Targeting protein biotinylation enhances tuberculosis chemotherapy

Abstract

Successful tuberculosis (TB) chemotherapy depends upon the unique contributions of its component drugs. Drug resistance poses a threat to the efficacy of individual agents and the regimens to which they contribute. Biologically and chemically validated targets capable of replacing individual components of current TB chemotherapy thus represent a major unmet need in TB drug development. We demonstrate that chemical inhibition of biotin protein ligase (BPL) can kill Mycobacterium tuberculosis (Mtb) and genetic silencing eliminates the pathogen efficiently from mice during acute and chronic infection. Partial chemical inactivation of BPL increases potency of two first-line drugs, rifampicin (RIF) and ethambutol (EMB), and genetic interference with protein biotinylation accelerates clearance of Mtb from lungs and spleens by RIF. These studies validate BPL as a vulnerable target that can serve as an alternate frontline target against Mtb.

Summary

One Sentence Summary

Biotin protein ligase (BPL) and protein biotinylation (PB) are druggable targets whose inhibition synergizes with rifampicin and has the potential to shorten tuberculosis chemotherapy.

Introduction

Tuberculosis (TB) is reemerging as an incurable infection due to drug-resistance. In 2013, approximately 480,000 people developed multidrug resistant tuberculosis (MDR-TB) (1). The discovery of an effective vaccine that prevents TB in adults remains an important goal but has been elusive (2). Consequently, our ability to control TB depends primarily on the development of more efficient chemotherapies for drug sensitive (DS) and drug resistant (DR) TB.

The cell envelope of mycobacteria is a selective permeability barrier containing several unique lipids which cause drug resistance and protect Mtb from the host immune system (3–7). The importance of mycobacterial lipid metabolism is underscored by the finding that over 250 genes are involved in lipid metabolism in Mtb as opposed to only 50 in Escherichia coli. The structurally diverse mycobacterial lipids are all derived from simple malonyl coenzyme A (CoA) building blocks, which are in turn prepared by acyl-CoA carboxylases (ACCs). Mtb encodes for three multimeric ACCs assembled from at least 10 different subunits (AccA1–3, AccD1–6, and AccE5), which together provide the malonyl-CoA, (methyl)malonyl CoA, and (long-chained alkyl)malonyl CoA building blocks required for synthesis of linear fatty acids, methyl-branched lipids, and mycolates, respectively (8–11). Each ACC must be post-translationally modified with the cofactor biotin (vitamin H) to become active. Blocking de novo biotin biosynthesis or biotin-ACC ligation, thus, has the potential to inhibit all lipid biosynthesis.

We previously reported the design and characterization of potent inhibitors of biotin protein ligase (BPL), the enzyme responsible for covalently ligating biotin onto the ACCs (12–19). This led to 5′-[N-(D-biotinoyl)sulfamoyl]amino-5′-deoxyadenosine (Bio-AMS), a BPL inhibitor with potent on-target whole-cell activity against drug-sensitive (DS) and drug-resistant (DR) Mtb. Nevertheless, several important questions remained concerning its mechanism of action, resistance, pharmacokinetic properties, synergy with other antitubercular drugs, and, importantly, consequences of BPL inhibition in vivo.

In this study, we first characterized the consequences of chemically inactivating BPL for viability of Mtb under different physiological conditions. Next, we determined the frequency and mechanism of resistance to BPL inactivation by Bio-AMS and used pharmacokinetic studies to characterize Bio-AMS metabolism in the host. Incomplete inhibition of BPL by fluctuating concentrations of Bio-AMS prevented growth of Mtb in a hollow fiber system. Furthermore, near-complete genetic inactivation of BPL killed Mtb during acute as well as chronic mouse infections. Finally, we established that partial genetic interference with protein biotinylation is sufficient to increase sensitivity of Mtb to killing by RIF during infection.

Results

Impact of chemical BPL inactivation on viability of Mtb in vitro and during macrophage infections

We first determined whether inhibition of BPL is sufficient to kill Mtb. INH, a first-line drug that inhibits the synthesis of mycolic acids (20), was used as a control. In biotin-free media, the BPL-inhibitor Bio-AMS was bactericidal at a concentration of approximately 5-fold the minimal inhibitory concentration (MIC) and killed Mtb with kinetics similar to that of INH (Fig. 1A). After 10 days of drug exposure, INH resistant mutants appeared whereas Bio-AMS continued to reduce colony-forming units (CFU) until viability of the culture was below the limit of detection of 25 CFU/mL.

