16-membered ring macrolides and erythromycin induce ermB expression by different mechanisms

Ribosome stalling on ermBL at the tenth codon (Asp) and mRNA stabilization are believed to be mechanisms by which erythromycin (Ery) induces ermB expression. Expression of ermB is also induced by 16-membered ring macrolides (tylosin, josamycin and spiramycin), but the mechanism underlying this induction is unknown. We introduced premature termination codons, alanine-scanning mutagenesis and amino acid mutations in ermBL and ermBL2. In this paper, we demonstrated that 16-membered ring macrolides can induce ermB expression but not ermC expression. The truncated mutants of the ermB-coding sequence indicate that the regulatory regions of ermB whose expression is induced by Ery and 16-membered ring macrolides are different. We proved that translation of the N-terminal region of ermBL is key for the induction of ermB expression by Ery, spiramycin (Spi) and tylosin (Tyl). We also demonstrated that ermBL2 is critical for the induction of ermB expression by erythromycin but not by 16-membered ring macrolides. The translation of ermBL and the RNA sequence of the C-terminus of ermBL are critical for the induction of ermB expression by Spi and Tyl.


Background
Macrolides have been used clinically for over 70 years. These antibiotics inhibit Gram-positive and several Gram-negative bacteria [1]. These antibiotics exert their therapeutic effects by antagonizing the growth of bacteria via the inhibition of protein synthesis by narrowing the nascent protein exit tunnel (NPET) in ribosomes [2][3][4]. However, macrolide antibiotics can also induce the expression of several resistance genes [5][6][7][8]. For example, macrolide antibiotics promote ribosome stalling on the regulatory leader peptide ermCL or ermBL and then induce the expression of ermC or ermB [6,7,9]. The ermC and ermB genes encode a ribosomal methylase that dimethylates a single adenine in 23S rRNA, dramatically Open Access † Weizhi He, Kai Jiang and Hua Qiu have contributed equally to this work and share first authorship. *Correspondence: hew12@fudan.edu.cn; liao@pan-support.com; show. me1988@163.com reduces the affinity of macrolides for the ribosome, causing a high level of macrolide resistance and cell survival [10,11]. The rapid emergence of drug resistance in bacterial pathogens makes many antibiotics, including macrolide antibiotics, ineffective. In response to this threat, other nonclinical large-scale antibiotics may be substitutes for clinical use. Till date, 16-membered ring macrolides are rarely used in the clinic; therefore, there is no large-scale emergence of bacteria that are resistant to 16-membered ring macrolides in the clinic. Such antibiotics can be used as a potential alternative approaches for treatment in the clinic. According to traditional understanding, 16-membered ring macrolides can't induce the expression of resistance genes [12]. With increasing research, it was found that 16-membered ring macrolides could also induce the expression of resistance genes. Yakhin et al. showed that ribosome stalling at an RYR arrest motif in the C-terminus of a leader peptide found upstream of yxjB (encodes an enzyme that methylates 23S rRNA) is believed to be the mechanism by which tylsoin induces yxjB expression [13]. All macrolide-lincosamide-streptogramin B (MLS B ) antibiotics, including 16-membered ring macrolides, act as inducers of ermB expression to various degrees [10,14]. However, the mechanism by which 16-membered ring macrolides induce ermB expression is not well known. We wanted to study how 16-membered ring macrolides induce the expression of ermB. The results of this study may avoid the large-scale emergence of drug-resistant bacteria in the clinical use of 16-membered macrolides.
The expression of ermB can be either constitutive (M19270) [15] or inducible (M11180) [16] depending on the regulatory region located upstream of the ermB gene. The regulatory region includes a short leader peptide, which is also called ermBL, with its own ribosome binding site (RBS1), a nontranslational loop-stem structure, several ermB (ermB') coding sequences, and its own ribosome binding site (RBS2). In addition, in our previous study, we also found that another leader peptide named ermBL2 is present in the regulatory region and is critical for erythromycin (Ery)-mediated induction of gene expression (Fig. S1) [17].
