Chronic exposure to ampicillin alters lung microbial composition in laboratory rat

Abstracts

Purpose

High-throughput sequencing technologies have revealed that the lungs contain a variety of low biomass microbiota associated with various lung diseases. Rat model is an important tool to understand the possible causal relationship between pulmonary microbiota and diseases. Antibiotic exposure can alter the microbiota, however, a direct influence of long-term ampicillin exposure on commensal bacteria of healthy lungs has not been investigated, which could be useful in the study of the relation between microbiome and long-term lung diseases, especially in animal model-making of lung diseases.

Methods

The rats were aerosolized ampicillin of different concentrations for five months, and then the effect on the lung microbiota was investigated using 16S rRNA gene sequencing.

Results

The ampicillin treatment by a certain concentration (LA5, 0.2 ml of 5 mg/ml ampicillin) administration leads to profound changes in the rat lung microbiota but not in the low critical ampicillin concentration (LA01 and LA1, 0.1 and 1 mg/ml ampicillin), when compared to the untreated group (LC). The genus Acidobacteria_Gp16 dominated the ampicillin treated lung microbiota while the genera Brucella, Acinetobacter, Acidobacteria_Gp14, Sphingomonas, and Tumebacillus dominated the untreated lung microbiota. The predicted KEGG pathway analysis profile revealed some difference in the ampicillin treated group.

Conclusions

The study demonstrated the effects of different concentrations of ampicillin treatment on lung microbiota of rats in a relatively long term. It could serve as a basis for the clinical use of antibiotic and the use of ampicillin to control certain bacteria in the animal model-making of respiratory diseases such as chronic obstructive pulmonary disease.

Keywords:

• 16S rRNA gene sequencing
• ampicillin
• antibiotics
• lung microbiome
• rat

Introduction

The host cell and symbiotic microorganisms form a superorganism in the human body, which collectively determines the disease and health of the body. The dysbiosis of symbiotic microbiota can modulate the progression of human disease or cause the disease.¹ The lungs are traditionally considered to be sterile, however, with the technological progress of high-throughput sequencing and bioinformatics analysis, a dynamic and complex microbial community has been found in the lung mucosa in both healthy and disease states.^2–5 It has been found that the lung bacterial community structure could respond to diseases such as asthma,³ infections by pathogens,^6,7 cystic fibrosis,⁸ chronic obstructive pulmonary disease (COPD),⁹ lung injury,¹⁰ or lung transplantation.¹¹

Disturbance of the microbial diversity and stability has been shown to correlate with immunologic, diverse metabolic, and inflammatory disorders.¹² Antibiotic exposure can alter the microbiota and is linked to the development and progression of disease. Most of the microbiome studies have focused on the gut microbiome and have examined the effects of various antibiotics on gut microbiota,^13–16 yet lung microbiome study is still in its infancy and facing a series of critical challenges.¹⁷ Aerosolization and nasal sprays are efficient and noninvasive methods to deliver molecules such as antibiotics to the specific body sites.¹⁸ In humans, inhaled antibiotics have been used to treat critical lung infections,¹⁹ and nasal spray antibiotics have been reported to prevent acute otitis caused by Streptococcus salivarius in children.²⁰ And therefore they represent a strategy to locally and efficiently modify the lung microbiota that could limit the exposure of other organs in the body. Ampicillin is the common clinical antibiotic used to treat a variety of bacterial infections. Indications include respiratory infections, urinary tract infections, meningitis, salmonellosis, and endocarditis. It can also be used to prevent group B streptococcal infection in newborns.

COPD is a common and frequently occurring disease of respiratory system, characterized by persistent respiratory symptoms and airflow limitation is one of the leading causes of morbidity and mortality worldwide.^21,22 The relationship between the changes in healthy lung microbiota and the progression of early COPD remains unclear. Antibiotic exposure can alter the microbiota. And a number of studies have found that long-term prophylactic antibiotic use in COPD for a period of up to 6 or 12 months, could reduce the number of patients with exacerbations and exacerbation frequency, and increase possibly health-related quality of life.^23,24 Animal models are valuable tools to investigate the lung microbiota as technical and ethical restrictions often limited the direct lung studies in humans. The rat is the most commonly used animal model. The COPD rat model-making needed to be exposed to smoke for 20 week.²⁵ During the time of the 20 week of model-making, using the antibiotic to intervene in the lung microbiota could benefit to uncover the relationship between the lung microbiota change in healthy rats and the formation of COPD. The maximal duration of continuous prophylactic antibiotics was 12 months, and for pulsed antibiotics it was 36 months.^23,24 There are no data on the impact of very long-term antibiotic use on the lung microbiota of healthy rats.