Fig. 1. Chemical inactivation of BPL kills Mtb and prevents growth in macrophages.

: (A) Impact of Bio-AMS on viability of Mtb in standard liquid culture. The dashed line indicates the limit of detection. (B) Impact of Bio-AMS on growth of Mtb with different carbon sources. Relative growth was calculated by dividing OD580 of the culture with Bio-AMS by the OD580 of the culture without Bio-AMS. (C) Impact of BioAMS on intracellular Mtb. Mouse bone marrow derived macrophages were infected and Bio-AMS or DMSO were added 24 h post infection. Data are representative of two independent experiments and averages 3 ± SEM (** P<0.01, Student’s t-test)

The anti-mycobacterial activity of several small molecules depends strictly on the primary carbon source used to cultivate Mtb (21). However, changes in the primary carbon source had little impact on the MIC (Fig. 1B) or the minimal bactericidal concentration (MBC, Fig. S1A) of Bio-AMS. Addition of biotin to the growth medium increased the MIC only 4-fold (Fig. S1B), but when we analyzed Bio-AMS with Mtb that had ceased to replicate due to starvation in phosphate-buffered saline (PBS) we observed no reduction in viability (Fig. S1C). In macrophages Bio-AMS inhibited growth of Mtb in a concentration dependent manner (Fig. 1C) and a tetrazolium reduction assay did not detect toxicity for macrophages (Fig. S1D). Bio-AMS also showed no signs of mitochondrial toxicity (Fig. S2).

Thus, chemical inactivation of BPL kills growing Mtb independent of the carbon source the bacteria are consuming and prevents the growth of intracellular Mtb, but is inactive against PBS-starved Mtb.

Resistance to chemical BPL inactivation

Bio-AMS resistant Mtb was isolated with a frequency of ~1 in 10⁷ CFU when Bio-AMS was present at a concentration of 10× the MIC (Fig. 2A). This frequency decreased to less than 1 in 10⁸ at 25× the MIC and we did not isolate any resistant Mtb from 10⁸ CFU at 50× the MIC of Bio-AMS. For INH we observed a resistance frequency of approximately 1 in 2.5 ´ 10⁶ CFU, which did not vary much with drug concentration. The Bio-AMS resistant clones expressed wild type amounts of BPL (Fig. S3A) and were free of mutations in birA, the gene encoding BPL. Whole-genome sequencing revealed all resistant isolates to contain mutations in rv3405c, most of which were predicted to inactivate rv3405c, and three strains harbored mutations only in rv3405c (Table S1). Rv3405c is a transcriptional repressor that controls rv3406 in M. bovis BCG (22). When we compared mRNAs from Bio-AMS resistant isolates with mRNA from WT, we detected a drastic change in rv3406 (Table S2), which resulted in overexpression of Rv3406 (Fig. S3B). Mtb carrying an Rv3406 overexpression plasmid (pGMEH-Ptb38-rv3406) was 64-fold more resistant to Bio-AMS than WT (Fig. 2B). We concluded that overexpression of Rv3406 is sufficient to cause Bio-AMS resistance.

Fig. 2. Bio-AMS resistance.

(A) Frequency of spontaneous resistance with Bio-AMS at concentrations of 1×, 25× and 50× the MIC. (B) Overexpression of Rv3406 causes resistance to Bio-AMS. H37Rv either contained the vector control (pTE-mcs) or the overexpression plasmid pGMEH-Ptb38-rv3406. One spontaneously resistant isolate (Bio-AMS-5R1) was included as a control. (C) Saturation curve used to determine the kinetic parameters for Bio-AMS oxidation by Rv3406. Data were fitted by non-linear regression to the Michaelis-Menten equation (left panel). Time course for Rv3406 catalyzed formation of UV active metabolite 3 as monitored by HPLC (right panel).