When the expression of ermB is inducible, ribosome stalling and mRNA stability are believed to control its expression [6,18,19]. In the Enterococcus faecalis strain DS16 transposon Tn917 (M11180) [16], ermB is preceded by a 258 nucleotide leader region, which contains two regulatory open reading frames, ermBL, which encodes a 27 amino acid-long leader peptide, and ermBL2, which encodes a 16 amino acid-long leader peptide [6,17,19,20]. This regulatory region has been well studied in previous research [6,19,20]. Ribosome stalling induced by erythromycin takes place on the tenth codon (Asp) of ermBL and induces a conformational switch in the mRNA, which exposes the ribosome binding site (RBS2) of ermB to the ribosome; then, ermB is translated (Fig. S1).
In addition to erythromycin, the expression of ermB is also induced by 16-membered ring macrolides (tylosin, josamycin, and spiramycin) [10,21]. However, two groups have demonstrated that ribosome stalling on the tenth codon (Asp) of ermBL by 16-membered ring macrolides has not been observed [6,22,23]. Mutations in the ermBL region exerted differential effects on the induction of gene expression by 14-and 16-membered ring macrolides [14,21], suggesting that different mechanisms by which 16-membered ring macrolides induce ermB expression may exist.
Fourteen and fifteen-membered ring macrolides have been effective in the clinic for many years. Therefore, there are a large number of drug-resistant bacterial species that are resistant to these antibiotics in the clinic. To date, over 40 published erm genes, constitutive or inducible, have been identified in bacteria, and inducible erm expression has been shown to be induced by 14-and 15-membered ring macrolide antibiotics. 16-membered macrolide antibiotics include a disaccharide at position C-5. They are mainly used in veterinary medicine and are rarely used in the clinic, and there is no large-scale emergence of drug-resistant bacteria in the clinic. Notably, 16-membered macrolides are generally reported not to induce the expression of inducible erm genes. Studying the mechanism by which 16-membered macrolides induce the expression of ermB is conducive to formulating more reasonable strategies of drug use and significantly delaying the emergence of drug-resistant bacteria. Therefore, our results provide important insights into the clinical potential of these underexplored 16-membered ring macrolide antibiotics for use against drug-resistant human pathogens. The study of the mechanism underlying inducible drug resistance is helpful for treatment and for the prevention of the emergence of strains resistant to 16-membered ring macrolide antibiotics.

Construction of the pGEX reporter plasmid
The pGEX reporter plasmid was constructed as described in a previous study [17]. In short, the pGEX vector replaced some new multiple cloning site sequences (SmaI-KpnI-XbaI-AflII-XhoI-TthIIII) with sequences between BspMI and TthIIII. The ermCL-ermC' cassette from Staphylococcus aureus plasmid pE194 (X03097) [25] included the tac promoter, the leader ORF and part of the ermC coding sequence (ermC'). The ermC-ermC' cassette was cloned between the XbaI site and AflII site of the pGEX vector to produce the translational fusion plasmid pGEX-Ptac-ermCL-ermC'. ErmC' is the N-terminus of the ermC gene. ErmC' has no ribosomal methylase activity but is essential for conformational changes in the proposed model of the induction of ermC expression by erythromycin [7]. In this paper, we used ErmC' as ten amino acids of the N-terminus of ErmC. The reporter gene lacZα was cloned into a vector following ermC' with AflII and XhoI.

Disc diffusion assay of pGEX reporter activation
The disc diffusion assay protocol was carried out as previously described [24]. Briefly, JM109 cells transformed with the pGEX reporter plasmid were shaken overnight in LB broth supplemented with 100 μg/mL ampicillin at 37 °C. The culture was diluted 1:100 in fresh LB broth supplemented with ampicillin (100 μg/mL) and IPTG (0.5 mM) and shaken at 37 °C until the OD 600 approached 0.2-0.5. Then, we mixed the cells into 8 ml of 0.8% LB agar at 50 °C. After brief mixing, the cell suspension was poured on top of a 1.5% LB agar plate (10 cm dish) supplemented with 100 μg/mL ampicillin, 0.5 mM IPTG, and 160 μg/ml X-Gal. After the soft agar had solidified, discs were placed on agar, and the appropriate antibiotic was added. The plates were incubated for 18 to 24 h at 37 °C. The disc diffusion experiment used in our study aimed to qualitatively analyze gene expression induction by antibiotics. The intensity of the color of the blue ring relative to the blue background of plate, which reflects the extent of induction, was qualitatively assessed. The concentrations of antibiotics used were as follows: Ery (64 mg/ml), Spi (128 mg/ml), Tyl (128 mg/ml) and Chl (64 mg/ml).

