ABSTRACT
Mycobacterium abscessus is a non-tuberculous Mycobacterium (NTM) that is often associated with respiratory infections in persons with cystic fibrosis. High levels of antibiotic resistance make M. abscessus infections difficult to treat, highlighting the urgent need for alternative therapeutic strategies. Mycobacteriophage therapy—treatment of infections with viruses that infect Mycobacterium hosts—shows considerable promise, but the narrow host specificity of the current mycobacteriophage repertoire limits this approach to personalized therapies. Here, we characterize the anti-M. abscessus activity of mycobacteriophage-encoded Lysin B enzymes (lysterases). Lysterases are serine-esterases that cleave the ester bonds linking mycolic acids to either arabinogalactan or trehalose in the mycobacterial cell envelope to complete the lysis process. In this study, we explored their ability to kill a spectrum of clinically isolated M. abscessus strains when administered exogenously. Six phylogenetically distinct lysterases were purified, characterized on the surrogate M. smegmatis, and their activity against a diverse panel of 24 M. abscessus clinical strains was assessed. The enzymes exhibit a range of activities against these strains, but most have broad-spectrum activity with only minor variations among the M. abscessus isolates. However, the activities can be quite potent, resulting in up to six orders of magnitude of reduction in viability within 4 h with one activity unit of the enzymes, and act synergistically with amikacin. Overall, mycobacteriophage-encoded lysterases show considerable potential for antimycobacterial control, especially for topical applications.
IMPORTANCE
Mycobacterium infections are clinically challenging, and there is a high failure of resolving Mycobacterium abscessus infections using antibiotics. Mycobacteriophages show promise in treating M. abscessus infections, but there is substantial variation among clinical isolates in their phage infection profiles. Mycobacteriophages are unusual in coding for lysis enzymes—lysterases—that target cell wall exposed constituents, and these can have potential antibacterial activity. Exploring the vast sequence diversity of thousands of phage-encoded lysterases identifies a subset that was expressed, purified, and tested for their ability to kill a broad range of M. abscessus clinical strains. Lysterases with the broadest and greatest activity offer substantial potential for treatment of M. abscessus infection, especially where they display synergism with current antibiotics.
KEYWORDS: mycobacteriophage, enzybiotics, lysterase, Mycobacterium abscessus
INTRODUCTION
Non-tuberculous mycobacteria (NTMs) are increasingly associated with respiratory infections in persons with cystic fibrosis (pwCF), as well as other immunocompromised cohorts. Prolonged antibiotic treatment often leads to the emergence of drug resistance, resulting in clinical challenges in treating NTM infections (1, 2). Mycobacterium abscessus is a common NTM species that taxonomically contains three different subspecies: abscessus, massiliense, and bolletii (1). M. abscessus grows quickly relative to many other mycobacterial strains (doubling time ~4 h) and exists environmentally and as an opportunistic pathogen (3–5). Multidrug regimens for M. abscessus, including use of dual beta-lactam, macrolides, and fluoroquinolones (6, 7), frequently fail, and treatment success is observed in only 40–50% of patients (8). Alternative therapeutic strategies are clearly needed, particularly for pwCF who fail to respond to or cannot tolerate antibiotic therapies.
The therapeutic use of bacteriophages is an alternative approach to antibiotic treatment that has been utilized in the former Soviet Union and Eastern European countries for nearly a century and has recently experienced a resurgence in Western Europe and America. Since 2019, phage therapy has been explored for compassionate use to control M. abscessus infections in CF patients, resulting in favorable outcomes and clinical improvements in a substantial proportion of patients (9–11). Through the Science Education Alliance Phage Hunters Advancing Genomics and Evolutionary Science (SEA-PHAGES) program (12), undergraduate researchers from 180 institutions have contributed to a collection of ~15,000 mycobacteriophages from environmental samples (11, 13). However, only 6–8 of these—or their mutant derivatives—efficiently infect and kill at least one M. abscessus isolate, and smooth colony morphotype strains are killed too insufficiently to warrant clinical interventions (14). In general, only about a quarter of M. abscessus strains are potential targets for phage therapies, and narrow host ranges of the phages are a major impediment to broader therapeutic use (14). Thus, although phage therapies for NTM infections hold some promise, alternative strategies are worth investigating.
Phage lysins are expressed during the late stages of phage infection to rupture the cell wall and release phage progeny. Most double-stranded DNA (dsDNA) tailed phages code for an endolysin that degrades peptidoglycan, and in Gram-negative bacterial hosts, a spanin or disruptin may also act to compromise the outer cell wall leaflet (15–17). In Gram-positive bacteria lacking an outer membrane, the endolysin can access the peptidoglycan from the outside environment, and these endolysins have been extensively investigated as potent antibiotics (enzybiotics) (15, 18–21). The mycobacteria present a particular challenge to phage endolysins because of their complex cell walls that contain a mycolic-rich outer layer referred to as the mycomembrane (22, 23). Mycobacteriophages typically have genes for two distinct lysins: Lysin A and Lysin B (24, 25). Lysin A acts as the endolysin to degrade peptidoglycan, and Lysin B (LysB) is a mycolylarabinogalactan esterase that cleaves the linkage of mycolic acids to the rest of the cell wall to complete cell lysis (25) and also cleaves trehalose dimycolate (TDM), another component of the mycomembrane (26, 27). Lysin B can thus be referred to as a “lysterase” to distinguish it from endolysins, the peptidoglycan-degrading enzymes.
Mycobacteriophage endolysins are not potently antimicrobial, as the mycomembrane limits access to their peptidoglycan targets. However, the lysterases do exhibit anti-Mycobacterium activity when added exogenously to M. smegmatis, M. tuberculosis, and M. ulcerans (28–32), possibly because of their promiscuous hydrolytic activity against TDM. The lysterase encoded by phage D29 has also been shown to be active against M. abscessus and to reduce bacterial burden during intrapulmonary treatment in mice (33). Combinations of endolysin, lysterase, and antimicrobial peptides have also been proposed as an anti-Mycobacterium platform (34). However, combinations of lysterases and phages may have only limited utility, as exogenous addition of D29 lysterase was shown to reduce the adsorption of phage D29 particles in M. smegmatis (35).
Here, we describe a bioinformatic approach to exploring the diversity of the >2,000 lysterases encoded by mycobacteriophages, which we used to guide the expression and purification of six sequence-diverse enzymes. We compared their relative activities against M. smegmatis in both exponential- and stationary-phase growth and determined their antimicrobial activities against a large set of diverse M. abscessus clinical isolates, including strains with either smooth or rough colony morphotypes, as well as those that are both antibiotic resistant and not susceptible to infection by any tested phages. We also demonstrate synergism between lysterases and amikacin against M. abscessus.
RESULTS
Diversity of mycobacteriophage lysterase proteins
The Actino_Draft database (as of June 2026) contains 2,127 lysterase proteins encoded by 2,803 Mycobacterium phages. Using MMseqs2 (36), we assembled these sequences and a small subset of M. abscessus prophage lysins into 49 amino acid sequence-defined groups, reflecting substantial diversity in group size, protein lengths, and amino acid sequences (Fig. 1A and B; Table S1). Group sizes varied considerably, ranging from one to two members to several hundred; protein lengths ranged from 240–500 amino acids, and most groups exhibited narrow length distributions consistent with conserved architectures within individual groups. Amino acid length distributions showed that most groups had a narrow interquartile range (IQR), a statistical measurement of data dispersion, consistent with a single domain architecture (Fig. 1A). However, some groups showed substantial length dispersion suggestive of structural variation (Fig. 1A). Groups 6, 8, and 9 displayed clear bimodal length distributions separated by approximately 60–140 amino acid residues, consistent with two architectural variants within the same sequence-defined family. These groups were subdivided into A and B subgroups, representing longer and shorter proteins, respectively. Group 5 showed a more moderate, but still bimodal, length dispersion (IQR ≈ 37 aa) and was also split into A and B subgroups.