Rv3406 is a dioxygenase that oxidizes 2-ethylhexyl sulfate (2-EHS, Fig. S3C) and its activity requires non-heme iron (II) and alpha-ketoglutarate (α-KG) (23). To test if Rv3406 could also oxidize Bio-AMS, we incubated Bio-AMS with recombinant Rv3406 and α-KG. This resulted in the time dependent formation of an UV-active product (Fig. 2C) with a λ_max of 254 nm and a molecular weight of 265, which are both consistent with the adenosine 5′-aldehyde 3 (Fig. S3D). This compound likely arises through oxidation of Bio-AMS at C-5′ of the nucleoside and spontaneous disproportionation of the resultant intermediate hemiaminal 2 into aldehyde 3 and N-(biotinoyl)sulfamide 4 (Fig. S3D). Rv3406 thus oxidizes Bio-AMS in the same manner it oxidizes an alkyl sulfate. This mechanism for enzymatic reactions of alkyl sulfates catalyzed by α-KG dependent dioxygenases is well established (23) and the identity of metabolite 4 was confirmed with an authentic standard (Fig. S3E). Steady-state kinetic analysis revealed that Bio-AMS is a poor substrate of Rv3406 and has a 48-fold higher K_M value and nearly 1,300-fold lower k_cat value than 2-EHS (Table S3). Nevertheless, degradation of Bio-AMS was dependent on the enzymatic activity of Rv3406 as degradation only occurred in the presence of α-KG (Fig. 2C). Furthermore, we found that the amount of Bio-AMS in Mtb was inversely correlated with expression of Rv3406 (Fig. S4A). Consistent with the in vitro data, we also measured higher amounts of N-(biotinoyl)sulfamide 4 when Rv3406 expression was increased (Fig. S4B). Collectively, these results demonstrated that the most frequent mechanism of spontaneous resistance to chemical inactivation of BPL is enzymatic cleavage of Bio-AMS.

BPL as a target to control Mtb infections

In mice, Bio-AMS was rapidly eliminated, in part due to hydrolysis at the acyl-sulfamide linkage, (Fig. S5, S6 and Table S4). In addition, the dose of Bio-AMS required to achieve concentrations above the MIC for a substantial fraction of the dosing interval was not tolerated by mice (Table S5). We therefore employed (i) a hollow fiber bioreactor system (HFS) to evaluate the impact of fluctuating Bio-AMS concentrations on growth of Mtb and (ii) a genetic approach to determine the consequences of inactivating Mtb’s BPL during infections.

In the HFS, medium is pumped from a central reservoir (CR) through a cylindrical bioreactor filled with tubular, semi-permeable membrane fibers. By manipulating the flow rate through the system, one can increase or decrease the drug concentrations to which the bacteria are exposed, creating defined, dynamic pharmacokinetic (PK) drug profiles (24). The fiber pore size (0.4 μM) ensures that the bacteria introduced into the bioreactor cartridge are retained in the extracapillary space (ECS). A log phase culture of Mtb H37Ra was introduced into an HF cartridge, allowed to grow in 7H9 for two days and then challenged with Bio-AMS. The Cmax was chosen to maintain concentrations above MIC90 for most of the dosing interval. The clearance flow rate was set to achieve a half-life for Bio-AMS of between 9 and 10 hours, which is a typical half-life of several antibiotics in clinical use (25, 26). The PK profiles of Bio-AMS were measured in both the CR and the ECS. Profiles taken on days 0 and 14 showed Cmax values of ~32 μM in the CR and ~17 μM in the ECS (Fig. 3A). Viability of Mtb H37Ra was monitored by CFU enumeration, which showed that Bio-AMS prevented growth in the HFS (Fig. 3B). The culture remained free of Mtb H37Ra resistant to Bio-AMS for 15 days, but resistant mutants emerged at around day 18. The appearance of Bio-AMS resistant Mtb H37Ra suggests that the PK profile analyzed here may lead to a higher frequency of resistance than observed in selections on agar plates. In the future, it will thus be important to define how PK parameters influence the frequency and type of mutations leading to resistance. Nevertheless, using Bio-AMS at a peak concentration and half-life similar to those of other clinically used cell wall-targeting antibiotics, the HFS simulations revealed that incomplete chemical inhibition of BPL achieves static growth inhibition across the dosing interval.

