IMMUNOLOGY

Structural basis of latent TGF-b1presentation and activation by GARPon human regulatory T cellsStéphanie Liénart1*, Romain Merceron2,3*, Christophe Vanderaa1, Fanny Lambert1,Didier Colau4, Julie Stockis1,5, Bas van der Woning6, Hans De Haard6,Michael Saunders6, Pierre G. Coulie1,5, Savvas N. Savvides2,3†, Sophie Lucas1,5†

Transforming growth factor–b1 (TGF-b1) is one of very few cytokines produced in a latentform, requiring activation to exert any of its vastly diverse effects on development, immunity,and cancer. Regulatory T cells (Tregs) suppress immune cells within close proximity byactivating latent TGF-b1 presented by GARP (glycoprotein A repetitions predominant) tointegrin aVb8 on their surface.We solved the crystal structure of GARP:latent TGF-b1 boundto an antibody that stabilizes the complex and blocks release of active TGF-b1. Thisfinding reveals how GARP exploits an unusual medley of interactions, including foldcomplementation by the amino terminus of TGF-b1, to chaperone and orient the cytokinefor binding and activation by aVb8. Thus, this work further elucidates the mechanismof antibody-mediated blockade of TGF-b1 activation and immunosuppression by Tregs.

Cytokines of the transforming growth factor–b(TGF-b) family exert widespread and di-verse effects on metazoan cells (1). Owingto their potency and pleiotropy, family mem-bers such as the closely related TGF-b1, -b2,

and -b3 isoforms are produced as latent, inactivecytokines, which require a tightly regulated stepof extracellular activation to acquire the ability tobind their cognate heterotetrameric receptor (2, 3).The mechanisms of TGF-b activation appear tobe isoform- and cell type–specific, but none havebeen fully elucidated on a structural basis.TGF-b1 is the predominant isoform expressed

by immune cells, including the immunosuppres-sive FOXP3+ CD4+ regulatory T cells known asTregs (4). Immunosuppression by Tregs preventsautoimmunity but is detrimental in cancer orchronic infections (5). Tregs suppress other im-mune cells within close proximity by activatinglatent TGF-b1 on their surface (6). This resultsfrom the following multistep process. Newly syn-thesized pro-TGF-b1 homodimers form disul-fide bonds with GARP (also known as LRRC32),a transmembrane protein retained at the sur-face of Tregs and a few other cell types (7–11).Pro-TGF-b1 is then cleaved by furin to producelatent TGF-b1, in which the C-terminal dimer, ormature TGF-b1 (mTGF-b1), remains noncovalentlyassociated to the N-terminal dimer known as LAP(latency associated peptide). Latent TGF-b1 is in-active because the two LAP monomers, by means

of their arm and “straightjacket” domains, forma ring around mTGF-b1, masking the interactionsites with the TbRI and TbRII receptor chains (12).TGF-b1 activation entails the release of mTGF-b1from LAP. Tensile force exerted by the cyto-skeleton and transmitted by integrin aVb6 to LAPis hypothesized to unfasten and elongate thestraightjacket, leading to the release and activa-tion of mTGF-b1 (13). However, this mechanism ofactivation is restricted to cells expressing aVb6.Tregs lack aVb6. Instead, latent TGF-b1 presentedby GARP on the surface of Tregs is activated byintegrin aVb8. This was shown using antibodiesagainst either the GARP:(latent)TGF-b1 complex orthe b8 integrin subunit, which block TGF-b1 acti-vation and immunosuppression by Tregs (6, 14).Despite these advances in our understandingof TGF-b1 activation, the mechanism by whichGARP structurally contributes to activation viaintegrin aVb8 has eluded the field. Thus, wesought to determine the crystal structure of GARPbound to latent TGF-b1 by using a specific helperantibody to stabilize the complex.We first coexpressed the extracellular domain

