Abstract
In France, human botulism is mainly food-borne intoxication, whereas infant botulism is rare. A total of 99 group I and II Clostridium botulinum strains including 59 type A (12 historical isolates [1947–1961], 43 from France [1986–2013], 3 from other countries, and 1 collection strain), 31 type B (3 historical, 23 recent isolates, 4 from other countries, and 1 collection strain), and 9 type E (5 historical, 3 isolates, and 1 collection strain) were investigated by botulinum locus gene sequencing and multilocus sequence typing analysis. Historical C. botulinum A strains mainly belonged to subtype A1 and sequence type (ST) 1, whereas recent strains exhibited a wide genetic diversity: subtype A1 in orfX or ha locus, A1(B), A1(F), A2, A2b2, A5(B2′) A5(B3′), as well as the recently identified A7 and A8 subtypes, and were distributed into 25 STs. Clostridium botulinum A1(B) was the most frequent subtype from food-borne botulism and food. Group I C. botulinum type B in France were mainly subtype B2 (14 out of 20 historical and recent strains) and were divided into 19 STs. Food-borne botulism resulting from ham consumption during the recent period was due to group II C. botulinum B4. Type E botulism is rare in France, 5 historical and 1 recent strains were subtype E3. A subtype E12 was recently identified from an unusual ham contamination. Clostridium botulinum strains from human botulism in France showed a wide genetic diversity and seems to result not from a single evolutionary lineage but from multiple and independent genetic rearrangements.
Introduction
Botulinum neurotoxins (BoNTs) are the most potent toxins which are responsible for a rare but severe neurological disorder called botulism. The disease is characterized by a flaccid paralysis and decreased secretions which result from the BoNT-dependent inhibition of acetylcholine release at neuromuscular junctions and other cholinergic endings. BoNTs are divided into seven toxinotypes based on their antigenicity properties in neutralization assay ( Popoff 1995 ; Peck et al. 2011 ; Barash and Arnon 2014 ; Dover et al. 2014 ). Thereby, neutralizing antibodies are specific of each toxinotype. All BoNT types induce similar pharmacological effects, which are characterized by flaccid paralysis, but with some differences between toxinotypes like duration and intensity of symptoms. For example, BoNT/A induces the longest and most severe forms of botulism, compared with BoNT/E which leads to shorter duration symptoms and BoNT/B which is most often responsible for mild botulism illness ( Eleopra et al. 1998 ; Keller et al. 1999 ; O’Sullivan et al. 1999 ; Foran et al. 2003 ; Meunier et al. 2003 ; Keller 2006 ). BoNT/B causes predominantly dysautonomic signs, whereas BoNT/A results in a pronounced paralytic effect on the respiratory muscles and most often leads to acute respiratory distress ( Jenzer et al. 1975 ; Hughes et al. 1981 ; Merz et al. 2003 ; Sobel 2005 ; Potulska-Chromik et al. 2013 ). In addition, each toxinotype is subdivided into several subtypes according to amino acid sequence variations. Variability is observed in bont genes, and amino acid sequence difference of at least 2.6% is assumed to define two distinct subtypes ( Smith et al. 2005 ). Variations in BoNT amino acid sequences might affect some aspects of their activity such as binding to target cells, efficiency of entry into cells, potency of enzymatic activity, duration of effects, neutralization efficiency by antibodies specific of type and subtype. For example, BoNT/A2 enters cells more rapidly and more efficiently, and is a more potent neuromuscular blocker in vivo than BoNT/A1 ( Pier et al. 2011 ; Torii et al. 2011 ). In addition, differences in in vitro and in vivo activity have been reported between subtypes BoNT/A1 to A5 using neuronal cell based and mouse assays, respectively ( Whitemarsh et al. 2013 ). In contrast to BoNT/A1, A2, A4 and A5, the persistence of BoNT/A3 in rat spinal neurons is shorter ( Whitemarsh et al. 2014 ). Moreover, antibodies against BoNT/A2 neutralize more efficiently BoNT/A2 than BoNT/A1, whereas antibodies against BoNT/A1 counteract efficiently both toxin subtypes ( Torii et al. 2013 ). However, the impact of the subtypes in naturally acquired botulism is not yet well understood. Moreover, the recent controversy about the reported new BoNT type H, rather referred as a new hybrid F/A, is mainly based on the existence or absence of protection with antibodies against already known BoNT types and subtypes, and further raises the importance of subtype determination ( Barash and Arnon 2014 ; Dover et al. 2014 ; Gonzalez-Escalona, Thirunavukkarasu, et al. 2014 ).
BoNTs are synthesized by Clostridium botulinum and atypical strains of C. butyricum and C. baratii. The bont genes are clustered with the genes of BoNT-associated nontoxic proteins (ANTPs) including hemagglutinins (HAs) or OrfX in a DNA fragment called the botulinum locus. The botulinum loci are located either on the chromosome, large plasmid, or phage depending on strains. Clostridium botulinum strains are heterogeneous at the phenotypic and genetic levels, and are classified into four groups (I–IV), group IV being assigned to a distinct species, C. argentinense . Two additional groups have been included, group V and VI which encompass BoNT/F-producing C. baratii strains, and BoNT/E-producing C. butyricum strains, respectively. In each group, C. botulinum strains exhibit genomic variability as evidenced by pulsed-field gel electrophoresis (PFGE), amplified fragment length polymorphism (AFLP), multilocus sequence typing (MLST), multilocus variable number tandem repeat analysis (MLVA), and single nucleotide polymorphism (SNP) based on whole genome sequencing (WGS) (review in Brüggemann et al. 2011 ; Hill and Smith 2013 ; Popoff and Bouvet 2013 ; Carter and Peck 2015 ; Smith, Hill, Raphael 2015 ). It is noteworthy that WGS is a powerful approach allowing not only to have access to detailed phylogenetic relatedness of neurotoxin-producing strains and toxin genes, but also to highlight the genomic modifications and genetic exchanges between strains ( Peck et al. 2011 ; Smith , Hill, Xie, et al. 2015 ).
Human botulism is rare but recurrently identified in France. From 6 to 31 human cases of botulism (6–27 outbreaks) are reported every year in our country ( Carlier et al. 2007 ; Mazuet et al. 2011 , 2014 ). Food-borne botulism is the most common form, and more especially botulism type B subsequently to the consumption of home-cured ham and pork products. However, since 2005, an increased number of botulism type A outbreaks was identified, which resulted in severe forms requiring hospitalization in intensive care units and mechanical respiratory ventilation for several weeks or months and one death was reported. Various food types were incriminated including home-made preparations, locally performed or imported from foreign countries, as well as industrial products. The incidence of food-borne botulism is variable in the European countries, and France with Germany, Spain and Italy are those with the highest number of reported cases within the period 1988–1998, not included the Eastern European countries ( There 1999 ). Differences in food practices and notably in use of homemade canned foods likely account for the variable incidence of botulism in Europe. In contrast, infant botulism and wound botulism were more rare in France, 12 cases and 1 case, respectively, within the period 2004–2013 ( Carlier et al. 2007 ; King et al. 2010 ; Mazuet et al. 2011 , 2014 ). Infant botulism is prevalent in some countries such as United States and Argentina, whereas the incidence of this form of botulism is low in Europe. However, infant botulism is more frequent in Italy than in other European countries, 26 cases in the period 1984–2006 versus 1–9 cases in the other countries ( Koepke et al. 2008 ). Differences in the dissemination of neurotoxigenic Clostridium in the environment which seems to be the major source of contamination of the babies or differences in feeding practices, notably the distribution of honey to newborns in some countries ( Aureli et al. 2002 ), likely reflect the differences in infant botulism incidence according to the countries. It is noteworthy that the comparison of epidemiological data from the different countries is not highly significant because the identification and notification of botulism cases are variable in each country.
