Exploration of Synergistic Action of Cell Wall-Degrading Enzymes against Mycobacterium tuberculosis

ABSTRACT

The major global health threat tuberculosis is caused by Mycobacterium tuberculosis. M. tuberculosis has a complex cell envelope—a partially covalently linked composite of polysaccharides, peptidoglycan, and lipids, including a mycolic acid layer—which conveys pathogenicity but also protects against antibiotics. Given previous successes in treating Gram-positive and -negative infections with cell wall-degrading enzymes, we investigated such an approach for M. tuberculosis. In this study, we aimed to (i) develop an M. tuberculosis microtiter growth inhibition assay that allows undisturbed cell envelope formation to overcome the invalidation of results by typical clumped M. tuberculosis growth in surfactant-free assays, (ii) explore anti-M. tuberculosis potency of cell wall layer-degrading enzymes, and (iii) investigate the concerted action of several such enzymes. We inserted a bacterial luciferase operon in an auxotrophic M. tuberculosis strain to develop a microtiter assay that allows proper evaluation of cell wall-degrading anti-M. tuberculosis enzymes. We assessed growth inhibition by enzymes (recombinant mycobacteriophage mycolic acid esterase [LysB], fungal α-amylase, and human and chicken egg white lysozymes) and combinations thereof in the presence or absence of biopharmaceutically acceptable surfactant. Our biosafety level 2 assay identified both LysB and lysozymes as potent M. tuberculosis inhibitors but only in the presence of surfactant. Moreover, the most potent disruption of the mycolic acid hydrophobic barrier was obtained by the highly synergistic combination of LysB, α-amylase, and polysorbate 80. Synergistically acting cell wall-degrading enzymes are potently inhibiting M. tuberculosis, which sets the scene for the design of specifically tailored antimycobacterial (fusion) enzymes. Airway delivery of protein therapeutics has already been established and should be studied in animal models for active TB.

INTRODUCTION

Pathogens of the genus Mycobacterium are the etiological agents of various severe diseases, the most notorious being tuberculosis (TB) caused by Mycobacterium tuberculosis. It is the most lethal infectious disease worldwide, and while it might be surpassed by COVID-19 in 2020 and 2021, it still poses a major global health threat, and, worryingly, multidrug resistance is on the rise. Treatment success of multidrug-resistant tuberculosis (MDR-TB) is currently only 57%, and the very protracted treatment comes with harsh side effects (1). Recent advances in drug development have drastically shortened the treatment of MDR-TB and provided more tolerable oral regimens composed of established TB drugs, novel compounds, and repurposed drugs (1–5). Still, MDR-TB treatment typically takes 9 to 20 months, and the search for more effective and quicker-acting drugs is ongoing. The main challenges are the slow growth of M. tuberculosis and the relatively impermeable, complex layered cell wall structure of mycobacteria (Fig. 1) in which the cell membrane is covered by a layer of peptidoglycan (PG) covalently linked to arabinogalactan. Mycolic acids are esterified to the AG and, together with free, intercalating glyco- and phospholipids and fatty acids, they constitute a near-impermeable outer “mycomembrane” (6–9). This mycomembrane is surrounded by a capsule comprised mainly of α-glucan (10, 11).

FIG 1.

Exploring the synergistic action of cell wall-degrading enzymes to permeabilize the Mycobacterium tuberculosis cell wall. The strategy is that the triple-auxotrophic M. tuberculosis mc²7902 strain is transformed with a bacterial Lux operon (33) to allow for the use of a bioluminescent assay to test drug sensitivity. Putative cell wall-degrading agents are added alone or in combination in order to destabilize the structural layers of the cell envelope, leading to permeabilization. Polysorbate 80 is a nonionic surfactant; α-amylase is an α-glucan-hydrolyzing enzyme; LysB is a mycolylarabinogalactan esterase, and lysozyme is a peptidoglycan-hydrolyzing enzyme. Intercalating glyco- and lipoproteins in the cell wall layers are not shown to maintain clarity.

Bacteriophage PG-degrading enzymes (endolysins) are increasingly investigated as antibacterial agents, with several products for topical use against Gram-positive infections already on the market (12; ClinicalTrials.gov registration no. NCT03163446) and progress being made for Gram-negative pathogens (13–15). Such products are rapidly gaining attention due to the looming antibiotics resistance crisis, together with improved know-how in biopharmaceutical protein production and formulation for nebulization or dry powder inhalation. In the context of respiratory diseases such as TB, it is encouraging that several recombinant biopharmaceutical protein treatments (e.g., dornase alpha to reduce viscosity of airway mucus) have been successfully developed for inhalation (16). However, due to their distinctive cell wall, bacteriophage-mediated lysis of mycobacteria is more complicated than that of other Gram positives. In addition to endolysin, mycobacteriophages employ a mycomembrane-targeting mycolylarabinogalactan esterase (LysB) to lyse mycobacteria (17–19).

