Characterization of toxin systems of Paenibacillus strains isolated from honeybees

Abstract

Paenibacillus is a large genus of Gram-positive bacteria isolated from a wide range of sources. Members of this genus have attracted significant research as plant growth promoters as well as pathogens of both plants and animals. Extensive studies have been conducted on Paenibacillus larvae, the causative agent of American Foul Brood in honeybees, but little is known about other closely related species associated with honeybees, such as Paenibacillus apiarius or Paenibacillus thiaminolyticus. These bacterial species have been isolated from dead honeybee larvae and are believed to be secondary invaders as opposed to primary pathogens. We utilized genome mining efforts to explore the number and diversity of toxins present in multiple species of Paenibacillus, with major emphasis on a potentially entomopathogenic clade that includes Paenibacillus alvei, P. apiarius, and P. thiaminolyticus. Our results show that strains of these species contain a variety of potentially insecticidal toxins, including highly conserved Spp-like and Mpp-like toxins. These toxins are absent in P. larvae. We also found several ADP-ribosylases which were shared between P. larvae and several strains from this entomopathogenic clade. Tc toxins, a class of insecticidal toxin not often found in Gram-positive bacteria were found in in P. apiarius and P. alvei. These findings suggest the potential for Paenibacillus species in this clade to be harnessed as sources of host specific and pollinator-friendly biopesticides.

Fig. 1

Phylogenetic Tree of Select Paenibacillus Species Reveals Potential Clade of Opportunistic Insect Pathogens. Maximum-likelihood tree was constructed using 117 Paenibacillus genomes spanning 42 species. The tree was constructed using VBCG (20 conserved proteins) run with default parameters and plotted using the ggtree library available in R. Species of potential pathogenic interest are highlighted in their own color. Brevibacillus forte was used as an outgroup and removed to conserve space.

Fig. 2

Putative Insecticidal Toxins Identified in 13 Paenibacillus Species. 117 Paenibacillus genomes were blasted against the Insecticidal Toxin Database containing 1427 curated insecticidal toxin protein sequences. Blast cutoffs were defined as follows: Identity Threshold > 40%, e-value Threshold < 1e-¹⁰, bit-score Threshold > 50. Protein sequences that met these parameters are included in the figure. Each individual rectangle (denoted by black border lines) represents an individual strain; the size of the rectangle indicates the number of toxins identified in that single strain.

Fig. 3

PCA of Putative Insecticidal Toxin Sequences. Amino acid sequences for proteins fulfilling BLAST parameters were converted into frequency tables depicting the position-specific frequency of a given amino acid at a given position in the sequence. These composition tables were then used to create the PCA. Colors of the individual points represent species identity and shapes represent toxin protein family identity.

Fig. 4

Genomic alignment of toxin clusters. (A) Genomic alignment of a subset of Spp-containing strains. The predicted Spp-encoding gene has been highlighted in a red box. (B) Genomic alignment of a subset of Mpp containing strains based on the conserved location found in P. alvei and P. apiarius. Predicted Mpp-encoding genes are highlighted in a red box. (C) Genomic alignment of same subset of Mpp-containing strains found in Panel B, but depicting the location conserved between two P. thaiminolyticus strains. Predicted Mpp-encoding genes are highlighted in a red box.

Fig. 5

Genomic alignment of Tc Toxin clusters. (A) Genomic alignment of P.apiarius and P. alvei to identify TcABC toxins. (B) Genomic alignment of P. apiarius and P. thiaminolyticus to identify TcABC toxins.(C) Genomic alignment of P. apiarius and P. larvae to identify TcABC toxins. For all panels the gene encoding TcB is highlighted by a red box.

Fig. 6

InterProScan Output of Domain Prediction in Putative P. apiarius NRS-1438 TcABC Toxin. (A) The domains of the predicted TcA protein. (B) The domains of the predicted TcB protein. (C) The domains of the predicted TcC protein.