DOI: 10.1126/science.1209740, 1723 (2011);334 Science

, et al.Christian M. KaiserThe Ribosome Modulates Nascent Protein Folding

This copy is for your personal, non-commercial use only.

clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others

here.following the guidelines

can be obtained byPermission to republish or repurpose articles or portions of articles

): January 17, 2012 www.sciencemag.org (this infomation is current as of

The following resources related to this article are available online at

http://www.sciencemag.org/content/334/6063/1723.full.htmlversion of this article at:

including high-resolution figures, can be found in the onlineUpdated information and services,

http://www.sciencemag.org/content/suppl/2011/12/21/334.6063.1723.DC1.html can be found at: Supporting Online Material

http://www.sciencemag.org/content/334/6063/1723.full.html#relatedfound at:

can berelated to this article A list of selected additional articles on the Science Web sites

http://www.sciencemag.org/content/334/6063/1723.full.html#ref-list-1, 17 of which can be accessed free:cites 61 articlesThis article

registered trademark of AAAS. is aScience2011 by the American Association for the Advancement of Science; all rights reserved. The title

CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience

on

Janu

ary

17, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

15. R. Amann, J. Peplies, D. Schüler, in Magnetoreceptionand Magnetosomes in Bacteria, D. Schüler, Ed.(Springer-Verlag, Berlin Heidelberg, 2007), pp. 25–36.

16. Materials and methods are available as supportingmaterial on Science Online.

17. D. Schüler, Int. Microbiol. 5, 209 (2002).18. C. T. Lefèvre et al., ISME J., published online 21 July

2011 (10.1038/ismej.2011.97).19. R. S. Wolfe, R. K. Thauer, N. Pfennig, FEMS Microbiol.

Lett. 45, 31 (1987).20. C. T. Lefèvre et al., Earth Planet. Sci. Lett. 312, 194

(2011).21. D. J. Lane, in Nucleic Acid Techniques in Bacterial

Systematics, E. Stackebrandt, M. Goodfellow, Eds. (Wiley,Chichester, 1991), pp. 115–175.

22. W. Manz, R. Amann, W. Ludwig, M. Wagner,K.-H. Schleifer, Syst. Appl. Microbiol. 15, 593 (1992).

23. F. P. Abreu, K. T. Silva, M. Farina, C. N. Keim, U. Lins,Int. Microbiol. 11, 75 (2008).

24. E. F. DeLong, R. B. Frankel, D. A. Bazylinski, Science 259,803 (1993).

25. U. Lins, C. N. Keim, F. F. Evans, P. R. Buseck, M. Farina,Geomicrobiol. J. 24, 43 (2007).

26. F. Abreu et al., ISME J. 5, 1634 (2011).27. H. Nakazawa et al., Genome Res. 19, 1801 (2009).28. C. T. Lefèvre et al., Appl. Environ. Microbiol. 76, 3740

(2010).29. C. T. Lefèvre, R. B. Frankel, F. Abreu, U. Lins,

D. A. Bazylinski, Environ. Microbiol. 13, 538 (2011).

Acknowledgments: This work was partially supported byU.S. National Science Foundation grant EAR-0920718(D.A.B.) and by a grant from the Fondation pour laRecherche Médicale SPF20101220993 (C.T.L.). Part of thetransmission electron microscopy characterization was carriedout in Ames Lab and was supported by the U.S. Departmentof Energy, Basic Energy Sciences, Materials Sciences andEngineering Division. The Ames Laboratory is operated forthe U.S. Department of Energy by Iowa State Universityunder contract DE-AC02-07CH11358. We thank F. Mahlaouiand M. L. Schmidt for help with sampling; the NationalPark Service staff of Death Valley National Park; and theteam at the Laboratório de Bioinformática, LaboratórioNacional de Computação Científica, Rio de Janeiro, Brazil,for their help in the sequencing and annotation of the genomeof strain BW-1. Samples at Death Valley National Park were

collected under Sampling Permit DEVA-2010-SCI-0038issued by the U.S. Department of the Interior NationalPark Service. 16S rRNA gene sequences (accession nos.JN015483 to JN015507, JN252194, and JN25219), genefor adenosine-5!-phosphate reductase (aprA; JN705544) andmam genes of BW-1 (JN830627 to JN830646 and JN845570to JN845575) are published in GenBank. Strain BW-1 hasbeen deposited to the Japan Collection of Microorganisms,RIKEN BioResource Center under the provisional nameCandidatus Desulfamplus magnetomortis and carriesaccession no. JCM 18010.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6063/1720/DC1Materials and MethodsSOM TextFigs. S1 to S6Tables S1 and S2References (30–39)Movies S1 and S2

12 August 2011; accepted 2 November 201110.1126/science.1212596

The Ribosome ModulatesNascent Protein FoldingChristian M. Kaiser,1,2 Daniel H. Goldman,3 John D. Chodera,1

Ignacio Tinoco Jr.,1,3 Carlos Bustamante1,2,3,4,5*

Proteins are synthesized by the ribosome and generally must fold to become functionallyactive. Although it is commonly assumed that the ribosome affects the folding process, this ideahas been extremely difficult to demonstrate. We have developed an experimental system toinvestigate the folding of single ribosome-bound stalled nascent polypeptides with opticaltweezers. In T4 lysozyme, synthesized in a reconstituted in vitro translation system, the ribosomeslows the formation of stable tertiary interactions and the attainment of the native state relativeto the free protein. Incomplete T4 lysozyme polypeptides misfold and aggregate when free insolution, but they remain folding-competent near the ribosomal surface. Altogether, our resultssuggest that the ribosome not only decodes the genetic information and synthesizes polypeptides,but also promotes efficient de novo attainment of the native state.

Proteins can spontaneously fold into theirnative structures under appropriate con-ditions (1). In vitro, some small proteins

and single domains attain their native structureswithin microseconds (2), whereas topologicallycomplex and larger proteins may require manyseconds to fold (3) and often populate foldingintermediates along the way (4, 5). In vivo, how-ever, folding is not necessarily limited to full-length proteins or domains. Proteins can begin tofold before they are fully synthesized and whilestill bound to the ribosome (6–12). Moreover,during protein synthesis on the ribosome (13),elongation rates are regulated by factors includ-ing tRNA abundance (14), codon order (15), andmRNA secondary structure (16). Decreasing these

rates locally (17, 18) or globally (19) can affectthe folding efficiency of newly synthesized pro-teins. The complete synthesis of even smallproteins (100 amino acids or less) requires atleast several seconds at a maximum rate of ~20amino acids per second in Escherichia coli (20),giving nascent-chain segments sufficient time toconformationally equilibrate in the environmentof the ribosome, perhaps adopting structures thatare distinctly different from the native protein fold(7, 21). However, the observation of folding tran-sitions in ribosome-bound nascent proteins hasnot been possible, and a detailed analysis has beenlimited to computational approaches (22, 23).

We have developed an experimental systemto directly probe the folding of single ribosome-bound nascent chains (24, 25) by subjecting themto force using optical tweezers (Fig. 1 and figs.S1 to S3). The force is applied between the nas-cent chain and the large ribosomal subunit. Be-cause force acts locally (26, 27), we can selectivelyperturb the stability of ribosome-bound nascentpolypeptides without disrupting the structural in-tegrity of the ribosome.We studied a cysteine-freeversion of T4 lysozyme (28), a monomeric cyto-solic protein composed of two globular regions,

or subdomains (Fig. 1, C and D). T4 lysozymefolding has been studied in ensemble (29–31)and single-molecule (27) experiments. The na-tive fold requires interactions between the N- andC-terminal sequences whose synthesis is sep-arated in time during vectorial translation by theribosome.

Using a reconstituted in vitro translation sys-tem supplemented with E. coli ribosomes (32),we first translated the protein with an unstruc-tured C-terminal extension of 41 amino acids suchthat the entire T4 lysozyme sequence emergesfrom the narrow ribosomal exit tunnel (33). Thisexperimental design allows us to study the fold-ing dynamics of the full-length protein on theribosome. When we stretched the molecule bycontinuously increasing the tension applied acrossthe nascent chain (“force ramp”), we observedsingle rips in the resulting force-extension traces,representing cooperative unfolding events (Fig.1E). Puromycin-release experiments confirmedthat these signals originated from ribosome-boundnascent proteins (fig. S4).

The mechanical unfolding pathways of freeand ribosome-bound full-length T4 lysozyme arevery similar: The latter unfolds at a mean force(FUnf) of 17.0 T 2.0 pN (N = 125 unfoldingevents) at a pulling speed of 100 nm/s (Fig. 1G).Using thewormlike chain (WLC)model (34), wecalculated a contour length increase upon unfold-ing (!LC) of 59.9 T 2.1 nm, consistent with thevalue expected for full-length T4 lysozyme (164amino acids ! 0.36 nm per amino acid – 0.9 nmcorresponding to the folded end-to-end distance =58.1 nm). Experiments with the free protein inthe absence of the ribosome (Fig. 1, F and H) re-vealed similar unfolding characteristics (FUnf =17.2 T 1.8 pN, DLC = 60.1 T 0.9 nm, N = 453),confirming that the entire protein is able to emergefrom the ribosomal tunnel by means of the 41–amino acid linker (33). Analysis of the unfoldingforce distributions (35) of free and ribosome-bound polypeptides (Fig. 1, G and H) alsoreveals similar distances to the transition state(!‡xfree = 2.3 T 0.5 nm, !x‡ribosome-bound = 2.0 T

1Institute for Quantitative Biosciences (QB3), University ofCalifornia, Berkeley, CA 94720, USA. 2Department of Physics,University of California, Berkeley, CA 94720, USA. 3Depart-ment of Chemistry, University of California, Berkeley, CA94720, USA. 4Department of Molecular and Cell Biology,University of California, Berkeley, CA 94720, USA. 5HowardHughes Medical Institute, University of California, Berkeley, CA94720, USA.

