Emergence and robustness of

multicellular behavior in bacteria

Joao B. Xavier

Computational and Systems Biology

Memorial Sloan-Kettering Cancer Center

Bacteria have many examples of social

interaction

What prevents evolutionary cheating?

Cooperator

Public good

"cheater"

+

•Strength by numbers

•Secretion of virulence factors

•Biofilm formation

•Quorum sensing

Swarming: collective motility in

Pseudomonas aeruginosa

Swarming: collective motility in

Pseudomonas aeruginosa

Swarming benefits the colony but requires

biosurfactant synthesis by cells

What prevents

evolutionary

cheating?

With: Wook Kim, Kevin Foster

Biosurfactant synthesis is well

characterized

Wild-type

rhlA-

(Non-cooperator)

Different genotypes are distinguishable

using neutral colors

GFP RFP mix

GFP RFP mix

ce

lls

in

co

lon

y

2x10 10

1x10 10

Biosurfactants are a

public good

Biosurfactant secretion is uncheatable

• Non-cooperators do

better than when alone…

• Not enough to distinguish

who wins, WT or rhlA-

• …but at expense of wild-

type

Measured relative fitness:

0.99 0.05

rhlA expression is delayed until

stationary phase

[h]

rhlAB expression ON rhlAB expression OFF

P. aeruginosa

PA14 rhlAB-GFP

Quorum sensing is necessary yet

not sufficient

[h]

Low density

rhlAB ON

rhlAB OFF

High density

rhlA

ΔlasI ΔrhlI without

autoinducers

Quorum sensing is necessary yet

not sufficient

Quorum sensing is necessary yet

not sufficient

ΔlasI ΔrhlI with

autoinducers

Expression of biosurfactant synthesis is

favored at lower nitrogen source levels

Carbon source: Glycerol (C3H5(OH)3)

Nitrogen source: (NH4)2SO4

rhlA regulation ensures

metabolic prudence

But only if there s a quorum

N

N C

C C

C

C

Medium both carbon

and nitrogen but

carbon is in excess

C C

CCC

C

C C

Cells grow while

there s

nitrogen…

C

C

C

…then use excess

carbon to secrete

rhamnolipids

Inducible rhlAB bypasses

metabolic pudence mechanism

No inducer

(behaves like non-

cooperator)

Inducer present

(strict cooperator)

Biosurfactant secretion in strict cooperator

is cheatable Day 1 Day 2 Day 3 Day 4

Metagenomics: probing the microbiome

without bacterial culturing

Lee, et al. (2010)

Single dose of clindamycin can perturb

mice microbiota up to 28 days…

Phylotype color scheme:

Ileum (Clindamycin treated mice)

Cecum (Clindamycin treated mice)

Rela

tive a

bun

dance

R

ela

tive a

bun

dance

Ileum (control)

Cecum (control)

Rela

tive a

bun

dance

R

ela

tive a

bun

dance

With: Charlie Buffie, Eric Pamer

…and greatly increases the risk of

Clostridium difficile colitis

Rela

tive a

bun

dance

R

ela

tive a

bun

dance

Clindamycin treated mice

Control mice

Phylotype color scheme:

Ileum (clindamycin + C. difficile)

Cecum (clindamycin + C. difficile)

A minimal ecological model of microbial

interactions in the intestine

Vanni Bucci, Serena Bradde

Antibiotic therapy and competition explain multi-

stability and hysteresis in intestinal microbiota

(competitive ability of sensitives)

(sensitiv

ity to a

ntibio

tic)

Problem:

Noise-free model predicts dominance states can last indefinitely. Can noise

describe return to sensitive dominated state?

Simulation time Simulation time

Exposure to environmental microbes

explains microbiota recovery

(No noise)

Antibio

tic p

uls

es

Antibiotic tolerants

Antibiotic sensitives

DN=3.3x10-4

(noise level)

DN=1x10-3

(noise level)

Can we test the model with

metagenomic data?

Dethlefsen & Relman (2011) PNAS

Question:

Can we separate OTUs into sensitives and tolerants according to their

response to ciprofloxacin?

Antibiotic sensitive or antibiotic tolerant dynamics

identified from singular value decomposition

Data from subject D Data from subject E Data from subject F

Sample (day) Sample (day) Sample (day)

OTUs that

correlate with

PC1

OTUs that

correlate with

PC2

Summary

• Social interaction is key in microbial evolution and ecology

• Multicellular cooperative traits are open to exploitation…

• …and therefore must have evolved with mechanisms for robustness

• We can find the mechanisms stabilizing bacterial multicellularity such as metabolic prudence

• Next gen sequencing and ecological modeling can unveil the human microbiome

• And lead to applications for human health

Clinical application:

the microbiome of bone marrow transplantation

Microbiota states: Timelines of bone marrow transplants:

State transitions:

With: Ying Taur, Eric Pamer

Xavier lab at MSKCC (est Dec 2009):

Dave van Ditmarsch

Vanni Bucci

Will Chang

Laura de Vargas Roditi

Carlos Carmona-Fontaine

Kerry Boyle

http://cbio.mskcc.org/xavierlab/

Foster lab at Oxford:

Kevin Foster

Wook Kim

Acknowledgments Eric Pamer

Charlie Buffie

Carles Ubeda

Ying Taur

Serena Bradde (now at Curie Institute)

Giulio Biroli (Institut Physique Theorique Saclay)

Thanks also to:

Peter Greenberg, Pradeep Singh,

Deborah Hogan, Les Dethlefsen

Lucille Castori Center for Microbes Inflammation

and Cancer

Funding: