Clinical Microbiology and 23 (2017) 560e566

Contents lists available at ScienceDirect

Clinical Microbiology and Infection

journal homepage: www.clinicalmicrobiologyandinfection.com

Original article innocuum is a significant vancomycin-resistant for extraintestinal clostridial infection

y J.-H. Chia 1, 2, , 7,Y.Feng3, 4, 7, L.-H. Su 1, 2, 5, T.-L. Wu 1, 2, C.-L. Chen 5, Y.-H. Liang 5, * C.-H. Chiu 5, 6,

1) Department of Laboratory Medicine, Chang Gung Memorial Hospital, Taoyuan, Taiwan 2) Department of Medical Biotechnology and Laboratory Science, Chang Gung University, Taoyuan, Taiwan 3) Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, Hangzhou, China 4) Institute of Translational Medicine, Zhejiang University School of Medicine, Hangzhou, China 5) Molecular Infectious Disease Research Centre, Chang Gung Memorial Hospital, Chang Gung University College of Medicine, Taoyuan, Taiwan 6) Division of Paediatric Infectious Diseases, Department of Paediatrics, Chang Gung Children's Hospital, Chang Gung University College of Medicine, Taoyuan, Taiwan article info abstract

Article history: Objectives: Extra-intestinal clostridial infection (EICI) is rare but can be fatal. Traditional phenotypic Received 16 October 2016 methods can only assign many of the Clostridium species to the genus level. Received in revised form Methods: A total of 376 non-repetitive Clostridium isolates from sterile sites were collected and subjected 15 February 2017 to matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) Biotyper analysis and 16S Accepted 22 February 2017 rRNA sequencing. Antimicrobial susceptibility was determined, and clinical characteristics of the patients Available online 28 February 2017 were assessed. Clostridium innocuum isolates were characterized by genome sequencing and genotyping. Editor: P.T. Tassios We used molecular and cellular methods to explore the virulence and resistance mechanisms of C. innocuum. Keywords: Results: Clostridium innocuum was the second most common species to cause EICI, only next to Clos- Clostridium innocuum tridium perfringens. All Clostridium isolates showed susceptibility to , , peni- Extra-intestinal clostridial infection cillin, piperacillin and ampicillin-sulbatam, while C. innocuum isolates were invariably resistant to MALDI-TOF vancomycin. Among 24 patients with EICI caused by C. innocuum, two (8.3%) had diarrhoea, three (12.5%) 16S rRNA sequencing had soft-tissue infection, six (25%) had appendicitis and four (16.7%) each had shock and gastrointestinal Vancomycin resistance perforation. The 30-day mortality was 16.7%. The C. innocuum isolated from different sites could not be separated from one another by genotyping. No known genes were identified in the genome of C. innocuum but the species expressed cytotoxicity to epithelial cells. D-Alanine-D-alanine ligase, alanine racemase and D-alanyl-D-alanine carboxypeptidase are three main genes responsible for vancomycin resistance in C. innocuum. Conclusions: Vancomycin-resistant C. innocuum is a previously unrecognized, yet prominent, cause for EICI. Genome analysis showed that the species could carry a lipopolysaccharide-like structure that is associated with cytotoxicity to cells in vitro. J.-H. Chia, Clin Microbiol Infect 2017;23:560 © 2017 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved.

