A novel alkane monooxygenase (alkB) clade revealed by massive genomic survey and its dissemination association with IS elements

Identification of alk-gene cluster

An initial database of AlkBFGHJKL, alkN and alkST (alk-genes hereafter) was constructed using protein sequences described in the study by Van Beilen et al. (2001). All available prokaryotes genomes (both bacteria and archaea) were downloaded from NCBI (http://www.ncbi.nlm.nih.gov) on March 2, 2022, representing for 200 phyla (146 out of which were candidatus). More specifically, the genome sequences of over 390,000 isolates or MAGs from 57,335 species belonging to 3,687 genera were included in this analysis. At the same time, the taxonomy and niches information corresponding to these genomes were also downloaded. The sequences from genomes with assembly levels of “Complete” or “Chromosome” were marked as chromosomal or plasmid according to their corresponding description. The rest genome sequences were marked as “Contig/Scaffold”. Subsequently, gene prediction was performed using Prokka v1.13 (Seemann, 2014). The presence of each gene in the genome was screened using blastp with an e-value of 1e−10 (Camacho et al., 2009).

Identification of putative gene clusters was performed using a two-step method. For the first, a set of four genes, that were alkBGHJ were used as markers for identification of putative alk-fragment (DNA fragment containing alk-genes) location using criteria as follows: (1) alkB must exist, (2) alkGHJ were present within 10 kb flanking region of alkB gene, (3) sequence of DNA segments flanking the alkB gene, ranging from 10 kb upstream to 28 kb downstream of the starting codon, were extracted from genome sequence for next step. Second, alk-fragments were re-predicted using prokka (Seemann, 2014), and then screened by each gene of AlkBFGHJKL, alkN and alkST using blastp in blast package v2.9.0 (Camacho et al., 2009). All alk-fragments were verified manually and the fake ones were deleted. The alk-fragments at the end of contigs, which contain truncated gene clusters, were excluded in this analysis. The IS elements in alk-fragments were identified using ISEScan v1.7.2.2 (Xie & Tang, 2017). Organization of alk-genes and IS elements were plotted using R v4.1.2 (R Core Team, 2021) with ggplot.

Phylogeny of alk-gene cluster-carrying genomes

To build the reference phylogenetic tree for this analysis, bcgTree v1.2.0 (Ankenbrand & Keller, 2016) was performed. The bcgTree pipeline builds phylogenetic trees from bacterial core genomes, these were 107 essential genes. Briefly, gene prediction was performed by prodigal v2.6.3 (Hyatt et al., 2010), then the core genes were identified using hmmsearch in HMMER v3.2.1 (Eddy, 1998; Eddy, 2009) and aligned using muscle v3.8.31 (Edgar, 2004). The unreliable alignment regions were removed by using Gblocks v0.91b (Castresana, 2000). The resulting alignment was used to construct a maximum likelihood phylogenetic tree using the RAxML v8.2.12 (Stamatakis, 2014) with rapid bootstrap 1000. The final tree was midpoint rooted and visualized using iTOL (Letunic & Bork, 2021).

Phylogeny of the alkB genes

To infer the phylogenetic structure of the alkB gene, we manually downloaded nucleotide sequences coding alkane 1-monooxygenase from KEGG (http://www.kegg.jp) with ortholog number K00496, along with corresponding taxonomy information for each sequence. The alkB genes with identical nucleotide sequences and taxonomy compared with alkB genes in alk-gene clusters were eliminated.

The retained alkB gene sequences from KEGG (http://www.kegg.jp) and the alkB gene sequences retrieved from alk-fragment were combined for phylogenetic analysis. All the alkB gene sequences were aligned using mafft v7.407 (Katoh & Standley, 2013). And then maximum likelihood phylogenetic tree was constructed using the RAxML v8.2.12 (Stamatakis, 2014) with rapid bootstrap 1000. The xylM, which encodes the xylene monooxygenase subunit, was used as the outgroup, as described in the studies by Nie et al. (2014) and Karthikeyan et al. (2021).

Organization structure of alk-genes

The sequence fragments were extracted from the genome sequence for each alk-gene cluster, from 10kbp upstream of the alkB gene to 28kbp downstream, using inhouse perl scripts. These sequence fragments were defined as alk-fragments. The corresponding gene prediction and annotation information were extracted at the same time. The organization structure was plotted using ggplot2 and gggenes packages in R v4.1.2.