Fig. 3. Activity of Bio-AMS in a HFS.

(A) Pharmacokinetic profile of Bio-AMS. Bio-AMS concentrations were measured in samples taken from the central reservoir (CR) and the extracapillary space (ECS). The dotted line represents the MIC90 of Bio-AMS for H37Ra. (B) Impact of Bio-AMS on viability of H37Ra. The dotted line indicates the lower limit of detection.

Next, we used a previously described dual-control switch (27, 28) to construct an Mtb mutant, BPL-DUC, in which anhydrotetracycline (atc) induces both transcriptional silencing of birA and proteolytic inactivation of BPL (Fig. S7). Exposure of BPL-DUC to atc (i) depleted BPL below the limit of detection of immunoblots within 24 h (Fig. S8A), (ii) caused a decrease in biotinylated proteins (Fig. S8B), (iii) prevented growth (Fig. 4A), and (iv) decreased viability of Mtb (Fig. 4B). C57BL/6 mice were infected with BPL-DUC and fed the mice doxycycline (doxy) containing food either from the beginning of the infection, after 14 days, after 35 days or not at all (Fig. 4C, d). Mice that were infected with H37Rv and received doxy-supplemented food served as a control. As early as 14 days post infection, we failed to recover CFU for BPL-DUC from the lungs of mice that received doxy throughout the infection (Fig. 4C). Similarly, when doxy food was given during the acute infection (i.e. on day 14), we also did not recover any CFU two weeks later. When BPL was inactivated during the chronic phase by switching to doxy food on day 35 we observed a decrease in CFU of more than 2 log₁₀ by day 56 and lungs were free of CFU by the next time point (day 112). On day 169 post-infection, we did not recover CFU from any of the groups that received doxy-supplemented food (Fig. 4C). Decreases in CFU that occurred after doxy treatment were accompanied by reduced lung tissue pathology (Fig. S9). Depleting BPL had a similar impact on CFU in spleens as it had in lungs (Fig. 4D). We conclude that depletion of BPL kills replicating Mtb in vitro and efficiently eliminates viable Mtb from mice during acute and chronic infections.

Fig. 4. Consequences of depleting BPL.

Impact of BPL depletion on growth (A) and survival (B) in liquid media, mouse lungs (C), and spleens (D). Data are representative of three (A, B) or two (C, D) two independent experiments and averages of three (A, B) or four (C, D) samples ± SEM.

Importance of Mtb protein biotinylation for the activity of rifampicin, isoniazid and ethambutol

The cell envelope of mycobacteria is a highly selective barrier that contributes greatly to the intrinsic drug resistance of Mtb (3, 4). The integrity of this envelope requires fatty acids and lipids, which depend on biotinylated ACC enzymes for synthesis. We therefore analyzed how inhibition of protein biotinylation affects Mtb’s acid fastness and susceptibility to RIF, INH and ethambutol (EMB). First, we treated Mtb with a lethal dose of Bio-AMS and analyzed the cell envelope using acid-fast staining before and after the amounts of biotinylated proteins were reduced (Fig. S10). This detected an increase in acid-fast negative bacteria on day 5 post exposure (when biotinylated proteins just began to decrease) and after 10 days of exposure to Bio-AMS most bacteria stained as acid-fast negative. Next, we measured if a sublethal dose of Bio-AMS changes susceptibility of Mtb to RIF, INH, and EMB. As it takes several days before exposure to Bio-AMS affects Mtb’s protein biotinylation pattern, we first grew Mtb with and without Bio-AMS for three days and then exposed these cultures to RIF, INH, or EMB – in each case, with or without Bio-AMS. Sublethal doses of Bio-AMS reduced the MIC of RIF and EMB but left the MIC of INH unchanged (Fig. 5ABC). We then determined if Bio-AMS could also enhance the cidal activity of RIF and EMB. For these experiments, we used Bio-AMS at 1 μM, RIF at 6.5 nM, and EMB at 12.5 μM. As expected, none of the compounds were cidal at these concentrations when used individually (Fig. 5DE). However, the combined use of either rifampicin or ethambutol with Bio-AMS killed more than 99.7 and 99.9% of the inoculum after 20 days of treatment.

