Microbiological changes and diversity in autochthonous non-toxigenic Corynebacterium diphtheriae isolated in France


  • E. Farfour,

    1. Institut Pasteur, Unité Prévention et Thérapies Moléculaires des Maladies Humaines, Centre National de Référence des Corynébactéries du Complexe Diphtheriae, Paris, France
    2. CNRS-URA 3012, Paris, France
    Search for more papers by this author
  • E. Badell,

    1. Institut Pasteur, Unité Prévention et Thérapies Moléculaires des Maladies Humaines, Centre National de Référence des Corynébactéries du Complexe Diphtheriae, Paris, France
    2. CNRS-URA 3012, Paris, France
    Search for more papers by this author
  • S. Dinu,

    1. ‘Cantacuzino’ National Institute of Research-Development for Microbiology and Immunology, Molecular Epidemiology Laboratory, Bucharest, Romania
    Search for more papers by this author
  • S. Guillot,

    1. Institut Pasteur, Unité Prévention et Thérapies Moléculaires des Maladies Humaines, Centre National de Référence des Corynébactéries du Complexe Diphtheriae, Paris, France
    2. CNRS-URA 3012, Paris, France
    Search for more papers by this author
  • N. Guiso

    Corresponding author
    1. Institut Pasteur, Unité Prévention et Thérapies Moléculaires des Maladies Humaines, Centre National de Référence des Corynébactéries du Complexe Diphtheriae, Paris, France
    2. CNRS-URA 3012, Paris, France
    • Corresponding author: N. Guiso, Institut Pasteur, Unité Prévention et Thérapies Moléculaires des Maladies Humaines, Centre National de Référence des Corynébactéries du Complexe Diphtheriae, CNRS-URA 3012, 25, rue du Dr Roux 75015 Paris, France

      E-mail: nicole.guiso@pasteur.fr

    Search for more papers by this author


Autochtonous toxigenic Corynebacterium diphtheriae have disappeared in mainland France, but non-toxigenic C. diphtheriae are still circulating. Using phenotypic and molecular tools, we retrospectively characterized 103 non-toxigenic C. diphtheriae collected in mainland France and highlight several changes. The proportion of C. diphtheriae belfanti increased between 1977 and 2011 and it is the most frequent biotype recovered in recent years. Resistance to ciprofloxacin has increased and most isolates with decreased sensitivity belong to the belfanti biotype. Using multilocus sequence typing, we demonstrate that French isolates are distributed in a large number of sequence types and identify three distinct lineages. C. diphtheriae mitis and gravis form lineage I while C. diphtheriae belfanti forms lineages II and III. Almost all isolates of lineage II are part of a unique clonal complex or are very close to it. Most French isolates have a dtxR sequence homologous to that of toxigenic isolates, suggesting that if lyzogenised by a corynephage, they can express diphtheria toxin.


The introduction of vaccination with diphtheria toxoid in the early 20th century changed diphtheria epidemiology substantially. The disease has become rare in countries with high vaccination coverage but it remains endemic in several other countries, especially in tropical areas. Over the last 30 years, several significant changes have been observed in countries with vaccination programmes. Firstly, toxigenic C. diphtheriae infections have become very much less frequent, and many of the cases observed are imported from endemic areas. Secondly, autochtonous toxigenic infections are caused by C. ulcerans, mostly transmitted by pets [1-3]. Furthermore, a recent study demonstrated an asymptomatic carriage of 8% for dogs in Osaka [4]. Contrasting with the decreased prevalence of toxigenic C. diphtheriae, recent autochtonous cases have been found to be associated with various non-toxigenic C. diphtheriae isolates. In England and Wales, the number of non-toxigenic isolates increased between 1986 (four isolates) and 2006 (68 isolates), peaking at 294 isolates in 2000 (http://www.hpa.org.uk/). Other countries with vaccination programmes have reported similar observations [5-10]. Clinical features of the infections due to non-toxigenic isolates may be indistinguishable from those due to toxigenic isolates, such as cutaneous infections and pharyngitis. Moreover, they have been associated with invasive infections such as bacteraemia, endocarditis, and bone and joint infections [7, 11, 12].

