Whole-Genome Sequence of Chryseobacterium oranimense, a Colistin-Resistant Bacterium Isolated from a Cystic Fibrosis Patient in France

Abstract

INTRODUCTION

Cystic fibrosis (CF) is the most common autosomal recessive genetic disease and results from mutations within the gene coding for the cystic fibrosis transmembrane regulator (CFTR) protein. This life-threatening disease affects all racial and ethnic groups, though it is more common among Caucasians (1, 2). CF is characterized by hyperproduction of viscous mucus by the affected glands, resulting mainly in impaired respiratory and pancreatic functions. The most common complication of CF involves the chronic respiratory infections caused by bacterial pathogens (3), which are the main reason for the high morbidity and mortality of the disease (4). Traditionally, only a few bacteria were involved in CF lung infections, including Staphylococcus aureus, Pseudomonas aeruginosa, Haemophilus influenzae, and Streptococcus pneumoniae. However, many new or emerging opportunistic bacteria have been described in CF patients over the past decade, for instance, Burkholderia cepacia complex, Stenotrophomonas maltophilia, Achromobacter xylosoxidans, Pandoraea spp., Ralstonia spp., Inquilinus limosus, and nontuberculosis mycobacteria, as well as fungi (5). Chronic microbial infection, along with P. aeruginosa infections, leads to excessive airway inflammation and the eventual loss of pulmonary function. Colistin is an extremely important antibiotic used in patients with CF upon the first acquisition and for maintenance of chronic Pseudomonas infections. Consequently, polymyxin-resistant P. aeruginosa clinical isolates are increasingly being reported in CF patients (6, 7). However, although aggressive antimicrobial therapy has often helped to eradicate or minimize the deterioration of lung infections, it has eventually led to the emergence of new and/or atypical multidrug resistance bacteria, including colistin-resistant bacteria in CF. Several colistin resistance bacteria have been reported recently in CF patients, such as I. limosus (8), Brevundimonas diminuta (9), Ochrobactrum anthropi (9), S. maltophilia, and A. xylosoxidans (8,–12).

Members of the genus Chryseobacterium, mainly Chryseobacterium indologenes, have been documented as opportunistic pathogens known to be associated with nosocomial infections in infants and immunocompromised patients of all age groups and are resistant to colistin (13, 14). There are about 283 reported cases of infections associated with C. indologenes (15, 16). In a report by Chen et al., 215 clinical isolates of multidrug-resistant C. indologenes were identified after the increasing clinical use of colistin and tigecycline (16), a risk for patients who have undergone extensive administration of antibiotics for a long period (17). Although the source of infection of this microbe is not clear, it has been reported to be acquired nosocomially via medical devices and contaminated water supplies in hospitals (18). C. indologenes was also reported from a cohort of CF patients in Italy (19). Thirty-five clinical isolates of Chryseobacterium spp. (C. indologenes, Elizabethkingia meningoseptica [formerly Chryseobacterium meningosepticum], and Chryseobacterium gleum) were reported from CF patients who were also coinfected by one of the dominating pathogens of CF (P. aeruginosa or Burkholderia cepacia complex) (20). Furthermore, Chryseobacterium spp. only susceptible to cotrimoxazole and quinolones were reported in Italian CF patients who had received colistin therapy because of coinfection with P. aeruginosa or B. cepacia (21). The genetic basis of these multidrug-resistant bacteria remains unknown. Nonetheless, bacterial whole-genome sequencing is an economically feasible tool for deciphering the resistome (22) and has provided unprecedented insight into the evolution of antibiotic resistance (AR) (23).

Here, we report the whole-genome sequence used to decipher the resistome and genomic properties of Chryseobacterium oranimense G311, a colistin-resistant Gram-negative bacterium isolated for the first time from the sputum of a 26-month-old child with CF. It should be noted that the patient was coinfected with S. maltophilia and P. aeruginosa and had received colistin treatment prior to the isolation of this colistin-resistant bacterium. We speculate that colistin therapy led to the selection of this colistin-resistant bacterium; however, we could not isolate any other strain to perform the comparison. The true significance of isolating C. oranimense G311 in terms of clinical evolution is difficult to establish; however, it could be clinically significant, especially in immunocompromised patients. We also performed a comparison of the C. oranimense G311 genome with the genomes of closely related C. gleum ATCC 35910 and Chryseobacterium sp. strain CF314.

MATERIALS AND METHODS

RESULTS

Phenotypic properties.