Fig. 5. Bio-AMS enhances the activity of rifampicin and ethambutol.

Impact of Bio-AMS on the activities of RIF (A, D), EMB (B, E) and INH (C). Mtb was grown with Bio-AMS (1 μM in DMSO) or DMSO only for 3 days after which the bacteria were exposed to the other drugs. Data are representative of two independent experiments and averages of three ± SEM.

We next asked if the increased potency of RIF and EMB caused by Bio-AMS could be replicated by limiting access to biotin. We grew the biotin auxotroph Mtb ΔbioA in media with defined biotin concentrations and measured the MIC of EMB and RIF. This showed that decreasing concentrations of biotin decreased the MIC of RIF but left the MIC of EMB unchanged (Fig. S11AB). When we compared the amount of RIF that accumulated in WT and Mtb ΔbioA we found increased amounts of RIF in Mtb ΔbioA when the mutant was grown with low amounts of biotin (Fig. S11C). Thus, Bio-AMS improves the potency of RIF by interfering with synthesis of a normal cell envelope, as indicated by the reduction in acid fastness, which in turn facilitates RIF uptake. In addition, Bio-AMS decreases the MIC of EMB by a yet to be determined mechanism.

Interfering with protein biotinylation enhances the activity of RIF during Mtb infection

Together with INH, RIF forms one of the cornerstones for treatment of drug-sensitive TB. We therefore sought to determine if sublethal interference with Mtb protein biotinylation could also improve the potency of RIF during infections. It would have been practically impossible to answer this question using BPL-DUC because this mutant is cleared rapidly from mice at the standard dose of doxy and the dose-response curve of the DUC switch is so steep (28) that it is infeasible to identify a dose of doxy that leads to intermediate silencing in animals. In previous work, however, we had constructed a panel of bioA-TetON mutants, in which transfer from media containing atc or doxy to atc/doxy-free media silences expression of BioA to different degrees (29). One of these mutants, bioA TetON-1, expresses ~1,000% BioA with atc and ~5% BioA without atc compared to WT, but grows almost like WT without supplemental biotin (29). We therefore measured RIF sensitivity of bioA TetON-1 and found the mutant to be more sensitive than WT to RIF without atc and more resistant than WT with atc (Fig. 6A).

Fig. 6. Partial inhibition of biotin synthesis increases susceptibility of Mtb to RIF.

Susceptibility of bioA TetON-1 to RIF in vitro (A) and during mouse infection (B, C). Data in B and C are means from eight mice ± SEM (* P<0.05, Mann Whitney test).

Next, we infected C57BL/6 mice with bioA TetON-1 by aerosol and divided the mice into two groups, one of which received doxy-supplemented food while the other was fed doxy-free food. Twenty-one days post infection, RIF was administered by oral gavage 5 days a week at a dose of 10 mg/kg for either 4 weeks or 8 weeks. At the end of RIF treatment, the mice were kept without RIF for 3 days and then CFU were determined in lungs and spleens. In agreement with our previous experiments, there was no difference in bacterial titer obtained from groups on doxy or regular chow diet in the absence of RIF, which demonstrates that the strain can sustain infection in mice even after expression of BioA is reduced (Fig. 6BC). Both treatment groups showed progressive reduction of CFU after rifampicin treatment as compared to the controls at 4 weeks and 8 weeks. However, significantly fewer CFU (p <0.05) were obtained from the lungs of mice when BioA expression was reduced due to the lack of doxy than when BioA was expressed constitutively (Fig. 6B). Similar results were obtained for the spleens of these mice (Fig. 6C). Importantly, without doxy, treatment with RIF alone was sufficient to kill Mtb bioA TetON-1 in lungs by more than 200-fold within 4 weeks. This efficacy is similar to that achieved by combining RIF with INH (30, 31).

To test if doxy could have indirectly changed exposure of RIF, we performed a drug-drug interaction study in which RIF was administered with and without doxy to uninfected mice, and measured both doxy and RIF plasma concentrations following a single RIF dose and at steady state. Seven daily doses of RIF caused a small but statistically significant decrease (~30%, p<0.05) of doxy exposure over the 8 h period during which plasma concentrations were measured. However, RIF exposure was not affected by co-administration of doxy (Table S6). In addition, drug distribution studies with doxy administered in the diet to Mtb-infected rabbits showed that doxy accumulates in lung lesions relative to plasma, with lesion/plasma ratios ranging from approximately 4 to 6 (Table S7). Accumulation into tissue was independent of the doxy concentration in the diet (200 or 400 ppm). Thus doxy concentrations in lung lesions are at least as high as those measured in plasma.

