Introduction

Global spread of multidrug-resistant M. tuberculosis (Mtb) strains and emergence of extensively drug-resistant tuberculosis (TB) call for innovative approaches for treating TB. As conventional first- and second-line anti-TB drugs are gradually rendered less useful due to drug resistance,1 and the number of drug candidates in the anti-TB pipeline is low,2 novel targets and agents inhibiting them need to be explored. Bacterial primase DnaG, a DNA-dependent RNA polymerase that synthesizes oligonucleotide primers during DNA replication, has been known for several decades to be essential for both chromosomal DNA replication and cell division.3, 4, 5, 6 DnaG is conserved in bacteria and is distinct from eukaryotic and archaeal primases.7 In addition, the catalytic TOPRIM (topoisomerase-primase) domain of DnaG shares a structural fold with DNA gyrase, a target of the clinically useful fluoroquinolone antibiotics.8 Therefore, DnaG is an attractive potential target for discovery and development of novel antibacterial agents. The requirements of radioactivity as a detection method and accessory DNA replication factors to boost primase activity in traditional in vitro primase activity assays have impeded facile identification and characterization of DnaG inhibitors, and, until recently, no potent (low-μM or better) inhibitors of DnaG were reported.

We recently developed a coupled colorimetric primase–pyrophosphatase assay for measurement of DnaG activity and applied this assay in a high-throughput screening (HTS) of small-molecule libraries to identify inhibitors of DnaG and another essential enzyme, inorganic pyrophosphatase (PPiase).9 HTS applications of this assay to Mtb DnaG9 and Bacillus anthracis DnaG10 yielded low-μM inhibitors of these enzymes. Doxorubicin, an anthracycline antibiotic of bacterial origin used in clinics as an anticancer drug, was identified in these studies as a potent inhibitor of both DnaG enzymes. The cytotoxic activity of doxorubicin against cancer cells originates from its inhibition of topoisomerase II,11 a eukaryotic homolog of gyrase, by creating a ternary complex with topoisomerase II and dsDNA. Even though the antibacterial mechanism of action of doxorubicin has not been extensively investigated, early studies with this compound reported it as a DNA replication inhibitor,12 whereas its inhibition of gyrase was shown to be too weak to explain its antibacterial potency.13 Our recent findings, taken together with these earlier observations, suggest that doxorubicin inhibits bacterial cell growth by inhibiting DnaG.

In our search for other potentially therapeutically useful inhibitors of DnaG, we explored several anthracycline-based DNA intercalators as well as less toxic natural anthranoids. We investigated the inhibitory potency of these agents against the activity of purified Mtb DnaG, as well as measured their minimum inhibitory concentrations (MICs) in the in vitro cultures of Mtb strain H37Rv and M. smegmatis str. mc2 155 (Msm), a fast-growing Mycobacterium model. Msm and Mtb DnaGs are nearly identical (82% sequence identity), with most differences exhibited in the C-terminal, replicative helicase-binding domain, which is not required for the primer synthesis activity of DnaG in vitro.8 A correlation between the in vitro antagonism of Mtb DnaG and the inhibition of the mycobacterial cell growth for these compounds strongly suggests that DnaG inhibition contributes significantly to their antimicrobial activity.