Fig 1.

Bioinformatic analysis and selection of diverse lysterase enzymes. (A) A total of 2,127 protein sequences are grouped by MMseqs2 to visualize within-group variation. Distribution of protein length (number of amino acids) is shown on the y-axis, and some groups with bimodal distributions are divided into “A” and “B” subgroups as indicated. (B) Tanglegram comparing dendrograms derived from structural alignment (left; FoldMason) and amino acid sequence analysis (right; IQ-TREE) of representative lysterase proteins. Red stars indicate groups from which the representative lysterases are selected and characterized; gray stars indicate groups from which proteins are purified but not studied further. Lines connect identical proteins between trees. Terminal labels correspond to the sequence-defined clusters identified by MMseqs2; A/B labels indicate architectural subgroups. Rulers below each tree indicate branch-length scales. Branch lengths reflect relative sequence divergence in the IQ-TREE phylogeny and relative structural divergence in the FoldMason dendrogram. (C) The SDS-PAGE gel shows the tested enzymes were purified to near homogeneity. Each lane contains 1 μg of enzyme. (D) AlphaFold2-predicted structures of the six lysterase proteins characterized in detail here, with group and molecular mass information indicated below. The initial methionine and C-terminal amino acid are labeled for each protein, and the β-sheets of the α/β sandwich are colored blue, situating the active site of lysterase.
Representative lysterase proteins were modeled using AlphaFold2 (37) and were structurally aligned using ChimeraX (Fig. S1) and assembled into a structural dendrogram using FoldMason (38) (Fig. 1B). Representatives used for structural comparison and tree generation corresponded to the median-length protein within each group or validated architectural subgroup, such that the selected proteins reflect the central architecture of each family. Despite sequential diversity, the predicted protein folds all share a conserved core that includes the catalytic triad that defines serine esterases (Fig. S1). Comparison of structural and sequence-derived dendrograms showed broadly consistent grouping patterns (Fig. 1B). Major groupings were preserved between the two trees, indicating that sequence similarity generally reflects structural relationships among lysterase proteins. However, several groups differ in relative placement between the structural and sequence-derived trees, suggesting that structural comparisons may capture relationships that are not fully resolved by sequence identity alone. Structural analysis of the A and B subgroups (Fig. S1B) showed architectural differences between longer and shorter variants, respectively, supporting subdivision. Across all comparisons, shorter and longer variants shared a conserved core (25), with structural differences localized outside the catalytic domain and including N- and C-terminal extensions, internal insertions, and additional structural elements. Groups 8 and 9 primarily differed through terminal extensions, whereas Group 5 contained a localized internal insertion and Group 6 had both a terminal extension and an internal insertion. Despite these differences, all subgroup pairs maintained a conserved catalytic core, though minor deviations in the histidine placement were observed within both subgroups of Group 8.
Expression and purification of six representative lysterases
We initially purified 19 putative lysterases as shown in Fig. 1B. Six representative proteins reflecting the overall diversity—Bongo gp38 (Group 18), Bxb1 gp9 (Group 3), Constella gp41 (Group 13), D29 gp12 (Subgroup 9B), MissWhite gp11 (Subgroup 6B), and phiGD57-1 gp25 (Subgroup 6A)—were highly soluble and selected for further characterization and assessment of antimycobacterial activity. These lysterases range in molecular mass from 29.4 to 41.6 kDa (Table 1, Fig. 1C), and AlphaFold structures predict one active site (α/β repeats and their linkers [25]) with or without an N-terminal domain (Fig. 1D). Overexpression plasmids were constructed and transformed into Escherichia coli, and His-tagged proteins were purified to near homogeneity (Fig. 1C). All six proteins are highly soluble, with average yields ranging from 4.3 to 19.6 mg/g of cell dry mass (Table S2). Lysterase stocks were stored in buffer with 50% glycerol at −20°C.
TABLE 1.
Lysterases characterized in this study
| Lysin sourcea | Clusterb | Molecular massc | GenBank accession |
|---|---|---|---|
| D29 | A2 | 30,031 | AAC18452 |
| MissWhite | A2 | 29,356 | ASZ75177 |
| Bxb1 | A1 | 38,217 | AAG59714 |
| Constella | J | 31,310 | AXF51531 |
| Bongo | M1 | 35,697 | YP_009604896 |
| phiGD57-1 | HC | 41,599 | QPO17264 |
Phage from which lysterase enzyme was purified.
Phage cluster/subcluster designation.
Molecular mass of the lysterase protein in daltons.
Bactericidal activity of lysterases against M. smegmatis
To facilitate comparison of lysterase activities, we first showed that incubation of high concentrations of each enzyme with M. smegmatis in exponential growth resulted in substantial loss of viability within one hour (Fig. 2A). We then determined the concentration of enzyme that conferred a 50% reduction in viability in one hour (EC50) and therefore defined 2 × EC50 of enzyme as one unit of activity. In general, losses in viability determined by colony counts (Fig. 2A) fit to non-linear regressions (Fig. 2B), and the specific activities vary among the enzymes, with D29 being the least active (EC50: 462 nM) and Bxb1 having the highest activity (EC50: 97 nM) in this assay (Fig. 2C). It is notable that variations in activity differ for the different enzymes as a function of protein concentration (Fig. 2B).
Fig 2.

Lysterase-mediated killing of M. smegmatis and establishing half-maximal effective concentration. (A) M. smegmatis (105 CFU) was incubated at 37°C with enzymes at different concentrations for one hour before 100-fold dilution and spreading onto plates and incubated for 4 days prior to colony counting. (B) Titration curves for each enzyme with the calculated log (CFU/mL) after a 60-min incubation (from panel A) plotted as a function of log (enzyme concentration). These data sets were fit in GraphPad Prism using a five-parameter logistic curve with a constrained maximum (from control plates with no enzyme) and minimum (lower limit of detection). The half-maximal value was determined from the fit. (C) Three biological replicates of the experiment shown in panels A and B were performed, and the half-maximal effective concentration determined from these are plotted for each replicate as concentration in nM on the y-axis for each lysterase as labeled on the x-axis. The bar height represents the average half-maximal effective concentration, and the error bars indicate the standard deviation.
We then compared the lysterase activities on exponentially growing and stationary-phase M. smegmatis cultures and observed that the reduction in viability was not substantially influenced by growth phase (Fig. 3A and B). However, when viability was determined at multiple time points, we did observe some differences in the kinetics of antibacterial activity (Fig. 3C and D). Specifically, the Constella, MissWhite, and phiGD57-1 enzymes appear to kill stationary-phase cultures more rapidly than log-phase cultures, and we note that the phiGD57-1 enzyme not only acts more slowly on log-phase cultures but also kills them incompletely, which may reflect enzyme decay over the incubation time. We note that at the enzyme concentrations used, the near-complete loss of viability was quite rapid; for example, with the Bongo enzyme, most of the cells were killed in less than 20 min.
Fig 3.

