Sampling microbial diversity using fragment analysis:

When is an OTU an ESU?

Patterns of microbial diversity and community composition• What factors regulate community diversity and composition?• How do humans influence microbial communities?

• Does microbial diversity change along environmental gradients?• Do communities from different locations (treatments) differ? • Are there patterns of community structure?

Sequence based diversity estimates• which gene?• how much to sequence?• how to define OTUs? (95%, 97%, 99%--non transitive relationships)

What are we measuring?

Non-sequence based diversity estimates• tRFLP, ARISA, SNPs, AFLP• Often quick, cheap, easy • Only a proxy for actual sequence diversity present• For tRFLP, ARISA one peak = 1 OTU (or does it?)

Operational Taxonomic Units?

Species?

Ecotypes?

Evolutionarily Significant Unit

Compare OTUs to ESUs

GGCCCCGG

GGCCCCGG

GGTCCCAG

GGCCCCGG

GG CCCC GG

GG CCCC GG

GGTCCCAG

GG CCCC GG

Type A Type B

Separate by size

Smaller fragments

Type A Type B markers

1000 bp

500 bp

750 bp

250 bp

Restriction Fragment Length Polymorphisms

MIT

9303

SS

120

MIT

9211

MIT

9313

MIT

9312

MIT

9302

MIT

9202

MIT

9201

MIT

9215

ME

D4

1500 bp

100 bp

1000 bp

500 bp

high B/A low B/A

RsaI digests of the ITS-23S rRNA distinguish Prochlorococcus ecotypesRFLPs of individuals

t-RFLP= terminal Restriction Fragment Length Polymorphism

Use 1 Fluorescently labeled primer in PCR

Digest with restriction enzyme

Run fragments on gel—Visualize using fluorescence detector markers

Type A

Type B

1000 bp

500 bp

750 bp

250 bp

Only bands containing fluorescent primer are seen

Rel

ativ

e F

luor

esce

nce

Uni

ts

DNA (bp)

Each sequence in pool represented by single peak(barring incomplete digestion issues)

Each peak treated as OTU What is the level of sequence diversity within a peak?

DEPENDS…..

Based on length differences in rRNA intergenic spacer

ARISA---Automated Ribosomal Intergenic Spacer Analysis

Fisher and Triplett AEM 1999

16S rRNA 23S rRNA

Rocap et al 2002 AEM

Intergenic spacer lengthscan be variable at finephylogenetic scales

Strategy: Amplify intergenic spacer using primers (universal or group specific) targeted to conserved positions in rRNA genes

ARISA---Automated Ribosomal Intergenic Spacer Analysis

16S rRNA 23S rRNA

Separate by size

Smaller fragments

markers

1000 bp

500 bp

750 bp

250 bp

sample

or

DAx 7.2: MMBL 11/8/2006 3:48:56PM

0 500 1000 DNA (BP)

0

2000

4000

RFU

2.83

3.25

14.46

0.37

0.45 0.30

0.01

2.56

7.04

0.35

1.00

7.68

6.06

1.39

0.03

2.73

33.49

0.04

8.40

7.65

DNA (base pairs)

Rel

ativ

e flu

ore

scen

ce u

nits

(R

FU

s)

Community analyses

Richness - how many?Does richness vary with the environment?

Community A B C

0

10

20

30

40

50

60

0 10 20 30

Salinity

(PSU)O

TU

s

Regression

R2 = 0.96p = 0.0019

What can we do with these OTUs?

Treat each OTU as a “taxonomic unit of richness”

Community Analyses

Community A B C

Site A Site B Site C Site D Site E Site A Site B Site C Site D Site E

OTU 1 1 1 1 0 1 Site A 1

OTU 2 1 1 0 1 0 Site B x 1

OTU 3 0 1 1 0 0 Site C x x 1

OTU 4 0 0 1 1 0 Site D x x x 1

OTU 5 0 1 1 0 0 Site E x x x x 1

OTU 6 1 0 1 0 0

Incidence matrix Similarity matrix

2a

2a + b + cX 100

where

a = # OTUs in both

b = # OTUs in X only

c = # OTUs in Y only

Sorensen similarity coefficient

Community analyses

Community composition - who? Does composition vary with the environment?

Community A B C

Pasture composition differed from forest & plantation composition (p<0.001).

