PFGE%20and%20Beyond:%20PulseNet%20in%20the%20Next%20Decade

1
PFGE and Beyond PulseNet in the Next Decade

• Bala Swaminathan, Ph.D.
• Centers for Disease Control and Prevention

2
Why Next Generation Subtyping Methods?

• PFGE (and other RFLP-based methods) are difficult
  to standardize
• Comparability of patterns within and between
  laboratories requires strict adherence to a
  standard protocol
• Normalization of patterns is complex
• PFGE is labor-intensive and requires high
  concentrations of a pure culture
• In some instances or for some pathogen groups,
  discrimination may not be adequate

3
Clinical isolate clusters with no demonstrable
epidemiologic links Example 1
4
Clinical isolate clusters with no demonstrable
epidemiologic links Example 2
5
Requirements for the next generation subtyping
method for PulseNet

• Broad applicability
• Rapid results (lt 24 h)
• Inexpensive
• Better discrimination than PFGE
• Quantitative relatedness between strains
• Accurate snapshot of the genome diversity
• Backward compatibility with PFGE data
• Easy to perform on a routine basis
• Amenable to automation
• Results should be readily comparable within and
  between laboratories

6
Methodologic Approaches

• Multi-locus sequence typing (MLST)
• Multi-locus Variable-Number Tandem Repeat
  Analysis (MLVA)
• High throughput SNP analysis

7
Multi-Locus Sequence Typing

• Based on the nucleotide sequence of internal
  regions of housekeeping loci
• Housekeeping loci should be conserved with only
  minimal nucleotide changes due to conserved
  protein function
• Multiple loci are targeted in this subtyping
  method
• Sequence variation allows for the assignment of
  alleles
• Isolate A ATTCGGCAT allele 1
• Isolate B ATTCGCCAT allele 2
• A combination of alleles for all loci provides an
  allele profile which can then be assigned to a
  sequence type (ST)
• Isolate A (1, 5, 6, 3, 4, 3, 1) ST-5
• Isolate B (1, 5, 6, 3, 3, 3, 1) ST-51
• Sequence types are grouped into clonal complexes
  based on similarity to a central allelic profile

8
Subtyping Campylobacter jujuni

• Three published MLST schemes
• Dingle at al (2001)
• 194 isolates
• 155 sequence types
• 51 unique STs
• Suerbaum et al (2001)
• 32 isolates plus NCTC 11168
• 31 unique allele profiles
• Frequent recombination
• Manning et al (2003)

9
Origin of replication
glt
asp
tkt
unc
MLST loci
nuoH
pgm
atpA
yphC
gly
fumC
1,641,481 bp
eftS
asd
ddlA
gln
10
Subtyping Campylobacter jujuni

• Sails et al (2003)
• Comparison of MEE, MLST and PFGE
• MLST is not as discriminatory as PFGE
• MLST plus a variable locus
• MLST and flaA SVR provides similar discrimination
  to PFGE

11
MLST studies with enterics

• Listeria monocytogenes Additional variable gene
  targets need to be included in MLST (MLST) to
  obtain acceptable discrimination
• Cai et al. 2002
• Zhang et al. 2004
• Salmonella enterica (Kotetishvili et al, 2002)
• MLST is more discriminatory than PFGE
• Escherichia coli (Whittam Laboratory)
• Distinguish pathovars of E. coli/Shigella groups
• Distinguish clonal lineages within pathovars
• E. coli O157H7 is too clonal for MLST subtyping
  (Noller et al, 2003)

12
Multilocus VNTR Analysis(MLVA)

• MLVA (Multi Locus VNTR Analysis)
• Variable Number Tandem Repeats (VNTRs)
• Conserved repeat motif found in the genome
• Example TAACCG
• Variable numbers of repeat units among isolates
  of the same species
• MLVA examines the number of repeats at multiple
  loci to determine genetic relationships

13
Development of E. coli O157 MLVA protocol

• Contract awarded to the Massachusetts Department
  of Public Health / State Laboratory Institute in
  fall 2001
• Collaboration with Dr. Paul Keim (The Northern
  Arizona University)

14
Development of E. coli O157 MLVA protocol (contd)

• Keys, C., S. Kemper, and P. Keim. 2005. Highly
  diverse variable number tandem repeat loci in the
  E. coli O157H7 and O55H7 genomes for
  high-resolution molecular typing. J. Appl.
  Microbiol. 98 928-940.
• 29 VNTR loci polymorphic in O157H7 serotype
  identified