Phylogenetic diversity of genomes carrying alk-gene clusters

To capture as much diversity as possible, >390 000 genomes which include both MAGs and isolate genomes, were screened for the presence of alk-gene clusters. Totally 117 genomes (Table S1) carrying at least one set of alk-gene clusters were observed, which formed 2 distinct clades (Fig. 1). Clade A was composed of 56 strains, all in α-proteobacteria at class level, representing for 6 orders and at least 26 genera. While clade B was composed of 61 strains, all in γ-proteobacteria at class level, representing for at least 3 orders and 6 genera. Strains in genus Pseudomonas, located in clade B, had consistent and complete alk-genes sets. While all the other species had incomplete alk-genes set, more or less. Longer branch length reflected distant phylogenetic relationship between species in clade A, suggesting wider distribution of alk-gene clusters compared with those in clade B.

Niches distribution of strains carrying alk-gene clusters

These alk-gene clusters carrying strains had extensive niches distribution, including marine, soil, water, oil field and human as inferred in Fig. S1. But the niches distribution has a strong correlation with taxonomy, rather than alk-genes. Soil niched strains were enriched in α-proteobacteria, accounting for the major of Sphingomonadales and Hyphomicrobiales in clade A. Freshwater niched strains were limited in α-proteobacteria. While urban water niched strains enriched in γ-proteobacteria, account for almost half of Pseudomonas. It is speculated that alk-gene clusters had less impact on niches adaptation. On the other hand, they had the potential for dissemination in a wide range of environments.

Alk-fragments distribution on genome

Totally 120 alk-fragments were extracted from 117 genomes, as two-copies of alk-fragments exist in three genomes. All of the three two-copies alk-fragments carrying strains are located in clade B (γ-proteobacteria). Marinobacter nauticus VT8, a species of Marinobacter isolated from oil producing well, had both alk-fragments on a single chromosome, separated by ∼150 kb distance. The other two strains were Marinobacter salarius strain HL2708#2, a species of Marinobacter isolated from Lava and Alcanivorax sp. N3-2A, a species of Alcanivorax isolated from coastal sediments, had one alk-fragment on chromosome and the other on plasmid.

Among all of the 120 alk-fragments observed, there were 24 cases located on chromosome, while only 5 cases were on plasmid. And the remaining 91 cases were located on contig or scaffold, owning to assemble level. Although alk-gene clusters were initially found on plasmids, they were also observed on chromosomes (Fig. S2).

Phylogenetic diversity of alkB from alk-fragments

As previously reported by Nie, alkB genes were clustered into eight clusters (Nie et al., 2014). The cluster IV had more taxonomy diversity, which were from α-, β- and γ-proteobacteria. Consistent phylogenetic structure of alkB genes was obtained in this study, as shown in Fig. 2. Interestingly, the 120 alkB genes located in alk-gene clusters were limited to a mono-clade, which formed a subclade of cluster IV.

As mentioned above, alk-gene clusters contained a complete gene set for alkane degradation, representing for lower adaptability requirements of the recipient in the transfer process between species, suggesting its potential for wider dissemination. But the results here showed that the spread of alk-gene clusters was limited. Their phylogenetic relationship was very conservative. Together with its limitation within a certain class of species as results described above, it was suggested that there might be restrictive factors which limit its dissemination.

Organization diversity of alk-genes and IS elements distribution

The organization diversity of alk-genes was related to taxonomy lineage, as inferred from Fig. 3. Species with similar genetic relationships showed similar organization types of alk-genes. The alk-genes in Pseudomonas had the highest integrity, including alkBFGHJKL, alkN and alkST. As a distinctive feature, almost all alkF and alkN were distributed in the species of Pseudomonas. The species in genus of Marinobacter located in clade B and the species in orders of Rhodobacterales, Hyphomonadales and Maricaulales located in clade A had relatively lower alk-genes integrity, these were alkST and alkBGHJKL. Most species in order of Sphingomonadales further lost alkK, and alkT. The alkS and alkBGJH were dominated in the genus of Alcanivorax and in order of Hyphomicrobiales. This phenomenon suggested that species barrier played an essential role in alk-genes dissemination, which needs to be broken in the transmission process.

At the same time, many IS elements existed near alk-genes, as shown in Fig. 3. These IS elements belong to 11 families. Previous reports suggested that IS elements might play a role in mobilizing the alk-gene clusters (Van Beilen et al., 2001). We further explored the distribution of IS elements among different taxonomy. As shown in Fig. 4, IS elements are distributed unevenly in different taxonomy.

The alk-fragments from Pseudomonas and Marinobacter, Hyphomicrobiales, Rhodobacterales, Hyphomonadales and Sphingomonadales had amount of IS elements belonging to diverse families. The alk-fragments from Halomonas contained IS elements belonging to only one family, while almost no IS element was observed in alk-fragments from Thalassolituus Alcanivorax and Maricaulales. The results showed that in addition to IS elements, there might be other ways to transmit alk-gene clusters. In order to explore these factors for further research, larger scale data may be required.