β-galactosidase assay
The β-galactosidase assay method was used as described previously [17]. E. coli strains carrying the pGEX reporter plasmid were grown in LB until OD600 ≈ 0.2. Cultures were split and treated with a series of concentrations of antibiotics. The cultures were incubated at 37 °C for 1.5 h with shaking at 220 rpm. The OD600 of the resuspended cells was measured. The cells were centrifuged and then thoroughly suspended with 1 ml Z buffer (60 mM Na 2 HPO4, 40 mM NaH 2 PO4, 10 mM KCl, 1 mM MgSO 4 , pH 7.0) and shaken vigorously to lyse the cells by the addition of 100 μl chloroform and 50 μl 0.1% SDS. The assays were performed at 28 °C with 200 μl ONPG (O-nitrophenol-β-D-galactopyranoside, 4 mg/ml, Sigma). The reaction was stopped after a sufficient yellow color developed by adding 0.5 mL of 1 M Na 2 CO 3 . Then, the samples were centrifuged and monitored at 420 nm. β-Galactosidase activities were calculated in Miller Units using the following formula: β-Galactosidase activity = A420 × 1000 × min -1 × ml -1 × A600 -1 [26]. At least three independent biological replicates were performed. Unless otherwise specified, the concentrations of antibiotics used were as follows: Ery (256 μg/ml), Spi (1024 μg/ ml) and Tyl (1024 μg/ml).

16-membered macrolides could specifically induce the expression of ermB
In a previous study, we constructed a pGEX-ermBL-ermB'-lacZα reporter plasmid that has a tac promoter (Ptac) following a regulatory region ermBL-ermB' operon and translational fusion with lacZα ( Fig. 1A, S2) [17]. We also constructed a new pGEX-ermCL-ermC'-lacZα reporter plasmid in this study. This reporter system allows for the easy monitoring of the induction of gene expression by antibiotics, either by measuring β-galactosidase enzyme activity (quantitative analysis) or by observing X-Gal hydrolysis on plates using a disc diffusion assay (qualitative analysis). Erythromycin is a 14-membered ring macrolide, while spiramycin and tylosin are 16-membered ring macrolides (Fig. 1B). Tylosin and spiramycin are 16-membered macrolide antibiotics that include a disaccharide at position C-5, with tylosin also bearing a mycinose substituent at position C-23.
In a previous study, ermB expression was found to be strongly induced by subinhibitory concentrations (approximately 25% MIC) of Ery, Spi and Tyl [10,18]. We first determined the minimal inhibitory concentration (MIC) of antibiotics for E. coli carrying the pGEX-ermBL-ermB'-reporter plasmid (Table S1). We next used our plasmid to qualitatively and quantitatively investigate the difference in reporter activation by a number of antibiotics at subinhibitory concentrations. Ery, Spi and Tyl activated the ermBL-ermB' reporter at a broad range of concentrations (Fig. 1C, D). As expected, chloramphenicol (Chl) and the antibiotic solvent DMSO could not induce ermB expression (Fig. 1C). These results showed that Ery, Spi and Tyl induced the expression of ermB. In a previous study, a well-studied model of ermCL expression from the Staphylococcus aureus plasmid pE194 revealed that ermCL expression was also strongly induced by erythromycin but not spiramycin or tylosin [24,[27][28][29]. As expected, in our reporter system, Ery but not Spi and Tyl could induce the expression of ermC (Fig. 1C, E). ErmC expression could not be induced by 16-membered ring macrolide antibiotics or chloramphenicol (Fig. 1C, E). In summary, Ery can induce the expression of ermB and ermC, while 16-membered ring macrolides can specifically induce the expression of ermB.

ErmBL translation is key for the induction of ermB expression by 16-membered ring macrolides
Ribosome stalling on the tenth codon of ermBL is believed to be the major mechanism by which erythromycin induces ermB expression [6,20]. One hypothesis suggested that small ORFs located in the 5' untranslated region may act as cis mRNA stabilizers, increasing the half-life of the downstream transcripts [30,31]. To evaluate whether ermBL translation is critical for the induction of ermB expression by Spi and Tyl, we used various constructs in the regulatory region of ermBL described in a previous study ( Fig. 2A) [17]. We found that Ery, Spi and Tyl could no longer induce the expression of downstream genes with RBS1 (GGA GGG ) deletion (Fig. 2B). We also found that the induction of expression by Ery, Spi and Tyl was severely impaired when the start codon of ermBL was mutated to a stop codon (ATG to TAA) (Fig. 2C), which means that ermBL expression is critical for the induction of ermB expression by Ery, Spi and Tyl. When we changed the 19th codon of ermBL to a stop codon (ACT19:TGA) and used the previously constructed plasmid with the 20th codon of ermBL mutated to a stop codon (AAA20:TGA) [17], the C-terminus of ermBL contained premature termination. ermB expression could also be induced by Ery, Spi and Tyl, which means that the translation of the N-terminus of ermBL is necessary for the induction of ermB expression by Ery, Spi and Tyl, while translation of the C-terminus (K20-K27) of ermBL is not important (Fig. 2D, E). The leader ORF ermBL encodes 27 long amino acids, and we changed the stop codon TAA to AAA, so ermBL will encode a 30-amino acid long peptide. Even though the length of ermBL was changed, induction of its expression still occurred, which shows that the length of the regulation leader peptide does not affect the induction of ermB expression by Ery, Spi and Tyl (Fig. 2F). Taken together, these results verify that ermBL translation is key for the induction of ermB expression by 16-membered ring macrolides.

Induction of ermB expression by Ery and 16-membered ring macrolides occurs via different mechanisms
The tenth codon Asp of ermBL is a key amino acid for ribosome stalling, and ribosome stalling in the tenth codon is critical for the induction of ermB expression by Ery [6,18]. In previous study, several groups showed that ribosome stalling in the tenth codon of ermBL could not be observed by in vitro toe-printing [6,22,23]. To investigate whether the tenth codon Asp of ermBL is a key amino acid for the induction of ermB expression by 16-membered ring macrolides, we used reporter constructs that had been constructed in our previous study in which the tenth codon was mutated to other amino acids [17]. Gupta et al. showed that replacement of Asp10 with tyrosine, cysteine, glutamine or valine abolished Ery-mediated translation arrest at codon 10 [19]. When the tenth codon Asp of ermBL was mutated to these amino acids that affect Ery-specific ribosome stalling in vitro, we wondered whether the induction of ermB expression by Ery, Spi and Tyl was affected in vivo. When the Asp10 codon of ermBL was replaced with tyrosine, cysteine, glutamine or valine, the induction became Spi-and Tyl-dependent, while the induction of reporter gene expression by Ery was abolished (Fig. 3A, B, C, D), indicating that the mechanisms underlying the induction of gene expression by Ery and 16-membered ring macrolides are different. To evaluate whether the RNA sequence of ermBL is important as an amino acid, we changed the Asp10 codon to valine encoded by different codons (GTT, GTC, GTA, GTG). The β-galactosidase assay and disc diffusion assay showed that induction of reporter gene expression by Ery was severely impaired by all the D10V mutations, while the reporter gene expression was also induced by Spi and Tyl (Fig. 3D), indicating that the difference between Ery and 16-membered ring macrolides was not RNA sequence dependent. Gupta1 et al. showed that the induction of ermB expression by Ery and telithromhycin could be affected by single amino acid changes in the ermBL sequence that switch the specificity of recognition of distinct antibiotics [19]. To verify whether the different mechanisms by which Ery and 16-membered ring macrolides induce ermB expression are also caused by the different antibiotics recognizing different amino acids in ermBL, we made mutations in the ermB regulatory region but not in ermBL. In a proposed classic model of mechanisms underlying induction of gene expression, the N-terminus of ermB (ermB') also contributed to the change in secondary structure (Fig. S1). We generated truncated mutations of ermB' (Fig. S2) and surprisingly found that the regions required for the induction of expression by these two types of antibiotics were different (Fig. 3E). In this paper, when ermB' equals the first ten amino acids of the N-terminus of ermB, we found that reporter gene expression could be well induced by Ery, Spi and Tyl (Fig. 1C, B, D, E). When ermB' contains only one amino acid (methionine), ermB expression could not be induced by these two kinds of antibiotics. When ermB' has two amino acids (MN), ermB expression could be moderately