of human GARP (GARPECD) and latent TGF-b1 andpurified a soluble form of the GARPECD:TGF-b1binary complex for x-ray crystallography. As thiscomplex proved recalcitrant to crystallization,we reasoned that a ternary assembly with a Fabfragment might be more amenable to crystalliza-tion. We chose to use the Fab derived from themonoclonal antibody MHG-8 because it bindsGARP:TGF-b1 complexes but not free GARP andblocks TGF-b1 activation by Tregs (6). Resolvingsuch a complex would also provide key insightsinto therapeutically relevant epitopes.Preparative reconstitution of GARPECD:TGF-

b1:MHG-8Fab complexes yielded crystals thatenabled determination of a structure to 3.1-Åresolution (fig. S1 and table S1). The structurefeatures a tripartite assembly obeying a 1:1:1 stoi-

chiometry, comprising one molecule of GARPECD,one complex of latent TGF-b1 (i.e., one LAPA/LAPB and one mTGF-b1A/mTGF-b1B dimer), andone copy of MHG-8Fab (Fig. 1A and table S2).The structure provided a snapshot of how latentTGF-b1 is sequestered by GARP via an intricatecombination of covalent and noncovalent inter-actions. Tethering of latent TGF-b1 by GARPoccurred on the opposite side from the RGDintegrin-binding motifs in LAP (9, 13), in anorientation compatible with integrin aVb8 bind-ing (Fig. 1A).GARPECD adopted a solenoid horseshoe struc-

ture with a nearly perfect circular curvature andfeatured 20 leucine-rich repeats (LRRs, R1 toR20) capped by N- and C-terminal motifs (Fig. 1B).Each repeat, comprising a consensus sequenceof 21 to 29 residues, manifested a hydrophobiccore running through the horseshoe solenoid(Fig. 1C). The concave side of the horseshoe waslined by an extended b sheet of parallel b strandscontributed by each of the 20 repeats and thetwo capping motifs (Fig. 1, B to D). This sidealso displayed N-linked glycosylation at four ofthe five predicted sites (Fig. 1B). The convex sidedisplayed an irregular mix of secondary structuresconnected by loops of various lengths (Fig. 1Band fig. S2). The GARP scaffold was furtherstabilized by two intrachain disulfide bonds:one (Cys26-Cys35) anchoring a b hairpin in theN-terminal cap and a second (Cys425-Cys436) an-choring the ends of an otherwise disordered loopprotruding from the convex side of R15 (Fig. 1B).This structure of GARP may serve as a struc-tural prototype for LRRC33, the only other LRRprotein known to bind and present latent TGF-b1 on cell surfaces and that shares high levels ofsequence identity with GARP (fig. S2) (15, 16).The most mechanistically relevant insights

from our structure center on the GARPECD:TGF-b1interface (Fig. 2 and figs. S2 to S4). Half (28 of56) of the GARP residues interacting with latentTGF-b1 were conserved in LRRC33 (fig. S2). Thus,LRRC33 may interact with latent TGF-b1 in amanner similar to GARP. Most residues (22 of 33)of latent TGF-b1 involved in the interaction withGARP were conserved in TGF-b2 and -b3, whichalso associate with GARP (figs. S4 and S5, A andB) (17). Thus, GARP is poised to actively regulatethe bioavailability of all three TGF-b isoforms.The GARPECD:TGF-b1 complex unmasked rare-

ly seen combinations of structural features, notonly due to the diversity and extent of its inter-action landscape but also because it involved atwofold-symmetric growth factor (latent TGF-b1),bound by a molecular partner (GARP), that lacksany element of symmetry. Our structure revealedthat latent TGF-b1 uses both copies of LAP andonly one of the two mTGF-b1 modules to saddlethe side of the GARP horseshoe. This results inthree main interfaces, namely GARP:mTGF-b1B,GARP:LAPA, and GARP:LAPB (Fig. 2A). The bind-ing footprint of latent TGF-b1 to GARP com-prised the ascending lateral and convex surfacesroughly halfway into the horseshoe (R4 to R11)and buried a total of ~1750 Å2 of molecularsurface area. Latent TGF-b1 interacted with a

