Probiotic Lactobacillus Paracasei Expressing a Nucleic Acid-Hydrolyzing Minibody (3D8 Scfv) Enhances Probiotic Activities in Mice Intestine as Revealed by Metagenomic Analyses

Probiotics are well known for their beneficial effects for animals, including humans and livestock. Here, we tested the probiotic activity of Lactobacillus paracasei expressing 3D8 scFv, a nucleic acid-hydrolyzing mini-antibody, in mice intestine. A total of 18 fecal samples derived from three different conditions at two different time points were subjected to high-throughput 16S ribosomal RNA (rRNA) metagenomic analyses. Bioinformatic analyses identified an average of 290 operational taxonomic units. After administration of L. paracasei, populations of the probiotics L. paracasei, Lactobacillus reuteri, and Pediococcus acidilactici increased, whereas the population of harmful bacteria such as Helicobacter species decreased. Furthermore, continuous administration of L. paracasei resulted in L. paracasei emerging as the dominant probiotic after competition with other existing probiotics. Expression of 3D8 scFv protein specifically increased the population of P. acidilactici, which is another probiotic. In summary, our results showed that L. paracasei expressing 3D8 scFv protein enhanced probiotic activity in mice intestine with no observable side effects. Thus, the system developed in this study may be a good tool for the expression of recombinant protein using probiotics.


Introduction
The microbiota in the gastrointestinal (GI) tracts of mammals are complex, and their composition can be influenced by several factors, including diets and environmental changes. Among known microbiota in the GI tracts, some bacteria species are regarded as probiotics, which can be defined as microorganisms providing beneficial effects when consumed by humans and animals [1]. Of known probiotics, many members of Lactobacillus spp. are consumed as health supplementary foods [2] and are used in the food fermentation industry to ferment cereals inhibiting pathogenic bacteria [3]. Lactobacillus are gram-positive bacteria with a rod-like shape that participate in glucose fermentation resulting in the production of lactic acid as well as small amounts of acetic and succinic acids [4]. Lactobacillus usually inhabits the mouth and GI tract [5]. In particular, the GI tracts of diverse mammals are preferentially colonized by Lactobacillus spp. such as Lactobacillus brevis, Lactobacillus casei, Lactobacillus acidophilus, Lactobacillus plantarum, Lactobacillus fermentum, and Lactobacillus salivariu [6]. To date, more than 70 different species of Lactobacillus have been identified.
The 3D8 single-chain variable fragment referred as 3D8 scFv is a nucleic acid hydrolyzing mini-antibody without sequence specificity [7]. The 3D8 scFv protein that was purified from Escherichia coli was subsequently shown to penetrate the cytosol of HeLa cells via caveolae-mediated endocytosis. Previous studies demonstrated that 3D8 scFv exhibits antiviral effects against a broad range of viruses, including the herpes simplex virus (HSV), pseudorabies virus (PRV), classical swine fever virus (CSFV), murine norovirus (MNV) and H1N1 influenza virus in various hosts [8][9][10][11]. Therefore, it is clear that 3D8 scFv engages in direct antiviral activities against various DNA and RNA viruses by penetrating into cells and directly hydrolyzing the viral genome based on previous findings. In addition, a previous study has shown that Lactobacillus paracasei expressing 3D8 scFV can be used as a preventive probiotic against norovirus infection [8]. In order to use L. paracasei expressing 3D8 scFV as a preventive probiotic, it is necessary to examine the possible impact of L. paracasei expressing 3D8 scFV in microbiota of the GI tracts. Because nothing is known about whether L. paracasei expressing 3D8 scFV will be beneficial or harmful to other microbiota. For this reason, metagenomics might be a good approach.
The rapid development of next-generation sequencing (NGS) has promoted studies associated with metagenomics, which reveals the composition of living microorganisms in a wide range of environmental samples [12]. In the case of bacteria, amplification of a partial 16S ribosomal RNA (rRNA) gene sequence followed by NGS is widely used as a culture-independent microbiological method. NGS-based 16S rRNA target sequencing allows identification of numerous bacteria species present in the intestine of animals. For example, it is estimated that there are approximately 40,000 bacterial species in the gastrointestinal microbiome of humans [13], and a recent study generated a reference mouse gut metagenome derived from fecal samples of 184 mice with different genetic backgrounds revealing 541 bacteria species [14]. Of known NGS systems, many previous studies used the FLX 454 system (454 Life Sciences, Branford, Connecticut, US), which produces long length sequence reads [15], whereas recent studies preferentially adopted the MiSeq system (Illumina, San Diego, California, US), which can produce reads up to 300 bases in length for 16S rRNA target sequencing [16].
In this study, we carried out high-throughput 16S rRNA metagenomic analyses to investigate the effects of the probiotic L. paracasei ATCC334 expressing 3D8 scFv in microbiota of the mouse intestine. A total of 18 fecal samples derived from three different conditions at two different time points were subjected to high-throughput 16S rRNA metagenomic analyses. Our results showed that L. paracasei expressing 3D8 scFv promoted probiotic activity in the mouse intestine.