Kinetics of lysterase-mediated killing of M. smegmatis mc2155. M. smegmatis cells (105 CFU) from exponentially growing and stationary-phase cultures were incubated with 2 × EC50 value calculated from experiments like those shown in Fig. 2, and incubated at 37°C. Samples were collected at different time points and promptly diluted 100-fold before spreading onto agar plates, which were then incubated at 37°C for 5 days to quantify viable colony-forming units. (A) Cells taken from the exponential phase culture experiment immediately after mixing lysterase and cells (top panels) and after 42 min of incubation (bottom panels). (B) Cells taken from the stationary phase culture experiment immediately after mixing lysterase and cells (top panels) and after 42 min of incubation (bottom panels). (C and D) Log (CFU/mL) of viable colony-forming units are plotted as a function of time for exponentially growing (C) and stationary-phase (D) cells. Controls with no enzyme are shown at the top, and enzyme conditions are labeled at the top of each plot. The horizontal dashed line represents the limit of detection. The vertical dashed line indicates the time to reach the half-maximal value for each condition; this time is shown as a text insert for each plot. No T50 value is available for phiGD57-1 log-phase curve. Biological replicate data are shown in Fig. S2.
To visualize lysterase-mediated antimycobacterial activity, we stained cultures treated with the Bxb1 enzyme using a live-dead stain and examined them microscopically (Fig. 4A). After 1-h incubation of M. smegmatis cells with the Bxb1 enzyme, the vast majority of cells were stained with propidium iodide, which is consistent with loss of viability, with few remaining live cells observed. These live cells were typically observed within cell clumps, suggesting that lack of complete cell dispersion could interfere with activity. The propidium iodide-stained cells appear to maintain their cell shape, suggesting that the peptidoglycan layer is not disrupted (Fig. 4B).
Fig 4.

Microscopic analysis of lysterase treatment of M. smegmatis. (A) M. smegmatis (5 × 107 CFU) was treated with a high concentration of Bxb1 lysterase (approximately 19 units) for 1 h at 37°C. PBST and cells without enzyme were used as negative control. The samples were stained with Syto 9 (green-live) and propidium iodide (PI; red-dead), confocal 60× obj. Scale bar is 30 µm. (B) Enlarged images of M. smegmatis cells illustrating cell shape integrity after lysterase treatment under the same conditions as panel A.
Lysterase killing of clinically derived M. abscessus strains
To determine the antibacterial activity of the lysterases against clinically relevant Mycobacterium strains, we tested 24 M. abscessus clinical isolates. These were chosen based on their genomic diversity, phage infection profiles, and colony morphotypes (10, 39). Seventeen of the strains have poor or no susceptibility to infection by a panel of phages used therapeutically against NTM clinical isolates; 13 strains have a smooth colony morphotype, 10 are rough, and one (GD30) has an intermediate morphotype. All three subspecies were also represented: M. abscessus subsp. abscessus (i.e., GD03, 33A, 100A, and 100B), subsp. bolletii (GD37 and 91), and subsp. massiliense (GD30) (Table 2). We also included two isogenic pairs of rough/smooth strains (GD100A/B; GD213B/C).
TABLE 2.
M. abscessus clinical isolate profiles
| Designator | Strain | DCCa | Morphotypeb | Phage susceptibilityc |
|---|---|---|---|---|
| GD03 | M. abscessus complex | Non-DCC | S | N |
| GD19 | M. abscessus abscessus | 5 | R | Y |
| GD28 | M. abscessus complex | 4 | S | N |
| GD30 | M. abscessus massiliense | 7 | I | Poor |
| GD33A | M. abscessus abscessus | 5 | S | Poor |
| GD36 | M. abscessus abscessus | 1 | S | Y |
| GD37 | M. abscessus bolletii | N/A | S | Poor |
| GD44 | M. abscessus abscessus | 3 | S | N |
| GD55 | M. abscessus abscessus | 4 | S | N |
| GD60 | M. abscessus abscessus | 6 | R | N |
| GD72 | M. abscessus abscessus | 5 | S | Poor |
| GD85 | M. abscessus abscessus | 5 | S | N |
| GD87 | M. abscessus abscessus | Non-DCC | R | N |
| GD91 | M. abscessus bolletii | N/A | S | N |
| GD95 | M. abscessus abscessus | Non-DCC | R | N |
| GD97 | M. abscessus abscessus | 4 | S | Poor |
| GD100A | M. abscessus abscessus | 1 | S | N |
| GD100B | M. abscessus abscessus | 1 | R | N |
| GD102 | M. abscessus abscessus | 2 | R | Y |
| GD111 | M. abscessus abscessus | 1 | R | Y |
| GD158 | M. abscessus abscessus | Non-DCC | R | Y |
| GD180 | M. abscessus abscessus | Non-DCC | R | Y |
| GD213B | M. abscessus abscessus | 4 | R | Y |
| GD213C | M. abscessus abscessus | 4 | S | N |
DCC, dominant circulating clone; Non-DCC, strains that do not belong to the analyzed clades (1-7); N/A, not applicable, as subspecies bolletii is not a part of DCC.
S, smooth; R, rough; I, intermediate.
N, none of tested phages infect the strain; Y, yes, at least one tested phage efficiently infects and kills the strain; Poor, at least one tested phage infects and kills the strains, but either with numerous survivors or low efficiency of plaquing.
To test the M. abscessus-specific bactericidal activity of each of the six lysterases, we incubated one activity unit of each lysterase with the M. abscessus strains for 4 h, and surviving cells were stamped on solid media at different time points (Fig. 5A). Ten of 24 strains (GD19, 30, 36, 37, 44, 72, 95, 97, 102, and 111) were observed to be killed by all tested lysterases, while only one strain (GD91) was resistant to killing by all but the D29 lysterase (Fig. 5A and Fig. S4). Based on the reduction in bacterial growth over the incubation period, the D29 and MissWhite lysterases were the most effective lysterases tested, efficiently killing 24 and 23 M. abscessus strains, respectively (Fig. 5A). Bxb1 and Constella lysterases completely killed 18 and 19 strains after 4 h, respectively, while Bongo was effective against 16 strains, followed by phiGD57-1, which was able to kill only 10 strains out of 24 (Fig. 5A).
Fig 5.

Lysterase killing of Mycobacterium abscessus clinical isolates. (A) Twenty-four M. abscessus clinical isolates were diluted to 106 CFU/mL and mixed with one unit of enzyme in PBST for the indicated times. At each time point, 1 µL of each sample was plated directly on solid medium without dilution and grown for 12 days at 37°C to visualize the presence of viable colony-forming units. Two no-enzyme controls (Crtl1, Crtl2) are included. The number in parentheses after each enzyme indicates the total number of strains that were killed after the treatment. Each data set includes 24 M. abscessus isolates, and the layout of the strain designations (GDxxx) is shown at the right. (B) Representatives of killing curves were generated from the data shown in panel A. The growth area for each strain, using CellProfiler, was plotted using a nonlinear curve fit (Asymmetric [five parameters], X is concentration) in GraphPad Prism. Two curves shown on the graph were selected based on killing efficiency: the lowest (circle) and the highest (square). More killing curve data are shown in Fig. S3. (C) The time to reduce bacterial growth by 50% was determined by quantifying bacterial growth in panel A, plotting these areas as a function of time, and identifying T50, the time to reach the half-maximal area. For example, D29’s lysterase reduces growth of GD87 by 50% within 17 min, whereas the phiGD57-1 lysterase did not achieve a 50% reduction of GD87 growth by 240 min. Average T50 values are shown for each strain and each enzyme in the last row and column.
