JCM Accepted Manuscript Posted Online 24 June 2015 J. Clin. Microbiol. doi:10.1128/JCM.00143-15 Copyright © 2015, American Society for Microbiology. All Rights Reserved.
1
Genomic epidemiology of Clostridium botulinum isolates from temporally related cases of
2
infant botulism in New South Wales, Australia.
3 4
Nadine McCallum2,3#, Timothy J Gray1, Qinning Wang1,2, Jimmy Ng1, Leanne Hicks1, Trang
5
Nguyen1, Marion Yuen1, Grant A. Hill-Cawthorne3,4, Vitali Sintchenko1,2,3
6 7
1
8
Pathology and Medical Research, Westmead Hospital, Westmead, NSW, Australia.2Centre
9
for Infectious Diseases and Microbiology – Public Health, Westmead Hospital, Westmead,
10
NSW, Australia. 3The Marie Bashir Institute for Infectious Diseases and Biosecurity, Sydney
11
Medical School, the University of Sydney, Sydney, NSW, Australia.4School of Public Health,
12
the University of Sydney, Sydney, NSW, Australia
Centre for Infectious Diseases and Microbiology Laboratory Services, Institute of Clinical
13 14
Running title: Genomic epidemiology of C. botulinum from NSW
15 16
#Corresponding author:
17
Dr Nadine McCallum
18
Centre for Infectious Diseases and Microbiology, level 3, ICPMR
19
Westmead Hospital, Westmead, NSW2145
20
AUSTRALIA
21
Phone: + 61 2 9845 6255 | F: + 61 2 9891 5317
22
Email:
[email protected]
23
24
Abstract
25
Infant botulism is a potentially life-threatening paralytic disease that can be associated with
26
prolonged morbidity if not rapidly diagnosed and treated. Four infants were diagnosed and
27
treated for infant botulism in NSW, Australia, between May 2011 and August 2013. Despite
28
the temporal relationship between cases, there was no close geographical clustering or
29
other epidemiological links. C. botulinum isolates, three of which produced botulism
30
neurotoxin serotype A (BoNT/A) and one BoNT serotype B (BoNT/B), were characterised
31
using whole-genome sequencing (WGS). In silico multi-locus sequence typing (MLST) found
32
that two of the BoNT/A-producing isolates shared an identical novel sequence type, ST84.
33
The other two isolates were single locus variants of this sequence type (ST85 and ST86). All
34
BoNT/A-producing isolates contained the same chromosomally-integrated BoNT/A2
35
neurotoxin gene cluster. The BoNT/B-producing isolate carried a single plasmid-borne
36
bont/B gene cluster, encoding BoNT subtype B6. Single nucleotide polymorphism (SNP)-
37
based typing results corresponded well with MLST, however, the extra resolution provided
38
by whole-genome SNP comparisons showed that the isolates differed from each other by
39
over 3,500 SNPs. WGS analyses indicated that the four infant botulism cases were caused by
40
genomically distinct strains of C. botulinum that were unlikely to have originated from a
41
common environmental source. Isolates did however cluster together, when compared with
42
international isolates, suggesting that C. botulinum from environmental reservoirs
43
throughout NSW have descended from a common ancestor. Analyses showed that the high
44
resolution of WGS provided important phylogenetic information that would not be captured
45
by standard seven-loci multi-locus sequence typing (MLST).
46
47
Introduction
48
Genomic epidemiology has provided novel insights into the genetic characteristics and
49
phylogenetic diversity of botulinum neurotoxin (BoNT)-producing Clostridium species (1-8).
50
BoNT are responsible for causing the serious paralytic disease botulism and their potent
51
neuro-paralytic activity make them one of the top, Tier 1, agents considered to pose a
52
significant threat to public health if used for bioterrorism (Electronic Code of Federal
53
Regulations – Title 42; Part 73: http://www.ecfr.gov/cgi-
54
bin/retrieveECFR?r=PART&n=42y1.0.1.6.61) (9, 10). The same properties also make them a
55
powerful tool for both medical therapeutic and cosmetic applications (11).
56
Botulism is a very rare disease in Australia (National Notifiable Diseases Surveillance System:
57
NNDSS- http://www9.health.gov.au/cda/source/cda-index.cfm), with only twenty cases
58
reported since 1991. However, a global survey found that Australia had one of the highest
59
numbers of notified cases of infant botulism in the world (12, 13). Infant botulism results
60
from the ingestion of Clostridium botulinum spores which germinate and temporarily
61
colonise the infant’s colon and vegetative cells begin to grow and produce BoNT (14-16).