The aim of this study is to analyze the genetic diversity of C. botulinum strains type A, B and E isolated in France in the recent years (1985–2013) versus historical strains (1947–1961) and some strains from other countries.
Materials and Methods
Bacterial Strains, Growth, Toxinotyping and DNA Preparation
The isolation of C. botulinum strains from clinical and food samples was performed with conventional culture method using fortified cooked meat medium (Becton Dickinson, France) and agar selective medium ( Mazuet, Sautereau, et al. 2015 ). Clostridium strains were grown in Trypticase yeast extract glucose (TGY) broth in an anaerobic atmosphere at 37 °C. Phenotypic identification was performed with reference methods ( Jousimies-Somer et al. 2002 ).
Toxin production and BoNT typing of each strain were determined by mouse test using anti-BoNT sera of the National Reference Center of Anaerobic bacteria and Botulism. All animal experiments were conducted with the approval of Institut Pasteur (agreement of laboratory animal use no 2013-0116). The presence of corresponding bont genes was confirmed using PCR methods as previously described ( Vanhomwegen et al. 2013 ).
Total genomic DNA was extracted from C. botulinum cultures by lysozyme and proteinase K treatment as described previously ( Dineen et al. 2003 ).
PCR Amplification and DNA Sequencing
Overlapping pairs of primers covering the whole bont sequences were designed for PCR amplification using sequence data available in GenBank. PCR amplifications were performed in 50 μl reaction buffer containing 200 ng DNA, 1 μM of each primer (Sigma Aldrich, Germany), 1.5 mM MgCl 2 , 250 mM of each dNTP, 20 mM Tris–HCl, pH 8.4, 50 mM KCl and 1 μ Taq Polymerase (In Vitrogen, France). The PCR cycles consisted of 90 °C for 45 s, 50 °C for 45 s, and 72 °C for 1 min were repeated 30 times. A final extension step of 72 °C for 10 min was added. Amplicons were sequenced by EUROFINS/MWG, Germany.
Whole Genome Sequencing
Whole genome sequencing using the NEBNext Ultra DNA Library Prep kit for Illumina (New England Biolabs) were performed using MiSeq machine (Illumina) in paired-end reads of 250 bases or on HiSeq2000 machine (Illumina) in single reads of 96 bases or 101 bases. Sequence files were generated using Illumina Analysis Pipeline version 1.8 (CASAVA). After quality filtering, reads were assembled using CLC software version 4 (CLC Bio).
Multilocus Sequence Typing Analysis of Group I Strains
Multilocus sequence typing of group I strains was based on seven housekeeping genes ( aroE, mdh, aceK, appB, rpoB, recA , and hsp ) as previously described ( Jacobson et al. 2008 ). Partial nucleotide sequences of the seven housekeeping genes were obtained by PCR amplification and subsequent sequencing according to the method described by Jacobson et al. (2008) . For some strains, nucleotide sequences of housekeeping genes were deduced from WGS ( table 1 ). MLST data were submitted to the C. botulinum MLST database ( http://pubmlst.org/cbotulinum /, last accessed May 10, 2016). Allelic numbers and MLST sequence types (STs) were identified by querying C. botulinum MLST database. More than one nucleotide difference was regarded as a criterion for a different ST. New alleles and new STs were submitted to the C. botulinum MLST database.
Clostridium botulinum Strains, Typing and Origin
 |
 |
 |
|
|
|
N ote .—The colors indicate C. botulinum A strains belonging to the same ST. Vertical bars on the left side indicate that the strains were from a same outbreak.
*All the strains have been isolated in France, except the strains noted with *: B697 (Sweden), 424-86 (Brazil), 181-02, 126-07 and 211-02 (Switzerland), 277-00 (Iran) ( Mazuet, Bouchier, et al. 2015 ), BL5 and BL6 (United Kingdom), HV and HV2 (United States, San Francisco, Mayer’s collection), Ped2 (Greenland). Strains Ba657 ( Hatheway et al. 1981 ), NCTC7273 ( Bowmer 1963 ), and ATCC9009 were included as reference strains.
Strain nonisolated. DNA sequences obtained by PCR with DNA extracted directly from samples and/or after enrichment culture of the samples.
WGS, whole genome sequencing; nd, not defined; ST, sequence type.
Clostridium botulinum Strains, Typing and Origin
|
|
|
|
|
|
N ote .—The colors indicate C. botulinum A strains belonging to the same ST. Vertical bars on the left side indicate that the strains were from a same outbreak.
*All the strains have been isolated in France, except the strains noted with *: B697 (Sweden), 424-86 (Brazil), 181-02, 126-07 and 211-02 (Switzerland), 277-00 (Iran) ( Mazuet, Bouchier, et al. 2015 ), BL5 and BL6 (United Kingdom), HV and HV2 (United States, San Francisco, Mayer’s collection), Ped2 (Greenland). Strains Ba657 ( Hatheway et al. 1981 ), NCTC7273 ( Bowmer 1963 ), and ATCC9009 were included as reference strains.
Strain nonisolated. DNA sequences obtained by PCR with DNA extracted directly from samples and/or after enrichment culture of the samples.
WGS, whole genome sequencing; nd, not defined; ST, sequence type.
Multilocus Sequence Typing Analysis of Group II Strains
Multilocus sequence typing of group II strains was based on 12 housekeeping gene sequences ( oppB, pta, recA, rpoB, pyc, gyrB, lepA, ilvD, guaA, atpD, trpB, 23S ) as described in ( Macdonald et al. 2011 ; Weedmark et al. 2015 ). Allele sequences were concatenated and were aligned with Muscle (default parameters). Maximum Likelihood phylogenetic tree was built based on the Kimura 2-parameter model ( Kimura 1980 ). About 1,000 boostrap experiments were performed to assess the robustness of the topology and percent are indicated on nodes. Evolutionary analyses were conducted in MEGA ( Tamura et al. 2013 ).
Nucleotide Sequence Accession Numbers
All the novel MLST alleles described in this project have been deposited at GenBank database: hsp60 , KF681499 to KF681507; recA KF681489 to KF681498; rpoB , KF681478 to KF681488; oppB KF681460 to KF681477; acek , KF681440 to 681459 ; mdh, KF681425 to 681439; aroE , KF681410 to KF681424.
The project has been deposited at GenBank under the accession numbers KF929215 and KM370319 for bont sequences type E12 ( Mazuet, Sautereau, et al. 2015 ), JQ954969 for bont/A7 ( Becher et al. 2007 ), and KF667385 for bont/A8 .
Results
Clostridium botulinum Strains
The C. botulinum strains investigated in this study are listed in table 1 . A total of 99 strains were analyzed including 59 C. botulinum type A (54 isolates from France, 4 from other countries, and 1 reference strain), 31 C. botulinum type B, and 9 C. botulinum type E strains ( table 1 ).
The C. botulinum type A strains included 12 historical strains (Prévot’s collection, 1947–1953) including 11 from French animal or food samples and 1 from Sweden, 1 reference strain isolated in 1976, as well as 46 strains isolated during the period 1986–2013: 26 strains were associated with food-borne botulism (13 outbreaks) in France and 1 in Switzerland, 14 with infant botulism (9 cases, France), 3 from food (France), 1 from a wound botulism in Switzerland, and 1 from cattle (Brazil) ( table 1 ).