*To whom correspondence should be addressed. E-mail:carlos@alice.berkeley.edu

www.sciencemag.org SCIENCE VOL 334 23 DECEMBER 2011 1723

REPORTS

on

Janu

ary

17, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

0.2 nm) and native-state lifetimes (fig. S5). Thus,the ribosome-bound protein folds to the same na-tive structure as the free protein and unfolds throughthe same pathway in the pulling experiments.

Whereas the unfolding transitions of free andribosome-bound T4 lysozyme are indistinguish-able in our experiments, the refolding exhibitsmarked differences: During repeated cycles ofpulling and relaxation, the free protein virtuallyalways refolds, but the ribosome-bound proteinrefolds only 28% of the time and lacks a well-defined refolding transition in the force-rampmeasurements (fig. S6). We carried out “force-clamp” experiments (fig. S7) to better resolve thefolding transition. From measurements of the re-folding time, trefold (Fig. 2A), we estimated ap-parent refolding rates at 3.6 pN (Fig. 2B). Thefree protein folds rapidly with an apparent rate(kapp) of 5.4 s"1 [95% confidence interval (CI):4.4 s"1, 6.5 s"1]. Remarkably, the folding rate ofthe ribosome-bound protein (with a 41–aminoacid C-terminal linker) is 0.012 s"1 (95% CI:0.004 s"1, 0.024 s"1), more than two orders ofmagnitude slower than the free protein.

The 41–amino acid linker is long enough toallow the entire protein to emerge from the exittunnel but restricts the folding environment to theimmediate vicinity of the ribosomal polypeptideexit site. Increasing the linker length by 19 aminoacids provides ~2.1 nm at 3.6 pN (calculatedbased on a WLC model) of additional separationfrom the ribosomal surface and increases the fold-ing rate: This construct, harboring a 60–aminoacid linker, folds with kapp = 0.24 s"1 (95% CI:0.21 s"1, 0.28 s"1) (Fig. 2B), significantly fasterthan the rate observed with the shorter linker,

but still more slowly than the free protein (Fig.2B, also compare Fig. 2C with 2A). We usedthis construct to characterize the folding of theribosome-bound protein in more detail.

Given the high negative charge density ofthe ribosomal RNA (rRNA) moiety, the observeddeceleration in folding is likely mediated byelectrostatic interactions of the ribosomal surfacewith charged residues in the nascent chain (36).Increasing the potassium chloride concentrationfrom 150 to 500 mM, which results in more ef-fective screening of electrostatic interactions (32),increased the folding rate of the ribosome-boundprotein, whereas the folding rates of the free pro-tein were less sensitive to salt (Fig. 2D). Thus, theeffect of the ribosome on the folding of the nas-cent polypeptide is mediated, at least in part, byelectrostatic interactions.

Close inspection of the force clamp traces re-vealed that, before folding to the native state (N),the protein transiently and reversibly samples anintermediate (I) from the unfolded state (U) (Fig.2, A and C), exhibiting bistability, or “hopping.”In almost all traces examined, folding to the na-tive state then proceeds from the intermediatestate (Fig. 2A, arrowhead), indicating that the lat-ter is an on-pathway state and, therefore, oblig-atory for productive attainment of the native state.The I-to-N transition is practically irreversible atthe refolding force; thus, the folding pathway canbe written as

U ⇄ I → N

From the extension changes in the force-clamp traces, we estimate that between 96 and

108 residues participate in the formation of thefolding intermediate. These dimensions are con-sistent with a folding intermediate that is formedlargely by the C-terminal T4 lysozyme subdo-main (32) and might be related to a previouslydescribed “hidden intermediate” (37).

The probability of occupying the state I (rel-ative to U ) during hopping (fig. S8) is similarin the presence and absence of the ribosome(Fig. 2E). From the force at which U and I areequally populated (F1/2 # 3.6 pN) and the ob-served change in extension between U and I atthat force (!xU-I # 10 nm), we estimate that I isstabilized by a Gibbs free energy of unfolding[!G0

unfolding (intermediate)] of = 3.0 kcal/mol(5.1kBT, where kB is the Boltzmann constant andT is temperature) relative to U (38) for both thefree and the ribosome-bound protein. This valueis small compared with the stability of the foldedprotein [!G0

unfolding (native) = 14.1 kcal/mol (30)],indicating that the intermediate is lacking manyof the interactions that stabilize the natively foldedprotein. It is also smaller than the reported stabil-ity of the “hidden” intermediate, perhaps becausethe N-terminal A helix is not part of the interme-diate observed in our experimental geometry (32).

