Whole-genome analysis reveals the evolution and transmission of an MDR DH/NAP11/106 Clostridium difficile clone in a paediatric hospital

Abstract

Background

Clostridium difficile strain DH/NAP11/106, a relatively antibiotic-susceptible strain, is now the most common cause of C. difficile infection (CDI) among adults in the USA.

Objectives

To identify mechanisms underlying the evolution and transmission of an MDR DH/NAP11/106 clone.

Methods

WGS (Illumina MiSeq), restriction endonuclease analysis (REA) and antibiotic susceptibility testing were performed on 134 C. difficile isolates collected from paediatric patients with CDI over a 2 year period.

Results

Thirty-one of 134 (23%) isolates were REA group DH. Pairwise single-nucleotide variant (SNV) analyses identified a DH clone causing seven instances of CDI in two patients. During the 337 days between the first and second CDI, Patient 1 (P1) received 313 days of antibiotic therapy. Clindamycin and rifaximin resistance, and reduced vancomycin susceptibility (MIC 0.5–2 mg/L), were newly identified in the relapsed isolate. This MDR clone was transmitted to Patient 2 (P2) while P1 and P2 received care in adjacent private rooms. P1 and P2 each developed two additional CDI relapses. Comparative genomics analyses demonstrated SNVs in multiple antibiotic resistance genes, including rpoB (rifaximin resistance), gyrB and a gene encoding PBP; gyrB and PBP mutations did not consistently confer a resistance phenotype. The clone also acquired a 46 000 bp genomic element, likely a conjugative plasmid, which contained ermB (clindamycin resistance). The element shared 99% identity with the genomic sequence of Faecalibacterium prausnitzii, an enteric commensal.

Conclusions

These data highlight the emergence of MDR in C. difficile strain DH/NAP11/106 through multiple independent mechanisms probably as a consequence of profound antibiotic pressure.

Introduction

Clostridium difficile (recently renamed Clostridioides difficile^1,2) is a pathogen that has been classified by the US CDC as an urgent public health threat.³ This distinction was made because of the increased morbidity, mortality and healthcare costs associated with C. difficile infection (CDI), as well as the emergence of multidrug resistance in epidemic strains, particularly strain BI [by restriction endonuclease analysis (REA)/NAP1 (by PFGE)/027 (by PCR ribotyping)].^4,5

In our cohort of children diagnosed with CDI between 2011 and 2013,⁶ BI/NAP1/027 was uncommon, and REA group DH (also known as NAP11 and ribotype 106) predominated.⁷ Although ribotype 106 was previously uncommon in the USA, the CDC recently reported that it is now the most common strain causing CDI among US adults.^8,9 Ribotype 106 had been the second-most prevalent ribotype in the UK until 2009,¹⁰ after which yearly declines in incidence were noted.¹¹

In our paediatric cohort, there was significant genomic diversity among strains causing CDI, and direct transmission between symptomatic children was uncommon.¹² Although antibiotic resistance was relatively uncommon among the DH strains causing paediatric CDI at our institution (L. K. Kociolek, A. R. Hauser, D. N. Gerding, D. W. Hecht, E. A. Ozer, unpublished data), we identified a single MDR DH clone that caused relapsing CDI in two paediatric patients over a 20 month period. The objective of this study was to determine the mechanisms that contributed to the transmission of this clone and the evolution of its MDR phenotype through a rigorous comparative genomics analysis. Because fluoroquinolone resistance was a major contributor to the pathogenesis and global dissemination of epidemic strain BI/NAP1/027,^4,5,13 elucidating the mechanisms of antibiotic resistance in newly emerging strains is of particular interest.

Patients and methods

Patients and setting

Paediatric patients diagnosed with CDI by tcdB PCR between 2011 and 2013 at the Ann & Robert H. Lurie Children’s Hospital of Chicago were included. The clinical microbiology laboratory restricts testing for toxigenic C. difficile by the tcdB (toxin B gene) PCR Xpert C. difficile assay (Cepheid, Sunnyvale, CA)¹⁴ to unformed stools obtained from children ≥12 months old. Patient hospital and antibiotic exposure data were extracted from the electronic medical record. Antibiotic days of therapy (DOT) was defined as the aggregate sum of days of administration of each specific antibiotic. For example, either receiving a single antibiotic on 2 consecutive days or two unique antibiotics only for 1 day would each be representative of 2 DOT.

Ethics

The Lurie Children’s Institutional Review Board (IRB) approved this study (IRB reference number 2014-15675) and waived informed consent.

