Whole Genome Sequencing (WGS)– a Powerful Tool for Traceback Analysis of Foodborne Pathogens

6th Annual American Food Sure Summit

Maria Hoffmann, Ph.D.Genomics Research Microbiologist

2

The Complex Etiology of Foods

Shrimp – IndiaCilantro – MexicoRomaine – Salinas, CACheddar – WisconsinCarrots – IdahoGruyere – SwitzerlandPecans – GeorgiaSprouts – ChicagoRed Cabbage - NY

Shrimp – IndonesiaImitation Crab – AlaskaTuna Scrape – IndiaFish Roe – SeychellesSalmon – Puget SoundSoy Sauce – China Rice – ThailandSeaweed Wrap – CAAvocado – MexicoCucumber – MarylandWasabi – JapanPepper – Vietnam

Watermelon – DelawareBlackberries – GuatemalaBlueberries – New JerseyPineapple – GuamGrapes – CaliforniaKiwi – New ZealandApples – New YorkPears – OregonCantaloupe – Costa RicaHoneydew – ArizonaPapaya – MexicoBanana – Costa Rica

Salad Sushi Fruit platter

3

Some perspective on the food supply

•Tracking and Tracing of food pathogens • Almost 200,000 registered food facilities (2/14)

• 81,574 Domestic and 115,753 Foreign• More than 300 ports of entry• More than 130,000 importers and more than 11 million

import lines/year• In the US there are more than 2 million farms

4

Gold standard method for pathogen identification

PFGE: banding patterns determine discrimination within serovar.

PulseNet, est. 1996http://www.cdc.gov/pulsenet/

5

• WGS is high resolution∙ 3-5 million data points are collected for each isolate

• WGS analyses are statistically robust ∙ Unlike PFGE patterns, WGS data can be analyzed in its evolutionary context. Accurate and stable genetic changes within pathogen genomes enable us to pin point specific common sources of outbreak strains (farms, processing plants, food types, and geographic regions)

PFGE v/s WGS

6

Pedigree vs Phylogeny

7

DNA based pathogen surviellance not new

• Flu: 1990s – flu vaccines predicted from phylogenetic trees

• HIV: 1990s – early tracking of HIV transmission using phylogenetics

http://evolution.berkeley.edu/evolibrary/news/081101_hivorigins

8

• CDC investigated a multistate (29 states) outbreak

• 410 confirmed cases between January 1st and July 7th, 2012

• Among the 326 case patient, 55 (17%) had been hospitalized

• Yellowfin tuna was implicated as source of this outbreak

• This product had been imported from an Indian corporation and was used to make spicy tuna sushi for restaurants and grocery stores

• At this time no reference genome

was available at NCBI

Salmonella enterica serovar Bareilly

9

PFGE identical in red

NGS distinguishes geographical structure among closely related Salmonella Bareilly strains

10

11/1

7/10

Sh

ell-

on

Sh

rim

p S

ri L

anka

01/1

4/10

Fro

zen

Fis

h

Ind

ia

12/0

6/04

Cru

shed

Ch

ilis

Ind

ia

10/1

9/07

Co

rian

der

Po

wd

er In

dia

03/1

2/01

Raw

Sh

rim

p V

ietn

am

Envi

ron

men

tal U

SA

09/1

8/08

San

d G

ob

y Fi

sh V

ietn

am

12/2

7/02

Fro

zen

Sh

rim

p In

dia

11/1

3/09

Bab

glad

esh

i Fre

sh W

ater

Fis

h (

Bac

ha)

