Intrapulmonary Treatment with Mycobacteriophage LysB Rapidly Reduces Mycobacterium abscessus Burden

ABSTRACT

Intrinsic and acquired antibiotic resistance in Mycobacterium abscessus presents challenges in infection control, and new therapeutic strategies are needed. Bacteriophage therapy shows promise, but variabilities in M. abscessus phage susceptibility limits its broader utility. We show here that a mycobacteriophage-encoded lysin B (LysB) efficiently and rapidly kills both smooth- and rough-colony morphotype M. abscessus strains and reduces the pulmonary bacterial load in mice. LysB aerosolization presents a plausible treatment for pulmonary M. abscessus infections.

TEXT

Mycobacterium abscessus infections are becoming more prevalent and are often nonresponsive to antibiotics due to intrinsic and acquired resistance (1, 2). M. abscessus infections are relatively common in people with cystic fibrosis (CF) and are common contraindicators for lung transplantation (3). There is an urgent need for new therapeutic approaches, and engineered mycobacteriophages show promise in several compassionate use cases (4–7). However, the relatively narrow host phage preference and M. abscessus strain diversity present challenges to the broader implementation of this approach (8). Phage-encoded lytic enzymes (lysins) have been proposed as antimicrobials for strains such as Streptococcus and Staphylococcus (9) and may also be applicable to M. abscessus infections.

Mycobacteriophages are unusual in coding for two lytic enzymes, lysin A (LysA) and lysin B (LysB; GenBank protein accession no. NP_046827.1), and there is enormous sequence variation in both (10, 11). LysA enzymes hydrolyze peptidoglycan and are essential for cell wall degradation, whereas LysB hydrolyzes the mycolyl ester linkages in mycolyl-arabinogalactan-peptidoglycan (mAGP) (10) and, to a lesser extent, in trehalose-6,6′-dimycolate (TDM) (12, 13); mAGP is a core constituent of the inner cell wall, and TDM is an abundant peripheral constituent of the cell envelope (14). Although LysA-like endolysins have antibacterial activity against Gram-positive bacteria when administered exogenously, the mycolic acid-rich mycobacterial outer envelope prevents external access to their peptidoglycan substrate. In contrast, the enzyme LysB inhibits growth when applied exogenously to Mycobacterium smegmatis, Mycobacterium ulcerans, and Mycobacterium tuberculosis cells (15, 16).

Exposure to M. smegmatis TDM hydrolase (Tdmh), a bacterial enzyme related to LysB that specifically hydrolyzes the ester linkage of TDM, causes lysis of mycobacteria (12, 17), implying that TDM is exposed on the surface of mycobacteria and is vulnerable to degradation by exogenous enzymes, including LysB. While comparing the effects of Tdmh and phage D29 LysB (10) on M. smegmatis and M. tuberculosis (strain mc²7000), we observed concentration-dependent killing by D29-LysB of more than 99% of 10⁷ cells of both species within 30 min of exposure (Fig. 1A and B). However, Tdmh acts relatively slowly, and in 30 min, there is little killing in either of the species (Fig. 1B). Such a rapid killing by D29-LysB was unexpected, although the timing was strongly corroborated by rapid ATP release due to lysis (Fig. 1C), as well as rapid loss of TDM (Fig. 1D and E). However, it is unclear whether TDM loss or mAGP complex hydrolysis is directly involved in lysis. While the antimycobacterial activity of D29-LysB has been noted previously (15, 16), this is the first demonstration of its rapid action.

FIG 1.