Herein, as the lung diseases such as the chronic obstructive pulmonary disease (COPD) were the long-term disease, we expected to study the relationship between lung microbial community change and the lung disease, by long-term interference using antibiotic. We investigated the effect of long-term and different concentrations of low doses of ampicillin on the pulmonary microbiota in a healthy rat model in the present study. We hypothesized that a long duration of antibiotic may lead to substantial changes in the pulmonary microbiota and the different concentrations of ampicillin may shape different lung microbiota. It could provide basis for further research on the relationship between the changes in bacterial flora structure and the occurrence and development of early COPD, and also provide some reference for clinically rational long-term prophylactic antibiotic use.

Materials and methods

Rats and sample collection

Male specific-pathogen-free (SPF) Sprague-Dawley (SD) rats (6 week in age) weighing 200 ± 20 g were obtained from Guangdong Medical Laboratory Animal Center (Guangdong, China) Biotechnology Co., Ltd. (Beijing, China). The animal care and experiments in this research were in accordance with the guiding principles for the care and use of laboratory animals recommended by the Chinese Association for Laboratory Animal Science Policy. It was approved by the Institutional Animal Care and Use Committee at the Guangzhou Medical University of China (Guangzhou, China; No. 2019-159). The rats were kept in a specific pathogen-free (SPF) room on alternate 12-h light/dark cycles at maintained temperature control and 50% humidity. They had free access to standardized food and water.

All rats were allowed to adjust to their environment for one month in four-animal cages in a common animal housing room on the free-standing shelves in the same one cabinet sharing the ventilation system before the experiments. To account for cage effect,^25,26 rats were assigned randomly into four groups: LC (n = 4), LA01 (n = 8), LA1 (n = 8), and LA5 (n = 8). Considering the previous microbiome-related studies by using the animal models and the complicated long-term antibiotic treatment with different concentrations which was difficult to accomplish for a big sample size, the number of 8 consistent genetic background rats were set for each group to explore effect of ampicillin antibiotic on lung microbiome. The LA01, LA1, and LA5 groups were administered aerosolized of 0.2 ml ampicillin with the concentration of 0.1 mg/ml (LA01), 1 mg/ml (LA1), and 5 mg/ml (LA5) once daily for one month. Then the antibiotic doses were adjusted to every other day until six months. The dosage utilized was determined according to the other similar researches about antibiotic disturbance in animal models.^27–30 The control group of LC did not receive aerosolization of ampicillin. The aerosolization of ampicillin was performed by the same experienced researcher using Micro Sprayer Aerosolizer (YUYAN Instrument, Shanghai, China), which was the standard aerosolization equipment.³¹ In order to avoid the continuous blockage of the rats’ noses and mouths by using isoflurane combined with a small animal anesthesia machine,^32,33 the rats were anesthetized with isoflurane by putting the cotton balls with 1–2 ml of isoflurane in the rat cage for 1–2 min. As described previously, the rats were held supine on a 60° incline board by having their upper incisor teeth after anesthesia. The mouth of the rats was gently opened due to the weight of the rats, and the rats were intubated through the mouth, and trachea using Micro Sprayer Aerosolizer inserted 10 mm into the trachea with a bulb-headed laryngoscope.³¹ We then used the nebulizing needle to aerosolize ampicillin into the airway through the epiglottis. After the treatment every time, the rats were housed in three or four-animal cages in sequential weeks over a 20 week period in the same room on free-standing shelves in the same one cabinet sharing the ventilation system.