Ban

glad

esh

Clin

ica

l MD

05/0

8/72

Fe

ath

er M

eal U

SA

02/0

8/07

Fro

zen

Bai

la

Ban

glad

esh

07/1

2/02

Fro

zen

Un

dev

ein

ed S

hri

mp

Ind

ia

03/1

7/08

Co

rian

der

Po

wd

er In

dia

04/2

2/05

Ses

ame

Seed

In

dia

05/0

9/07

Gin

ger

Po

wd

er In

dia

12/2

9/04

Fro

zen

Sh

rim

p In

dia

06/0

1/09

Ch

ili P

ow

der

Ind

ia

04/0

6/10

Fis

h S

tom

ach

Vie

tnam

09/1

7/03

Co

rian

der

Po

wd

er In

dia

Clin

ica

l MD

Envi

ron

men

tal U

SA

1975

-07-

Fro

g Le

gs U

nkn

ow

n

02/0

5/08

Kh

eer

Mix

Pak

ista

n

11/1

8/05

Cay

enn

e P

epp

er I

nd

ia

08/0

8/05

Fro

zen

Wh

ole

Tila

pia

Th

aila

nd

08/1

7/06

Lo

bst

er T

ails

Tai

wan

05/0

2/72

Po

ult

ry M

eal U

SA19

74-0

8-N

on

fat

Dry

Milk

Un

kno

wn

05/1

4/09

Irr

igat

ion

Wat

er U

SA

02/2

6/04

Fro

zen

Raw

Pee

led

Sh

rim

p In

dia

ATC

C 9

115

12/0

5/05

Fro

zen

Ro

ck L

ob

ster

Tai

ls U

nit

ed A

rab

Em

irat

es

07/0

6/05

Fre

sh C

anal

ou

pe

USA

Envi

ron

men

tal U

SA

07/0

9/01

Pab

da

Fish

Ban

glad

esh

03/1

4/05

Co

rian

der

B

angl

ades

h

08/1

8/11

Co

con

ut

Ind

ia

06/2

6/00

Sca

llop

s In

do

nes

ia

10/1

7/11

Pu

nja

bi C

heo

le S

pic

e In

dia

05/0

1/72

Po

ult

ry F

eat

her

Mea

l USA

02/1

7/11

Red

Ch

ili P

ow

der

Pak

ista

n

04

/17/

08 F

en

nel

See

ds

Un

ited

Ara

b E

mir

ates

07/3

0/01

Wh

iske

r Fi

sh V

ietn

am

11/1

6/05

Hils

a Fi

sh T

hai

lan

d

09/0

8/08

Ch

ili P

ow

der

Th

aila

nd

1109

/30/

10 S

esam

e Se

eds

Ind

ia

11/0

8/02

Cu

min

Po

wd

er In

dia

03/2

7/02

Sh

rim

p In

dia

12/2

3/02

Fro

zen

Raw

Eso

mu

s Sw

aiso

n W

ho

le

Vie

tnam

10/1

2/01

Fro

zen

Ro

hu

Ris

h In

dia

10/1

7/00

Sh

rim

p In

dia

09/0

9/10

Ch

ili P

ow

der

Ind

ia

02/2

2/10

Gro

un

d R

ed P

epp

er U

SA

06/2

7/11

Org

anic

Bla

ck P

epp

er In

dia

05/1

3/03

Fro

zen

Raw

Sh

rim

p In

dia

03/0

1/06

Fro

zen

Cra

b w

ith

Cla

ws

Sri

Lan

ka

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l MD

Clin

ica

l NY

Clin

ica

l MD

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l MD

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l MD

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l MD

Clin

ica

l MD

Clin

ica

l NY

Clin

ica

l MD

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l NY

Clin

ica

l NY

03/1

0/10

Co

rian

der

Mex

ico

08/0

1/06

Tu

rmer

ic P

ow

der

Ind

ia

Different PFGE than the outbreak pattern

Same PFGE but not part of the

outbreak

Outbreak IsolatesMD isolates – in greenNY isolates – in purple

11

Same PFGE cluster together (120 SNPs) outbreak isolates cluster together with 100% bootstrap Closest neighbor differ by 20 SNPs

0-6

11720

12

GenomeTrakr

Network of State, Federal, Hospital, University, Contract, and International laboratories sequencing and sharing data on pathogens.

1. Increased resolution of whole genome sequencing (WGS)

2. Both genomic data and metadata from foodborne pathogens

3. Data and filtered metadata housed in public databases at the National Center for Biotechnology Information (NCBI)

4. Publicly accessible

5. Real time comparison and analysis

14

FDA’s GenomeTrakr

• Distributed network of labs to use whole genome sequencing

• Contributing members:

• 13 FDA labs

• 11 PulseNet labs (state public health labs)

• 5 Dept. of Agriculture labs

• 7 University labs

• 1 U.S. hospital lab

• 2 international labs (Argentina, Mexico)

• 3 private contracting labs

• Data curation and bioinformatic support/analyses provided by National Center for Biotechnology Information (NCBI) and FDA-CFSAN.

15

16

17

GenomeTrakr verses PulseNet?

clinicals -> PNfood/env -> GT

clinicals -> PNfood/env -> GT

NCBI

18

GenomeTrakr verses PulseNet?

clinicals -> PNfood/env -> GT

clinicals -> PNfood/env -> GT

NCBI

19

Publicizing data

NCBI:Sequences and metadata

– fastq files in SRA DB, annotated assemblies in GenBank– metadata in BioSample DB (taxonomy, collected by, country and state, year,

isolation source)– Private: city, county, zipcode, firm names, product names, patient data (age,

sex, etc)

Analyses– Phylogenetic trees for each pathogen published daily at NCBI:

http://www.ncbi.nlm.nih.gov/projects/pathogens

GitHub:– CFSAN SNP Pipeline: http://snp-pipeline.readthedocs.org/en/latest/index.html

20http://www.ncbi.nlm.nih.gov/projects/pathogens

21

New isolate check - Salmonella

SNPs distance to same category

SNPs distance to different category

22

Look at close matches within SNP cluster

23

Biosample: Isolate metadata

24

1. Are there any new outbreaks or new clinical-food/environmental links?

2. Which clinical isolates belong to the same outbreak?3. Does this clinical isolate match any food/environmental

isolate?4. Does this food/environmental isolate match any clinical

isolate?5. Is this a resident strain? How long has it been in the

facility?