Three C. botulinum type B strains (historical strains) were from the Prévot’s collection (1953–1961) including 2 from food-borne botulism and 1 from cattle botulism. A reference strain subtype B1 isolated in United Kingdom before 1947 is included. 23 C. botulinum type B strains were isolated in France during the period 1986–2013, 19 from food-borne botulism (18 outbreaks), 1 from food, and 3 from infant botulism. In addition, 4 C. botulinum type B strains from other countries (2 from United Kingdom, 1 from Switzerland, and 1 from Middle East,) were included ( table 1 ).
Botulism type E is very rare in France. A total of 9 strains were investigated including 1 reference strain, 5 historical strains originated from France, United States, and Greenland (Prévot’s collection), and 3 from recent human botulism cases in France (2 from food-borne botulism and 1 from intestinal colonization) ( table 1 ).
Clostridium botulinum Type A and Subtypes
BoNT and ANTP genes were sequenced via specific PCR DNA amplification and sequencing and/or by WGS. Variations in bont genes were analyzed at the nucleotide and amino acid levels. Both nucleotide and amino acid sequence variations yielded similar level of subtype subdivision in each C. botulinum type ( table 1 andfig. 1 ).
Phylogenetic relatedness of BoNT/A nucleotide sequences and deduced proteins. The dendograms were constructed using the UPGMA method. The genetic distances were computed by using the Kimura two-parameter model. The scale bar indicates similarity values. The numbers next to each node indicate the cophrenetic correlation. Evolutionary analyses were conducted in Bionumerics (V.6.6 Applied Maths). Strains noted * , ** , # were isolated from a same botulism outbreak.
The 58 C. botulinum type A isolates were identified as subtype A1 (15 strains), A2 (18 strains), A1(B) (17 strains), A5(B′) (3 strains), A1(F) (1 strain), A2(B) (2 strains) and 2 recently identified subtypes termed A7 (1 strain) and A8 (1 strain). No subtype A3 or A4 was detected. All the 12 historical strains belong to the subtype A1, whereas only 2 strains of 46 from the recent period were assigned to the subtype A1. However, these 2 strains contain bont/A1 in OrfX locus, versus ha-bont/A1 locus in the historical strains. In contrast, the strains isolated from food-borne botulism and food during the recent period were distributed in several subtypes (A1, A2, A5, A7, A8), and most of them (22 of 30) were bivalent strains [A1(B), A1(F), A2(B), A5(B′)]) ( table 1 and fig. 1 ). The strains from a same outbreak shared the same bont type and subtype, except in two outbreaks where samples contained a mix of several C. botulinum types and subtypes, A1(B)/B5f2 and B/E12, respectively ( table 1 ). Clostridium botulinum A1(B) was responsible for two large outbreaks of botulism in 2011 due to the consumption of commercial food with green olives ( Pingeon et al. 2011 ), and for two other food-borne outbreaks subsequently to the ingestion of home-made ham and egg plant preparation, respectively ( table 1 ). Clostridium botulinum A2 was associated with several food-borne botulism outbreaks during the recent period ( table 1 ). Notably, a severe outbreak including one decease and four patients hospitalized in intensive care unit with long-term mechanical ventilation was due to the ingestion of home-made canned beans contaminated with C. botulinum A2 in 2010 ( Oriot et al. 2011 ) ( table 1 ).
Two strains (1141-11 and 1430-11) from food-borne botulism contained a bont subtype A5 and a truncated bontB gene. The strain 1430-11 was responsible for a sporadic case of botulism subsequently to the consumption of a commercial ready-to-eat preparation (pasta carbonara), and the strain 1141-11 was isolated from stool of a food-borne botulism case of unknown origin ( table 1 ). A third strain (126-07) A5 subtype was originated from wound botulism in Switzerland. BoNT/A5 sequences of strains 1141-11 and 1430-11 were identical at the nucleotide and amino acid level to that identified in an infant botulism case in California ( Dover et al. 2010 ) and to that of a unknown origin strain which was characterized by the E Johnson’s laboratory ( Jacobson et al. 2011 ), whereas BoNT/A5 sequence of the strain 126-07 was similar to that of the strain isolated from wound botulism in United Kingdom ( Carter et al. 2010 ) ( table 2 and supplementary fig. S1 , Supplementary Material online). The two strains 1141-11 and 1430-11 differed by their truncated bont/B . The strain 1430-11 contained a truncated bont/B2 identical to the corresponding gene of United Kingdom and Swiss strain, whereas the strain 1141-11 harbored a truncated bont/B3 gene related to that of the US strain ( supplementary fig. S1 , Supplementary Material online and table 2 ).
Clostridium botulinum A5 Strains
 |
|
N ote .—The colors indicate the strains with identical sequence type (ST).
nd, not determined; IP, Institut Pasteur.
Clostridium botulinum A5 Strains
|
|
N ote .—The colors indicate the strains with identical sequence type (ST).
nd, not determined; IP, Institut Pasteur.
The two recently identified bont subtypes A7 and A8 showed 6–15% amino acid difference with the other known subtypes ( table 3 and fig. 1 ). Clostridium botulinum A7 was isolated from an outbreak including two severe cases of botulism due the ingestion of a commercial ready-to-eat product containing chicken enchiladas ( King 2008 ). The C. botulinum A8 strain was recovered in France in 2012 from stool of a patient who was living alone and who was used to consume out-of-date industrial food products and was similar to that described by ( Kull et al. 2015 ). The source of the contamination was not identified.
Nucleotide/Amino Acid Sequence Identities of bont/A Genes and Predicted Proteins of Clostridium botulinum A Subtypes
| Subtype | A2 | A3 | A4 | A5 | A6 | A7 | A8 |
|---|---|---|---|---|---|---|---|
| Kyoto F | Loch Maree | Ba 657 | H0 4402065 | CDC41370 | 148.08 a | 217.12 a | |
| A1 ATCC3502 | 95/90 | 92/85 | 94.5/89 | 99/97 | 98/96 | 97/94 | 97/93 |
| A2 | 97/93 | 94/88 | 95/90 | 96/92 | 94.5/90 | 96/93.5 | |
| A3 | 92/84 | 93/85 | 93/86 | 92/85 | 94/88 | ||
| A4 | 94/87 | 94/88 | 93/87 | 94/89 | |||
| A5 | 98/95 | 97/94 | 97/93.5 | ||||
| A6 | 96/93 | 96/93 | |||||
| A7 | 96/91 |
| Subtype | A2 | A3 | A4 | A5 | A6 | A7 | A8 |
|---|---|---|---|---|---|---|---|
| Kyoto F | Loch Maree | Ba 657 | H0 4402065 | CDC41370 | 148.08 a | 217.12 a | |
| A1 ATCC3502 | 95/90 | 92/85 | 94.5/89 | 99/97 | 98/96 | 97/94 | 97/93 |
| A2 | 97/93 | 94/88 | 95/90 | 96/92 | 94.5/90 | 96/93.5 | |
| A3 | 92/84 | 93/85 | 93/86 | 92/85 | 94/88 | ||
| A4 | 94/87 | 94/88 | 93/87 | 94/89 | |||
| A5 | 98/95 | 97/94 | 97/93.5 | ||||
| A6 | 96/93 | 96/93 | |||||
| A7 | 96/91 |
N ote .— a Recently identified subtype ( King 2008 ; Kull et al. 2015 ).