Animals and Bacteria
Six-week-old female specific pathogen-free (SPF) BALB/c mice (Orient Bio Laboratories, Seongnam, Korea) weighing 18-20 g were housed under standard laboratory conditions. All animal procedures performed in this study were reviewed, approved, and supervised by the Institutional Animal Care and Use Committee (IACUC) of Kunkuk University (Ethical code: KU16080) and all experimental procedures performed here were in accordance with the guidelines of the Institute of Laboratory Animal Resources (ILAR). L. paracasei ATCC 334 was kindly provided by Dr. Jos Seegers (Falcobio, Lieden, Netherland). Wildtype (WT) L. paracasei and transgenic (TG) L. paracasei were anaerobically cultured in antibiotic-free de Man, Rogosa, and Sharpe (MRS) media and MRS media supplemented with 3 µg/mL chloramphenicol, respectively.

Sample Preparation and DNA Extraction
Fecal samples were collected from the mice 7 days after administration. Genomic DNA was extracted from a total of 18 samples, including three biological replicates for each condition, using ExtractMaster™ Fecal DNA Extraction Kit according to the manufacturer's instructions (Epicentre, Madison, WI, U.S.A.).

Preparation of Libraries for 16S Ribosomal RNA Sequencing and Paired-End Sequencing Using the MiSeq System
Libraries were constructed following the Illumina 16S metagenomics sequencing library preparation guide. In brief, the V3 and V4 regions of the 16S rRNA gene were targeted for PCR amplification using isolated DNA as a template. The 7-bp barcoded PCR primers were used to label 16S rRNA amplicons for each individual sample. Paired-end sequencing (2 × 300 bp) was conducted using the MiSeq system (Illumina Inc., San Diego, CA, USA) at Macrogen (Seoul, Korea). The raw images generated by Illumina MiSeq system were processed using MiSeq Control Software (MCS, v2.2, Illumina Inc.) for system control with base calling by integrated primary analysis software called Real Time Analysis (RTA, v1.18, Illumina Inc.). The binary base calls (BCL) is converted into FASTQ

Sample Preparation and DNA Extraction
Fecal samples were collected from the mice 7 days after administration. Genomic DNA was extracted from a total of 18 samples, including three biological replicates for each condition, using ExtractMaster™ Fecal DNA Extraction Kit according to the manufacturer's instructions (Epicentre, Madison, WI, USA).

Preparation of Libraries for 16S Ribosomal RNA Sequencing and Paired-End Sequencing Using the MiSeq System
Libraries were constructed following the Illumina 16S metagenomics sequencing library preparation guide. In brief, the V3 and V4 regions of the 16S rRNA gene were targeted for PCR amplification using isolated DNA as a template. The 7-bp barcoded PCR primers were used to label 16S rRNA amplicons for each individual sample. Paired-end sequencing (2 × 300 bp) was conducted using the MiSeq system (Illumina Inc., San Diego, CA, USA) at Macrogen (Seoul, Korea). The raw images generated by Illumina MiSeq system were processed using MiSeq Control Software (MCS, v2.2, Illumina Inc.) for system control with base calling by integrated primary analysis software called Real Time Analysis (RTA, v1.18, Illumina Inc.). The binary base calls (BCL) is converted into FASTQ using the Illumina package MiSeq Reporter (MSR). The obtained raw data were deposited in Sequence Read Archive (SRA) database in National Center for Biotechnology Information NCBI with project number PRJNA352304.

