Characterization of the Self-Resistance Mechanism to Dityromycin in the Streptomyces Producer Strain

The World Health Organization has identified antimicrobial resistance as a substantial threat to human health. Because of the emergence of pathogenic bacteria resistant to multiple antibiotics worldwide, there is a need to identify the mode of action of antibiotics and to unravel the basic mechanisms responsible for drug resistance. Antibiotic producers’ microorganisms can protect themselves from the toxic effect of the drug using different strategies; one of the most common involves the modification of the antibiotic’s target site. In this work, we report a detailed analysis of the molecular mechanism, based on protein modification, devised by the soil microorganism Streptomyces sp. strain AM-2504 to protect itself from the activity of the peptide antibiotic dityromycin. Furthermore, we demonstrate that this mechanism can be reproduced in E. coli, thereby eliciting antibiotic resistance in this human commensal bacterium.

Dityromycin is a cyclic decapeptide antibiotic (Fig. 1A) discovered as a secondary metabolite produced by Streptomyces sp. strain AM-2504 (5,6). Its structure is almost identical to that of GE82832, a translocation inhibitor produced by Streptosporangium spp. (7)(8)(9). Both molecules interact with the same ribosomal site and display the same mechanism of action, even if they are produced by different microorganisms. The interaction of dityromycin with the ribosome is peculiar compared to other ribosomal inhibitors, since it interacts exclusively with the ribosomal protein S12 without a direct interaction with the 16S rRNA ( Fig. 1B and C) (8,9). Indeed, the binding of dityromycin to the ribosome is mediated by its interaction with five amino acids at the ribosomal protein S12, namely, arginine 30, valine 32, arginine 55, histidine 76, and valine 78 (Fig. 1C). Dityromycin inhibits protein synthesis by blocking the translocation of tRNA through the ribosome, while it has almost no effect on the accommodation of aminoacyl-tRNA into the P-site or the A-site, in the context of either the 30S ribosomal subunit or the 70S ribosome (7). Recent studies have shown that dityromycin blocks the EF-G-dependent translocation of peptidyl tRNA and mRNA without preventing the ribosomal binding of the elongation factor. In particular, dityromycin hampers the interaction between domain III of EF-G and protein S12, which is located near the decoding center of the ribosome, at the interface of two subunits (9,10).
A strict requirement to ensure that the antibiotic producer cell protects itself when the active metabolites are produced, accumulated, and released is to possess a self-resistance mechanism(s) (11). Microorganisms can adopt several strategies to develop self-resistance, such as (i) chemical modification of the target, as in the case of rRNA methylation or mutation (12,13); (ii) activation of efflux pumps (14,15); (iii) sequestration of the natural product by proteins keeping the active compound in a bound and inactive state (16,17); (iv) production of the antibiotic in a prodrug form, activated through selective cleavage by peptidases found in the periplasmic space, and subsequently released outside the cell (18,19); and (v) oxidative inactivation of the matured prodrug following export (20).
Biochemical and genetic similarities demonstrate that such resistance mechanisms have prefigured those found subsequently in antibiotic-resistant pathogens, which have acquired specific genes or gene clusters originally developed by antibiotic producers (21).
In this work, we have investigated the self-resistance mechanism developed by Streptomyces sp. strain AM-2504 toward the antibiotic activity of its natural product dityromycin. To prevent self-toxicity, this microorganism has evolved a strategy based on the production of a modified and self-resistant ribosomal protein variant. Our structure of the 30S ribosomal subunit (gray) and localization of dityromycin (red) in contact with the ribosomal protein S12 (green). A, P, and E tRNA binding sites are indicated in blue, yellow, and violet, respectively. (C) Close-up of dityromycin (red) bound to S12 (green) missing the C-terminal domain. The amino acid residues of S12 involved in the interaction with the antibiotic are also indicated. (Modified from reference 9.) Fabbretti et al.
approach consisted in analyzing three aspects, namely, identification and biochemical characterization of the self-resistance mechanism, characterization of the amino acid residues of the ribosomal protein S12 involved in resistance by the heterologous expression of site-directed mutants in Escherichia coli, and a phylogenetic analysis of the dityromycin binding pocket in the ribosomal protein S12.

RESULTS
Identification of the mechanism conferring self-resistance. To determine the self-resistance mechanism of Streptomyces sp. strain AM-2504 to dityromycin, we set up a simple experiment evaluating whether the resistance mechanism relied on possible modifications/secretion of the drug (prodrug or efflux pumps) or was based on mutations/methylations/modifications of the target.
