IMMUNE ENGINEERING Selective targeting of engineered · IMMUNE ENGINEERING Selective targeting of engineered T cells using orthogonal IL-2 cytokine-receptor complexes Jonathan T.

IMMUNE ENGINEERING

Selective targeting of engineeredT cells using orthogonal IL-2cytokine-receptor complexesJonathan T. Sockolosky,1,2 Eleonora Trotta,3* Giulia Parisi,4* Lora Picton,1 Leon L. Su,1

Alan C. Le,5 Akanksha Chhabra,5 Stephanie L. Silveria,3 Benson M. George,2,5,6

Indigo C. King,7 Matthew R. Tiffany,8 Kevin Jude,1 Leah V. Sibener,1,9

David Baker,7 Judith A. Shizuru,5 Antoni Ribas,4,10

Jeffrey A. Bluestone,3,10 K. Christopher Garcia1,2,10,11†

Interleukin-2 (IL-2) is a cytokine required for effector T cell expansion, survival, andfunction, especially for engineered T cells in adoptive cell immunotherapy, but itspleiotropy leads to simultaneous stimulation and suppression of immune responses as wellas systemic toxicity, limiting its therapeutic use. We engineered IL-2 cytokine-receptororthogonal (ortho) pairs that interact with one another, transmitting native IL-2 signals, butdo not interact with their natural cytokine and receptor counterparts. Introduction oforthoIL-2Rb into T cells enabled the selective cellular targeting of orthoIL-2 to engineeredCD4+ and CD8+ T cells in vitro and in vivo, with limited off-target effects and negligibletoxicity. OrthoIL-2 pairs were efficacious in a preclinical mouse cancer model of adoptivecell therapy and may therefore represent a synthetic approach to achieving selectivepotentiation of engineered cells.

Adoptive transfer of tumor-reactive T cellshas evolved into a clinically useful therapycapable of inducing antitumor immunityin patients (1, 2). However, the broad ap-plication of adoptive T cell transfer (ACT)

therapies to treat cancer has several limitations,including the production of sufficient quantitiesof cells for infusion and the failure of transferredT cells to persist and remain functional in vivo.In the clinic, the concomitant administration ofthe T cell growth factor interleukin-2 (IL-2) im-proves the survival, function, and antitumor ac-tivity of transplanted T cells (3, 4). However, theuse of IL-2 to potentiate ACT is complicated bythe pleiotropic nature of IL-2,which induces both

immune stimulatory and suppressive T cell re-sponses as well as potentially severe toxicities (5).This is governed by the interaction between IL-2and the IL-2 receptor (IL-2R), which consists ofa, b, and g subunits (6). IL-2Rb and the commong-chain (IL-2Rg) together form the signaling dimerand bind IL-2 with moderate affinity, whereasIL-2Ra (CD25) does not signal but increases theaffinity of IL-2 for the binary (bg) IL-2 receptorto sensitize T cells to low concentrations of IL-2.The activity of IL-2 as an adjuvant to ACT is

dependent on the balance between activation oftransplanted and endogenous T cell subsets bear-ingnatural IL-2 receptors, aswell as host responsesthat cause dose-limiting toxicities. Strategies toovercome these limitations could improve T cellimmunotherapy (7, 8). Recognizing the need fornew approaches that afford precise targeting ofIL-2–dependent functions to a specific cell typeof interest, we devised a strategy to redirect thespecificity of IL-2 toward adoptively transferredT cells. This method, based on receptor-ligandorthogonalization, uses a mutant IL-2 cytokineand mutant IL-2 receptor that bind specificallyto one another but not to their wild-type counter-parts (Fig. 1A).We focused on the murine IL-2/IL-2Rb inter-

action to enable in vivo characterization in syn-geneic mouse models. The IL-2Rb chain waschosen as the mutant receptor because the bchain is required for signal transduction andcan bind IL-2 independently. We devised a two-step approach to engineer orthogonal IL-2/IL-2Rb pairs informed by the crystal structure of theIL-2 high-affinity receptor complex (6) (Fig. 1B).First, point mutations of the IL-2Rb chain wereidentified from inspection of the interface be-tween IL-2 and IL-2Rb that abrogated binding

to wild-type IL-2 (Fig. 1, C to E). The IL-2Rb hot-spot residues His134 and Tyr135 make numerouscontacts with IL-2 that contribute a majority ofthe binding free energy between IL-2 and IL-2Rb(6) (Fig. 1E). A double mutant IL-2Rb [His134 →Asp (H134D) and Tyr135→ Phe (Y135F)], referredto herein as orthoIL-2Rb, lacked detectable bind-ing to IL-2 (Fig. 1D), even in the presence of CD25(fig. S1) (7, 9).Next, we used yeast display-based evolution to

mutate, and thus remodel, the wild-type IL-2interface region that was opposing (or facingthe site of) the IL-2Rb mutations in the crystalstructure, in order to create amolecule that boundto orthoIL-2Rb but not to wild-type IL-2Rb. IL-2residues in proximity to the orthoIL-2Rb bindinginterfacewere randomlymutated andwere chosenon the basis of a homology model of the mouseIL-2/IL-2Rb complex (Fig. 1E) derived from thecrystal structure of the human IL-2 receptor com-plex (6). A library of ~108 unique IL-2 mutantswas displayed on the surface of yeast (fig. S2) andsubjected to multiple rounds of both positive(against orthoIL-2Rb) and negative (against IL-2Rb) selection (figs. S2 and S3). This collection ofyeast-displayed IL-2 mutants bound the orthoIL-2Rb, but not wild-type IL-2Rb, and retained CD25binding (Fig. 1D). Sequencing of yeast clones fromthe evolved IL-2 libraries revealed a consensusset of mutations at IL-2 positions in close struc-tural proximity to the orthoIL-2Rb mutations(fig. S4). Interestingly, a Gln30→ Asn (Q30N)mu-tation was highly conserved across three inde-pendent mutant IL-2 yeast libraries, whereasall other IL-2 positions used a restricted but notspecific mutational signature. We found thatIL-2 mutations Q30N, Met33 → Val (M33V), andAsp34→ Leu orMet (D34L/M) appear to form asmall nonpolar pocket to compensate for theIL-2Rb Y135F mutation, whereas Gln36 → Thr,Ser, Lys, or Glu (Q36T/S/K/E) andGlu37→ Tyr orHis (E37Y/H)mutations present a polar or chargedsurface to compensate for the IL-2Rb H134D mu-tation (Fig. 1F).Because of the affinity-enhancing effects of

CD25 expression on the interaction of IL-2 withthe binary (bg) IL-2 receptor (10), IL-2 mutantswith negligible binding to IL-2Rb alone may stillform a functional signaling complex on cells thatalso express CD25 (8). Therefore, we used a yeast-based functional screen to further triage IL-2mu-tants that bound specifically to the orthoIL-2Rband signaled selectively on T cells that express theorthoIL-2Rb (Fig. 1G and fig. S5), and producedrecombinant formsof select IL-2mutants (orthoIL-2)for characterization (figs. S6 to S8).We focused our efforts on two orthoIL-2 mu-

tants, 1G12 and 3A10. OrthoIL-2 1G12 and 3A10share the consensus Q30N, M33V, and D34Lmutations but differ at positions Glu29, Gln36,Glu37, and Arg41 (Fig. 1I). OrthoIL-2 1G12 and3A10 bound the orthoIL-2Rb with an affinitycomparable to that of the wild-type IL-2/IL-2Rbinteraction and displayed little to no detectablebinding to wild-type IL-2Rb (Fig. 1H and figs. S7and S8) but differed in their ability to activateIL-2Rb signaling in CD25-positive wild-type