induced by 16-membered ring macrolides but not by Ery. When ermB' includes three amino acids (MNK), ermB expression could be completely induced by 16-membered ring macrolides but not Ery. When ermB' has four amino acids (MNKN), ermB expression could be completely induced by 16-membered ring macrolides and moderately induced by Ery. The ermB regulatory region that is necessary for the induction of ermB expression by Ery and 16-membered ring macrolides is different. In summary, we showed that 16-membered ring macrolides and Ery induce the expression of ermB via different mechanisms, which is not because of different stalling efficiencies due to one amino acid change in the leader peptide sequence. Fig. 2 Translation of the N-terminus of ermBL is critical for the induction of ermB expression by 16-membered ring macrolides. A The detailed RNA and amino acid sequence of ermBL and its own ribosome binding site (RBS1). B-E β-Galactosidase activity assay and disc diffusion assay of various ermBL mutants exposed to erythromycin, spiramycin and tylosin. Miller units of β-galactosidase activity are shown on the Y-axis. The number on the top of each bar represents the fold change in beta-gal activity between the antibiotic and DMSO group. The error bars correspond to the SEM of three independent experiments. All the β-Galactosidase activity of antibiotics group compared to DMSO group is significantly. P < 0.05; (unpaired two-tailed Student's t test)

Alanine-scanning mutagenesis of ermBL confirms that 16-membered ring macrolides and Ery induce ermB expression via different mechanisms
In a previous study, the seventh codon to eleventh codon of ermBL (R7-K11) were found to be key amino acids for the translation arrest induced by erythromycin [6,22]. Our data showed that ermBL translation is critical for the induction of ermB expression, while Asp10 is not important for the induction of ermB expression by 16-membered ring macrolides. To distinguish which amino acids of ermBL are necessary for the induction of ermB expression by 16-membered ring macrolides, we used an alanine-scanning mutagenesis assay (Fig. 4A). Similar to the previous conclusion, the M1A mutation disrupts the translation of ermBL, so the induction of ermB expression by Ery, Spi and Tyl is impaired (Fig. 4B). L2-M6 (the second codon to the sixth codon of ermBL) is not important for the induction of ermB expression by Ery, and it is not important for the induction of ermB expression by 16-membered ring macrolides. In contrast to erythromycin, alanine-scanning mutagenesis of ermBL demonstrated that R7-K11 (the seventh codon to the eleventh codon of ermBL) is not critical for the induction of ermB expression by spiramycin and tylosin because the induction of ermB expression by spiramycin and tylosin is less impaired than that by erythromycin once these key amino acids are mutated to alanine (Fig. 4B). Unexpectedly, we found that the C-terminus of ermBL (N21, Y24, K27) is critical for the induction of ermB expression by 16-membered ring macrolides (Fig. 4B) (Fig. S4). However, Fig. 2 shows that the translation of the last codons of ermBL is not critical for the induction of ermB expression by 16-membered ring macrolides. These contradictory results indicate that the C-terminus of ermBL is crucial for antibiotic-induced ermB expression, not because it is the amino acid of ermBL, but for other reasons. When we changed an amino acid in the the C-terminus of ermBL to alanine (Fig. 4B), we changed not only the amino acid sequence but also the RNA sequence of the C-terminus of ermBL. Figure 2 shows that the amino acids of the last codons of ermBL are not critical for represents the fold change in beta-gal activity between the antibiotic and DMSO group. The error bars correspond to the SEM of three independent experiments. All the β-Galactosidase activity of antibiotics group compared to DMSO group is significantly. P < 0.05; (unpaired two-tailed Student's t test). B β-Galactosidase activity assay and disc diffusion assay of ermB' truncated mutants exposed to erythromycin, spiramycin and tylosin. Miller units of β-galactosidase activity are shown on the Y-axis. The number on the top of each bar represents the fold change in beta-gal activity between the antibiotic and DMSO group. The error bars correspond to the SEM of three independent experiments. All the β-Galactosidase activity of antibiotics group compared to DMSO group is significantly. P < 0.05; (unpaired two-tailed Student's t test) the induction of ermB expression by 16-membered ring macrolides. Therefore, we have reason to think that the RNA sequence, not the amino acid sequence, of the C-terminus of ermBL, is critical for the induction of ermB expression by 16-membered ring macrolides.
In our previous work, we showed that the last sequence of ermBL (N21-K27) function as N-terminus of ermBL2 are very important for Ery-mediated induction of ermB expression [17]. However, S22A and D23A mutations maintained the induction of ermB expression by 16-membered ring macrolides, but these mutations reduced the effect of Ery, which further shows that the two kinds of antibiotics have different mechanisms by which they induce ermB expression.