To explore the ribosomal effect within thefolding pathway, we used a Bayesian HiddenMarkov model (BHMM) approach to conduct akinetic analysis of the force-clamp data (Fig. 3A)(32). The values for the rates of the U-to-I andI-to-U transitions (kU-I and kI-U, respectively)are essentially the same for the free and theribosome-bound protein (60–amino acid linker).In contrast, kI-N of the ribosome-bound protein issmaller by at least one order of magnitude com-

10 15 200

0.1

0.2

0.3

0.4

Rupture force (pN)

Pro

babi

lity

10 15 200

0.1

0.2

0.3

0.4

Rupture force (pN)

Pro

babi

lity

tunnelexit

L17

N-terminalsubdomain C-terminal

subdomain

NC

A

B C

N-term. C-term.N C

Avi linker (SGGGG)149 +20 +41 +60164

(f.l.)

T4 lysozymeD

polystyrenebead

DNA handle

suctionpipette

optical trap

biotin/streptavidin

ribosome

nascentchain

F

mRNA

ribosome-nascent chaincomplex

0

10

20

30

Extension (nm)

Forc

e (p

N)

0

10

20

30

Extension (nm)

Forc

e (p

N)

E

G

F

H

50 nm 50 nm

free+41

free+41

Fig. 1. (A) Schematic of the molecular assembly for optical-tweezers ex-periments. Force can be applied to the nascent polypeptide bymoving the opticaltrap relative to the pipette. (B) Surface representation of the ribosome [ProteinData Bank identification numbers (PDB IDs): 2aw4 and 2avy] showing theopening of the ribosomal exit tunnel in the large subunit and the location ofribosomal protein L17, serving as the attachment site in the optical-tweezersexperiments. Ribosomal RNA, pink; ribosomal proteins, white; L17, blue. Thesmall subunit is shown in a semitransparent rendering. (C) Cartoon diagram of T4lysozyme (PDB ID: 4lzm). The N-terminal subdomain (light orange) is composed

of residues 13 to 59; the C-terminal subdomain (dark orange) comprises residues1 to 12 and 60 to 164. (D) Primary structure diagram of the protein constructtranslated for optical-tweezers experiments (32). Red spheres indicate stallingpositions along the sequence. f.l., full length. (E and F) Force-extension traces ofT4 lysozyme unfolding (red) and refolding (blue) on the ribosome (41–aminoacid linker) (E) and free in solution (F). The protein unfolds in one cooperativetransition near 17 pN. (G and H) Rupture-force histograms (gray bars) forunfolding of the ribosome-bound (G) and free (H) protein. Red lines, ruptureforce distribution reconstructed from the force-dependent lifetimes (fig. S5).

23 DECEMBER 2011 VOL 334 SCIENCE www.sciencemag.org1724

REPORTS

on

Janu

ary

17, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

pared with that of the free protein (Fig. 3A). TheBHMM analysis also yields estimates of the ex-tensions of each species (fig. S9). Interestingly,the unfolded protein, both free and ribosome-bound, is more compact than predicted from aWLCmodel (34) at forces below 4 pN (Fig. 3B).This deviation from the model is consistent witha compaction of the unfolded protein, perhapsreflecting transient local secondary structureformation at those low forces before coopera-tive folding transitions. For the free protein, theunfolded-state extensions gradually increase tothe values predicted by the WLC model whenthe force is raised to 5 pN. Notably, the unfoldedstate remains more compact in the ribosome-bound protein (over the limited force range ac-cessible in these measurements; Fig. 3B). Thus,the ribosomal interactions appear to have a dualeffect: deceleration of native tertiary-structure for-mation and stabilization of a compacted state be-fore folding (Fig. 3C), presumably mediated atleast in part by electrostatic interactions.

How might the ribosome contribute to ef-ficient de novo protein folding? So far, we havedescribed folding of the entire protein. To studyfolding before completion of synthesis, we trans-lated the N-terminal 149 amino acids of T4 ly-sozyme, so that ~110 to 120 residues of the fullT4 lysozyme sequence (164 residues) exited theribosome (~70% of the full protein), whereas 30

to 40 amino acids remained within the exit tun-nel. Upon stretching the ribosome-bound poly-peptide, the force-extension curves did not revealany cooperative folding or unfolding transitions(fig. S10). Even when we extended the T4 lyso-zyme sequence by a 20–amino acid linker (sothat ~144 to 154 out of the 164 amino acids of T4lysozyme are outside the tunnel), we did not de-tect folding (Fig. 4A), presumably because theC-terminal residues, which interact with the N-terminal A helix in the native structure, are stillsequestered within the ribosomal exit tunnel.

