Correspondence: Terence I. Moy, Cubist Pharmaceuticals Inc., 65 Hayden Avenue, Lexington, MA 02421, USA. Tel.: +1 781 860 8799; fax: +1 781 861 1164; e-mail: firstname.lastname@example.org
Myxopyronin B (MyxB) binds to the switch region of RNA polymerase (RNAP) and inhibits transcriptional initiation. To evaluate the potential development of MyxB as a novel class of antibiotic, we characterized the antimicrobial activity of MyxB against Staphylococcus aureus. Spontaneous MyxB resistance in S. aureus occurred at a frequency of 8 × 10−8, similar to that of rifampin. The MyxB-resistant mutants were found to be altered in single amino acid residues in the RNAP subunits that form the MyxB-binding site. In the presence of human serum albumin, the MyxB minimum inhibitory concentration against S. aureus increased drastically (≥128-fold) and 99.5% of MyxB was protein bound. Because of the high serum protein binding and resistance rate, we conclude that MyxB is not a viable starting point for antibiotic development.
The RNA polymerase (RNAP) core enzyme of Staphylococcus aureus consists of two α-subunits, one β-subunit, and one β′-subunit, which are encoded by the rpoA, rpoB, and rpoC genes, respectively. The primary σ-factor in S. aureus is encoded by the sigA gene (Deora & Misra, 1996). Rifampin, an RNAP inhibitor in clinical use, binds to the β-subunit of RNAP within the DNA/RNA channel and blocks the elongation of RNA when the transcript becomes two to three nucleotides in length (Campbell et al., 2001). Rifampin is commonly used in combination with other antibiotics due to rapid resistance development from single amino acid mutations in the rifampin-binding site (Campbell et al., 2001).
Myxopyronin B (MyxB) is a natural product isolated from Myxococcus fulvus strain Mxf50 that has activity against Gram-positive bacteria including rifampin-resistant S. aureus. MyxB binds to a pocket deep inside the switch region of RNAP that is distinct from the binding site of rifampin and inhibits transcriptional initiation (Mukhopadhyay et al., 2008; Belogurov et al., 2009). One proposed mechanism of action of MxyB is that it locks the RNAP clamp in a closed conformation, thereby preventing the interaction of RNAP with promoter DNA (Mukhopadhyay et al., 2008). Another proposed mechanism of action is its inhibition of the propagation of promoter DNA melting (Belogurov et al., 2009). In this study, we characterized the effect of MyxB in an in vitro transcription assay, the antimicrobial properties of MyxB, and the development of single-step resistance to MyxB.
Materials and methods
Rifampin was purchased from USP Pharmacopeia (Rockville, MD). MyxB and the desmethyl derivative of MyxB (dMyxB) were synthesized as described previously (Hu et al., 1998; Doundoulakis et al., 2004; Lira et al., 2006). An in vitro transcription assay using Escherichia coli RNAP holoenzyme (Epicentre Biotechnologies, Madison, WI) was used to determine IC50 values as described previously (Marras et al., 2004). Binding to human serum proteins was measured using an ultracentrifugation-based method: 2 μM of compound was incubated with human serum, centrifuged at 100 000 g for 4 h at 37 °C, and the free fraction of the compound in the supernatant was quantified by LC–MS/MS.
Antibacterial activity of the compounds was tested against three strains of S. aureus (ATCC 29213, ATCC 12600, and MW2) as described previously (Friedman et al., 2006). To determine the frequency of spontaneous resistance, cultures were grown in Mueller–Hinton broth, plated onto Mueller–Hinton agar containing the appropriate compound, and incubated at 37 °C for 48 h. Resistant colonies were passaged three times on drug-free plates and tested for minimum inhibitory concentrations (MICs) in broth or on agar, the latter being prepared using a spiral plater according to the manufacturer's protocol (Spiral Biotech Inc., Norwood, MA). The rpoA, rpoB, rpoC, and sigA genes were sequenced from the strains by SeqWright (Houston, TX) as described (Friedman et al., 2006).
Results and discussion
In confirmation of previous reports, MyxB and dMyxB inhibited the transcription and growth of S. aureus. MyxB and dMyxB inhibited transcription, with IC50 values of 24 and 14 μM, respectively. This 1.7-fold increase in the potency of dMyxB compared with MyxB is comparable to the 2.7-fold increase in potency as reported by Lira et al. (2007). For comparison, rifampin has an IC50 value of 0.1 μM in this assay. MyxB and dMyxB MICs against S. aureus ranged from 0.5 to 1.0 μg mL−1, in agreement with previously published values (Irschik et al., 1983; Doundoulakis et al., 2004).
Previous studies have shown that MyxB lacks in vivo efficacy in a mouse infection model (Irschik et al., 1983). To investigate this lack of efficacy, we determined the effect of human serum albumin (HSA) on the antibacterial potency of MyxB and dMyxB. MyxB MIC values were 1, 16, 32, 64, and 128 μg mL−1 in the presence of 0%, 0.5%, 1%, 2%, and 5% HSA, respectively. dMyxB MICs followed a similar trend and at the physiologically relevant concentration of HSA of 5%, the dMyxB MICs increased by ≥128-fold. Using an ultracentrifugation-based method of measuring human serum protein binding, we determined that 99.5% of MyxB and 99.6% of dMyxB were protein bound. Taken together, these data indicate that binding to serum proteins reduces the antibacterial activity of these compounds in vivo.