The latter part of the ermB regulatory region is a key component for the induction of ermB expression by spiramycin and tylosin
To determine which region of ermBL is key for the induction of ermB expression by spiramycin and tylosin, we engineered several hybrid leaders using the feature that 16-membered ring macrolides induce ermB expression rather than ermC expression. This difference provides a good model for investigating the mechanism by which 16-membered ring macrolides induce ermB expression.
Codons 9 (isoleucine, I) and 10 (serine, S) of WT ermCL are P-and A-site codons in ribosomes that are stalled at ermCL by erythromycin, while Codons 10 (asparagine, D) and 11 (lysine, K) of WT ermBL are P-and A-site codons in ribosomes that are stalled at ermBL by erythromycin [6,7]. We mutated the Asp10 and Lys11 codons of ermBL to Ile and Ser, respectively (ermBL DK [10,11] IS). We found that it had no effect on the induction of ermB expression by sipramycin and tylosin (Fig. 5A), but the induction by erythromycin was impaired (Fig. 5A). ErmBL DK [10,11] IS changed IS codons from the ninth and tenth codons of ermCL to the tenth and eleventh codons of ermBL, and changing the length of the nascent peptide dramatically reduced the efficiency of stalling in IS, as shown in a previous study [7]. We also mutated the ninth (I) and tenth (S) codons of ermCL to Asp and Lys, respectively (ermCL IS [9,10] DK), and we found that the expression of ermC was not induced by spiramycin and tylosin as ermCL WT (Fig. 5B, 1C, and E). Taken together, the results suggest that D10-K11, which is key for the induction of ermB expression by erythromycin, is not a key amino acid for the induction of ermB expression by spiramycin and tylosin.
To distinguish which part is key for the induction of ermB expression by spiramycin and tylosin, we tailored  (Fig. 5C) (Fig. S3). We found that spiramycin and tylosin did not induce the expression of the gene in this fusion plasmid (Fig. 5C). This result illustrated that the first part of ermBL does not determine the induction of expression by 16-membered ring macrolides. We also tailored the ERY-inducible ermBL-lacZa reporter fusion to the ermCL-controlled ribosome stalling part (M1-S10) named CL-BL (Fig. 5D) (Fig. S3) and found that spiramycin and tylosin induced gene expression similar to erythromycin (Fig. 5D). In summary, these data showed that the latter part of the ermB regulatory region (T12 of ermBL to ermB') is key for the induction of expression by spiramycin and tylosin.

Fig. 5
The latter part of the ermB regulatory region is key for the induction of ermB expression by spiramycin and tylosin. A Codons 10(Asp) and 11(Lys) of WT ermBL were mutated to isoleucine and serine, respectively, to mimic the identified P-and A-site codons in ribosomes stalled at ermCL. B Codons 9 (isoleucine) and 10 (serine) of WT ermCL were mutated to aspartic acid and lysine, respectively, to mimic the identified P-and A-site codons in ribosomes stalled at ermBL. C The first part of the ERY-inducible ermCL-lacZα reporter plasmid is a substitute for ErmBL-controlled ribosome stalling (M1-K11), named BL-CL. D The first part of the ERY-inducible ermBL-lacZα reporter plasmid is a substitute for ErmCL-controlled ribosome stalling (M1-S10), named CL-BL. β-Galactosidase activity assay and disc diffusion assay of these mutants exposed to erythromycin, spiramycin and tylosin. Miller units of β-galactosidase activity are shown on the Y-axis. The number on the top of each bar represents the fold change in beta-gal activity between the antibiotic and DMSO group. The error bars correspond to the SEM of three independent experiments. All the β-Galactosidase activity of antibiotics group compared to DMSO group is significantly. P < 0.05; (unpaired two-tailed Student's t test)

Translation of ermBL2 is not critical for the induction of ermB expression by spiramycin and tylosin
The above results show that the translation of ermBL and the C-terminal region of ermBL are key for the induction of ermB expression by erythromycin, spiramycin and tylosin. In addition, the mechanism underlying the induction of expression by Ery and 16-membered ring macrolides is different. Additionally, the amino acid sequence of the C-terminal region of ermBL is not critical for ermB induced by spiramycin and tylosin. In our previous work, we showed that a new leader peptide, ermBL2, is present in the ermB regulatory region and is critical for the induction of ermB expression by Ery (Fig. 6A) [17]. Therefore, we hypothesized that the importance of the C-terminus of ermBL is either because the amino acid sequence of ermBL2 or the RNA sequence itself affects the induction of ermB expression by 16-membered ring macrolides.