Rapid lysis of mycobacteria by D29-LysB. (A) Change in the viability of M. tuberculosis (mc²7000) upon exposure to D29-LysB at the indicated concentrations for 10 and 30 min. ND, no colonies detected. (B) Activity of D29-LysB compared to that of Tdmh against M. smegmatis (mc²155) and M. tuberculosis (mc²7000). Cells were mixed with 8 μM purified recombinant D29-LysB, Tdmh, or both proteins and incubated at 37°C for the indicated time periods. Then, dilutions were plated on Middlebrook 7H11 solid growth medium supplemented with oleic acid-albumin-dextrose-catalase (7H11OADC) (for M. tuberculosis) or 7H10ADC (for M. smegmatis). (C) Release of ATP upon incubation of M. tuberculosis (mc²7000) with D29-LysB. Approximately 10⁷ cells were mixed with 8 μM D29-LysB, and after incubation at 37°C for the indicated time, ATP in the supernatant was measured using a luciferase-based-assay kit (ENLITEN; Promega) and a luminometer (Veritas). Values are expressed as relative light units (RLU). Cells without the enzyme and the enzyme with the buffer were background references. A catalytic mutant of D29-LysB (S82A) was also used as a control. Data represent the mean ± SD from three biologically independent experiments. (D) Radio-thin layer chromatography (TLC) showing TDM levels in ¹⁴C-labeled M. tuberculosis (mc²7000) cells exposed to 8 μM D29-LysB for 30 min. Cells were heat inactivated prior to exposure. After the exposure, lipids were extracted in 2:1 chloroform/methanol and resolved on TLC using chloroform/methanol/water (90:10:1) as the solvent phase. Cells exposed to the buffer were used as the control. Purified ¹⁴C-labeled TDM and TMM were used as markers. A lipid of unknown identity (denoted with a question mark) accumulated with concomitant loss of TDM. (E) TDM profile of mc²7000 cells exposed to 8 μM either D29-LysB or Tdmh over a shorter time period.

Because of the rapid and efficient lytic activity of D29-LysB, we evaluated the enzyme for its antimicrobial activity on M. abscessus, a rapidly growing and intrinsically drug-resistant mycobacterium related to M. smegmatis. In a 2-h exposure of 10⁷ CFU/mL bacilli of the type strain ATCC 19977 to various concentrations of recombinant D29-LysB, a concentration-dependent killing was observed that reached saturation at 4 μM (Fig. 2A). We next tested the efficacy of D29-LysB at 2× saturating concentration (8 μM) over a range of exposure times against ATCC 19977 (smooth morphotype), its rough-colony morphotype CIP104536^R, and 6 rough and 6 smooth clinical isolates of M. abscessus. D29-LysB was effective in killing more than 99% of bacilli for both morphotypes within 30 min of exposure (Fig. 2B and C). Most of the clinical isolates of the smooth morphotype displayed sensitivity to the exposure in a manner similar to that of ATCC 19977 (Fig. 2B), although the rough clinical strains were somewhat less sensitive, and one (Mab606) had only a modest sensitivity (Fig. 2C). The data indicate that while most smooth strains are highly sensitive to D29-LysB, the rough morphotypes display a variable degree of susceptibility to the enzyme. The sensitivity of the smooth morphotypes to D29-LysB was somewhat unexpected, given that these variants are generally more resistant to mycobacteriophages and less efficiently killed than rough variants (8). In addition, a clear zone of growth inhibition was observed when 40 μM D29-LysB solution was spotted onto detergent-free agar plates containing ~10⁶ CFU of strains ATCC 19977, CIP104536^R, or Mab103 (Fig. 2D), indicating that the presence of Tween 80 was not an obligate requirement for D29-LysB activity. However, potentiation of the activity by the detergent, suggested previously in other mycobacteria (15, 16), cannot be ruled out. Finally, a measure of ATP release confirmed the lysis of D29-LysB-exposed M. abscessus cells (Fig. 2E).

FIG 2.

Lysis of M. abscessus by D29-LysB in vitro. (A) Concentration-dependent killing of ATCC 19977 by D29-LysB. Approximately 10⁷ CFU/mL of bacilli were mixed with the indicated concentrations of recombinant D29-LysB in 250 μL of phosphate-buffered saline (PBS) containing Tween 80; the mixture was incubated at 37°C for 2 h, and dilutions were plated on 7H10OADC plates. (B and C) Change in the viability of the smooth (B) and rough (C) variants of M. abscessus after in vitro treatment with 8 μM D29-LysB for the indicated time periods. ATCC 19977 is the type strain, CIP104536^R is a glycopeptidolipids (GPL)-negative mutant of CIP104536 that forms rough colonies, and all other strains were isolated from clinical specimens. Approximately 10⁷ CFU/mL of bacilli of each strain was mixed with the enzyme under the conditions described in panel A, and cells were diluted and plated on 7H10OADC plates at specified intervals. (D) Activity of D29-LysB when spotted on a lawn of one smooth (ATCC 19977) and two rough (CIP104536^R and Mab103) variants of M. abscessus grown on a 7H10OADC agar plate. (E) Luciferase-based measurement of ATP released from 10⁷ ATCC 19977 cells exposed to either D29-LysB (8 μM) or the protein storage buffer over the indicated time period. See Fig. 1C for the assay conditions. Data represent the average of two biologically independent experiments.