We quantified cell survival over time and fit the data using non-linear regression curves (Fig. 5B and Fig. S3). Most killing activity was complete within 3 h. The time required to reduce bacterial growth by 50% (T50) revealed substantial differences in susceptibility between the enzymes and the clinical isolates. One activity unit of the D29 lysterase efficiently killed all strains within 4 h of treatment (Fig. 5C), reducing GD87 growth by 50% in as little as 17 min, while requiring 185 min to eliminate 50% of GD60. Interestingly, GD60, as well as GD03, could only be eradicated at long times by the lysterases of D29 and MissWhite, but not by the others. Similarly, MissWhite lysterase had a wide range of bactericidal activity against these clinical isolates. GD95 was the most susceptible to MissWhite (T50: 32 min), while GD91 was challenging to kill by this lysterase. The MissWhite and D29 lysterases, in addition to killing the largest number of strains, also killed the fastest, with average T50 values across all susceptible strains of 81 and 90 min, respectively. In contrast, the phiGD57-1 enzyme was the slowest enzyme, with an average T50 of 129 min, and the least effective, as 14 clinical isolates were resistant to this enzyme. The performance of Bxb1, Constella, and Bongo-derived lysterases fell between these two extremes: more than half of the strains were killed by each enzyme, with average T50 values of 103–117 min. These experiments demonstrate that a diverse panel of M. abscessus clinical isolates is susceptible to killing by phage-encoded lysterases.
Lysterases and amikacin act synergistically
Because lysterases and commonly used antibiotics for NTM infections act on different targets, it is plausible that they act synergistically, especially as even relatively low (and sub-lethal) levels of lysterases could weaken the cell wall and facilitate antibiotic access. To test this, we examined the impact of combining the D29 lysterase with amikacin over a range of concentrations of each (Fig. 6A). Synergy is observed, and reductions in viability are seen at concentrations where neither agent alone kills efficiently. The fractional inhibitory concentration index (FICI) of 0.31 is strongly indicative of synergy. A similar behavior was observed when combining Bxb1 lysterase and amikacin (Fig. 6B). In contrast, when D29 lysterase and Bxb1 lysterase were combined, only an additive effect (FIC 0.75) was seen (Fig. 6C).
Fig 6.

Lysterases act synergistically with amikacin. The effect of combining lysterases with amikacin with M. abscessus ATCC19977 is shown using a resazurin microtiter assay in which different concentrations (in µm) of (A) D29 lysterase and amikacin, (B) Bxb1 lysterase and amikacin, or (C) a combination of D29 and Bxb1 lysterases were tested. Each concentration range is twofold, and the end concentrations are indicated. The well showing inhibition with the lowest concentration of each drug upon combination is marked with yellow circle, and the corresponding fractional inhibitory concentration index (FICI) of that well is shown below. FICI below 0.5 indicates synergistic interaction, while FICI between 0.5 and 1.0 indicates an additive effect. Another biologically independent replicate of the experiment is shown in Fig. S4. Table S3 shows FIC of every well.
DISCUSSION
Mycobacteriophage lysterases exhibit extensive sequence diversity while maintaining a conserved catalytic core. Although most lysterase groups encode a single dominant structural form, there is substantial variation in sequence and length. Most prior studies determining antimycobacterial activity have focused on the D29 lysterase, and perchance, this was one of the broadest-acting lysterases against M. abscessus isolates. Nonetheless, it was not the most effective lysterase for all M. abscessus clinical isolates, and some (e.g., GD100B and GD180, Fig. 5C) were more effectively killed at shorter times by other lysterases such as MissWhite. The range of activities we observed suggests that cocktails or small combinations of lysterases may be more clinically useful than a single preparation.
Although there is substantial variation in the response of different clinical isolates to the lysterases, this does not appear to be systematic in a way that correlates with either M. abscessus subspecies, colony morphotype, or dominant circulating clone (Table 2). This is an important observation as it suggests that the variation in sensitivities can be largely accommodated by using lysterase combinations without any specific subset of strains. It is plausible that there is M. abscessus diversity that is not encompassed within the tested strains and that there are additional dominant circulating clones in other globally distributed geographies. Even if these were not as susceptible to the most active enzymes tested here, the broad range of available lysterases suggests that this potential limitation could be readily overcome.
Here, we only examined the activities of six purified lysterase enzymes, which were chosen to reflect the overall diversity of mycobacteriophage lysterases. But the variations in lysterase sequence, length, and structure suggest that additional lysterases might be worth evaluating experimentally, as they could have desirable attributes, such as stability, rapid killing of both growing and non-growing mycobacteria, and broad activity against clinically relevant isolates. The variations include distinct A/B subgroups possessing localized insertions, extensions, or additional structural elements outside the conserved catalytic domain, and subgroups spanning multiple phage clusters likely arising from modular exchange.
Lysterase bactericidal activities were observed against both growing and non-growing M. smegmatis, which is noteworthy, as prior reports showed that endolysins kill log-phase Salmonella cells more effectively than stationary-phase bacteria (40), and that endolysin activity against stationary Bacillus anthracis is somewhat enzyme-dependent (41). We note that during Mycobacterium infections, non-growing mycobacterial cells pose substantial challenges to pathogen clearance as they are typically unresponsive to most antibiotic therapies (42, 43). The ability of lysterases to kill growing and non-growing bacteria suggests a potential treatment advantage.
Although we only examined lysterase-antibiotic synergies on one strain and with one antibiotic, the observations are promising (Fig. 6). For example, addition of the D29 lysterase or Bxb1 lysterase at only 0.06 µM shows strong activity in combination with ~27 or ~14 nM amikacin respectively, whereas those concentrations of amikacin or lysterase alone show little to no activity (Fig. 6). This is also a key finding, as many clinical isolates are resistant to amikacin and display high MICs. The synergy provided by the lysterase-amikacin synergy could rejuvenate broader use of amikacin to treat M. abscessus infections.
In summary, mycobacteriophage-encoded lysterase enzymes show promise as enzybiotic candidates for control of NTM infections. The variations in sensitivities of M. abscessus clinical isolates can be circumvented by exploring the diverse landscape of lysterase enzymes, as there are many pockets of this diversity yet to be evaluated. Lysterases may not be suitable for intravenous administration but could be effectively applied topically, including in bone, joint, and soft tissue infections. Amikacin is commonly administered by aerosolization of a liposome preparation, and the inclusion of lysterases in such treatments could enhance antimycobacterial activity through synergistic action (44).
MATERIALS AND METHODS
Identification of lysterase proteins
Lysterase (LysB) protein sequences were extracted from the Actino_Draft MySQL database (Phamerator database; version 607; accessed from https://databases.hatfull.org/), using custom Python scripts. Genes annotated as “lysin B” or related variants were retrieved from the gene and phage tables, and only phages infecting hosts within the genus Mycobacterium were retained. Predicted amino acid sequences were exported in FASTA format for downstream analyses. We also included lysterases from seven M. abscessus prophages. The complete list of proteins is shown in Table S1.
Sequence grouping and architectural classification
Protein sequences were grouped using MMseqs2 (36) v15.6f452 (linclust algorithm) with a minimum sequence identity threshold of 50% and 80% alignment coverage to define broad lysterase families. Groups were subsequently evaluated for architectural heterogeneity using protein length distributions. Groups showing substantial length dispersion (range ≥ 40 aa or interquartile range ≥ 25 aa; minimum cluster size ≥ 5) were examined for multimodal structure. Subdivision into architectural subgroups required ≥60 aa separation between subgroup median lengths and improved silhouette support (Δ ≥ 0.03). Very small subgroups (<5 sequences or <15% of the parent cluster) were collapsed to avoid over-partitioning.
Representative selection and structural modeling
For each final group or architectural subgroup, a representative sequence was selected based on median protein length. Representative proteins were structurally modeled using AlphaFold2 (37). Structural alignments were generated in ChimeraX v.1.11.1 using MatchMaker, and pairwise sequence alignments were used to identify conserved and subclass-specific regions. Structural visualizations were generated using a common orientation based on the conserved catalytic domain to facilitate comparison among subgroup pairs.
Structural alignment and phylogenetic inference
Predicted structures were aligned using FoldMason v4.dd3c235 (38) in easy-msa mode to generate a structure-based multiple sequence alignment. The Newick tree produced by FoldMason was used as the structure-derived tree for downstream comparison.
The FoldMason-derived amino acid alignment was used as input for maximum-likelihood inference in IQ-TREE 2 v2.3.6 with automatic model selection (ModelFinder). Branch support was assessed using 1,000 ultrafast bootstraps and 1,000 SH-aLRT replicates.
Comparison of structure- and sequence-derived trees
Structural and amino acid trees were midpoint-rooted, converted to binary where necessary, and ladderized. Trees were converted to dendrogram objects using cophenetic distances and average-linkage hierarchical clustering to enable direct comparison. Tanglegram visualization was performed in RStudio v2025.05.1 using dendextend v1.19.1 with topology-preserving untangling (step2side method; fixed random seed). Group coloring was defined by cutting the structural dendrogram at a relative height of 0.25.
Cloning, lysterase expression, and purification
Lysterase genes (Bxb1 gp9, Bongo gp38, Constella gp41, D29 gp12, MissWhite gp11, and phiGD57-1 gp25) were cloned into pET21a(+) with a C-terminal 6xHis-tag (#69740, Sigma-Aldrich) resulting in plasmids pKF40, pKF27, pKF31, pLAM3, pKF36, and pKF39, respectively. Each plasmid was transformed into Rosetta 2(DE3) Competent Cells (#71397, Sigma-Aldrich) via heat shock, and transformants were selected on LB plates (#240110, Difco), containing 50 µg/mL carbenicillin and 15 µg/mL chloramphenicol. Individual colonies were picked and inoculated into LB broth (#240230, Difco) starter culture, which was incubated at 37°C overnight. Expression cultures using ZYP-5052 media (45) containing the aforementioned antibiotics were inoculated at a 1:100 ratio. Auto-induced cultures were grown shaking at 17°C until the cell density reached an OD600 of 10, at which point the cells were pelleted at 5,000 × g for 10 min. Approximately 5 g of the pellet was frozen at −80°C to be used for protein extraction and purification.
Cell pellets were thawed on ice and incubated at room temperature for 30 min in lysis buffer (50 mM Tris pH 8.5, 500 mM NaCl, 10% glycerol, 2 mM β-ME, 10 mM imidazole, and 1 mg/mL lysozyme) prior to homogenization (EmulsiFlex – C3, Avestin) and extraction of the crude sample. Protein purification was carried out using HisPur Ni-NTA Magnetic Beads (#88831, ThermoFisher). After washing unbound materials using varied concentrations of salt (0.2-1 M NaCl), His-tagged lysterases were eluted using elution buffer (20 mM Tris pH 8.5, 20 mM NaCl, 10% glycerol, 2 mM β-ME, and 500 mM Imidazole pH 8). Subsequently, samples were further purified on the HiTrap Q HP column (#17115401, Cytiva), with a salt gradient ranging from 20 mM to 1 M NaCl Tris buffer (50 mM Tris pH 8.5, 10% glycerol, 2 mM β-ME, and varying NaCl concentration). Purified samples were dialyzed using a 10K MWCO Slide-A-Lyzer Dialysis Cassette Kit (#66807, ThermoFisher) against a storage buffer (PBST; PBS pH 7.4, 0.05% Tween-80, and 50% glycerol). The purified proteins were stored at −20°C. Heat-denatured samples were collected, loaded into 4%–20% Mini-PROTEAN TGX Precast Protein Gels (#4561094, BioRad), and electrophoresed at 250 V for 15 min. Gels were subsequently stained with Imperial Protein Stain (#24615, ThermoFisher) and destained with deionized water, according to the manufacturer’s protocol. The stained protein bands were visualized under a GelDoc system (Universal Hood II, Bio-Rad) and Image Lab software (v 3.0 build 11).
Mycobacterium growth conditions
M. abscessus clinical strains were obtained from the Hatfull lab collection. Selected strains were cultured in Middlebrook 7H9 medium supplemented with 10% Oleic Acid-Albumin-Dextrose (OADC) solution and 0.05% Tween-80. The cultures were grown at 37°C with orbital shaking for 4–5 days. The saturated cultures were sonicated using Qsonica sonicators at 500 W and 20 kHz in 15-s pulses with 10-s rests, for a total of 4 min of sonication before each use. To determine cell density, the cultures were 10-fold serially diluted in Phosphate Buffered Saline with 0.05% Tween-80 (PBST), and 3 µL of each dilution was spotted on 7H10 agar. The cells were grown at 37°C for 7–10 days before the concentration of viable cells was calculated.
M. smegmatis mc2155 cultures were grown to saturation from a single colony and were incubated for 3 days at 37°C in Middlebrook 7H9 supplemented with albumin-dextrose (ADC), carbenicillin (CB; 50 µg/mL), cycloheximide (CHX; 10 µg/mL), and Tween-80 (0.05%). Saturated cultures were subcultured 1:10 into 7H9/ADC/Tween-80 and incubated for ~24 h until OD600 reached 0.4–0.8 absorbance (abs).
Lysterase activity unit determination
In biological triplicate, liquid cultures of M. smegmatis mc2155 were grown to mid-log phase (OD600 = 0.4–0.7 abs), and incubated with seven concentrations of each lysterase, with shaking (300 rpm) at 37°C, for 1 h. After incubation, the mixtures were diluted 100-fold, and 100 µL of the diluted reaction was spread onto 7H10/ADC/CB/CHX plates and incubated at 37°C for 4 days. Titration curves were generated for all enzymes, and each biological replicate and fit to an asymmetric, five-parameter curve in GraphPad Prism (v10.5.0), with S = 1, Bottom = 2.7, Top = 5.1–5.4 (average of the controls for each replicate), and Hill Slope constrained to be >−5. Based on the titration curves, the EC50 for each enzyme was determined, and the activity unit for each enzyme was set equal to 2 × EC50.
Microscopy
In a total volume of 200 µL, log-phase M. smegmatis (5 x 107 CFU) was treated with 19 units of Bxb1 lysterase in PBST for 1 h at 37°C with 250 rpm orbital shaking (Fisher Inc #02217759). PBST and cells only were used as negative controls. Following incubation, samples were centrifuged at 17,000 × g for 1 min, and the supernatants were removed. A modified protocol was used to stain live and dead cells with Syto 9 (green) and propidium iodide (PI; red) supplied by ThermoFisher #L7012. The cell pellet was stained with 100 µL of a staining solution (1 mL of Tris buffer saline (TBS) pH 7.5, 1.5 µL of 3.34 mM Syto 9, and 1.5 µL of 20 mM PI) for 10 min at RT in the dark. Subsequently, the sample was centrifuged at 17,000 × g for 1 min to remove excess dye and resuspended in 20 µL of TBS. The stained sample was visualized and imaged using a Nikon A1R confocal microscope with a 60 × 1.4 numerical aperture oil-immersion objective. Both live and dead cells stained with Syto 9 and PI were excited with the 488 nm laser for both red and green channels with a pinhole size of 1.0 Airy unit. Then, 1,024 × 1,024-pixel images were obtained from NIS-Elements AR software v5.21.03. Dead and live cells are displayed in red and green, respectively.
