In Vitro and In Vivo Activities of Macrolide Derivatives against Mycobacterium tuberculosis

Abstract

Existing macrolides have never shown definitive clinical efficacy in tuberculosis. Recent reports suggest that ribosome methylation is involved in macrolide resistance in Mycobacterium tuberculosis, a mechanism that newer macrolides have been designed to overcome in gram-positive bacteria. Therefore, selected macrolides and ketolides (descladinose) with substitutions at positions 9, 11,12, and 6 were assessed for activity against M. tuberculosis, and those with MICs of ≤4 μM were evaluated for cytotoxicity to Vero cells and J774A.1 macrophages. Several compounds with 9-oxime substitutions or aryl substitutions at position 6 or on 11,12 carbamates or carbazates demonstrated submicromolar MICs. For the three macrolide-ketolide pairs, macrolides demonstrated superior activity. Four compounds with low MICs and low cytotoxicity also effected significant reductions in CFU in infected macrophages. Active compounds were assessed for tolerance and the ability to reduce CFU in the lungs of BALB/c mice in an aerosol infection model. A substituted 11,12 carbazate macrolide demonstrated significant dose-dependent inhibition of M. tuberculosis growth in mice, with a 10- to 20-fold reduction of CFU in lung tissue. Structure-activity relationships, some of which are unique to M. tuberculosis, suggest several synthetic directions for further improvement of antituberculosis activity. This class appears promising for yielding a clinically useful agent for tuberculosis.

Tuberculosis (TB) has been recognized as a major public health problem worldwide, exacerbated greatly by the human immunodeficiency virus pandemic. The length and complexity of antibiotic therapy for tuberculosis and the emergence of multidrug-resistant strains make a compelling case for the development of new efficacious anti-TB drugs.

The development of new members of classes of established antibiotics, such as the macrolides, which already possess many desirable pharmacological properties, is one approach to rapidly add new drugs to the existing anti-TB armamentarium.

Although they are antibiotics of choice for several respiratory pathogens and have demonstrated efficacy in other mycobacterioses, such as leprosy (6, 15, 17) and Mycobacterium avium (14, 26, 27) infections, the macrolides developed to date have lacked potency against the tubercle bacillus (24). The clinical use of clarithromycin in treating tuberculosis is limited to multiple-drug-resistant cases in which there are few remaining treatment options (21).

The current development goal for macrolides has been primarily to overcome ribosome modification and drug efflux, the major mechanisms of resistance to this antibiotic class (22). Methylation of A2058, located in the peptidyl transferase loop of domain V of the 23S rRNA subunit, by specific methylases (erm) decreases macrolide binding affinity, rendering the organisms resistant to macrolides, lincosamides, and streptogramins, known as the MLS_B phenotype (28, 29). Macrolides in which the cladinose group is replaced with a keto group (ketolides) avoid efflux-mediated resistance and fail to induce erm, while specific substitutions at positions 6, 9, and 11,12 enhance binding to the ribosome.

Telithromycin, the first clinically approved ketolide, is effective in treating respiratory tract infections, including those caused by erythromycin-resistant bacteria with an inducible MLS_B phenotype. However, not all erythromycin-resistant bacteria are susceptible to telithromycin, including gram-positive organisms with extensively dimethylated A2058 as well as Mycobacterium tuberculosis (24). Although innate macrolide resistance in M. tuberculosis also appears to be mediated by erm (3), there is not yet direct evidence of whether this results in mono- or dimethylation. In order to gain a better understanding of structure-activity relationships (SAR) with respect to this organism and to identify new macrolides or ketolides with superior activity against M. tuberculosis, more than 30 macrolides and ketolides with a variety of substitutions were evaluated.

MATERIALS AND METHODS

Compounds.

Rifampin was purchased from Fisher Scientific, the macrolides listed in Table 1 were purchased from Sigma, cethromycin (ABT-773) and A323348 were provided by Abbott Laboratories, telithromycin (RU647 and HMR3647) and other RU macrolides were provided by Aventis, and clarithromycin was a gift from Alexander Mankin. ITR macrolides were synthesized in-house as previously described (1).

TABLE 1.

