Generally, the clinical antibiotic resistance in human pathogens belongs to acquired resistance, with only a small fraction under the category of innate resistance. The origins of these acquired resistance genes have been traced to antibiotic-producers in natural environments. In antibiotic-producing microbes, self-resistance is a prerequisite for the synthesis of antibiotics. Antibiotic biosynthetic gene clusters (BGCs) contain one or more resistance genes to achieve self-protection, and these genes are considered to be the reservoirs of resistance genes, which may transfer to human pathogens by conjugation, transformation or transduction. Consequently, the elucidation of self-resistance mechanisms from antibiotic-producing microbes will not only reveal the action model of antibiotics and guide the discovery of new natural products, but also provide key clues for the studies of clinical antibiotic resistance.
To avoid suicide, antibiotic-producers have developed several mechanisms, including efflux pumps, chemical modification, prodrugs, compound sequestration, (sub)cellular location, target modification, and damage repair to shield the toxicity of antibiotics, thereby achieving self-protection. Different strategies are adapted depending on the structure of antibiotics, molecular target and producer species. For instance, within the enediyne producers, apoproteins are known to afford self-protection to the producers by binding the nine-membered enediyne chromophore, whereas the strategies of chemical modification guided by self-sacrifice proteins and sequestration mediated by drug binding proteins are utilized for the detoxification of ten-membered enediynes. To prevent self-harm from bleomycin, its producer Steptomyces verticillus employs bleomycin N-acetylation and sequestration mechanisms to protect itself. The antitumor agent mitomycin C producer, Steptomyces lavendulae, has developed several mechanisms, including prodrug, efflux pump, drug sequestration, and reoxidation of the active reduced mitomycin C to ensure self-resistance.
2. Resistance Widespread in Nature
Self-resistance determinants are not confined to the antibiotic producers. Instead, some of them are widely prevalent in the clinical pathogens and environmental bacteria
[1]. Recent studies of self-resistance mechanisms against enediyne antitumor antibiotics revealed that an unprecedented sequestration mechanism for the anthraquinone-fused enediynes has been evolved in their producers and the homologs of these resistance elements are widely distributed in nature
[2]. Within the gene cluster of tiancimycin (
Figure 1), resistance genes
tnmS1,
tnmS2, and
tnmS3 play a role in the sequestration of tiancimycin. The homologs of TnmS1, TnmS2, and TnmS3 are widespread in anthraquinone-fused enediynes producers and other bacteria, from different body sites, including the human microbiome
[2]. The expression of homologous genes from the gene clusters encoding enediyne biosynthesis has been reported to endow
E. coli BL21(DE3) with cross-resistance to anthraquinone-fused enediynes, while the homologs from human microbiome confer specific resistance to tiancimycin A
[2]. These results further highlight that the resistance elements responsible for anthraquinone-fused enediynes sequestration are widely distributed in nature, although little is known about how these resistance genes disseminate in the environment. Similarly, homologous resistance genes encoding following enzymes are widespread in nature and perform conserved biological functions, such as GyrI-like cyclopropane hydrolases that mediate cyclopropyl moiety opening of DNA-alkylating agents YTM/CC-1065
[3], AlbA-like drug-binding proteins that guide resistance to albicidin
[4][5], and NapW-like short-chain dehydrogenase/reductase that catalyze hemiaminal pharmacophore inactivation for tetrahydroisoquinoline antibiotics (
Figure 1)
[6]. In addition to antibiotic resistance, non-antibiotic drug resistance is also widespread. As direct evidence for this conception, Acbk-like kinase, inactivating a clinically used non-antibiotic antidiabetic drug acarbose by phosphorylation, is widely distributed in the human gut and oral microbiome (
Figure 1)
[7]. The specific kinase AcbK derived from
Actinoplanes sp. SE50/110, is located within the gene cluster for acarbose synthesis. It phosphorylates acarbose at the O6A hydroxyl and serves as the self-resistance mechanism for acarbose production
[8]. Recently, Donia et al. performed a metagenomics-based investigation of the human microbiome and found that homologues of AcbK are widespread in the bacteria from the human gut and oral microbiome and provide acarbose resistance, indicating the phosphorylation strategy of acarbose has disseminated in the human microbiome as a resistance mechanism
[7]. Therefore, research on these widely distributed resistance elements will contribute to predicting and combating clinical drug resistance.
Figure 1. Representative natural products that resistance determinants are widespread in nature.