Enzyme activities
Log-phase M. smegmatis mc2155 cultures were diluted to 106 CFU/mL in PBST. Cells were mixed with one activity unit of lysterase isolated from phages D29, MissWhite, Bxb1, Constella, Bongo, and phiGD57-1, as well as controls consisting of cells mixed with PBST, and were prepared in duplicate in a 96-well plate. Mixtures were incubated at 37°C, shaking at 300 rpm, and aliquots were removed at predetermined time points (0, 7, 14, 21, 28, 35, 42, 72, 102, and 132 min). Aliquots were removed and diluted in PBST at a 1:10 ratio. Dilutions (100 µL) were spread onto large 7H10/ADC plates, using glass beads. Plates were incubated at 37°C for 5 days, imaged in GelDoc system (Universal Hood II, Bio-Rad) with Image Lab software (v 3.0 build 11), and analyzed in CellProfiler v4.2.8 (46). Lysterase kinetics experiments were performed in triplicate.
Twenty-four M. abscessus saturated cultures were diluted to 106 CFU/mL in PBST. In a 96-well plate, the cells were treated with one unit of lysterase derived from phages D29, MissWhite, Bxb1, Constella, Bongo, and phiGD57-1, with cells and PBST only as a growth control. The samples were incubated at 37°C with shaking. During the incubation period, 1 μL of each sample was stamped on 7H10 agar at each time point (0, 15, 30, 45, 70, 120, and 240 min) using a 96-pin replicator. The plates were incubated at 37°C for 12 days. The area of bacterial growth was determined using CellProfiler v4.2.8 (46). GraphPad Prism was used to generate the curves and to calculate the T50 using Richard’s five-parameter dose-response equation (47).
CellProfiler processing
Plate images (.scn) taken on GelDoc system were input into an automated image analysis pipeline modified from yeast colony classification on CellProfiler v4.2.8 (46). Masked plates were analyzed for colonies within pixel ranges of 7-80 units and filtered based on the maximum measurement value. Identified colonies were numbered and outlined, and a summarizing .csv file was produced. The .csv output from CellProfiler was converted to Excel format, and the number of colonies identified by the pipeline was manually checked against the image of the plate and adjusted when colonies were incorrectly identified. Colony counts were converted to CFU/mL, and plates with no detected colonies were scored at the limit of detection. The limit of detection was set at one-half of the CFU/mL for a colony count equal to one. The CFU/mL was calculated for each plate and plotted on GraphPad Prism.
REMA-based measurement of lysterase activity in combination with amikacin
Synergistic activities were tested by the two-dimensional dilution checkerboard method using a resazurin microtiter assay (REMA). Briefly, a total of 50 µL of PBST was dispensed into each well of a 96-well microdilution plate. Amikacin (AMK) was serially diluted by twofold in the indicated concentration range in 50 µL of PBS and added to each well along the column of the plate. Lysterase from the indicated phage was similarly twofold serially diluted in the indicated concentration range in 50 µL of PBST and added along the row of the plate. As necessary, AMK was replaced with another lysterase to evaluate the combination of two lysterases. Lastly, 50 µL of PBST containing 5 × 104 CFU/well M. abscessus ATCC19977 was added to each well. This density was typically obtained as follows. An overnight culture of M. abscessus strain grown in Middlebrook 7H9/OADC/Tween-80 medium was harvested by centrifugation at 2,500 × g, washed twice with PBST, and adjusted to an OD600 of 0.3 (~1 × 107 CFU/mL) with PBST. The density was further diluted by 10-fold, and 50 µL of this suspension was added to each well. The plate was incubated at 37°C for 3 h in a shaking incubator at 100 rpm. After incubation, 50 µL of Middlebrook 7H9 medium supplemented with 10% OADC and 17 µL of 0.05% (wt/vol) resazurin solution were added to each well, bringing the final volume to 167 µL. The plates were further incubated at 37°C for 3 days. A set without any drug was set up as positive control, while another set without any cells was set up as negative control. If necessary, sterile water was added to the empty peripheral wells to prevent evaporation. Fluorescence of each well was measured using a spectrofluorometer at Ex of 530 nm/Em of 590 nm setting. MIC for amikacin or lysterase was individually determined as the concentration at which the fluorescence value decreased by 90% or more below the positive control. The fractional inhibitory concentration index (FICI) of a given well containing amikacin and lysterase was calculated with the following equation:
ΣFIC = (MIC of drug A in combination/MIC of drug A alone) + (MIC of drug B in combination/MIC of drug B alone)
FICI values of ≤0.5, >0.5–1.0, >1.0–4.0, and >4.0 were interpreted as “synergistic,” “additive,” “indifference,” and “antagonism,” respectively (48). All experiments were performed in at least two biologically independent sets.
ACKNOWLEDGMENTS
The authors thank Colin Lewis for excellent technical assistance, Chaowei Shang for confocal microscopy training and technical assistance, Ching-Chung Ko for reviewing the manuscript, and Crystal Petrone for administrative work that supports this project. A.K.S. acknowledges support from the ICMR-DHR Long-Term International Fellowship for Young Indian Biomedical Scientists (2023–24), Indian Council of Medical Research (ICMR), Government of India.
This work was supported by grants from the National Institutes of Health (GM131729) and from the Howard Hughes Medical Institute (GT17580) to G.F.H.; from the National Institutes of Health (AI187284) to A.K.O.; and from the National Institutes of Health (AI173544) to K.G.F.
Conceptualization: K.G.F., G.F.H.; Data curation: J.M., E.D.A., M.T.M., C.M., R.A.G., A.K.S., P.G., and A.K.O.; Formal analysis: J.M., K.G.F., E.D.A., M.T.M., C.M., R.A.G., A.K.S., P.G., and A.K.O.; Funding acquisition: K.G.F., G.F.H., and A.K.O.; Investigation: J.M., K.G.F., E.D.A., J.M.K., E.S., M.V., M.T.M., C.M., R.A.G., and A.K.S.; Methodology: J.M., K.G.F., R.A.G., A.K.S., P.G., and A.K.O.; Project administration: J.M., K.G.F., and G.F.H.; Resources: A.K.O. and G.F.H.; Software: C.M. and R.A.G.; Supervision: G.F.H.; Validation: J.M., K.G.F., and G.F.H.; Visualization: J.M., E.D.A., C.M., R.A.G., and A.K.S.; Writing – original draft: J.M., E.D.A., M.T.M., R.A.G., and A.K.S.; Writing – review and edits: J.M., K.G.F., E.D.A., J.M.K., E.S., M.V., M.T.M., C.M., R.A.G., A.K.O., P.G., A.K.O., and G.F.H.
Contributor Information
Graham F. Hatfull, Email: gfh@pitt.edu.
Patricia A. Champion, University of Notre Dame, Notre Dame, Indiana, USA
DATA AVAILABILITY
Phage genome sequences for Bongo, Bxb1, Constella, D29, MissWhite, and phiGD57-1 are available in GenBank, and the accession numbers are shown in Table 1.
SUPPLEMENTAL MATERIAL
The following material is available online at https://doi.org/10.1128/jb.00410-26.
Structural comparisons of representative lysterase A/B subgroups.
Biological replicates of M. smegmatis killing kinetics.
Replicates of the clinical isolate killing kinetics screen.
Clinical isolate cell titration after the lysterase treatment.
Effect of combining lysterases and amikacin.
Lysterase groupings.
Yield of purified lysterases.
Plate readings to determine FIC values.
ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.
