Transcriptome Profiling of Cellular Response to Kanamycin B in E.coli


 Background: Aminoglycosides are not only antibiotics but also have wider and diverse non-antibiotic cellular functions. No genome-wide study focusing on the changes of gene expression by aminoglycosides in E.coli has been reported. Here, we report transcriptome-profiling analysis of E.coli with or without Kanamycin B to elucidate the understanding of non-antibiotic cellular functions. Results: The differentially expressed genes (DEGs) at two given concentrations of Kanamycin B were identified. The results indicated that Kanamycin B does not affect the expression of the majority of the genes. Functional classification of the DEGs revealed that they were mainly related to microbial metabolism including two-component systems, biofilm formation, oxidative phosphorylation and nitrogen metabolism in diverse environments. Conclusions: Kanamycin B treatment causes diverse changes in the transcriptional profile of E. coli JM109, that are not directly associated with the antibiotic activity of Kanamycin B.

treatment. The results provide an insight into the wider non-antibiotic function of aminoglycosides and provide a foundation for a better understanding of aminoglycosides in clinical use.

Overview of the RNA-seq data
To investigate the effect of Kanamycin B on E. coli JM109, we performed transcriptome-pro ling analysis by RNA-seq in the absence or presence of Kanamycin B. Since Kanamycin B is a bactericidal antibiotic, initial cultures in the absence or presence of a broad range of Kanamycin B were cultivated to select concentrations that are neither lethal nor inhibit growth ( Figure 1A). 0.5μM or 1μM of Kanamycin B showed minimal effects on cell growth, while 2μM of Kanamycin B clearly inhibited growth. As a result, 0.5μM and 1μM of Kanamycin B were chosen for transcriptome pro ling analysis. Total RNA was extracted from cells at mid-log phase that had been treated with 0, 0.5μM and 1μM of Kanamycin B for RNA-seq analysis. The transcriptome-pro ling data was uploaded to the National Center for Biotechnology Information Sequence Read Archive (accession number PRJNA756617). With three biological repeats of each sample, a total of 9 cDNA libraries were constructed containing 340.20 million raw reads; 339.64 million clean reads (accounting for 99.83% of raw reads) were recorded after removing adapter sequences and reads of low quality and those with more than 5% N bases. The average number of clean reads per sample was about 37.74 million and the clean Q20 (sequencing error rate < 1%) base rate was > 97.86% for each sample. Ultimately, 336.74 million high-quality reads (accounting for 99.14% of clean reads) were mapped to the Escherichia coli str. K-12 substr. MG1655 genome.
To investigate their reproducibility, the biological repeats were analysed by principal component analysis (PCA). The results revealed that PC1 and PC2 had values of 26.1% and 14.4%, respectively, and accounted for 40.5% of the principal components ( Figure 1B). PCA showed consistency between the three replicates at each Kanamycin B concentration; samples from the biological repeat clustered together, re ecting minimal differences between them. PCA suggested that the three biological repeats in this study were reasonably reproducible ( Figure 1B).

Gene expression pro le in response to Kanamycin B treatment
The cellular response to Kanamycin B was revealed by the changes in the levels of gene expression. The differentially expressed genes (DEGs) in response to treatment by Kanamycin B were identi ed by selecting gene expression levels with |fold changes| ≥2 and signi cant differences in p-values of < 0.05 at 0.5 μM or 1 μM. Figure 2A shows a cluster analysis of the DEGs (for a full list of DEGs see additional le 1). There are two main clusters: one with genes that were induced and one with genes that were repressed in response to Kanamycin B treatment. We identi ed 83 or 136 induced DEGs and 112 or 155 repressed DEGs upon treatment with 0.5 μM or 1 μM of Kanamycin B respectively compared to no Page 4/25 treatment ( Figure 2B). There are 65 upregulated and 58 downregulated DEGs in the 1 μM / 0.5 μM group ( Figure 2B). Fewer numbers of DEGs were observed in the 1 μM / 0.5 μM group compared to the 0.5 μM / 0 μM and 1 μM / 0 μM groups suggesting that 0.5 μM or 1 μM Kanamycin B causes smaller cellular responses. We next compared the similarities and differences of the DEGs in response to 0.5 μM or 1 μM Kanamycin B treatment. The numbers of overlapping genes between the groups (0.5 μM / 0 μM, 1 μM / 0 μM and 1 μM / 0.5 μM) are shown in the Venn diagram for upregulated or downregulated DEGs ( Figure   2C, 2D) and overlapping groups listed (additional le 2). There is considerable overlap between 0.5 μM / 0 μM and 1 μM / 0 μM DEG groups; 30 genes and 57 genes were co-induced or co-repress in 0.5 μM and 1 μM of Kanamycin B treatment. All of the DEGs represent 8.8% of the whole genome transcripts. In addition, the results show that the magnitude of the transcriptional responses in DEGs varied signi cantly from gene to gene. For example, some genes are upregulated over 10-fold upon Kanamycin B treatment while the expression of many other genes are induced only by about 2-fold (Table 1 and additional le 1). Functional classi cation of differentially expressed genes (DEGs) To investigate the function of the DEGs, we carried out a Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of DEGs. Table 1 (1 μM / 0 μM), additional le 3 (0.5 μM / 0 μM) and Figure 3 shows the DEGs classi ed into known or predicted functional groups. These genes encode proteins that are known or predicted to be involved in the following cellular functions: microbial metabolism in diverse environments; two-component systems, nitrogen metabolism, butanoate metabolism, Arginine and proline metabolism and Inositol phosphate metabolism. Genes that encode proteins that participate in cellular degradation, including chloroalkane and chloroalkene degradation, naphthalene degradation, chlorocyclohexane and chlorobenzene degradation, uorobenzoate degradation, toluene degradation and degradation of aromatic compounds. Genes that encode proteins involved in oxidative phosphorylation, aminoacyl-tRNA biosynthesis or the TCA cycle. KEGG pathway analysis of DEGs showed that many genes are classi ed in more than one functional group (Table 1). Further analysis of the genes that belong to the ve relatively large functional classes (two-component systems, oxidative phosphorylation, nitrogen metabolism, microbial metabolism in diverse environments and butanoate metabolism) revealed the linkage network of these DEGs ( Figure 4, Table 1), suggesting an overlap between these main functional classes through these DEGs.