Introduction vulnerable to EICI is the elderly with underlying diseases such as gastrointestinal diseases, diabetes mellitus, liver cirrhosis, malig- Extra-intestinal clostridial infection (EICI) occurs infrequently nancy and chronic kidney diseases requiring haemodialysis [5,6]. but is associated with a high mortality [1e4]. A population Furthermore, most of the patients usually have a history of recent antimicrobial use [2,7]. is the most frequent * Corresponding author. C.-H. Chiu, Division of Paediatric Infectious Diseases, species for EICI, followed by Clostridium septicum, Clostridium Department of Paediatrics, Chang Gung Children's Hospital, Chang Gung University difficile and Clostridium innocuum, according to previous reports College of Medicine, No. 5, Fu-Hsin Street, Kweishan 333, Taoyuan, Taiwan. [5,8,9]. Clostridium innocuum may attract special attention in that E-mail address: [email protected] (C.-H. Chiu). all of the C. innocuum isolates found so far are exclusively resistant 7 J-HC and YF contributed equally to this work. y Current address: Anti-infectious Group, Eurofins Panlabs Ltd., Taipei, Taiwan. to vancomycin [10,11], whereas the other Clostridium species are http://dx.doi.org/10.1016/j.cmi.2017.02.025 1198-743X/© 2017 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved. J.-H. Chia et al. / Clinical Microbiology and Infection 23 (2017) 560e566 561 susceptible [12e14]. Although the name ‘innocuum’ was chosen MALDI-TOF MS with FLEX CONTROL 3.0 software and the MALDI BIO- because of the organism's proposed lack of virulence, the very TYPER v3.1. limited literature based on few clinical cases indicated that the mortality rate caused by this organism was even higher than other Clinical data collection and definitions Clostridium species [15]. Due to factors such as Gram-stain variability, rare spores and This study was approved by the Institutional Review Board (101- atypical clostridial colonial morphology, traditional phenotypic 4682B) of Chang Gung Memorial Hospital, Taiwan. Clinical infor- methods can only assign many of the Clostridium species to the mation, including demographic data, underlying diseases, clinical genus level and yet exhibit a low level of selectivity for species diagnoses and outcome, of the patients with culture-confirmed identification [16], especially for Clostridium ramosum, C. innocuum C. innocuum infection were retrospectively collected. In addition and Clostridium clostridioforme, the so-called RIC group. Hence, 16S to the patients with EICI, clinical information of ten patients with rRNA sequencing has become the reference standard for species diarrhoeal disease with faecal samples growing C. innocuum in identification. Recently, the technology of matrix-assisted laser C. difficile selective agar and CDC-ANA blood agar (BD BBL, Sparks, desorptioneionization time-of-flight mass spectrometry (MALDI- MD, USA) were collected and used for comparison. The species TOF MS) was also used for bacterial species identification in the identification was accomplished by 16S rRNA sequencing. The clinical setting. To date, few studies have systematically charac- history of antimicrobial use for at least 5 days before the onset of terized EICI by genotypic method [9]. In this study we applied both the in these patients was also reviewed. phenotypic and genotypic methods to characterize 376 Clostridium isolates collected from a medical centre in Taiwan during Antimicrobial susceptibility testing 2007e2011. We found that C. innocuum was the second most common aetiological agent for EICI, and the clinical features of EICI Antimicrobial susceptibilities to clindamycin, metronidazole, caused by this organism were different from those described pre- , piperacillin and ampicillin-sulbatam were tested by the viously. We also sequenced the genome of a clinical C. innocuum break-point agar dilution method according to the CLSI criteria isolate and explored the putative mechanism of virulence and M11-A8 for anaerobic [17]. The interpretive criteria for vancomycin resistance in this species. susceptibility of the CLSI M100S were used for calls of susceptible or resistant: clindamycin, susceptible, 2 mg/L, resistant, 8 mg/L; Materials and methods metronidazole, susceptible, 8mg/L, resistant, 32 mg/L; peni- cillin, susceptible, 0.5 mg/L, resistant, 2 mg/L; piperacillin, Bacterial isolates 32 mg/L, resistant, 128 mg/L; ampicillin-sulbatam, susceptible, 8/4 mg/L, resistant, 32/16 mg/L; vancomycin, resistant, 8 mg/L From 2007 to 2011, a total of 3191 non-repetitive specimens [18]. from normally sterile sites were positive in anaerobic culture at Chang Gung Memorial Hospital, a 3700-bed medical centre in Genome sequencing and annotation northern Taiwan. The specimens included ascites, blood and pleural fluid. A total of 376 non-repetitive Clostridium isolates collected One C. innocuum isolate, designated AN88-98, was selected for from these sterile specimens, including 226 from ascites, 144 from whole genome sequencing. The genomic DNA was extracted using blood and six from pleural fluid, were studied from 2007 to 2011. QIAamp DNA Mini Kit (Qiagen, Valencia, CA, USA) and then Clostridium isolates were identified using rapid ID 32A system sequenced using an Illumina GAII instrument (Illumina Inc., San (BioMerieux, Durham, NC, USA), MALDI-TOF Biotyper (Bruker Diego, CA, USA) following a 2 75-bp pair-end protocol. The raw Daltonik GmbH, Bremen, Germany), and 16S rRNA sequencing. All reads were assembled using CLC GENOMICS WORKBENCH software (CLC the isolates were cultured on CDC-ANA agar plates and incubated in bio, Aarhus, Denmark). The assembled draft genome has been anaerobic conditions at 37C for 2e5 days before examination. deposited into GenBank under the Accession Number JQIF0000- 0000. The genome annotation of the sequenced isolated AN88-98 Species identification by 16S rRNA sequencing was performed using NCBI Prokaryotic Genome Annotation Pipeline. A PCR assay targeting an 800-bp fragment of the 16S ribosomal The genomes of C. difficile strain 630 (NC_009089.1) and gene of the isolate using primers 6F (50-GAAGAGTTTGATCMTGGC- C. perfringens strain ATCC13124 (NC_008261.1) were downloaded TC-30) and 809R (50-GCGTGGACTACCAGGGTATC-30) was per- for comparison. The NCBI BLASTP program was used to determine formed. PCR products were purified and subjected to sequencing in the orthologous relationship between the Clostridium species. The both directions on a 3100-Avant Genetic Analyzer (Applied Bio- threshold was set as follows: e-value must be smaller than 1e-10, systems, Foster City, CA, USA). The 16S ribosomal DNA sequence and the aligned length must be longer than 50% of the protein obtained was compared with those deposited in the NCBI 16S ri- sequence. bosomal RNA sequences (Bacteria and Archaea) database using the The BLASTP were also run for searching known clostridial viru- Basic Local Alignment Search Tool (BLAST) through the internet lence genes in C. innocuum. The protein sequences of the virulence (available at http://www.ncbi.nlm.nih.gov/BLAST/). A 99% iden- genes were downloaded from the Virulence Factor DataBase (VFDB; tity was used to determine a specific species. available at http://www.mgc.ac.cn/VFs/).