RESEARCH

Liénart et al., Science 362, 952–956 (2018) 23 November 2018 1 of 5

1de Duve Institute, UCLouvain, 1200 Brussels, Belgium.2Department of Biochemistry and Microbiology, GhentUniversity, 9052 Ghent, Belgium. 3VIB Center for InflammationResearch, 9052 Ghent, Belgium. 4Ludwig Cancer Research,Brussels, Belgium. 5Walloon Excellence in Lifesciences andBiotechnology, 1300 Wavre, Belgium. 6argenx, 9052 Zwijnaarde,Belgium.*These authors contributed equally to this work.†Corresponding author. Email: sophie.lucas@uclouvain.be (S.Lu.);savvas.savvides@ugent.be (S.N.S.)

on March 15, 2020

http://science.sciencem

ag.org/D

ownloaded from

predominantly hydrophobic cushion in GARPvia van der Waals contacts mediated by the N-terminal loop of LAPA and LAPB, by the loopcomprising residues 323 to 337 in mTGF-b1B,and by helix a3 in mTGF-b1B (Fig. 2A, i).Two structural features were particularly no-

table in the interfaces of GARP with the LAPmodules. First, latent TGF-b1 was covalentlytethered to GARP via two disulfide bonds: onebetween Cys33 in LAPA and Cys211 in GARP, theother between Cys33 in LAPB and Cys350 in GARP(Fig. 2A, ii and iii) (9). Covalent linkage betweendistinct proteins is an unusual mode of interac-tion when redox regulation is not at play. TheLAPA–GARP disulfide was nested in a hydropho-bic pocket capped by Trp297 of GARP (Fig. 2A, ii).Second, the N-terminal loop of LAPA inserted

into a hydrophobic pocket formed by strandsb12 in R6 and b14 in R7 of GARP (Fig. 2A, ii).This unanticipated structural feature led to com-plementation of the GARP horseshoe fold by ab-strand addition contributed by LAPA (strandb1A), supporting a role of GARP as a molecularchaperone for latent TGF-b1 biogenesis. GARPsealed this unusual interaction by inserting a bhairpin formed by residues 290 to 299. Notably,this hairpin was the only ordered part within along segment (residues 280 to 317) that origi-nates at R10 and was predicted to be intrinsi-cally disordered (Fig. 1B and fig. S2). A bindingmode of this nature was proposed to partiallycompensate for the high entropic cost associ-ated with the binding of intrinsically disorderedsegments to partner molecules. However, struc-

tural snapshots of this phenomenon have onlyrarely been experimentally observed (18).Our structure unveiled the structural plastic-

ity of latent TGF-b1 and conformational changesthat underlie its sequestration by GARP. Thestructure of latent TGF-b1 bound to GARP wasglobally similar to that of unbound porcine latentTGF-b1 (12), with structural disorder in the bow-tie region that can associate with aVb6 integrin(13). However, it displayed three drastic localconformational changes compared with allTGF-b1 structures known to date. First, helix a3in mTGF-b1B became a 310 helix element (h1) andwas rotated upward by 45° (Fig. 2B). Second, thepreceding loop with residues 323 to 337 wassubstantially restructured toward the GARP con-cavity (Fig. 2B). Such extensive conformational