Attempts to express soluble fragments inE. coli were unsuccessful (fig. S11). Thus, weused puromycin-modified DNA oligonucleotides(39) to release the in vitro translated 149–aminoacid fragment from the ribosome (32). In contrastto the lack of transitions observed on the ribo-some, the released fragment unfolded at a rangeof forces and contour length changes (Fig. 4B),indicating that it adopts a highly heterogeneousensemble of structures, some of which exhibitconsiderable (mechanical) stability. Given thehomogeneous unfolding of the complete pro-tein, it is unlikely that all of these structuresrepresent productive, on-pathway species. Rath-er, the diverse behavior of this fragment is prob-ably due to misfolding, aggregation, entanglementof several polypeptides immobilized in close vi-cinity, or interactions of the misfolded protein

with the bead surface, none of which are ob-served for the ribosome-bound nascent protein.Thus, the ribosome appears to prevent misfoldingof the incomplete protein through a kinetic mech-anism, effectively acting as a molecular chaper-one for nascent polypeptides. This chaperoneactivity is probably mediated by the surface sur-rounding the polypeptide exit tunnel and is dis-tinct from the previously described “protein foldingactivity of the ribosome” mediated by the 26SrRNA (40). Mechanisms that keep proteins in afolding-competent conformation may be particu-larly important if C-terminal residues, whichare synthesized last, are required for productivefolding, as in T4 lysozyme and other proteins(21, 41–43).

In the cell, molecular chaperones interact withnascent polypeptides during their synthesis (44).The in vitro experiments described here suggestthat the ribosome contributes to efficient de novofolding in several ways. Polypeptide compaction,an early event during protein folding (45), is pro-moted by the ribosome and may, in conjunctionwith the spatial arrangement within polysomes(46), limit aberrant interactions among nascentchains. Rather than acting as an inert “wall,”which would be expected to increase folding ratesentropically (47, 48), the ribosome slows foldingof T4 lysozyme, presumably by attracting posi-tively charged residues and repelling negatively

Time (s)

Ext

ensi

on (n

m)

A

I

U

N

B

C

3.5 4 4.5 5

0.01

0.10

1.00

Force (pN)

PIn

term

edia

te

FreeRNC

+41 +60 free10-3

10-2

10-1

100

101

kap

p (1

/s)

E

0 1Time (s)

0.5

!refold

D

free

+60

4 4.5 5

10-2

10-1

100

Force (pN)

kap

p (1

/s)

Fig. 2. (A) Example extension-versus-time trace of T4 lysozyme refolding atconstant force (5 pN). Before transitioning to the native state (N) at ~0.5 s,the protein visits a folding intermediate (I). Enlargement of the first secondof the trace (red dashed box) reveals that the intermediate is visited im-mediately before folding to the native state (arrowhead), demonstrating thatthe intermediate is on-pathway. U, unfolded state. (B) Apparent refoldingrates for ribosome-bound T4 lysozyme with 41–amino acid (+41) and 60–amino acid (+60) linkers and for the free protein (free). Error bars: 95% CIs.

(C) Example trace of ribosome-bound protein (60–amino acid linker) re-folding at 5 pN. (D) Apparent refolding rates of free protein (blue circles)and +60 linkers (red triangles) at 150 mM (filled symbols) and 500 mM(open symbols) KCl. Increasing the salt concentration mitigates the effect ofthe ribosome on the refolding rate. (E) Population of the folding inter-mediate before refolding. The unfolded and intermediate states are equallypopulated (P = 0.5) at a force of ~3.6 pN, both in the free and the ribosome-bound protein.

www.sciencemag.org SCIENCE VOL 334 23 DECEMBER 2011 1725

REPORTS

on

Janu

ary

17, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

charged oneswithin the same nascent polypeptidechain. Thus, it can bias the conformational searchof the protein and its folding rate. This mecha-nism may complement the mode of action of

other molecular chaperones that bind their sub-strates through hydrophobic interactions andmaybe particularly pronounced for T4 lysozyme andother basic proteins that represent a large group in

most proteomes (49). However, such amechanismshould be operative even for proteins harboring anet negative charge and may apply at least tocytosolic proteins, which are held in close prox-imity to the ribosomal surface during synthesis.Our findings may represent a paradigm for howthe ribosome can, in principle, affect nascent-chain folding. The system is easily adaptable forinvestigating proteins other than T4 lysozymeand should be amenable to observing nascent-chain elongation in real time. These future exper-iments will shed more light onto how proteinfolding is tuned to synthesis.

References and Notes1. C. B. Anfinsen, Science 181, 223 (1973).2. J. Kubelka, T. K. Chiu, D. R. Davies, W. A. Eaton,

J. Hofrichter, J. Mol. Biol. 359, 546 (2006).3. K. W. Plaxco, K. T. Simons, D. Baker, J. Mol. Biol. 277,

985 (1998).4. D. J. Brockwell, S. E. Radford, Curr. Opin. Struct. Biol. 17,

30 (2007).5. J. B. Udgaonkar, Annu. Rev. Biophys. 37, 489 (2008).6. J. P. Ellis, C. K. Bakke, R. N. Kirchdoerfer, L. M. Jungbauer,

S. Cavagnero, ACS Chem. Biol. 3, 555 (2008).7. M. S. Evans, I. M. Sander, P. L. Clark, J. Mol. Biol. 383,

683 (2008).8. S. T. Hsu et al., Proc. Natl. Acad. Sci. U.S.A. 104, 16516

(2007).9. A. Khushoo, Z. Yang, A. E. Johnson, W. R. Skach,

Mol. Cell 41, 682 (2011).10. J. Lu, C. Deutsch, Nat. Struct. Mol. Biol. 12, 1123 (2005).11. C. A. Woolhead, P. J. McCormick, A. E. Johnson, Cell 116,

725 (2004).12. S. Bhushan et al., Nat. Struct. Mol. Biol. 17, 313 (2010).13. R. Green, H. F. Noller, Annu. Rev. Biochem. 66,

679 (1997).14. S. Varenne, J. Buc, R. Lloubes, C. Lazdunski, J. Mol. Biol.