Comparative in vitro activities of macrolides

Macrolide	MIC (μM)
Older (pre-1985)
Tylosin tartrate	64
Erythromycin	128
Erythromycin-ethylsuccinate	128
Erythromycin-estolate	128
Erythromycin-stearate	128
Oleandomycin	>128
Troleandomycin	>128
Spiramycin	>128
Rosamicin	>128
Midecamycin	>128
Expanded spectrum (since 1985)
Clarithromycin	8
Roxithromycin	32
Azithromycin	128
Cethromycin	4
Telithromycin	128
Rifampin	0.07

MICs.

The MICs of macrolides for M. tuberculosis were determined by the microplate Alamar blue assay (MABA) (7). Rifampin was included as a control. Macrolide stock solutions were prepared in dimethyl sulfoxide at a final concentration of 10.24 mM, and the final testing concentrations ranged from 128 to 2 μM. For the most active compounds, the stock concentration and final testing concentration range were lowered to 640 μM and to 8 to 0.125 μM, respectively. M. tuberculosis H₃₇Rv (ATCC 27294; American Type Culture Collection, Rockville, Md.) was grown to late log phase (70 to 100 Klett units) in Middlebrook 7H9 broth supplemented with 0.2% (vol/vol) glycerol, 0.05% Tween 80, and 10% (vol/vol) oleic acid-albumin-dextrose-catalase. Cultures were centrifuged for 15 min at 4°C at 3,150 × g, washed twice, and resuspended in phosphate-buffered saline. Suspensions were then passed through an 8-μm-pore-size filter to remove clumps, and aliquots were frozen at −80°C. The number of CFU was determined by plating on 7H11 agar plates.

Twofold dilutions of macrolides were prepared in Middlebrook 7H12 medium (7H9 broth containing 1 mg of Casitone/ml, 5.6 μg of palmitic acid/ml, 5 mg of bovine serum albumin/ml, and 4 mg of catalase/ml; filter sterilized) in a volume of 100 μl in 96-well, black, clear-bottom microplates (BD Biosciences, Franklin Lakes, N.J.). M. tuberculosis H₃₇Rv (100 μl continuing 2 × 10⁴ CFU) was added, yielding a final testing volume of 200 μl. The plates were incubated at 37°C; on the seventh day of incubation, 12.5 μl of 20% Tween 80 and 20 μl of Alamar blue were added to all wells. After incubation at 37°C for 16 to 24 h, the fluorescence of the wells was read at an excitation of 530 nm and emission of 590 nm. The MIC was defined as the lowest concentration effecting a reduction in fluorescence of ≥90% relative to the mean of replicate bacteria-only controls.

Clinical isolates were obtained from Ramathibodhi Hospital, Bangkok, Thailand. Isolates were initially cultured on solid medium, and inocula were standardized by using a McFarland standard of 1 and then further diluted. Susceptibility to clinical TB agents was determined by using the proportion technique as well as a MABA with visual readings (12). The latter was also used to determine the MICs of macrolides.

CLogP.

The CLogP values (estimation of hydrophobicity or partition between octanol and water) were calculated on the CLogP website of Daylight Systems Inc. (http://www.daylight.com/daycgi/clogp) by using the SMILES structure format obtained through ACD/ChemSketchFreeware.

Cytotoxicity.

Vero cells (ATCC CRL-1586) were cultured in 10% fetal bovine serum (FBS) in minimum essential medium Eagle (4). J774A.1 cells were cultured in 10% FBS in Dulbecco's modified Eagle's medium (DMEM). The cells were incubated at 37°C under 5% CO₂ until confluent and then diluted with phosphate-buffered saline to 10⁶ cells/ml. In a transparent 96-well plate (Falcon Microtest 96), threefold serial dilutions of the macrolide stock solutions resulted in final concentrations of 102.4 to 0.42 μM in a final volume of 200 μl. After incubation at 37°C for 72 h, medium was removed and monolayers were washed twice with 100 μl of warm Hanks' balanced salt solution (HBSS). One hundred microliters of warm medium and 20 μl of freshly made MTS-PMS [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium and phenylmethasulfazone] (100:20) (Promega) were added to each well, plates were incubated for 3 h, and absorbance was determined at 490 nm.

Macrophage assay.