Nucleotide/Amino Acid Sequence Identities of bont/A Genes and Predicted Proteins of Clostridium botulinum A Subtypes
| Subtype | A2 | A3 | A4 | A5 | A6 | A7 | A8 |
|---|---|---|---|---|---|---|---|
| Kyoto F | Loch Maree | Ba 657 | H0 4402065 | CDC41370 | 148.08 a | 217.12 a | |
| A1 ATCC3502 | 95/90 | 92/85 | 94.5/89 | 99/97 | 98/96 | 97/94 | 97/93 |
| A2 | 97/93 | 94/88 | 95/90 | 96/92 | 94.5/90 | 96/93.5 | |
| A3 | 92/84 | 93/85 | 93/86 | 92/85 | 94/88 | ||
| A4 | 94/87 | 94/88 | 93/87 | 94/89 | |||
| A5 | 98/95 | 97/94 | 97/93.5 | ||||
| A6 | 96/93 | 96/93 | |||||
| A7 | 96/91 |
| Subtype | A2 | A3 | A4 | A5 | A6 | A7 | A8 |
|---|---|---|---|---|---|---|---|
| Kyoto F | Loch Maree | Ba 657 | H0 4402065 | CDC41370 | 148.08 a | 217.12 a | |
| A1 ATCC3502 | 95/90 | 92/85 | 94.5/89 | 99/97 | 98/96 | 97/94 | 97/93 |
| A2 | 97/93 | 94/88 | 95/90 | 96/92 | 94.5/90 | 96/93.5 | |
| A3 | 92/84 | 93/85 | 93/86 | 92/85 | 94/88 | ||
| A4 | 94/87 | 94/88 | 93/87 | 94/89 | |||
| A5 | 98/95 | 97/94 | 97/93.5 | ||||
| A6 | 96/93 | 96/93 | |||||
| A7 | 96/91 |
N ote .— a Recently identified subtype ( King 2008 ; Kull et al. 2015 ).
The type A strains from infant botulism in France mainly belonged to the A2 subtype (11 out of 15 isolated strains), which was involved in 6 out of 10 type A infant botulism outbreaks. Two infant botulism strains were identified as subtype A1(B) and one A1 in OrfX locus ( table 1 ).
BoNT/A1 sequences of C. botulinum A1 (HA or OrfX locus), A1(B), A1(F) strains were highly conserved at the nucleotide and amino acid level, whereas BoNT/A2 sequences showed significant variations. However, BoNT/A2 sequences from strains of a same outbreak were identical ( table 1 and fig. 1 ).
Clostridium botulinum Type B and Subtypes
The 31 C. botulinum type B strains of this study consisted of 22 group I and 9 group II strains. Most of the group I C. botulinum B strains were subtype B2. Three historical C. botulinum B strains, two of them from food-borne botulism in France due to consumption of ham and one from cattle botulism, were subtype B2. In the recent period in France, 11 monovalent group I C. botulinum B strains have been isolated from food-borne botulism, and have been assigned to subtype B2, except one strain which was subtype B3. Clostridium botulinum B2 strains were most often (8 of 11) from canned vegetables (asparagus, spinach, bell pepper), and three from foods of animal origin (ham, homemade paté, tuna). Two additional strains from food-borne botulism in United Kingdom and Iran ( Pourshafie et al. 1998 ; Mazuet, Bouchier, et al. 2015 ), respectively, were also B2 subtype ( table 1 ). Two monovalent group I C. botulinum B strains (B2 and B5) were isolated from infant botulism ( table 1 ). An additional strain responsible for infant botulism was a bivalent B5f2 strain. BoNT/B2 sequences from all C. botulinum B2 strains from food-borne botulism, food and infant botulism were closely related at the nucleotide and amino acid levels ( fig. 2 ).
Phylogenetic relatedness of BoNT/B nucleotide sequences and deduced proteins. The dendograms were constructed using the UPGMA method. The genetic distances were computed by using the Kimura two-parameter model. The scale bar indicates similarity values. The numbers next to each node indicate the cophrenetic correlation. Evolutionary analyses were conducted in Bionumerics (V.6.6 Applied Maths). Strains noted * , ** , ° , # were isolated from a same botulism outbreak. For the A1(B) strains, the sequences of the truncated BoNT/B (nucleotides 1–384, 127 amino acids) were analyzed. These short BoNT/B sequences cannot be significantly compared with the other full length BoNT/B subtypes.
All the 8 C. botulinum B strains which have been isolated from ham responsible for food-borne botulism during the recent period, were nonproteolytic (Bnp or B4) strains of group II ( table 1 ). An additional strain from food in United Kingdom was also a B4 subtype ( table 1 ). BoNT/B4 sequences at the nucleotide and amino acid level were identical and were distantly related to those of group I C. botulinum B strains ( fig. 2 ). BoNT/B4 showed 93–94.5% identity at the amino acid level with the other BoNT/B subtypes, whereas the relatedness between BoNT/B of group I strains ranged from 94% to 98% identity ( table 4 ).
Nucleotide/Amino Acid Sequence Identities of bont/B Genes and Predicted Proteins of Clostridium botulinum B Subtypes
| Subtype | B2 | B3 | Bnp (B4) | B5 | B6 | B7 | B8 |
|---|---|---|---|---|---|---|---|
| BL6 | 472-00 | 815-12 | 338-05 | Osaka 05 | NCTC 3807 | Maehongson 2010 | |
| B1 NCTC 7273 | 98/96 | 98/96 | 96/93 | 97.5/95.5 | 98/96 | 98/95 | 98/95.5 |
| B2 | 99/98 | 96/94 | 98/96 | 99/98 | 98/96 | 98/96 | |
| B3 | 97/94.5 | 98/96 | 98/97 | 98/96 | 98/96 | ||
| B4 | 96/93 | 96/93.5 | 96/94 | 96/93 | |||
| B5 | 97/95 | 97/94 | 97/94.5 | ||||
| B6 | 97/95 | 98/96 | |||||
| B7 | 97/95 |
| Subtype | B2 | B3 | Bnp (B4) | B5 | B6 | B7 | B8 |
|---|---|---|---|---|---|---|---|
| BL6 | 472-00 | 815-12 | 338-05 | Osaka 05 | NCTC 3807 | Maehongson 2010 | |
| B1 NCTC 7273 | 98/96 | 98/96 | 96/93 | 97.5/95.5 | 98/96 | 98/95 | 98/95.5 |
| B2 | 99/98 | 96/94 | 98/96 | 99/98 | 98/96 | 98/96 | |
| B3 | 97/94.5 | 98/96 | 98/97 | 98/96 | 98/96 | ||
| B4 | 96/93 | 96/93.5 | 96/94 | 96/93 | |||
| B5 | 97/95 | 97/94 | 97/94.5 | ||||
| B6 | 97/95 | 98/96 | |||||
| B7 | 97/95 |
Nucleotide/Amino Acid Sequence Identities of bont/B Genes and Predicted Proteins of Clostridium botulinum B Subtypes
| Subtype | B2 | B3 | Bnp (B4) | B5 | B6 | B7 | B8 |
|---|---|---|---|---|---|---|---|
| BL6 | 472-00 | 815-12 | 338-05 | Osaka 05 | NCTC 3807 | Maehongson 2010 | |
| B1 NCTC 7273 | 98/96 | 98/96 | 96/93 | 97.5/95.5 | 98/96 | 98/95 | 98/95.5 |
| B2 | 99/98 | 96/94 | 98/96 | 99/98 | 98/96 | 98/96 | |
| B3 | 97/94.5 | 98/96 | 98/97 | 98/96 | 98/96 | ||
| B4 | 96/93 | 96/93.5 | 96/94 | 96/93 | |||
| B5 | 97/95 | 97/94 | 97/94.5 | ||||
| B6 | 97/95 | 98/96 | |||||
| B7 | 97/95 |
| Subtype | B2 | B3 | Bnp (B4) | B5 | B6 | B7 | B8 |
|---|---|---|---|---|---|---|---|
| BL6 | 472-00 | 815-12 | 338-05 | Osaka 05 | NCTC 3807 | Maehongson 2010 | |
| B1 NCTC 7273 | 98/96 | 98/96 | 96/93 | 97.5/95.5 | 98/96 | 98/95 | 98/95.5 |
| B2 | 99/98 | 96/94 | 98/96 | 99/98 | 98/96 | 98/96 | |
| B3 | 97/94.5 | 98/96 | 98/97 | 98/96 | 98/96 | ||
| B4 | 96/93 | 96/93.5 | 96/94 | 96/93 | |||
| B5 | 97/95 | 97/94 | 97/94.5 | ||||
| B6 | 97/95 | 98/96 | |||||
| B7 | 97/95 |
The bont/B sequence was the same in all C. botulinum A1(B) strains and contained a premature stop codon leading to a putative protein of only 127 amino acids. Therefore, the bont/B of the bivalent A1(B) strains cannot be significantly aligned with the other full length BoNT/B subtypes ( fig. 2 ). Three C. botulinum bivalent B5f2 strains were isolated from food or stool of patient of a food-borne botulism outbreak due to the ingestion of vegetable preparations (olive, tomato, pumpkin) ( table 1 ). It is noteworthy that two food samples containing the B5f2 strains (olive, tomato) were also contaminated with C. botulinum A1(B) ( table 1 ). BoNT/B sequences of the C. botulinum B5f2 strains including an infant botulism strain were identical and closely related to the corresponding BoNT/B sequence of the bivalent B5a4 strain ( fig. 2 ). But, they were distantly related to the BoNT/B2 sequences ( figs. 1 and 2 ).
