A methoxylated flavone from Artemisia afra kills Mycobacterium tuberculosis

Abstract

Tuberculosis, caused by Mycobacterium tuberculosis (Mtb), is a deadly and debilitating disease globally affecting millions annually. Emerging drug-resistant Mtb strains endanger the efficacy of the current combination therapies employed to treat tuberculosis; therefore, there is an urgent need to develop novel drugs to combat this disease. Artemisia afra is used traditionally in southern Africa to treat malaria and recently has shown anti tuberculosis activity. This genus synthesizes a prodigious number of phytochemicals, many of which have demonstrated human health effects. Transcriptomic analysis revealed that A. afra exerts different effects on Mtb compared to A. annua or the well-known antimalarial artemisinin, suggesting other phytochemicals present in A. afra with unique modes of action. A biochemometric study of A. afra resulted in the isolation of a methoxylated flavone (1), which displayed considerable activity against Mtb strain mc²6230. Compound 1 had an MIC of 312.5 μg/mL and yielded no viable colonies after 6 days of treatment. In addition, 1 was effective in killing hypoxic Mtb cultures, with no viable cultures after 2 days of treatment. This suggested that A. afra is a source of potentially powerful anti-Mtb phytochemicals with novel mechanisms of action.

Figure 1.. A. afra, A. annua, and artemisinin have distinct transcriptomic impacts on Mtb.

Aerobically growing Mtb was treated with each extract or compound at lethal doses for four hours or at inhibitory and sub-inhibitory doses for 24 hours. Untreated cultures were harvested at the same two time-points and RNAseq was used to generate transcriptomic profiles. All conditions were tested in quadruplicate. A. PCA was done on the read count tables from each sample in each condition, revealing that each treatment clustered separately. B. Expression of each gene in each treatment was compared to that in the time-matched control, and differentially expressed genes were subject to Gene Set Enrichment Analysis. Gene sets were GO Biological Process gene lists obtained from AmiGO. “Activated” gene sets had higher expression in the treated samples compared to the controls while “suppressed” genes sets had lower expression. “GeneRatio” indicates the proportion of genes in the set that were differentially expressed in the indicated condition. “p.adjust” is the P value of the overrepresentation of genes within the set among the differentially expressed genes, after correction for multiple comparisons.

Figure 2.. Time-kill curves of Mtb treated with an A. afra extract and constituent fractions and subfractions.

In all cases, treatment was applied to liquid cultures at Day 0 and aliquots were plated on drug-free solid media at the indicated timepoints. Colonies were enumerated after 3–4 weeks of growth in order to quantify the number of colony forming units (CFU) that had survived drug treatment at each timepoint. All no-drug samples were given DMSO as a vehicle control. A. Mtb was exposed to crude DCM extract of A. afra and the three flash chromatography fractions with the lowest MICs (see Table 1). B. Mtb was exposed to crude DCM extract of A. afra and subfraction HAB9. No viable colonies were recovered from cultures treated with HAB9 after six days. The limit of detection is indicated with a dashed line. C. Mtb was sealed in vials with 17 mL of culture and 11 mL of headspace. The cultures became hypoxic approximately eight days after sealing as evidenced by methylene blue decoloration. After the cultures had been sealed for 14 days, subfraction HAB9 or the vehicle control DMSO were injected by needle to prevent introduction of oxygen. This was considered day 0 of treatment. No colonies were recovered from cultures treated with HAB9 after two or six days of treatment. The limit of detection is indicated with a dashed line.

Figure 3.. Biochemometric identification of 1.

(A) S-line plot illustrating the covariance (y-axis) and correlation (coloration) between features and classification of the OPLS-DA model. Greater values in the covariance and correlation represent features that hold a larger influence in the model generation and discrimination between active and inactive samples. Compound 1 is highlighted, along with ions from another A. afra subfraction (HAB4). (B) Extracted ion chromatogram of 1, m/z 299.0841 from subfraction HAB9.

Figure 4.

Structure of compound 1, 5-hydroxy-4’,7-dimethoxyflavone.