Whereas endolysin derived from mycobacteriophage Ms6 has been reported not to inhibit Mycobacterium smegmatis or M. tuberculosis growth when added therapeutically (20), the smaller hen egg white lysozyme (which also cleaves PG) has long been shown to weakly inhibit both M. smegmatis and M. tuberculosis (21–24). For M. smegmatis, a growth-inhibitory effect of LysB enzymes has been described in the presence of surfactant or membrane-destabilizing cationic peptides (17–19, 25). Hen egg white lysozyme, RipA (an M. tuberculosis PG-endopeptidase), RpfE (an M. tuberculosis putative transglycosylase), or hydrolase-30 (an M. smegmatis cell wall hydrolase) had moderate inhibitory effects on the growth of M. smegmatis, but this effect was significantly increased if the enzymes were administered in combination with various antibiotics, confirming the role of the mycomembrane as a drug barrier (24). This illustrates how cell wall-weakening enzyme treatments have the potential to significantly shorten the treatment regimens of M. tuberculosis chemotherapy.

Previous studies on the use of cell wall-degrading enzymes against Mycobacteria suffer from two main limitations. First, these studies often use Mycobacterium smegmatis, a rather distant relative to M. tuberculosis, as the test organism, with known differences in cell wall composition (26, 27). This organism is noninfectious and grows very rapidly, hence its popularity. However, for cell wall-degrading antibiotics research, the pathogen should be used. Second, M. tuberculosis clumps heavily in cultures due to the composition of its cell wall. As this has precluded most high-throughput drug test formats, the surfactant polysorbate 80 is routinely added to M. tuberculosis culture medium to support finely dispersed planktonic growth. Polysorbate 80 does not restrict M. tuberculosis propagation in cultures that use glycerol as the carbon source (28, 29). Still, it is known to alter mycobacterial drug susceptibility (30, 31) and to destabilize the outer capsule (11). Obviously, this could affect the validity of results obtained with cell wall-degrading enzymes and precludes exploration of surface-active components in the enzyme cocktail. In this study, we solve both problems by developing a bioluminescent derivative of a biosafety level 2 triple-auxotrophic M. tuberculosis (32). The bacterial lux operon used does not require cellular uptake of any cofactor (33), which allows for quantifying metabolic activity in static, clumped cultures.

We set out to systematically assess the potency of enzymes that degrade the various layers of the M. tuberculosis cell wall, using cultivation conditions that keep the M. tuberculosis cell wall/biofilm intact. Apart from PG and the mycomembrane, we targeted the outer capsule of M. tuberculosis. Considering that α-glucan is a main constituent of this capsule (10), we speculated that Aspergillus oryzae α-amylase, an α-glucan-hydrolyzing enzyme already used in human medicine to treat pancreatic insufficiency, could have a capsule-destabilizing effect and potentially synergize with the known capsule destabilization imparted by polysorbate 80. Furthermore, we hypothesized that the anti-TB potency of cell wall-degrading enzyme treatments would be enhanced by using the enzymes in cocktails, which could potentially “peel” the cell envelope layer by layer and synergistically weaken it (Fig. 1). Amylase-induced hydrolysis of the capsular layer may, for instance, increase permeability for enzymes such as LysB that degrade the mycomembrane and/or for enzymes such as lysozyme that degrade the PG layer underneath.

RESULTS

Enzyme production and characterization.

Mycobacteriophage Ms6 and D29 mycolic acid esterases (LysB-Ms6 and LysB-D29) were produced in Escherichia coli and purified via immobilized-metal affinity chromatography (IMAC) and size exclusion chromatography (SEC) to yields of 3.12 and 1.34 mg/liter E. coli culture (i.e., 81 and 44 nmol/l), respectively (Fig. 2a and b). Minor low-molecular-weight bands were observed in purified Lys-Ms6, probably indicating low-level protein degradation. Enzymatic activity on a pNPB substrate seemed highly dependent on the bacteriophage from which the enzyme was derived, with LysB-D29 displaying 3-fold higher specific activity than LysB-Ms6 (Table 1; not significant, P = 0.2, Mann-Whitney test). Enzymatic activity of commercially available lysozyme was validated using a Micrococcus PG degradation assay (Table 1, bottom).

FIG 2.

Production of LysB-Ms6 (a) and LysB-D29 (b). After expression in E. coli BL21(DE3), proteins were purified via nickel chromatography and size exclusion chromatography (SEC; chromatogram shown). After SEC, peak fractions containing the recombinant protein at highest purity were pooled, indicated in this figure by blue rectangles on SEC chromatograms and Coomassie-stained SDS-PAGE.

TABLE 1.