The rapid and efficient killing of M. abscessus cells by D29-LysB in vitro led us to test the in vivo efficacy of the enzyme on pulmonary M. abscessus infection in a mouse model using an IACUC-approved protocol (Colorado State University [CSU] protocol number 1650). In two independent experiments, SCID mice were infected with 1 × 10⁶ CFU of Mab103 through intrapulmonary aerosolization using an FMJ-250 high-pressure syringe device (PennCentury, Philadelphia, PA, USA) with an attached MicroSprayer (model IA-C; PennCentury), as previously described (18). Briefly, the mice were anesthetized using a mixture of isoflurane and oxygen. Each mouse was quickly placed in a stand with its teeth suspended up at a 45° angle, and its tongue gently rolled out using a cotton tip. The MicroSprayer tip was then inserted into the trachea and bacteria inoculum sprayed out into the lungs. Two days postinfection (dpi), a group of mice were treated daily with 50 μL of aerosolized endotoxin-free D29-LysB at 40 μM for 6 days, using the FMJ-250 device in a procedure similar to the one described above for infection. At 2 and 6 dpi, mice from the untreated group were euthanized; the lungs were homogenized and plated on Middlebrook 7H11 agar supplemented with oleic acid-albumin-dextrose-catalase (7H11OADC) to determine the bacterial burden during the normal course of infection. One day after the last treatment with D29-LysB (9 dpi), all treated and untreated mice were euthanized, and the bacterial burdens in the lungs were compared (Fig. 3A). While the bacterial burden in the lungs of the untreated group decreased by about 10-fold over the 9-day period, the administration of D29-LysB caused an additional 20-fold decrease in the bacterial load (Fig. 3B). Mice in the treatment group, however, showed weight loss of less than 10% (Fig. 3C) but without any noticeable change in their overall activities.

FIG 3.

Therapeutic potential of D29-LysB against M. abscessus in a mouse model. (A) Schematic illustration of the experimental design (dpi, days postinfection). (B) Pulmonary burden of strain Mab103 in the lungs of untreated SCID mice at 2, 6, and 9 dpi and in D29-LysB-treated mice at 9 dpi. Daily, for 6 days (from 2 dpi to 8 dpi), 50 μL of aerosolized D29-LysB solution (40 μM) was delivered through the intrapulmonary route. Data are an aggregate of two independent experiments, with each data point representing one mouse. **, P < 0.01; ****, P < 0.0001; Kruskal-Wallis test. (C) Average change in the body weight of the animals in the untreated and D29-LysB-treated groups at the experimental endpoint relative to the initial weight. *, P < 0.05 (unpaired t test).

In summary, we report a potent bactericidal activity of recombinant D29-LysB against M. abscessus. Moreover, D29-LysB demonstrated strong efficacy when administered directly into the lungs of mice with an acute Mab103 infection during short periods of therapy. Further studies are needed to test the efficacy of D29-LysB during chronic stages of infection and during prolonged periods of administration. Given that bronchial tubes are the primary sites of bacterial colonization in CF patients, aerosolized D29-LysB can be a promising therapeutic agent to rapidly reduce M. abscessus loads in such patients. D29-LysB may therefore be a potent adjuvant for use with either antibiotic or bacteriophage therapies and could help to moderate toxicities arising from extended antibiotic use. We also note that D29-LysB may be particularly useful for treatment of smooth-colony morphotype strains, for which no bacteriophages are yet available (8). Because of the substantial sequence diversity among LysB enzymes encoded by mycobacteriophages, additional enzymes may have greater activity against strains such as Mab606 that are less efficiently killed by D29-LysB. Further characterization of LysB in CF infection models, as well as the discovery of new phage lysins, will expand the treatment options for M. abscessus infections.