62
This form of the disease only occurs in infants, generally under one year old, as they have
63
relatively low levels of gastric acid and poorly developed normal gut microflora (17). It is
64
epidemiologically distinct from foodborne botulism which results from the ingestion of
65
preformed BoNT from contaminated food (18)
66
(http://www.cdc.gov/ncidod/dbmd/diseaseinfo/files/botulism.PDF).
67
The first clinical signs in infant botulism are usually constipation and lethargy, with
68
progression to poor feeding, muscular weakness, abnormal eye movements and, in the
69
absence of intervention, progressive symmetrical flaccid paralysis and respiratory arrest due
70
to neuromuscular junction blockage (19, 20). A number of risk factors have been identified,
71
including the consumption of honey (21-24), herbal infusions (25), contamination of infant
72
food such as powdered formula (26), exposure to household dust (27) and disrupted soil
73
(20, 28). Contact with pet terrapins has also recently been identified as a risk factor for
74
infant botulism caused by Clostiridium butyricum producing BoNT/E (29). However, sources
75
of infection or specific risk factors have generally not been identified for either sporadic or
76
clustered cases (30). Infant botulism has a low mortality rate but can be associated with
77
significant morbidity. However, the timely administration of Human Botulism Immune
78
Globulin Intravenous (BIG-IV) has been shown to reduce morbidity, as evidenced by a
79
significant reduction in the mean length of hospital stay (19, 20, 28, 31, 32).
80
The diagnosis of infant botulism is confirmed by isolation of C. botulinum from the stool,
81
followed by confirmation of toxin production by the mouse neutralization assay (33).
82
Emerging evidence suggests that the genotyping of strains can help to link clustered cases
83
and identify potential shared sources of infections (34-38). Unfortunately the resolution of
84
standard typing methods and the frequent absence of environmental samples often make it
85
difficult to conclusively link cases or identify environmental reservoirs. The use of genome
86
sequencing (WGS) to compare entire bacterial genomes at the single nucleotide level does
87
however have the capacity to greatly improve clustering of bacterial pathogens (39).
88
There are seven well characterised BoNT serotypes (A to G) and a potential eighth type was
89
recently reported (H) (40, 41) but still to be confirmed. The neurotoxins are produced by
90
four genetically and physiologically distinct groups of C. botulinum (Groups I to IV) (42, 43)
91
or less frequently by Clostridium butyricum or Clostridium baratii (44-47). Different
92
serotypes of toxin can cause neuromuscular paralysis of significantly different durations and
93
are further classified into subtypes based on bont gene sequence variation. The majority of
94
human botulism cases are caused by C. botulinum Group II, non-proteolytic strains that
95
produce toxin serotypes B, E or F; or by C. botulinum Group I, proteolytic strains that
96
produce toxins A, B or F. Group I strains are closely related to non-toxogenic Clostridium
97
sporogenes isolates (7, 37, 48). Some strains are bivalent and express more than one type of
98
BoNT (e.g. serotype BoNT/Ab), whereby the uppercase letter denotes the dominant toxin
99
serotype (2). Some BoNT/A strains also carry a silent serotype B toxin, with the presence of
100
the inactive toxin indicated in parentheses (e.g. BoNT/A(B)) (2). BoNT is produced by
101
neurotoxin gene complexes that are located on mobile genetic elements (MGE) which can
102
move horizontally between strains with diverse genetic backgrounds, making the bont-gene
103
or BoNT-complex sequences an unreliable target for phylogenetic comparisons (35, 49, 50).
104
The largest epidemiological studies of infant botulism have been carried out in Japan, where
105
31 cases have been reported since 1986 (4), and California in the US, where 978 cases have
106
been reported since 1976, with 20-50 cases reported each year (34). Dabritz, et al., applied
107
amplified fragment-length polymorphism (AFLP) analysis to genotype isolates and BoNT-
108
specific real-time PCR to characterise BoNT subtypes. Comparative analyses identified
109
several distinct clades associated with spatiotemporal clusters or clusters containing both
110
infant and environmental (i.e. honey) isolates, implicating common-source exposure (34).
111
Considerable diversity was documented in the genomic backgrounds of isolates expressing
112
the same toxin subtypes, indicating extensive horizontal transfer of neurotoxin gene clusters
113
(34). The most in-depth genomic characterisation of C. botulinum isolates causing infant
114
botulism was carried out on all isolates from reported cases of infant botulism in Japan
115
between 2006 and 2011 (4). Comparative genomic analysis of the ten sequenced isolates
116
and 13 reference sequences available in GenBank revealed that only ~33% of the C.
117
botulinum genomes represented a core-genome, reflecting the large phylogenetic distances
118
separating lineages of C. botulinum (49). Core-genome SNP analysis was shown to improve
119
cluster resolution, separating the largest group of serotype A(B) isolates into two distinct
120
lineages (4).