Preprocessing of Raw Data, Clustering, and Taxonomic Assignment
Data analyses consisted of three steps: preprocessing and clustering, taxonomic assignment, and statistical analyses for diversity. Paired-end reads were merged using the FLASH program [18]. The filtered clean reads were subjected to clustering at 100% identity to identify operational taxonomic units (OTUs) using the CD-HIT-OUT program [19]. Clustering was comprised of three steps. First, short reads and extra-long tails were removed from the raw data. The second step was to identify error-free reads. In the third step, the remaining representative reads from non-chimeric clusters were clustered using a greedy algorithm into OTUs with 97% identity at species level. Taxonomical composition from phylum to species level for each sample was analyzed using the quantitative insight into microbial ecology (QIIME) pipeline [20]. Community richness and diversity (α diversity) were analyzed by diverse α diversity metrics: Chao1, Shannon, Simpson, and Good's coverage using the QIIME pipeline.

Statistical Analyses
We have conducted a power test using the G*Power program (version 3.1.9.2) to calculate the power of the study [21]. Post hoc testing to compute the achieved power was conducted by one-way analysis of variance (ANOVA) using an F-test with the following parameters: Repeated measures, within-between interaction, effect size f 0.4 (large), α error probability 0.05, number of groups 3, number of measurements 2, correlation among replicate measures 0.05, nonsphericity correction 1, resulting in a power of 0.7882055 (78.82). In order to achieve a power higher than 0.8 (80) with the same parameters, the sample size should be more than 42, with an expected power of 0.8034136.
In addition, a t-test was conducted on five major species of Helicobacter sp., Pediococcus acidilactici, L. paracasei, L. reuteri, and Bifidobacterium using the following parameters: difference between two dependent means (matched pairs), correlation between groups 0.5, α error probability 0.05, total sample size 6, and respective effect size. Effect sizes for two different groups was determined from group parameters (mean value group 1, mean value group 2, standard deviation (SD) value group 1, SD value group 2) using the G*Power program.
Next, we conducted a two-way ANOVA with replication using Excel's Data Analysis (version 2016, Microsoft, Redmond, WA, USA) to find bacteria species significantly changed under three different conditions (negative (N), TG and WT) and two time points.
The differences in abundance of bacterial genera among 18 samples, the ANOVA was performed using the Statistical Analysis of Metagenomic Profiles (STAMP v. 2.1.3) software package [22]. For comparison between two different conditions, two-sided Fisher's exact test with confidence interval (CI) method (DP: Asymptotic-CC with 0.95) was conducted using STAMP software package.

Sample Preparation and High-Throughput 16S ribosomal RNA Metagenomic Analyses Using MiSeq System
Negative samples were obtained from the mice fed with PBS buffer or fed a diet containing WT samples were from mice fed L. paracasei, and TG samples were from mice fed L.paracasei expressing 3D8 scFv, as established in a previous study [8]. To examine changes in the microbiota in mice intestines among different conditions, we harvested fecal samples at two different time points (day 0 and day 7) for each condition ( Figure 1 and Table 1). Finally, a total of 18 samples derived from three different conditions at two different time points with three biological replicates were subjected to metagenome profiling. A total of 5,089,513 reads (2,302,478,088 bp) were obtained from 18 libraries (Table 1). Between 224,927 reads (102,570,095 bp) and 326,110 reads (148,592,201 bp) were obtained from each sample ( Table 1). The power of our study was 0.7882055 (78.82) with three number of groups (three conditions), two number of measurements (two time points), and three replicates for each group.

