Changes in gut bacteria after orlistat treatment in obese mice

Eight-week-old male C57BL/6J mice, weighing 18–20 g and classified as specific-pathogen-free (SPF), were housed in an SPF animal facility. The facility maintained a controlled environment with a room temperature of 22–24°C, 60% humidity, and a 12-hour light/dark cycle. Mice were health-screened before the start of the experiment to ensure they were free from any significant diseases or abnormalities. All mice were provided by The Animal Center at the Second Hospital of Harbin Medical University (Harbin, China). The experimental procedures adhered to the guidelines set forth by the Harbin Medical University's Guide for the Care and Use of Laboratory Animals, and all protocols were approved by the Institutional Animal Care and Use Committee (IACUC). For one week, the mice were acclimatized and fed a complete diet comprising 40–43% corn, 26% bran, 29% bean cake, 1% salt, 1% bone meal, 1% lysine, 1% vitamins, and trace elements. During this period, their basal body weights were recorded to establish a baseline. After acclimatization, the mice were randomly divided into three groups: a normal control group (N, n=10) and a model group (n=20). The normal control group continued to receive the standard diet, while the model group was switched to a high-fat diet. The high-fat diet consisted of 60% fat and was composed of 2 kg of a mixture that included 1.2 kg of lard, 100 g of milk powder, and 0.8 kg of maltose, along with other components to ensure a balanced but high-fat intake. Following the establishment of the model, the 20 model mice were further randomly allocated into two subgroups: a positive control group (F, n=10) and an orlistat treatment group (A, n=10). The positive control group (F) received the high-fat diet without any additional treatments, while the orlistat group (A) received the high-fat diet supplemented with orlistat, a known anti-obesity drug, to evaluate its effects on weight gain and metabolic parameters.

Initially, the A received a 60% fat diet along with orlistat (30 mg/kg, twice daily, bid). The orlistat, provided by Hangzhou Zhongmei Huadong Pharmaceutical Co. Ltd. (China), was thoroughly pulverized and mixed with 0.5% carboxymethylcellulose (CMC) to ensure a uniform suspension. The mixture was subjected to repeated pipetting and ultrasonication to achieve a homogenous orlistat suspension, ensuring consistent dosing throughout the treatment period. The F group received a 60% fat diet and an equivalent volume of 0.5% CMC, which served as a vehicle control. This ensured that any observed effects in the orlistat group could be attributed to the drug itself rather than the vehicle. The N received a standard basic feed and an equivalent volume of distilled water, maintaining consistency in the experimental design. The treatment period spanned 9 weeks, during which all groups had unrestricted access to water and food.

Body weights were recorded weekly at the same time to ensure consistency and minimize variability. Initially, all mice underwent a 16-hour fast to standardize their metabolic state, followed by a 9-week feeding regimen according to their respective dietary groups. Capillary blood samples were collected from the distal third of the mice's tails at baseline and after 9 weeks. Blood collection was performed under rapid ether anesthesia to minimize stress and discomfort. FPG levels were measured using a blood glucose meter (Jiangsu Yuyue Medical Equipment & Supply Co. Ltd., China).

At the conclusion of the experiment, anesthesia was induced with 3% isoflurane mixed with 30% oxygen and 70% nitrous oxide using an anesthetic chamber. Anesthesia was maintained with 1.5% isoflurane via a facemask to ensure that the mice remained fully anesthetized throughout the procedure. Under full anesthesia, cardiac blood collection was performed to obtain a sufficient volume of blood for further analysis. The mice were then promptly sacrificed by cervical dislocation to ensure a quick and painless death, minimizing any potential distress and ensuring the accuracy of the experimental results. Heart blood samples were collected, and serum was isolated by allowing the blood to clot at room temperature for 4 hours. The clotted blood was then incubated at 4°C for 12 hours and subsequently centrifuged at 3,000 rpm for 15 minutes to separate the serum. The isolated serum was stored at -80°C until further analysis to maintain the integrity of the samples. Serum GLP-1 and GIP levels were determined using an enzyme-linked immunosorbent assay (ELISA) kit (Bioswamp, Wuhan, China).