Analysis of Similarity(ANOSIM & NMDS)

Carney et al. (2004) Ecology Letters

The sample set: active and inactive carbonate chimneys

Young, hot, actively chimneys:40-85ºC, pH 9-11

Old chimneys with little or no venting:<10ºC, pH 8

Chimney containing minerals exposed to both hot fluids and background SW

Brazelton et al. 2006

Ecological succession of Lost City Archaea

(WJ Brazelton, 2006)

120 150 160170 190

142RF

U’s

DNA (BP)

Fluorescent size standard

Fluorescently labelled PCR product

What size is a peak anyway?

What size is a peak anyway?

210.0

167.8

233.2

?

Diffusive spread of larger peaksDAx 7.2: MMBL 7/11/2007 12:50:21AM

605-S3_FLB:T *

0 500 1000 DNA (BP)

0

500

1000

1500 RFU

0.15

51.62

4.16

2.67

2.52

8.59

1.42

4.03

3.87 1.85

16.29

2.99

DAx 7.2: MMBL 7/11/2007 12:50:21AM605-S3_FLB:T *

0 500 1000 DNA (BP)

0

500

1000

1500 RFU

0.15

51.62

4.16

2.67

2.52

8.59

1.42

4.03

3.87 1.85

16.29

2.99

Solutions?

Thresholds for peak calling--signal:noise cutoff% area or % peak height cutoffhelps to standardize amount of DNA loaded onto gel (ie 10ng)

Binning--at minimum data must be binned in 1bp binsso 233.2 and 233.3 are counted as the same fragment

typically 1-3 bp bins < 500 bp3-5 bp 500-1000 bp10 bp > 1000 bp

Hewson and Fuhrman 2006Microbial Ecology

Use maximal similarity between two samples to address null hypothesis that communities are different

Iterative binning strategy

T-RFLP specific issues : which enzymes to pick? How many to use?

Engebretson and Moyer, 2003 AEM

In silico analysis of 4603 bacterial SSU rRNA sequences and 18 common REs

But what if……?

You already have a clone library from the same/related environment

You are using group specific primers to differentiate subtypes of interest

You are using a locus other than the SSU rRNA (functional gene, ITS etc….)

Then how do you choose enzymes?

REPK: Restriction enzyme picker

Collins and Rocap, 2007 NAR

http://rocaplab.ocean.washington.edu/tools/repk

Rationally chooses enzymes that discriminate user-defined sequence groups

REPK: Restriction enzyme picker

http://rocaplab.ocean.washington.edu/tools/repk

http://rocaplab.ocean.washington.edu/tools/repk

REPK: Output

Terminal fragment lengths of all sequences with all enzymes

REPK: Output

http://rocaplab.ocean.washington.edu/tools/repk

Visual view of which enzymes are capable of distinguishing each user-defined group

REPK: Output

http://rocaplab.ocean.washington.edu/tools/repk

Enzymes are binned together based on their ability to discriminate groups

REPK: Final Output

http://rocaplab.ocean.washington.edu/tools/repk

Sets of enzymes that together can distinguish all user defined sequence groups

Automated Ribosomal Intergenic Spacer in silico Analysis (ARISISA)

355 complete bacterial genomes in GenBank

All Bacteria

0

10

20

30

40

50

60

70

294 378 462 546 630 725 865 1010 1290 1570

ARISA fragment lengths (base pairs)

Sequences

Wrabel and Rocap

Fragment lengths: predominant bacterial phyla

Genera/species

0 250 500 750 1000 1250 1500 1750

ARISA fragment length (base pairs)

Alpha-ProteobacteriaBeta-ProteobacteriaDelta-Proteobacteria

Epsilon-ProteobacteriaGamma-Proteobacteria

ActinobacteriaBacteroidetes/Chlorobi

ChlamydiaeCyanobacteria

Firmicutes

A single ARISA peak

can represent more than

one group or species of bacteria.

Some groups display minimal overlap.

ARISA fragment lengths (bps)