Here, we report a description and analysis of the diversity of non-toxigenic C. diphtheriae causing autochtonous infections in mainland France.

Materials and Methods

Clinical isolates

Non-toxigenic C. diphtheriae isolates of the collection of the French National reference centre (NRC) for Corynebacteria of the diphtheriae complex were included in this study (n = 103). These isolates, consecutively collected throughout mainland France between 1977 and 2011, were either sent by clinical laboratories for confirmation of identification and toxigenic assessment, or provided by the Institut Pasteur Collection. Twenty-four of these isolates have previously been characterized [7, 13].

Molecular identification, phenotypic characterization and antibiotic susceptibility

All isolates were identified using a specific PCR for the dtxR gene [14]. Biotypes were determined using the API Coryne strip test (product no. 20900; bioMérieux, Marcy l'Etoile, France) and interpreted according to the World Health Organisation guidelines [15]. Toxigenic status was assessed using a PCR for the tox gene [16]. The sensitivities to 16 antibiotics were determined with a disc diffusion method according to the French reference guidelines (CA-SFM) [17]. As there are no specific guidelines for Corynebacterium species, the sensitivity was interpreted using the non-species-related breakpoint of the CA-SFM. For isolates with decreased sensitivity, the minimum inhibitory concentrations (MICs) were evaluated by E-test (bioMérieux)

Multilocus sequence typing (MLST)

Multilocus sequence typing was performed as previously described [18]. Sequence types (STs) were assigned by comparing the sequences of each locus with similar sequences deposited in the online PubMLST database (http://pubmlst.org/cdiphtheriae/). New alleles and new STs were submitted to this database.

A single isolate for each ST identified in this work, as well as previously published STs, was included in this analysis [6, 8, 18]. Clonal analysis was performed using the eBurst program (www.mlst.net). A clonal complex or eBurst group was defined as a cluster of isolates sharing six of seven alleles.

Phylogenetic analysis

Phylograms were constructed using the Jukes-Cantor algorithm and the neighbour-joining method with the MEGA program version 5 [19].

Sequencing of the dtxR gene

The dtxR gene was amplified and sequenced using the protocol from Nakao et al. [20] and the TAQ PCR core kit (Qiagen, Courtaboeuf, France). Amino-acid sequences were deduced from nucleotide sequences using EMBOSS software (www.mobyle.pasteur.fr) [21]. Both nucleotide and amino-acid sequences were compared with those available in the GenBank database and new sequences were submitted to the EMBL database.


We investigated 103 isolates collected between 1977 and 2011. They were mainly isolated from respiratory (n = 42), blood (n = 30) and cutaneous (n = 16) samples (Table 1). Other sites included lymphatic node (n = 1), eye (n = 1), ear (n = 1) and bone or joint (n = 5). For seven isolates, the origin of the biological samples was not mentioned.

Table 1. Origin and biotype of non-toxigenic C. diphtheriae
Clinical samplesBiotypes
Bone or joint410

Microbiological characteristics

Corynebacterium diphtheriae mitis was the most common biotype (n = 50; 48.5%) followed by belfanti (n = 37; 36.0%) and gravis (n = 16; 15.5%). Belfanti isolates were mostly collected from respiratory samples (31/37) and many mitis isolates were from sterile sites, blood (25/50) and bone or joint (4/50), and cutaneous (10/50) samples (Table 1). Gravis and mitis isolates were obtained throughout the period, whereas most belfanti isolates were from cases after 1996 (29 of 37 isolates). Similarly, 16% of isolates before 1995 and 52% after this date were belfanti (Fig. 1). There was a peak of mitis isolates, including almost all isolates collected from blood, bone and joint, between 1990 and 1993.

Figure 1.