J774A.1 cells were grown to confluency in 75-cm² cell culture flasks in DMEM medium containing 10% FBS. Using a cell scraper, the cells were detached and centrifuged at 200 × g for 5 min at room temperature, and the pellet was suspended to a final concentration of 1 × 10⁵ to 3 × 10⁵ cells/ml. One-milliliter aliquots of cell suspension were distributed into 24-well plates (Falcon Multiwell 24 well) containing 13-mm coverslips (Nalge Nunc International), and the plates were incubated at 37°C in a 5% CO₂ incubator for 16 h. M. tuberculosis Erdman (ATCC 35801) frozen cultures were thawed, sonicated for 15 s, and diluted to a final concentration of 1 × 10⁵ to 3 × 10⁵ CFU/ml with DMEM, and 500 μl of the dilution was dispensed to each well of a new 24-well plate. J774A.1 cells on coverslips were transferred to the 24-well plates containing M. tuberculosis Erdman, and the plates were incubated at 37°C for 2 h to allow for phagocytosis. Coverslips were rinsed with HBSS to remove the extracellular bacteria, and the coverslips were transferred to new 24-well plates with 1 ml of fresh medium in each well. Cultures were incubated at 37°C under 5% CO₂ for 16 h and then transferred to 1 ml of fresh medium containing macrolides at 0.06, 0.32, 1.6, and 8 μM. All experimental conditions were set up in triplicate.

At T₀ (for untreated controls) and after 7 days of incubation, medium was removed and macrophages were lysed with 200 μl of 0.25% sodium dodecyl sulfate. After 10 min of incubation at 37°C, 200 μl of fresh medium was added. The contents of the wells were transferred to a microtube and sonicated (Branson Ultrasonics model 1510; Danbury, Conn.) for 15 s, and 1:1, 1:10, 1:100, and 1:1,000 dilutions were plated on 7H11 (Difco) agar plates. Colonies were counted after incubation at 37°C for 2 to 3 weeks.

Maximum tolerated dose.

Prior to assessing in vivo efficacy, RU66252 and RU69874 were administered by oral gavage to pairs of female BALB/c mice once daily for 5-day cycles, and mice were observed for overt signs of toxicity (weight loss, ruffled fur, huddling, etc.), after which the mice were rested for 1 or 2 days and the dosage was increased for another 5-day cycle. The maximum administered dosages of RU69874 and RU66252 were 500 and 1,000 mg/kg of body weight, respectively.

Low-dose aerosol model of acute infection.

Eight-week-old female BALB/c mice were infected via aerosol with a suspension of 5 × 10⁶ CFU of M. tuberculosis Erdman/ml by using a Glas-Col inhalation system. Macrolides were dissolved or suspended in 0.5% carboxymethyl cellulose (CMC) and administered by oral gavage in a maximum volume of 200 μl. Mice were treated daily during the acute phase of infection from day 10 until day 30 postinfection. Each treated group was composed of 5 or 7 mice, while the control group, which received only CMC, was composed of 7 to 10 mice.

The mice were sacrificed the day after the last day of treatment, and the lungs were removed, homogenized, and serially diluted 10-fold in HBSS. One hundred microliters was plated on 7H11 agar in duplicate. The plates were incubated at 37°C for 2 to 3 weeks. The CFU data were analyzed and plotted with SAS software, version 8.2.

RESULTS

In vitro SAR.

A comparative study of older and expanded-spectrum macrolides suggests a pattern of improved activity with newer agents, as exemplified by MICs of clarithromycin (8 μM) and cethromycin (4 μM) (Table 1).

Additional macrolides and ketolides were evaluated for anti-TB activity and cytotoxicity for Vero and J774A.1 cells (Tables 2 -4) Comparison of three sets of macrolide-ketolide counterparts revealed 16-, 337-, and >16-fold decreases in MICs of macrolides versus their ketolide counterparts for RU66252 and RU004 (Table 4), RU69874 and telithromycin (Table 3), and clarithromycin and RU56006 (Table 2), respectively.

TABLE 2.

In vitro activity and selectivity (μM) of 11,12-diol erythromycin derivatives

^aSI, selectivity index = IC₅₀/MIC.

TABLE 4.

In vitro and selectivity (μM) of 11,12-carbazate derivatives

^aSI, selectivity index = IC₅₀/MIC.

TABLE 3.

In vitro activity and selectivity (μM) of 11,12-carbamate derivatives

^aSI, selectivity index = IC₅₀/MIC.

Among 11,12 diols, substitution of the 9-carbonyl of erythromycin with 9-oximes, as represented by roxithromycin and RU29558, yielded increases in activity of 4- and 18-fold, respectively (Table 2). In contrast, substitution of the inactive 9-carbonyl-substituted ketolide 56006 with a 9-oxime-substituted analog, RU54615, failed to increase activity. Although a 9-oxime piperidine (ITR003) also did not increase activity significantly, further modification of the piperidine moiety enhanced activity, with the decyl-substituted analog (RU60887) demonstrating a >1,000-fold increase in potency. Among the four compounds with a piperidine substitution, activity appeared to correlate with calculated hydrophobicity (CLogP) of the R9 substituent.

A variety of aryl-substituted 11,12 carbamate macrolides and ketolides demonstrated low or submicromolar MICs (Table 3), telithromycin excluded. However, most of these compounds also had relatively low 50% inhibitory concentrations (IC₅₀s) for Vero and/or J774A.1 cells, indicating relatively poor selectivity. The macrolides RU60856 and RU69874 were notable exceptions, the former with less cytotoxicity and the latter demonstrating better anti-TB activity. Among three compounds tested with an unsubstituted 11,12 carbamate, the 6-quinolinyl propenyl carbamate (cethromycin; ABT-773) was 32-fold more active than the allyl-substituted RU192803. However, A323348, with a fluorine at position 2 and a quinolinyl isoxazolyl propynyl at R6, demonstrated the best activity and selectivity among these compounds. Introduction of a carbazate at position 11,12 also resulted in several compounds with relatively low MICs (Table 4). The macrolide RU66252 had slightly better activity than its carbamate analog, RU66080 (Table 3), and 16-fold better activity than its carbazate ketolide, RU004. The latter, in turn, had 4-fold-better activity than its 2-fluoro analog (RU3562). RU004 and RU69697 differ only in linker length but have the same MIC.

Three of the most active compounds, along with clarithromycin, were tested against a panel of 10 pan-sensitive and drug-resistant isolates of M. tuberculosis. Similar to that observed in previous studies (16, 19, 25), there was considerable variability among the isolates with respect to clarithromycin MICs (Table 5). Overall, the newer macrolides had lower MICs than clarithromycin. Exceptions were three isolates for which clarithromycin MICs were submicromolar; the MIC of RU66252 was equivalent to the clarithromycin MIC and those of RU69874 and RU60887 were equivalent or slightly higher.

TABLE 5.

MICs of macrolides for drug-sensitive and resistant clinical isolates of M. tuberculosis^a

Isolate resistance profile	MIC (μM)
	CLR	RU66252	RU69874	RU60887
Pan sensitive (H37Rv)	4.0	0.5	0.5	1.0
Pan sensitive	0.25	0.25	0.25	0.5
Pan sensitive	2.0	0.5	0.5	0.5
Pan sensitive	2.0	0.25	0.25	2.0
Pan sensitive	0.5	0.5	0.5	1.0
STR	0.125	0.125	0.25	0.5
STR	4.0	2.0	1.0	1.0
INH	16.0	4.0	4.0	4.0
INH, RIF, STR	2.0	0.5	0.5	1.0
INH, RIF	32.0	1.0	4.0	4.0
INH, RIF, EMB, STR	16.0	1.0	4.0	2.0

^aINH, isoniazid; RIF, rifampin; STR, streptomycin; EMB, ethambutol; CLR, clarithromycin. MABA MICs were determined by visual reading.

Compounds with the best activity and selectivity were evaluated for the ability to inhibit M. tuberculosis Erdman, since this strain grows better than H₃₇Rv in macrophages and mice and was subsequently used for these assays. The Erdman strain had a susceptibility similar to that of the H₃₇Rv strain for these compounds; the MICs of RU66252, A323348, RU69874, and RU60887 for this strain were 0.5, 0.75, 0.38, and 1 μM, respectively.

Activity against M. tuberculosis in macrophage culture.

Dose-responsive bactericidal activity against M. tuberculosis Erdman in J774A.1 macrophages was observed for RU66252, A323348, and cethromycin as well as for clarithromycin and rifampin controls (Fig. 1). RU66252 and A323348 effected a 1 to 2 log reduction in viability at 1.6 to 8 μM, while cethromycin and clarithromycin demonstrated more modest activity. RU69874 and RU60887 showed little activity, and the apparent activity of the latter at 8 μM may have been due to cytotoxicity, as the IC₅₀ for J774A.1 cells was less than 5 μM (Table 3). A preliminary experiment revealed a dose-dependent reduction in growth inhibition in axenic medium when FBS was added at 1 and 5% (vol/vol), thus offering one possible explanation of the reduced activity in cell culture which contains 10% FBS (data not shown).

FIG. 1.

Dose response of macrolides against M. tuberculosis Erdman in J774A.1 cells. Triplicate cultures were treated with macrolides at 0.06, 0.32, 1.6, and 8 μM. The mean (and standard deviation [SD]) CFU/well at the beginning and end of treatment was 1.3 × 10³ (0.42 × 10³) and 1 × 10⁴ (0.29 × 10⁴), respectively.

In vivo activity.

In preliminary tolerance tests, no signs of overt toxicity were noted at dosages up to 1 g of RU66252/kg administered q.d. for five consecutive days. A preliminary evaluation of the activities of RU66252, RU69874, RU60887, and clarithromycin was performed in a low-dose aerosol infection mouse model. Dose-dependent activity at 100 and 200 mg/kg was observed for the carbamate RU69874 and the carbazate RU66252, with the latter effecting a more pronounced inhibition (Table 6). RU60887 was not significantly active at 100 mg/kg and was not evaluated at higher dosage due to toxicity concerns based on the IC₅₀ for J774A.1 cells (Table 2). The reduction in CFU observed with 200 mg of clarithromycin/kg was modest, with a P value of just under 0.05.

TABLE 6.

Activities of macrolides against M. tuberculosis Erdman in BALB/c mice^a

Compound (mg/kg)	Log10 CFU
	Mean	SD	CV (%)	P
CMC	4.86	4.73	75
RIF (25)	<1.48			<0.0001
CLR (200)	3.92	3.77	70	0.0455
RU66252 (200)	<2.57			<0.0001
RU66252 (100)	4.24	4.37	136	0.0099
RU69874 (200)	4.16	4.33	148	0.0010
RU69874 (100)	4.8	4.62	66	0.9990
RU60887 (100)	4.39	4.34	90	0.9750

^aCV, coefficient of variation; RIF, rifampin; CLR, clarithromycin.

Dose-dependent activity of RU66252 was confirmed in a second experiment (Fig. 2), with significant activity observed at dosages of 100, 150, and 200 mg/kg (P values of 0.0003, <0.0001, and < 0.0001, respectively).

FIG. 2.

Dose response of RU66252 against M. tuberculosis Erdman in BALB/c mice. All posttreatment lung CFU were determined at day 31 postinfection; however, data points are offset for clarity.

DISCUSSION

Although it appears that M. tuberculosis may share at least one mechanism of resistance with macrolide-resistant Staphylococcus and Streptococcus species (8), the first approved ketolide, telithromycin, fails to demonstrate activity against the tubercle bacillus (Tables 1 and 3) (24). However, other analogs, including both macrolides and ketolides bearing various substitutions at the 6, 9, and 11,12 positions, had MICs of ≤4 μM, which represents an improvement over earlier macrolides.

Consistent with observations in Escherichia coli (10), in the absence of aryl substitutions at positions 6 or 11,12, the lack of a cladinose moiety and the substitution with a keto group at C-3 resulted in a decrease in activity against M. tuberculosis of 16-fold (RU56006 versus clarithromycin). The presence of an aryl group at positions 6 or 11,12 has been associated with improved activity against MLS-resistant strains (10, 20) in which domain V of the 23S ribosome is modified by either mutation or methylation. Improved activity is apparently the result of enhanced binding in domain II. In both wild-type and A2058G and C2611U (MLS_B resistance phenotype) mutants of E. coli, the ketolide RU004, with an aryl-substituted 11,12 carbamate, was twice as active as the corresponding macrolide, RU66252, overcompensating for the loss of the cladinose group (10). However, in M. tuberculosis, aryl-substituted 11,12 carbamate or carbazate macrolides (including RU66252) were more active than their ketolide counterparts (Table 3 and 4). Therefore, the cladinose-containing macrolides, even in the presence of an aryl group at position 11,12, have superior activity relative to their ketolide analogs against M. tuberculosis.

Macrolides with the cladinose group can induce MLS_B, while the ketolide counterparts do not (2). However, ketolides have not been effective in strains constitutively expressing MLS resistance (MLS_B) (20), and considered together with the results of this study, this finding may suggest that TB ribosomes are constitutively methylated or that requirements for induction differ. The ketolide cethromycin, with activity clearly superior to that of erythromycin against M. tuberculosis, has been shown to bind to methylated ribosomes better than erythromycin in Streptococcus pneumoniae (5), but this has yet to be directly determined in M. tuberculosis.

Macrolides with 9-oxime substitutions have been studied previously (13, 18). 9-Oxime-substituted analogs were more active than erythromycin against erythromycin-resistant organisms, such as the M. avium complex, Staphylococcus aureus (23), and Mycobacterium leprae (11). Quantitative SAR studies revealed that the substitution of 9-oximes with short alkyl groups gives activity comparable to that of clarithromycin, while the introduction of more-bulky alkyl groups improves activity against S. aureus (23). Results with macrolides for M. tuberculosis were consistent with these observations, in that both of the C-9 oxime-substituted erythromycin analogs, roxithromycin and RU29558, were markedly more active than erythromycin (9-carbonyl) (Table 2). Conversely, the 9-oxime-substituted ketolide, RU54615, did not have improved activity relative to its carbonyl analog, RU56006 (both MICs were >128 [Table 2]). Among the 11,12 diols with 9-oxime substitutions, there was a direct correlation between the hydrophobicity of the piperidine-based substituents and activity. Hydrophobicity has previously been associated with improved activity of clarithromycin versus erythromycin in M. avium and Mycobacterium smegmatis, at least partially due to enhanced penetration and ribosome binding (9).

Because only three C-6-substituted 11,12-unsubstituted carbamates were evaluated in this study, elucidation of the SAR at position 6 was limited to the observation that the quinolinyl propenyl markedly enhanced activity compared to the allyl substitution alone. The promising activity of A323348, however, suggests that the evaluation of closely related analogs may be fruitful.

The intracellular activities of the macrolides evaluated in this study (in contrast to clarithromycin and rifampin) appeared somewhat less potent than those observed against M. tuberculosis in axenic medium (as represented by the MICs). Possible factors that may have contributed to this reduced activity include protein binding by FBS, lower intracellular pH, and in the case of RU60887, a higher MIC for strain Erdman than for H₃₇Rv.

Previous studies of the activity of clarithromycin against M. tuberculosis in mouse models indicated that at 200 mg/kg, this macrolide is either not significantly active (17) or is weakly active (19). Results from this study were consistent with these observations, indicating weak activity at the threshold of statistical significance (Table 6). The substituted carbazate RU66252 was significantly more active than clarithromycin in both the macrophage model (Fig. 1) and in the mouse (Table 6), and activity in the latter was confirmed to be significant and dose responsive (Fig. 2). The differences in activity of RU66252 in the two in vivo experiments is likely related to an apparent difference in the growth kinetics, since the numbers of CFU in untreated postinfection tissue were similar (data not shown) but day 10 and day 31 CFU were approximately 1 log₁₀ lower in the preliminary experiment. Notably, mice tolerated dosages 5 to 10 times higher than those effecting significant growth inhibition. Since there was a general correlation between activities in the macrophage and mouse models, it would also be of interest to evaluate A323348 in the mouse model. This was not done in the present study due to lack of sufficient compound.

Overall, the findings of this study suggest directions for the rational design of additional macrolides and ketolides that are active against the tubercle bacillus at physiologically relevant concentrations. More comprehensive SAR studies, currently being conducted in our institute, should greatly facilitate the development of a highly efficacious anti-TB agent from this class.