After the six months, rats were then sacrificed under sodium pentobarbital anesthesia by neck removal to minimize suffering for lung tissue collection. The lung tissues of each rat were collected into 15 ml centrifuge tubs with 5 mL phosphate-buffered saline (PBS), immediately placed in liquid nitrogen, and stored at −80 °C for future analysis. The surgical scissors were cleaned and rinsed in ethanol, PBS and flamed between each organ. PBS control specimens from surgical scissors rinsed with clean instruments were included as procedural negative controls for whole lung tissue. The PBS on the sampling bench during the whole working time and blank DNA extraction without any sample template served as negative controls.

Lung microbial DNA extraction and 16S rRNA gene sequencing

Isolating the total DNA from lung tissue was accomplished using the Qiagen genomic DNA isolation kit according to the manufacturer’s instructions. The V3-V4 regions of the 16S rDNA were amplified using the primers 341 F (5′-CCTACGGGNGGCWGCAG-3′) and 806 R (5′-GACTACHVGGGTATCTAATCC-3′) combined with the adapter and barcode sequences. All PCR reactions were carried out in 10 μL reactions with 5 μL of Phusion® High-Fidelity PCR Master Mix (New England Biolabs), 0.3 μM of forward and reverse primers, and about 50 ng template DNA. The reaction started with initialization at 95 °C for 5 min, followed by 25 cycles of denaturation at 95 °C for 30 s, annealing at 50 °C for 30 s, and elongation at 72 °C for 40 s. And the reaction was completed with a final elongation step at 72 °C for 7 min. Three technical replicates for every sample were prepared and pooled during purification with OMEGA DNA purification Kit and Monarch DNA Gel Extraction Kit. Purified PCR products were quantified using Qubit®3.0 (Life Invitrogen) and mixed equally. The sequencing of pooled 16S rRNA gene was performed on an Illumina NovaSeq 6000 platform using the 250 paired-end protocol at Biomarker Technologies Corporation (Beijing, China). Sequence data is available at the ENA Accession number: PRJNA862086.

Bioinformatic and statistical analysis

Raw data were demultiplexed into individual samples according to the unique barcodes and then quality controlled using FastQC.³⁴ Forward and reverse end sequences of each sample were merged using Vsearch v2.21.1.³⁵ The sequences and the chimeric sequences were removed by Vsearch v2.21.1. Then the high-quality sequence analysis was performed by QIIME2 pipeline.³⁶ The Unoise3³⁷ was used to generate Amplicon Sequence Variants (ASVs). Sequences that occurred less than ten times were discarded and sequences with 99% similarity present in negative control samples were removed prior to diversity analyses in order to avoid misleading conclusions.³⁸ The SINTAX algorithm³⁹ with RDP training set v18 reference database^40,41 was used to annotate taxonomic information. The ASV table was sub-sampled randomly (the smallest number of reads in the sample) to obtain equal sequencing depths among samples. The rarefaction analysis, Observed species, Chao1, ACE, Shannon diversity estimators, Simpson’s diversity index, and Goods’ coverage were measured by R software⁴² with the vegan package.⁴³ Principal coordinates analysis (PCoA), and permutational multivariate ANOVA (PERMANOVA) tests were conducted to evaluate differences in community structures between the samples based on weighted Unifrac distance metrics using the QIIME2 pipeline.³⁶ Linear discriminant analysis effect size (LEfSe)⁴⁴ with FDR correction^45,46 was used to explore significant differences among groups in the abundance of the taxon. According to the normal distribution and homogeneity of variance, the Analysis of Variance Kruskal-Wallis test was used for statistical analysis, with the significance set at p < 0.05 and q < 0.05 by R package.⁴² The PICRUSt2 algorithm⁴⁷ was used to estimate probable Kyoto Encyclopedia of Genes and Genomes (KEGG)⁴⁸ function of the rat lung microbiome.

Results

Sequencing output and filtering

Experimental rats showed no obvious behavioral or physical changes during the course of the research. The 16S ribosomal RNA genes were sequenced by high-throughput sequencing technology to characterize the microbiota composition and community structure of rat lung microbiota. A total of 3,962,972 microbial sequences (sequences ranged between 40,758 up to 136,338 per animal sample) were detected across samples after discarding the sequences with less than ten.