Biotype distribution among C. diphtheriae isolated between 1977 and 2011. The biotypes of Corynebacterium diphtheriae isolates collected between 1977 and 2011 by the French National Reference Centre (NRC) for Corynebacteria of the diphtheriae complex were determined as indicated in 'Materials and Methods'

Antimicrobial susceptibilities

Almost all isolates were susceptible to penicillin G (99%) and erythromycin (98.1%). Lower sensitivity or resistance was detected for 12 of 16 antibiotics tested. No isolate was resistant to amoxicillin or other beta-lactams tested. The least active antibiotics were tetracycline, rifampicin and ciprofloxacin. All but one of the isolates with reduced sensitivity to ciprofloxacin were biotype belfanti, isolated since the year 2000. Multiple resistance was rare and all the isolates resistant to macrolides were susceptible to penicilin G. One isolate was intermediate to penicillin G (MIC, 2 μg/mL), cefotaxime (MIC, 2 μg/mL) and cotrimoxazole (MIC of cotrimoxazole, 3 μg/mL; MIC of sulphamethoxazole, >1024 μg/mL and MIC of trimethoprim, 8 mg/μL) and resistant to tetracycline (MIC, 32 μg/mL). Isolates collected since 1997 had a decreased sensitivity to 10 antibiotics (Fig. 2).

Figure 2.

Antibiotic resistance of C. diphtheriae isolates before and after 1997. The antibiotic resistance of C. diphtheriae isolates collected between 1977 and 2011 by the French National Reference Centre (NRC) for Corynebacteria of the diphtheriae complex was determined as indicated in 'Materials and Methods'

Multilocus sequence typing

Multilocus sequence typing analysis allowed us to identify 63 STs, including the eight previously described among a series of 24 C. diphtheriae isolates causing invasive infections in France (ST128, ST130, ST153, ST156, ST193, ST194, ST195 and ST196) (14). Clonal analysis classified these French isolates into six eBurst groups, three of which (eBurst groups 1, 3 and 11) were previously described by Bolt et al. [18]. ST5 and ST32 were clustered in the eBurst groups 11 and 1, respectively.

Almost all C. diphtheriae belfanti isolates (18 out of 25) are part of eBurst group 3. The remaining belfanti isolates are closely related to this eBurst group. Indeed, four belfanti isolates (i.e. ST93, ST180, ST207 and ST215) are double locus variants (DLV) of various isolates of eBurst group 3 and ST214 forms an eBurst group 14 with ST215 (Fig. 3; Table 2). The eBurst groups 12 and 13 and 14 are also each composed of two STs.

Table 2. Sequence type (ST) and eBurst group of non-toxigenic C. diphtheriae
eBurst groupSTNb isolatesYear of isolation
Figure 3.

Minimum spanning tree of C. diphtheriae belfanti isolates from France and other sequence types (STs) of eBurst group 3. Each circle corresponds to a sequence type. The relationships between isolates are indicated by the connections between them and numbers between the circles indicate the number of allelic differences. Black lines connecting pairs of STs are related to the number of alleles they differ by: one or two alleles (thick lines), and three to seven alleles (dashed lines). Grey zones between circles indicate that they belong to the same clonal complex.

There was no linkage between ST or eBurst group and clinical presentation or biotype, except for ST130 and eBurst group 3: ST130 accounted for 23 of 24 isolates of the mitis peak between 1990 and 1993 and most isolates were involved in bacteraemia between 1991 and 1993; eBurst group 3 was mostly associated with respiratory infections.

Phylogenetic analysis

Three distinct lineages were observed by phylogenetic analysis (Fig. 4). Lineage I was composed of mitis and gravis isolates, and all belfanti isolates were in lineages II and III. To assess the degree of recombination within the bacterial population, the index association (IA) was calculated for the overall population and for each lineage [22]. The overall IA value was 0.348 (p <0.001); the IA values were 0.0841 (p 0.002) for lineage I and 0.1113 (p <0.001) for lineage II, suggesting that C. diphtheriae lineage I and II are in linkage disequilibrium and may be considered as clonal.

Figure 4.

