structures significantly changed following Mtb infection ( Figure 3A). Taxa at the genus level with abundances greater than 0.001 were analyzed to compare the differences among groups. The communities of Lactobacillus, Oscillibacter, Ruminiclostridium, Alistipes, Bifidobacterium, and Bilophila were significantly changed ( Figure 3B), with greatest variation occurring within the Lactobacillus, Alistipes, and Bifidobacterium. Lactobacillus were suppressed on day 18 following Mtb infection and recovered to original levels by day 90; Alistipes and Bifidobacterium remained unchanged on day 18 but were suppressed on day 90 ( Figure 3B). Bacterial community structures within groups C and D also significantly changed following Mtb infection ( Figure 3C and 3E).   However, the significantly changed genera within these groups were quite different from those of group B. Parasutterella, Parabacteroides, Alistipes, Lactobacillus, Bifidobacterium were the most abundance genera present within mice in group C; Akkermansia, Klebsiella, Escherichia, Shigella, Blautia, Anaerostipes, Bilophila, Thalassospira, Lachnoclostridium, Bacteroides, Alistipes, Alloprevotella were the most abundant within group D mice ( Figure 3D and 3F). Alistipes and Lactobacillus showed similar behavior to group B, while Bifidobacterium was stimulated following Mtb infection on day 18 post-infection within group C ( Figure 3B and 3D). In group D, Alistipes prevalence was increased on day 18 and recovered to original levels on day 90, while the abundances of Lactobacillus and Bifidobacterium were unaffected by Mtb infection ( Figure 3F). A marked increase in Bacteroides and Lachnoclostridium levels were observed following Mtb infection ( Figure 3F).

Differences in intestinal flora within immunized mice following Mtb infection
The bacterial community structure was separated according to group ( Figure 4A). Alphaproteobacteria, Actinobacteria, Erysipelotrichia differed among groups ( Figure 4B). Bacterial diversity within infected samples was similar result to that of samples collected prior to Mtb infection ( Figure 4C). On day 90 post-infection, the bacterial community still significantly differed between the groups ( Figure 5A). Betaproteobacteria was the only class that differed significantly between groups B and D ( Figure 5B). Chao1, Shannon and PD indexes showed no differences among groups on day 90 ( Figure  5C).

Discussion
The gut microbiota affect mammalian both during homeostasis and disease states [23]. The relationship between humans and gut microbiota is vital for the maintenance of health. Vaccines provide active acquired immunity to particular pathogens, and their effectiveness is critical for disease prevention.  Therefore, gut microbiota composition may affect how individuals respond to vaccinations. Evidence indicates that many factors related to gut microbiota could affect the host response to vaccines. Kim et al. (2017) showed that malnutrition and poor sanitation, which affects the composition of gut microbiota, can impair the efficacy of vaccines. It was shown that antibiotic-driven intestinal dysbiosis could lead to decreased responses to adjuvant and live vaccines, and that restoration of the gutmicrobiota could recover these responses [24]. Evidence from human and mouse models both suggested that the gut microbiome could affect immune responses to vaccines [25]. However, limited studies have investigated the influence of the intestinal microbiome on vaccine responses [26]. Tuberculosis, commonly known as consumption for many years, results from infection by Mtb. The vaccine currently available for TB prevention is only BCG [27]. However, it was reported that the BCG vaccine has several limitations such as a low rate of protection [28,29]. Therefore, the development of novel vaccines that are more effective and safer should be considered as a priority. In this study, we selected two vaccination variations: ESAT6 and ESAT6 + TLR8 agonists. The effect of immunization on intestinal flora may affect its response to Mtb infection. We found that vaccination with ESAT6 + TLR8 agonists had a great effect on gut bacterial community and bacterial diversity compared to ESAT6 alone (Figure 2 and Figure S2). Toll-like receptor 8 is a protein in the TLR family in humans [30]. Agonists of the TLR family have been proven to enhance the protective effect of the Mtb vaccine [31][32][33]. Our previous study demonstrated that vaccination with ESAT6 + TLR8 agonists could strengthen the protective efficacy of Mtb immunization [13]. A diverse bacterial community could contribute to the improvement of protective efficiency. It has been shown that members of Bacteroidetes, which contain a great number of beneficial commensal microorganisms, decreased in patients experience repeated TB infections [34]. We found that the abundance of Bacteroidia increased following vaccination with ESAT6 + TLR8 agonists ( Figure 2B), indicating that Bacteroidia played an important role in vaccination efficacy. Mtb infection could also affect human sputum microbiota [35]. Murine gut microbiota significantly changes following Mtb infection [11]. Our results also confirmed that Mtb infection could alter the gut microbiota following challenge with Mtb infection (Figure 3A), as Lactobacillus and Bifidobacterium were both affected by Mtb infection ( Figure 3B).
Reduction of probiotics could indicate a decreased benefit of the presence of gut microbiota. Conversely, the abundance of Bifidobacterium in mice vaccinated with ESAT6 increased after Mtb infection ( Figure 3D) and no change occurred within the group immunized with ESAT6 + TLR8 agonists, possibly indicating no apparent negative influence of Mtb infection. However, the gut microbial community of vaccinated mice within both groups changed after Mtb infection ( Figure 3C and 3E). Probiotics could benefit host by several mechanisms such as restoration of disordered microbiota [36]. It was found that probiotics benefit the immune response to oral vaccinations which could improve the vaccine efficacy [26]. Thus, application of probiotics accompanied by vaccination may increase the efficiency of Mtb infection prevention. Differences in gut microbiota between groups B, C, and D remained following Mtb inoculation on day 18 ( Figure 4A), with the trend of bacterial diversity remaining unchanged ( Figure 4C). Significantly different taxa at the class level decreased over time ( Figure 4B, 5B, and Figure  S3), and the bacterial diversity indices of the groups became indistinguishable ( Figure 5C). We also found that several genera were significantly correlated with inflammation factors, including Alloprevotella, Clostridium_sensu_stricto_1, Romboutsia, Ruminococcaceae_UCG_013, Ruminococcus_1, Thalassospira, and Turicibacter (Table S1). These genera could possibly be considered as enhancers of immunity. The host-microbiota could initially protect lung from Mtb colonization [37]. Gut commensal bacteria could affect the outcome of BCG vaccination [38], and treatment of TB significantly alters the composition of the gut microbiota [39]. It was demonstrated that the addition of certain bacteria was an effective strategy for the process of prevention and treatment of tuberculosis.

Conclusions
In sum, vaccination is an effective method for prevention of tuberculosis. Our study found that different methods of vaccination distinctly influenced mice intestinal bacteria. TLR8 agonists vaccinated in combination with ESAT6 significantly affected gut microbiota and bacterial diversity. Behaviors of probiotics within the gut exhibited a significantly different response following Mtb challenge. The gut microbiota communities between the vaccinated groups also showed different patterns throughout the course of infection. All these data confirmed that vaccination strategy can affect gut microbiota structure and result in a shift of gut microbial communities following Mtb infection, and probiotics may be considered as a functional treatment to recover gut microbiota function.
These results could ultimately facilitate advances in TB treatment.

Data availability
All bacterial 16S rRNA gene sequences generated for the present study were deposited in the National Centre for Biotechnology Information Sequence Read Archive (http://www.ncbi.nlm.nih.gov/sra) under the accession number PRJNA588481.