Liénart et al., Science 362, 952–956 (2018) 23 November 2018 2 of 5

Fig. 1. Structure of the GARPECD:TGF-b1:MHG-8Fab complex. (A) (Left) Cartoonrepresentations of the GARPECD:TGF-b1:MHG-8Fabstructure as determined by x-ray crystallography.Latent TGF-b1 is composed of two LAP molecules(LAPA and LAPB) and two mTGF-b1 molecules(mTGF-b1A and mTGF-b1B). The red dashed linedenotes the linker between the extracellulardomain and the transmembrane domain ofGARP. (Top right) Cartoon representation of adifferent view of the GARP:TGF-b1 complex,with MHG-8Fab not illustrated for clarity.(Bottom right) Schematic representation ofprotein components participating in theGARPECD:TGF-b1 complex. Yellow circles repre-sent interchain disulfide bonds. (B) Cartoonrepresentation and close-up view of theGARPECD. Secondary structure elements,modeled N-glycans, and intramolecular disulfidebonds are depicted. (C) Cartoon representationof a typical LRR fragment (R6) in GARP. Byconvention, the loop fragments connecting theconcave to the convex side of the repeats formthe lateral ascending face, whereas the loopsconnecting the convex to the concave side formthe lateral descending face. (D) Cartoonrepresentation and b-strand complementationobserved in the GARP structure. The hydrogenbonding network is represented with blackdashed lines. IC, intracellular; EC, extracellular;HC, heavy chain; LC, light chain; N-ter,N terminus; C-ter, C terminus; R, LRR(leucine-rich repeat); RNT, LRR N terminus;RCT, LRR C terminus; RGD, RGDLXX(I/L)binding motif for RGD-binding integrins.

A

B C

D

RESEARCH | REPORTon M

arch 15, 2020

http://science.sciencemag.org/

Dow

nloaded from

switching resolved otherwise severe steric clashesand introduced Tyr336 of mTGF-b1B into a newstructural context stabilized by a stacking in-teraction with Tyr137 of GARP and by hydrogenbonding between its main-chain atoms andSer160 of GARP (Fig. 2A). Third, helix a1 inLAPB, which mediates interactions between

LAPB and GARP, underwent marked unwindingwhile remaining physically tethered to GARPvia the LAPB-GARP disulfide (Fig. 2, A and C).Unwinding of helix a1 was also seen in LAPA.Collectively, such extensive structural plasticityin latent TGF-b1 in a region far from the integrinbinding site may explain why it can associate

with two structurally very distinct classes of pro-teins. Namely, GARP or LRRC33 on one handand LTBPs (latent TGF-b binding proteins) onthe other (15).The MHG-8Fab used to generate the crystals

blocked TGF-b1 activation by human Tregs as ef-ficiently as the full MHG-8 antibody (Fig. 3A). Thus,our crystal structure also revealed how an antibodycan block TGF-b1 activation from GARP:TGF-b1complexes and antagonize Treg suppression (6).MHG-8 bound a mixed conformational epitope,concomitantly contacting residues from GARPand latent TGF-b1 (Figs. 1A and 3, B to D). Spe-cifically, CDR1 and CDR3 of the MHG-8 lightchain and CDR3 of the heavy chain contactedthe convex side of GARP, whereas the heavy chainalso contacted the C-terminal arm domain ofLAPB and the 310 helix h1 of mTGF-b1B via itsCDR2, CDR3, and framework regions (Fig. 3, Cand D, and figs. S6 to S8). We confirmed involve-ment of certain residues by measuring MHG-8binding to GARP:TGF-b1 complexes includinghemagglutinin (HA)–tagged mutant forms ofGARP or latent TGF-b1 expressed in human em-bryonic kidney (HEK) 293T cells. Mutations atthree sites in GARP (Leu140→Lys and Glu142→Leu, Arg143→Ala, or Arg163→Glu) and one site inmTGF-b1 (Lys338→Glu) substantially decreasedbinding of MHG-8 (Fig. 3B). A majority (9 of 17) ofLAP or mTGF-b1 residues in contact with MHG-8were not conserved in the corresponding po-sitions of TGF-b2 and -b3. This explains whyMHG-8 does not bind GARP:TGF-b3 complexesand only binds GARP:TGF-b2 with very low af-finity (figs. S4 and S5C).The antagonistic activity of MHG-8 on Tregs

