Binary and quaternary cytolytic complexes of bacterial origin in which aegerolysin-like proteins are combined with larger, non-aegerolysin-like protein partner(s), described to date, include: Cry16Aa/Cry17Aa/Cbm17.1/Cbm17.2 from C. bifermentas subsp. malaysia [40]; Cry34Ab1/Cry35Ab1 from B. thuringiensis [130,134]; and AflP-1A/AflP-1B from A. faecalis [143]. In fungus P. ostreatus, PlyA forms a pore embedded in the membrane together with PlyB (Figure 1) [60]. Similar cytolytic effects were observed when PlyB was combined with other Pleurotus-derived aegerolysins, e.g., OlyA6, PlyA2 and EryA [78]. These heteromeric aegerolysin-based cytolytic complexes have been exploited as potent biopesticides for specific pests, with Cry16Aa/Cry17Aa/Cbm17.1/Cbm17.2 acting against Aedes mosquitoes, and Cry34Ab1/Cry35Ab1, AflP-1A/AflP-1B, OlyA6/PlyB, PlyA/PlyB, PlyA2/PlyB, or EryA/PlyB acting against Coleoptera species, especially the western corn rootworm.

Unexpectedly, partner proteins can be classified into five groups (Table 2, Figure 2): (1) PlyB and EryB have a similar MACPF fold; (2) the remaining models, including BlyB, showed a reasonably good superposition; BlyB has been shown to best align the structure of bacterial GNIP1Aa, another MACPF domain-containing protein [52,121]; (3) the Cry35Ab1 structure and (4) the AfIP-1B model do not superimpose with the PlyB structure or with each other; for AfIP-1B, the MACPF domain was found to be insignificant [52]; (5) Cry16Aa and Cry17Aa only superimpose with each other.

Proteins containing a membrane-attack complex/perforin (MACPF) domain are transmembrane pore-forming proteins important for both human immunity and pathogen virulence. Little is known about the function of MACPF-like domain proteins in filamentous fungi and their taxonomic distribution. The number of putative MACPF proteins in a single fungal species ranges from zero to ten or more [155]. The identification and annotation of putative MACPF-like proteins is generally more error-prone because these genes have a higher number of introns [155]. The sequences can be divided into two groups. The proteins in the first group, such as PlyB or EryB, were assigned to the MACPF domain PF01823, which was confirmed by the 13-amino acid signature Y/F-G-X₂-F/Y-X₆-G-G typical of this domain, and they are grouped with human perforin. The second group of sequences is not recognized by the Pfam tool but still contains the typical MACPF/CDC signature, such as BlyB [155]. Fungal MACPF proteins probably contribute to various specific processes. While some of them were found in secretomes (without a typical signal sequence being recognized), others are intracellular, and some of them might be involved in pathogenesis, although probably not all [155]. The large number of introns, the sporadic taxonomic distribution, and the different number of MACPF proteins per fungal species might indicate the involvement of horizontal gene transfer mechanisms in specific ecological niches [155]. Putative MACPF-like domains were identified in the genomes of fungal species with different lifestyles; some of them are pathogenic, such as plant pathogen M. perniciosa, the caterpillar fungus C. militaris, or the nematode trap fungus Arthrobotrys oligospora. Some of the species are also saprophytic, such as the white rot fungus P. ostreatus, grassland-litter decomposers P. eryngii, saprophytic and food-producing A. oryzae, or saprophytic soil fungus A. nidulans [155].

However, the binary protein partner Cry35Ab1 does not have a MACPF domain associated with it. Instead, it contains the C-terminal domain toxin 10 (PF05431), which is typical of a family of insecticidal crystal toxins of Bacillus, named after the insecticidal crystal toxin P42. Cry35Ab1 also has an additional N-terminal ricin-type β-trefoil lectin domain (PF00652) (classified as insignificant only) [52,130].

Cry16Aa and Cry17Aa are delta endotoxins composed of three distinct structural domains, endotoxin N, M, and C, respectively. An N-terminal helical bundle domain is involved in membrane insertion and pore formation (PF03945), a central β-sheet domain is involved in receptor binding (PF00555), and the C-terminal β-sandwich domain interacts with the N-terminal domain to form a channel (PF03944). During sporulation, the bacteria produce crystals of delta endotoxins. When an insect ingests these proteins, they are activated by proteolytic cleavage. For all such proteins, the N-terminus is cleaved, and for some members, a C-terminal extension is also cleaved. After activation, the endotoxin binds to the intestinal epithelium and, by the forming cation-selective channels, causes cell lysis, which leads to death [137,156].

Surprisingly, it was observed in the genome of P. ostreatus that the two genes encoding the pair of pore-forming proteins also form a bidirectional pair with 5′–5′ orientation of plyA (priA) and plyB [91,92]. Similar gene pairs encoding putative bicomponent toxins have been observed previously, such as Asp-HS with Asp-HSB, NigA2 with NigB1, and BlyA with BlyB from A. fumigatus, A. niger, and B. bassiana, respectively [52,92]. Moreover, such gene pairs have been also identified in other entomopathogenic fungi besides in B. bassiana, in the genomes of Cordiceps militaris, Metarizium acridum, M. anisopliae, M. robertsii, and Ophiocordiceps sinensis, but the involvement of these protein pairs in pore formation remains to be confirmed [52]. Here, we identified two additional putative pairs: PlyA2 with EryB from P. eryngii and L152 with L152B from A. geisen (Figure 3). The gene locations of the aegerolysin and putative MACPF-like genes in the genomes of some fungi and bacteria are schematically shown in Figure 3. A total of six fungal (putative) bicomponent toxins form a bidirectional gene pair with 5′–5′ orientation of the adjacent gene, and distances between the two genes vary (Figure 3). Clusters of genes encoding fungal secondary metabolites may also contain some reverse-oriented genes that may nevertheless show coordinate expression. In contrast to the fungal gene pairs, the AfIP-1A/AfIP-1B gene pair encoded on the bacterial chromosome has the same sense orientation (Figure 3). The bicomponent toxin gene pair Cry34Ab1/Cry34Ab1 is known to be plasmid-encoded, but the variation of the plasmid in the bacterium B. thuringiensis is exceptionally high to see if these two genes are adjacent. The Cry operon in plasmid pCryO of C. bifermentans subsp. malaysia contains four genes downstream of the promoter pCyt: Cry16Aa is located 91 base pairs (bp) downstream, followed by Cry17Aa, which is located 426 bp downstream, followed by Cbm17.1, and 1022 bp downstream, followed by Cbm17.2 [40]. Their placement at the same locus or under the control of the same promoter does not necessarily lead to their joint action, but rather indicates coordinated expression.

Some of the aegerolysins included in this review are also encoded in genomes that do not have an adjacent (MACPF domain-containing) partner protein; such aegerolysins are the fungal EryA, Aa-Pri1, Asp-HS-like, Ter, AoHlyA, and NigA1, bacterial RahU, insects P23 and the ascoviral TnAV2c gp029 (Table 2). For some of them, insufficient data are available, such as for the MpPRIA1, MpPRIA2, and MpPRIB, because the sequenced countings are too short, or no genomic data were found for GME7309 (Table 2). However, a combined action with MACPF domain-containing proteins encoded elsewhere in the genome cannot be excluded, as some aegerolysins, such as PlyA2 and EryA from P. eryngii, have been shown to exhibit cytolytic activity when combined with their non-native partner with component B, PlyB, from P. osteratus [78].