The literature in recent years has provided limited descriptions of anthrax lysins active against B. anthracis. Although new anthrax phages are being reported, their lysins are often not produced or studied, as seen in the cases of phages Negev_SA, Tavor_SA, and Carmel_SA (Alkalay et al. 2018). In our work, we reviewed the enzymatic activity of all lysins from 41 known complete B. anthracis phage genomes (https://www.ncbi.nlm.nih.gov/). Among these, the majority (31 out of 41) exhibited amidase activity, suggesting a potential trend in B. anthracis–specific lysins. The remaining lysins included two with peptidase activity, one with lysozyme activity, and five whose enzymatic activities remain unverified.

We conducted a comprehensive literature search and compiled all lysins, to the best of our knowledge, that exhibit the capability to eliminate B. anthracis cells. Due to the different protocols employed for lysin activity assays in vitro and diverse ways of presenting the results, it is hard to compare the lysins featured in this review one to one and choose the best-acting one. Nevertheless, for certain lysins (AmiBA2446, PlyB, PlyBt33, LysBC17, LysB4, PlyP56, PlyN74, and PlyTB40), biochemical characterization, including thermal stability, pH range, or salt concentration impact, was also performed, which makes it easier to choose and adjust them to the desired purposes.

In addition to phage lysins, bactericidal enzymes include lytic enzymes encoded by bacteria, called autolysins. These two groups exhibit numerous advantageous properties, as they are quite similar in many respects (Mehta et al. 2013). Notably, a well-described autolysin with established activity against B. anthracis, not included in Table 1, is worth mentioning in this review. AmiBA2446 (NP844822) is an autolysin of bacterial origin, identified in the B. anthracis Ames strain A2012 (Mehta et al. 2013). This amidase (245 aa) had potent antibacterial activity against B. anthracis (ΔSterne, Sterne 34F2), B. cereus (ATCC 4342, Frankland and Frankland 1887 AL), and B. thuringiensis subsp. kurstaki ATCC 33,679 strains, which was measured spectrophotometrically. For all these strains except for B. cereus Frankland and Frankland 1887 AL, an almost 100% killing effect was observed (Mehta et al. 2013). Another interesting example of a bacteriolytic enzyme we can recall is Ply67 (ALN97746.1) whose activity was confirmed for B. anthracis endospores (Fu et al. 2020). Ply67 (217 aa) was identified in Bacillus pumilus phage BpSp. Unlike most phage lysins, this hydrolase does not have amidase/peptidase activity but is a spore cortex-lytic enzyme. Ply67 joins some features of phage lysins and cortex-lytic activity of GSLEs (germination-specific cortex-lytic enzymes) identified in Bacillus and Clostridium species. Its amino acid sequence exhibits 55% similarity to Bacillus phage AR9 endolysin and 40% homology to the B. anthracis GSLE, SleB (Fu et al. 2020). In the phage BpSp genome, there are two genes encoding hydrolases. A product of the gene gp067, Ply67, leads to the death of spores of B. anthracis, B. cereus, and B. thuringiensis rather than their germination, causing spore surface damage and their shrinking. The gene gp019, in turn, encodes N-acetylmuramoyl-l-alanine amidase, which could be the regular phage endolysin, but so far, it has not been described in detail (Yuan and Gao 2016).