We wanted to determine whether the expression of ermBL2 was critical for the induction of ermB expression by Spi and Tyl. We used previous constructs in which the start codon (GTG) of ermBL2 was mutated to alanine (GCG, GCT, GCC, GCA) [17]. In vivo β-galactosidase assays and disc diffusion assays showed that erythromycin did not induce ermB expression with all alanine mutations, while spiramycin and tylosin lost the ability to induce the expression of ermB with most alanine mutations except the GTG1:GCG mutation (Fig. 6B, C, D, E). We also used previous constructs in which the start codon (GTG) of ermBL2 was mutated to other stronger or weaker start codons (ATG, TTG, CTG) [17]. These single nucleotide mutations eliminated the effect of Ery on the induction of ermB expression while maintaining the effect of Spi and Tyl on the induction of ermB expression (Fig. 6F, G, H). We then used previous constructs in which the start codon of ermBL2 (GTG) was mutated to a stop codon (TAA, TAG, TGA) [17] and found that the induction of ermB expression by Ery was impaired by all stop mutations, while spiramycin and tylosin maintained the ability to induce the expression of ermB with GTG1:TAG mutations (Fig. 6I, J, K). In summary, we concluded that the translation of ermBL2 is necessary for the induction of ermB expression by ery rather than Spi and Tyl. Additionally, GTG1: (TAA, TAG, TGA) mutations had different effects on the induction of ermB expression by Spi and Tyl, indicating that the RNA sequence itself, rather than the amino acid sequence of ermBL2, is critical for the induction of ermB expression by Spi and Tyl.

Discussion
The slow discovery of new antibiotics and the emergence of a large number of antibiotic-resistant bacterial species have led to the possible use of some unpopular antibiotics. Macrolide antibiotics are used to treat infections caused by gram-positive and gram-negative bacteria [1]. These antibiotics have been effective in clinical use for over 70 years. Most of the macrolides currently used in the clinic are semisynthetic erythromycin derivatives composed of a 14-or 15-membered macrolactone ring. Therefore, there are a large number of drug-resistant bacterial species that are resistant to these antibiotics in the clinic. Because of the substantial clinical application of macrolides, they are not easily discarded. A simple strategy is to use macrolides that are not commonly used or not used in the clinic, such as 16-membered macrolides. We wanted to know whether 16-membered ring macrolides induce the expression of resistance genes similar to 14-or 15-membered ring macrolides and whether the mechanism of inducing expression is the same as that of 14-or 15-membered ring macrolides.
Four erm genes (ermA, ermB, ermC, ermD) whose expression is induced by erythromycin have been well studied [5,7,32]. Only the expression of ermB could be induced by 16-membered-ring macrolides in previous study [10,21]. The ermB gene encodes the ribosomal methylase that dimethylates a single adenine in 23S rRNA, which leads to high macrolide resistance and bacterial survival [10,11]. All the MLS B antibiotics induced the expression of ermB [10]. This feature of ermB is different from other classes of erm (ermA, ermC and ermD), and the expression of ermA, ermC or ermD is induced by certain specific MLS B antibiotics [7,32]. In our reporter system, we found that 16-membered ring macrolides specifically induced the expression of ermB rather than ermC. Till date, there are three major mechanisms to control the expression of inducible macrolide resistance genes by different kinds of macrolide antibiotics. (1) Ribosome stalling on the leader peptide is the main mechanism to control the expression of inducible macrolide resistance genes, such as ermAL [32], ermBL [6], ermCL [7] and ermDL [5] induced by erythromycin. (2) Macrolide antibiotic-induced stabilization of resistance gene mRNA is another mechanism to control the expression of several inducible resistance genes [18,22,33]. (3) Regulation of ermC gene expression by ketolides is controlled by ribosomal frameshifting [34]. Translational attenuation and mRNA stabilization are mechanisms by which ermB expression is induced by erythromycin. The proposed translational attenuation model of the induction of ermB expression by erythromycin has been studied (Fig. S1) [17]. In the absence of erythromycin, ermB expression is repressed because the ribosome binding site 2 (GGAG) and AUG start codon of the ermB mRNA are sequestered in a stem-loop structure. An alternative stem-loop structure is changed in the presence of erythromycin, exposing the RBS2 and start codon of the ermB gene and causing the induction of ermB expression (Fig. S1).