Species identification by MALDI-TOF MS Multi-locus sequence typing and phylogenetic analysis

Samples of the Clostridium species were prepared for analysis by Seven housekeeping genesdadk, aroE, tpi, ddl, dxr, recA, the Bruker Biotyper MALDI-TOF MS system (MALDI BIOTYPER v3.1 glyAdwere selected for the multilocus sequence typing scheme. database; Bruker Daltonik GmbH, Bremen, Germany). The extrac- Primers for amplifying the seven fragments are listed in the Sup- tion method using formic acid was performed. A bacterial test plementary material (Table S1). The genomic DNA of 29 standard was included in each plate to calibrate the instrument and C. innocuum isolates, including 24 EICI isolates and five faecal iso- validate the run. The Bruker system used comprised a Microflex lates, were used as the templates, and PCR conditions were in 562 J.-H. Chia et al. / Clinical Microbiology and Infection 23 (2017) 560e566 accordance with a previous publication [19]. The derived sequences We then performed 16S rRNA sequencing on these isolates and were firstly multi-aligned by CLUSTALW and then concatenated for further identified over 30 Clostridium species. Of them, C. innocuum phylogenetic analysis. The alignment was imported into MEGA5 accounted for 24 cases, which outnumbered C. difficile to become (available at http://megasoftware.net/) and a maximum-likelihood the second most predominant species (Table 1). The species tree was constructed using the default parameter. C. bifermentans, C. ramosum, C. paraputrificum, C. hathewayi, C. septicum and C. butyricum followed according to the isolation rate. Limulus amoebocyte lysate assay MALDI-TOF MS was also used for species identification but its identification power was not satisfactory when 16S rRNA The endotoxin was detected using a ToxinSensor™ Chromo- sequencing was taken as a reference standard. Clostridium metal- genic LAL Endotoxin Assay Kit (GenScript, Piscataway, NJ, USA). lolevans was mis-identified to be C. glycolicum, and C. botulinum was Clostridium innocuum AN88-98 was tested, along with Pseudo- mis-identified to be C. sphenoides. The isolates of C. peptidivorans, monas aeruginosa PAO1 as a positive control and Staphylococcus C. aciditolerans, C. neonatale and C. tunisiense, which were identified aureus ATCC25923, C. difficile AN99-3 and endotoxin-free distilled by 16S rRNA sequencing, could not be identified to the species level water as negative controls. The assay was performed in accordance by MALDI-TOF MS. Meanwhile, C. hastiforme was identified by with the manufacturer's instructions. At the end of the assay, we MALDI-TOF MS as Tissirella praeacuta, a synonym of the same or- read the absorbance of each reaction at 545-nm wavelength. The ganism [20]. experiment was repeated three times. The normally sterile site samples were predominantly ascites and blood, with the former accounting for 60.1% and the latter Vancomycin-resistance mechanism 38.3%. Interestingly, the 12 C. hathewayi isolates came exclusively from ascites, whereas 90% of C. septicum were from blood. Other subtilis WB800 was used as the recipient strain, and the Clostridium species did not show as strong a preferential infecting plasmid pHY300PLK was used as the expression vector. Two DNA site as C. hathewayi and C. septicum (Table 1). fragments were amplified from genomic DNA of C. innocuum AN88- The antimicrobials tested included clindamycin, metronidazole, 98 and introduced into the recipient, respectively: one contains the penicillin, piperacillin and ampicillin-sulbatam. The collected genes encoding D-alanine-D-alanine ligase and alanine racemase, Clostridium isolates were susceptible to nearly all antimicrobials, the other the gene encoding D-alanyl-D-alanine carboxypeptidase. except clindamycin and penicillin, the resistance rates to which The primers for amplifying the two fragments are shown in the were both 7% (Table 1). Different species may have completely Supplementary material (Table S2). The two fragments were also different resistance profiles. The resistance rate to clindamycin was ligated together and then introduced to the recipient. The recipient 41.7% for C. paraputrificum but <15% for other species. Most (91.7%) carrying only pHY300PLK and the three transformants were sub- of the C. hathewayi isolates were resistant to penicillin, followed by jected to antimicrobial susceptibility to vancomycin using E-test C. difficile, which showed a resistance rate of 22.2% to penicillin. (bioMerieux). Clinical characteristics of EICI caused by C. innocuum Statistical analysis A total of 24 EICI cases, including five bacteraemia and 19 intra- Data analyses were performed using SPSS software (SPSS 17, abdominal infections were caused by C. innocuum. No poly- SPSS Inc., Chicago, IL, USA). The Student's t test and the chi-square microbial infections were detected in the 24 patients. Ten in- test were used whenever appropriate. All statistical significance dividuals with culture-confirmed gastroenteritis caused by was set at p <0.001. C. innocuum who were treated during the same period were used as controls in the analysisdC. innocuum was isolated from faecal Results specimens of these patients. Among the 24 EICI patients, 8.3% had diarrhoea, 12.5% had soft-tissue infection, 25% had appendicitis and Identification and antimicrobial susceptibility of Clostridium species 16.7% each had shock and gastrointestinal perforation. Overall, the 30-day mortality was 16.7%. There was no mortality in 30 days Using the Rapid ID32 system, we found Clostridium in 376 among patients with bacteraemia or gastroenteritis, whereas the specimens (12%). Of them, 190 were C. perfringens,18were mortality of intra-abdominal infection caused by C. innocuum was C. difficile, and most of the others were assigned to Clostridium spp. 21.1% (Table 2).

