N E W S A N D V I E W S

To discriminate nonself from self, theinnate immune system uses a set of

germline-encoded pattern-recognition recep-tors (PRRs) that recognize conserved micro-bial structures called pathogen-associatedmole-cular patterns (PAMPs)1. In mice andhumans, the Toll-like receptors (TLRs) form afamily of ten members that typically representsuch PRRs. TLRs are able to sense a plethoraof PAMPs that are structurally unrelated. Forexample, TLR2 and TLR4 recognize sugar-bearing molecules (peptidoglycans andlipopolysaccharides, respectively), and TLR3and TLR9 sense nucleic acids (double-stranded RNA and unmethylated CpG DNA).Unexpectedly, TLR5 recognizes a protein, fla-gellin, from both Gram-positive and Gram-negative bacteria2. The recognition of proteinby PRRs was unexpected, as PAMPs are con-served, but proteins are notoriously easy tomutate in response to selection pressures. Inthis issue of Nature Immunology, Smith et al.provide an explanation to this conundrum bydemonstrating that TLR5 recognizes the verysame motif in flagellin that is also necessaryfor propulsion, which is required for success-ful propagation of the bacteria3.

Toll was originally discovered in Drosophilamelanogaster through its involvement indorsoventral patterning during early develop-ment of the embryo. The receptor was subse-quently shown to be required for the control ofinfection by Gram-positive bacteria or fungi inadult flies, paving the way for the discovery oforthologous TLRs in vertebrates. TLRs are

type I integral transmembrane glycoproteins.In their cytoplasmic tail, they share with theinterleukin 1 receptor a large conserved stretchof approximately 150 amino acids, called theToll–interleukin 1 receptor domain, whichmediates downstream signaling through dif-ferent adaptor proteins. These signaling path-ways ultimately activate the transcriptionfactor NF-κB and produce an array of inflam-matory cytokines, adhesion molecules andeffectors such as antimicrobial peptides4,5. TheTLR extracellular domain contains 19–25copies of a leucine-rich repeat (LRR) motif.The structures of all of these ectodomainsremain to be solved, but some closely relatedLRR structures (such as ribonuclease inhibitorand CD42b) serve to construct the model forthe extracellular domain of the TLRs. Themodel suggests that LRR motifs build a horse-shoe-shaped solenoid with an extended con-cave β-sheet. Different amino acid insertionsin the repeats could account for recognition ofvarious ligands, yielding binding surfaces on

the concave site that are 10 times larger thanthe binding surfaces of antibodies or T cellreceptors. It is assumed that this might explainthe variety of structures accommodated by theTLRs’ recognition cleft6.

Bacteria swim by using flagella that areattached to the membrane by a flexible hookand a disc that is part of a rotating ‘motor’. Theflagellar filament is a hollow cylinder formedby 11 protofilaments (Fig. 1). Counter-clock-wise rotation of the disc allows all the protofil-aments to draw together into a bundle, andthe bacterium propels itself smoothly. After awhile, the disc must reverse direction, and theprotofilaments untwist, causing the bacteriumto tumble in a disorderly way. Protofilamentsare composed almost entirely of flagellinmonomers. Each monomer has roughly theshape of an arrowhead with four domains,D0–D4. The protofilaments are made ofmonomers that are packed by axial interac-tions between the deeply buried D1 concaveand convex surfaces7 (Fig. 1).

Jean-Marc Reichhart is at the Centre National de la

Recherche Scientifique UPR9022, Institut de

Biologie Moléculaire et Cellulaire, 15 rue René

Descartes, 67084 Strasbourg, France.

e-mail: jm.reichhart@ibmc.u-strasbg.fr

TLR5 takes aim at bacterial propellerJean-Marc Reichhart

The Toll-like receptor (TLR) family targets pathogen-derived molecules in regions unlikely to change under selectionpressures. For TLR5, which recognizes the protein flagellin, the function of the targeting motif is key.

NATURE IMMUNOLOGY VOLUME 4 NUMBER 12 DECEMBER 2003 1159NATURE IMMUNOLOGY VOLUME 4 NUMBER 12 DECEMBER 2003 1159

Flagellum

Protofilament

Flagellin polymer

Flagellin monomer

TLR5

Figure 1 Bacteria swim by rotatinga flagellum that is attached by aflexible hook to the ‘motor’, partof which is a disc contained in themembrane. The propeller, up to15 µm in length, is composed of 11 protofilaments. Eachprotofilament is nearly exclusivelya polymer of flagellin. Themonomers are packed throughrelatively small but deeply buriedaxial interactions between theconcave (green) and convex (red)surfaces of the D1 domain. TLR5recognizes the flagellin monomerat the very same surface, normallyhidden in the filament, andactivates the immune system.

K. R

.

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reim

mu

no

log

y

The evidence is becoming stronger andstronger that mast cells can contribute to

remodeling and other changes such asepithelial proliferation in tissues at sites ofpersistent mast cell activation1,2. However,

Stephen J. Galli and Susumu Nakae are in the

Department of Pathology, Stanford University,

L-235, 300 Pasteur Drive, Stanford, California

94305-5324, USA.

e-mail: sgalli@stanford.edu

Mast cells to the defenseStephen J Galli & Susumu Nakae

Mast cells are not only important in IgE-associated disorders but also contribute to host defense against bacteria.One way they do this is by enhancing T cell recruitment and lymph node enlargement during bacterial infection.