15
Development of E. coli O157 MLVA protocol (contd)

• MA protocol based on 25 VNTR loci
• Amplified in four multiplex PCR reactions
• Fluorescently labeled PCR amplicons sized using
  capillary electrophoresis system (CEQ 8000,
  Beckman Coulter, Fullerton, CA)
• Internal validation at the CDC PulseNet Methods
  Development and Validation Laboratory started in
  summer 2004

16
E. coli O157 strains used in the initial
validation

• 152 isolates analyzed by both MLVA and PFGE using
  XbaI
• Geographically diverse sporadic isolates with
  unique XbaI PFGE patterns (UPP collection)
• Outbreak isolates from eight well characterized
  outbreaks
• Epidemiologically unrelated isolates clustered by
  PFGE
• A subset of 54 isolates were further
  characterized with BlnI

17
Nine VNTR loci included in the final MLVA
protocol for E. coli O157
VNTR Alternative name1 Repeat size (bp) No. of repeats No. of repeats No. of alleles Inside ORF
Minimum Maximum
VNTR-3 Vhec3, TR5 6 4 23 20 Yes
VNTR-9 Vhec4, TR1 6 5 20 17 No
VNTR-10 Vhec1, TR2 6 10 68 39 Yes
VNTR-17 TR3 6 2 18 11 Yes
VNTR-19 TR7 6 4 10 7 Yes
VNTR-25 TR4 6 1 20 8 No
VNTR-34 Vhec2, TR6 18 5 10 6 Yes
VNTR-36 Vhec7 7 3 15 14 No
VNTR-37 6 3 19 14 Yes
1 Vhec loci are form Lindstedt et al. (2003) TR
loci are from Noller et al. (2003)
18
MLVA protocol steps

• Boiled whole cell DNA templates prepared from
  overnight cultures
• Nine VNTR sites amplified in three PCR reactions
• Diluted (160) PCR products mixed with sample
  loading solution and 600 bp DNA size standard
• PCR products sized using CEQ 8000 capillary
  electrophoresis system (Beckman Coulter)
• Fragment list exported to BioNumerics (Applied
  Maths, Kortijk, Belgium) for analysis

19
Discriminatory power of MLVA compared to PFGE

• 152 isolates
• 133 unique MLVA patterns
• 126 unique XbaI PFGE patterns
• A subset of 54 isolates were characterized by
  PFGE using two enzymes
• 35 unique MLVA patterns
• 39 unique XbaI-BlnI PFGE patterns

20
Clustering of 152 E. coli O157H7 isolates by
MLVA
Cluster II
Cluster I
Sakai
EDL933
21
Clustering of 43 E. coli O157H7 isolates by
MLVA and by PFGE using combined XbaI-BlnI data
MLVA II
MLVA Ib
MLVA Ia
PFGE III
PFGE I
PFGE II
22
Clustering of outbreak isolates and some selected
sporadic isolates by MLVA
GA water park outbreak
CT apple cider outbreak
CO outbreak
NJ outbreak
Western States outbreak
WI restaurant outbreak
NY County Fair
MI outbreak
23
Clusters 0411ml-1c and 0501ml-1c PFGE pattern
combination EXHX01.0086/EXHA26.0576
24
Conclusions from the on-going validation of the
E. coli O157 MLVA protocol

• Overall, MLVA slightly less discriminating than
  PFGE with two enzymes
• MLVA can further discriminate some of the most
  common PFGE patterns
• Epidemiological congruence of the MLVA data
  better than that of PFGE
• Development of interpretation guidelines may pose
  a challenge

25
Future plans

• 2005
• Complete the CDC internal validation of the E.
  coli O157 MLVA protocol
• Custom-made 1 kb standard for the locus VNTR-10?
• Reagent evaluation
• Fine-tuning of the BioNumerics scripts
• Begin collaborative validation of the E. coli
  O157 MLVA protocol by transferring the protocol
  to four PulseNet laboratories

26
Future plans (contd)

• 2006
• Expand the implementation of the protocol to at
  least four more PulseNet laboratories
• Establish a national database with a pattern
  naming strategy
• Establish interpretation criteria

27
SNP-based Typing of E. coli O157
28
AAGGTTA ATGGTTA
29

• SNPs as genotyping markers
• Unambiguous data
• Easy to exchange/compare in database
• Good potential for automation
• Amenable to high-throughput platforms
• Useful for long-term epidemiology/population
  genetics
• Alternative for typing highly clonal species,
  serotypes