Taken together, these experiments demonstrate that interfering with Mtb’s ability to biotinylate proteins increases sensitivity to RIF and strongly suggest that Bio-AMS or other inhibitors targeting biotin metabolism will be excellent drug candidates that can be expected to synergize with RIF in vivo.

Discussion

Tuberculosis remains a major killer in the developing world. The stop TB partnership envisions a TB free world by 2050 (1). Continued improvement of TB chemotherapy including the development of new drugs will be required to achieve this goal. New TB drugs should ideally be active against DS and DR Mtb and shorten the treatment of DR and DS TB. Mtb BPL was shown to be susceptible to chemical inhibition by Bio-AMS, which can prevent growth of DS and DR Mtb (17). This suggested Mtb BPL as a potentially attractive target for TB drug development, but its importance for bacterial viability under different physiological conditions and during infections remained to be determined. It also remained unclear how inhibition of BPL impacts the potency of current first line drugs and how it might affect the length of treatment. By answering these questions, we sought to further evaluate the potential of Bio-AMS as a lead and BPL as a target for TB drug development.

We found chemical inhibition of BPL to (i) be cidal for Mtb during growth with different carbon sources (Fig. 1AB, Fig. S1A), (ii) prevent growth in macrophages (Fig. 1C, Fig. S1C), (iii) have a low frequency of low-magnitude resistance and an undetectable frequency of high-magnitude resistance (Fig. 2A), (iv) prevent growth of Mtb in HFS simulating fluctuating Bio-AMS concentrations typically observed in in vivo systems (Fig. 3), and (v) synergize with other TB drugs (Fig. 5AB). To determine the consequences of BPL inhibition during infections we constructed Mtb BPL-DUC (Fig. S7), which enabled rapid depletion of BPL by means of a genetic approach (Fig. S8A). Characterization of this mutant confirmed our conclusions from chemical BPL inactivation experiments and revealed that depletion of BPL is sufficient to rapidly kill Mtb during acute and chronic infection in mice (Fig. 4CD, Fig. S9). Interestingly, depletion of BPL eradicated Mtb not only during acute infections but also if it was initiated on day 35 after the bacteria had already established a chronic infection. This is in contrast to both the lack of Bio-AMS activity against nonreplicating Mtb in vitro and the potency of INH, which is higher if administered during the acute phase (e.g. beginning on day 3 post infection) rather than during the chronic phase (e.g. beginning on day 28 post infection) (32). That BPL inactivation efficiently kills Mtb during the chronic phase of mouse infection is consistent with data demonstrating that Mtb is still replicating during this stage of the infection even though the CFU count remains stable (33). Moreover, it suggests that BPL might also be required for ACCases to synthesize the building blocks for the cell wall remodeling that seems to occur during chronic infection (34).

One main limitation of the work that characterized Mtb BPL so far is that it did not yet lead to a drug-like inhibitor with sufficient bioavailability. It thus seems worth noting that susceptibility to cleavage at the biotin-adenosine linker both contributes to its limited bioavailability and is the primary mechanism of spontaneous resistance to Bio-AMS. Ongoing experiments that aim to improve bioavailability of Bio-AMS therefore should focus on modifications that enhance the stability of the acyl-sulfamide linker region of the molecule connecting the biotin and nucleoside moieties or analogues wherein the acyl-sulfamide is replaced with a suitable bioisostere. Using this approach, it may be possible to not only improve bioavailability, but to also further decrease the frequency of resistance by overcoming Rv3406-mediated destruction of Bio-AMS. The enzyme(s) that cleaves Bio-AMS in mice also remains unknown; however, we were able to identify the alkyl sulfatase Rv3406 as the enzyme responsible for Bio-AMS degradation in Mtb. Interestingly, Cole and co-workers have shown that an unrelated class of 2-carboxyquinoxalines that target DprE1 are also inactivated by Rv3406, but via an alternative mechanism whereby the compounds mimic the substrate α–KG (35). This suggests that Rv3406 enzyme may have a more general role in xenobiotic metabolism.