Microbiology and WGS

As previously described, saved stools from children with CDI underwent anaerobic culture^6,15 and isolates underwent REA^6,16 at the Microbiology Research Laboratory at Edward Hines Jr. VA Hospital. Genomic DNA was extracted from C. difficile isolates using the BiOstic Bacteremia DNA Isolation Kit (MO BIO Laboratories, Carlsbad, CA). Paired-end sequencing libraries were prepared using the Nextera XT DNA Library Prep Kit (Illumina, San Diego, CA), and WGS was performed using Illumina MiSeq to produce paired 300 bp reads. De novo genome assembly was performed using SPAdes (v3.6.2; http://cab.spbu.ru/software/spades/).¹⁷In silico MLST¹⁸ was performed using PubMLST (https://pubmlst.org/cdifficile/), which permitted clade assignment for each isolate.

Isolate genetic relatedness was determined by performing pairwise comparisons of single-nucleotide variants (SNVs) among strains, as previously described.¹⁹ Illumina reads were trimmed and filtered for low-quality bases and adapter sequences using Trimmomatic v0.36.²⁰ Reads were then aligned to the chromosomal sequence of the clade 1 reference strain 630 (GenBank accession number AM180355.1) using Stampy (v1.0.29) with an expected substitution rate setting of 0.01. SNVs relative to the reference were called using the mpileup function of samtools (v0.1.19-44428 cd) with the following options: -E (recalculate extended BAQ), -M 0 (cap mapping quality at 0), -Q 25 (skip bases with BAQ <25), -q 30 (skip alignments with mapQ <30), -m 2 (minimum gapped reads for indel candidates of 2), -D (output per-sample DP in BCF), -S (output per-sample strand bias P value in BCF) and -g (generate BCF output). SNVs were filtered if they failed to meet one or more of the following criteria: minimum SNV quality score of 200, minimum read consensus of 75%, minimum of five reads covering the SNV position, maximum of 3× the median read depth of the total alignment, minimum of one read in either direction covering the SNV position, homozygous under the diploid model and not within a repetitive region as determined by BLAST alignment of fragments of the reference strain 630 sequence against itself. For each strain, the reference strain 630 sequence was used as the base sequence. Any positions with SNVs that passed the above filters were changed to the SNV base. Any positions with SNVs that did not pass the above filters were changed to a missing base character. Any non-SNV position with coverage of fewer than five reads was changed to a missing base character. After filtering, positions with a base in <100% of the genomes were excluded (i.e. only core genome sequence was included). To minimize the impact of recombination on SNV counts, ClonalFrameML (v1.0-16-g30da94a) was used to identify regions of potential recombination, which were then masked in the alignment. Isolates were considered to be isogenic if there were zero to two or zero to three SNVs (depending on the time between isolate collection) between strains, as previously described.¹⁹

Antibiotic susceptibility

MICs of the following antibiotics were measured by the agar dilution method at the Loyola University Chicago Stritch School of Medicine: metronidazole, vancomycin, rifaximin, clindamycin and moxifloxacin (i.e. high-level fluoroquinolone resistance).^21,22 Ampicillin MICs were measured by broth microdilution in the Special Infectious Diseases Laboratory at Lurie Children’s Hospital. MIC breakpoints were set for metronidazole (≥32 mg/L), clindamycin (≥8 mg/L), moxifloxacin (≥8 mg/L) and ampicillin (≥2 mg/L) based on CLSI breakpoints²² and for vancomycin (≥4 mg/L) based on the EUCAST epidemiological cut-off value.²³ The rifaximin resistance breakpoint (≥32 mg/L) was previously described.²⁴

Comparative genomics analysis

To identify the mechanisms leading to the emergence of multidrug resistance in a C. difficile clone, a comparative genomics analysis was performed among a cluster of seven isogenic isolates. The core genome of these C. difficile isolates, defined as genomic regions present in all seven isolates, was determined using Spine (v0.2; http://vfsmspineagent.fsm.northwestern.edu/cgi-bin/spine.cgi), which is a program that defines the core genome within a group of isolates based on the identified shared genomic sequences after alignment of the sequences in the user-defined group.²⁵ Thus, the remaining accessory genome of each isolate comprises genomic sequences that are absent from at least one of the six other isolates. These discrete accessory genomic element (AGE) sequences were identified using ClustAGE (v0.7; http://vfsmspineagent.fsm.northwestern.edu/cgi-bin/clustage.cgi). AGEs were annotated using Prokka (v1.11; http://www.vicbioinformatics.com/software.prokka.shtml).