180, 549 (1984).15. G. Cannarozzi et al., Cell 141, 355 (2010).16. X. Qu et al., Nature 475, 118 (2011).17. C. Kimchi-Sarfaty et al., Science 315, 525 (2007).18. G. Zhang, M. Hubalewska, Z. Ignatova, Nat. Struct.

Mol. Biol. 16, 274 (2009).19. E. Siller, D. C. DeZwaan, J. F. Anderson, B. C. Freeman,

J. M. Barral, J. Mol. Biol. 396, 1310 (2010).20. S. T. Liang, Y. C. Xu, P. Dennis, H. Bremer, J. Bacteriol.

182, 3037 (2000).21. C. C. Chow et al., Biochemistry 42, 7090 (2003).22. A. H. Elcock, PLoS Comput. Biol. 2, e98 (2006).23. E. P. O’Brien, S. T. Hsu, J. Christodoulou, M. Vendruscolo,

C. M. Dobson, J. Am. Chem. Soc. 132, 16928 (2010).24. J. R. Moffitt, Y. R. Chemla, S. B. Smith, C. Bustamante,

Annu. Rev. Biochem. 77, 205 (2008).25. A. Ashkin, J. M. Dziedzic, J. E. Bjorkholm, S. Chu,

Opt. Lett. 11, 288 (1986).26. H. Dietz, M. Rief, Proc. Natl. Acad. Sci. U.S.A. 103,

1244 (2006).27. E. A. Shank, C. Cecconi, J. W. Dill, S. Marqusee,

C. Bustamante, Nature 465, 637 (2010).28. M. Matsumura, B. W. Matthews, Science 243, 792

(1989).29. J. Cellitti et al., Protein Sci. 16, 842 (2007).30. M. Llinás, S. Marqusee, Protein Sci. 7, 96 (1998).31. W. A. Baase, L. Liu, D. E. Tronrud, B. W. Matthews,

Protein Sci. 19, 631 (2010).32. Materials and methods and additional information are

available as supporting material on Science Online.33. N. R. Voss, M. Gerstein, T. A. Steitz, P. B. Moore,

J. Mol. Biol. 360, 893 (2006).34. C. Bustamante, J. F. Marko, E. D. Siggia, S. Smith,

Science 265, 1599 (1994).35. O. K. Dudko, G. Hummer, A. Szabo, Proc. Natl. Acad.

Sci. U.S.A. 105, 15755 (2008).36. P. Weinkam, E. V. Pletneva, H. B. Gray, J. R. Winkler,

P. G. Wolynes, Proc. Natl. Acad. Sci. U.S.A. 106, 1796(2009).

0

20

40

60

Extension (nm)

Forc

e (p

N)

0

20

40

60

Extension (nm)

Forc

e (p

N)

A B

C

0 20 40 600

20

40

60

"L C,Unfolding (nm)

F Unf

oldi

ng (p

N)

fragment (free)

full length (free)

full length (RNC)

50 nm 50 nm

Fig. 4. (A) Force-extension curves of a T4lysozyme (164 amino acids) with a 20–aminoacid C-terminal extension bound to the ribo-some (~150 amino acids outside the ribosomaltunnel), exhibiting no defined unfolding tran-sitions. (B) A free protein fragment of 149amino acids misfolds into a heterogeneousensemble of structures that unfold over a widerage of forces. (C) Comparison of unfoldingevents recorded for the free protein and the149–amino acid fragment. The full-lengthprotein unfolds within a narrow, stochasticrange of forces and extension changes, where-as the unfolding transitions of the isolatedfragment are highly heterogeneous.

unfolding

3.6 3.8 4.0 4.2 4.4 4.6

0.1

1

10

100

Force (pN)

k U!I

(1/s

)

3.6 3.8 4.0 4.2 4.4 4.6

0.1

1

10

100

Force (pN)

k I!U

(1/s

)

3.6 3.8 4.0 4.2 4.4 4.6

0.1

1

10

100

Force (pN)

k I!N

(1/s

)

A

B

0 10 20 30 400

5

10

15

20

"x (nm)

F (p

N)

10 15 20 253.5

4

4.5

5

"x (nm)

F (p

N)