RESEARCH

Sockolosky et al., Science 359, 1037–1042 (2018) 2 March 2018 1 of 6

1Departments of Molecular and Cellular Physiology andStructural Biology, Stanford University School of Medicine,Stanford, CA 94305, USA. 2Stanford Cancer Institute,Stanford University School of Medicine, Stanford, CA 94305,USA. 3Diabetes Center and Department of Medicine,University of California, San Francisco, CA 94143, USA.4Division of Hematology-Oncology, Department of Medicine,David Geffen School of Medicine, and Jonsson ComprehensiveCancer Center, University of California, Los Angeles, CA90095, USA. 5Department of Blood and Marrow Transplantation,Institute for Stem Cell Biology and Regenerative Medicine,and Ludwig Center for Cancer Stem Cell Research and Medicine,Stanford University School of Medicine, Stanford, CA 94305,USA. 6Stanford Medical Scientist Training Program, StanfordUniversity, Stanford, CA 94305, USA. 7Department ofBiochemistry, Howard Hughes Medical Institute, and Institute forProtein Design, University of Washington, Seattle, WA 98195,USA. 8Department of Pediatrics and Genetics, StanfordUniversity School of Medicine, Stanford, CA 94305, USA.9Immunology Graduate Program, Stanford University Schoolof Medicine, Stanford, CA 94305, USA. 10Parker Institute forCancer Immunotherapy, 1 Letterman Drive, Suite D3500, SanFrancisco, CA 94129, USA. 11Howard Hughes Medical Institute,Stanford University School of Medicine, Stanford, CA 94305,USA.*These authors contributed equally to this work.†Corresponding author. Email: [email protected]

on March 21, 2020

http://science.sciencem

ag.org/D

ownloaded from

and orthoIL-2Rb T cells. Stimulation of orthoIL-2Rb T cells (fig. S5B) with orthoIL-2 1G12 resultedin dose-dependent phosphorylation of STAT5(pSTAT5), a hallmark of IL-2R signaling, withpotency similar to that of wild-type IL-2, but alsoinduced pSTAT5 on wild-type T cells, albeit withsignificantly reduced potency relative to IL-2(Fig. 1, G and I, and fig. S6). By comparison,orthoIL-2 3A10 was specific for orthoIL-2RbT cells, but with a weaker potency relative to IL-2(Fig. 1, G and I, and fig. S6). We speculated thatorthoIL-2 1G12 activity on wild-type T cells is aconsequence of weak residual binding to wild-type IL-2Rb (fig. S7). Low-affinity interactionswith IL-2Rb alone are enhanced in the presenceof CD25 (8). Indeed, orthoIL-2 1G12 exhibitedbinding to wild-type IL-2Rb when first capturedby CD25, with limited binding in the absence ofCD25 (figs. S1 and S8). OrthoIL-2 3A10 did notbind appreciably to IL-2Rb even in the presenceof CD25, in agreement with its negligible bio-logical activity on CD25-positive T cells. Interac-tion of orthoIL-2 1G12 and 3A10with orthoIL-2Rbwas significantly enhanced in thepresenceofCD25,with apparent binding affinities of the ternaryCD25/orthoIL-2Rb/orthoIL-2 complex that cor-relate with their respective potency on orthoIL-2Rb T cells (fig. S1).In clinical ACT regimens, patient-derivedT cells

for ACT are expanded in IL-2 before re-infusion inorder to obtain sufficient numbers of therapeuticcells with the desired genotype/phenotype (2).Weexplored the in vitro activity of orthoIL-2 on ac-tivated primary mouse CD8+ T cells engineeredto express the orthoIL-2Rb and a yellow fluores-cent protein (YFP) to distinguish modified (YFP+)and unmodified (YFP–) cells (Fig. 2A). The tran-scription factor STAT5 is phosphorylated uponIL-2 engagementwith the IL-2R and translocatesto the nucleus, where it promotes the prolifera-tion and cell cycle progression of T cells (11).Wild-type IL-2 induced the phosphorylation of STAT5(pSTAT5) in bothwild-type and orthoIL-2Rb CD8+

T cells with similar potency and signaling am-plitude, indicating functional signal transduc-tion through the wild-type receptor but notorthoIL-2Rb (Fig. 2B). By comparison, orthoIL-21G12 potently activated STAT5 on orthoIL-2Rb–transduced T cells, with a potency increase by afactor of ~5 relative to wild-type T cells.OrthoIL-23A10 induced somewhat weaker, albeit selectivepSTAT5 on orthoIL-2Rb–expressing but not wild-type T cells (Fig. 2, B, D, andE). These resultswereconsistent with the biased binding of the orthoIL-2s to the orthoIL-2Rb, which translated into theselective or specific expansion of orthoIL-2RbT cells cultured ex vivo in orthoIL-2 1G12 or 3A10,respectively (Fig. 2, C and D). The orthoIL-2Rb–transduced T cells cultured in saturating concen-trations of orthoIL-2 3A10 became enriched tonear homogeneity after 3 to 5 days (Fig. 2F).IL-2 is indispensable for the development and

function of regulatory T cells (Tregs) (12), whichare sensitive to IL-2 as a result of constitutiveexpressionofCD25 and require IL-2Rb–dependentactivation of STAT5 signaling for survival andfunction (13). Both orthoIL-2 1G12 and 3A10 re-

tained specificity for Tregs modified to express theorthoIL-2Rb, with potency similar to that onCD8+