Genomic sequences found to be present in the MDR isolates but absent in the antibiotic-susceptible isolate were considered potential antibiotic resistance-associated AGEs. These sequences were screened against the BLAST non-redundant protein sequence database (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and the ICEberg database for integrative and conjugative elements (ICEs).²⁶ To assess for the acquisition of potentially biologically significant variants in the MDR isolates, non-synonymous SNVs were identified from whole-genome alignments of the MDR strains. Non-synonymous SNVs were manually validated by visualization of read alignments using Tablet v1.16.09.06.²⁷

To assess the relationship between bacterial cell wall morphology and vancomycin susceptibility (which has previously been noted in vancomycin-intermediate Staphylococcusaureus²⁸), four isolates underwent transmission electron microscopy (TEM) during the stationary phase of growth. Three isolates from this isogenic cluster were selected, one with vancomycin MIC 0.5 mg/L and two with vancomycin MIC 2 mg/L. For comparison, a genetically unrelated isolate with vancomycin MIC 0.25 mg/L was also imaged.

Results

Of the 134 C. difficile isolates that underwent WGS, 108 (81%) were assigned to clade 1 by MLST. Among these clade 1 strains, 31 (29% of clade 1 strains, 23% of all sequenced isolates) of which had been previously identified as REA group DH, a cluster of seven isogenic DH strains was identified based on the presence of only zero to two SNVs between isolate pairs. These seven isolates had been collected from two patients with multiply relapsing CDI, and this cluster was investigated further. The nucleotide sequences for the seven genomes included in this study have been deposited at DDBJ/ENA/GenBank under the following accession numbers: 5624 (MOQM00000000); 5559 (MOQJ00000000); 6609 (MOSY00000000); 7456 (MPEW00000000); 7462 (MPGF00000000); 7482 (MPFN00000000); and 7491 (MPFP00000000).

Figure 1 illustrates the CDI chronology of these two patients, their hospital exposures, and the serial antibiograms of the isolates. Patient 1 (P1), a teenage patient with recurring medical device-related infections requiring surgical management, experienced a considerably prolonged hospitalization, and during the 337 inpatient days between the first and second CDI, P1 received 313 DOT (Table 1). Clindamycin and rifaximin resistance, as well as reduced vancomycin susceptibility, were newly identified in the relapsed CDI isolate. Although these MIC changes were noted in all subsequent isolates in this cluster, the change in ampicillin MIC was inconsistent among subsequent isolates. The MDR clone was then transmitted to Patient 2 (P2), a 3-year-old patient with cancer, while P1 and P2 simultaneously received care in adjacent private rooms in the paediatric ICU (PICU). In P1 the minimum infectious period was 57 days, and in P2 the CDI incubation period was 39 days. P1 and P2 each developed two additional CDI relapses after transmission. P1 received metronidazole to treat the first and second CDI and vancomycin for the third and fourth CDI. P2 received metronidazole to treat the first and second CDI and vancomycin for the third CDI.

Figure 1.

Patient C. difficile infection (CDI) chronology and hospital exposures, along with isolate relatedness and serial antibiograms of an isogenic cluster of seven C. difficile isolates in two paediatric patients with relapsing CDI. SNVs are counted in comparison with the previous strain. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC. DOT, days of therapy; SNV, single-nucleotide variant.

Table 1.

Patient 1 monthly antibiotic exposure between relapsing CDIs

Month	Antibiotics received
March 2012	ceftriaxone, metronidazole, vancomycin (iv)
April 2012	cefazolin, ceftriaxone, clindamycin, meropenem, vancomycin (iv)
May 2012	ceftriaxone, gentamicin, trimethoprim/sulfamethoxazole, vancomycin (iv)
June 2012	ceftriaxone, vancomycin (iv)
July 2012	ciprofloxacin, meropenem, metronidazole, trimethoprim/sulfamethoxazole, vancomycin (iv)
August 2012	ceftriaxone, clindamycin, vancomycin (iv)
September 2012	clindamycin, trimethoprim/sulfamethoxazole
October 2012	ceftriaxone, clindamycin, gentamicin, vancomycin (iv)
November 2012	ceftriaxone, rifampicin, vancomycin (iv)
December 2012	ceftriaxone, gentamicin, rifampicin, vancomycin (iv)
January 2013	ceftriaxone, gentamicin, meropenem, vancomycin (iv)
February 2013	ceftriaxone, ciprofloxacin, clindamycin, gentamicin, meropenem, metronidazole, trimethoprim/sulfamethoxazole, vancomycin (iv)

iv, intravenously.