Successful therapy of TB depends on drug combinations and new drugs should ideally synergize with the existing frontline drugs. Partial inhibition of BPL by sublethal concentrations of Bio-AMS increased the potency of RIF and EMB, revealing a very attractive feature of BPL as a target for TB drug development (Fig. 5). Limiting Mtb’s access to biotin also allowed us to determine how reducing Mtb protein biotinylation affects the potency of RIF during infections. This showed that partial inhibition of Mtb protein biotinylation, which by itself did not reduce growth of Mtb in lungs, accelerated killing of Mtb by RIF in mice (Fig. 6BC). The magnitude of this effect is noteworthy. The potency we observed for treatment of an Mtb mutant with impaired protein biotinylation by RIF alone is similar to the potency reported for treatment of WT Mtb by the combination of RIF with INH (30, 31).

RIF and INH are two of the most important drugs for treatment of TB. INH has the most potent bactericidal activity during the early treatment and RIF is most effective in preventing relapse (36–38). This importance of the INH and RIF combination is evidenced by the fact that TB caused by Mtb that is resistant to INH and RIF is classified as multidrug-resistant (MDR), irrespective of resistance to other drugs. Inactivation of BPL resembles inhibition of InhA, the target of INH, in that both interfere with cell envelope biosynthesis and kill growing Mtb rapidly both in vitro and during infections. Potentially, drugs targeting Mtb BPL might be as effective as INH for treating TB and could help to further shorten TB chemotherapy by improving the potency of RIF.

Materials and Methods

Study design:

The overall objective of this study was first to provide a more thorough evaluation of the previously described BPL inhibitor Bio-AMS and then to further validate Mtb’s BPL as a target for TB drug development. Bio-AMS was characterized with respect to its activity under a variety of growth conditions, primary mechanism of resistance, pharmacokinetic properties, and interaction with existing TB drugs. Genetic approaches were applied to evaluate BPL as a target for drug development by determining the consequences of depleting BPL in vitro and during infection. Animals were randomly allocated into groups and identifiable with respect to their treatment during the experiments. All studies were carried out in accordance with the guide for the fair and ethical use of Laboratory Animals of the National Institutes of Health, with approval from the Institutional Animal Care and Use Committee (IACUC) of the New Jersey Medical School, Rutgers University, Newark or the IACUC of Weill Cornell Medical College. Animals were maintained under pathogen-free conditions and fed water and chow ad libitum; all efforts were made to minimize suffering or discomfort. All experiments with Mtb were carried out in biosafety level 3 facility and approved by the relevant institutional biosafety committees.

Materials and Reagents:

Middlebrook’s 7H9 medium, 7H10 medium and Middlebrook OADC (oleic albumin dextrose catalase) growth supplement were from Difco. Hygromycin (Thermofisher Scientific) Kanamycin (Sigma) and Zeocin (Invitrogen) were used at a concentration of 50 μg/mL, 25 μg/mL and 25 μg/mL respectively. Anti-BPL antiserum was generated using purified BPL protein by Covance and used at a dilution of 1:2,500. Anti-rv3406 antiserum was a kind gift from Dr. Leila de Mendonça Lima and used at a dilution of 1:1,000. Strains are listed in Table S8.

Antibacterial activity measurements:

MIC measurements were performed as described (18); CFUs were used as a readout to assess bactericidal activities in liquid culture and during infections of bone marrow derived macrophages. Carbon sources were used at a concentration of 0.1%. PBS starvation was used as a model for nonreplicating bacteria as described previously (28).

Bio-AMS resistance:

Approximately 10⁸ bacteria were cultured on 7H10 agar plates containing drug at a concentration of 10×, 25× or 50× the MIC. Frequency of resistance was calculated as number of CFUs/10⁸ bacteria plated. mRNA analyses of Bio-AMS resistant strains were performed as described in our previous work (39).