After filtering for low-quality reads, comparison between the lung samples and negative controls were performed to confirm the effect of potential contamination. As shown in Figure S1, bacterial taxa detected in lung specimens were all collectively distinct from those detected in negative control specimens (p = 0.001 for all comparisons; PERMANOVA). The most abundant ASVs detected in lung specimens and negative controls were very different. These results confirmed that the detection of bacteria in lung specimens was not significantly influenced by potential contamination.

In order to remove only sequences that represented contaminants from the data set, sample sequences that matched any sequence from a negative control at the level of 99% identity over at least 200 bp were removed. In removing single sequences, we also removed ASVs that were present in less than three samples. After the above filtering process, 1,498,257 sequence reads were used for subsequent analyses and resulted in 2677 ASVs (Table S1). Collectively, these sequences represented 21 phyla, 52 classes, 74 orders, 159 families, and 334 genera. The pattern of the most abundant ASVs of the lung samples and the negative controls are similar.

Microbial richness and diversity (alpha diversity)

To evaluate alteration in the microbiota community structure between the groups, the microbial alpha diversity was compared as shown in Figure 1A. Rarefaction curve can verify whether the obtained sequencing data cover all taxa in the sample. As shown in Figure 1A, the rarefaction curve of each group tended to be flat which suggested that the sequencing depth was sufficient to reflect more than 99.3% (Table S1) of the microbiome taxa present. The number of samples extracted was appropriate, and the depth of sequencing data basically included enough taxa of the samples. Rank-abundance curve can represent the abundance and evenness of taxa in the samples. As depicted in Figure 1B, the curve of control group (LC) dropped faster than that in ampicillin groups particularly the LA5 group, indicating lower taxa abundance in LC group.

Figure 1. Alpha diversity comparison between the lung different groups. (A) Rarefaction curve for microbial communities in groups. (B) Rank-abundance distribution curve for microbial communities in groups. (C) The comparison of gut microbiota alpha diversity between each group, including Chao1, ACE, Shannon, and Simpson. Starred samples. Pairwise significance was determined by (C) pairwise and corrected for multiple comparisons using the Benjamini–Hochberg method. Significance key: *p ≤ 0.05; **p ≤ 0.01.

Figure 1C displayed the alpha diversity including Observed species, Chao1, ACE, Shannon, Simpson, and Goods’ coverage (Table S1), of lung microbiota in the four groups of rats administered different concentrations of antibiotics and controls. The result indicated that alpha diversity of the lung microbiota in LA5 group was higher than that of other groups including LC, LA01, and LA1 groups in terms of the richness index, indicating the increasing microbial taxa richness in rats challenged with 5 mg/mL ampicillin (Figure 1C). The Simpson index in LA5 group was lower than in other groups which also represent the higher microbial taxa richness in LA5 group as the above other alpha diversity indexes, though the difference was not significant in statistical analysis. However, when comparing the other ampicillin concentration groups (LA01, LA1) with control group (LC), no significant difference was found in Chao1, ACE, Shannon, and Simpson indexes.

Microbial community composition (beta diversity)

The weighted UniFrac distance metric was created from the subsampled even ASV table to explore structural differences in bacterial communities among the four groups. Principal coordinate analysis (PCoA) showed partial separation of microbiota communities by group (Figure 2A). Samples derived from the LA5 (p = 0.001) exhibited a distinct microbial structure and were completely separated from the LC control group. While microbiota communities in the LC, LA01, and LA1 groups were clustered closely, meaning that their bacterial communities were similar to each other, but separated from those in the LA5 group. The first three principal axes (PC1, PC2, and PC3) explained the most considerable variation of 51.99%. PERMANOVA test confirmed a significant association between the different concentrations of ampicillin treatment groups and bacterial community structures (Figure 2A). The PERMANOVA comparing all modes of ampicillin treatment groups to non-ampicillin group showed a significant difference (p = 0.001), supporting the view that the significant result of the PERMANOVA was due to true differences induced by antibiotic treatment, and could not be explained by differences in dispersion.