Validation by real-time PCR (RT-PCR)
To verify the accuracy and reproducibility of the transcriptome results by an alternative method, we performed quantitation by RT-PCR. We selected 8 genes that were differentially expressed upon treatment with Kanamycin B. A comparative analysis of all the selected genes showed a similar expression pattern in the RT-PCR analysis as observed in RNAseq data ( Figure 6). Thus, the RT-PCR experiments con rm the reliability of the transcriptome sequencing data.

Discussion
We It has been reported that the uptake of aminoglycosides into bacterial cells needs the proton motive force (PMF) that is produced by electron ow through the respiratory chain of oxidative phosphorylation. The PMF is mainly generated by the respiratory complex I that contains membrane proteins and/or Fe-S clusters and oxidoreductases [29]. The mechanism of aminoglycoside uptake and the bacterial cell response to the aminoglycosides is still unclear. Aminoglycoside binding riboswitches have been characterized [7] and a randomly selected Kanamycin B riboswitch has been reported [30]. The transcription levels of some genes associated with oxidative phosphorylation change upon treatment of Kanamycin B in this study. These observations together raise the possibility that Kanamycin B may regulate the expression of oxidative phosphorylation genes through riboregulatory interactions with the transcripts. Further studies will be required to examine this speculation.
In this study, treatment by Kanamycin B induces transcription of nitrate reductase (narK, narG, narI and narH) and nitrite reductase (nirD and nirB) genes (Table 1), which are key genes that are involved in nitrogen metabolism. These reductases participate in the conversion of NO 3to NO 2to NO in cells and are associated with wider essential metabolic processes such as energy production, amino acid metabolism, bio lm formation, antibiotic resistance and bacterial pathogenesis. Nitrogen metabolism is also closely interlinked with bio lm formation that is also linked with the metabolism of glutamic acid, glutamine and arginine. It is widely recognized that NO plays an important role in modulating the architecture of bio lms [31,32]. Other antibiotics have been reported to be able to induce bio lm formation [33]. Therefore, it is possible that Kanamycin B may induce the expression of genes associated with nitrogen metabolism through mimicking the effects of a natural ligand.
Genome-wide transcriptome pro ling of Mycobacterium tuberculosis upon treatment of kanamycin has been reported [28]. Comparison with the functional classi cation and pathways of the DEGs identi ed in the Mtb study with those identi ed in this study indicated more differences than similarities between the two studies. This may due to the following reasons: (1) Mycobacterium tuberculosis is a clinical strain that causes tuberculosis. The genotype of this strain may have evolved to escape the host immune system and the strain itself may have accumulated multiple mutations through the transmission of the tuberculosis. The genotype of the Mtb is fundamentally different from E. coli JM109 in our study. (2) The RNA-seq analysis with Mtb was treated with kanamycin, however the related aminoglycoside kanamycin B was used to treat the E. coli JM109 in this study. Although kanamycin B and kanamycin are analogue compounds, they also have a different resistance pro le between the two organisms. The aminoglycoside riboswitch and many other regulatory RNAs can distinguish structural analogues through different binding a nities [7,34,35]. The RNA-seq is a sensitive technology that readily detects RNAs that respond to structural analogues.

Bacterial strains and growth condition
The strain used in this study was Escherichia coli K-12 derived JM109 strain. JM109 was grown in rich medium LB at 0μM, 0.5μM, 1μM, 2μM Kanamycin B. OD600 was measured using a SpectraMax® M5.

Sequence reads mapping and assembly
The raw reads of fastq format were initially ltered by removing reads containing adaptors, poly-N and low-quality reads. The high-quality clean reads were obtained (reads contain 20% base quality lower than Q20). Q20, Q30, GC-content and sequence duplication level of the clean data were calculated. The analyses were performed by using high quality clean reads.

Differential expression analysis (DEGs) and Functional annotations of DEGs
The expression levels of genes were counted by using HTseq in each sample [37]. The DEGs are ltered to meet the threshold level with a |fold change| ≥ 2, and false discovery rate p-value < 0.05 in each pairwise comparison by using edgeR [38].