From the onset of the study (week 0) to its conclusion (week 9), mice from the three groups were placed into respective metabolic cages at predetermined intervals each week. Fresh feces from 10 mice across the groups were collected within 2 hours using sterile forceps, then stored in aseptic centrifuge tubes. The tubes were sealed and labeled with sealing film, and the samples were immediately frozen at -80°C to preserve the integrity of the microbial DNA. Fecal samples collected at weeks 0, 3, 6, and 9 were used for DNA analysis. DNA was extracted from the fecal samples using the QIAamp^® Fast DNA Stool Mini Kit (QIAGEN Biotechnology) according to the manufacturer's instructions. The purity of the extracted DNA was assessed by measuring the absorbance of 12 randomly selected DNA samples using a Thermo NANODROP 2000C DNA detector. The DNA samples were then stored at -20°C in labeled centrifuge tubes to prevent degradation. The Qubit 2.0 DNA test kit was employed to quantify the isolated genomic DNA, which was subsequently used for PCR amplification. The PCR utilized primers fused with V3-V4 universal primers compatible with the MiSeq sequencing platform. The specific primers used were: i) 341F primer: CCCTACACGACGCTCTTCCGATCTG CCTACGGGNGGCWGCAG; 805R primer: GACTGGAGTTCCTTGGCACCCGAGAATTCCAGACTACHVGGGTATCTAATCC. PCR products from bacterial and archaeal DNA, with amplicon sizes over 400 bp, were treated with 0.6x volume of Agencourt AMPure XP magnetic beads to purify and size-select the DNA fragments. For fungal PCR products and other PCR products with amplicon sizes under 400 bp, 0.8x the volume of magnetic beads was used. This step ensures the removal of primer dimers and other non-specific PCR products, resulting in high-quality DNA for sequencing. The Qubit 2.0 DNA detection kit was used to accurately quantify the recovered DNA. The purified PCR products were then equally mixed at a 1:1 ratio to achieve a total of 10 ng of DNA per sample. This pooled DNA was prepared to a final sequencing concentration of 20 pmol, ensuring optimal conditions for subsequent sequencing on the MiSeq platform.

First, sample sequences were differentiated using barcodes. Each sample sequence then underwent a rigorous quality control process to remove non-specific amplifications and chimeric sequences (27, 28). Next, the raw reads were demultiplexed and allocated to their respective samples based on the barcode information. Vsearch (Version 2.10.4) was then utilized to merge paired-end reads from the initial DNA fragments. The merging process helps to reconstruct the full-length sequences from the shorter reads, which is essential for accurate downstream analysis. The merged raw tags were further refined into clean tags through the Vsearch quality control process. This refinement includes filtering out low-quality reads, trimming adapter sequences, and removing any remaining artifacts. The resulting clean tags are of high quality and suitable for subsequent analyses. To eliminate chimeric sequences, USEARCH (version 5.2.236) was employed using the de novo method. This approach identifies and removes chimeras that may have been formed during the PCR amplification process. Additionally, the Silva database was used to further screen and remove any remaining chimeric sequences (29).

Operational taxonomic units (OTUs) were analyzed by clustering sequences based on sequence similarity. Sequences were grouped into OTUs using a ≥97% similarity threshold, which is a common criterion for defining identical OTUs (30). A Venn diagram was used to display the number of shared and unique OTUs across the three groups, providing a clear illustration of their similarities and overlaps. This visual representation helps in understanding the distribution and uniqueness of OTUs among the different sample groups. The diversity of OTUs and their similarity indices were also evaluated to provide a quantitative measure of the microbial community structure. Additionally, a sample clustering tree diagram (dendrogram) was generated to provide a visual representation of the similarity and disparity among the samples. The dendrogram, depicted through its branching structure, shows how closely related the samples are to one another.

The alpha (within samples) and beta (among samples) diversities were analyzed using in-house Perl scripts, and the ggplot2 package in R (Version 3.2) was used for visualization. For each OTU, representative sequences were selected, and the GreenGenes database (Release gg_13_5) was used for taxonomic annotation. Dominant species classification based on abundance in each sample was performed using GraPhlAn 0.9.7 and iTOL 3.2.1 (31, 32). The predicted functional genes were categorized into Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and Kyoto Orthology (KO) groups. The differences in these functional profiles among the three groups were compared using STAMP (33). To determine the association between enriched taxa and significant functional metagenomes, Spearman's correlation coefficients were calculated.

Principal Component Analysis (PCA) was employed to reduce the dimensionality of the dataset while preserving the most influential features that account for the variation. This method helps to discern key elements and structures, streamlining the complexity and elucidating underlying patterns. The PCA was implemented using the vegan package in R (Version 3.2). Additionally, a co-occurrence network analysis was conducted to explore the relationships and interactions between different microbial taxa. The nine microbiota OTUs with the highest relative abundances were selected and consolidated based on the lowest common taxonomy assignments at the genus level. Spearman correlation analysis was performed using non-rarified sequence data to identify significant associations between bacterial genera. An edge (connection) was created between two bacterial genera if the Spearman correlation coefficient met the criteria of P < 0.05 and |r| > 0.7. These thresholds ensure that only strong and statistically significant correlations are included in the network. The network analysis was conducted using R software (Version 3.2), which provides robust tools for visualizing and interpreting the complex relationships within the microbial community.