The induction of ermB expression by 16-membered ring macrolides is not well documented. The purpose of this work was to investigate the mechanism by which 16-membered ring macrolides induce the expression of ermB. We used spiramycin and tylosin as standard 16-membered ring macrolides. We first constructed a reporter plasmid to sense antibiotics. Spiramycin and tylosin specifically induced the expression of ermB rather than ermC. In this study, introduction of a premature termination codon showed that the translation of the N-terminus of ermBL is necessary for the induction of ermB expression by Spi and Tyl, while the C-terminus of ermBL is not important for the induction of ermB expression by Spi and Tyl.
We found that Ery and 16-membered ring macrolides induced the expression of ermB via different mechanisms for the following reasons: (1) Ribosome stalling on ermBL at the tenth codon (Asp) is believed to be the major mechanism by which Ery induces ermB expression. We changed the tenth codon (Asp) to other amino acids and found that the majority of the mutations rendered the capacity of induction by 16-membered ring macrolides rather than erythromycin. (2) Gupta et al. revealed the ribosome as a highly selective sensor of Ery and telithromycin. Its ability to recognize and discriminate between Ery and telithromycin could be directly modulated by minor variations in the sequence of the nascent peptide [19]. If the mechanism by which 16-membered ring macrolides induce ermB expression are still relevant to this situation, then when we changed the other parts of the ermB regulatory region, Ery and 16-membered ring macrolides should show similar inducing effects. However, when we changed the length of ermB' (N-terminal region of ermB), the induction effect was different between these two kinds of antibiotics (Fig. 3E), which means that the different mechanisms by which Ery and 16-membered ring macrolides induced ermB expression are not the same as the different mechanisms by which Ery and telithromycin induce ermB expression. (3) Alanine-mutational analyses have also shown that changes in the R7-D10 amino acids of ermBL had little influence on induction effects of Spi and Tyl, while gene expression induction by Ery was severely impaired. The latter part of ermBL is important for the induction of ermB expression by Ery and 16-membered ring macrolides. However, S22A and D23A had little influence on the induction effect of Spi and Tyl but severely impaired the induction effect of Ery (Fig. 4B). This further shows that the mechanisms by which these two kinds of antibiotics induce expression are different. (4) Furthermore, hybrid CL-BL or BL-CL constructions showed that the latter part of the ermB regulatory region is critical for the induction of ermB expression by Spi and Tyl.
ErmBL2, which exists in the ermB regulatory region, is critical for the induction of ermB expression by Ery [17]. The C-terminus of ermBL and the N-terminus of ermBL2 share base sequence but not amino acid sequence. ermBL2 is a (+ 1) frameshift compared with ermBL in the common area. We mutated the start codon (GTG) of ermBL2 to alanine, start codon or stop codon and found that the translation of ermBL2 is not important for the induction of ermBL expression by Spi and Tyl. GTG1:TAG mutation maintained the induction effect of Spi and Tyl, while the GTG1:TAA and GTG1:TGA mutations decreased the induction effect of Spi and Tyl, indicating that the RNA sequence itself, rather than the amino acid sequence, of ermBL2 is critical for the induction of ermB expression by Spi and Tyl. GTG1: alanine (GCG, GCT, GCC, GCA) mutations also confirmed this conclusion.
Here, we used a well-studied ermBL-ermB operon (M11180) [16] LacZa fusion reporter plasmid as a model to investigate the detailed mechanism underlying expression induction by 16-membered ring macrolides [17,18]. In summary, the translation of ermBL and the RNA sequence of the C-terminus of ermBL are critical for the induction of ermB expression by Spi and Tyl. The detailed mechanism needs further study, and the study of the mechanism underlying the drug resistance induced by 16-membered ring macrolides will be helpful for the treatment and prevention of the emergence of 16-membered ring macrolide-resistant strains.

Conclusion
The translation of ermBL and the RNA sequence of the C-terminus of ermBL are critical for the induction of ermB expression by Spi and Tyl. The study of the mechanism underlying the drug resistance induced by 16-membered ring macrolides will be helpful for the treatment and prevention of the emergence of drug-resistant strains.