Table 1 Distribution of Clostridium species from sterile sites and antimicrobial susceptibility profile

Clostridium species Total Sites of isolation Resistant isolates (%)

AS PL BL DA MTZ PEN PRL SAM

C. perfringens 190 111 2 77 8 (4.2%) 0 0 0 0 C. innocuum 24 19 0 5 1 (4.2%) 0 3 (12.5%) 0 0 C. difficile 18 13 1 4 0 1 (5.6%) 4 (22.2%) 0 0 C. bifermentans 18 12 0 6 2 (11.1%) 0 0 0 0 C. ramosum 17 12 0 5 2 (11.8%) 0 0 0 0 C. paraputrificum 12 6 0 6 5 (41.7%) 0 0 0 0 C. hathewayi 12 12 0 0 2 (16.7%) 0 11 (91.7%) 0 0 C. septicum 101090 0 0 00 C. butyricum 106040 0 0 00 C. sordellii 64 020 0 0 00 Other species 58 30 2 26 6 (10.3%) 1 (1.7%) 6 (10.3%) 0 0 Total 375 226 5 144 26 (6.9%) 2 (0.5%) 24 (9.1%) 0 0

Abbreviations: AS, ascites; BL, blood; DA, clindamycin; MTZ, metronidazole; PEN, penicillin; PL, pleural fluid; PRL, piperacillin; SAM, ampicillin-sulbatam. J.-H. Chia et al. / Clinical Microbiology and Infection 23 (2017) 560e566 563

Soft-tissue infection was relatively more common in the pa- three strains shared 819 genes in common, which might constitute tients with bacteraemia (40%), whereas acute appendicitis was the clostridial core genome. relatively more common in patients with intra-abdominal infection According to the genome sequence, we developed a multilocus (31.6%). This result is different from a previous report that showed sequence typing scheme for C. innocuum by sequencing seven fever of unknown origin was the most common presenting symp- housekeeping genes. The sequences of the seven genes were tom for C. innocuum bacteraemia, followed by gastrointestinal concatenated and used for constructing the phylogenetic tree. As a symptoms [15]. Diarrhoea is the only significant symptom that result, C. innocuum exhibited a clear clonal structure, with its iso- distinguished EICI cases with gastroenteritis; all gastroenteritis lates being grouped into three lineages (Fig. 1). The nucleotide patients (100%) and yet only two EICI patients (8.3%) suffered from divergence was <1% within lineage and 2%~7% between lineages. diarrhoea (p <0.001, Table 2). Clostridium innocuum isolates from blood, ascites and faeces could All patients with bacteraemia had underlying diseases: 40% of not be separated from one another by their phylogenetic positions. them had diabetes mellitus and 60% had malignancy. On the other hand, 52.7% of the patients with intra-abdominal infection had Virulence mechanism underlying diseases (Table 2). The patients with gastroenteritis had the lowest rate of underlying diseases. By sequence comparison, we did not identify in the genome of All of the 24 C. innocuum isolates were resistant to vancomycin C. innocuum any genes encoding that are reported in other (MIC 8 mg/L). Apart from that, the antimicrobial susceptibility Clostridium species, including C. perfringens, C. botulinum, C. difficile, profile of C. innocuum was not significantly different from that of C. tetani and C. septicum. Previously we found that the cell pellet of other Clostridium species (Table 1). EICI caused by C. innocuum C. innocuum was cytotoxic to human colon carcinoma cell line HT- seemed to be unrelated to previous exposure to antimicrobials. 29 and kidney epithelial cell line Vero cells whereas the superna- Only one patient with bacteraemia (20%) and eight with intra- tant was not (Chia JH, 2016, unpublished data). The result implies abdominal infection (42%) received antimicrobial therapy before that C. innocuum may contain a cell-wall-embedded factor like the occurrence of EICI. In contrast, all patients with gastroenteritis lipopolysaccharide in Gram-negative . To verify this hypoth- had a history of antimicrobial therapy for >5 days within 2 weeks. esis, we tested the presence of endotoxin in C. innocuum and C. difficile, taking P. aeruginosa as a positive control and S. aureus as a Phylogenetic relationship negative control. The Limulus amoebocyte lysate assay showed that the signal of endotoxin from C. innocuum was much lower than for One C. innocuum isolate was selected for genome sequencing, P. aeruginosa but significantly higher than C. difficile, S. aureus and and a 4.6-Mb draft genome was obtained. The average GC content endotoxin-free control. No significant difference was detected of this organism was 44%, which was dramatically higher than among the last three groups (Fig. 2). C. difficile (29%) and C. perfringens (28%). A total of 4131 protein- coding genes were predicted. In all, 2075 (50%) were shared with Vancomycin resistance mechanism C. difficile strain 630, with an average protein identity of 41.6 ± 12.5%; 1780 (43%) were shared with C. perfringens strain In a previous study, D-alanine-D-alanine ligase and racemase ATCC13124, with an average protein identity of 39.3 ± 11.3%. The have been attributed to the intrinsic resistance of C. innocuum to