Mouse infection with Mtb BPL-DUC and bioA TetON-1:

4–6-week-old female C57BL/6 mice were infected with ~100 CFUs of BPL-DUC strain and divided into various groups. The control group was maintained on a regular diet whereas test groups received 2000 ppm doxy rodent chow (Research Diets, St. Louis, MO) when indicated. At each time point 4 mice per group were sacrificed, lung and spleen were homogenized in PBS, and dilutions were cultured on antibiotic-free agar plates. C57BL/6 mice were infected with bioA TetON-1 as for BPL-DUC. The control group received doxy throughout the experiment; the test group received the regular chow. RIF was administered at a dosage of 10 mg/kg by oral gavage 5 days a week. 8 mice from each group were sacrificed 4 and 8 weeks after RIF treatment and organs processed as described above.

Pharmacokinetic profiling:

Groups of 4 mice received Bio-AMS formulated in 0.9% saline according to the following dosing scheme: 5 mg/kg via the intravenous (IV) route, 25 mg/kg via the intraperitoneal (IP) route and 25 mg/kg via the oral (PO) route. Blood samples were collected in heparinized tubes, pre-dose and 5 min, 15 min, 30 min, 1 h, 1.5 h, 3 h, 5 h, and 8 h post-dose following IV and IP injections, and pre-dose, 5 min, 30 min, 1 h, 3 h, 5 h, and 8 h following oral gavage. Blood samples were centrifuged to recover plasma and quantify Bio-AMS and its major metabolites by LC/MS-MS as described in the supplementary materials.

Tolerability:

Groups of 3 mice received either 50 mg/kg, 100 mg/kg, 250 mg/kg or 500 mg/kg of Bio-AMS via intraperitoneal injection at 4 mL/kg. They were observed continuously for the first half hour post injection, then at 24 h post-injection.

Biochemical characterization of Rv3406:

His-tagged Rv3406 was overexpressed in BL21 (DE3) E. coli, purified as described (23). Enzymatic activities were established as described in the supplementary materials.

Intra-bacterial pharmacokinetics:

A previously described (28) experimental set up was used to study accumulation of Bio-AMS and its degradation products inside Mtb. Briefly, Mtb laden filters were grown in Middlebrook 7H10 agar plates for 5 days followed by exposure to Bio-AMS for 18 h in GAST medium containing 25 μM Bio-AMS or an equivalent amount of DMSO. After 18 h the filters were incubated on GAST medium for 24 h without any antibiotics and samples were collected and processed as described in the supplementary materials.

Activity of Bio-AMS in a hollow fiber bioreactor system (HFS):

Mtb H37Ra (ATCC 25177) was grown in Middlebrook 7H9 medium supplemented with OADC for 4 days at 37°C. High flux polysulfone HF cartridges (C2011, FiberCell Systems Inc., New Market, MD 21774) were equilibrated for 3 days with 7H9 OADC. Initial PK data was obtained for Bio-AMS diluted in 7H9 OADC and infused by syringe pump into the HFS. A clearance flow rate was used that simulated a Bio-AMS half-life of 9 to 10 hours in the ECS. To determine the susceptibility of H37Ra to Bio-AMS under defined PK parameters, a new HF cartridge pre-equilibrated with 7H9 OADC was inoculated with 20 mL of a bacterial suspension of 10⁴ to 10⁶ CFU/mL. The culture was allowed to establish in the cartridge ECS (extra-capillary space) under a constant flow of 7H9 OADC (30 mL/h) for 2 days before the first Bio-AMS infusion. To achieve a Cmax of at least 9 μM in the ECS of the HF bioreactor, 1.42 mg Bio-AMS was infused for 60 min at a flow rate of 30 mL/h. This infusion was repeated on an almost daily basis for 18 days. Bio-AMS concentrations in HF samples were quantified by mass spectrometry as detailed below (see “Quantitation of Bio-AMS and biotin in mouse plasma” ) except that only 1 μL volumes of extracted samples were injected. PK analysis of drug concentrations attained in both CR (central reservoir) and ECS compartments was performed on days 0 and 14.

Statistical analysis:

Averages were used as a measure of central tendency. Data from continuous variables were analyzed using Mann-Whitney and Student’s t tests. Differences with P less or equivalent to 0.05 were considered statistically significant. All statistical analyses were performed using GraphPad Prism 7.0 (GraphPad Software Inc.)