Figure 2. Lung microbiome composition in rats of different groups. (A) The microbiome composition of Lung tissue clusters by different concentrations of ampicillin together by principal component analysis of the ASVs using weighted Unifrac distances. The resulting p values and significance keys are tabulated. Individual data points represent specimens grouped by sample or control type. The top 1% relative abundance of detected phyla (B) and genera (C) in all groups. The y-axis denotes relative abundance (%). *p ≤ 0.05; **p ≤ 0.01.

According to 16S rRNA gene sequences, 19 phyla, 41 classes, 65 orders, 127 families, and 216 genera were identified. At the phylum level (Figure 2B), Proteobacteria (35.08%), Firmicutes (20.19%), Acidobacteria (14.91%), Actinobacteria (11.46%), candidate_division_WPS-2 (3.72%), Bacteroidetes (3.51%), Chloroflexi (2.72%) and Candidatus_Saccharibacteria (1.31%) dominating in both the ampicillin group (93.87%), and the untreated group (90.02%). At the genus level (Figure 2C), Brucella (6.02%), WPS-2_genera_incertae_sedis (3.72%), Acidobacteria_Gp1 (3.65%), Acinetobacter (2.46%), Acidobacteria_Gp14 (2.33%), Alicyclobacillus (2.15%), Sphingomonas (1. 80%), Acidobacteria_Gp6 (1.79%), Acidobacteria_Gp3 (1.61%), Bradyrhizobium (1.55%), Tumebacillus (1.51%), Planosporangium (1.47%), Clostridium_sensu_stricto (1.32%), Saccharibacteria_genera_incertae_sedis (1.31%), Staphylococcus (1.24%), Lactobacillus (1.18%), Acidobacteria_Gp16 (1. 16%) and Pseudomonas (1.16%) was dominated (39.11% in the ampicillin group, and 42.22% in the non-ampicillin group).

To further determine which taxa were significantly affected, we performed statistical analysis employing the relative abundance at the phyla to genera levels (Figures 2B, C and 3). As the LA01 and LA1 groups had no significant difference from the LC group of untreated lungs, indicating that ampicillin needed to reach a certain concentration to cause changes in the microbiota, we compared the LA5 group to find the taxon that potentially be influenced by the ampicillin. LEfSe analyses were conducted to detect the differences between the LC group and ampicillin-treated groups at the ASV, genus, or higher level. Comparison of the LC control and LA5 groups showed the presence of 97 taxa (including 36 genera) with significant differences (Figure 3A). At the genus level, Acidobacteria_Gp16, Ramlibacter, Nitrospira, Gaiella, Aggregatilinea, Latescibacteria_genera_incertae_sedis, and Reyranella were significantly increased in the LA5 group versus the LC control group, whereas Brucella, WPS_2_genera_incertae_sedis, Acinetobacter, Sphingomonas, Acidobacteria_Gp14, Tumebacillus, Acidobacteria_Gp2, Acidobacteria_Gp2, Neobacillus, and Ornithinimicrobium, Virgibacillus were significantly decreased. Besides, there were 8 taxa at the class level, 11 taxa at the order level, and 12 taxa at the family level that were significantly different when compared to the LC with LA5 groups.

Figure 3. Linear discriminant effect size (LEfSe) analyses comparing differentially abundant taxa between different groups. Microbes associated with the effect of different concentrations of ampicillin are shown. (A) Taxonomies differentiating bacterial microbiota in non-ampicillin control treated group (LC) and 5 mg/ml ampicillin treated (LA5) groups, LC and 0.1 mg/ml ampicillin treated (LA01) groups (B), LC and 1 mg/ml ampicillin treated (LA1) groups.

Though the whole microbial community structure, represented by the alpha and beta diversity indexes, had not significantly different between the LA01/LA1 and LC groups, there were still some taxa that were significantly different between them determined by LEfSe analysis. In our comparison of the LA01 and LC groups, the abundances of a total of four taxa (related to two genera) were found to be significantly changed (Figure 3B). At the genus level, Psychrobacter was significantly increased in the LA01 group as compared with the LC group, whereas Actinoallomurus and Acidobacteria_Gp16 were significantly decreased. Finally, in comparison of the LA1 and LC groups, the abundances of a total of nine taxa (including five genera) were found to be significantly changed (Figure 3C). At the genus level, Streptococcus and Novosphingobium significantly increased in the LA1 group as compared with the LC group, whereas Pontibacter, Sphingomonas, and Pseudoclavibacter were significantly decreased.