Table 2 Demographic, clinical and laboratory characteristics of Clostridium innocuum infection

Characteristics Bacteraemia Intra-abdominal infection Gastroenteritis

Isolate number (sites of isolation) 5 (blood) 19 (ascites) 10 (faeces) Age (years) 70.4 ± 11.8 59.1 ± 25.0 44.8 ± 33.5 Male (%) 3 (60%) 10 (52.6%) 6 (60%) White blood cell count 11340 ± 7759.7 8205.3 ± 6484.0 NA 30-day mortality 0 (0%) 4 (21.1%) 0 (0%) Clinical presentations Shock 1 (20%) 3 (15.8%) 0 (0%) Soft-tissue infection 2 (40%) 1 (5.3%) 2 (20%) Diarrhoea 1 (20%) 1 (5.3%) 10 (100%)a Acute appendicitis 0 (0%) 6 (31.6%) 0 (0%) Peritonitis 1 (20%) 3 (15.8%) 0 (0%) Gastrointestinal perforation 1 (20%) 3 (15.8%) 0 (0%) Underlying diseases Liver cirrhosis 1 (20%) 0 (0%) 0 (0%) Intracranial haemorrhage 1 (20%) 0 (0%) 0 (0%) Diabetes mellitus 2 (40%) 4 (21.1%) 2 (20%) Hypertension 1 (20%) 4 (21.1%) 2 (20%) Brain infarction 1 (20%) 1 (5.3%) 0 (0%) Malignancy 3 (60%) 6 (31.6%) 0 (0%) Previous antimicrobials 5 days Gentamicin 0 (0%) 3 (15.8%) 2 (20%) Cefazolin 0 (0%) 2 (10.5%) 0 (0%) Ceftazidime 0 (0%) 1 (5.3%) 0 (0%) Flomoxef 1 (20%) 0 (0%) 0 (0%) Teicoplanin 0 (0%) 1 (5.3%) 2 (20%) Imipenem 0 (0%) 1 (5.3%) 0 (0%) Ciprofloxacin 0 (0%) 0 (0%) 2 (20%) Oxacillin 0 (0%) 0 (0%) 2 (20%) Trimethoprim-sulfamethoxazole 0 (0%) 0 (0%) 2 (20%)

a This indicates a significant difference between extra-intestinal clostridial infection (bacteraemia plus intra-abdominal infection) and gastroenteritis patients (p <0.001). 564 J.-H. Chia et al. / Clinical Microbiology and Infection 23 (2017) 560e566

Fig. 2. Endotoxin activity of Clostridium innocuum assessed by Limulus amoebocyte lysate assay. ***p <0.001; ns, non-significant (p >0.05).