C. oranimense G311 (Collection de Souches de l'Unité des Rickettsies [CSUR] reference no. P277) (10⁴ CFU/ml), along with P. aeruginosa (10⁴ CFU/ml) and S. maltophilia (10⁵ CFU/ml), was isolated from the sputum sample of a 26-month-old girl in August 2012. The isolate was pigmented yellow when isolated on Columbia agar with 5% sheep blood COS (bioMérieux) medium at 37°C after 24 h of incubation and correctly identified by MALDI-TOF as C. oranimense with a good score (>2.0). The colony size of the CF isolate varied from 0.5 to 1 mm in diameter and was capsulated (Fig. 1). This aerobic, Gram-negative, nonmotile bacillus grew well at 29°C, and growth was also observed at 18°C, 10°C, and 4°C after 2 days, 4 days, and 8 days of incubation, respectively. The isolate was able to grow microaerophilically and in the presence of 5% CO₂ but not under an anaerobic condition. Growth was also observed on Trypticase soy agar, extended spectrum-β-lactamase-producing Enterobacteriaceae, and cepacia media at 37°C and 29°C. The isolate also grew at a salt concentration of up to 2% after 24 h of incubation. Although Chryseobacterium species have been described to be present in CF patients, C. oranimense has not been reported so far, and it was not isolated from any other CF clinical sample in our lab as well. 16S rRNA PCR amplification and sequencing confirmed that the sequence was 99.7% similar to that of C. oranimense H8^T (30). The genome sequence confirmed the identification. The phylogenetic tree based on the 16S rRNA sequence was constructed to show the phylogenetic position of CF C. oranimense G311 (Fig. 2).

Open in a separate window

FIG 1

(A) Chryseobacterium oranimense G311 yellow isolate on Columbia agar with 5% sheep blood (bioMérieux) at 37°C; (B) Gram staining image of Chryseobacterium oranimense G311 viewed at ×100 magnification; (C) transmission electron microscopic image of Chryseobacterium oranimense G311 using a Morgani 268D TEM (Philips) at an operating voltage of 60 kV.

Open in a separate window

FIG 2

Phylogenetic tree based on 16S rRNA sequence highlighting the phylogenetic position of CF Chryseobacterium oranimense G311. Myroides odoratimimus was used as an outgroup. Sequences were aligned using ClustalX, and phylogenetic inferences were obtained using the neighbor-joining method within Mega 5 software. Bootstrap values are expressed by percentage of 1,000 replicates with a Kimura 2 parameter test and shown at the branching points. The branches of the tree are indicated by the genus and species name of the type strains followed by the NCBI gene accession numbers.

General features of the C. oranimense G311 genome.

A total of 2,764,904 reads were obtained, leading to 511,490,430 bp of sequence data. The size of the C. oranimense G311 genome is 4,457,049 bp, comprising one circular chromosome, with a 37.7% GC content, assembled into 15 contigs. No plasmid was detected. A total of 4,475 genes were predicted, including 4,387 protein-coding genes (EBI accession number CEJ67725 to CEJ72111) and 88 RNAs (3 rRNA operons and 85 tRNAs). Of the 4,387 protein-coding genes, 3,004 (68.47%) were assigned a putative function, whereas 216 (4.92%) genes were identified as ORFans and 1,100 were annotated as hypothetical proteins (26.60%). As many as 541 genes had a signal peptide, and 896 transmembrane proteins were detected. A comparison among the general features of the genomes of three Chryseobacterium spp. is shown in Table 1. The average nucleotide identities of the C. oranimense G311 genome with C. gleum ATCC 35910 and Chryseobacterium sp. CF314 genomes were 80.68% and 79.72%, respectively. The distribution of genes into COG functional categories and the comparison of C. oranimense with C. gleum ATCC 35910 and Chryseobacterium sp. strain CF using the BLAST Ring Image Generator (31) (BRIG) are presented in Fig. 3, and the distribution of COG categories is presented in Table S1 in the supplemental material.

TABLE 1

General features of the Chryseobacterium oranimense G311 genome in comparison to the Chryseobacterium gleum ATCC 35910 and Chryseobacterium sp. strain CF314 genomes

Species	Database accession no.a	Genome size (bp)	% GC content	No. of CDSb	No. of RNA	Avg nt identity	No. of C. oranimense G311 proteins with:
							Any similarity	Up to 80% similarity
Chryseobacterium oranimense G311	CDHM01000001–CDHM01000015	4,457,049	37.70	4,387	88
Chryseobacterium gleum ATCC 35910	ACKQ02000001–ACKQ02000007	5,569,640	36.80	5,304	79	80.68	1,355	2,435
Chryseobacterium sp. strain CF314	AKJY01000001–AKJY01000119	4,484,672	36.62	4,182	54	79.72	1,449	1,952

Open in a separate window

^aThe genome sequences were deposited in the Whole Genome Sequence (WGS) database: 15 contigs for Chryseobacterium oranimense G311, 7 contigs for Chryseobacterium gleum ATCC 35910, and 119 contigs for Chryseobacterium sp. strain CF314.