In vitro activity testing of potential antimycobacterial enzymes^a

Compound	Substrate	In vitro substrate-degrading activity (U/nmol)	n
LysB-Ms6	pNPB	338 ± 7.7	3
LysB-D29		1375 ± 173.7	2
Lysozyme (hen egg white)	Micrococcus peptidoglycan	771 ± 26.1	3
Lysozyme (human)		589 ± 13.2	3

^aLipolytic activity of LysB enzymes was determined in vitro on a pNP-butyrate substrate. One LysB enzyme unit (U) will release 1 nmol of p-nitrophenol per minute at pH 8 at 37°C. Peptidoglycan (PG)-degrading activity of human and hen egg white lysozyme was determined on a fluorescein-labeled Micrococcus lysodeikticus cell wall substrate. One lysozyme enzyme unit (U) will produce a 0.001 U per minute change in the absorbance at 450 nm at pH 6.24 and 25°C.

Cell wall-degrading enzymes inhibit mycobacterial growth.

Lux operon-induced bioluminescence has already been shown to correlate very well with (inhibition of) mycobacterial growth in drug susceptibility assays (33, 34). We therefore used this bioluminescence to reliably evaluate the antimycobacterial activity of recombinant LysB-Ms6 and LysB-D29 as well as commercially available lysozyme (human and chicken egg white) and α-amylase (A. oryzae) and antibiotics isoniazid (INH) and rifampin (RIF). Bioluminescence was determined after 1, 4, 7, 10, and 14 days of incubation. When modeling dose-response curves, inhibition often appeared incomplete at day 1 (lower asymptote not zero), and after excluding that time point, the highest percentage of explained variance was consistently obtained at day 4. Therefore, MICs were determined at this time point (Table 2 and Fig. 3b).

TABLE 2.

MIC values of compounds against M. tuberculosis mc²7902_Lux^a

Compound	MIC (± SE [μM]) in:		Fold decrease in MIC
	Assay medium	Assay medium plus 0.05% polysorbate 80
Rifampin	0.3 ± 0.06	≤0.02	≥10
Isoniazid	0.1 ± 0.03	0.2 ± 0.03	0.6
α-Amylase (A. oryzae)	≥250	≥250	NA
Lysozyme (hen egg white)	≥250	40 ± 5	≥6
Lysozyme (human)	≥250	33 ± 4	≥8
LysB-D29	≥0.4	0.08 ± 0.01	≥5
LysB-Ms6	≥0.4	0.2 ± 0.07	≥2

^aLuminescence drug susceptibility assay (n = 3 with technical duplicates). MIC was determined according to Lambert and Pearson (48). If no inhibition was observed, the MIC was defined as larger than or equal to the highest concentration tested. If inhibition was observed but no sigmoid could be fitted, the MIC was defined as lower than or equal to the lowest value for which all data points were lower than 200 RLU (corresponding to half of the inoculum). NA, not available.

FIG 3.

The effect of various antimycobacterial compounds on bioluminescence of M. tuberculosis mc²7902_Lux. (a) Scheme of the generalized logistic function as applied to a bioluminescence growth assay and its main parameters. (b) Effect of antimycobacterial agents in absence (purple circles) and presence (cyan triangles) of 0.05% polysorbate 80 (PS80). Luminescence detected after 4 days of incubation at 37°C. If data allowed, the generalized logistic model and the Gompertz model were fitted to the luminescence data (the model explaining the highest percentage of variance is shown). The MIC calculated from regression parameters according to Lambert and Pearson (48) is indicated as a dotted vertical line. If no inhibition was observed, the MIC was defined as larger than or equal to the highest concentration tested (not shown in graph). If inhibition was observed but no sigmoid curve could be validly fitted, the MIC was defined as lower than or equal to the lowest value for which all data points were lower than 200 RLU (corresponding to half of the inoculum). Data shown were derived from three independent experiments, with each data point an average of duplicate plates within a repetition of the experiment. Mean and standard deviation are shown. (c) For several conditions from the high and low ends of inhibition curves of LysB-Ms6 (circles) and human lysozyme (triangles), both in the presence of 0.05% PS80, viability was determined by CFU plating on solid medium after measuring luminescence. Data points shown are averages of technical plating replicates.

We calculated an MIC of INH in the same order of magnitude as described in Andreu et al. at day 4 (33). The MIC we calculated for RIF was approximately 10-fold higher, but it should be noted that we used a different strain of M. tuberculosis (pathogenic H37Rv in literature versus mc²7902 in our study), and our definition of MIC is more stringent than that used in literature (1-log reduction in bioluminescence).