3. Resistance-Guided Natural Products Discovery
The rapid development of bioinformatics tools and genome sequencing technologies has brought a revolution in the discovery of natural products, leading to a transformation from traditional bioactivity-guided fractionation to modern genome-based target mining
[9][10]. The enormous amount of genome data that is now available has revealed that microorganisms harbor more natural product BGCs than those observed under laboratory cultivation conditions, and most of them gene clusters encoding unknown products
[11][12]. However, how to deal with the increasing BGCs and how to mine the desired products from the huge resources have become a major focus. Recently, researchers found that the self-resistance genes co-localized with BGCs can be used as a potential tool to link BGCs with molecular targets for mining natural products with desired activity (
Figure 2)
[13][14][15]. For example, Tang and coworkers successfully discovered a natural herbicide with a new mode of action from
Aspergillus terreus by a putative self-resistance gene,
astD, encoding a dihydroxyacid dehydratase (DHAD) homolog
[16]. The DHAD is an essential enzyme that catalyzes the last step of branched-chain amino acid biosynthesis, and is therefore effectively targeted for herbicide development
[16]. However, no compounds that target this enzyme have been reported in planta. Fungal genomes scanning of a DHAD homologue revealed that a BGC encoding a sesquiterpene cyclase homologue and a DHAD homologue was present in the genome of
Aspergillus terreus. Subsequent experiments demonstrated that the aspterric acid encoded by this BGC is indeed a competitive inhibitor of DHAD and effectively functions as a herbicide, and the DHAD variant AstD functions as a self-resistance enzyme in the BGC for aspterric acid
[16]. Similarly, the Müller group discovered a novel group of topoisomerase inhibitors, including pyxidicycline A and B, by putative self-resistance genes encoding topoisomerase-targeting pentapeptide repeat protein
[17]. Wright et al. identified the caseinolytic protease (ClpP) inhibitor clipibicyclene from
Streptomyces cattleya using ClpP as putative antibiotic resistance gene
[18]. Ge et al. discovered a novel tetracycline, hainancycline, by using the common tetracycline antibiotics resistance enzyme TetR/MarR-transporter as probe
[19].
Figure 2. Structures of representative natural products that discovered by using antibiotic resistance gene as a probe.
In addition to discovering natural products with desired activity, self-resistance genes can also be used in determining the biomolecular target of known antibiotics (
Figure 3). For instance, the β-lactone obafluorin isolated from
Pseudomonas fluorescens ATCC 39502 shows potent antibacterial activity against both Gram-positive and Gram-negative pathogens
[20]. The mechanism of action of obafluorin, however, was unknown as this molecule was reported to cause an unusual cell-elongation phenotype compared to other β-lactone antibiotics. During comparative genomic analysis of obafluorin BGCs, an open reading frame,
obaO, was identified and speculated to be an immunity gene
[20]. ObaO was shown to be a homologue of threonyl-tRNA synthetase and conferred resistance to obafluorin-sensitive strains and obafluorin producer when expressed. Subsequently, in vitro enzyme assays demonstrated that the obafluorin did indeed fully inhibit
E. coli threonyl-tRNA synthetase with an IC
50 of 92 ± 21 nM, thus indicating the target of this compound
[20]. In another example, harzianic acid is a N-methylated tetramic acid isolated from
Trichoderma harzianum in 1994. Although it displays excellent antifungal activity, including against plant pathogens
Sclerotinia sclerotiorum and
Rhizoctonia solani, the molecular target of harzianic acid remains unknown
[21]. Recently, Tang et al. discovered that the harzianic acid is an inhibitor of acetohydroxyacid synthase (AHAS, the first enzyme on branched-chain amino acid biosynthesis pathway), which was guided by a truncated AHAS homolog resided within the BGC that was demonstrated to be the self-resistance enzyme
[22]. A similar biomolecular target discovery scenario is also observed in determining the mode of action of polyketide rumbrins, which further revealed their promising potential to be HIV inhibitors
[23].
Figure 3. Structures of representative natural products that determine the biomolecular target through antibiotic resistance genes.
Although the above examples have successfully confirmed the potential of self-resistance genes in directed genome mining for natural products with known or predicted biomolecular targets, the development of compounds with novel mechanisms of action is also an urgent need to solve the ongoing antibiotic crisis. Recently, Wright et al. reported that the method of combining the absence of known self-resistance genes with phylogenetic analysis of biosynthetic genes could be effective in finding natural products with new modes of action
[24]. They applied this approach to the glycopeptide family of antibiotics and successfully discovered a novel functional class of glycopeptide antibiotics composed of complestatin and corbomycin (
Figure 2), which have a new mechanism of action that inhibits peptidoglycan re-modelling. This research outcome again indicated that self-resistance determinants are useful for prioritizing BGCs than just function in the self-protection. Other examples of employing a self-resistance determinant in natural products discovery are reviewed elsewhere
[13][25][26]. Taken together, self-resistance genes can be a bridge between the bioactivity-guided and genome-based methods for natural products discovery. Studying the complex self-resistance strategies from a temporal-spatial shielding perspective will allow researchers to further understanding the evolutionary relationship between natural product biosynthesis and resistance, thereby facilitating discovery of new drug candidates with high activity.