may result from its ability to impair the releaseof mTGF-b1 induced by combined pulling byintegrin aVb8 on one side and resistance orsynergistic pulling by GARP on the other side ofLAP (Figs. 3C and 4). By simultaneously con-tacting amino acids in the arm of LAPB and inGARP, MHG-8 may reinforce the fastening of thestraightjacket, preventing the elongation re-quired to release mTGF-b1 (Figs. 3C and 4). Al-ternatively, by cross-linking amino acids in GARPwith the 310 helix h1 of mTGF-b1B, MHG-8 mayprevent the release of mTGF-b1 and immobilizeits 310 helix h1 in a position that is incompatiblewith binding to the TbRI chain of its receptor,even if the straightjacket is elongated and re-laxed. In either scenario, MHG-8 sterically pre-vents mTGF-b1 release by engaging both GARPand latent TGF-b1 while still allowing binding ofaVb8, as demonstrated previously (14).Thus, we elucidate the role of GARP in TGF-

b1 biogenesis and activation, as well as themechanistic principles underlying antibody-mediated blockade of TGF-b1 activation andimmunosuppression by human Tregs. Visualiza-tion of this large molecular assembly illustratesthe feasibility of blocking TGF-b1 activity ema-nating from a precisely defined and restrictedcellular source, such as the surface of Tregs. Thiswork also suggests how other TGF-b1–presentingproteins, such as LTBPs deposited in the extra-cellular matrix (2, 19), may operate as alternative

Liénart et al., Science 362, 952–956 (2018) 23 November 2018 3 of 5

Fig. 2. GARP:TGF-b1 interfaces and plasticity of latent TGF-b1. (A) Cartoon representation andclose-up views on the three main GARP:TGF-b1 interfaces, outlined by dashed boxes. Details of theGARP:mTGF-b1B, GARP:LAPA, and GARP:LAPB interfaces are depicted in panels (i), (ii), and (iii),respectively. Proteins are shown in cartoon representation, and secondary structure elements areindicated.The main interfacing residues are shown as sticks, and polar interactions are depicted withblack dashed lines. Only contacts observed in both copies of the complex in the asymmetric unitare shown. Disulfide bonds are represented by sticks and spheres, and sulfur atoms are colored yellow.(B) Close-up view on the GARP:mTGF-b1B interface, with structural superposition of the porcinemTGF-b1 from porcine latent TGF-b1 (chain D, PDB code 3RJR) showing the main conformationalchange observed in our structure. (C) Close-up view on the GARP:LAPA interface, with structuralsuperposition of LAPB in our structure and the two LAP in porcine latent TGF-b1 (chains C and D,PDB code 3RJR) showing a second main conformational change in our structure. This is observed asan unwinding of the LAP helix a1. The disulfide bond between LAPA and GARP is depicted.

RESEARCH | REPORTon M

arch 15, 2020

http://science.sciencemag.org/

Dow

nloaded from

Liénart et al., Science 362, 952–956 (2018) 23 November 2018 4 of 5

Y13

7H

R14

3AT

162D

R16

3ET

165A

Y33

6A

GARP

LAP

S13

8GL

140K

/E14

2L

R16

6M/H

167E

E18

9A

E10

0AE

146R

Q26

9YH

270Y

Q27

2YY

284A

S33

7AK

338E

mTGF- 1

latent TGF- 1

STIM:

-F

GT-it

na (mA

b 1D

11)