Both human and chicken egg white lysozyme were slightly increasing mycobacterial growth at low concentrations and only marginally inhibiting growth at concentrations beyond 100 μM but became effective (no significant difference between the two, P = 0.318; t test) upon addition of polysorbate 80 as a second compound. Activities were enhanced 6- to 8-fold (Fig. 3b and Table 2), while polysorbate 80 supplementation in the absence of cell wall-degrading enzymes did not hamper mycobacterial growth (Fig. S3 in the supplemental material). This demonstrates that by omitting surfactant polysorbate 80 (PS80) from the standard susceptibility assay protocol, its modulation of the effects of antimycobacterial enzymes can be studied effectively. α-Amylase, on the other hand, did not considerably affect growth in the concentration range tested, neither in the presence nor in the absence of surfactant (MIC ≥ 250 μM). LysB-D29 and LysB-Ms6 could inhibit growth effectively in the presence of polysorbate 80 (MICs of 0.08 ± 0.01 μM and 0.2 ± 0.07 μM, respectively; no significant difference between the two by t test, P = 0.189). It should be noted that the dose-response curve shapes differ, with LysB-D29 showing a sharp dose-dependent effect and stronger polysorbate 80 dependence than LysB-Ms6.

Several conditions at the high and low ends of the LysB-Ms6 and human lysozyme inhibition curves were selected for a repeat luminescence assay followed by CFU determination, which demonstrated a strong correlation between RLU and CFU (Fig. 3c; Spearman correlation coefficient = 0.93, P = 1.6 × 10⁻⁵), thereby validating the use of the luminescence assay. Upon administration of 250 μM human lysozyme in the presence of 0.05% PS80, viability was vigorously decreased after 4 days at 37°C (10³ CFU/well versus 10⁷ CFU/well after sub-MIC treatment), and upon administration of the highest concentrations of LysB-Ms6 with 0.05% PS80, no viable M. tuberculosis was observed at all (Fig. 3c).

Synergy.

Considering the layered structure of the mycobacterial cell wall, we hypothesized that cell wall-degrading enzymes can display synergy if administered jointly. Using Combenefit (35), we analyzed various combinations of antimycobacterial compounds in two-dimensional “checkerboard” assays (Table 3) in assay medium containing either zero or 0.05% polysorbate 80 as a potential susceptibility-modulating agent. The following three tested compound combinations scoring highest for synergy were selected for in-depth analysis: LysB-Ms6 with human lysozyme and 0.05% polysorbate 80, and LysB-Ms6 with α-amylase in presence or absence of 0.05% polysorbate 80. When LysB-Ms6 was combined with either human lysozyme or A. oryzae α-amylase, both synergistic and antagonistic regions were observed in the dose-response surface with few significant combinations and an overall neutral effect (Fig. 4a and b). However, when 0.05% polysorbate 80 was added to the combination of LysB-Ms6 and A. oryzae α-amylase, the shape of the dose-response surface changed drastically, displaying a deep valley of synergy in which 18 out of 36 conditions scored significant for synergy (P < 0.05 to P < 0.0001; Fig. 4c).

TABLE 3.

Synergy and antagonism of antimycobacterial compounds as determined by the Combenefit (50) tool^a

Compound			Max conc (μM)		IC50 (μM)		Sum antagonism score			Sum synergy score
A	B	C	A	B	A	B	Loewe	Bliss	HSA	Loewe	Bliss	HSA
α-Amylase	LysB-Ms6		125	0.2	125.00	0.13	−11.51	−10.63	−10.37	27.73	33.09	35.10
α-Amylase	LysB-Ms6	Polysorbate 80	125	0.2	0.04	0.05	0.00	0.00	0.00	117.06	100.37	117.31
Lysozyme	LysB-D29	Polysorbate 80	125	0.2	14.90	0.04	−34.14	−24.17	−18.05	0.00	1.34	2.93
Lysozyme	LysB-Ms6		125	0.2	125.00	0.04	−49.85	−48.63	−48.63	7.93	8.71	8.95
Lysozyme	LysB-Ms6	Polysorbate 80	125	0.2	18.75	0.14	−17.71	−17.88	−17.63	36.78	46.89	53.27

^aLuminescence drug susceptibility assay on M. tuberculosis mc²7902_Lux (n = 3 with technical duplicates). Compounds A (human lysozyme or Aspergillus oryzae α-amylase) and B (LysB variants) were added checkerboard-wise in combined dilution series, of which the highest concentration is indicated. If present as compound C, polysorbate 80 is at 0.05%. The half-maximal inhibitory concentration (IC₅₀) was derived from a Hill equation fit of the data. Synergy and antagonism scores were determined using either the Loewe, Bliss, or HSA model. Bold indicates three highest scoring synergy conditions.

FIG 4.