Microbial Composition in Mice Intestine
We carried out bioinformatics analyses for NGS data as described in methods section. Because of the stringent criteria for filtering sequenced reads, approximately 21% to 27% of sequenced raw data were further used for the identification of OTUs. We obtained from 48,511 reads (W0R2) to 69,639 reads (N7R2), which was enough for OTU analyses (Figure 2A). There was some bias for the number of filtered reads among replicates. In total, we identified 396 OTUs from 18 libraries (Table S1). In detail, between 242 OTUs (N7R2) and 320 OTUs (T0R1) were identified, with an average of 290 OTUs ( Figure 2B). Based on our results, there was no correlation between number of reads and number of OTUs.
Among 13 identified known classes, the proportion of bacteria belonging to an unknown class ranged from 0.5% (W0R1) to 26.10% (N7R1) ( Figure 3A). The proportion of the class Bacteroidia ranged from 42.6% (T0R2) to 72.7% (N7R3). The second dominant bacterial class was Clostridia ranging from 11.5% (N7R1) to 38.2% (T0R2). The class Epsilonproteobacteria was also abundantly present in all samples. The class Deferribacteres was identified in most samples except W7R2, and compared with other samples, the proportion of class Deferribacteres was very high in the N7R2 sample.  Among the 31 known bacterial families, only 0.8% of bacteria were identified as unknown ( Figure 3B). The three bacterial families Porphyromonadaceae, Bacteroidaceae, and Rikenellaceae were abundantly present in all samples. Rarefaction curves based on the Chao1 index were generated for 18 libraries and showed that the rarefaction curves nearly reached a plateau (Figure 4). Sequencing coverage for the 18 samples ranged from 0.9991079 to 0.999704 (Table 2). Both rarefaction and sequencing coverage indicate sufficient coverage of 16S rRNA sequencing. In addition, we calculated the richness estimate and diversity index using the Mothur program [23]. Although there was Among the 31 known bacterial families, only 0.8% of bacteria were identified as unknown ( Figure 3B). The three bacterial families Porphyromonadaceae, Bacteroidaceae, and Rikenellaceae were abundantly present in all samples. Rarefaction curves based on the Chao1 index were generated for 18 libraries and showed that the rarefaction curves nearly reached a plateau (Figure 4). Sequencing coverage for the 18 samples ranged from 0.9991079 to 0.999704 (Table 2). Both rarefaction and sequencing coverage indicate sufficient coverage of 16S rRNA sequencing. In addition, we calculated the richness estimate and diversity index using the Mothur program [23]. Although there was significant bias for the richness estimate and diversity index among biological replicates in the same condition, we did not find any significant differences among conditions (Table 2). *OTUs: Operational taxonomic unit is an operational definition of a species or group of species that is often used when only DNA sequence data are available *Chao1: The Chao calculator returns the Chao1 richness estimate for an OTU definition *Shannon: The Shannon index takes into account the number and evenness of species. *Simpson: The Simpson index represents the probability that two randomly selected individuals in the habitat will belong to the same species. *Goods Coverage: Goods Coverage calculated as C=1-(s/n), where s is the number of unique OTUs and n is the number of individuals in the sample. This index gives a relative measure of how well the sample represents the larger environment

Comparison of Bacterial Compositions Among the Three Different Conditions
To identify condition-specific bacterial genera, we analyzed the composition of bacterial genera in three conditions at two different time points ( Figure 5). A total of 49 and 50 genera were identified at day 0 and day 7, respectively. Most of the identified bacterial genera were commonly identified in all three conditions. For example, 43 genera (87.8%) and 44 genera (88%) were commonly identified at day 0 and day 7, respectively. At day 0, the two genera Akkermansia and Enterococcus were specific for the negative condition whereas Atopostipes and Pediococcus were specific for TG and WT, respectively. The read counts for condition specific bacterial genera were very low, ranging from only 1 count to 22 counts. At day 7, the genera Oceanobacillus and Olsenella were specific for negative condition. Interestingly, the read count for the genus Oceanobacillus was very high (2815 reads). The genus Akkermansia was commonly identified in negative and TG conditions relatively high read counts (N: 570, TG: 1443). Furthermore, between TG and WT conditions, the three genera Pediococcus  The Simpson index represents the probability that two randomly selected individuals in the habitat will belong to the same species; Goods Coverage: Goods Coverage calculated as C = 1 − (s/n), where s is the number of unique OTUs and n is the number of individuals in the sample. This index gives a relative measure of how well the sample represents the larger environment.