REFERENCES
- 1. Adekambi T, Sassi M, van Ingen J, Drancourt M. 2017. Reinstating Mycobacterium massiliense and Mycobacterium bolletii as species of the Mycobacterium abscessus complex. Int J Syst Evol Microbiol 67:2726–2730. doi: 10.1099/ijsem.0.002011 [DOI] [PubMed] [Google Scholar]
- 2. Olivier KN, Weber DJ, Wallace RJ, Faiz AR, Lee JH, Zhang Y, Brown-Elliot BA, Handler A, Wilson RW, Schechter MS, Edwards LJ, Chakraborti S, Knowles MR, Cystic Fibrosis Study G. 2003. Nontuberculous mycobacteria. I: multicenter prevalence study in cystic fibrosis. Am J Respir Crit Care Med 167:828–834. doi: 10.1164/rccm.200207-678OC [DOI] [PubMed] [Google Scholar]
- 3. Honda JR, Virdi R, Chan ED. 2018. Global environmental nontuberculous Mycobacteria and their contemporaneous man-made and natural niches. Front Microbiol 9:2029. doi: 10.3389/fmicb.2018.02029 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Kavaliunaite E, Harris KA, Aurora P, Dixon G, Shingadia D, Muthialu N, Spencer H. 2020. Outcome according to subspecies following lung transplantation in cystic fibrosis pediatric patients infected with Mycobacterium abscessus. Transpl Infect Dis 22:e13274. doi: 10.1111/tid.13274 [DOI] [PubMed] [Google Scholar]
- 5. Degiacomi G, Sammartino JC, Chiarelli LR, Riabova O, Makarov V, Pasca MR. 2019. Mycobacterium abscessus, an emerging and worrisome pathogen among cystic fibrosis patients. Int J Mol Sci 20:5868. doi: 10.3390/ijms20235868 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Renna M, Schaffner C, Brown K, Shang S, Tamayo MH, Hegyi K, Grimsey NJ, Cusens D, Coulter S, Cooper J, Bowden AR, Newton SM, Kampmann B, Helm J, Jones A, Haworth CS, Basaraba RJ, DeGroote MA, Ordway DJ, Rubinsztein DC, Floto RA. 2011. Azithromycin blocks autophagy and may predispose cystic fibrosis patients to mycobacterial infection. J Clin Invest 121:3554–3563. doi: 10.1172/JCI46095 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Nguyen M-V, Haas MK, Kasperbauer SH, Calado Nogueira de Moura V, Eddy JJ, Mitchell JD, Khare R, Griffith DE, Chan ED, Daley CL. 2024. Nontuberculous mycobacterial pulmonary disease: patients, principles, and prospects. Clin Infect Dis 79:e27–e47. doi: 10.1093/cid/ciae421 [DOI] [PubMed] [Google Scholar]
- 8. Dartois V, Dick T. 2024. Toward better cures for Mycobacterium abscessus lung disease. Clin Microbiol Rev 37:e0008023. doi: 10.1128/cmr.00080-23 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Dedrick R.M, Guerrero-Bustamante CA, Garlena RA, Russell DA, Ford K, Harris K, Gilmour KC, Soothill J, Jacobs-Sera D, Schooley RT, Hatfull GF, Spencer H. 2019. Engineered bacteriophages for treatment of a patient with a disseminated drug-resistant Mycobacterium abscessus. Nat Med 25:730–733. doi: 10.1038/s41591-019-0437-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Dedrick RM, Smith BE, Cristinziano M, Freeman KG, Jacobs-Sera D, Belessis Y, Whitney Brown A, Cohen KA, Davidson RM, van Duin D, et al. 2023. Phage therapy of Mycobacterium infections: compassionate use of phages in 20 patients with drug-resistant mycobacterial disease. Clin Infect Dis 76:103–112. doi: 10.1093/cid/ciac453 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Hatfull GF. 2025. All the world’s a phage. Proc Natl Acad Sci USA 122:e2523344122. doi: 10.1073/pnas.2523344122 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Jordan TC, Burnett SH, Carson S, Caruso SM, Clase K, DeJong RJ, Dennehy JJ, Denver DR, Dunbar D, Elgin SCR, et al. 2014. A broadly implementable research course in phage discovery and genomics for first-year undergraduate students. mBio 5:e01051-13. doi: 10.1128/mBio.01051-13 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Russell DA, Hatfull GF. 2017. PhagesDB: the actinobacteriophage database. Bioinformatics 33:784–786. doi: 10.1093/bioinformatics/btw711 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Hatfull GF. 2022. Mycobacteriophages: from petri dish to patient. PLoS Pathog 18:e1010602. doi: 10.1371/journal.ppat.1010602 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Cahill J, Young R. 2019. Phage lysis: multiple genes for multiple barriers. Adv Virus Res 103:33–70. doi: 10.1016/bs.aivir.2018.09.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Young R. 2014. Phage lysis: three steps, three choices, one outcome. J Microbiol 52:243–258. doi: 10.1007/s12275-014-4087-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Holt A, Cahill J, Ramsey J, Martin C, O’Leary C, Moreland R, Maddox LT, Galbadage T, Sharan R, Sule P, Cirillo JD, Young R. 2021. Phage-encoded cationic antimicrobial peptide required for lysis. J Bacteriol 204:JB00214-21. doi: 10.1128/JB.00214-21 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Murray E, Draper LA, Ross RP, Hill C. 2021. The advantages and challenges of using endolysins in a clinical setting. Viruses 13:680. doi: 10.3390/v13040680 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Pastagia M, Schuch R, Fischetti VA, Huang DB. 2013. Lysins: the arrival of pathogen-directed anti-infectives. J Med Microbiol 62:1506–1516. doi: 10.1099/jmm.0.061028-0 [DOI] [PubMed] [Google Scholar]
- 20. Hermoso JA, García JL, García P. 2007. Taking aim on bacterial pathogens: from phage therapy to enzybiotics. Curr Opin Microbiol 10:461–472. doi: 10.1016/j.mib.2007.08.002 [DOI] [PubMed] [Google Scholar]
- 21. Nelson D, Loomis L, Fischetti VA. 2001. Prevention and elimination of upper respiratory colonization of mice by group A streptococci by using a bacteriophage lytic enzyme. Proc Natl Acad Sci USA 98:4107–4112. doi: 10.1073/pnas.061038398 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Zuber B, Chami M, Houssin C, Dubochet J, Griffiths G, Daffé M. 2008. Direct visualization of the outer membrane of mycobacteria and corynebacteria in their native state. J Bacteriol 190:5672–5680. doi: 10.1128/JB.01919-07 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Hoffmann C, Leis A, Niederweis M, Plitzko JM, Engelhardt H. 2008. Disclosure of the mycobacterial outer membrane: cryo-electron tomography and vitreous sections reveal the lipid bilayer structure. Proc Natl Acad Sci USA 105:3963–3967. doi: 10.1073/pnas.0709530105 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Garcia M, Pimentel M, Moniz-Pereira J. 2002. Expression of Mycobacteriophage Ms6 lysis genes is driven by two sigma(70)-like promoters and is dependent on a transcription termination signal present in the leader RNA. J Bacteriol 184:3034–3043. doi: 10.1128/JB.184.11.3034-3043.2002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Payne K, Sun Q, Sacchettini J, Hatfull GF. 2009. Mycobacteriophage Lysin B is a novel mycolylarabinogalactan esterase. Mol Microbiol 73:367–381. doi: 10.1111/j.1365-2958.2009.06775.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Ojha AK, Trivelli X, Guerardel Y, Kremer L, Hatfull GF. 2010. Enzymatic hydrolysis of trehalose dimycolate releases free mycolic acids during mycobacterial growth in biofilms. J Biol Chem 285:17380–17389. doi: 10.1074/jbc.M110.112813 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Gil F, Grzegorzewicz AE, Catalão MJ, Vital J, McNeil MR, Pimentel M. 2010. Mycobacteriophage Ms6 LysB specifically targets the outer membrane of Mycobacterium smegmatis. Microbiology (Reading) 156:1497–1504. doi: 10.1099/mic.0.032821-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Li J, Li J, Zhang Z, Zeng C, Liu X, Wu H, Wu H. 2026. Exploring the molecular properties and bactericidal potential of Lysin B from mycobacteriophage WXIN. BMC Microbiol 26:174. doi: 10.1186/s12866-025-04683-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Subramanyam B, Sivaramakrishnan GN, Dusthackeer A, Nagamiah S, Kumar V. 2012. Phage lysin as a substitute for antibiotics to detect Mycobacterium tuberculosis from sputum samples with the BACTEC MGIT 960 system. Clin Microbiol Infect 18:497–501. doi: 10.1111/j.1469-0691.2011.03601.x [DOI] [PubMed] [Google Scholar]