(a, b, and c) Synergy of various antimycobacterial compounds (Bliss model) mapped to the experimental dose-response curve, and the corresponding synergy and antagonism matrix. Data were derived from a checkerboard bioluminescence drug susceptibility assay of M. tuberculosis mc²7902_Lux. Bioluminescence was detected after 4 days of incubation at 37°C. Data shown were derived from 3 independent experiments, each with duplicate technical replicate plates. The synergy score shown in the matrix is the percentage of growth inhibition by the drug combination that is not explained by the Bliss-modeled reference dose response, as calculated using the Combenefit (50) tool. Results are colored according to the obtained synergy score only if the result is significant following a one-sample t test. *, P < 0.05; **, P < 0.001; ***, P < 0.0001. (d and e) Mixed model of the synergy of LysB-Ms6 with α-amylase (A. oryzae) in the presence or absence of 0.05% polysorbate 80. A linear mixed model was applied to the bioluminescence data of two sets of drug combinations from panel c to evaluate the separate and combined effects of each component (LysB-Ms6, α-amylase, or PS80) on the variance. For each set of drug combinations, the resulted predicted values are plotted as follows: the effect of α-amylase alone or in the presence of PS80 (left) and the combined effect of LysB-Ms6 and a set concentration of α-amylase in absence (middle) or presence (right) of PS80.

The significance of synergy between LysB-Ms6, α-amylase, and polysorbate 80 was further assessed by applying a linear mixed model to bioluminescence data of two sets of drug combinations (Fig. 4d and e; details in Fig. S5). In the presence of 0.05% polysorbate 80, a significant interaction (P = 0.012 in the mixed model) was identified between a dilution series of LysB-Ms6 and 7.8 μM α-amylase, confirming synergy (Fig. 4e). In the absence of polysorbate 80, no such significant interaction was observed. In the presence of 3.9 μM α-amylase (Fig. 4d), a similar trend was observed, but it was of borderline significance (P = 0.058 in the mixed model). From combined results of bioluminescence assays and synergy analyses (Fig. S6), we conclude that the combination of dilution series of LysB-Ms6 and A. oryzae α-amylase in the presence of 0.05% polysorbate 80 leads to a strong, synergistic effect on the growth of M. tuberculosis mc²7902, validating the hypothesis that targeting multiple layers of the cell wall can lead to more effective antimycobacterial activity.

DISCUSSION

The results of this study can be summarized as a methodological improvement in anti-M. tuberculosis drug screening and a main discovery with regard to the synergistic antimycobacterial effect of cell wall-degrading enzymes.

The methodological improvement pertains to the rather mundane, but the vexing problem is that M. tuberculosis grows in clumps when its cell envelope is left unperturbed. While adding surfactants in high-throughput microtiterplate-based drug studies enables spectrophotometry-based readouts of growth, it likely invalidates susceptibility tests of cell wall-targeting enzymes and drugs, as accessibility to the cell wall substrates of such enzymes will be altered, which we confirmed in our studies. Other researchers have deferred to the use of model organism M. smegmatis, which clumps less and grows more rapidly than M. tuberculosis and is noninfectious, allowing work outside highly restricted and expensive biosafety level 3 laboratories. Nevertheless, the cell wall of M. smegmatis differs significantly from that of the pathogenic slow-growing mycobacteria, and it is questionable whether results can be extrapolated to M. tuberculosis. For this reason, we resorted to the use of M. tuberculosis engineered with a bacterial lux operon. In a parallel project, other luciferase systems were also implemented, and whereas higher specific luminescence could be obtained with firefly luciferase, the bacterial lux operon does not require addition of the luciferin cofactor. This is a major advantage, as it eliminates doubts about false results caused by bacterial clumping limiting cofactor diffusion or cell wall-degrading enzyme treatment “opening up” access for such cofactor. Moreover, we combined this system with biosafety convenience by implementing it in a triple-auxotrophic M. tuberculosis derivative generated in and generously provided by the Jacobs lab (32). The luminescent auxotrophic M. tuberculosis strain and associated protocols should be useful in a great variety of drug screening studies.

Using this strain, we discovered that the combination of LysB-Ms6 and A. oryzae α-amylase was highly synergistic in inhibiting mycobacterial growth in the presence of polysorbate 80. Remarkably, α-amylase itself has no anti-M. tuberculosis effect, either with or without 0.05% polysorbate 80. The most straightforward explanation is that polysorbate 80 and α-amylase collaborate in disrupting the α-glucan capsule (nonessential for in vitro M. tuberculosis growth) (11), clearing the way for LysB to hydrolyze the linkage between underlying mycolic acids and the arabinogalactan layer, thus disrupting the mycomembrane. This finding lays the first stone for the creation of specifically tailored antimycobacterial (fusion) enzymes and also for more extensive synergy studies of both enzymes and classical antibiotics that affect cell wall integrity.