A translational test programmed with poly(U) was used to compare cell extracts prepared from either E. coli (S30 E.coli ) or Streptomyces sp. strain AM-2504 cells (S30 Str ) in the presence or absence of the antibiotic. Dityromycin, an antibiotic targeting the translation elongation phase, completely inhibited protein synthesis with S30 E.coli at concentrations ranging between 1 and 10 M, whereas it did not show any effect on the poly(U) translation carried out with S30 Str (Fig. 2A). The lack of inhibition with the S30 Str extract clearly excluded prodrug or efflux pumps as a possible self-resistance mechanism.
To test whether this self-resistance mechanism was caused by protective mutations in the ribosomes or by an enzymatic activity able to modify/deactivate the antibiotic, 70S ribosomes were separated by ultracentrifugation from the post-ribosomal supernatant (S100) and the fractions were tested by crossed in vitro poly(U) translation assays. In particular, the effect of dityromycin on Streptomyces 70S ribosomes (70S Str ) was assayed in the presence of E. coli S100 (S100 E.coli ). Vice versa, Streptomyces S100 extract (S100 Str ) was tested with E. coli 70S ribosomes (70S E.coli ). The results of this experiment (Fig. 2B) show that the reaction mixtures consisting of S100 Str plus 70S E.coli and S100 E.coli plus 70S E.coli were totally inhibited by dityromycin, whereas those formed by S100 E.coli plus 70S Str and S100 Str plus 70S Str were not. These results strongly indicate that the modifications causing the self-resistance of the producer strain lie on the ribosome.
Since the target of dityromycin is the ribosomal protein S12 (9), 30S ribosomal subunits from both E. coli and Streptomyces sp. strain AM-2504 were purified. Increasing concentrations of these isolated particles were used in a pulldown assay conducted in the presence of a fixed concentration of the antibiotic. After ultracentrifugation at 300,000 ϫ g, the supernatants were tested in a translational system based on E. coli cell extracts programmed with poly(U) to detect the residual presence of unbound dityro- . Streptomyces 70S incubated with E. coli S100 or Streptomyces S100 is indicated in blue or black, respectively. E. coli 70S incubated with either E. coli S100 or Streptomyces S100 is indicated in red or green, respectively. (C) Residual translation inhibition activity of the supernatants obtained from ultracentrifugation of 30S ribosomes preincubated with dityromycin. After centrifugation of increasing amounts of E. coli 30S (black) and Streptomyces sp. strain AM-2504 30S (red), the residual inhibition activity of the supernatants was determined in a poly(U) translation test (further details are provided in Materials and Methods). mycin (see Materials and Methods). As reported in Fig. 2C, the supernatant resulting from the pulldown of E. coli 30S subunits incubated with dityromycin did not interfere with poly(Phe) synthesis. Conversely, the supernatant obtained after centrifugation of Streptomyces 30S subunits incubated with dityromycin inhibited the reaction. This result indicates that dityromycin is not bound by the Streptomyces 30S subunits and is not pulled down during ultracentrifugation, even at high concentrations of ribosomes. In contrast, dityromycin stably interacts with the E. coli 30S subunits and is thus removed from the supernatant during the centrifugation step (Fig. 2C). In conclusion, the present results point to a self-resistance mechanism based on a mutation/modification residing in the small ribosomal subunit.
Sequencing of the S12 gene reveals three amino acid substitutions in the dityromycin binding site. As shown in the crystallographic structure of Fig. 1, dityromycin binds the 30S subunit by interacting exclusively with five critical positions of the ribosomal protein S12, namely, Arg30, Val32, Arg55, His76, and Val78 (Fig. 1C). To clarify the role played by S12 in reducing the binding affinity of dityromycin to the 30S Str subunit (Fig. 2), we sequenced the rpsL gene, encoding the S12 protein of Streptomyces sp. strain AM-2504. In addition, we sequenced the entire genome of the producer strain, thus obtaining its first annotated draft genome sequence (22). The comparison of Streptomyces sp. strain AM-2504 genome with the sequences available in the NCBI genome data bank allowed us to verify that the strain named AM-2504 described in this study exhibits 98.6% nucleotide identity with Streptomyces kasugaensis. Therefore, it is conceivable that Streptomyces sp. strain AM-2504 can be classified as S. kasugaensis.