Phylogenic representation using the neighbour-joining method corrected with the Jukes-Cantor algorithm for C. diphtheriae isolates. A representative isolate of each sequence type (ST) recovered in mainland France was included. C. diphtheriae belfanti isolates are underlined. The eBurst group identified by Bolt et al[18] and a new eBurst group are indicated: (▼) eBurst group 1, (●) eBurst group 3, (♦) eBurst group 11, (■) eBurst group 12, (▲) eBurst group, (○) eBurst group 14.

dtxR gene analysis

Among the 63 representative sequence types, we identified 12 different nucleotide sequences of the dtxR gene, corresponding to six different amino-acid sequences. One DtxR polypeptide is encoded by six different nucleotide sequences whereas the others are each encoded by a single nucleotide sequence.

Four protein sequences are identical to those of toxigenic isolates, i.e. the reference strain PW8, isolate NCTC13129 and two clinical isolates from Romania (C. diphtheriae 9337 and 25298) [23-25]. The GeneBank database contains no sequence identical to the remaining proteins (accession numbers HE980017 and HE980018). Each protein sequence is harboured by a single isolate.


The disappearance of autochthonous diphtheria due to toxigenic C. diphtheriae in mainland France and in countries with high vaccination coverage is a direct consequence of the large-scale use of diphtheria toxoid [1]. A switch in the populations causing autochtonous infections from toxigenic to non-toxigenic isolates in the 1990s and the 2000s has been documented [5, 6, 26, 27]. Here, we describe recent changes in the French C. diphtheriae population, and draw conclusions about the circulation of these bacteria.


Corynebacterium diphtheriae belfanti isolates increased as a proportion of the population during the study period. This biotype seems to have a respiratory tropism whereas other biotypes mostly cause cutaneous or invasive infections. Interestingly, C. diphtheriae belfanti has rarely been reported in the literature, and almost exclusively in vaccinated countries (Canada, the United States and France) where there is surveillance of such infections [18]. As most toxigenic isolates are mitis and gravis, it is possible that the change of biotype is a consequence of the disappearance of toxigenic isolates at the end of the twentieth century. C. diphtheriae belfanti might have an advantage over non-toxigenic mitis and gravis isolates for colonizing or infecting the upper respiratory tract.

Antibiotic susceptibility

Most autochthonous French isolates are sensitive to first-line antibiotics prescribed for C. diphtheriae infection, confirming the validity of the current therapeutic guidelines [28-30]. However, we observed a trend among recent isolates for diminished susceptibility to ten antibiotics, indicating that antibiotic surveillance should be continued. Only one isolate displayed cross-resistance to several classes of antibiotics (penicillin G, cefotaxim, tetracycline and cotrimoxazol). Multidrug resistance seems to be rare among C. diphtheria; only one such isolate has previously been described in Canada [31]. Almost 14% of our isolates displayed reduced sensitivity to ciprofloxacin, and all were biotype belfanti collected in the 2000s. This apparent emerging resistance could be a consequence of the use of antimicrobial agents, as described for Streptococcus pneumoniae and the increased use of fluoroquinolones [32, 33]. Similarly, the treatment of unrecognized C. diphtheriae upper respiratory tract infections with fluoroquinolones, or the use of fluoroquinolones in general, may be having an effect on the C. diphtheriae bacterial population. These possibilities are supported by the rare use of ciprofloxacin for C. diphtheriae treatment and the observation of decreased sensitivity only among recently collected isolates. This in turn suggests that asymptomatic carriage of C. diphtheriae is not uncommon, but C. diphtheriae throat carriage has been described to be rare or absent among patients with upper respiratory tract infections [34] and asymptomatic individuals [35]. However, these studies involved few socially/professionally active and healthy subjects [35] and bacterial flora of the upper respiratory tract in infected patients may not be representative of that in the general population [34]. Asymptomatic throat carriage may also concern only a particular sub-population or may be transient, and cutaneous carriage has not been evaluated. In temperate countries, skin infections mostly affect people with predisposing conditions and those who are socio-economically disadvantaged [6, 7].