Materials and Methods

Expression and purification of Mtb DnaG

The Mtb DnaG protein was expressed and purified by a modified version of our previously published protocol,9 as follows. Protein expression was carried out in Escherichia coli BL21 (DE3) cells cultured in LB broth supplemented with ampicillin (100 μg ml−1). A 2-l culture was grown to an attenuance at 600 nm of 0.2 and induced with 0.5 mM of IPTG and incubated for 16 h at 18 °C. (Note: All purification steps were carried out at 4 °C, without freezing the bacterial pellet). The cells were harvested and the pellet was suspended in 50 ml of lysis buffer (40 mM Tris pH 8.0, 600 mM NaCl, 5% v/v glycerol, 1 mM PMSF, 2 mM MgCl2 and 2 mM β-mercaptoethanol). The cells were disrupted by sonication on ice and clarified by centrifugation at 40 000 × g for 40 min. The supernatant was filtered through a 0.45-μm Millex-HV PVDF filter (Millipore, Billerica, MA, USA) and applied to a 1-ml Ni-IMAC HisTrap FF column (GE Healthcare Life Sciences, Pittsburgh, PA, USA) equilibrated with lysis buffer. The column was washed with 20 ml of lysis buffer containing 50 mM imidazole, and the protein was eluted with 11 ml of lysis buffer containing 500 mM imidazole. The fractions containing protein were loaded onto a size-exclusion S-200 column (GE Healthcare Life Sciences) equilibrated in gel filtration buffer (40 mM Tris pH 8.0, 600 mM NaCl, 5% v/v glycerol and 2 mM of β-mercaptoethanol), and the protein-containing fractions were pooled and concentrated using an Amicon Ultra-15 centrifugal filter device (Millipore) to 3 mg ml−1 final concentration. The protein was then flash frozen in 30-μl aliquots in liquid nitrogen and stored at −80 °C. The freezing process did not affect the protein activity; the aliquots were used immediately upon thawing and were not reused later or refrozen. The presence of the N-terminal His-tag did not have any effect on the protein activity, as compared with the previously published results.9 Mtb PPiase was expressed and purified as previously described.9

Dose–response assays

The dose–response assays were performed in 96-well plates as previously described.9 Primase activity measurements were performed in triplicate. The dose–response curves were analyzed by nonlinear regression with SigmaPlot 9.0 (SysStat Software, San Jose, CA, USA). The following general dose-response equation was used in nonlinear regression data fitting of the relative protein activity, f, as a function of inhibitor concentration, [I], with IC50 and Hill coefficient, n, as fitting parameters:

Hill coefficients greater than unity were observed for the anthracyclines and the mitoxantrone derivative, an indication of multiple molecules of these inhibitors interacting with DNA/DNA-DnaG complex or self-association of these compounds, likely through ring-ring stacking interactions. Inhibition of Mtb PPiase alone by all the compounds of interest was tested by an analogous assay with 50 μM of sodium pyrophosphate as a substrate, in the absence of Mtb DnaG, DNA and NTP. Absorbance was measured at 30 s (when approximately half of the PPi was cleaved), and at 4 min. The reactions were carried out without a compound and at 100 μM of each compound. No inhibition of Mtb PPiase was observed by any of the molecules tested.

Determination of MIC values against mycobacterial culture growth

Determination of MIC values against Mtb by the alamar blue assay

The MIC for each compound was determined using the microplate alamar blue assay as previously described with slight modifications.14 All compounds were diluted to 20 μM working stocks in 7H9 medium. The working stocks were twofold serially diluted to achieve compound concentrations between 640 and 0.156 μM. A volume of 200 μl sterile distilled water (ddH2O) was added to all perimeter wells of 96-well test plates to reduce evaporation from test wells. The 7H9 medium without compound served as a growth control and uninoculated 7H9 medium was included as a sterility control. A volume of 100 μl 7H9 medium containing test compounds at the various concentrations was added to the other wells. Mtb strain H37Rv was inoculated from frozen stocks into Middlebrook 7H9 broth supplemented with 10% albumin-dextrose-catalase (ADC, BD Biosciences, San Jose, CA, USA), 0.05% Tween80 (Sigma-Aldrich, St Louis, MO, USA) and 0.4% v/v glycerol and incubated at 37 °C until turbid. Cultures were then diluted to an attenuance at 600 nm of 0.2 in fresh 7H9 medium, then additionally diluted 1 : 25 in 7H9, added to 50 ml polypropylene tubes containing glass beads, vortexed for 30 s, allowed to settle for 10 min and 100 μl was distributed into the wells of columns 2–10 of the test plate making the final concentration of test compounds 0.078–10 μM. Compounds with initial MICs >10 μM were similarly tested at compound concentrations between 10 and 320 μM. Compound screening was carried out on biological replicates in triplicate. After the plates were incubated at 37 °C in a humid environment for 5–6 days, 40 μl of alamar blue diluted 1 : 2 in 10% Tween80 was added to each well and the plates were incubated at 37 °C. The color of each well was preliminarily evaluated 24 h after the addition of alamar blue, with a final evaluation after 48 h. Alamar blue changes from indigo blue to pink as a result of bacterial growth. The lowest concentration of compound that resulted in no color change was recorded as the MIC for each compound.