DNA sequence data demonstrate that three out of five amino acids located in the dityromycin binding site of S12, namely, Val32Thr, Arg55Lys, and Val78Ile, are different from those found in the model organism E. coli (Fig. 3). These substitutions are likely the molecular determinants of the self-resistance mechanism to the antibiotic, preventing dityromycin binding to the ribosome (Fig. 2C).
Expression of the S12 mutant proteins in E. coli confers resistance to dityromycin. To verify whether these mutations are responsible for the self-resistance of the producer strain, single substitutions at Val32, Arg55, and Val78 were introduced into the E. coli rpsL gene, and the mutant genes were provided in trans to the E. coli BL21 strain grown in the presence of increasing concentrations of dityromycin. The experimental system was based on gene expression from the high-copy-number plasmid pET11a, which contained the rpsL gene isolated from E. coli MC4100 carrying the original mutation Lys42Arg, which confers resistance to the antibiotic streptomycin. As shown in Fig. 4A, E. coli cells expressing a wild-type copy of S12, in the absence of the plasmid, are sensitive to streptomycin, but they become resistant once transformed. Moreover, the presence of either Val32Thr, Arg55Lys or Val78Ile substitutions did not affect the streptomycin-resistant phenotype. This demonstrates that the ribosomes effectively incorporate the S12 protein encoded by the plasmid and respond to the mutant genetic background supplied in trans. Furthermore, the results shown in Fig. 4B demonstrate that the ribosomal protein S12 carrying the Val32Thr or Arg55Lys substitution confer dityromycin resistance to E. coli, whereas the cells expressing the point mutation Val78Ile proved to be susceptible to this antibiotic.

DISCUSSION
The growing number of bacterial pathogens uniquely adapted to survive in the presence of one or more antibiotics is an extremely serious threat to global health. The term "antibiotic resistome," a large repertoire of genes that has been originally developed by bacteria and is currently remodeled by the selective pressure imposed by drug use/abuse, has been coined to describe the genetic determinants involved in antibiotic resistance (23,24).
Resistance to antibiotics is tightly linked to their production, since microorganisms producing these molecules must protect themselves against their action. In many cases, mechanisms of self-resistance have been shown to be mechanistically similar to the ones identified in clinical pathogens (25). Resistance genes are often carried by the same gene cluster of the antibiotic biosynthetic enzymes, and their expression is temporally coupled. Indeed, antibiotic producer cells need to express resistance determinants prior to, or concomitantly with, synthesis of the antibiotic molecule (26,27). In recent years, the analysis of several genomes suggested that microorganisms produce only a fraction of the secondary metabolites that they encode, and new approaches have been used to identify biosynthetic gene clusters whose expression can be induced. The colocalization of resistance genes and biosynthetic genes in the genomes of underexploited groups of secondary metabolite producers, for example, has been used in pursuit of new biosynthetic pathways. This genome mining approach, guided by the search for self-resistant determinants, has been successful for topoisomerase inhibitors whose biosynthetic gene cluster has been identified by looking for putative pentapeptide repeat proteins, which are known to confer self-resistance to these molecules (28).
More than half of the antibiotics currently used inhibit protein synthesis by binding to the functional centers of the ribosome. Recently, important insights into the mechanism of these antibiotics have been obtained by X-ray crystallography and cryoelectron microscopy (Cryo-EM) analysis, opening the possibility of developing new molecules by structure-based drug design (29)(30)(31)(32)(33)(34)(35)(36).