Clostridium botulinum Type E and Subtypes
The historical C. botulinum type E strains (two from France, three from United States and Greenland environment samples) and one (188-09) of three from the recent period were subtype E3 ( table 1 and fig. 3 ). The strain 188-09 was isolated from the gastric juice of a French patient out of three who ingested vacuum packed hot-smoked whitefish of Canadian origin and processed in Finland ( King et al. 2009 ). BoNT/E3 sequences of the historical strains and strain 188-09 were highly similar to that of the reference strain Alaska E43 ( fig. 3 and supplementary table S1 , Supplementary Material online).
Phylogenetic relatedness of BoNT/E nucleotide sequences and deduced proteins. The dendograms were constructed using the UPGMA method. The genetic distances were computed by using the Kimura two-parameter model. The scale bar indicates similarity values. The numbers next to each node indicate the cophrenetic correlation. Evolutionary analyses were conducted in Bionumerics (V.6.6 Applied Maths). The strain ATCC9009 was included as reference strain subtype E1.
A recently identified C. botulinum E subtype, termed E12, was isolated from ham which was responsible for botulism in two persons. Ham sample contained both BoNT/B and BoNT/E but only a C. botulinum E strain (84-10) was isolated. BoNT/B and BoNT/E have been evidenced in the serum of the two patients ( Mazuet, Sautereau, et al. 2015 ). BoNT/E12 from strain 84-10 showed 91–96% identity at the amino acid level with the other BoNT/E subtypes ( supplementary table S1 , Supplementary Material online).
The third botulism type E case was a botulism by intestinal colonization in a 10-year-old children having a Meckel’s diverticulum. The boy had a history of chronic constipation and diplopia two years before an episode of intense asthenia, dry mouth, dysphonia followed by a cardio-respiratory arrest which was controlled by intubation and mechanical ventilation. Bont/E was PCR amplified (referred as 639-11, table 1 ) from stool, and matched with the subtype E5 ( fig. 3 ), indicating the presence of neurotoxigenic C. butyricum rather than C. botulinum E. However, isolation of a viable neurotoxigenic Clostridium strain was unsuccessful. BoNT was evidenced neither in the serum nor in stool of the patient.
Diversity of the BoNT-Associated Nontoxic Proteins
Genomic comparison of the antp genes has been performed in the 40 strains for which WGS has been performed ( table 1 ). NTNH sequences clustered according to the locus type A, B, E or F ( supplementary fig. S2 , Supplementary Material online). In each locus type, NTNH sequence variations reflected the BoNT subtypes.
HA33 sequences showed variations also related to BoNT subtypes ( supplementary fig. S2 , Supplementary Material online). Phylogenetic analysis of HA33 nucleotide and amino acid sequences showed the same clade division of strains, except that the two C. botulinum A5 strains 1141-11 and 1430-11 were closely associated with C. botulinum A2b2 strains at the nucleotide level, and with C. botulinum B2 strains at the amino acid level ( supplementary fig. S3 , Supplementary Material online). In bivalent strains, ha33 is located in the bont/B locus and HA33 sequence variations correlated with bont/B subtypes. However, the two C. botulinum A5 contained the same HA33 sequence, whereas bont/B subtypes were different, B3′ and B2′ ( supplementary fig. S1 , Supplementary Material online).
Phylogenetic analysis of HA17 and HA70 showed the same distribution of variants according to nucleotide and amino acid sequences, and that each clade corresponded to distinct bont subtype ( supplementary figs. S4 and S5, Supplementary Material online). However, HA17 sequences were conserved in subtypes A1(B) and Bf2, whereas HA70 sequences were phylogenetically separated in these two subtypes ( supplementary figs. S4 and S5, Supplementary Material online).
OrfXs showed variations at the nucleotide and amino acid levels which corresponded to the distinct bont subtypes A1(B), A2b2, Bf2, A2, A7 and A8. However, albeit OrfX variants clustered with individual bont subtypes, OrfX heterogeneity was observed in C. botulinum A2 and reflected the MLST diversity. OrfXs from C. botulinum E12 (strain 84-10) were distantly related to OrfX sequences of group I strains ( supplementary figs. S6–S8, Supplementary Material online). OrfX2 is more conserved than OrfX1 and OrfX3 and shows lower phylogenetic difference than the two other OrfXs ( supplementary figs. S6–S8, Supplementary Material online).
P47 sequences clustered with bont types and subtypes and showed no variation inside each bont subtype except in C. botulinum A2 strains where two P47 subgroups could be distinguished ( supplementary fig. S9 , Supplementary Material online). BotR also showed genetic variations at the nucleotide and amino acid level which clustered with the bont subtypes ( supplementary fig. S10 , Supplementary Material online).