- + + + +

Antibodies: - - MH

G-8

mA

b

MH

G-8

Fab

mo

use

IgG

1is

otyp

e co

ntr o

l

+pSMAD2

-ACTIN 40 kDa

55 kDa

0.0

0.5

1.0

1.5

Rat

iop

SM

AD

2/-A

CT

IN

CDR1

CDR2

CDR3

Heavy chainLight chain

CDR1

CDR2

CDR3

i. MHG-8VL:GARP ii. MHG-8VH:GARP iii. MHG-8VH:LAPB and :mTGF- 1B

Ser116

Gly118

Thr113Thr162

Thr165

Trp71CDR2

Ser75CDR2

His167

Leu145

Arg143

3

4

His48CDR1

Lys50CDR1

Trp112CDR3

Trp52CDR1

Tyr11CDR3

Tyr123CDR3

Tyr123CDR3

Tyr336

Lys338

Asn92

Glu100

Tyr120CDR3

Asp122CDR3

Ser138

Tyr137Arg143

Arg166

Glu189

Arg163

Leu140

Asp124CDR3

Asn119CDR3

A

B

C

D

Res

idu

al b

ind

ing

(vs

WT

)

MHG-8VH

MHG-8VL

LAPA

mTGF- 1B

mTGF- 1A

GARPECD

LAPB

Binding footprint

1

3

2

5

6

7

0%

20%

40%

60%

80%

100%

Fig. 3. Structural description of the MHG-8Fab:GARP and MHG-8:TGF-b1 interfaces. (A) Immunoblot analysis of SMAD2 phosphorylation(pSMAD2) as a readout for TGF-b1 activation in human Tregs stimulated inthe presence or absence of the indicated antibodies (20 mg/ml) or Fab(13 mg/ml). STIM, stimulation with anti-CD3– and anti-CD28–coated beads;mAb, monoclonal antibody; IgG1, immunoglobulin G1. (B) Binding ofMHG-8 to transfected HEK293T cells expressing GARP:TGF-b1 complexescomprising the indicated N-terminally HA-tagged mutant forms of GARP orTGF-b1. Bar graphs indicate residual binding to a given mutant complex bycomparison with the wild-type (WT) complex taken as 100% (meanof three to five experiments). Error bars represent the 99% confidenceinterval around the mean. Mutations in red are associated with significantlyreduced binding compared with the WT, as their confidence interval doesnot contain the 100% binding threshold indicated by the dashed horizontalline. (C) (Left) Close-up view on the MHG-8Fab:GARP and MHG-8:TGF-b1

interfaces. GARP and TGF-b1 are shown in surface representation, and thevariable region of MHG-8Fab is depicted as a cartoon. CDR1, -2, and -3 loopsare displayed in orange, yellow, and black, respectively. (Right) Schematicrepresentation of the overall structure and contact area by MHG-8VH (darkblue) and MHG-8VL (light blue). (D) Details of the MHG-8VL:GARP, MHG-8VH:LAPA, and MHG-8VH:LAPB:mTGF-b1B interfaces are depicted in panels (i),(ii), and (iii), respectively. Proteins are represented as cartoons, and sec-ondary structure elements are indicated. Main interfacing residues are shownas sticks, and polar interactions are indicated by black dashed lines. Onlycontacts observed in both copies of the complex in the asymmetric unit areshown. CDR, complementarity determining region; VH, variable domain ofthe heavy chain; VL, variable domain of the light chain. Single-letterabbreviations for amino acid residues are as follows: A, Ala; C, Cys; D, Asp;E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro;Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

Fig. 4. Model for the integrin- and GARP-dependentmechanical activation of TGF-b1 and itsantagonism by monoclonal antibody MHG-8.A schematic representation of the GARPECD:TGF-b1:MHG-8Fab complex, with the MHG-8 bindingfootprint depicted in blue, is shown. Visualizationof the covalent sequestration of latent TGF-b1 byGARP suggests how GARP may facilitate TGF-b1activation by integrin aVb8 on the Treg surface.Structural analyses of a pro-TGF-b1 monomerbound by the integrin aVb6 head show howcytoskeletal forces elicit reshaping of latent TGF-b1to release mTGF-b1 from LAP (13). Maximal har-nessing of such forces depends on the directionof the force vector (20). Thus, GARP may providestrong anchoring to the cell surface, with itsjuxtamembrane region providing the necessaryflexibility that allows for the optimal alignment ofpulling-force vectors. Our crystal structure alsosuggests how an antibody can block TGF-b1activation from GARP:TGF-b1 complexes on Tregs bycross-linking GARP to one LAP and one mTGF-b1monomer, which sterically prevents mTGF-b1 releasewhile still allowing integrin aVb8 binding.