Discussion

Clostridium perfringens accounted for nearly half of the EICI cases according to previous studies [4,9]. Over 30 Clostridium species were responsible for the rest in this study. Although such diversity reflects the great discriminating power of 16S rRNA sequencing, it also suggests that the entire Clostridium genus could be the aeti- ology for EICI. Except for C. perfringens and C. difficile, the com- mercial instruments such as Rapid ID32A and even MALDI-TOF were unable to make an accurate species identification for Clos- tridium. The difficulties lie mainly in the limited knowledge on such Fig. 1. The maximum-likelihood tree based on concatenated sequences from seven a large number of Clostridium species. If these species do not share housekeeping genes of Clostridium innocuum. Scale bar under the tree indicates genetic the same clinical features, then a typing method with higher res- distance. The numbers near the nodes represent bootstrap reliability. The sources of olution, such as 16S rRNA sequencing, is needed. Further addition of sample collection are represented by different shapes: diamond, isolates collected from blood; triangle, isolates from ascites; circle, isolates from faeces. the information of C. innocuum into the MALDI-TOF MS database is also helpful for rapid identification and clinical diagnosis and treatment of the clostridial infection. vancomycin because the organism synthesized the pentapeptide The major clinical specimen from which Clostridium was isolated precursors with low affinity for vancomycin [11]. The two respon- was ascites. The second most common was blood. The EICI was sible genes are adjacent to each other on the chromosome and are considered to be of enteric origin: the clostridial organisms living in conserved among all C. innocuum isolates. We cloned the two genes the gut translocate from the intestinal lumen to the mesenteric and transferred the recombinant plasmids to B. subtilis, which was lymph nodes, and from there to the systemic circulation and other used to express the two recombinant proteins. MIC of vancomycin distant sites [22]. The multilocus sequence typing data on of this recipient strain increased from 0.38 to 1.5 mg/L (Table 3). C. innocuum constructed in this study also provide evidence for the According to genome annotation, we found a gene encoding D- theory of enteric origin that EICI isolates cannot be separated from alanyl-D-alanine carboxypeptidase, which was supposed to elimi- the intestinal isolates phylogenetically. nate the high-affinity vancomycin pentapeptide precursor [21]. In theory, gastrointestinal disorders, such as gastrointestinal The carboxypeptidase-encoding gene had a different genomic bleeding and perforation, or advanced liver failure, can introduce location from the above two genes, but all of the three genes were substantial damage of normal mucosal barriers, which makes pa- present in all C. innocuum isolates by PCR screening. Further tients susceptible to EICI. There have been many studies reporting introduction of this gene into B. subtilis increased MIC from 1.5 to that most of the patients suffering from EICI experienced gastro- 3 mg/L (Table 3). Taken together, this carboxypeptidase, along intestinal diseases concomitantly [6]. However, the results of this with D-alanine-D-alanine ligase and alanine racemase, may be study do not suggest taking intestinal gastrointestinal disorder to responsible for the vancomycin resistance in C. innocuum. be a prerequisite for EICI, at least for those caused by C. innocuum. Contribution from multiple genes also suggests that such resis- From our results, only 8.3% of the patients suffered from diarrhoea tance is unique and intrinsic to C. innocuum and unlikely to be and 16.7% had gastrointestinal perforation. Being a species of laterally transferred to other organisms. Clostridium, C. innocuum is remotely related to C. difficile and J.-H. Chia et al. / Clinical Microbiology and Infection 23 (2017) 560e566 565

Table 3 Contribution of genes of Clostridium innocuum to vancomycin resistance in Bacillus subtilis WB800

Plasmids Inserted genes (accession number) MIC (mg/L)

None None 0.38 pHY300PLK None 0.38 pHY300PLK þ Insert1 D-alanine-D-alanine ligase (KGJ51321.1) 1.5 Alanine racemase (KGJ51322.1) pHY300PLK þ Insert2 D-alanyl-D-alanine carboxypeptidase (KGJ52971.1) 0.38 pHY300PLK þ Insert1 þ Insert2 D-alanine-D-alanine ligase (KGJ51321.1) 3 Alanine racemase (KGJ51322.1) D-alanyl-D-alanine carboxypeptidase (KGJ52971.1)