FreeRibosome-bound

Cunfolding

refoldingrefolding

FreeRibosome-bound

xNative

xUnfolded

xIntermediate

freeribosome-bound

extension (x)

ener

gykI-N

kU-I

kI-U

Fig. 3. (A) Kinetic rates along the T4 lysozyme folding pathway.The rate of the final, irreversible step (kI-N) along the foldingpathway is significantly slower for the ribosome-bound protein(60–amino acid linker). (B) Distance changes upon unfolding (N!U,open symbols) and refolding (U ! N, filled symbols) at variousforces. At low forces, the distance is shorter than expected from a

WLCmodel (gray line), indicating a partial compaction of the polypeptide that does not resist forces above4 pN. The compact structure is stabilized in the ribosome-bound protein. Error bars: 95% CIs. (C)Schematic folding energy landscapes of free and ribosome-bound T4 lysozyme. The height of the barrierfrom I to N is affected by the ribosome, resulting in a decrease in kI-N.

23 DECEMBER 2011 VOL 334 SCIENCE www.sciencemag.org1726

REPORTS

on

Janu

ary

17, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

37. J. Cellitti, R. Bernstein, S. Marqusee, Protein Sci. 16,852 (2007).

38. C. Bustamante, Y. R. Chemla, N. R. Forde, D. Izhaky,Annu. Rev. Biochem. 73, 705 (2004).

39. R. Liu, J. E. Barrick, J. W. Szostak, R. W. Roberts,Methods Enzymol. 318, 268 (2000).

40. C. Voisset, S. J. Saupe, M. Blondel, Biotechnol. J. 6,668 (2011).

41. G. de Prat Gay et al., J. Mol. Biol. 254, 968 (1995).42. J. L. Neira, A. R. Fersht, J. Mol. Biol. 285, 1309 (1999).43. H. Taniuchi, J. Biol. Chem. 245, 5459 (1970).44. F. U. Hartl, M. Hayer-Hartl, Science 295, 1852 (2002).45. B. Schuler, E. A. Lipman, W. A. Eaton, Nature 419,

743 (2002).

46. F. Brandt et al., Cell 136, 261 (2009).47. H. X. Zhou, K. A. Dill, Biochemistry 40, 11289 (2001).48. J. Mittal, R. B. Best, Proc. Natl. Acad. Sci. U.S.A. 105,

20233 (2008).49. J. Kiraga et al., BMC Genomics 8, 163 (2007).

Acknowledgments: We thank S. Marqusee for the T4lysozyme source plasmid and for helpful discussions,O. Dudko for help with the analysis of force-ramp data,and J. R. Moffitt and members of the Bustamante and Tinocolaboratories for helpful discussions on the manuscript.C.M.K. acknowledges support from the NIH K99 Award5K99GM086516; C.M.K. and J.D.C. acknowledge fundingfrom the QB3 Institute, Berkeley (Distinguished Postdoctoral

Fellowship); D.H.G. acknowledges the NSF’s GraduateResearch Fellowship; I.T. acknowledges support from NIHgrant 5R01GM10840 and the Human Frontiers of ScienceProgram; and C.B. acknowledges support from NIH grant5R01GM32543.

Supporting Online Materialwww.sciencemag.org/cgi/content/full/334/6063/1723/DC1Materials and MethodsFigs. S1 to S11References (50–63)

13 June 2011; accepted 3 November 201110.1126/science.1209740

The Hedgehog Pathway PromotesBlood-Brain Barrier Integrity andCNS Immune QuiescenceJorge Ivan Alvarez,1* Aurore Dodelet-Devillers,1* Hania Kebir,1 Igal Ifergan,1 Pierre J. Fabre,2

Simone Terouz,1 Mike Sabbagh,1 Karolina Wosik,1 Lyne Bourbonnière,1 Monique Bernard,1

Jack van Horssen,3 Helga E. de Vries,3 Frédéric Charron,2 Alexandre Prat1†

The blood-brain barrier (BBB) is composed of tightly bound endothelial cells (ECs) and perivascularastrocytes that regulate central nervous system (CNS) homeostasis. We showed that astrocytessecrete Sonic hedgehog and that BBB ECs express Hedgehog (Hh) receptors, which togetherpromote BBB formation and integrity during embryonic development and adulthood. Usingpharmacological inhibition and genetic inactivation of the Hh signaling pathway in ECs, we alsodemonstrated a critical role of the Hh pathway in promoting the immune quiescence of BBB ECsby decreasing the expression of proinflammatory mediators and the adhesion and migration ofleukocytes, in vivo and in vitro. Overall, the Hh pathway provides a barrier-promoting effect andan endogenous anti-inflammatory balance to CNS-directed immune attacks, as occurs inmultiple sclerosis.

The blood-brain barrier (BBB) confers ho-meostasis to the central nervous system(CNS) and limits the entry of blood-borne

molecules and circulating leukocytes. The BBBis composed of specialized endothelial cells (ECs)held together by multiprotein complexes knownas junctional proteins (1, 2). Astrocytes, whichare closely apposed to the CNS vasculature, helpmaintain BBB integrity and immune quiescencethrough contact-dependent mechanisms and byreleasing soluble factors (2–5). BBB disruptionis a central and early feature of multiple sclero-sis (MS) that allows leukocytes to enter theCNS, leading to demyelination and neuronal dam-age (6–8).