Determination of MIC values against Msm

MIC values were determined using the double-dilution method starting at 150 μg ml−1 in a total volume of 200 μl. Dilutions of a Msm str. mc2 155 culture were added to the solutions of the tested compounds dissolved in Mueller-Hinton broth. Bacteria were grown at 37 °C until cultures became turbid, 2 days. MIC values were determined as the last dilution to have no bacterial growth as determined by alamar blue (resazurin; 5 μl of a 7 mg ml−1 solution) staining for 2 days. Wells showing no color change from the standard blue color were determined to have no bacterial growth.

Results

Inhibition of Mtb DnaG by a series of DNA intercalator agents

By using our recently developed colorimetric coupled primase–pyrophosphatase assay with purified Mtb DnaG and Mtb PPiase, we investigated inhibition of purified Mtb primase DnaG by seven compounds: the anthracyclines doxorubicin, daunorubicin and idarubicin, the natural anthranoids aloe-emodin and rhein, an anthracenedione di-glucosyl mitoxantrone and the fluoroquinolone ofloxacin (Figure 1). These compounds are structurally diverse DNA intercalating agents, previously reported to have some antibacterial activity. Consistent with previously observed low-μM inhibition of Mtb DnaG by doxorubicin,9 structurally related daunorubicin and idarubicin also displayed IC50 values for Mtb DnaG inhibition in the low-μM range: IC50=7.2±0.3 and 8.2±1.1 μM for daunorubicin and idarubicin, respectively (Figure 2a and Table 1). Aloe-emodin, a plant-derived compound with cathartic and modest anticancer activities,15 exhibited significant, albeit weaker inhibition of this enzyme with IC50=19±2 μM (Figure 2b and Table 1). A close structural analog of aloe-emodin, also with cathartic and anticancer activities, rhein,16 however, did not inhibit Mtb DnaG to a measurable degree (IC50>100 μM) (Figure 2b and Table 1). Another DNA intercalator structurally distinct from any of the above compounds, a derivative of an anticancer drug, mitoxantrone, inhibited Mtb DnaG about twofold more weakly (IC50=38±1 μM) than aloe-emodin (Figure 2c and Table 1). Finally, the fluoroquinolone ofloxacin, which targets DNA gyrase, did not show any observable inhibition of Mtb DnaG (Table 1).

Figure 1
figure 1

Structures of small molecules used in this study.

Full size image
Figure 2
figure 2

Primase–PPiase dose-response assays with (a) daunorubicin (white circles) and idarubicin (black circles), (b) aloe-emodin (black squares) and rhein (white squares) and (c) a glycosylated mitoxantrone derivative (white triangles). The assays were performed with 1.25 μM of DNA, 110 μM of NTP and 0.6 μM of Mtb DnaG. Daunorubicin (IC50=7.2±0.3 μM) and idarubicin (IC50=8.2±1.1 μM) showed low-μM range inhibition. Daunorubicin and idarubicin could be well modeled by using the Hill coefficients (n) of 2.1±0.1 and 2.8±0.8, respectively. Aloe-emodin displayed weaker inhibition of the enzyme (IC50=19±2 μM), and rhein did not show any significant inhibition (IC50>100 μM). The mitoxantrone derivative showed much weaker inhibition (IC50=38±1 μM) with n=5.0±1.8.

Full size image
Table 1 Mtb DnaG and Mtb PPiase IC50 values and mycobacterial MIC values for DNA intercalators tested
Full size table

Inhibition of in vitro growth of Mtb and Msm by the DNA intercalators

To test whether the identified Mtb DnaG inhibitors halt growth of mycobacteria, and not only function on purified enzyme, we determined the MIC values for all seven molecules studied against two strains of mycobacteria, Msm mc2 155 and Mtb H37Rv (Table 1). Doxorubicin inhibited Msm and Mtb with an MIC of 8 and 5 μM, respectively, comparable with the IC50 values in the primase assay. Daunorubicin displayed more potent growth inhibition of Msm and Mtb, with an MIC of 0.5 and 1.25 μM, respectively. On the other hand, idarubicin strongly inhibited Msm (MIC=0.6 μM), but not Mtb (MIC=80 μM). Aloe-emodin displayed intermediate potency against Msm (MIC=70 μM) and did not show inhibition of Mtb up to 320 μM. Interestingly, rhein did not show any inhibition of either Msm (MIC>260 μM) or Mtb (MIC>320 μM). Similarly, the di-glucosylated mitoxantrone derivative did not display inhibition of either Msm (MIC>100 μM) or Mtb (MIC>320 μM).

Discussion

Primase DnaG has recently emerged as a promising potential target for the discovery of antibacterial agents.9,10,17 HTS-driven search for inhibitors of Mtb DnaG and Bacillus anthracis DnaG yielded doxorubicin, a Streptomyces natural product and a known bacterial DNA replication inhibitor12 that is a highly efficient antineoplastic agent commonly used in the treatment of hematopoietic and solid tumors. These studies strongly suggest that doxorubicin evolved to target DnaG and that anthracyclines in general and other DNA intercalators may have antimycobacterial activity and efficiently inhibit DnaG. Indeed, we showed here that the anthracycline daunorubicin, like doxorubicin, is an inhibitor of DnaG with similar potent antibacterial activity against both Msm and Mtb. In fact, daunorubicin is 4 to 16-fold more potent than doxorubicin in growth inhibition of both mycobacteria. This may be attributed to an additional mode of action such as oxidative damage induced by this anthracycline, more efficient penetration through the cell envelope or more potent inhibition of DnaG in the context of the entire replication machinery. On the other hand, idarubicin, similar to daunorubicin, is more potent than doxorubicin against Msm, but it appears to be less potent than doxorubicin in inhibiting Mtb. Specific structural features of the cell envelope of Mtb may prevent idarubicin from penetrating into the cytoplasm of Mtb.

Another DNA intercalator of a different structural class, aloe-emodin, displayed MIC values against Msm in a mid-μM range, but exhibited no growth inhibition of Mtb strain H37Rv. A recent study showed that aloe-emodin is a weak (~100 μM) inhibitor of H37Ra strain of Mtb.18 The weaker activity of aloe-emodin against Mtb than against Msm may be explained by differences in the cell envelopes of these two mycobacteria. It has been shown that the presence of surface-exposed C-type glycopeptidolipids (GPLs) in Msm increases cell-wall barrier permeability. These GPLs are species-specific and found only in nontuberculous mycobacterial species. The GPL-deficient mutant strain exhibited an increase of the cell hydrophobicity.19 The many polar groups of emodin and its small size may enable it to cross the outer membrane of Msm more efficiently. This phenomenon is thought to account for the resistance of the outer membrane of Msm to rifampicin, a hydrophobic antibiotic.20 Furthermore, the presence of porins, mainly MspA, in the outer membrane of Msm may allow the transport of small molecules decorated with hydrophilic groups. A mutant strain with a deletion of mspA exhibited a ninefold and fourfold reduction in permeability for cephaloridine and glucose, respectively.21 No sequence homologs to MspA were identified in Mtb.

In contrast, a close analog of aloe-emodin, rhein (US patent US5652265, 1997), which contains a carboxylic acid moiety instead of hydroxymethyl groups of aloe-emodin, shows no measurable growth inhibition of either Msm or Mtb. In semiquantitative agreement with these MIC measurements, aloe-emodin is a mid-μM range inhibitor of Mtb DnaG, whereas rhein does not inhibit this enzyme to an observable extent. These data strongly suggest that inhibition of DnaG has a major role in the antibacterial activity of aloe-emodin. An analog of aloe-emodin, emodin, and rhein appear to cross a variety of bacterial cell envelopes of nonmycobacteria equally well,22 although we cannot exclude the possibility that the inactivity of the charged rhein against both mycobacteria may be due to some specific features of the mycobacterial envelope, such as multidrug resistance pumps.23 A recent study demonstrated that emodin is an ~100 μM inhibitor of gyrase (whose catalytic domain shares the catalytic TOPRIM fold with DnaG) and topoisomerase I,8 which, together with our results, suggests that the emodin scaffold is efficient in targeting gyrases and bacterial primases. In the same study, halogenated emodins or haloemodins were shown to inhibit bacterial topoisomerases much more potently than the parent compound.24 Extending the correlation of MIC and IC50 of DnaG inhibition among the DNA intercalators tested here, the glycosylated derivative of mitoxantrone was inert toward Msm and Mtb and it was very weakly inhibitory to Mtb DnaG (IC50=38±1 μM). Notably, a recent study identifies mitoxantrone as an inhibitor of E. coli gyrase (IC50=80 μM).25 Therefore, mitoxantrones appear to exhibit an analogous target profile to that of emodins. Another DNA intercalator of a distinct structural class, the fluoroquinolone ofloxacin, a known potent antimycobacterial agent (MIC=3 μM against both Msm26 and Mtb27) acting as an inhibitor of another DNA-binding enzyme, gyrase, did not have an observable inhibition effect on Mtb DnaG (IC50>100 μM), serving as a control in this study.

The mechanism of action of DNA-binding antibiotics in bacteria has not been well understood, except for (i) agents that result in DNA breaks indirectly through binding to DNA breaking enzyme–DNA complexes, such as quinolones, and (ii) agents that directly result in DNA breaks, such as bleomycin or calicheamicin, or crosslink DNA, like mitomycin. It is clear that DNA binding alone cannot explain the mechanism of action of DNA intercalators and other DNA-binding agents, because the equilibrium constants (Kd) for rather nonspecific binding of these compounds to DNA are often much larger than MIC values. The biological activity of DNA-binding compounds in many cases must rely on inhibition of a particular mechanism, likely one involving DNA interaction with a specific DNA-binding protein. The current work in conjunction with other recent studies from our group argues that doxorubicin and other anthracyclines as well as aloe-emodin inhibit mycobacterial growth by inhibiting DnaG. Historically, anthracyclines have not been considered as leads as first-line anti-TB drugs owing to their toxicity and suppression of the immune system. Nevertheless, an anthracycline analog or its formulation with less toxicity against human cells, but still potent as a DnaG inhibitor, could be potentially useful as an antibacterial, especially against strains resistant to conventional anti-TB therapy. For example, the anthracycline aclacinomycin has long been known to be much less toxic than doxorubicin.28 The recent development of anticancer derivatives of doxorubicin with lower toxicity has been promising. Peptide–doxorubicin conjugates, such as AEZS-108, demonstrate higher anticancer selectivity and far lower toxicity against noncancer cells than doxorubicin.29, 30, 31, 32 Analogous antibacterial targeting strategies for discovery of an antracycline analog could be envisioned. 4'-Iodo-4'-deoxydoxorubicin, methoxymorpholinyl doxorubicin and 3'-azido doxorubicin are other examples of doxorubicin analogs with improved toxicity profiles.33, 34, 35 Liposomal formulations of doxorubicin have shown lower cardiac toxicity than doxorubicin. In summary, this study showcases DnaG as a potential target for future investigations with DNA intercalators and other inhibitory compounds.