30
E. coli O157 genes are highly conserved

• Mosaic genome 5.59Mb
• Genomic diversity by PFGE MLVA
• gt99.9 homology in orthologous genes
• MLST didnt work well for typing O157
• Noller et al 7 housekeeping 2 membrane
  protein genes
• 77 isolates, gt18 PFGE
  types, 2 STs
• (1 SNP in ompA)
• Foley et al 7 virulence 1 housekeeping
  genes
• 92 isolates, 72 PFGE
  types, 5 STs
• (2 SNPs in eaeA, 1 in
  hlyA, 10 in uidA)

31
In silico genome comparison

• Anchor Sakai query EDL933
• Most genes are 100 identical
• 100 loci bearing SNPs
• (phageborne, sequencing errors,
• or paralogous)
• Need a better strategy to identify
• novel SNPs

http//www.genome.wisc.edu/ http//genome.gen-info
.osaka-u.ac.jp/ http//colibase.bham.ac.uk/ http/
/snpsfinder.lanl.gov/
32
NimbleGen CGR microarray
Mutation Mapping
Resequencing
Singh-Gasson et al. 1999. Nat. Biotechnol.
17974-978 Nuwaysir et al. 2002. Genome Res.
121749-1755
33
Selection of genes for CGR

• Conserved among different E. coli O157
  isolates
• Single-copy in the genome
• Re-sequencing capacity per slide 1.2Mb
  (1,200 genes)

• 376 O157-specific genes in 95 size-conserved
• S-loops (including many virulence factors)
• 69 housekeeping genes with putative SNPs
• 754 additional backbone genes
  randomly-selected
• throughout the entire genome
• Large virulence plasmid (pO157)

Ohnishi et al. 2002. PNAS. 9917043-17048
34
O157 strains for resequencing
Strain Origin Year Characteristics PFGE pattern
Sakai Japan 1996 stx1, stx2 0373
F5733 Georgia 1998 stx1, stx2 0224
G5289 Washington 1994 stx2, Phage type 31 0238
01-577 Virginia 2001 stx2, PFGE type 0047 0047
N0436 Colorado 2002 stx1 1315
N0303 New York 2001 stx1, stx2 0264
N0587 North Carolina 2001 stx2 0390
F6141 Georgia 1998 stx1, stx2 0224
F8768 Colorado 2002 stx2 1264
G5101 Washington 1993 stx1, stx2, Mug, Urea 2529
493/89 Germany 1989 stx2, Sorbitol, O157H- 2528
35
Total no. of SNPs in test strains 836
Strain Characteristics PFGE pattern Total no. of SNPs Strain-specific SNPs
Sakai stx1, stx2 0373 - -
F5733 stx1, stx2 0224 0 0
G5289 stx2, Phage type 31 0238 9 1
01-577 stx2, PFGE type 0047 0047 16 0
N0436 stx1 1315 30 4
N0303 stx1, stx2 0264 45 6
N0587 stx2 0390 110 21
F6141 stx1, stx2 0224 150 18
F8768 stx2 1264 164 25
G5101 stx1, stx2, Mug, Urea 2529 351 92
493/89 stx2, Sorbitol, O157H- 2528 473 197
No. of unique SNPs common in G5101 493/89
138 Average SNPs between any of two O157H7
65 No. of informative SNPs to differentiate
between any of two O157H7 139
36

• Polymorphic genes/regions
• 836 SNPs in 503 genes, 65 gene gt3 SNPs
• ECs1934 backbone, putative exonuclease VIII
  (RecE)
• prophage CP-933U 22 SNPs
• ECs1205 Shiga-toxin II subunit A (6 SNPs in
  960-bp)
• ECs1206 Shiga-toxin II subunit B (0 SNPs
  in 270-bp)
• ECs2973-2974 Shiga-toxin I (1 SNP in
  subunit B)
• Conserved genes/regions
• S-loops related to adhesion/invasion
• LEE (Locus of enterocyte enfacement) Type III
  secretion system
• Backbone regions, i.e. between S270-S276

37

• Data analysis in progress
• Backbone vs. S-loops
• Transition vs. transversion
• Synonymous vs. non-synonymous
• Insertions/deletions
• Phylogenetic analysis

38
Conclusions

• PFGE will continue to be an essential subtyping
  method for PulseNet
• MLVA may provide additional discrimination for E.
  coli O157H7 and some Salmonella serotypes
• MLVA protocol for E. coli O157 H7 will be
  transferred to selected PulseNet laboratories in
  2005
• SNP is the subtyping method of the future SNP
  may be used in combination with MLVA
• Much work needs to be done on new subtyping
  methods for PulseNet