To trace the biological evolutionary sequence and comprehend potential mechanisms, a phylogenetic tree was constructed to analyze sequence differences at the taxonomic level. The multiple sequence alignment was performed using MUSCLE (Version 3.8.31) (34), which is a widely used tool for aligning multiple sequences with high accuracy and speed. Boxplots were generated to compare the distance distributions within and among groups. This visualization, created using R (Version 3.2), helps to illustrate the variability and central tendency of the distances, making it easier to identify any significant differences in the distribution of sequence distances between and within the sample groups. Lastly, the Linear Discriminant Analysis Effect Size (LEfSe) method was employed to identify differences in the relative abundance of taxa across the groups (35). LEfSe is a powerful tool that combines statistical significance with biological relevance by using linear discriminant analysis (LDA) to detect features (taxa) that are significantly different between groups.

Body weight and metabolic indices are presented as mean ± standard deviation. One-way ANOVA was applied to compare the levels of body weight, FPG, intestinal hormones among different groups, Post-hoc pairwise comparisons were conducted using the Bonferroni correction method to control for the family-wise error rate and to identify specific differences between groups. All P values reported are two-tailed, with a significance threshold established at 0.05. Quantitative statistical analyses were executed using R software (Version 3.2).

The average weekly weight gain in the A group (0.48 ± 0.13 g vs 0.21 ± 0.12 g, P<0.0001) and the F group (0.78 ± 0.14 g vs 0.21 ± 0.12 g, P<0.0001) was significantly higher compared to the N group. In addition, the weight gain in the A group was significantly lower than that in the F group (0.48 ± 0.13 g vs 0.78 ± 0.14 g, P<0.0001; Figure 1). A significant elevation in FPG levels was observed in both F and A groups, with the A group showing a comparatively lower increase in FPG than the F group (Figure 1). Moreover, the levels of GLP-1 and GIP in the F and A groups were significantly reduced compared to the N group. Notably, the A group showed increased levels of GLP-1 and GIP relative to the F group, independent of weight changes (Figure 2).

From 12 fecal samples (four from each group), a total of 360,612 valid sequences were procured. After data trimming and quality control, 343,894 high-quality sequences were retained, averaging 28,658 sequences per sample. Figure 3A displays the similarity and overlap in the number of OTUs across the three groups, while the clustering tree diagram shown in Figure 3B demonstrates a high similarity between the F0 and N0 groups. The microbial diversity and richness were higher in the F group compared to the N group. Nevertheless, orlistat treatment led to a decline in both diversity and richness, as determined by the Shannon index (Figure 4A) and Richness index (Figure 4B). Echoing the trends seen in Figure 3A, the Rarefaction curve suggested lower Shannon index values (Figure 4C) and a reduced count of OTUs (Figure 4D) in the orlistat-treated mice. These findings suggest that a 60% fat diet correlates with increased microbial diversity and richness, while orlistat administration can diminish these parameters. The microbial composition of fecal samples from the A, F, and N groups was comparatively analyzed, and bacteria of relatively high abundance are illustrated in Figure 5. A notable decrease in Bacteroidetes proportion was noted in mice fed on a 60% fat diet, which was further reduced by orlistat treatment. In contrast, the Helicobacter proportion declined in the F group, whereas orlistat treatment was linked to a higher abundance of Helicobacter. Additionally, the proportion of Allobaculum increased in response to a 60% fat diet, with a subsequent increase upon orlistat administration.

Figure 6 illustrates the taxonomic tree of gut microbial composition, revealing shifts in the abundance of Bacteroidetes, which bifurcate into two branches. These branches include genera such as Macellibacteroides, Tannerella, Odoribacter, Barnesiella, Coprobacter, Alloprevotella, Alistipes, and Bacteroides.

Functional variations in the gut microbiota across the groups are depicted in Figure 7. Notably, the "cell motility" and "Neurodegenerative Diseases" pathways were substantially enriched in the orlistat-treated mice, whereas the "Glycan Biosynthesis and Metabolism" and "Transport and Catabolism" pathways were less expressed in these mice.

The PCA outcomes, shown in Figure 8, indicate alterations in the OTU similarity in the F group, which appeared to normalize following orlistat treatment. Spearman's rank correlations were employed to examine the co-occurrence patterns of microbial communities at the phylum level, aiming to discern shifts in bacterial ecosystem structure (Figure 9). A relatively intricate network of correlations was observed in the microbiota of orlistat-treated mice, in contrast to a more simplified network in mice fed on a 60% fat diet. Orlistat was also correlated with a reduced density in the bacterial correlation network.

The phylogenetic tree results are displayed in Figure 10. Distances between multiple samples within the three groups were computed, and the variance in these distances was assessed within as well as across groups (Figure 11). The findings indicated an expanded sample distance within the model group, while the range of fluctuation in the A group was akin to that of the N group. The LEfSe results highlighted the significance of Coprobacter, Parabacteroides, Acetatifactor, Lachnospiraceae, Romboutsia, and Peptostreptococcaceae in mice on a 60% fat diet. After orlistat treatment, differences were observed in Actinobacteria, Macellibacteroides, Parasutterella, Sutterellaceae, Burkholderiales, and Betaproteobacteria (Figure 12).