Comparison of Bacterial Compositions Among the Three Different Conditions
To identify condition-specific bacterial genera, we analyzed the composition of bacterial genera in three conditions at two different time points ( Figure 5). A total of 49 and 50 genera were identified at day 0 and day 7, respectively. Most of the identified bacterial genera were commonly identified in all three conditions. For example, 43 genera (87.8%) and 44 genera (88%) were commonly identified at day 0 and day 7, respectively. At day 0, the two genera Akkermansia and Enterococcus were specific for the negative condition whereas Atopostipes and Pediococcus were specific for TG and WT, respectively. The read counts for condition specific bacterial genera were very low, ranging from only 1 count to 22 counts. At day 7, the genera Oceanobacillus and Olsenella were specific for negative condition. Interestingly, the read count for the genus Oceanobacillus was very high (2815 reads). The genus Akkermansia was commonly identified in negative and TG conditions relatively high read counts (N: 570, TG: 1443). Furthermore, between TG and WT conditions, the three genera Pediococcus (TG: 9697, WT: 4), Aerococcus (TG: 274, WT: 17), and Enterococcus (TG: 7, WT: 11) were identified. In particular, the number of reads associated with the genus Pediococcus was very high in the TG condition.  We further analyzed the identified bacteria according to species in order to decipher changes in bacterial composition among the three different conditions (Table 3). In total, 1,032,761 reads were identified from 18 samples. Of these, 89.37% of reads were derived from uncultured bacteria, followed by uncultured Helicobacter sp. (5.63%), and unknown species (2.35%). At the species level, there were dramatic changes in microbial composition among conditions ( Figure 6). L. paracasei that was used for feeding was not found in N0, N7, T0, or W0, as expected (Table 3 and Figure 6A-6C). At 7 days after feeding, a large number of L. paracasei were found in both TG and WT samples ( Figure  6B and 6C). The number of L. paracasei in WT at day 7 (4,790 reads) was about five times larger than that in TG at day 7 (953 reads). Interestingly, we also identified another lactobacillus, L. reuteri that was not fed to the mice. The amount of L. reuteri was decreased in negative and transgenic samples at day 7 compared day 0, whereas the amount of L. reuteri was increased in wildtype at day 7 ( Figure  6B and 6C). In addition, we identified P. acidilactici, which is a known probiotic. Surprisingly, the population of P. acidilactici was dramatically increased in the transgenic condition at day 7 ( Figure  6D). However, the number of uncultured Bifidobacterium species was decreased in all three conditions at day 7 compared with day 0 (Figure 6A-6C). We further analyzed the identified bacteria according to species in order to decipher changes in bacterial composition among the three different conditions (Table 3). In total, 1,032,761 reads were identified from 18 samples. Of these, 89.37% of reads were derived from uncultured bacteria, followed by uncultured Helicobacter sp. (5.63%), and unknown species (2.35%). At the species level, there were dramatic changes in microbial composition among conditions ( Figure 6). L. paracasei that was used for feeding was not found in N0, N7, T0, or W0, as expected (Table 3 and Figure 6A-C). At 7 days after feeding, a large number of L. paracasei were found in both TG and WT samples ( Figure 6B,C). The number of L. paracasei in WT at day 7 (4,790 reads) was about five times larger than that in TG at day 7 (953 reads). Interestingly, we also identified another lactobacillus, L. reuteri that was not fed to the mice. The amount of L. reuteri was decreased in negative and transgenic samples at day 7 compared day 0, whereas the amount of L. reuteri was increased in wildtype at day 7 ( Figure 6B,C). In addition, we identified P. acidilactici, which is a known probiotic. Surprisingly, the population of P. acidilactici was dramatically increased in the transgenic condition at day 7 ( Figure 6D). However, the number of uncultured Bifidobacterium species was decreased in all three conditions at day 7 compared with day 0 ( Figure 6A-C).