C. perfringens. At least it does not possess any known clostridial when a specific species of Clostridium cannot be assured. In addi- toxins, unlike C. difficile and C. perfringens. It is possible that tion to metronidazole, piperacillin and ampicillin-sulbatam also C. innocuum possesses a unique virulence mechanism to cause appear to be appropriate for treating EICI. gastrointestinal as well as extra-intestinal infections, such as the In conclusion, the technology of MALDI-TOF MS or 16S rRNA lipopolysaccharide-like structure we identified in this study. Clos- sequencing allows us to explore the diversity of Clostridium species tridium difficile, a Gram-positive bacterium, also contains surface causing EICI. Although C. perfringens remains the most common, lipocarbohydrate, which has a similar activity to lipopolysaccharide the previously ignored C. innocuum was found to be the second in Gram-negative bacteria [23]. But this hypothesis needs experi- most common species. Due to the intrinsic vancomycin resistance mental verification. in C. innocuum, metronidazole, piperacillin and ampicillin- Recent antimicrobial use was considered to be another signifi- sulbatam instead of vancomycin are recommended as the first- cant risk factor for EICI, regardless of the species for the infection line agents to treat EICI. being C. difficile or C. septicum [2,7]. Antimicrobial use leads to an alteration of the intestinal microbiota, giving rise to overgrowth of Funding endogenous Clostridium or allowing colonization by nosocomial Clostridium. In our study, however, the rate of previous exposure to This study was supported by research grants, CMRPG3F0341, antimicrobials was not high. In fact, community-associated clos- OMRPG3A0031 and CLRPG3D0052 from Chang Gung Memorial tridial infection has been increasingly recognized, much of which Hospital and 101-2320-B-182A-006 from Ministry of Science and was not associated with previous administration of antimicrobials Technology, Taiwan. in the patients [24]. Most of the literature described how the 30-day mortality of Transparency declaration clostridial bacteraemia, caused by C. perfringens, C. difficile or by C. septicum [5,25],was 40% in general. The mortality from All authors declare that they have no competing interests. C. innocuum infections was reported to be even higher than for these species in some of the literature [7,15]. In the five cases of Appendix A. Supplementary data bacteraemia that we studied, there was no mortality. However, two important, but perhaps contradictory, issues should be further Additional Supporting Information may be found in the online considered. First, many of the EICI appeared to be polymicrobial version of this article can be found at http://dx.doi.org/10.1016/j. [26,27]. Accordingly, the deadly pathogen may not necessarily be cmi.2017.02.025. Clostridium. Second, most of the bacteraemic patients were elderly and have suffered from malignancy and diabetes mellitus before the infection [5,6,8,28]. In contrast, few patients with enteric References infection had underlying diseases. Hence, mortality of EICI may [1] Foroulis CN, Gerogianni I, Kouritas VK, Karestsi E, Klapsa D, Gourgoulianis K, primarily result from the underlying disease instead of from clos- et al. Direct detection of Clostridium sordellii in pleural fluid of a patient with tridial bacteraemia. Before the virulence of C. innocuum is clarified, pneumonic empyema by a broad-range 16S rRNA PCR. Scand J Infect Dis timely and appropriate antimicrobial therapy remains necessary in 2007;39:617e9. [2] Mattila E, Arkkila P, Mattila PS, Tarkka E, Tissari P, Anttila VJ. Extraintestinal terms of life-saving, especially in immunocompromised patients or Clostridium difficile infections. Clin Infect Dis 2013;57:e148e53. in those with multiple co-morbidities. [3] Young PE, Dobhan RR, Schafer TW. Clostridium perfringens spontaneous bac- Our data revealed that Clostridium isolates were generally sus- terial peritonitis: report of a case and implications for management. Dig Dis Sci 2005;50:1124e6. ceptible to antimicrobials. The resistance rate to clindamycin was [4] Chen YM, Lee HC, Chang CM, Chuang YC, Ko WC. Clostridium bacteremia: only 7%, with 100% susceptible in C. difficile. A survey in Canada emphasis on the poor prognosis in cirrhotic patients. J Microbiol Immunol showed that 74% of C. difficile isolates were resistant to clindamycin Infect 2001;34:113e8. [5] Leal J, Gregson DB, Ross T, Church DL, Laupland KB. Epidemiology of Clos- [5], which we attributed to the geographic difference. Currently the tridium species bacteremia in Calgary, Canada, 2000e2006. J Infect 2008;57: recommended treatment for C. difficile infection is metronidazole 198e203. or vancomycin [29,30]. The major advantage of metronidazole is its [6] Lee NY, Huang YT, Hsueh PR, Ko WC. Clostridium difficile bacteremia, Taiwan. e low cost, whereas that of oral vancomycin is its favourable phar- Emerg Infect Dis 2010;16:1204 10. [7] Libby DB, Bearman G. Bacteremia due to Clostridium difficiledreview of the macokinetic profile. Metronidazole is usually preferred for mild literature. Int J Infect Dis 2009;13:e305e9. cases and vancomycin for moderate and severe cases. Clostridium [8] Patel SB, Mahler R. Clostridial pleuropulmonary infections: case report and e innocuum is the only known Clostridium species expressing reduced review of the literature. J Infect 1990;21:81 5. fi [9] Woo PC, Lau SK, Chan KM, Fung AM, Tang BS, Yuen KY. Clostridium bacter- susceptibility to vancomycin. This study con rmed that aemia characterised by 16S ribosomal RNA gene sequencing. J Clin Pathol C. innocuum carries the genes necessary to express vancomycin 2005;58:301e7. resistance phenotype in the genome. Due to the relatively high [10] Ackermann G, Tang YJ, Jang SS, Silva J, Rodloff AC, Cohen SH. Isolation of Clostridium innocuum from cases of recurrent in patients with prior percentage of this species in causing EICI, we suggest that vanco- Clostridium difficile associated diarrhea. Diagn Microbiol Infect Dis 2001;40: mycin not be used routinely in the treatment of EICI, especially 103e6. 566 J.-H. Chia et al. / Clinical Microbiology and Infection 23 (2017) 560e566