^bCDS, coding sequences.

Open in a separate window

FIG 3

(A) Graphical circular map of the Chryseobacterium oranimense G311 genome. Circle range is from 1 (outer) to 5 (inner). Circle 1, positive strand; circle 2, negative strand; circle 3, tRNA (red) and rRNA (green); circle 4, GC; circle 5, GC skew. All genes are color coded according to Cluster of Orthologous Group (COG) functions, shown in the table with number of genes for each COG using BRIG software. (B) Comparison of the Chryseobacterium oranimense G311 (CO) genome with Chryseobacterium gleum ATCC 35910 (CG) and Chryseobacterium sp. strain CF314 (CF) genomes using RAST.

Specific features of the C. oranimense G311 genome.

We found an operon (20,162 bp) comprised of modular polyketide synthase genes and a zeaxanthin glycosyltransferase gene. We noted a similar arrangement in the size of 18,518 bp in C. gleum ATCC 35910 (Fig. 4), though we did not find this arrangement in the Chryseobacterium sp. strain CF314 genome. The presence of zeaxanthin, a carotenoid pigment, in the genome likely explains the yellowish color of our CF isolate. We located the O-antigen biosynthesis cluster in the genome of the C. oranimense isolate, as shown in Fig. S1 in the supplemental material (open reading frame [ORF] 569 [CEJ68283] to ORF 601 [CEJ68319]). Similar clusters are present in the genomes of C. gleum ATCC 35910 and Chryseobacterium sp. strain CF314 (see Fig. S1 in the supplemental material). In addition, a new capsular polysaccharide (CPS) biosynthesis was identified in the genome of C. oranimense (see Fig. S2 in the supplemental material) (ORF 1284 [CEJ68992] to ORF 1317 [CEJ69025]). We could not find a similar cluster in the other genomes analyzed. The list of all the genes of O-antigen-like and K-antigen-like clusters is shown in Table S2 in the supplemental material.

Open in a separate window

FIG 4

Polyketide synthase operon arrangement in Chryseobacterium oranimense G311 and Chryseobacterium gleum ATCC 35910.

DISCUSSION

Chryseobacterium species are found in a variety of habitats and essentially ubiquitous, though some are opportunist pathogens (32). C. oranimense has been reported to be isolated from raw milk (30), yet this is the first report of C. oranimense in humans, i.e., from a CF patient. Chryseobacterium species are multidrug resistant, with most intrinsically resistant to penicillin, first- and second-generation cephalosporin, aztreonam (14, 33), and colistin (16), and have been reported to be acquired nosocomially via medical devices and contaminated water supplies in hospitals (18). Using a polyphasic approach, some studies have reported the presence of unusual bacteria, such as Acinetobacter spp., Bordetella spp., Comamonas spp., Rhizobium spp., Herbaspirillum spp., Moraxella spp., I. limosus, and Chryseobacterium spp., in the sputum samples from CF patients (34). Although the emergence of new multidrug-resistant, Gram-negative bacteria in CF lung infections has been relatively low, the incidence is increasing considerably, presenting a serious challenge for the development of effective and appropriate antibiotic therapies when they are misidentified. It is known that Chryseobacterium spp. cause infections in immunocompromised patients (13, 14), and their existence in CF airways has been reported over the last 10 years (34). One study reports E. meningoseptica and C. indologenes as the most frequently isolated species, followed by C. gleum and coinfections with at least one Gram-negative bacterium, such as P. aeruginosa, A. xylosoxidans, S. maltophilia, or B. cepacia complex in CF patients (34). Many of the isolates in the above-named study were found to be resistant to imipenem but were not checked for resistance to colistin, whereas our multidrug isolate was found to be resistant to both imipenem and colistin. As the life expectancy of CF patients has increased, antimicrobial pressure has also experienced an increase, and, consequently, more multidrug-resistant microorganisms are being isolated from the CF lung microbiota. Importantly, as these bacteria have developed multiple mechanisms of antibiotic resistance, they must be identified correctly for designing therapeutic treatments.