Genetic Diversity of Group I Strains
MLST analysis based on seven housekeeping genes have been found useful to elucidate phylogenetic lineages in group I C. botulinum ( Jacobson et al. 2008 ; Luquez et al. 2012 ; Raphael et al. 2014 ). We used the MLST method developed by Jacobson et al. (2008) and the CDC data bank of C. botulinum MLST profiles. MLST based on PCR amplification and sequencing as well as on WGS was performed in 81 strains of group I [59 C. botulinum A and A(B), and 22 C. botulinum B, Ba, and Bf] ( table 1 ). MLST analysis revealed an extreme diversity of group I C. botulinum strains involved in human botulism in France. Out of 81 group I strains, 41 STs were identified, 16 matching with previously reported STs in the CDC data bank and 25 being novel STs ( supplementary table S2 , Supplementary Material online). Strains isolated from the same botulism outbreak shared identical ST ( table 1 ) except in one outbreak where the strain isolated from food was a C. botulinum A2 ST41 and that from patient’s stool was C. botulinum A2 ST42. However, two outbreaks in two distant places in France shared the same type of strains C. botulinum B2 ST53 ( table 1 ). Only a low number of strains from different origin retained a conserved MLST profile, notably C. botulinum Bf2 ST14 was found in two independent food-borne outbreaks and one infant botulism case, and C. botulinum B2 ST38 and A1(B) ST10 were isolated in two independent foods and food-borne outbreaks, respectively ( table 1 ).
Most historical C. botulinum A strains were assigned to ST1 (9 out of 10) and were mainly from animal origin. Clostridium botulinum A strains from the recent period showed diverse STs (25 profiles) distinct from ST1. The two C. botulinum A5 strains isolated in France from food-borne botulism belonged to distinct MLST profiles ( table 1 ). The strain 1141-11 shared the same ST with that of the strain described by Johnson et al. (2011) evoking a possible common origin, whereas the strain 1430-11 showed a unique ST. The C. botulinum type A5 strain (126.07) isolated from wound botulism in Switzerland and that described by Carter et al. from wound botulism in United Kingdom which likely originated from Afghanistan ( Carter et al. 2011 ) showed identical ST ( table 2 ) suggesting a common origin, which remains to be determined.
Clostridium botulinum type B from group I also belonged to diverse STs (19 strains, 11 STs). Clostridium botulinum B2 strains or bivalent strains from France were in a distinct ST than those of strains from other countries such as C. botulinum B1 NCTC7273 or strains from United Kingdom (BL6) or Iran (277.00) ( table 1 ). The four C. botulinum B5f2 strains from three different outbreaks belonged to the same ST (ST14), and two historical strains shared identical ST different from those of the recent strains ( table 1 ).
Clostridium botulinum strains from infant botulism were also diverse, 10 STs were identified in strains from 12 infant botulism cases. A same strain, C. botulinum A2 ST26, was recovered in three infants at different period (2006, 2011, and 2013) and in distinct locations in France (Paris, Grenoble, Toulouse), whereas in the other infant botulism cases a unique ST was identified. In most cases, STs of strains from food-borne botulism were different from those of infant botulism supporting a distinct source of contamination in the two forms of botulism. However, two STs (ST7 and ST14) were shared by strains from both botulism forms ( table 1 ).
Relations within STs by clustering analysis using the maximum parsimony tree constructing method are shown in fig. 4 . STs of 74 group I isolates from France are distributed into 5 complexes and 19 singletons organized in 4 large clusters ( fig. 4 A). Figure 4 B shows that the C. botulinum A and B subtypes are diversely spread in STs. Cluster 1 contained mainly strains from food-borne botulism, and cluster 2 mostly consisted of strains from animal origin. However, strains from food-borne botulism were spread into the four clusters. Clostridium botulinum strains from infant botulism were mainly assigned to cluster 3, but some of them belonged to cluster 1 and 4. The historical strains (isolation between 1947 and 1961) were disseminated into different clusters, notably cluster 2, and the strains from the recent period showed a wider ST distribution. Thereby, C. botulinum strains did not evolved from a single lineage but from multiple phylogenetic pathways. Comparison of French isolates with group I strains from other countries available on databases was performed by phylogenetic MLST analysis ( fig. 5 ). Clostridium botulinum strains from France as well as from other countries were distributed in the four clusters, and no unique evolutionary lineages could be delineated for strains from a same location.
Genetic relationship based on MLST analysis in 74 group I Clostridium botulinum isolates from France. The circles represent the different STs, and the sizes correspond to the number of strains with a particular genotype. Circles are colored according to the origin of strain (food and food-borne botulism, infant botulism, animal botulism or unknown origin). Similar types are connected by lines and halos depict MLST complexes. Heavy lines connecting two circles denote single locus variants, thin lines connect double locus variants, dotted lines show triple locus variants. Evolutionary analyses were conducted in Bionumerics (V.6.6 Applied Maths). ( A ) Strain relatedness of 74 group I C. botulinum isolates. ( B ) Strain relatedness according to types and subtypes of the C. botulinum isolates. ( C ) Strain relatedness according to the date of isolation. Historical strains (1947–1961) are indicated by the year written in blue.
Genetic relationship based on MLST analysis in 151 group I Clostridium botulinum isolates from France and worldwide strains. The circles represent the different STs, and the sizes correspond to the number of strains with a particular genotype. Circles are colored according to the origin of strain (food and food-borne botulism, infant botulism, animal botulism, wound botulism, soil or unknown origin). Similar types are connected by lines and halos depict MLST complexes. Heavy lines connecting two circles denote single locus variants, thin lines connect double locus variants, dotted lines show triple locus variants. Evolutionary analyses were conducted in Bionumerics (V.6.6 Applied Maths). The 151 strains include 74 French isolates ( fig. 4 ), 7 strains from Switzerland, Brazil, Sweden, Iran, and United Kingdom ( table 1 ) and 2 reference strains (Ba657 and NCTC7273, from United States and United Kingdom, respectively) MLST sequences of which have been determined in this study, as well as 68 group I C. botulinum sequences available in GenBank. Ar, Argentina (40); Br, Brazil (1); Cn, China (1); Fr, France (74); Jp, Japan (10); Ir, Iran (1); Pr, Puerto Rico (1); Se, Sweden (1); Ch, Switzerland (3); UK, United Kingdom (4); Ug, Uganda (1); US, United States of America (14). Numbers in brackets indicate the numbers of strains.
MLST Analysis of Group II Strains
MLST analysis has been determined from WGS of 4 nonproteolytic C. botulinum group II strains ( table 1 ). As the house-keeping gene sequences of group I strains cannot been used in group II strains due to sequence diversity, we exploited the sequences of 12 house-keeping genes as previously defined ( Macdonald et al. 2011 ; Weedmark et al. 2014 ). Dendogram of 12 concatenated house-keeping gene sequences from the four strains of this study and group II strains of databases is shown in fig. 6 . The three C. botulinum B4 strains isolated from ham in France belonged to the same phylogenetic branch together with another B4 strain from United States, a C. botulinum F6 strain and two C. botulinum E strains from North America ( fig. 6 ). The C. botulinum E12 strain was on a distinct phylogenetic lineage ( fig. 6 ).
MLST molecular phylogenetic analysis for Clostridium botulinum group II isolates. Concatenated nucleotide sequences of 12 MLST loci were aligned with Muscle (default parameters) and the phylogenetic tree reconstructed by using the Maximum Likelihood method based on the Kimura 2-parameter model ( Kimura 1980 ). About 1,000 boostrap experiments were performed to assess the robustness of the topology and percents are indicated in nodes. The scale bar indicates the genetic distance (number of nucleotide substitutions per site). Evolutionary analyses were conducted in MEGA6 ( Tamura et al. 2013 ). Strains described in this study are in bold. BoNT subtype, year, location, sample type and origin are indicated as follows: orange, BoNT/B4; turquoise, BoNT/F6; red, BoNT/E3; purple, BoNT/E9; green, BoNT/E12; and black, nontoxigenic (NT). Abbreviations: ND, no data; Arg, Argentina; Can, Canada; Que, Quebec; Env, environmental. The new pta allele identified for the strain 1012-10 has been deposited in GenBank (KU 351681). The new alleles (GenBank KM370304 to KM370318) for 15 MLST loci of the strain 84–10 are available on supplementary table of Mazuet, Sautereau, et al. (2015) .