Actinobacteria

0

10

20

30

40

50

60

70

294 378 462 546 630 725 865 1010 1290 1570Bacteroidetes/Chlorobi

0

10

20

30

40

50

60

70

294 378 462 546 630 725 865 1010 1290 1570

Chlamydiae

0

10

20

30

40

50

60

70

294 378 462 546 630 725 865 1010 1290 1570Cyanobacteria

0

10

20

30

40

50

60

70

294 378 462 546 630 725 865 1010 1290 1570

Firmicutes

0

10

20

30

40

50

60

70

294 378 462 546 630 725 865 1010 1290 1570

Actinobacteria

Bacteroidetes/Chlorobi

Chlamydiae

Cyanobacteria

Firmicutes

Alpha-Proteobacteria

0

10

20

30

40

50

60

70

294 378 462 546 630 725 865 1010 1290 1570

Beta-Proteobacteria

0

10

20

30

40

50

60

70

294 378 462 546 630 725 865 1010 1290 1570Delta-Proteobacteria

0

10

20

30

40

50

60

70

294 378 462 546 630 725 865 1010 1290 1570Epsilon-Proteobacteria

0

10

20

30

40

50

60

70

294 378 462 546 630 725 865 1010 1290 1570Gamma-Proteobacteria

0

10

20

30

40

50

60

70

294 378 462 546 630 725 865 1010 1290 1570

Alpha-Proteobacteria

Beta-

Delta-

Epsilon-

Gamma-

70

0

Seq

uen

ces

0

25

50

75

100

125

150

175

200

225

1 2 3 4 5 6 7 8 9

ARISA peaks

Bacterial strains

Intraspecific variability in ARISA peaks due to multiple operons

Number of ribosomal operons: 1 – 15

Predicted ARISA peaks per strain: 1 - 9!!

Rel

ativ

e fl

uo

resc

ence

un

its

(RF

Us)

DNA (base pairs)

Time to play ARISA detective …A

B

C

0 1500

A Photobacterium profundum

B 2 Prochlorococcus strains, 3 Synechococcus strains.

C Colwellia psychrerythraea

0

25

50

75

100

125

150

175

200

225

1 2 3 4 5 6 7 8 9

ARISA peaks

Bacterial strains

Intraspecific variability in ARISA peaks due to multiple operons

Geobacillus kaustophilus,

Photobacterium profundumVibrio

cholerae

Colwellia psychrerythraea

7 Prochlorococcus, 7 Synechococcus

Number of ribosomal operons: 1 – 15

Predicted ARISA peaks per strain: 1 - 9!!

Bacteria associated with the diatom Pseudo-nitzchia

Kaczmarska et al. 2005 Harmful Algae

Automated Ribosomal Intergenic Spacer Analysis (ARISA)

2 ng

10 ng PCR product, purified

gram + bacteria DNA extraction

amplify ITS with fluorescent label

ITS (variable)16S

23S

run on capillary sequencer with internal size standard

DAx 7.2: MMBL 11/8/2006 3:48:56PM

0 500 1000 DNA (BP)

0

2000

4000

RFU

2.83

3.25

14.46

0.37

0.45 0.30

0.01

2.56

7.04

0.35

1.00

7.68

6.06

1.39

0.03

2.73

33.49

0.04

8.40

7.65

DNA (base pairs)R

elat

ive

fluo

resc

ence

un

its (

RF

Us)

3 µm AB

0.2 µm FLB

exponential stationary

Are diatom-associated bacterial assemblages found among field data?

P. delicatissima P. pungens Chaetoceros socialis Ditylum brightwellii Thalassiosira sp.

Isolates + field filters from 3 stations:

P1, P17, P38

YES

19 Pseudo-nitzschia spp. 8 Ditylum brightwellii 2 Thalassiosira sp. 1 Chaetoceros socialis

P. calliantha P. cuspidata (tentative ID) P. delicatissima P. multiseries P. pungens C. socialis D.brightwellii Thalassiosira sp.

Origins of diatom isolates

P. multiseries

P. australis, P. fraudulenta

Bacterial assemblages differ with Pseudo-nitzschia species.

Patterns related to Pseudo-nitzschia species?

Null hypothesis: Bacterial assemblages are the same between Pseudo-nitzschia species. R = 1; p = 0.01

species + size fractionP. australis ABP. australis FLBP. delicatissima ABP. delicatissima FLBP. multiseries ABP. multiseries FLBP. pungens ABP. pungens FLB

Similarity10

2D Stress: 0.15

Michele Wrabel

Conclusions

Fragment analysis methods offer a means to interrogate a large number of samples fairly rapidly and inexpensively

Ability to ask (and answer!) ecological questions about microbial communities

Important to be aware of both the strengths and limitations of the techniques to interpret data correctly

Thanks to the F.A.M.O.U.S. gang

Billy Brazelton

Eric Collins

Colleen Evans

Clara Fuchsman

Claire Horner-Devine

Kate Hubbard

Andrew Opatkiewicz

Jess Silver

Michele Wrabel

ARISA---Automated Ribosomal Intergenic Spacer Analysis

Fisher and Triplett AEM 1999

Partial ARISA profiles of the bacterial communities in Crystal Bog Lake (A), Lake Mendota (B), and Sparkling Lake (C) during the summer of 1998. In each panel, the red and black electropherograms represent duplicate PCRs that were performed on a single sample from each site.