Just a few B. anthracis lysins, namely PlyB, PlyG, PlyPH, and LysB4, have been examined in animal models, with all demonstrating satisfactory outcomes in mice (Table 2). Specifically, a lysin combination (the cocktail of PlyB and PlyG) was observed to enhance efficacy, potentially reducing therapeutic doses and treatment duration (Schuch et al. 2002). Bacillus strains used in the presented in vivo studies are well-established models of anthrax. Nevertheless, a more virulent strain is desirable due to the need to reflect infection and treatment accurately. Moreover, in most discussed studies, growing vegetative cells of Bacillus were used (spores were applied only in the LysB4 antimicrobial activity test). Application of Bacillus spores would be more suitable because of its expected usage in bioterrorism attacks (Park et al. 2018). Even though the mouse model is approved as a standard, it cannot fully simulate an infection of the human body due to the differences in anatomy, physiology, and the mechanism of action of the immune system. Mice exhibit features such as a weak or absent cough reflex, a lack of bronchioles, and a proportionally larger nasal surface area with less airway branching compared to humans. Moreover, substantial differences exist in the functions and effectors of murine and human immune systems (Welkos et al. 2016). Therefore, another animal model is recommended to test the efficacy of lysins. Primates seem more appropriate as they are most related to the human species (Park et al. 2018). Also, the discussed studies did not use an aerosol challenge model, which is desirable since the inhalational route of administration reflects the natural transmission of anthrax bacilli and is most likely to be applied in potential bioterrorism incidents (Loving et al. 2007).

The comparison of amino acid sequences within the variable cell-binding domains of the 17 lysins revealed notable similarities and differences, allowing us to categorize 16 of them into three main groups (Fig. 1). The sequence of the remaining Ply67 lysin stands out due to its low similarity level. This discrepancy appears consistent with Ply67 characterization as a partially cortex-lytic protein, distinguishing it distinctly from the other lysins. In the proposed Block I and II, the SH3 domain was identified as a common CBD type. SH3b domain(s) are widely utilized by many lysins as their CBD and are found in a variety of proteins with enzymatic activity. These domains typically adopt the characteristic SH3 β-barrel fold, consisting of 5–7 β-strands. The folds often form two tightly packed antiparallel β-sheets joined by the linker (Broendum et al. 2018). The closer relatedness of the lysins from these two blocks is also evident in the phylogenetic tree, where these two blocks emerge from a single branch (Fig. 2). Amidase02_C module, in turn, appears only in phages that infect A1α or A1γ bacterial peptidoglycan chemotype; therefore, the occurrence of this domain is not surprising since the peptidoglycan A1γ-chemotype is widely distributed among Bacillaceae (Archibald et al. 1993; Vázquez et al. 2021).

Two amino acid residues in the PlyG CBD that were reported as critical for the binding to B. anthracis, L190 and Q199 (Kikkawa et al. 2007), are repeated in the most closely related lysins found within Block III (highlighted in purple in Fig. 1). This homology observed among the binding domains of PlyG, PlyPH, and AmiBA2446 may contribute to a similarity in their host range and thereby the specificity of these enzymes. Both, PlyG and PlyPH, demonstrate specific lytic activity exclusively against B. anthracis strains and a surrogate strain, B. cereus ATCC 4342 (Table 1). Additionally, AmiBA2446 exhibits antimicrobial activity against B. thuringiensis subsp. kurstaki ATCC 33,979 which shares a close genetic resemblance with B. anthracis (Mehta et al. 2013). In contrast, PlyL has only one of these critical amino acid residues, L190, which potentially accounts for variations in its host range, displaying notable activity against B. cereus, B. subtilis, and B. megaterium but lower against B. anthracis Sterne 34F2 (Low et al. 2005). However, given the differences in available data regarding the bacterial host ranges of all described lysins, direct comparisons are challenging, making it difficult to unequivocally determine whether the proposed blocks of lysins correlate with their specific host ranges.

For many phage researchers, there is a perspective that phages and their lytic enzymes can, in the future, completely replace antibiotics, especially against multi-drug-resistant bacteria (Mehmood Khan et al. 2023). Due to certain obstacles to lysin administration to patients, like possible proteolysis, methods of safe and effective delivery systems, including encapsulation, are being developed (Gondil and Chhibber 2021). Furthermore, advancements in the molecular engineering of lysins and their domains offer promising prospects for enhancing their efficacy (Seijsing et al. 2018). This suggests that native or engineered lysins could potentially serve as an alternative treatment for anthrax in animals or humans, especially in situations where the use of antibiotics is restricted or undesirable.