In this work, we characterized the mechanism of self-resistance of the dityromycin producer Streptomyces sp. strain AM-2504. Using E. coli and Streptomyces-based in vitro translation assays in combination with an ultracentrifugation pulldown assay, we demonstrated that the resistance determinants are localized in the Streptomyces 30S ribosomal subunit. Since the target of dityromycin is the ribosomal protein S12, we sequenced the gene coding for Streptomyces S12 and compared it with its E. coli homolog. Three amino acids in the Streptomyces protein (out of the five involved in the interaction with the antibiotic) differed from the E. coli protein sequence. Therefore, we suggest that the molecular basis underlying the self-resistance mechanism could be the reduced affinity of the antibiotic toward its target due to the amino acid substitutions on the ribosomal protein S12. To confirm this hypothesis, we introduced the observed amino acid substitutions into a plasmid-borne copy of E. coli rpsL gene harboring another mutation conferring resistance to the antibiotic streptomycin. The observed streptomycin resistance phenotype provides explicit, albeit indirect, evidence that the plasmid gene encoding ribosomal protein S12 is effectively incorporated into the 30S subunits. Even though the setup of the assay did not allow strict quantification of the number of copies of mutant protein incorporated, it showed unequivocally that this incorporation took place and that plasmid-encoded S12 bearing two specific mutations in the dityromycin binding site conferred resistance to the antibiotic. The results shown in Fig. 4B clearly indicate the effects of amino acid replacements at the three key sites of ribosomal protein S12 involved in the binding of the antibiotic. In particular, at position 32, the substitution of the nonpolar aliphatic valine with the polar amino acid threonine interferes with the hydrophobic interaction established with the dityrosine group of dityromycin according to the crystal structure of the dityromycin-30S complex. It has been pointed out by crystallographic analysis that the arginine at position 55 packs against the N,N-dimethyl valine of the antibiotic (9). Our results strongly indicate that the replacement of a positively charged side chain, as in the case of lysine, is not sufficient to maintain the correct structure of the binding site, which presumably requires the chemical properties of the guanidinium group of arginine. This finding is consistent with the high specificity demonstrated by the arginine side chain in the interaction of proteins with their substrates (37,38). On the other hand, we demonstrate that the substitution of Val78 with isoleucine does not interfere with the interaction of dityromycin with S12.
To identify the degree of conservation of the amino acid residues responsible for dityromycin resistance, we aligned the S12 sequences to 72 Streptomyces spp. randomly selected from the NCBI database. The alignment shows that, of the three residues considered, only Ile78 (Val in E. coli) is conserved in all Streptomyces sequences analyzed. The other two key residues are quite rare, with Thr32 (Val in E. coli) reported only in 7 out of 72 genomes (Streptomyces antioxidans, S. autoliticus, S. gilvosporeus, S. iranensis, S. malaysiensis, S. natalensis, and S. violaceusniger) and Lys55 (Arg in E. coli) in only two strains, S. gilvosporeus and S. natalensis. Therefore, the results of this comparative analysis show that only 2 of 72 Streptomyces strains share the same S12 residues conferring resistance to dityromycin as Streptomyces sp. strain AM-2504 (see Fig. S1 in the supplemental material). When the comparative analysis was extended to 30 bacterial strains randomly covering all relevant branches of the phylogenetic tree, it was found that all selected bacteria display a valine at position 32, embedded in a stretch of four amino acids (RGVC), which is highly conserved and it is part of the "core" of a pseudo beta-barrel of S12 (Fig. S2A  and S2B). The other two residues are less conserved, since Lys55 and Ile78 are found in 4 strains and 11 bacterial strains, respectively.
To our knowledge, the results of this work represent the first example of a selfresistance mechanism mediated by a ribosomal protein in the producer strain. Resistance mechanisms identified by isolation of single mutations are extremely rare among ribosomal proteins. Point mutations in rpsL, the gene encoding S12, have been identified as the key factor conferring streptomycin resistance (39) and at the same time activate the expression of cryptic antibiotics, like the blue-colored actinorodhin (40,41). This intriguing finding, which is the first reported evidence of an effect induced on gene expression by a point mutation located on a ribosomal protein, is still lacking a detailed explanation at the molecular level. A few other examples include protein L3, where a single mutation has been shown to confer resistance to tiamulin, the semisynthetic derivative of pleuromutiline, reducing its binding affinity for the ribosome and mutations in L16 conferring resistance to the oligosaccharide antimicrobial agents avilamycin (42) and evernimicin (43). In Pasteurella multocida, a bacterium responsible for zoonotic diseases, a combination of amino acid changes on ribosomal protein S5 (point mutation at Ser32 and amino acid deletion at Phe33) and a single transversion in helix 34 of 16S rRNA have been identified as the molecular determinants of spectinomycin resistance (44).
In summary, we can conclude that Streptomyces sp. strain AM-2504 has developed a successful strategy to protect its own ribosomes from the effect of dityromycin, combining the effects of three amino acid substitutions, which occur as isolated point mutations in other Actinomyces strains and are compatible with the structural function of ribosomal protein S12. Although this resistance mechanism does not fall into the category of antibiotic resistance transfer and spreading, in a strict sense, the characterization of this strategy at the molecular level adds a new piece to the complex puzzle underlying the coevolution of antibiotic production and self-resistance mechanisms.
Culture conditions of Streptomyces sp. strain AM-2504. For the production and isolation of spores, Streptomyces sp. strain AM-2504 was grown in ISP2 medium for 7 to 10 days at 28°C. The white aerial mycelium, consisting of spores, was stripped from the upper layer of the plate using a spatula and dispersed into 1 ml of sterile water. The spore suspension was filtered through a wad of absorbent cotton, and the spores were concentrated by centrifugation at 4,000 rpm for 15 min. The spores were resuspended in 100 l of 20% glycerol and stored at -20°C.