Of the seven isolates, the first isolate from this cluster (5624) was determined to be antibiotic susceptible, and the subsequent six were MDR. Table 2 demonstrates the genomic changes that were acquired by the MDR isolates. By comparative genomics analysis, the MDR isolate was noted to have acquired a 46 000 bp AGE (Figure 2a) that contained 39 genes (Figure 2b, Table 3), including ermB (rRNA adenine N-6-methyltransferase). The element was flanked by tRNA genes, indicating integration of the element within the C. difficile chromosome. The element contained several genes commonly found in conjugative plasmids, namely type IV secretion genes (TRAG family protein and TraE protein)²⁹ and relaxase,²⁹ although these genes can also be found in ICEs.³⁰ This large AGE shared 99% sequence identity (by BLAST query) with the genomic sequence of Faecalibacterium prausnitzii (GenBank accession number FP929045.1), an enteric commensal that is an important component of the healthy adult microbiome.³¹ Of note, this was the only organism that shared close sequence identity with this AGE. We were unable to identify a similar plasmid or ICE in the BLAST or ICEberg databases. TEM failed to demonstrate differences in cell wall morphology among the four DH isolates that had various vancomycin MICs (Figure 3).

Table 2.

Antibiotic resistance-associated genomic changes

Patient, CDI no.	Genomic change	Associated phenotypic change
1, 2	acquisition of 46 200 bp element that contains 39 genes, including ermB	clindamycin resistance
1, 2	rpoB missense mutation (G1514A, Arg-505→Lys)	rifaximin resistance
1, 2	penicillin-binding protein gene missense mutation (G2834T, Ser-945→Ile)	ampicillin susceptibility reduced 0–2 dilutions
2, 3	gyrB missense mutation (G1423A, Gly-475→Ser)	moxifloxacin susceptibility unchanged

Figure 2.

(a) Comparative genomics analysis of an isogenic cluster of C. difficile isolates comparing the distribution of accessory genomic elements (AGEs) among an antibiotic-susceptible isolate (blue) and six MDR isolates (red). The outer ring (alternating orange and green) indicates individual AGEs, ordered by decreasing size. A 46 000 bp element (bin1) is absent from the antibiotic-susceptible isolate and present in all MDR isolates. (b) Gene annotation of this AGE demonstrates 39 genes. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC. *annot*, annotation.

Table 3.

Gene annotation of the antibiotic resistance-associated accessory genomic element

Gene annotation
Replication initiation protein
Chromosome partitioning ATPase
ParB-like partition protein
Hypothetical protein
Transcriptional regulator
Putative ATPase of the AAA superfamily protein
TRAG family protein
Hypothetical protein
Hypothetical protein
Conjugative transposon membrane protein
TraE protein
NLP/P60 protein
Isocitrate/isopropylmalate dehydrogenase
SH3 type 3 domain-containing protein
Hypothetical protein
DNA topoisomerase III
Hypothetical protein
DNA primase
Hypothetical protein
SNF2-related protein
Relaxase/mobilization nuclease family protein
Heat shock protein HtpX
Hypothetical protein
Prophage head protein
Hypothetical protein
Zeta toxin
rRNA adenine N-6-methyltransferase
Helix-turn-helix protein
Integrase family protein
Regulatory protein
Putative Zn peptidase
Putative transcriptional regulator
Hypothetical protein
Putative transcriptional regulator
Growth inhibitor
Helix-turn-helix domain protein
Periplasmic molybdate-binding protein
Integrase
Hypothetical protein

Figure 3.

TEM of four C. difficile isolates demonstrating similar cell wall morphology amongst the isolates. (a) Isogenic DH isolate from Patient 1 (vancomycin MIC 0.5 mg/L). (b) Isogenic DH isolate from Patient 1 (vancomycin MIC 2 mg/L). (c) Isogenic DH isolate from Patient 2 (vancomycin MIC 2 mg/L). (d) Non-isogenic DH isolate from an unrelated paediatric patient (vancomycin MIC 0.25 mg/L).

Discussion

These data demonstrate the evolution and transmission of an MDR C. difficile clone in a paediatric hospital. Because fluoroquinolone resistance was a major contributor to the pathogenesis and global dissemination of epidemic strain BI/NAP1/027,^4,5,13 elucidating the mechanisms of antibiotic resistance in newly emerging strains such as DH/NAP11/106 is particularly important. Through rigorous analysis of antibiograms and whole genomes of strains causing multiply relapsing CDI in individual patients, an analysis that to our knowledge has not been previously described, we were able to identify multiple independent genomic changes that together conferred an MDR phenotype in a C. difficile DH strain. This MDR isolate was subsequently transmitted to another patient, and both patients suffered multiply relapsing CDI with this MDR strain.

The profound antibiotic exposure likely contributed to these events, and these data further highlight the importance of antibiotic stewardship in CDI prevention. Prolonged antibiotic exposure to multiple classes of antibiotics presumably provided selective pressure for these genomic changes. Furthermore, antibiotic exposure likely caused a sustained dysbiosis and loss of C. difficile colonization resistance. This seemingly led to prolonged carriage of the MDR strain after CDI recovery, increasing the probability of hospital transmission. Although no other patients were identified in this transmission cluster, we cannot rule out transmission to other patients who become colonized, but not infected, with C. difficile, and thus were not identified by this study of paediatric patients with CDI. We also cannot rule out continued transmission of this isolate after completion of the study period.

The mechanisms by which antibiotic resistance developed in this cluster of isogenic isolates are consistent with those previously described in C. difficile.³² For example, rifaximin resistance was conferred by the well-described rpoB G1514A missense mutation leading to a substitution of arginine for lysine at amino acid 505 in the RpoB RNA polymerase protein.³² Although a G1423A missense mutation in gyrB (causing a substitution of glycine for serine at amino acid 475) was identified in the second relapsed isolate in P2, this was not associated with a change in moxifloxacin MICs.

Clindamycin resistance in C. difficile is often mediated through acquisition of mobile genetic elements, and several elements have been described.³² However, the specific integrated genetic element containing ermB described in the present study has not been previously reported. While the element was noted to have integrated into the C. difficile chromosome (i.e. flanked by tRNA genes), the element contained genes that are commonly found in plasmids and ICEs. Although we cannot definitively determine whether the element is an integrated plasmid or ICE, we favour the former for several reasons. For example, in one of the six MDR isolates, the element was assembled as a duplicated sequence totalling 92 000 bp in length. This suggests that in this strain, the element may have existed in both integrated and excised forms among the C. difficile genomes extracted from culture, leading to this technical artefact. Furthermore, the element also contained a gene annotated as a prophage head protein. This may suggest that the element resulted from fusion of a plasmid and a bacteriophage, as previously described.³³ Nonetheless, this study further highlights the importance of horizontal gene transfer in C. difficile evolution and diversity.³³ Because BLAST analysis revealed that this element was only found in F.prausnitzii, these data also highlight the potential role of the healthy intestinal microbiota in modifying the C. difficile resistome. F. prausnitzii, like C. difficile, is a member of the Clostridiaceae family. F. prausnitzii has been well documented to be a prominent component of the healthy adult microbiome, and depletion of this organism may be associated with various gastrointestinal and/or inflammatory conditions.³¹ However, its contribution to antibiotic resistance is poorly understood.

Because of the important role that vancomycin has in CDI treatment, particularly relapsing and severe CDI, the change in vancomycin susceptibility was of particular interest. However, the vancomycin MIC of 2 mg/L noted in the MDR isolates is still below the EUCAST epidemiological cut-off value for resistance (i.e. 4 mg/L).²³ Furthermore, because oral vancomycin is non-absorbable, vancomycin gut concentrations are typically several orders of magnitude greater than this MIC and thus the clinical significance of this MIC elevation is negligible. Nonetheless, the mechanism of the elevation in vancomycin MIC could not be completely elucidated in this study. Acquisition of known vancomycin resistance genes was not identified, and cell wall morphological changes, as previously identified in vancomycin-intermediate S. aureus,²⁸ were not visualized in our C. difficile DH strains.

Of note, the patients in this study were diagnosed with CDI by tcdB PCR, which is the sole method for CDI diagnosis at our clinical microbiology laboratory. Additional toxin enzyme immunoassay testing, a more specific test for CDI, was not performed.³⁴ The diagnostic predictive value of PCR has recently been questioned because patients with C. difficile carriage with alternative diarrhoeal aetiologies are increasingly identified.³⁴ Because our laboratory only performs C. difficile PCR on unformed stools, presence of diarrhoea can be inferred, but we cannot definitively rule out C. difficile carriage in the presence of another diarrhoeal aetiology. Thus, the prevalence of CDI in this study may have been overestimated.

In summary, these data highlight the emergence of MDR C. difficile through multiple mechanisms, presumably as a consequence of profound antibiotic pressure, leading to the development of multiply relapsing CDI in two paediatric patients. Additional research in a larger cohort of patients is needed to better understand the frequency of emergence of MDR among strains causing paediatric CDI, as well as the role of gene exchange between intestinal commensals in the development of antibiotic resistance.