11
We identified not only probiotics, but also potential pathogenic bacteria such as uncultured

12
Firmicutes bacterium, uncultured Helicobacter species, and uncultured Shigella species. Of these, the

24
We examined the relative abundance of the 52 known bacteria genera in three conditions at two

28
were highly present in WT and TG, but not in the N sample (Table S2). Genera Oceanobacillus (p =  We identified not only probiotics, but also potential pathogenic bacteria such as uncultured Firmicutes bacterium, uncultured Helicobacter species, and uncultured Shigella species. Of these, the number of Helicobacter species was decreased in all conditions at day 7 compared with day 0 ( Figure 6A-C).
We calculated the power based on five major species of Helicobacter sp., P. acidilactici, L. paracasei, L. reuteri, and Bifidobacterium. As a result, we obtained a power greater than 0.8 following three bacteria species in a given comparison, such as a power of 0.8556975 (T0 vs. T7) and 0.8820874 (W0 vs. W7) for Helicobacter sp., a power of 0.9914611 (N7 vs. T7) and 0.9825466 (W0 vs. W7) for P. acidilactici, and a power of 0.8329246 (W0 vs. W7), 0.9139413 (T7 vs. W7), and 0.9370286 (W7 vs. N7) for Lactobacillus reuteri. Furthermore, a two-way ANOVA test demonstrated that the abundance of Helicobacter sp. (p = 0.020955727) was significantly changed among the three conditions, while P. acidilactici showed significant changes in abundance by time (p = 0.007410375), condition (p = 0.00245994), and by the interaction between time and condition (p = 0.002458016).
We examined the relative abundance of the 52 known bacteria genera in three conditions at two different time points. The relative abundance of selected bacteria at day 7 was compared to that at day 0. The heat map shows changes in bacteria abundance under the three conditions (Figure 7). Bacteria belonging to the genera Aeroccus (p = 0.237), Pediococcus (p = 0.093), and Enterococcus (p = 0.060) were highly present in WT and TG, but not in the N sample (Table S2). Genera Oceanobacillus (p = 0.332) and Olsenella (p = 0.006) were only present in N samples; however, Oceanobacillus (p = 0.332) was highly abundant at day 7, whereas Olsenella (p = 0.006) was highly abundant at day 0. The abundance of bacteria belonging to the genera Anaeroplasma   shown the effects of antibiotic treatment on the mouse intestinal microbiome [24][25][26]. In addition, 46 some studies have demonstrated the influence of dietary ingredients on the fecal microbial community in mice [27,28]. Moreover, the change of mouse intestinal microbiota in response to 48 different environmental changes such as irradiation [29], acclimatization-induced stress [30], and 49 exposure to arsenic and iron [31] has been studied. Gut microbiota from mice with different genetic 50 backgrounds such as wildtype mice and gnotobiotic mice have also been previously examined 51 [32,33].

52
In this study, we performed metagenome sequencing of a mouse model to examine the impact 53 of consumption of lactobacillus expressing a recombinant protein on mouse intestine microbiota.

54
We identified approximately 300 species from a single source mouse which represents about 60% of 55 known bacteria species in a reference mouse gut metagenome derived from 184 mice [14]. Based on 56 this result, our metagenomic approach was successful for revealing the microbiota in the intestine of 57 a single source mouse. Among three biological replicates used for each condition, the microbial 58 compositions among different replicates for the same condition were similar, but the abundance of Figure 7. Relative abundance of known bacterial genera in three different conditions. Read numbers for 52 known bacteria genera in the same condition were combined. For each condition, the number of reads at day 7 was compared with the number of reads at day 0. The fold changes were converted into log 2 and imported into the Genesis program. Hierarchical clustering was performed using the complete linkage method. Red and green colors indicate up-and down-regulation, respectively. A gray color indicates that the number of reads associated with the respective bacterium was 0.