T cells (Fig. 2G and fig. S9, A and B). In additionto cells that naturally respond to IL-2, activationof orthoIL-2Rb signaling pathwayswith orthoIL-2

could, in principle, be achieved in any cell typethat also expresses the IL-2Rg. Activated mouseB cells expressed the IL-2Rg but lacked appre-ciable levels of IL-2Rb (14, 15) and were relativelyinsensitive to IL-2–dependent STAT5 activation

Sockolosky et al., Science 359, 1037–1042 (2018) 2 March 2018 2 of 6

Control

WTT74

YT74

V

H134D

Y135F

R189E

H134D

Y13

5F0

50001000015000

200000

400000

600000

IL-2

bin

din

g (

MF

I)

orthoIL-2R

AA # 29 30 33 34 36 37 41 WT

EC50 (pM)

ORTHO EC50 (pM)

RATIO (WT/ortho

EC50)

WT E Q M D Q E R 3 3 1 1G12 N V L T H K 300 10 30 3A10 D N V L K A 1000 ortho

orthoIL-2R

wt IL-2R Strep-647 orthoIL-2R

WT IL-2 Mutant

IL-2 Library Evolved

IL-2 Library

CD25

IL-2R

IL-2R IL-2R

IL-2R (CD25)

IL-2

orthoIL-2R :1G12 orthoIL-2R :3A10

Q30

Y135

E29

H134

Q36

D34

E37

D29

L34

A37

K36

M33

N30

V33 E29

L34

H37

T36

F135

D134

N30

V33

F135

D134

WT-WT Ortho-WT Ortho-Ortho WT-Ortho

Yeast evolution Site-directed

IL-2R mutagenesis Discard

WT binders

WT-WT

IL-2R orthoIL-2R orthoIL-2R

Ortho-WT Ortho-Ortho

IL-2R :IL-2

IL-2 1G12 3A10

pSTAT5

SS

C-A

ortho T cell

WT T cell

Fig. 1. Engineering and characterization of orthogonal IL-2 and IL-2R pairs. (A) Schematicoverview of orthogonal IL-2/IL-2R pairs, consisting of a mutant IL-2 cytokine and mutant IL-2Rthat interact specifically with each other but do not cross-react with their wild-type counterparts.(B) Strategy used to engineer orthogonal IL-2/IL-2Rb pairs. (C) Wild-type and mutant IL-2Rb tetramerbinding to wild-type IL-2 displayed on yeast by fluorescence-activated cell sorting. MFI, mean fluo-rescence intensity. Data are representative of two independent experiments. (D) Histograms of wild-typeIL-2Rb (blue), orthoIL-2Rb (red), or CD25 (purple) binding to yeast-displayed wild-type IL-2, the naïvemutant IL-2 yeast library, or mutant IL-2 yeast clones after in vitro evolution. In vitro evolution of threeindependent mutant IL-2 yeast libraries (fig. S4) yielded similar results. (E) Homology model of themouse IL-2/IL-2Rb structure and the site I interface of IL-2 (gray) and contacts with IL-2Rb His134 andTyr135 (teal). Dashed lines indicate potential polar contacts. (F) Model of the orthoIL-2/orthoIL-2Rbinteractions. (G) Off-yeast pSTAT5 functional screen of IL-2 mutant activity on wild-type and orthoIL-2RbCTLL-2 T cells. (H) Representative surface plasmon resonance (SPR) sensograms of wild-type andorthoIL-2 binding to wild-type IL-2Rb or orthoIL-2Rb. Data are representative of two independent experi-ments. KD, dissociation constant. (I) Sequences of wild-type (WT) IL-2, orthoIL-2 1G12, and orthoIL-2 3A10and corresponding in vitro bioactivity (pSTAT5 EC50) on wild-type and orthoIL-2Rb CTLL-2 Tcells. Amino acidcodes: A, Ala; D, Asp; E, Glu; F, Phe; H, His; K, Lys; L, Leu; M, Met; N, Asn; