Biological Identification of the Strain AF18
A patient with obstructive jaundice who suffered an infection two days after the percutaneous transhepatic cholangial drainage (PTCD) surgery was admitted to Fujian Province Hospital in 2014. From the bile sample of the patient, two types of colonies were isolated after serial dilutions and isolations on MacConkey agar plates. One type was mucous, entirely pink, and of 4-5mm in diameter, which was finally identified as a K. pneumoniae clone sensitive to common antibiotics (Table 1); the other type was composed by small (2-3mm in diameter) red-centered colonies with clear and transparent edges (Fig. 1A). The bacteria of the small colonies seemed prone to adhere to the cells of K. pneumoniae and were not able to be isolated until extensive dilutions. The taxonomy of the small colonies was not immediately identified by the microbiological laboratory in the hospital, and we designated it as strain AF18. AF18 exhibited resistance to most β-lactam antibiotics in antimicrobial susceptibility testing (Table 1). As the infection was rather intractable and finally cured by intravenous amikacin, the final diagnosis for the patient was a co-infection caused by a sensitive K. pneumoniae strain and a multidrug-resistant strain of unknown species.
Microscope observation showed that AF18 was a Gram-negative bacillus (Fig. 1B), and its cells were surrounded by flagella under a transmission electron microscope (Fig. 1C). The scanning electron microscope confirmed the tubular shape of AF18 and a smooth surface with no polysaccharide particles (Fig. 1D), in line with the mucus-free characteristics of its colonies. VITEK-II in the hospital laboratory did not identify any bacterial species with identical biochemical properties to AF18 (Table S1), whereas the API20E biochemical identification system suggested AF18 as Pantoea sp. but with low reliability. The mass spectrometry which scans the protein profile of samples did not identify the species of AF18 either.
Complete Genome of Enterobacteriaceae bacterium AF18
To determine the taxonomy and genetic features of AF18, we performed whole-genome sequencing using two platforms, Illumina Hiseq (generates short-reads) and PacBio sequencer (generates long-reads), obtaining a high-quality completed genome sequence. AF18 possessed a circulated chromosome and two plasmids (Table 2).
By using Mash [24] to search the publicly available bacterial genomes and drafts with a cutoff of mutation distance < 0.25, we identified 33 non-redundant close relatives of AF18, all of which were in the Enterobacteriaceae family. The average nucleotide identity (ANI) matrix of the 34 strains (Fig. 2A) shows that the closest five with identity > 98.5% ( > 95% regarded as strains of the same species [28]) are nominated as [Kluyvera] intestini (GCA_001856865.3), Metakosakonia sp.(GCA_003925915.1), Enterobacter sp. (GCA_000814915.1, GCA_900168315.1 ), and just Enterobacteriaceae bacterium (GCA_002903045.1). The phylogenetic relationship of these relatives was further inferred with core genome SNPs (Fig. 2B), which confirmed the relationships inferred from the ANI matrix and indicated the novel species, including AF18, possibly represents another genus than Kluyvera. Herein, we temporarily nominated our stain as Enterobacteriaceae bacterium AF18 as the nomenclature of its genus and species is still undefined.
We predicted seven copies of 16S rDNA sequences in AF18. We aligned them to the 33 genomes we picked using BLASTN and calculated the average identity. We removed the genomes which do not contain high quality 16S rDNA sequence. The result shows a good congruence of 16S rDNA and whole-genome comparisons (Table S4). However, considering cutoffs commonly used for intra-species classification by whole-genome ANI > 95% [28] and 16S rDNA identity > 99% [39], 16S rDNA classification found two more strains of the species, namely Enterobacteriaceae bacterium ENNIH1, and Phytobacter ursingii strain CAV1151 (Table S4). Thus, we think that 16S rDNA can also be used as a marker gene to clarify the taxonomy of isolated strains, but we need to examine the identity cutoff we used carefully.
The chromosome of AF18 possesses 5651 protein-coding genes whose functions facilitate the survival and adaptation of AF18 in various habits (Table S2, Table S3). For example, motility-related genes, including a complete flagellar gene cluster that encodes all components of flagellar, csg gene cluster that encodes curli assembly proteins to mediate adhesion, and other genes of ompA, pilRT, ibeB, icaA, htpB and fimB, together confer the ability of adhesion, invasion, chemotaxis, and escape to the host strain. Efflux pump genes which confer resistance to macrolides, quinolones and aminoglycosides were also identified. Meanwhile, the AF18 genome possesses 20 genomic islands, 11 prophages, and five CRISPR sequences (Table S3), suggesting the active transfer of stress-adaptive genes by these mobile genetic elements in this species. More importantly, markers of soil-inhabiting bacteria, including a complete nitrogen fixation gene cluster and ksgA—— a pesticide-resistant gene, were found in AF18 genome, which suggests that AF18 is able to colonize natural environments. The mobility of this strain may potentiate its dissemination to various habits.
Analysis of conserved genes in plasmids shows that most of the antibiotic-resistant genes of AF18, including qnrS, dfrA, and blaCTX-M-3, are carried by the smaller plasmid pAF18_2 (Fig. 3, Table S2) which is, in major part, responsible for the antibiotic resistance profile of AF18 (Table 1). Sequence alignment shows that pAF18_2 is similar to many plasmids from other Enterobacteriaceae species, such as E. coli (KF914891.1, KC788405.1, CP028486.1), K. pneumoniae (KX928750.1, CP026179.1), and C. freundii (KT989599.1), and they contain identical replication origins, replication and transcription systems, plasmid partition systems, and a partial gene cluster responsible for plasmid conjugation, which indicates that the plasmid might be compatible with all these Enterobacteriaceae host species. Besides, these plasmids share a common anti-restriction system that ensures they would not be destroyed by the restriction-modified system in other host strains. Specifically, the pAF18_2 contains an active transposase system with complete IS elements which had acquired the blaCTX-M-3 gene and an arsenical resistant system. Many other DNA manipulating enzymes, such as integrase and DNA invertase, were also identified in the plasmid, all of which could facilitate the plasmid in efficiently acquiring and transferring antibiotic-resistance genes and other stress-adaptive genes among Enterobacteriaceae strains. Unfortunately, due to constraints related to the outbreak of the 2019 novel coronavirus, we were unable to perform conjugation experiments.
Growth of AF18 in Co-cultures and Its Transcriptional Regulation
To disentangle the respective contribution of AF18 and the sensitive K. pneumoniae in the co-infection, we co-cultivated the two strain in various concentration of ceftriaxone, and found that addition of 1% of AF18 was able to elevate the MIC from 0.125 µg/ml of pure K. pneumoniae culture to 64 µg/ml. Furthermore, when spreading the co-culture onto the MacConkey agar containing ceftriaxone, the sensitive K. pneumoniae colonies were able to withstand 8 µg/ml ceftriaxone (Fig. 4A), indicating a strong protective effect of AF18 to the co-infected K. pneumoniae.
Although necessary in the co-infection for antibiotic-resistance, AF18 only took less than 1% in the initial sample. Even when equally input, the proportion of AF18 decreased to 1% of the co-culture if without antibiotic pressure (Fig. 4B). It seems that AF18 may be less aggressive, and its growth rate is much slower than the co-inhabited K. pneumoniae. It has been reported that plasmid carriage may slow down growth rate due to the cellular cost imposed [40], and thus we generate a new strain—AF18-NC by deleting the resistant plasmid of AF18. Then we measured the independent growth curve of the three strains— K. pneumoniae, AF18, and AF18-NC, respectively (Fig. 4C). As expected, AF18-NC did grow faster than its mother strain AF18 since it was relieved from the plasmid-caused cellular cost. However, the growth rate of AF18-NC was still much slower than that of K. pneumoniae, suggesting that slow growth is an inherent property of the novel species.
Next, we analyzed the genes involved in the regulation of the growth rate by a comparison between the transcriptomes of AF18 and AF18-NC. A total of 3,309 genes of chromosomal coding genes were significantly differentially expressed, with 1675 upregulated and 1634 downregulated in AF18 (Fig. 4D). Functional cluster analysis with GO database showed that most of the differentially expressed genes were in the categories of transcriptional regulation, biosynthesis regulation, metabolic process regulation, signal transduction, and flagellar motility (Fig. S1). Analysis of the non-coding sRNA expression profile identified a total of 15 sRNAs differentially expressed between AF18 and AF18-NC. Interestingly, two of the down-regulated sRNAs in AF18, namely sRNA00063 and sRNA00291 (Fig. S2), shared 98% of their predicted target genes which constitute up to 56% of those differentially expressed genes as mentioned above, suggesting that these two sRNAs have a key role in promoting growth. This result indicated the importance of the two sRNAs in global regulation of growth rate, and consequently, the contribution of the host AF18 in co-infections.