The genomic comparison of C. oranimense G311 with the available genomes of Chryseobacterium gleum ATCC 35910 and Chryseobacterium sp. strain CF314 (35) revealed similar genome sizes and GC contents, and none of them harbored any plasmid (Table 1). Apart from deciphering the resistome of this atypical bacterium, which will be discussed in details below, we identified three specific features in the C. oranimense G311 genome. First, the presence of PKS might play a role in the synthesis of zeaxanthin, a secondary metabolite imparting the yellowish pigmentation of the isolate. Flavobacterium multivorum has been widely studied for the production of the xanthophyll carotenoid zeaxanthin, as this species could be used as a commercial source of zeaxanthin (36). High intake of foods providing zeaxanthin is related with lower incidence of age-related macular degeneration (ARMD), mostly for ocular and retinal health. They are used as supplemental antioxidants in treating ARMD (37). The presence of this bacterium, which produces beta-carotene, in a clinical isolate in the context of CF was unexpected, and the role this bacterium may play in the lung microbiota remains to be studied in the future. Second, the lipopolysaccharide (LPS) cluster in the genome of this bacterium could be acquired laterally and to the best of our knowledge was unknown in this genus. This cluster consisted of glycosyltransferases (see Table S2 in the supplemental material) that likely contribute to modification of LPS (38), which is a well-known phenomenon associated with resistance to polymyxins (38,–41). Third, C. oranimense G311 also harbors a new capsular polysaccharide biosynthesis (K-antigen) gene cluster that was unique to this genome and also acquired laterally. Within this cluster, the wza and wzc genes have been described as outer membrane lipoprotein and integral inner membrane protein/protein tyrosine kinase, respectively, in some human pathogens, such as K. pneumoniae K2 and Escherichia coli K-12 (42,–45). Another gene product, ugd, was identified, which is involved in the production of UDP-4-amino-4-deoxy-l-arabinose, a compound that renders E. coli resistant to cationic antimicrobial peptides (46). The ugd produces UDP-glucuronic acid (UDPGA), which plays a role in the production of a sugar derivative, UDP-4-amino-4-deoxy-l-arabinose (l-Ara4N), which is vital for bacterial resistance to polymyxin (38, 47). Capsular clusters in the genus Flavobacterium have been reported in Flavobacterium columnare ATCC 43622 (48), Flavobacterium psychrophilum strain 259-93 (49), and Zunongwangia profunda SM-A87 (50).

The resistome of C. oranimense G311 comprises a reservoir of diverse β-lactamases, including a class A β-lactamase gene, bla_CME, and the cme-1 gene has been reported to be structurally divergent from other class A enzymes (51) in E. meningoseptica. The cme-1 gene encodes a clavulanic acid-susceptible extended-spectrum β-lactamase that hydrolyzes most of the cephalosporins, such as cefotaxime and ceftazidime, and monobactams, such as aztreonam, though it does not hydrolyze cephamycins and carbapenems. The C. oranimense G311 cme-like gene clustered with the cme-1 gene reported from E. meningoseptica (data not shown). Another class A β-lactamase gene, penA, encodes penicillin-binding protein PBP-2a, which is a mecA gene product that can result in ceftazidime and amoxicillin-clavulanic acid resistance if it is overproduced or mutated (52, 53). We discovered two genes for class B metallo-β-lactamases: a bla_GOB-like gene and a putative metallo-β-lactamase bla_IND gene. Class B lactamases (generically termed metallo-β-lactamases) employ one or two Zn(II) ions for cleaving the β-lactam ring. The Gob-18 is fully active against a broad range of β-lactam substrates and has been reported from E. meningoseptica (54); many more variants of gob genes have recently been reported from this species, which is known to be intrinsically resistant to most β-lactams, including carbapenems (55). The bla_IND-4 gene found in the C. oranimense G311 genome is 93.7% similar to bla_IND-4 from C. indologenes 009 (56), an enzyme that is able to hydrolyze carbapenems. We also discovered many class C extended-spectrum β-lactamases (ESBLs), such as bla_ACC-, bla_ampH-, bla_ACC-4-, and bla_CMY-like; however, certain genes, such as bla_ACC-like, showed similarities with the reported genes from plant sources (57).

The most common mechanism of resistance to colistin is modification of the LPS structure (58). Intrinsic resistance to polymyxins in Burkholderia cenocepacia and Proteus mirabilis has been linked to alterations in their lipid A structure with the addition at the 4′-phosphate moiety of the LPS of 4-amino-l-arabinopyranose and 4-amino-l-arabinose (l-Ara4N), respectively (59, 60). Such modifications have also been reported for K. pneumoniae and E. coli (58). In K. pneumoniae, the resistance to polymyxin is due to increased production of capsular polysaccharides (61). Recently, it has been demonstrated that in Acinetobacter baumannii or in Enterobacter aerogenes, acquired resistance to colistin may also be due to mutations in the pmrA-pmrB two-component systems (62, 63). Finally loss of LPS production by mutations in the three genes lpxA, lpxC, and lpxD has been associated with the resistance to Acinetobacter baumannii (64). Here, we found similar variants of pmrA (E8D) (63), pmrB (L208F, P360Q) (64), and lpxA (G68D) (Table 4) (65) that confer resistance to colistin in Acinetobacter baumannii (63,–65). Thus, we believe the intrinsic resistance of C. oranimense G311 to colistin is due to both alterations in LPS and production of capsular polysaccharides.

Supplementary Material

ACKNOWLEDGMENTS

Footnotes

REFERENCES