Clinical strains of Mycobacterium tuberculosis exhibit differential lipid metabolism-associated transcriptome changes in in vitro cholesterol and infection models

Abstract

Many studies have identified host-derived lipids, characterised by the abundance of cholesterol, as a major source of carbon nutrition for Mycobacterium tuberculosis during infection. Members of the Mycobacterium tuberculosis complex are biologically different with regards to degree of disease, host range, pathogenicity and transmission. Therefore, the current study aimed at elucidating transcriptome changes during early infection of pulmonary epithelial cells and on an in vitro cholesterol-rich minimal media, in M. tuberculosis clinical strains F15/LAM4/KZN and Beijing, and the laboratory H37Rv strain. Infection of pulmonary epithelial cells elicited the upregulation of fadD28 and hsaC in both the F15/LAM4/KZN and Beijing strains and the downregulation of several other lipid-associated genes. Growth curve analysis revealed F15/LAM4/KZN and Beijing to be slow growers in 7H9 medium and cholesterol-supplemented media. RNA-seq analysis revealed strain-specific transcriptomic changes, thereby affecting different metabolic processes in an in vitro cholesterol model. The differential expression of these genes suggests that the genetically diverse M. tuberculosis clinical strains exhibit strain-specific behaviour that may influence their ability to metabolise lipids, specifically cholesterol, which may account for phenotypic differences observed during infection.

Keywords: Mycobacterium tuberculosis; cholesterol; clinical strains; lipid metabolism; transcriptome.

Figure 1.

Differential expression of Mycobacterium tuberculosis(M. tuberculosis) lipid-associated virulence genes in Middlebrook 7H9 culture medium and during 48-h pulmonary epithelial cell infection. Selected genes have been identified as being vital in lipid metabolism in M. tb during infection. (A-E) qRT-PCR bar graphs depicting the gene expression values as a log2-fold change for the various lipid-associated virulence genes in various clinical strains of M. tb in standard Middlebrook 7H9 broth and during 48-h infection of PECs. Statistical significance is represented by one asterisk (P ≤ 0.05) and two asterisks (P ≤ 0.01) as per a non-parametric Mann-Whitney t-test. Relative gene expression was quantified using the 2^−ΔΔCt method.

Figure 2.

Graphical representation of the average biological assay 1, 2, and 3 bacterial colonies (CFU/ml) from clinical M. tuberculosis strains, Beijing and F15/LAM4/KZN, and the laboratory strain H37Rv, post-infection in pulmonary epithelial cells at 4 and 48 h, respectively.

Figure 3.

Growth curves and bacterial CFU/ml of clinical M. tuberculosis strains. (A) Growth curve for M. tuberculosis strains cultivated in 7H9; (B) growth curve for M. tuberculosis strains cultivated in the in vitro lipid model; (CandD) the corresponding CFU/ml for each type of environment. For the growth curve, growth was plotted on a log₁₀ scale by measurement of OD_600nm and recorded every 2 days until a stationary phase was reached. The bacterial CFU/ml counts were determined by plating serially diluted M. tuberculosis on Middlebrook 7H11 agar for a total of 21 days. The curves illustrated correspond to the mean ± SD of three biological replicates. The M. tuberculosis H37Rv laboratory strain and the M. tuberculosis KZN1435 clinical strains grew optimally in both culturing conditions, while the M. tuberculosis KZN605 and Beijing strains presented the slowest growth. * denotes differences that are statistically significant (P < 0.05) as per a non-parametric Mann-Whitney t-test.

Figure 4.

(A) Compilation of functionally categorised genes involved in the breakdown of cholesterol and its relative expression. The transcriptomic data of M. tuberculosis clinical strains KZN605 and Beijing were compared with the reference H37Rv laboratory strain in an in vitro lipid model and while additionally comparing conditions of M. tuberculosis strains (H37Rv, Beijing, KZN605) grown in cholesterol-supplemented minimal media against the same strains grown in 7H9/10% OADC medium. Genes were grouped together according to the varying functions each plays in the multiple stages of cholesterol degradation. (B) Characterisation of a fraction of the mycolic acid synthesis pathway in M. tuberculosis strains. Gene expression implicated in the mycolic acid synthesis pathway in M. tuberculosis strains KZN605, Beijing and H37Rv in an in vitro lipid model compared with the KZN605, Beijing and H37Rv strains in Middlebrook 7H9 medium. (C) The simultaneous depiction of triacylglycerol synthesis and degradation in M. tuberculosis strains and the concomitant differential expression of genes involved in these processes. Gene expression implicated in the triacylglycerol synthesis and degradation pathway in M. tuberculosis strains KZN605, Beijing and H37Rv in an in vitro lipid model compared with the KZN605, Beijing and H37Rv strains in Middlebrook 7H9 medium. (D) Gene expression implicated in the sulfolipid-1 synthesis pathway in M. tuberculosis. The differential expression of genes involved in the production of sulfolipids in M. tuberculosis strains H37Rv, Beijing and KZN605 in an in vitro lipid model compared with the same strains in 7H9 medium. The results are presented as log2-fold change in expression in a heatmap format. A colour key is provided to indicate log2-fold changes, and grey boxes indicate genes that were not detected from the transcriptome data.

Figure 5.

Differential expression of selected M. tuberculosis (M. tb) lipid-associated virulence genes in Middlebrook 7H9 culture medium and cholesterol-supplemented media for functional confirmation of RNA-sequencing analysis. (A-D) qRT-PCR bar graphs depicting the gene expression values as a log2-fold change for the various lipid-associated virulence genes in various clinical strains of M. tuberculosis in standard Middlebrook 7H9 broth and lipid media. Relative gene expression was quantified using the 2^−ΔΔCt method. The fold change for qRT-PCR analysis was calculated using the Ct values of the lipid genes grown in 0.01% cholesterol versus Ct values of lipid genes in clinical strains cultured in 7H9 broth. Statistical significance is represented by one asterisk (P ≤ 0.05) and two asterisks (P ≤ 0.01) as per a non-parametric Mann-Whitney t-test. Error bars represent the ± SD for three biological replicates for each strain. The 16S rRNA gene was used as an internal standard to normalise gene expression data.