Discussion
Among the 99 strains of group I and II investigated in this study, 20 were historical strains (1947–1961) (12 C. botulinum A, 3 C. botulinum B, and 5 C. botulinum E) and 69 (43 C. botulinum A, 23 C. botulinum B, and 3 C. botulinum E) were isolated during the recent period (1986–2013) in France from food or biological samples related to human botulism. Additional 3 C. botulinum A and 4 C. botulinum B strains originated from other countries in the recent period were included ( table 1 ). The main finding is the broad diversity of the C. botulinum subtypes and genomes of strains from the recent period as monitored by botulinum locus gene sequencing and MLST analysis. Indeed, the C. botulinum A strains encompassed the subtypes A1 ( ha-bont/A1 and orfX-bont/A1 locus), A1(B), A1(F), A2, A2b2, A5(B2′) A5(B3′), as well as the recently identified A7 and A8 subtypes, and were distributed into 23 STs ( table 1 ). In contrast, the historical C. botulinum A strains were more homogeneous. They all contained a ha-bont/A1 locus and mainly belonged to ST1. However, the low number of historical strains and their origin mainly from animal or environment prevented precise evolution analyses of C. botulinum A strains involved in human botulism in France. As human botulism type A was rare in France in the past, the recent and outbreaks with diverse C. botulinum A STs mainly resulted from increased importation of food preparations or food products including spices subsequently transformed and commercialized in France. However, food-borne botulism outbreaks with home-made preparations from local products and infant botulism rather reflected the prevalence and diversity of C. botulinum A in the environment in France which was underestimated until now.
Bivalent C. botulinum A strains, mainly A1(B), were frequently isolated from food-borne botulism and foods [15 A1(B), 5 other bivalent strains of 32 C. botulinum A from food-borne botulism and foods]. In United States, bivalent A1(B) strains were the most frequently identified strains. Indeed, between 2010 and 2013, 86% of 47 C. botulinum A strains isolated from food or food-borne botulism samples were highly genetically related A1(B) subtypes ( Raphael et al. 2014 ). In contrast, C. botulinum A1(B) strains from Japan were found to cluster in two lineages ( Kenri et al. 2014 ). Clostridium botulinum A1(B) was identified in 6 outbreaks in France (2 infant botulism, and 4 food and food-borne botulism) and belonged to 4 STs, two of them being common to two outbreaks. Indeed, an industrial ready meal and an eggplant preparation from Morocco contained an identical C. botulinum A1(B) ST10. A common ST (ST7) was also found between an infant botulism case and one large outbreak of food-borne botulism due to the consumption of green olives and dried tomatoes ( table 1 ). Bivalent A1(B) strains from United States belong to ST4, whereas those of France exhibited distinct STs suggesting different contamination sources in the two countries.
Clostridium botulinum A2 from various origin have been found to retain highly conserved (99.9–100% identity) bont/A2 and to be divided into 6 STs (ST2 being the most frequent, and then 7, 22, 26, 27 and 28) ( Luquez et al. 2012 ). In our study, 17 of 20 C. botulinum A2 strains shared conserved bont/A2 , and three showed more divergent sequences. BoNT/A2 from two strains exhibited 99.1% identity and one strain 97.7% identity with the other BoNT/A2 sequences at the amino acid level ( fig. 1 ). The genetic background of C. botulinum A2 seems to be more variable than previously reported ( Luquez et al. 2012 ). Clostridium botulinum A2 strains from 11 independent outbreaks were distributed into 10 STs. In one outbreak, two different STs (ST41 and 42) have been found in food and stool samples. Three STs (2, 22, and 26) isolated from infant botulism in France have also been found in patient and environment samples in other countries ( Luquez et al. 2012 ), whereas 7 were newly described STs ( table 1 ).
Two botulism cases due to C. botulinum A5 have been identified in France. One case was associated with the consumption of an industrial meal (pasta carbonara) and the strain was assigned to a unique ST (ST46) ( table 2 ), whereas the origin of the other case, possibly home-made preparations, has not been identified. The strain of the second case was identical (ST16) with a C. botulinum A5 strain from China characterized by Johnson ( Jacobson et al. 2008 , 2011 ). A common origin of these two strains is doubtful. The patients with C. botulinum A5 botulism complained about very long recovery (>2 years). This raises a possible long action of BoNT/A5 which has not been experimentally evidenced ( Whitemarsh et al. 2013 ).
The origin of the botulism case with the recently identified C. botulinum type A8 in France was not identified. This strain shares identical BoNT sequence with the recently reported C. botulinum A8 in Germany in an old man who consumed home-made green beans. BoNT/A8 was found to exhibit reduced ganglioside binding and enzymatic activity resulting in a lower biological activity compared with BoNT/A1 ( Kull et al. 2015 ). These findings showing that the genetic diversity of BoNTs might impact their functional activity and subsequently their clinical relevance, strengthen the importance of BoNT subtyping determination.
Most of group I C. botulinum B strains including 11 strains from food or food-borne botulism, one from infant botulism during the recent period as well as two historical strains, were subtype B2 ( table 1 ). It was already reported that C. botulinum B1 was most often originated from United States and associated with vegetables, whereas C. botulinum B2 strains were frequent in Europe and associated with animals or meat ( Hill et al. 2007 ). However, in our study C. botulinum B2 strains were not only associated with meat, but also with vegetables ( table 1 ). Similar findings were reported in Italy ( Franciosa et al. 2009 ). In Japan, C. botulinum B2 is prevalent and some strains, notably from infant botulism, are related to B2 strains from Europe on the basis of common MLVA ( Umeda et al. 2013 ). In our study, C. botulinum B2 strains from 14 independent outbreaks or food were divided into 11 STs which were distinct from those described in Japan ( Umeda et al. 2013 ). The numerous STs suggest multiple sources of C. botulinum B2 strains with distinct geographical localizations. Other C. botulinum B subtypes were rare. Clostridium botulinum B3 and B5 were identified in an infant botulism and food-borne botulism outbreak, respectively ( table 1 ). Clostridium botulinum B3 and B5 are rarely reported, one B3 strain was described in United States and B5 strain in Japan ( Hill et al. 2007 ; Kenri et al. 2014 ). In addition, bivalent C. botulinum B5f2 was involved in two food-borne outbreaks and one infant botulism case. All C. botulinum B5f2 strains shared the same ST ( table 1 ).
Clostridium botulinum strains responsible for infant botulism in France were also genetically diverse as strains from food-borne botulism: 6 subtypes [A1, A1(B), A2, B2, B5, Bf2], and 10 STs in strains from 12 outbreaks. Although bont subtypes are common with those found in food-borne botulism strains, most of STs from infant botulism strains were distinct suggesting different sources of contamination between the two botulism forms. This might also indicate that C. botulinum from certain STs are more adapted to induce an intestinal colonization. However, some STs such as ST7 and ST14 can be found in both botulism forms.