N E W S A N D V I E W S

Flagellated bacteria are responsible forepithelial infections, and flagella are a virulencefactor in both mammals and plants8,9. Flagellinactivates the vertebrate innate immune systemand leads to the production of cytokines such astumor necrosis factor-α and interleukin 6,making it a potent inducer of septic shock. It isalso involved in the expression of costimulatorymolecules such as CD80 and CD86 on den-dritic cells. Because of its expression by variousbacteria and its recognition by TLR5, flagellinhas all the hallmarks of a PAMP. But how canPRRs recognize a protein that is likely to changeunder selection pressure?

By expressing flagellin in hamster cells,Smith et al. eliminated the possibility thatTLR5 could recognize the flagellin proteinthrough a bacteria-specific sugar modifica-tion. Subsequently, through deletion analysis,transposon insertional mutation and alaninescan, the essential structure for TLR5 activa-tion was mapped to a discrete region in theflagellin D1 domain. Because this region cor-responds to a motif involved in the contactbetween flagellin monomers in the protofila-ment (Fig. 1), Smith et al. analyzed the effectsof their point mutations and showed goodinverse correlation to bacterial motility.Unexpectedly, the newly mapped recognitionsurface is completely hidden within theprotofilaments. However, by comparing thebinding of intact filaments to that of isolatedflagellin monomers, it was apparent that onlythe monomers were recognized by TLR5.Finally, immunoprecipitation experimentsalso demonstrated that TLR5 physically inter-acts with the flagellin monomer3.

Although our ultimate understanding ofthe direct interaction between TLR5 and fla-gellin awaits the cocrystallization of thereceptor’s ectodomain with flagellin, thework by Smith et al. sheds new light on theway TLRs bind to their ligands. Their find-ings also emphasize the similarities betweenthe defense mechanisms against pathogens inanimals and plants. In Arabidopsis, flagellinbinds to FLS2, the product of a resistancegene, and is a potent inducer of immuneresponses. This gene encodes a receptor-likeserine-threonine kinase containing an LRRectodomain and that activates a ‘down-stream’ mitogen-activated protein kinasepathway to induce typical defense reactionssuch as oxidative burst or production of ethylene and defense-related proteins9.Although both FLS2 and TLR5 recognizeconserved regions in the flagellin monomer,FLS2 senses a contiguous stretch of 22 aminoacids from the N terminus of flagellin, whichis different to the motif that is important forTLR5 recognition.

The findings by Smith et al. raise a new setof questions. The main form of flagellin on aliving infectious bacterium is the polymer-ized protofilament. Thus, how could flagellinmonomers become exposed to the sensingmechanism of the host? It is possible thatactivating flagellin monomers could becomeavailable during bacterial growth and repli-cation. Alternatively, flagellin monomersmay be exposed after protofilament depoly-merization, when bacteria are trapped in theacidic phagosomes of the activated cells.However, in addition to being expressed by

phagocytic cells such as monocytes andimmature dendritic cells, TLR5 is alsoexpressed on the apical side of epithelial cells.A second question pertains to the stoichiom-etry of the TLR5-flagellin interaction. Thefirst protein that was shown to bind a Tollreceptor was the cysteine-knot cytokineSpaetzle. This ligand is recognized as a dimerby two Toll molecules that are broughttogether, resulting in downstream signal-ing10. TLR5 binds a single flagellin monomerthat is highly asymmetric. How this bindingis able to trigger receptor dimerization andsignalling is unclear.

This exciting new finding by Smith et al.shows that TLRs target molecules that areimportant in accomplishing critical tasks,thus reinforcing the hypothesis that PRRsrecognize pathogen-associated moleculesthat are indispensable for their survival. Asthe PAMP moniker suggests, these moleculesare not only a pattern among the pathogensbut at least in the case of flagellin and flagel-lar structures, they are also patterns within apattern.

1. Janeway, C.A. Jr. Immunol. Today 13, 11–16 (1992).2. Hayashi, F. et al. Nature 410, 1099–1103 (2001).3. Smith, K.D. et al. Nat. Immunol. 4, 1247–1253

(2003).4. Akira, S. Curr. Opin. Immunol. 15, 5–11 (2003).5. Beutler, B. & Rietschel, E.T. Nat. Rev. Immunol. 3,

169–176 (2003).6. Bell, J.K. et al. Trends Immunol. 24, 528–33

(2003).7. Samatey, F.A. et al. Nature 410, 331–337 (2001).8. Zeng, H. et al. J. Immunol. 171, 3668–3674 (2003).9. Gomez-Gomez, L. & Boller, T. Trends Plant Sci. 7,

251–256 (2002).10. Weber, A.N. et al. Nat. Immunol. 4, 794–800

(2003).

1160 VOLUME 4 NUMBER 12 DECEMBER 2003 NATURE IMMUNOLOGY

such mast cell–dependent tissue changes aregenerally thought to contribute to disease. Inthis issue of Nature Immunology, McLachlanet al.3 now report that mast cell productionof tumor necrosis factor (TNF) can substan-tially enhance T cell recruitment to locallymph nodes and the accompanying lymphnode enlargement during experimentalinfection with Escherichia coli. Thus, mastcells can regulate tissue changes that confereither benefit or harm, depending on thespecific circumstances.

It is difficult to be certain of the mast cell’sspecific functions in host defense or tissueremodeling because immune responses aswell as the tissue changes that can accom-pany them may involve the coordinated andpotentially redundant or overlapping activi-ties of several cell types. As a result, charac-terizing the specific contributions of anysingle element that participates in such complex processes can be challenging. Butthis problem can be readily addressed byusing genetically mast cell–deficient mice

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reim

mu

no

log

y