Massive annotation of bacterial L-asparaginases reveals their puzzling distribution and frequent gene transfer events

Abstract

L-Asparaginases, which convert L-asparagine to L-aspartate and ammonia, come in five types, AI-AV. Some bacterial type AII enzymes are a key element in the treatment of acute lymphoblastic leukemia in children, but new L-asparaginases with better therapeutic properties are urgently needed. Here, we search publicly available bacterial genomes to annotate L-asparaginase proteins belonging to the five known types. We characterize taxonomic, phylogenetic, and genomic patterns of L-asparaginase occurrences pointing to frequent horizontal gene transfer (HGT) events, also occurring multiple times in the same recipient species. We show that the reference AV gene, encoding a protein originally found and structurally studied in Rhizobium etli, was acquired via HGT from Burkholderia. We also describe the sequence variability of the five L-asparaginase types and map the conservation levels on the experimental or predicted structures of the reference enzymes, finding the most conserved residues in the protein core near the active site, and the most variable ones on the protein surface. Additionally, we highlight the most common sequence features of bacterial AII proteins that may aid in selecting therapeutic L-asparaginases. Finally, we point to taxonomic units of bacteria that do not contain recognizable sequences of any of the known L-asparaginase types, implying that those microorganisms most likely contain new, as yet unknown types of L-asparaginases. Such novel enzymes, when properly identified and characterized, could hold promise as antileukemic drugs.

Figure 1

Abundance of l-asparaginases in bacteria. (a) The table shows the number of taxonomic groups of bacteria containing each of the five l-asparaginase types (i.e., AI, AII, AIII, AIV, and AV). For example, AI l-asparaginase is present in 13,130 species belonging to 3061 genera and 790 families. (b) The phylogenetic tree of Bacteria illustrates the presence or absence of the five types of l-asparaginases in the phylogenetic taxa. The tree encompasses 1665 bacterial families belonging to 28 most abundant phyla and covering 96% of all bacterial species. Clades of the four largest bacterial phyla (Proteobacteria, Actinobacteriota, Bacteroidota, and Firmicutes) cover 75% of species and are highlighted in the tree. The five outer rings provide information about the presence (filled with color) or absence (white) of each l-asparaginase type in a bacterial family. (c) List of the phyla shown in the tree and the percentage of species in each phylum containing enzyme types AI, AII, AIII, AIV, and AV.

Figure 2

Relation between sequence similarity of l-asparaginase proteins and phylogenetic distance between species. (a) Alignment score of orthologous l-asparaginases and phylogenetic distance separating the bacterial species. Protein sequence of each prototypic enzyme (EcAI, EcAII, EcAIII, ReAIV, and ReAV) was separately aligned to all its orthologous sequences from other bacterial species. The phylogenetic distance between bacterial species was obtained from the GTDB reference tree of bacteria. Each dot in the scatterplots represents a single comparison between a prototypic enzyme protein and an orthologous sequence from other bacterial species. (b) Fragment of the phylogenetic tree of AV proteins in bacterial species (n = 1672) showing close evolutionary relation of AV proteins between the species of Rhizobium (n = 23) and Burkholderia (n = 50). Bootstrap support values are shown on the main tree branching. (c) Global sequence alignment of AV proteins from Rhizobium bangladeshe (Rb) and Burkholderia ubonensis (Bu), with sequence identities (green), similarities (yellow), and differences (red) highlighted.

Figure 3

Putative horizontal gene transfer events of l-asparaginases in bacteria. HGT events of the five AI-AV l-asparaginase types (a–e, respectively) across 223 classes of bacteria belonging to 35 most abundant phyla. Rectangles represent classes of bacteria and mark the presence (filled with color) or absence (white) of a given l-asparaginase type in a given class. Arcs show horizontal gene transfer between two bacterial species. The height of the arcs marks the highest taxonomic rank that is different between the species (i.e., phylum, class, order, family). Arc widths are arbitrary and do not represent any taxonomic or evolutionary distance between bacteria.

Figure 4

Conservation of residues and the active sites of representative l-asparaginases. Residues are colored according to their conservation: red (highly variable: 0–30% identity), green (highly conserved, 80–100% identity), or yellow (30–80% identity). (a,b) The EcAI subunit A (a) and a covalent reaction intermediate (b) with a substrate molecule (cyan) in the active site (PDB ID: 2him). (c,d) The EcAII subunit A with the l-Asp product (cyan) bound in the active site (PDB ID: 3eca). (e,f) The EcAIII (ɑ + β)₂ homodimer with the l-Asp product (cyan) bound in the active site (PDB ID: 2zal). (g,h) A protomer of ReAIV predicted by the Robetta server with a detailed view (h) of the residues in the putative active site; residues potentially involved in Zn²⁺ coordination are marked by blue circles; the predicted S–S bridge between Cys188 (putative metal coordination ligand) and Cys106 that might be formed in the absence of a metal cation is marked by a yellow arrow. (i,j) The ReAV homodimer (i) and the active site (j) with the Zn²⁺ ion (dark blue sphere) coordinated close to the nucleophilic Ser48 (PDB ID: 7os5). In all panels, the nucleophilic residue (Thr or Ser) is conserved and colored light green.

Figure 5

Sequence characteristics of l-asparaginase domain in bacterial AII proteins. (a) Amino acid conservation along the l-asparaginase domain between orthologs and the reference EcAII protein. The catalytic residues are marked by blue bars and substrate-binding residues are shown in yellow. The bar charts show identity percentage, gaps percentage and information content (IC) at each site. (b) Position-specific score matrix (PSSM) calculated with reference to EcAII, showing how often a given residue is found at a specific position within the domain. Preferred residues (occurring more often than their expected frequency) are shown in green and avoided amino acids (occurring less often than their expected frequency) are shown in red. The complete PSSM profile across all l-asparaginase domain sites is shown in Table S12. (c) Most common arrangements of structurally and functionally important amino acid residues—catalytic (blue) and substrate binding (yellow)—found in bacterial AII proteins. The first two residue patterns are present in clinical drugs used to treat ALL (E. coli strain K12 and Erwinia chrysanthemi). The percentage numbers indicate the fraction of AII proteins containing a given residue pattern. Box plots (on the right) show sequence identity distribution of the full-length AII l-asparaginase domain across orthologs containing a given residue pattern. The box plot with sequence identity statistics of the whole l-asparaginase domain may be interpreted as a proxy of a dispersal range of a given motif among diverse bacteria.