When the resistant mutants were selected at 4 × MIC of MyxB, the average frequency of resistance was similar to rifampin for three strains of S. aureus. Similar frequencies of resistance were measured when the selection was performed at 8 × MIC (Table 1).
Table 1. Frequency of spontaneous resistance of RNAP inhibitors
Frequency of resistance
4 × MIC
4.2 × 10−8
6.0 × 10−8
4.1 × 10−8
8 × MIC
3.3 × 10−8
6.7 × 10−8
3.7 × 10−8
Myxopyronin B (MyxB)
4 × MIC
4.8 × 10−8
8.0 × 10−8
1.3 × 10−7
8 × MIC
2.3 × 10−8
6.4 × 10−8
1.1 × 10−7
Several of the single-step resistant mutants gained a high degree of resistance. For rifampin, MyxB, and dMyxB, the majority of the resistant isolates tested had an increase in MIC≥16-fold. Some of the resistant mutants were ≥12 800-fold more resistant to rifampin or ≥128-fold more resistant to dMyxB. Cross-resistance to dMyxB was detected for the MyxB-resistant isolates, but no cross-resistance was detected between rifampin- and MyxB-resistant isolates (data not shown).
The rpoB and rpoC genes were sequenced from 12 MyxB-resistant mutants. Additionally, the rpoA and sigA genes were sequenced from six of these mutants. While no mutations were found in rpoA or sigA, single nucleotide changes were found in either rpoB or rpoC for each of the 12 mutants (Table 2). A total of nine different amino acid changes were identified affecting seven residues. For the RpoB protein, E1079D, P1125L, S1127L, and S1127R mutations were identified. For the RpoC protein, K334N, T925R, A1141T, A1141V, and L1165R mutations were identified. Based on analysis of the crystal structure of the Thermus thermophilus RNAP holoenzyme bound to MyxB or dMyxB (Mukhopadhyay et al., 2008; Belogurov et al., 2009), all of the mutated residues are predicted to be located near the MyxB-binding site formed by the RpoB and RpoC subunits (Fig. 1). RpoB residue S1127 and RpoC residues K334 and A1141 are predicted to interact directly with MyxB. RpoB residues E1079 and P1125 and RpoC residue L1165 are one to two amino acids away from residues that are predicted to interact with MyxB. RpoC residue T925 is not present in the T. thermophilus RpoC protein, but the T. thermophilus residue in the corresponding position (I1223) is oriented towards the MyxB-binding site and is within 5 Å of MyxB (schematic in Fig. 1).
Table 2. dMyxB- and MyxB-resistant mutants isolated in this study
Fold increase in MIC
β (RpoB) mutation
β′ (RpoC) mutation
Broth microdilution MIC assay was used to measure fold increase in dMyxB MIC.
† Gradient MIC assay on agar media was used to measure fold increase in MyxB MIC.
Concurrent with our studies, Mariner et al. (2011) characterized corallopyronin A (CorA)-resistant mutants. CorA is a RNAP inhibitor that is structurally related to MyxB and has been reported to share the same binding site on RNAP as MyxB (Mukhopadhyay et al., 2008). The CorA-resistant mutants were found to be cross-resistant to MyxB and have single amino acid substitutions in residues located within the MyxB-binding site. The CorA- and MyxB-resistant mutants had slight to minimal changes in the generation time, indicating that the RNAP mutations cause a slight to minimal loss of fitness (Mariner et al., 2011).
Based on the structural and binding site differences between MyxB and rifampin, we and others (Mukhopadhyay et al., 2008) have speculated that myxopyronins could be developed as a new class of clinically relevant RNAP inhibitors that would overcome rifampin's deficiency of high resistance incidence. However, we found several fundamental challenges for the clinical development of the myxopyronins. First, the antibacterial activity of MyxB and dMyxB is drastically decreased in the presence of serum albumin. Binding to serum albumin is typically driven by hydrophobic interactions (Curry, 2009). Because the binding of dMyxB to RNAP is principally driven by hydrophobic interactions (Mukhopadhyay et al., 2008; Belogurov et al., 2009), it may be difficult to produce less hydrophobic MyxB analogs that retain RNAP inhibitory activity. The second issue is compound stability; the central core of the myxopyronins contains a Michael acceptor, which is generally regarded as undesirable due to its reactivity. We found that MyxB was unstable at pH 3 or after exposure to UV light (data not shown). Finally, similar to rifampin, resistance to MyxB occurs at a high frequency. We isolated MyxB-resistant mutants with single amino acid changes in seven different residues in the MyxB-binding site within RNAP, but we did not observe growth defects for these mutants, suggesting that the MyxB-binding site can be mutated in a way that does not significantly affect RNAP activity. While myxopyronins and rifampin have differences in the mechanism of action and binding sites (Campbell et al., 2001; Mukhopadhyay et al., 2008; Belogurov et al., 2009), the shared problem of resistance may represent an inherent limitation for practical uses of these RNAP inhibitors as monotherapies.
We gratefully acknowledge the assistance of Lihong Gao and Azard Mahamoon. We thank Katherine Mariner, Alex O'Neill, and Ian Chopra for communication of their work before publication.