The Hedgehog (Hh) pathway is involved inembryonic morphogenesis, neuronal guidance,

and angiogenesis (9, 10). In adult tissues, it par-ticipates in vascular proliferation, differentiation,and tissue repair (11–13). CNS morphogenicevents are primarily associated with Sonic Hh(Shh) signaling (14–16). Secreted Shh binds andinactivates the receptor Patched-1 (Ptch-1), allow-ing activation of Smoothened (Smo), which theninduces target genes through the Gli family oftranscription factors (17). Previous studies haveimplicated Shh signaling in MS and its animalmodel (18, 19). We therefore explored whethertheHh pathway contributes to themaintenance ofBBB functions, including its immune quiescence.

mRNA and protein analyses demonstrated ex-pression of Shh and its 45-kD uncleaved pre-cursor in human fetal astrocytes (HFAs) but noton primary cultures of BBBECs (fig. S1,A andB).In contrast, the highest levels of Ptch-1 and Smoexpression were observed in BBB ECs (fig. S1,A and B). Essential to Hh signaling is the auto-catalytic cleavage of the 45-kD Shh protein toyield a ~19-kD active form, which was present inastrocyte-conditioned media (ACM) (fig. S1B).We detected Shh in human and mouse astrocytesin vitro and in situ but not in ECs or pericytes(fig. S1, C to H). Conversely, Ptch-1 and Smowere detected on cultured BBBECs and in situ inhuman and mouse CNS ECs, but not on astro-

cytes or pericytes (fig. S1, C to H). Therefore, ourdata suggest that the Hh pathway is used by peri-vascular astrocytes to communicatewithBBBECs.

To determine whether astrocyte-secreted Shhinfluences BBB function, the transendothelialelectrical resistance (TEER) and permeability ofhuman BBB ECs were evaluated under condi-tions stimulating or abrogating the Hh pathway.We found that human recombinant Shh (hrShh)significantly increased TEER and decreased dex-tran 3-kD clearance, as well as permeability to14C-sucrose and bovine serum albumin–fluoresceinisothiocyanate (BSA-FITC) (Fig. 1, A toD). Thiseffect was comparable to the one induced byACM, and the response of BBB ECs to Shh didnot affect EC proliferation (fig. S2A). Activationof the Hh pathwaywas responsible for the barrier-promoting effect of ACM, because BBB ECstreated with the Smo agonist purmorphaminereproduced the effect of hrShh, whereas Smoantagonists cyclopamine (Fig. 1, A to D) andSANT-1 (fig. S2B) reversed the effect of ACM.These data reflect a combination of changes inTEER, paracellular permeability, and inducibletransport mechanisms, suggesting that Hh ligandscould affect both processes, possibly through theregulation of junctional proteins.

Hh activation is associated with a restric-tive permeability, therefore we determined theexpression of the transcription factors Gli-1 andsex-determining region Y-box–18 (SOX-18), animportant regulator of junctional protein expres-sion (i.e., claudin-5) and barrier formation (20).The number of BBB ECs expressing Gli-1 in-creased upon Hh activation, as compared to un-treated cells. However, cyclopamine reduced theexpression of Gli-1 induced by ACM (Fig. 1, Eand F). hrShh stimulation and ACM significantlyup-regulated Hip, Gli-1, and SOX-18 transcrip-tion (Fig. 1G and fig. S2C). SOX-18 expressionwas maximal 2 hours after Gli-1 activation, con-firming that SOX-18 is regulated by the Hh path-way, possibly through Gli-1.

To further confirm the role of the Hh pathwayin maintaining BBB properties in vivo, we in-jected mice with cyclopamine and studied theextent of BBB leakage. Cyclopamine inducedacute BBB disruption (6 hours after injection), asdemonstrated by the increased extravasation ofexogenous dextran and endogenous fibrinogen(Fig. 1, H and I). Cyclopamine treatment also

1Neuroimmunology Unit, Center of Excellence in Neuromics,Centre de Recherche du Centre Hospitalier de l’Université deMontréal, Faculty of Medicine, Université de Montréal, Mon-tréal, Quebec, H2L 4M1, Canada. 2Laboratory of Molecular Bi-ology of Neural Development, Institut de Recherches Cliniquesde Montréal and Department of Medicine, Université de Mon-tréal, Montréal, Quebec, H2W 1R7, Canada. 3Department of Mo-lecular Cell Biology and Immunology, VUMedical Center, 1007MB Amsterdam, Netherlands.

*These authors contributed equally to this work.†To whom correspondence should be addressed. E-mail:a.prat@umontreal.ca

www.sciencemag.org SCIENCE VOL 334 23 DECEMBER 2011 1727

REPORTS

on

Janu

ary

17, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m