Genetic variations also concerned the genes of the nontoxic proteins of the botulinum clusters. More particularly, ntnh variations matched the bont type and subtype variants and might be used in the determination of C. botulinum types and subtypes. MLST analysis is a powerful discriminative method which allowed to differentiate almost all the group I strains of this study at the outbreak level and presents benefits for comparing with strains on databases from other laboratories as already reported ( Jacobson et al. 2008 ; Umeda et al. 2009 ; Olsen et al. 2014 ; Raphael et al. 2014 ). The group I strains showed a high genetic diversity, and albeit they can be classified into four phylogenetic groups, no genetic lineages could be defined based on bont type and subtype, geographic location, date of strain isolation, or origin. These results support that the evolution of botulinum locus genes is independent of that of the core genome accounting of the diversity of bont type and subtype in strains with various STs ( Luquez et al. 2012 ; Raphael et al. 2014 ; Weedmark et al. 2014 ; Hill et al. 2015 ; Smith , Hill, Raphael 2015 ; Williamson et al. 2016 ).
The group II C. botulinum B strains showed identical bont sequences assigned to B4 subtype ( fig. 2 ). Group II C. botulinum B4 have been reported to form a homogeneous group with identical bont gene sequences and highly related genomic background ( Stringer et al. 2013 ). All the eight French C. botulinum B4 strains have been isolated from ham or food-borne botulism due to ham consumption ( table 1 ). This correlates with previous reports indicating that C. botulinum type B is a frequent inhabitant of the digestive tract of pigs ( Dahlenborg et al. 2001 ; Myllykoski et al. 2006 ). Clostridium botulinum B of group II seems to be a preferred contaminant of pig meat, only three ham samples contained group I C. botulinum A or B strains [two B2 and one A1(B) subtypes] ( table 1 ). WGS of three C. botulinum B4 strains showed that they contain identical botulinum locus, with 100% identity of ha, botR and ntnh sequences at the nucleotide and amino acid levels ( supplementary figs. S2, S3, S4, S5, and S10, Supplementary Material online). In silico MLST analysis showed that these three strains shared a common genetic background identical to that of four other group II strains from environment or fish originated from North America but it is questionable that these strains have a common origin ( fig. 6 ). Group II strains, notably C. botulinum E, are highly prevalent in marine sediments, fish and other seafood from Northern hemisphere areas including United States, Canada, Scandinavia and Russia, whereas the group II C. botulinum type B strains are associated with meat products and more rarely with fish ( Lindstrôm et al. 2006 ; Carter and Peck 2015 ). Albeit group II C. botulinum types B and E are highly related at the genomic level, they can be divided in two clades differing notably by genes involved in carbohydrate utilization or transport ( Stringer et al. 2013 ). The different physiologic properties likely account for the different habitats between nonproteolytic C. botulinum B and E.
Clostridium botulinum type E is rarely found in France. During the recent period, one botulism outbreak was associated with a hot-smoked fish of Canadian origin containing a C. botulinum type E3 strain, which is common in North America ( Macdonald et al. 2011 ). A recently identified C. botulinum type E (E12) was isolated from ham sample. This unusual contamination of ham could result from the seasoning used to prepare this product ( Mazuet, Sautereau, et al. 2015 ). Interestingly, MLST analysis showed that the C. botulinum E12 strain belonged to a distinct phylogenetic branch and that the most related strains were two strains isolated from salted whitefish ( fig. 6 ) supporting an environmental origin probably from salt. A third botulism type E case was an intestinal colonization with a toxigenic C. butyricum E5 from unknown origin ( table 1 ).
Altogether these results highlight the genetic diversity of C. botulinum strains isolated in France from human botulism. Comparison with the historical strains indicates that the C. botulinum strains of the recent period did not result from a single evolutionary lineage but from multiple and independent genetic rearrangements. Local evolution of strains and commercial exchanges of food products with other countries might account for the diversity of French C. botulinum strains. Recent investigations also show similar C. botulinum diversity in other countries such as in Italy ( Giordani et al. 2015 ), Japan ( Kenri et al. 2014 ), Australia ( McCallum et al. 2015 ) and other parts in the world (reviewed in Carter and Peck 2015 ; Hill et al. 2015 ; Smith, Hill, Raphael 2015 ; Smith, Hill, Xie, et al. 2015 ). Thereby, the BoNT-producing clostridia share a high genetic plasticity the mechanisms of which including acquisition or loss of genetic materials by mobile genetic elements (plasmids, phages) and genetic rearrangements by insertion, recombination, mutation events are not yet fully understood. Genetic rearrangement also results from horizontal gene transfer between neurotoxigenic Clostridium strains. Indeed, horizontal gene transfer of bont or the whole toxin gene cluster mediated by transposition through insertion sequence elements, exchange of plasmids or phages has already been documented in C. botulinum ( Hill et al. 2007 , 2009 ; Skarin and Segerman 2011 ; Smith, Hill, Xie, et al. 2015 ; Williamson et al. 2016 ). Identical botulinum gene clusters in strains with various genomic backgrounds and phylogenetic clusters as monitored by ST ( table 1 and fig. 4 ) support horizontal gene transfer between strains from group I and II. Moreover, different bont loci in a same genomic background such as bont/Ba4, bont/A1(B) , and bont/A2 loci in C. botulinum ST7 ( table 1 ) further argues for horizontal gene transfer from distinct neurotoxigenic strains in a same recipient strain. The mode of evolution of C. botulinum strains is still speculative. For example, the bontA1-ha locus seems to be the ancestor in group I C. botulinum strains, because all the historical strains contain this type of toxin gene cluster. The bont/ A1- orfX and bont/A2-orfX clusters are then observed in monovalent or bivalent isolates of a more recent period. The bont/A2-orfX locus is mainly found in strains of the phylogenetic cluster 4 and seems to have subsequently disseminated in strains of the other phylogenetic clusters ( fig. 4 ) suggesting horizontal gene transfer and recombination events. Several methods of genetic analysis have been used to address the genetic diversity of C. botulinum strains. Among them, MLST appears to be a robust discriminatory method which correlates in strain clustering with the other methods such as MLVA and PFGE ( Jacobson et al. 2008 ; Luquez et al. 2012 ; Macdonald et al. 2011 ; Umeda et al. 2013 ; Gonzalez-Escalona, Timme, et al. 2014 ; Olsen et al. 2014 ; Weedmark et al. 2014 ). However, MLST failed to distinguish C. botulinum type A(B) strains, which could be differentiated by SNP analysis ( Raphael et al. 2014 ). Compared with the standard MLST protocol based on PCR product sequencing, WGS allows MLST but also additional and complementary genetic analysis such as SNP analysis and investigation of genetic rearrangement events.
Acknowledgments
We thank M. Tichit for technical assistance. Sequencing was performed at the Genomics Platform, member of “France Génomique” consortium (ANR10-INBS-09-08). This publication used the C. botulinum MLST website ( http://pubmlst.org/cbotulinum , last accessed May 10, 2016) developed by K. Jolley, University of Oxford, developed and funded by the Wellcome Trust. This work was supported by Institut Pasteur and Institut national de Veille Sanitaire
Literature Cited
Author notes
Data deposition: All the novel MLST alleles described in this study have been deposited in GenBank database: hsp60, KF681499 to KF681507; recA KF681489 to KF681498; rpoB, KF681478 to KF681488; oppB KF681460 to KF681477; acek, KF681440 to 681459; mdh, KF681425 to 681439; aroE, KF681410 to KF681424. GenBank accession numbers for the bont sequences type E12 ( Mazuet, Sautereau, et al. 2015 ) were KF929215 and KM370319, for bontA7 ( Becher et al. 2007 ) JQ954969, and for bont/A8 KF667385.
Associate editor: Howard Ochman