The anti-M. tuberculosis potential of LysB-Ms6 itself and its improvement in the presence of polysorbate 80 had already been shown on the M. smegmatis model, and it was demonstrated that cell wall hydrolysis was the main mechanism behind inhibitory activity of LysB-Ms6 (30). We report a similar surfactant-dependent bacteriostatic effect of both LysB-Ms6 and LysB-D29 on M. tuberculosis. Additionally, we showed that while subtherapeutic concentrations of lysozymes appeared to stimulate growth, which is potentially explained by the mycobacterial need for peptidoglycan-degrading enzymes for cell division and growth of dormant cells (36), the addition of polysorbate 80 to human or chicken egg white lysozyme abrogated this effect and inhibited growth of M. tuberculosis.

The mycobacterial cell envelope is a complex nonproteinaceous (thus not directly genetically encoded) structure synthesized by an extensive enzyme machinery (7). This likely substantially reduces the chance that single spontaneous mutations lead to resistance against enzyme antimicrobials, contrary to small-molecule antibiotics, which bind to genetically encoded protein targets. In addition, it has been observed that in treating MDR-TB patients (the main target group for biopharmaceutical enzyme treatment), low-level antibiotics resistance can be overcome by administrating drug doses exceeding the MICs of the resistant strains (37). After weakening the Mycobacterium envelope with cell wall-degrading enzymes, classic antibiotics could potentially penetrate more readily, increasing the effective drug concentration intracytoplasmatically to levels above the MIC.

While promising as antimycobacterials, adequate delivery of the enzyme therapeutics to the site of infection is essential to therapeutic efficacy. We believe our study is timely, as several inhaled nebulized enzyme therapeutics are under clinical evaluation, providing evidence that lung delivery of therapeutic enzymes is feasible (16, 38). As an example, it was calculated that 8.8 ± 1.9 mg of an antibody could be delivered to the alveoli of nonhuman primates using a novel device for protein inhalation (39). Considering a conservative estimate of 70 ml lung lining fluid volume (40), this would amount to a dose of 0.13 mg/ml, or 4.3 μM of a 30-kDa protein, widely surpassing the MICs we found for LysB in the presence of surfactant (polysorbate 80, which has been successfully included in inhaled therapeutics) (16; https://www.accessdata.fda.gov/scripts/cder/iig/index.cfm). It should, however, be considered that mucus hypersecretion is a common feature of airway inflammation, possibly lowering the effective therapeutic dose when treating MDR-TB patients (41, 42). While optimizing lung delivery, topical application of antimycobacterial enzyme therapy might already suffice to treat the skin infection Buruli ulcer caused by Mycobacterium ulcerans (43).

The repeated inhalation of nonhuman proteins might lead to an immune response, although tolerance is the default response to antigens that come into contact with the mucosa in the deep airways (44). If necessary, enzymes could be administered in controlled-release formulations to reduce the number of administrations required.

Novel drug regimens have already drastically shortened MDR-TB therapy, but still, duration, complexity, and toxicity of (second-line) drug regimens are the main challenges in the worldwide fight to end TB (1). Enhancing or accelerating the effectiveness of conventional antibiotics by adding cell wall-degrading enzymes in two to three rounds of (controlled-release) enzyme administration could potentially lower the treatment duration critically and help alleviate the immense burden of TB therapy.

MATERIALS AND METHODS

Materials.

Chicken egg white lysozyme was purchased at Sigma-Aldrich (L6876, ≥90% protein), as were human lysozyme recombinant from rice (L1667, ≥90% pure) and Aspergillus oryzae α-amylase (A8220, ≥800 fungal amylase units/g). All were dissolved in phosphate-buffered saline (PBS), and protein concentrations were determined via spectrophotometry at 280 nm. Polysorbate 80 (Sigma-Aldrich) was dissolved at 0.2% in PBS. Antibiotics isoniazid (INH) and rifampin (RIF) (Sigma-Aldrich) were dissolved as stock solutions of 1 mg/ml (INH) in PBS and 10 mg/ml (RIF) in methanol. All solutions were 0.22 μm filter sterilized.

Plasmid construction and enzyme expression.

Sequences encoding LysB originating from mycobacteriophages Ms6 and D29 (LysB-Ms6 and LysB-D29, UniProt accession numbers Q9FZR9 and O64205, respectively) were ordered as synthetic DNA in the pUC57 vector (GenScript). Via PCR, the genes were cloned into the pLH36 vector in frame with an N-terminal His₆ tag flanked by a murine caspase-3 cleavage site (sequences in Fig. S1 in the supplemental material). The expression constructs were verified by Sanger sequencing (VIB Genetic Service Facility, Antwerp, Belgium) and transformed into E. coli BL21(DE3). After initial growth at 28°C in terrific broth medium (Sigma-Aldrich) supplemented with carbenicillin, expression was induced at a culture optical density at 600 nm (OD₆₀₀) of 0.6 by the addition of 0.6 mM isopropyl-β-d-thiogalactopyranoside. LysB-Ms6 was expressed overnight at 28°C and LysB-D29 for 4 h at 37°C. Bacteria were harvested by centrifugation (8,000 × g, 15 min at 4°C).