[11] David V, Bozdogan B, Mainardi JL, Legrand R, Gutmann L, Leclercq R. Mech- a chromogenic agar-based culture method. J Med Microbiol 2011;60: anism of intrinsic resistance to vancomycin in Clostridium innocuum NCIB 945e9. 10674. J Bacteriol 2004;186:3415e22. [20] Bae JW, Park JR, Chang YH, Rhee SK, Kim BC, Park YH. Clostridium hastiforme is [12] Chilton CH, Crowther GS, Baines SD, Todhunter SL, Freeman J, Locher HH, et al. a later synonym of Tissierella praeacuta. Int J Syst Evol Microbiol 2004;54: In vitro activity of cadazolid against clinically relevant Clostridium difficile 947e9. isolates and in an in vitro gut model of C. difficile infection. J Antimicrob [21] Reynolds PE, Arias CA, Courvalin P. Gene vanXYC encodes d,d-dipeptidase Chemother 2014;69:697e705. (VanX) and d,d-carboxypeptidase (VanY) activities in vancomycin-resistant [13] Citron DM, Kwok YY, Appleman MD. In vitro activity of oritavancin gallinarum BM4174. Mol Microbiol 1999;34:341e9. (LY333328), vancomycin, clindamycin, and metronidazole against Clostridium [22] McGill F, Fawley WN, Wilcox MH. Monomicrobial Clostridium difficile bac- perfringens, Propionibacterium acnes, and anaerobic Gram-positive cocci. teraemias and relationship to gut infection. J Hosp Infect 2011;77:170e1. Anaerobe 2005;11:93e5. [23] Sanchez-Hurtado K, Poxton IR. Enhancement of the cytotoxic activity of [14] Tyrrell KL, Citron DM, Warren YA, Fernandez HT, Merriam CV, Goldstein EJ. Clostridium difficile toxin A by surface-associated antigens. J Med Microbiol In vitro activities of daptomycin, vancomycin, and penicillin against Clos- 2008;57:739e44. tridium difficile, C. perfringens, Finegoldia magna, and Propionibacterium acnes. [24] Naggie S, Frederick J, Pien BC, Miller BA, Provenzale DT, Goldberg KC, et al. Antimicrob Agents Chemother 2006;50:2728e31. Community-associated Clostridium difficile infection: experience of a veteran [15] Crum-Cianflone N. Clostridium innocuum bacteremia in a patient with ac- affairs medical center in southeastern USA. Infection 2010;38:297e300. quired immunodeficiency syndrome. Am J Med Sci 2009;337:480e2. [25] Ingram CW, Cooper JN. Clostridial bloodstream infections. South Med J [16] Alexander CJ, Citron DM, Brazier JS, Goldstein EJ. Identification and antimi- 1989;82:29e31. crobial resistance patterns of clinical isolates of Clostridium clostridioforme, [26] Garcia-Lechuz JM, Hernangomez S, Juan RS, Pelaez T, Alcala L, Bouza E. Extra- Clostridium innocuum, and Clostridium ramosum compared with those of intestinal infections caused by Clostridium difficile. Clin Microbiol Infect clinical isolates of Clostridium perfringens. J Clin Microbiol 1995;33:3209e15. 2001;7:453e7. [17] Clinical and Laboratory Standards Institute (CLSI). Methods for Antimicrobial [27] Jacobs A, Barnard K, Fishel R, Gradon JD. Extracolonic manifestations of Susceptibility Testing of Anaerobic Bacteria; Approved Standard-Seventh Clostridium difficile infections. Presentation of 2 cases and review of the Edition. Wayne, PA: Clinical and Laboratory Standards Institute (CLSI), CLSI literature. Medicine (Baltimore) 2001;80:88e101. document M11-A7; 2007. [28] Bodey GP, Rodriguez S, Fainstein V, Elting LS. Clostridial bacteremia in cancer [18] Clinical and Laboratory Standards Institute (CLSI). Performance Standards for patients. A 12-year experience. Cancer 1991;67:1928e42. Antimicrobial Susceptibility Testing M100S 26th ed. Clinical Laboratory and [29] Debast SB, Bauer MP, Kuijper EJ. European Society of Clinical Microbiology and Standards Institute. Wayne, PA: Clinical and Laboratory Standards Institute Infectious Diseases: update of the treatment guidance document for Clos- (CLSI), CLSI document M100S26; 2016. tridium difficile infection. Clin Microbiol Infect 2014;20(Suppl 2):1e26. [19] Seo JY, Kim PW, Lee JH, Song JH, Peck KR, Chung DR, et al. Evaluation of [30] Tart SB. The role of vancomycin and metronidazole for the treatment of PCR-based screening for vancomycin-resistant enterococci compared with Clostridium difficile-associated diarrhea. J Pharm Pract 2013;26:488e90.