Routine Questions in Regulatory Investigations

25

Salmonella Braenderup 2014 pre-outbreak

• In 2014, FDA conducted baseline environmental sampling in nut butter processing facilities

• A few of the samples tested positive for S. Braenderup and a PFGE pattern matched several cases of recent salmonellosis without a common link

• WGS was performed on both environmental and clinical isolates and found to be extremely close (2 SNP differences)

26

Salmonella Braenderup

env. swab

clinical

27

Comparing Traditional and Retrospective Outbreaks in Nut Butters

❖ Salmonella Tennessee (Company A, Brand A Peanut Butter, 2006/2007): 715 cases, 129 hospitalizations, no deaths

❖ Salmonella Typhimurium (Company B, Brand B Peanut butter, 2008/2009): 714 cases, 166 hospitalizations, 9 deaths

❖ Salmonella Bredeney (Company C/Brand C Peanut butter, 2012): 42 cases, 10 hospitalizations, 0 deaths

Retrospective Outbreak Investigation

❖ Salmonella Braenderup (Company D/Brand D nut butter, 2014): 6 cases, 1 hospitalization, no deaths

Traditional Outbreak Investigations

28

0

5

10

15

20

25

30

35

40

4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68

Timeline for Traditional Approach to Foodborne Illness Investigation

Contaminated foodenters commerce

Identify contaminated foodand confirm that product or

environmental samplePFGE pattern matches the clinical

sample pattern

Identify illnesses and get PFGE pattern from

clinical samples

Source of contaminationidentified too late to prevent most illnesses

CDC FDA/FSIS

Nu

mb

er

of

Cas

es

Days

29

0

5

10

15

20

25

30

35

40

4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68

Timeline for Foodborne Illness InvestigationUsing Whole Genome Sequencing

Contaminated food enterscommerce

FDA, CDC, FSIS, and States use WGS in real-time and in parallel on clinical, food,

and environmental samples

Source of contaminationidentified early through WGS combined database queries

Averted Illnesses

Nu

mb

er

of

Cas

es

Days

30

• Earlier intervention means:

1) Reduced amount of recalled product;

2) fewer sick patients

3) less impact overall and minimal damage to brand recognition.

Immediate benefits of WGS to industry, growers, and distributers

31

The Fresh-cut Tomato Supply Chain is complex

32

WGS-based monitoring can pinpoint root causes

33

Benefits to industry, growers, and distributers (continued)

• Regular testing throughout network: 1) identifies specific suppliers that are introducing contaminants;

2) identifies whether contaminant is resident to a facility or transient;

3) knowledge of where contaminant is coming from allows industry to fix the problem based on scientific evidence.

•Shift costs to the supplier who has introduced the contaminant.

•How often is the root cause of the problem left unresolved

to occur again at a later date?

One Data Record - Many Possibilities

…..AAGCTTGGAGATCTACGTGTACCTAGTCGAAGACTGAGGCTCTA….

AMR

SNP

Serotype

wgMLST

Markers

Virulence

Resistance (Disinfectant, Heat,Heavy metal…)

Ecological FitnessBiofilm persistence

UnknownAdaptation

35

AMR genotype prediction

36

Resistome Tracker

• What is Resistome Tracker?An epidemiological tool that allows users to track the appearance and spread of resistance genes in bacteria from different sources around the world.

• What can Resistome Tracker help you do?Identify potential reservoirs for the dissemination of resistant bacteriaMay help speed the investigation of resistant outbreaks

• Who operates ResistomeTracker?The National Antimicrobial Resistance Monitoring System (NARMS)

37

NARMS Resistomics

13

v

15

39

EXPLORE

Geographic distribution of Salmonella with selected resistance gene, by year

Running sum of isolate number

40

DISCOVER

Lists genes that arenew to NCBI; updated weekly

Search for the first instance of a gene in Salmonella basedon collection date

41

Improving Food Safety

1. Identify source of foodborne outbreaks more quickly ~ WGS provides an integrated food safety surveillance system ~ permits international capacity building through integration of foreign

food safety entities into the GT network

2. Transparency of open data gives industry full access ~ Genome data made public in real-time~ Public software and analysis tools readily available to industry for viewing of results

3. Food Safety Modernization Act (2011) – preventive Controls, Improve Industry Practices

~ WGS compliments rapid testing methods with environmental

monitoring for repeat positives and problems w/ resident pathogens.

42

Acknowledgements

• FDA• Center for Food Safety and Applied

Nutrition• Office of Regulatory Affairs• Center for Veterinary Medicine

• National Institutes of Health• National Center for

Biotechnology Information

• USDA/FSIS• Eastern Laboratory

• CDC• Enteric Diseases Laboratory

CFSAN contributors:Ruth TimmeMaria Sanchez-LeonEric BrownErrol StrainJames PettengillYan LuoMarc Allard

CVM contributors:Patrick McDermottHeather TateShaohua Zhao