Purification of LysB enzymes.

After resuspension, bacterial pellets were incubated for 30 min in ice-cold lysis buffer (25 mM Tris-HCl, pH 8, 5 mM MgCl₂, 0.1% Triton X-100, and 1× cOmplete protease inhibitor [Roche]) and lysed by sonication at 70% amplitude using a Qsonica sonicator (4 s on, 8 s off for 9 min). After 30 min centrifugation (100,000 × g) at 4°C, cleared lysates were 0.22 μm filter sterilized. Enzymes were isolated from the cleared lysates via nickel affinity chromatography (IMAC) using a HisTrap column (GE Healthcare) and size exclusion chromatography (SEC) using a SuperDex 75 10/300 or 16/600 SEC column (GE Healthcare) calibrated with 20 mM HEPES, pH 7, and 150 mM NaCl. After quantification (bicinchoninic acid [BCA] assay, Pierce), enzymes were 0.22 μm filter sterilized, and aliquots were snap frozen in liquid nitrogen before storage at −80°C. All experiments were performed with enzyme preparations freshly thawed from this storage stock, and no enzymes were repeatedly freeze-thawed.

Evaluation of enzyme activity.

Lipolytic LysB activity was quantified using an assay adapted from literature (17, 45). Briefly, LysB was incubated with 5 mM p-nitrophenol butyrate substrate (pNPB) in 20 mM Tris-HCl, pH 8.0, 100 mM NaCl, and 0.1% Triton X-100. The release of p-nitrophenol after hydrolysis of the pNPB substrate was measured as increase in absorbance at 400 nm for 30 min, and the ΔA₄₀₀/minute was calculated from the linear region of the blank-corrected values after a lag phase. The specific enzyme activity was calculated using the micromolar extinction coefficient of p-nitrophenol at 400 nm at 37°C of 0.0148. One LysB enzyme unit releases 1 nmol of p-nitrophenol per minute at pH 8 at 37°C using pNPB as the substrate.

PG-degrading activity of human and hen egg white lysozyme was determined on a fluorescein-labeled Micrococcus lysodeikticus cell wall substrate using the EnzChek lysozyme assay kit (Thermo Fisher) in which one unit is defined as the amount of enzyme required to produce a 0.001 units/minute change in the absorbance at 450 nm of at pH 6.24 and 25°C, using a suspension of Micrococcus lysodeikticus as the substrate.

Mycobacterium strain and handling.

To enable a bioluminescence-based Mycobacterium drug susceptibility assay, we used the pMV306hsp+LuxG13 plasmid (a gift from Brian Robertson and Siouxsie Wiles; Addgene; catalog no. 26161), which contains a bacterial luciferase operon enhanced by using the G13 promoter in front of luxC (34). The integrase-free reporter plasmid pMV306DIhsp+LuxG13 (BCCM/GeneCorner accession number LMBP11308) was produced by deleting the int gene from pMV306hsp+LuxG13 via inverted PCR using phosphorylated primers (5′-Pi-GTCCATCTTGTTGTCGTAGGTCTG-3′ and 5′-Pi-TCTTGTCAGTACGCGAAGAACCAC-3′) (46), followed by ligation of the product. The use of an integrase-free reporter plasmid and a separate suicide vector containing integrase (pBlueScript-Integrase, a gift from Peter Sander, Institute of Medical Microbiology, University of Zurich) allowed for temporary integrase activity—to integrate the reporter plasmid—and subsequent selection of integrase-free reporter strains (47). The biosafety level 2-approved triple-auxotrophic ΔpanCD ΔleuCD ΔargB M. tuberculosis mc²7902 strain (32) (a gift from W.R. Jacobs, Jr.) was cotransformed with pMV306DIhsp+LuxG13 and pBlueScript-Integrase via electroporation (12.5 kV/cm, 25 μF capacitance, 800 Ω resistance). After selection on kanamycin, the obtained M. tuberculosis mc²7902_Lux strain was stored in 1-ml aliquots at OD₆₀₀ of 1 in 20% glycerol at −80°C. For each drug testing assay, two M. tuberculosis mc²7902_Lux aliquots were thawed, combined, and cultured at 37°C with shaking in a 60-ml square bottle in 20 ml of Middlebrook 7H9 broth (BD Diagnostics) supplemented with 0.05% (vol/vol) polysorbate 80, 10% oleic acid-albumin-dextrose-catalase supplement (OADC; BD Diagnostics), 0.5% (vol/vol) glycerol, 1 mM l-arginine, 50 μg/ml l-leucine, and 24 μg/ml l-pantothenate. At days 5 and 8 after thawing, bacteria were subcultured by 1:100 and 1:20 dilution, respectively, for use in setting up the antimicrobial assays on day 9. Middlebrook 7H10 agar (BD Diagnostics) supplemented with 10% OADC, 0.5% (vol/vol) glycerol, 1 mM l-arginine, 50 μg/ml l-leucine, and 24 μg/ml l-pantothenate was used for growth of M. tuberculosis on solid culture; colony counts after incubation at 37°C for up to 12 weeks were given as CFU per milliliter plated (CFU/ml).