Genomic DNA extraction. Streptomyces spores (20 l) were inoculated in 20 ml of JM medium at 28°C for 18 h with aeration. Cells were collected by centrifugation, and genomic DNA was extracted using the Invisorb spin plant minikit (Invitek).
DNA amplification and sequence of rpsL from Streptomyces sp. Oligonucleotide primers GSF1 and GSR1 (Table 1), designed to cover the coding region of the rpsL gene from Streptomyces, were used to amplify by PCR the target gene using genomic DNA as the template. The DNA sequence of the amplicon, covering approximately 95% of the coding region, was analyzed to identify the amino acid variants carried by the antibiotic producer strain.
Site-directed mutagenesis. Mutations in the E. coli rpsL gene were obtained using the QuikChange XL kit (Stratagene) and, as the template for the PCR, the plasmid vector pET11a carrying the coding sequence of S12, derived from E. coli MC4100 (rpsL150 [Str r ]). The DNA sequences of the oligonucleotide primers used to introduce codon variants are indicated in Table 1. E. coli DH5 cells were transformed with the mutagenized plasmids, and Sanger DNA sequencing confirmed the presence of mutated nucleotides.
Preparation of E. coli and Streptomyces S30 cell extracts. E. coli MRE600 cells were grown at 37°C in LB medium until they reached mid-log phase, while Streptomyces cells were grown at 28°C in T1 medium for 45 h. Both cell types were disrupted by grinding with precooled alumina in a chilled mortar and processed following a previously described protocol (45). Both S30 extracts were extensively dialyzed against a buffer containing 10 mM Tris-HCl (pH 7.7), 10 mM Mg acetate, 60 mM NH 4 Cl, and 0.5 mM dithiothreitol (DTT) and stored in small aliquots at -80°C.
Isolation of ribosomes and ribosomal subunits. The crude extract S30 was subjected to ultracentrifugation at 100,000 ϫ g for 17 h at 4°C. The supernatant, called S100, was stored at -80°C, while the pellet, consisting of 70S ribosomes, was resuspended and split in two aliquots, one for in vitro assays and one for preparation of 30S and 50S ribosomal subunits. 70S ribosomes were dissociated into subunits at 1 mM Mg 2ϩ concentration by ultracentrifugation at 25,000 rpm for 17h at 4°C in a swing bucket rotor (SW28), on a 10% to 30% sucrose density gradient (45).
In vitro translation assays. Cell-free systems programmed with either poly(U) or 027 model mRNA were used to investigate the effect of dityromycin on the bacterial translational apparatus. Translation activity tests were performed using either S30 extracts or S100 extracts and 70S ribosomes, or crisscrossed combinations of E. coli and Streptomyces sp. S100 extracts and 70S ribosomes (46).
Ribosome pulldown assay. The binding of dityromycin to either Streptomyces or E. coli 30S ribosomal subunits was monitored by an ultracentrifugation-based assay. Increasing concentrations of 30S subunits were incubated for 10 min at room temperature with a fixed amount of antibiotic (1 M) in 90 l of a solution consisting of 10 mM Tris-HCl (pH 7.7), 10 mM Mg acetate, 60 mM NH 4 Cl, and 0.5 mM DTT. After ultracentrifugation in a S-100 AT-3 rotor at 100,000 rpm for 1 h at 4°C, aliquots of 20 l of the supernatant were removed and tested in poly(U)-dependent in vitro translation assays.
Determination of MIC. To verify the incorporation of S12 mutants in E. coli ribosomes, a MIC plate (5 rows ϫ 12 wells) was set up with increasing concentrations of streptomycin, ranging from 0.0 g/l to 512.0 g/l. Each well of the plate contained 190 l of E. coli BL21(DE3)/pLysS cell culture. The plate rows contained the following: (i) no-plasmid control, (ii) pET11a expression vector with rpsL150, (iii) mutant V32/T32, (iv) mutant R55/K55, and (v) mutant V78/I78.
A MIC plate (4 rows ϫ 12 wells) with the four tester strains and a gradient of dityromycin (from 0.0 g/l to 400 g/l) was used to estimate the ability of the mutant proteins to counteract the presence of the antibiotic. Further details are provided in the legend to Fig. 4.