Differences gene function in all rats lung microbiota samples

The PICRUSt2 algorithm was further used to analyze the effects of ampicillin on lung microbiota function. A total of 10,543 KEGG orthologs (KOs)⁴⁸ and 382 pathways were detected. KEGG pathway comparisons were further performed to explore potential differences in the functional composition of the microbiome of antibiotic treatment groups versus no-antibiotic treatment controls by LEfSe analysis. We found that rat lung microbiota samples from the LA5 and LC control groups showed different KEGG profiles. The Drug metabolism cytochrome P450 (map00982), Phenylalanine metabolism (map00360), Fatty acid degradation (map00071), Non_homologous end joining (map03450) and beta-Alanine metabolism (map00410) had significantly lower inferred abundances in the LA5 group than in the LC control group (Figure 4A). The Base excision repair (map03410) and Central carbon metabolism in cancer (map05230) were up-regulated in the LA01 group than in the LC control group (Figure 4B), and the Streptomycin biosynthesis (map00521) had significantly higher inferred abundances in the LA1 group than in the LC group (Figure 4C).

Figure 4. Comparing Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways between different groups with Linear discriminant effect size (LEfSe) analyses. The list of LEfSe provides features that were differential among conditions of biological significance, ranking them according to the effect size. (A) Differential PICRUSt2 predicted KEGG pathways between LC and 5 mg/ml ampicillin treated (LA5) groups; 0.1 mg/ml ampicillin treated (LA01) groups (B); 1 mg/ml ampicillin treated (LA1) groups (C).

Discussion

The present study investigated the effects of ampicillin aerosolization on lung microbiota of Sprague-Dawley rats. Results of this study demonstrated that ampicillin could modify lung community compositions by the long-term low-dose nebulization of antibiotics at a certain concentration of LA5. Brucella, Acinetobacter, Acidobacteria_Gp14, Sphingomonas, Tumebacillus, and Acidobacteria_Gp16 were the dominant and significantly different taxa between the LA5 and LC groups (Figures 2 and 3). Furthermore, alterations were identified in potential metabolites associated with Drug metabolism cytochrome P450 (map00982), Phenylalanine metabolism (map00360), Fatty acid degradation (map00071), Non_homologous end joining (map03450), beta-Alanine metabolism (map00410) and so on (Figure 4A).

The bacterial signal was significantly different between the lung samples and the negative control samples (Figure S1). Potential contamination was eliminated from the data set to facilitate comparison between the different lung sample groups to thus avoid the contamination overwhelming the microbial signal for the low bacterial biomass lung tissue samples.^26,49 We used aerosolization to give the ampicillin to rats since it could achieve high antibiotic concentrations in the lower respiratory tract to maximize bacterial killing with minimal systemic side effects and had the minimum effect on gut microbiota,¹⁸ avoiding the gut dysbacteriosis that is related to lung microbiota.^25,50 Proteobacteria, Firmicutes, Acidobacteria, and Actinobacteria were the dominant phyla in LC group of healthy rats, which was consistent with previous reports.^51,52 These results indicated that aerosolization with sterile normal saline (control group in this study) did not change the microbiota composition in comparison to native lungs which was also confirmed by other researchers.⁵¹ Moreover, there were no significant differences in lung microbiota between LC, LA01, and LA1 groups, which indicated that only the concentration of ampicillin reached enough concentration that could influence the lung microbial compositions and the lung microbial structure could remain relatively stable for external disturbance.