Multilocus sequence Typing

Isolates from mainland France are distributed across a very large number of STs. Most of them have only been found in France and only seven (ST5, ST23, ST32, ST42, ST80, ST81 and ST128) of 63 have previously been identified in another country: Canada, the USA, Brazil, Poland, Russia and Kazakhstan [8, 18]. However, the MLST technique was only recently developed for C. diphtheriae typing and the number of isolates analysed using this technique is still small. Furthermore, despite the large number of STs identified, the isolates cluster in a small number of eBurst groups. Other than eBurst group 14, which has only been found on mainland France, all eBurst groups recovered in this study are geographically dispersed, and are represented in Western Europe (eBurst groups 1, 3, 11 and 13), Eastern Europe (eBurst groups 1 and 11), North America (eBurst groups 1, 3 and 12) and South America (eBurst group 13) [8, 18]. Interestingly, the larger eBurst group includes almost all C. diphtheriae belfanti isolates and is exclusively located in North America and Western Europe, geographical areas with high vaccination coverage and now almost no autochthonous toxigenic isolates. Note that MLST has mostly been applied to clinical isolates collected from Eastern Europe, Western Europe and North America. This may explain the geographical distribution of eBurst groups described and especially the absence of eBurst group representatives from Africa and Asia.

Phylogenic analysis

French C. diphtheriae isolates are distributed between three distinct lineages. Two lineages (lineage I and II) have already been described by Bolt et al. [18]. We demonstrate that almost all of our C. diphtheriae belfanti isolates belong to a single clonal complex and the IA suggests that they are clonal. Lineage III is composed of a single isolate (ST 181) and is phylogenetically distant from the two other lineages. It was collected from a woman presenting with rhinitis. ST181 does not share any allele with other French C. diphtheriae belfanti. We compared the nucleotide sequences at each locus by NCBI blast, and the rpoB sequence was homologous to that of a C. diphtheriae belfanti isolate collected from the ear of a cat [35]. The feline isolate carries the tox gene but does not express the diphtheria toxin. Indeed, the nucleotide sequence of the tox gene shows several substitutions and deletions, which prematurely terminate the peptide at amino acid 25. Furthermore, it has a higher sequence identity with the C. ulcerans tox gene than the C. diphtheriae tox gene. Unfortunately, the feline isolate was not characterized by MLST. It would be of interest to determine the sequence type of feline isolates to find out if lineage III is composed of animal isolates and if C. diphtheriae could be transmitted between humans and animals.


In toxigenic C. diphtheriae, the expression of the diphtheria toxin is regulated by the diphtheria toxin repressor (DtxR) [37, 38]. This transcriptional regulator binds in an iron-dependent way to operators of several genes, including those involved in iron uptake and the tox gene. Under conditions of low iron availability, as in the throat, DtxR does not repress the tox gene, which consequently can be transcribed. Several French non-toxigenic isolates harbour the same nucleotide or amino-acid sequence of the dtxR gene and protein as toxigenic isolates, suggesting that, if lysogenised by a beta-corynephage harbouring the tox gene, they could produce the diphtheria toxin.


We report evidence of recent changes in the C. diphtheriae population of mainland France. It appears that toxigenic C. diphtheriae mitis and gravis isolates were replaced at the end of the twentieth century by non-toxigenic C. diphtheriae mitis, gravis and above all belfanti biotypes. The belfanti biotype seems to have a respiratory tropism and its population structure is that of a large clonal complex distinct from other biotypes. There is increasing resistance to several antibiotics, including ciprofloxacin, suggestive of asymptomatic carriage or unrecognized C. diphtheriae infections. Finally, sequencing of the dtxR gene suggests that if lysogenised by a corynephage harbouring the tox gene, several French autochthonous isolates may express the diphtheria toxin.


We thank the Collection of the Institut Pasteur for providing some of the isolates. This work was supported by the Institut Pasteur Foundation (Paris, France), the CNRS-URA 3012 and the Institut National de Veille Sanitaire (Saint Maurice, France).

Transparency Declaration

We warrant that all authors have seen and approved the content and the order of their names on the title page and have agreed that the corresponding author may act on their behalf throughout the editorial review and publication process. We certify that all potential conflicts of interest have been disclosed.