A bioluminescence microtiter assay for mycobacterial growth inhibition.

To remove bacterial clumps before setting up the microtiter assay, mycobacteria were passed three times through a 27G needle. An M. tuberculosis mc²7902_Lux inoculum at an OD₆₀₀ of 0.008 (2 × 10⁶ CFU/ml) was prepared in assay medium (Middlebrook 7H9 broth supplemented with 10% OADC, 0.5% [vol/vol] glycerol, 1 mM l-arginine, 50 μg/ml l-leucine, and 24 μg/ml l-pantothenate) without any polysorbate 80 unless specifically stated otherwise.

Drug susceptibility was tested in a microtiter plate format based on the bioluminescence assay proposed by Andreu et al. (33). Briefly, duplicate 2-fold serial dilutions of various compounds were prepared in assay medium in sterile 96-well black opaque plates (CulturPlate-96 Black; PerkinElmer). To allow for synergy testing, horizontal 2-fold dilution series of one compound was combined with vertical 2-fold dilution series of another compound in a checkerboard fashion (Fig. S2). Afterwards, the M. tuberculosis mc²7902_Lux inoculum was added to 2 × 10⁵ CFU per well. Each plate contained three wells containing inoculum without drugs (positive growth controls) and three wells containing no inoculum (negative controls). All perimeter wells were filled with 200 μl ultrapure water to limit evaporation from sample wells. Assay plates were incubated at 37°C in an incubator providing 5% CO₂ and 80% humidity, which yielded a higher reproducibility than when plates were placed (in a sealed box or plastic bag) in a regular incubator (data not shown). Bioluminescence was measured after 1, 4, 7, 10, and 13 days of incubation in a GloMax 96 or GloMax Navigator microplate luminometer (Promega) with 1.0 s integration per well and expressed as relative light units (RLU).

Bioluminescence microtiter assay data processing.

Three independent repetitions (“true replicates”) were performed for each assay. Within these independent repetitions, each microtiter plate was set up in duplicate (technical replicates), and bioluminescence data of these technical duplicates were averaged in a first stage of data analysis. In a second stage, data of the independent repetitions were subjected to fitting of the symmetrical generalized logistic model (equation 1) and the asymmetrical Gompertz model (equation 2) (48, 49). Both models are sigmoid dose-response curves relating mycobacterial bioluminescence (y, in RLU) to the log of the antimicrobial concentration of the target compound (x), where A is the lower asymptote, B represents the slope, C is the distance between lower and upper asymptotes, and M is the x value at the inflection point (Fig. 3a). Curves were fitted to the data at each time point separately using GenStat 19.1 (VSN International Ltd.), and the percentage of variance accounted for by the fitted curve was used to judge the best dose-response model and optimal time point for each data set.

The MIC (equation 3 and Fig. 3a) (48, 49), defined as the intersection of the lower asymptote with the tangent to the inflection point of the curve, was calculated for the best-fitting dose-response curve of each data set. If no inhibition was observed, the MIC was set as larger than or equal to the highest concentration tested. If inhibition was observed but no dose-response curve could be fitted, the MIC was set as lower than or equal to the lowest value for which all data points were lower than 200 RLU (corresponding to the bioluminescence of half the inoculum).

$$y=A+\frac{C}{1+e^{-B(x-M)}}$$ (1)

$$y=A+C{e}^{-e^{-B(x-M)}}$$ (2)

$$\text{MIC}={10}^{M-\frac{1}{B}}$$ (3)

Correlation between CFU/well and RLU/well was assessed using the SciPy Stats package (50, 51).

Synergy evaluation using Combenefit.

Synergy was first evaluated using the Combenefit tool (35). Briefly, this tool fits a Hill equation dose-response curve for each of the two compounds and then predicts the additive effect according to several reference models (Loewe, Bliss, and highest single agent [HSA]) (52–54). Subsequently, it calculates the deviation of the real data from the predicted additive model, resulting in synergy and antagonism scores. The results shown here (Fig. 4) use the Bliss independence model for calculating the expected combined effect from the dose-response matrix data, assuming the two drugs act independently (55). Results obtained using the Loewe and HSA models correspond closely (Fig. S4).