Antibiotic therapy is often considered to diminish the microbial diversity and abundance, but we found that the alpha diversity index of rat lung microbiome in LA5 group was significantly higher than in other groups, and there was no significant difference in other antibiotic treatment groups versus no-antibiotic treatment controls groups. This phenomenon has been noticed in previous research.^53,54 It is likely attributable to the formation of communities of antibiotic-resistant bacteria after the long-term ampicillin treatment. And the differentiated taxon was increased with the increasing antibiotic concentration (Figure 3). The previous study found after the ampicillin treatment, the diversity of culturable strains decreased at the first 5 days but then the diversity increased with the long-term ampicillin treatment.¹⁷ This result was consistent with our study that the long-term ampicillin treatment may gradually enhance the antibiotic-resistant and the microbial diversity increased. The further study should also sample the short-term ampicillin treatment lung tissue to evaluate the microbial diversity. LEfSe revealed that Acidobacteria_Gp16, Ramlibacter, Nitrospira, Gaiella, and so on were the most unique microbes in the LA5 of high concentration ampicillin exposed group (Figure 3). The above taxon that enriched in the LA5 group mainly belonged to the phyla of Proteobacteria, Acidobacteria, and Actinobacteria. These three phyla have been reported to be the hosts of most antibiotic resistance genes (ARGs) and have the potential resistance to the ampicillin.^55–58 We speculated that the shift of the taxon commonly induced by antibiotic treatment, rather than specific taxa, is relevant to the reduced immunosuppression.^25,59

Also, the other reason may be that the balance of lung microbiota was disturbed by the certain ampicillin concentration, which increased the risk of bacterial infection and thereby increased microbial diversity. The long-term antibiotics treatment would kill certain bacteria and disturb the lung microbiota balance at the same time. The whole microbial diversity would change with the dysbiosis, including the chance of increasing diversity by the external bacterial infection. However, it is unclear whether antibiotic disturbance induced lung inflammation and then further aggravated the microbiome is altered, or antibiotic disturbance induces microbiome changes and then inflammation occurs. We will explore the relationship between inflammation and microbial changes after antibiotic exposure in future.

In our study, the Proteobacteria, candidate_division_WPS-2, Candidatus_Saccharibacteria, Bacteroidetesand, and Chloroflexi phyla and Bradyrhizobium, Planosporangium, Saccharibacteria_genera_incertae_sedis, Lactobacillus, Acidobacteria_Gp1,Acidobacteria_Gp3, and Alicyclobacillus genera were similar in antibiotic treatment groups versus no-antibiotic treatment controls groups, which was inconsistent with previous studies suggesting that lower microbial diversity favors Proteobacteria abundance,^9,60 as the Proteobacteria abundance was not significantly different. Interestingly, we observed no relationship between microbial diversity and Proteobacteria abundance that was consistent with the study of the effects of smoking on the lower respiratory tract microbiome in mice.⁶¹ In other words, although the microbial diversity index of LA5 was higher than in other groups (Figure 1C), the relative abundance of Proteobacteria had no significant difference (Figure 2). And the administration of different concentrations of ampicillin led to significant differences in the relative abundance of above taxon (Figures 2 and 3). However, in the lung of healthy humans, Bacteroidetes and Firmicutes were the two most abundant bacterial phyla, and changes in their abundance in rat lungs can reflect, to some extent, the internal conditions of the human body.⁶² This distinction is related to the fact that humans and rats are different taxa.

The functional predictions suggested that the administration of different concentrations of ampicillin led to changes in the metabolic pathways of lung microbiota in rats (Figure 4). Bacterial fitness was closely related to the ability of bacteria to absorb nutrients or amino acids provided by their respective host ecological niches, because they required energy and carbon to grow and replicate.⁶³ Consequently, pathogenic bacterial physiology must be adapted to these different physiological conditions to ensure the expression of adaptive and virulence factors.⁶⁴ The microbial community changed dynamically depending on the lung condition, followed by pneumococcal infection.⁶⁵ Compared with the different groups, LA5 showed more dramatic changes in the lung microbiota. Figure 4 showed that D_Alanine_metabolism (map00473), Lipopolysaccharide biosynthesis (map00540), Monoterpenoid biosynthesis (map00902), DNA replication (map03030), Carbon_fixation_in_photosynthetic _organisms (map00710), and so on increased significantly in the LA5 group, some of which were normal microbial metabolic functions (helping digestion and nutrient absorption).^53,66 In contrast, the functional pathways involved in Drug metabolism cytochrome P450 (map00982), Phenylalanine metabolism (map00360), Fatty acid degradation (map00071), Non_homologous end joining (map03450), beta-Alanine metabolism (map00410), and so on were reduced; although these alterations in microbiota genetic metabolic pathways were not necessarily associated with lung disease, they deserve the attention of pathology researchers and prescribing clinicians.⁶⁷

