330

NEWS & VIEWS NATURE|Vol 437|15 September 2005

by fusing with the lipid bilayer of target cells.Mashburn and Whiteley show that these vesi-cles can also release the PQS signal to otherbacteria, thereby sidestepping the problems ofpoor water-solubility and degradation.

Remarkably, P. aeruginosa has three quorum-sensing systems that use AHL signals. Apartfrom the PQS system, there is the LasR/LasIsystem, which signals using 3-oxo-dode-canoylhomoserine lactone (OdDHL), and theRhlR/RhlI system, which uses butyryl-homoserine lactone (BHL). Of these three signals, OdDHL and PQS are poorly water-soluble, and all three can be degraded by otherbacteria. Mashburn and Whiteley found thatPQS (but not OdDHL or BHL) is not dis-persed between bacteria as single, water-solu-ble molecules, but rather is concentratedwithin membrane vesicles, presumably buriedwithin the lipid bilayer.

PQS is one of at least 55 quinolones andquinolines made by P. aeruginosa, some ofwhich have potent antibacterial activity. Themembrane vesicles, in addition to signalling tokindred P. aeruginosa cells, showed potentantibacterial activity against the Gram-posi-tive bacterium Staphylococcus aureus. Most ofthe antimicrobial activity could be extractedby organic solvents, indicating that toxicitywas due to quinolines rather than to the pro-tein component of the vesicles.

The production of the vesicles seems to beregulated by PQS, because when PQS synthesiswas abolished by a mutation in the pqsH gene,

On page 422 of this issue, Mashburn andWhiteley1 describe the unexpected con-vergence of two seemingly unrelated areasof microbiological research: how bacteriatalk to their friends, and how they attacktheir enemies. The authors studied thebacterial pathogen Pseudomonas aerugi-nosa, which releases a hydrophobic mole-cule called the ‘pseudomonas quinolonesignal’ (PQS) to send messages to otherbacteria of the same species. The surpriseis that, rather than being secreted as singlemolecules, PQS is released in bubble-like‘vesicles’ that also contain antibacterialagents and probably toxins aimed at hosttissue cells as well.

Various groups of bacteria use diffusiblechemicals to signal to their own kind, andthis method of communication seems tohave evolved independently several times2.Pseudomonas aeruginosa is a Gram-negative proteobacterium and, like manyproteobacteria, it secretes molecules calledacylhomoserine lactones (AHLs). The concen-tration of these signal molecules increases withcell population density, enabling the bacteria tosense how many of their kind surround them— a phenomenon known as quorum sensing.Whole sets of genes are switched on only athigh cell densities, including genes for bio-luminescence, DNA transfer, pigment produc-tion, and genes required for infecting plants,animals and humans2. But AHL molecules arereadily broken down by other bacteria, andsome AHL signals are poorly soluble in water,so they cannot travel far in an aqueous envi-ronment — both factors that would seem tolimit their potential as communication signals.PQS, although not an AHL-type signal, is alsopoorly water-soluble and readily broken downby other bacteria. So how do these moleculesmanage to spread the word from one bacterium to another?

Mashburn and Whiteley found clues fromanother bacterial communication system —one aimed at enemies rather than friends.Many Gram-negative bacteria release minutevesicles (about 50 nm in diameter)3 from theirouter membranes as a means of delivering tox-ins to host cells and other bacteria (sometimesreferred to type VI protein export4). The vesi-cles consist of a lipid bilayer surrounding anaqueous core and they can therefore transportlipid-soluble toxins (lipopolysaccharide endo-toxin) on their surface and protein toxins intheir core. They release their poisonous cargo

MICROBIOLOGY

Bacterial speech bubblesStephen C. Winans

Many bacteria socialize using diffusible signals. But some of thesemessages are poorly soluble, so how do they move between bacteria? It seems they can be wrapped up in membrane packages instead.

vesicle production virtually ceased. Addition ofextra PQS restored vesicle formation, even ifprotein synthesis was blocked by an antibiotic.So, whatever PQS is doing, it is probably hap-pening at the post-translational level, ratherthan causing alterations in gene expression.Most of the added PQS was found in vesicles,which also contained other quinolones pos-sessing antimicrobial activity. It seems thatwhen PQS is inserted into the outer membrane

— whether it was made by that bac-terium or added from another source —it can destabilize the membrane andspontaneously form vesicles containingitself, other quinolones and presumablyendotoxin and periplasmic proteins.These vesicles can then signal to otherpseudomonads, kill unrelated bacteria,and deliver toxic lipopolysaccharidesand proteins to host cells.

Pseudomonas aeruginosa is an oppor-tunistic human pathogen that is a par-ticular problem in the lungs of cysticfibrosis patients. Greater understandingabout how these bacteria communicatewith each other and attack host cellscould lead to novel treatment strategiesfor P. aeruginosa infections. The way inwhich other bacteria generate theseubiquitous membrane vesicles, and whatthey might contain, is unknown, butmost proteobacteria do not synthesizePQS, so other mechanisms must be

involved. Moreover, hydrophobic long-chainAHLs were not found in membrane vesicles inthis study, so how these signalling moleculesdiffuse or are transported between bacteriahas still to be discovered. ■

Stephen C. Winans is in the Department ofMicrobiology, Cornell University, Ithaca, New York 14853, USA.e-mail: scw2@cornell.edu

1. Mashburn, L. M. & Whiteley, M. Nature 437, 422–425(2005).

2. Waters, C. M. & Bassler, B. L. Annu. Rev. Cell Dev. Biol. (in thepress)

3. Kadurugamuwa, J. L. & Beveridge, T. J. J. Antimicrob.Chemother. 40, 615–621 (1997).

4. Wai, S. N. et al. Cell 115, 25–35 (2003).

CorrectionIn the News & Views article “Cell biology:Powerful curves” by L. Mahadevan and T. J.Mitchison1, which discussed a paper publishedin the same issue2, the authors put forward anelastic-sheet model of microtubule structure.Such a model was previously proposed3 andapplied 3–5 by Flyvbjerg and colleagues in earlierpapers. The authors apologize for the oversightin not citing these earlier quantitative studies.

1. Mahadevan, L. & Mitchison, T. J. Nature 435, 895–897(2005).

2.Wang, H.-W. & Nogales, E. Nature 435, 911–915 (2005).3. Jánosi, I. M., Chrétien, D. & Flyvbjerg, H. Eur. Biophys. J.

27, 501–513 (1998). 4.Chrétien, D., Jánosi, I. M., Taveau, I. & Flyvbjerg, H. Cell

Struct. Funct. 24, 299–303 (1999). 5. Jánosi, I. M., Chrétien, D. & Flyvbjerg, H. Biophys. J. 83,

1317–1330 (2002).

Figure 1 | Friend or foe? Gram-negative proteobacteria such as Pseudomonas aeruginosa communicate with each other bysecreting chemical messengers. Mashburn and Whiteley1

show that they also release such signals in vesicles that containtoxins against other bacteria and the cells of their host.

EYE

OF

SCIE

NC

E/SP

L

15.9 News & Views 325 MH 9/9/05 5:42 PM Page 330

Nature Publishing Group© 2005