LAPA

Unfastened straightjacket

Pulling force

activation

MHG-8

MHG-8 binding footprint

+mTGF- 1B

mTGF- 1A

GARPECD

integrin

LAPB

EC

IC

Resistance or pulling force

EC

IC

RGDRGD

RESEARCH | REPORTon M

arch 15, 2020

http://science.sciencemag.org/

Dow

nloaded from

sources of active TGF-b1 and may also be spe-cifically targeted by antibody- or small-molecule–based reagents. Such structural and mechanisticinsights may facilitate the design of particularlyspecific approaches to treat various diseasesassociated with altered TGF-b1 activity and/ordysfunctional Treg responses, most notablyfor cancer immunotherapy.

REFERENCES AND NOTES

1. C. J. David, J. Massagué, Nat. Rev. Mol. Cell Biol. 19, 419–435 (2018).2. A. P. Hinck, T. D. Mueller, T. A. Springer, Cold Spring Harb.

Perspect. Biol. 8, a022103 (2016).3. I. B. Robertson, D. B. Rifkin, Cytokine Growth Factor Rev. 24,

355–372 (2013).4. M. A. Travis, D. Sheppard, Annu. Rev. Immunol. 32, 51–82 (2014).5. H. Nishikawa, S. Sakaguchi, Curr. Opin. Immunol. 27, 1–7 (2014).6. J. Cuende et al., Sci. Transl. Med. 7, 284ra56 (2015).7. E. Gauthy et al., PLOS ONE 8, e76186 (2013).8. J. Stockis, O. Dedobbeleer, S. Lucas, Mol. Biosyst. 13,

1925–1935 (2017).9. R. Wang et al., Mol. Biol. Cell 23, 1129–1139 (2012).10. J. Stockis, D. Colau, P. G. Coulie, S. Lucas, Eur. J. Immunol. 39,

3315–3322 (2009).11. D. Q. Tran et al., Proc. Natl. Acad. Sci. U.S.A. 106,

13445–13450 (2009).

12. M. Shi et al., Nature 474, 343–349 (2011).13. X. Dong et al., Nature 542, 55–59 (2017).14. J. Stockis et al.,Proc. Natl. Acad. Sci. U.S.A. 114, E10161–E10168 (2017).15. Y. Qin et al., Cell 174, 156–171.e16 (2018).16. A. C. Ng et al., Proc. Natl. Acad. Sci. U.S.A. 108 (suppl. 1),

4631–4638 (2011).17. B. X. Wu et al., J. Biol. Chem. 292, 18091–18097 (2017).18. T. Flock, R. J. Weatheritt, N. S. Latysheva, M. M. Babu,

Curr. Opin. Struct. Biol. 26, 62–72 (2014).19. I. B. Robertson, D. B. Rifkin, Cold Spring Harb. Perspect. Biol. 8,

a021907 (2016).20. Y. Chen, S. E. Radford, D. J. Brockwell, Curr. Opin. Struct. Biol.

30, 89–99 (2015).

ACKNOWLEDGMENTS

We thank S. Depelchin for editorial assistance and J. Cuende forearly technical assistance. Proteros biostructures GmbHperformed the crystallization and synchrotron radiationexperiments. Funding: This work was supported by grants fromWELBIO (grant BF-2014-01) and the ERC (Consolidator grantTARG-SUP 682818) to S.Lu., and from the VIB to S.N.S. S.Li.was supported by a Télévie fellowship (F.R.S.-FNRS). Authorcontributions: S.Li., R.M., B.v.d.W., H.D.H., M.S., P.G.C., S.N.S.,and S.Lu. analyzed the data; S.Li., R.M., S.N.S., and S.Lu. wrote themanuscript; S.Lu. conceived of the study; S.Li., C.V., F.L., D.C.,and J.S. performed the mutagenesis experiments, bindingexperiments and experiments with Tregs and produced and purifiedthe mAb; B.v.d.W., H.D.H., and M.S. produced the GARP:latent

TGF-b1 complexes; and R.M. performed crystallographicrefinement and structure analysis with contributions from S.N.S.Competing interests: The authors declare the following financialcompeting interests: B.v.d.W., H.D.H., and M.S. are full-timeemployees of argenx BVBA. Patents pertaining to the resultspresented in the paper have been filed under the PatentCooperation Treaty (International Publication Number WO2016/125017 A1), with S.Lu., S.Li., B.v.d.W., H.D.H., P.G.C., andM.S. as inventors and UCLouvain and argenx as applicants. S.Lu.has received research support from argenx and owns stockoptions in argenx. Data and materials availability: Atomiccoordinates and structure factor files have been deposited inthe Protein Data Bank (PDB) with accession code 6GFF. TheMHG-8 antibody is available from S.Lu. under a material transferagreement with the UCLouvain. All other data needed toevaluate the conclusions of this study are present in the paperor the supplementary materials.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/362/6417/952/suppl/DC1Materials and MethodsFigs. S1 to S8Tables S1 to S3References (21–24)

25 May 2018; accepted 10 October 2018Published online 25 October 201810.1126/science.aau2909

Liénart et al., Science 362, 952–956 (2018) 23 November 2018 5 of 5

RESEARCH | REPORTon M

arch 15, 2020

http://science.sciencemag.org/

Dow

nloaded from

cells1 presentation and activation by GARP on human regulatory TβStructural basis of latent TGF-

Hans De Haard, Michael Saunders, Pierre G. Coulie, Savvas N. Savvides and Sophie LucasStéphanie Liénart, Romain Merceron, Christophe Vanderaa, Fanny Lambert, Didier Colau, Julie Stockis, Bas van der Woning,

originally published online October 25, 2018DOI: 10.1126/science.aau2909 (6417), 952-956.362Science 

, this issue p. 952Science activity, including cancer immunotherapy.reg1 functionality and dysfunctional Tβwith altered TGF-

1 release. These structural and mechanistic insights may inform treatments of diseasesβmembrane-associated TGF-monoclonal antibody (MHG-8) that binds to the complex. In so doing, they also demonstrate how MHG-8 prevents

1, using a crystal structure in which the complex was stabilized using a Fab fragment from aβinteracts with TGF- reveal how GARPet al.GARP, which acts to chaperone and orient the cytokine for activation at the cell surface. Liénart

1 homodimers form disulfide bonds with the transmembrane proteinβ1). Within the cell, newly synthesized pro-TGF-β1 (TGF-β−mechanism involves the activation of a surface-bound latent form of the cytokine transforming growth factor

) can suppress immune responses through a variety of mechanisms. One suchregsRegulatory T cells (T1 regulation by GARPβVisualizing TGF-

ARTICLE TOOLS http://science.sciencemag.org/content/362/6417/952

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2018/10/24/science.aau2909.DC1

CONTENTRELATED

http://stke.sciencemag.org/content/sigtrans/12/570/eaav5183.fullhttp://stm.sciencemag.org/content/scitransmed/6/258/258ra142.fullhttp://stm.sciencemag.org/content/scitransmed/7/315/315ra189.fullhttp://stm.sciencemag.org/content/scitransmed/5/215/215ra174.fullhttp://stm.sciencemag.org/content/scitransmed/3/83/83ra42.full

REFERENCES

http://science.sciencemag.org/content/362/6417/952#BIBLThis article cites 24 articles, 9 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Science. No claim to original U.S. Government WorksCopyright © 2018 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of

on March 15, 2020

http://science.sciencem

ag.org/D

ownloaded from