However, the exact interactions between lung microbiota and metabolite changes require further investigation. The KEGG pathway analysis indicated that the dysbiosis of gut bacteria in lung cancer was strongly associated with dysregulation of basic metabolic processes such as energy metabolism and transport.⁶⁸ In our present study, several metabolism pathways were found to be decreased after ampicillin treatment, such as Fatty acid metabolism, Tyrosine metabolism, Biosynthesis of unsaturated fatty acids, and so on. Therefore, consistent with the previous study,⁶⁸ we speculated that lung microbiota affected the host via the metabolites or immune system, which provides an improved understanding of perturbations of the microbiome-metabolome interface in lung health and may identify some potential diagnostic and therapeutic targets. However, the function of the microbiome was not necessarily entirely dependent on any of these interactions, and changes in these relationships may affect the lung disease system and cause subtypes.

Although our study yielded some novel insights, there still are some limitations. The gender-specific differences were not investigated which could examine the impact of sex hormones on the rat lung microbiome development. Also, more sampling time points should be considered to track the development of the lung microbiome in rats. At the same time, the effects of fungi and viruses should also be considered.

Conclusions

This study found that a long-term enough concentration of ampicillin exposure by aerosolization can alter the microbial composition in the rat lung. The alpha and beta diversity of Lung microbiota in relatively high concentration ampicillin treated LA5 group were significantly different from and LC control group. The percentage of Acidobacteria_Gp16 in lung tissue changed in the LA5 group after inhalation of ampicillin compared with the LC control group. The D_Alanine_metabolism, Lipopolysaccharide biosynthesis, Monoterpenoid biosynthesis, DNA replication, and Carbon_fixation_in_photosynthetic_organism were elevated and Drug metabolism cytochrome P450, Phenylalanine metabolism, Fatty acid degradation, Non_homologous end joining, and beta-Alanine metabolism were decreased in the LA5 group after atomizing ampicillin compared with the LC control group. Furthermore, defining the composition of the resident microbiome and how microbial communities shift with exposure may help to explain their role in responding to ampicillin. This study could serve as a basis for clinical use of ampicillin in the treatment of respiratory diseases. Subsequent studies are required to determine if these changes in the lung microbiota induced by ampicillin might have an effect on the development and progression of diseases. Also, the impact of the sex hormones, more developmental time points, and fungi and viruses on the lung microbiome of healthy rats, as well as lung microbial functions determining by protein or metabolism levels should be included. At the same time, the effect of the lung microbiome to the gut microbiome and the lung or brain function through the lung-gut or lung-brain microbiota axis should also be considered.

Author contributions

P.R., Y.Z., B.L., and G.L. conceived the study, directed the project, and designed the experiments. H.J. obtained the samples. T.H. performed the microbial sequencing analysis. P.C. interpreted the results, and wrote the manuscript. All authors contributed to the article and approved the submitted version.

Consent to participate and for publication

The authors declare no conflict of interest.

Ethics approval

The animal care and experiments in this research were in accordance with the guiding principles for the care and use of laboratory animals recommended by the Chinese Association for Laboratory Animal Science Policy. It was approved by the Institutional Animal Care and Use Committee at the Guangzhou Medical University of China (Guangzhou, China; No. 2019-159).

Availability of data and material

Sequence data is available at the ENA Accession number: PRJNA862086.

Code availability

Not Applicable.

Disclosure statement

The authors report no conflict of interest.

Additional information

Funding

This study was supported by the NSFC-Guangdong Joint Fund Project [U1901210], Science and Technology Planning Project of Guangzhou China [202102021133], Medical Scientific Research Foundation of Guangdong Province [20201123142953643], Local Innovative and Research Teams Project of Guangdong Pearl River Talents Program [2017BT01S155].