Discussion
Mice are frequently used as a model animal to examine the changes in microbiota in various tissues in response to diverse environmental stimuli. Modulation of the microbiota under a given condition results in a beneficial or harmful effect on the host life. For instance, many studies have shown the effects of antibiotic treatment on the mouse intestinal microbiome [24][25][26]. In addition, some studies have demonstrated the influence of dietary ingredients on the fecal microbial community in mice [27,28]. Moreover, the change of mouse intestinal microbiota in response to different environmental changes such as irradiation [29], acclimatization-induced stress [30], and exposure to arsenic and iron [31] has been studied. Gut microbiota from mice with different genetic backgrounds such as wildtype mice and gnotobiotic mice have also been previously examined [32,33].
In this study, we performed metagenome sequencing of a mouse model to examine the impact of consumption of lactobacillus expressing a recombinant protein on mouse intestine microbiota. We identified approximately 300 species from a single source mouse which represents about 60% of known bacteria species in a reference mouse gut metagenome derived from 184 mice [14]. Based on this result, our metagenomic approach was successful for revealing the microbiota in the intestine of a single source mouse. Among three biological replicates used for each condition, the microbial compositions among different replicates for the same condition were similar, but the abundance of identified bacteria was diverse. These results indicate that each individual mouse has a unique complex of microbiota, as previously shown [14].
Of the known top core genera in the mouse gut metagenome, three genera, Bacteroides, Alistipes, and Lactobacillus, were identified in our study; however, we additionally identified Barnesiella, Mucispirillum, and Oscillibacter genera. Of these, bacteria in the genus Bacteroides are clinical pathogens present in most anaerobic infections and have a beneficial relationship with the host [34]. In addition, Barnesiella is an obligatory anaerobic bacteria belonging to the Bacteroidetes and is one of the bacteria most frequently identified in the gut microbiota of BALB/c mice, which were used in our study [35].
Numerous studies have demonstrated that many factors influence the composition of microbiota. Of factors known to influence the composition of microbiota, probiotics are the main living factors that positively affect host health. For example, a transcriptome-based study showed that B. bifidum, a probiotic, reduced the cholesterol level in mice [36]. Colonization of B. bifidum in the intestine of different mice was monitored by quantitative real-time PCR [37]. Other studies using lactobacillus species showed a reduction in obesity [38] and inhibition of alcohol-induced pathogenic activity in the liver [39]. In addition, several previous studies using metagenomic approaches showed that probiotics modulate the intestinal microbiota [40]. Furthermore, a recent study demonstrated that the probiotic L. casei expressing human lactoferrin confers antibacterial activity in the GI tract of mouse [41].
In this study, we found that the administration of lactobacillus resulted in an increased population of the introduced L. paracasei. However, the populations of other probiotics L. reuteri and uncultured Bifidobacterium spp., which were already present in the intestine of the mouse, were decreased. We suppose that this result reflects competition among different probiotics, with the introduced L. paracasei becoming the dominant probiotic as a result of its continuous administration. In contrast, the population of uncultured Helicobacter sp., which are harmful bacteria, was decreased by administration of lactobacillus. These results are highly consistent with the known ability of probiotics to increase populations of probiotics and decrease populations of harmful bacteria.
The heat map identified several bacteria genera whose abundance was dramatically increased by the administration of lactobacillus. In particular, several round-shaped bacteria such as the genera Aerococcus, Pediococcus, Enterococcus, and Ruminococcus were identified. In general, members of the genera Pediococcus and Enterococcus are lactic acid bacteria of the phylum Firmicutes [42]. Therefore, we cautiously presume that the identified bacteria belonging to the genera Aerococcus and Ruminococcus might be similar to other lactic acid bacteria.
We also identified a large number of uncultured bacteria that are unknown. Although these included several bacteria species that might be involved in probiotic activity, most of them were unknown bacteria for which genome and basic characteristics are not currently available. Therefore, identification and characterization of unknown bacteria residing in the mouse intestine should be intensively conducted.
The most interesting result was a dramatic increase in the population of P. acidilactici, which is a probiotic in the T7 condition. The number of reads for P. acidilactici was one in W0 and four in W7 conditions. This result indicates the presence of P. acidilactici in the BALB/c mouse. Interestingly, administration of L. paracasei expressing 3D8 scFv facilitated growth by P. acidilactici, whereas administration of L. paracasei without 3D8 scFv did not change the population of P. acidilactici. Based on this result, we cautiously propose a positive relationship between expressions of 3D8 scFv and P. acidilactici.
We hypothesized that the newly developed system composed of lactobacillus expressing 3D8 scFv would not affect the microbial composition in the mouse intestine. However, the influence of lactobacillus expressing 3D8 scFv was very species specific. It might be of interest to further examine the correlation between expression of 3D8 scFv and P. acidilactici. Furthermore, it would also be interesting to test other recombinant proteins for the ability to change the microbiota in the intestine of mice. We believe that the newly established system expressing a recombinant protein might be a useful tool for probiotics-related research or industry.
Taken together, our results showed that probiotic L. paracasei expressing 3D8 scFv protein enhances probiotic activity in the mice intestine without any observable side effects. Thus, the system developed in this study might be a good tool for expression of recombinant protein using probiotics in the near future.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4425/9/6/276/s1, Table S1: Classification of identified 396 OTUs identified from 18 samples with respective bacteria taxonomy and read number., Table S2: Statistical analyses of 53 bacterial genera in 18 different samples by ANOVA implemented in STAMP program., Table S3: Read number, mean value, and standard deviation for the identified 396 species in each condition.
Author Contributions: S.C. and S.L. conceived and designed the experiments; S.C. and D.K. carried out the experiments; S.C., Y.L., E.-J.K., and M.-J.C. analyzed the data; S.-J.B. contributed reagents/materials/analysis tools; S.C. and W.K.C. drafted the paper. All authors contributed to subsequent drafts of the manuscript. All authors read and approved the final manuscript.