121
Currently there is no information about the genomic characteristics of C. botulinum in
122
Australia. Therefore, whole genome sequencing (WGS) was used to perform comparative
123
genomic analyses on C. botulinum isolates from a potential cluster of four infant botulism
124
cases that occurred in NSW within a 16-month period.
125
126
Materials and methods
127
Case reports and isolate characterisation. In Australia clinical cases of infant botulism are
128
rare with only six cases notified in NSW in the last 15 years. Four of these cases were
129
reported between September 2011 and August 2013. These four cases were documented in
130
infants born at term after uncomplicated pregnancies. All were admitted to the intensive
131
care units of paediatric referral hospitals in Sydney and required tracheal intubation. Infant
132
botulism was suspected following characteristic electromyography features. Only two cases
133
were treated with Botulism Immune Globulin Intravenous (Human) (BIG-IV) (19) and all
134
recovered. Public health investigations were not able to identify epidemiologic links
135
between the cases and no risk factors or potential sources of infection were identified. The
136
clinical course of the cases are summarised in table 1, and clinical reports on cases 1 and 2
137
have been previously published (31).
138
C. botulinum was isolated from stool samples of all cases. Isolates were cultured on Botulism
139
Selective Media and identified using API 20A (Biomerieux, Hazelwood, MO), Rapid ID 32A
140
(Remel, Lenexa, KS) and 16S rDNA sequencing. Mouse bioassay was used to confirm the
141
presence of C. botulinum neurotoxin and to identify the toxin serotype (33). Monovalent
142
neutralising anti-toxins were supplied by the Centers for Disease Control and Prevention,
143
Atlanta, GA.
144
DNA extraction and WGS. Genomic DNA was extracted from pure cultures using the DNeasy
145
Blood & Tissue Kit (QIAGEN) and 200-bp fragment Libraries were prepared using the
146
IonXpress Fragment Library Kit and IonXpress Barcode Adapters, following the protocol for
147
100 ng-DNA input library preparation (Life Technologies, USA). Samples were then pooled
148
and sequenced together on an Ion 318 Chip in an Ion Torrent PGM, using the Ion PGM
149
Template OT2 200 Kit and Ion PGM 200 Sequencing Kit (Life Technologies, USA).
150
Bioinformatic analyses. Sequence data were processed and analysed using the Torrent
151
Suite 4.0.2. Sequencing reads were mapped to the reference genomes: C. botulinum ATCC
152
3502 A1 (accession: AM412317.1), C. botulinum A2 str. Kyoto (accession: NC_012563.1) and
153
C. botulinum B1 str. Okra (accession: CP000939.1) using the Torrent Alignment plugin v4.0-
154
r77189 and variants detected with the variantCaller plugin v4.0-r76860, using default
155
settings: positions used to call single nucleotide polymorphisms (SNPs) had to have a
156
minimum Phred-scaled call quality ≥10, a minimum read fold-coverage ≥6 and a maximum
157
strand bias of 0.95. De novo assembly was also performed on sequencing reads using the
158
Assembler plugin v3.4.2.0 plugin, which uses MIRA Version 3.9.9-Development to assemble
159
contigs. Contigs were then further assembled into scaffolds using Mauve v2.3.1 (51).
160
WebACT (http://www.webact.org/WebACT/home) was used to produce BLAST comparisons
161
of sequenced isolates against each other and the reference genome and plasmid sequences.
162
Seven-loci multiple locus sequence typing (MLST) (35) was performed in silico, by uploading
163
de novo-assembled contig files to the PubMLST Clostridium botulinum query page (http://
164
http://pubmlst.org/cbotulinum/). The MLST allele sequences obtained from NSW1_A2 were
165
then downloaded and concatenated to create an MLST pseudomolecule (supplementary
166
table 1). Reads from all four strains were then aligned to this pseudomolecule to confirm
167
the allele sequences, which were then submitted to the PubMLST Clostridium botulinum
168
sequence query page to confirm the allele profiles. MLST allele profiles were then compared
169
to the list of allele profiles available for download from PubMLST and a phylogenetic tree
170
was created using the PubMLST Treedrawing tool (http://pubmlst.org/cgi-
171
bin/mlstanalyse/mlstanalyse.pl?site=pubmlst&page=treedraw&referer=pubmlst.or).
172
De novo-assembled contigs were also uploaded to CSI Phylogeny 1.0a
173
(https://cge.cbs.dtu.dk/services/CSIPhylogeny/), a webserver which identifies SNPs from
174
whole genome sequencing data, filters and validates the SNP positions and then infers
175
phylogeny based on concatenated SNP profiles (52). C. botulinum ATCC 3502 was used as
176
the reference strain. SNPs were excluded if they were in regions with a minimum fold
177
coverage of