- 30. Abouhmad A, Korany AH, Grey C, Dishisha T, Hatti-Kaul R. 2020. Exploring the enzymatic and antibacterial activities of novel mycobacteriophage lysin B enzymes. Int J Mol Sci 21:3176. doi: 10.3390/ijms21093176 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Singh AK, Gangakhedkar R, Thakur HS, Raman SK, Patil SA, Jain V. 2023. Mycobacteriophage D29 Lysin B exhibits promising anti-mycobacterial activity against drug-resistant Mycobacterium tuberculosis. Microbiol Spectr 11:e04597-22. doi: 10.1128/spectrum.04597-22 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Fraga AG, Trigo G, Murthy RK, Akhtar S, Hebbur M, Pacheco AR, Dominguez J, Silva-Gomes R, Gonçalves CM, Oliveira H, Castro AG, Sharma U, Azeredo J, Pedrosa J. 2019. Antimicrobial activity of mycobacteriophage D29 lysin B during Mycobacterium ulcerans infection. PLoS Negl Trop Dis 13:e0007113. doi: 10.1371/journal.pntd.0007113 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Hurst-Hess K, Walz A, Yang Y, McGuirk H, Gonzalez-Juarrero M, Hatfull GF, Ghosh P, Ojha AK. 2023. Intrapulmonary treatment with Mycobacteriophage LysB rapidly reduces Mycobacterium abscessus burden. Antimicrob Agents Chemother 67:e0016223. doi: 10.1128/aac.00162-23 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Abouhmad A, Kassaliete J, Davids C, Grimon D, Dishisha T, Abdelkader K, Eldin ZE, Clarsund M, Briers Y, Godaly G, Hatti-Kaul R. 2026. Engineered phage-derived lysins effectively kill mycobacterial pathogens. Trends Biotechnol 44:2067–2101. doi: 10.1016/j.tibtech.2026.02.015 [DOI] [PubMed] [Google Scholar]
- 35. James S, Nair G, Nath P, Jain V. 2026. Concentration-dependent dual activity of a phage-encoded lysin modulates infection dynamics in mycobacteriophage D29. MBio 17:e00732-26. doi: 10.1128/mbio.00732-26 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Steinegger M, Söding J. 2017. MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets. Nat Biotechnol 35:1026–1028. doi: 10.1038/nbt.3988 [DOI] [PubMed] [Google Scholar]
- 37. Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, Tunyasuvunakool K, Bates R, Žídek A, Potapenko A, et al. 2021. Highly accurate protein structure prediction with AlphaFold. Nature 596:583–589. doi: 10.1038/s41586-021-03819-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Gilchrist CLM, Mirdita M, Steinegger M. 2026. Multiple protein structure alignment at scale with FoldMason. Science 391:485–488. doi: 10.1126/science.ads6733 [DOI] [PubMed] [Google Scholar]
- 39. Hatfull GF. 2023. Phage therapy for nontuberculous mycobacteria: challenges and opportunities. Pulm Ther 9:91–107. doi: 10.1007/s41030-022-00210-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Oliveira H, Thiagarajan V, Walmagh M, Sillankorva S, Lavigne R, Neves-Petersen MT, Kluskens LD, Azeredo J. 2014. A Thermostable Salmonella Phage Endolysin, Lys68, with Broad Bactericidal Properties against Gram-Negative Pathogens in Presence of Weak Acids. PLoS One 9:e108376. doi: 10.1093/nar/gku734 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Paskaleva EE, Mundra RV, Mehta KK, Pangule RC, Wu X, Glatfelter WS, Chen Z, Dordick JS, Kane RS. 2015. Binding domains of Bacillus anthracis phage endolysins recognize cell culture age-related features on the bacterial surface. Biotechnol Prog 31:1487–1493. doi: 10.1002/btpr.2172 [DOI] [PubMed] [Google Scholar]
- 42. Ojha AK, Baughn AD, Sambandan D, Hsu T, Trivelli X, Guerardel Y, Alahari A, Kremer L, Jacobs WR, Hatfull GF. 2008. Growth of Mycobacterium tuberculosis biofilms containing free mycolic acids and harbouring drug-tolerant bacteria. Mol Microbiol 69:164–174. doi: 10.1111/j.1365-2958.2008.06274.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Dulberger CL, Guerrero-Bustamante CA, Owen SV, Wilson S, Wuo MG, Garlena RA, Serpa LA, Russell DA, Zhu J, Braunecker BJ, Squyres GR, Baym M, Kiessling LL, Garner EC, Rubin EJ, Hatfull GF. 2023. Mycobacterial nucleoid-associated protein Lsr2 is required for productive mycobacteriophage infection. Nat Microbiol 8:695–710. doi: 10.1038/s41564-023-01333-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Rose SJ, Neville ME, Gupta R, Bermudez LE. 2014. Delivery of aerosolized liposomal amikacin as a novel approach for the treatment of nontuberculous mycobacteria in an experimental model of pulmonary infection. PLoS One 9:e108703. doi: 10.1371/journal.pone.0108703 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Studier FW. 2005. Protein production by auto-induction in high density shaking cultures. Protein Expr Purif 41:207–234. doi: 10.1016/j.pep.2005.01.016 [DOI] [PubMed] [Google Scholar]
- 46. Stirling DR, Swain-Bowden MJ, Lucas AM, Carpenter AE, Cimini BA, Goodman A. 2021. CellProfiler 4: improvements in speed, utility and usability. BMC Bioinformatics 22:433. doi: 10.1186/s12859-021-04344-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Di Veroli GY, Fornari C, Goldlust I, Mills G, Koh SB, Bramhall JL, Richards FM, Jodrell DI. 2015. An automated fitting procedure and software for dose-response curves with multiphasic features. Sci Rep 5:14701. doi: 10.1038/srep14701 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Taneja NK, Tyagi JS. 2007. Resazurin reduction assays for screening of anti-tubercular compounds against dormant and actively growing Mycobacterium tuberculosis, Mycobacterium bovis BCG and Mycobacterium smegmatis. J Antimicrob Chemother 60:288–293. doi: 10.1093/jac/dkm207 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Structural comparisons of representative lysterase A/B subgroups.
Biological replicates of M. smegmatis killing kinetics.
Replicates of the clinical isolate killing kinetics screen.
Clinical isolate cell titration after the lysterase treatment.
Effect of combining lysterases and amikacin.
Lysterase groupings.
Yield of purified lysterases.
Plate readings to determine FIC values.
Data Availability Statement
Phage genome sequences for Bongo, Bxb1, Constella, D29, MissWhite, and phiGD57-1 are available in GenBank, and the accession numbers are shown in Table 1.
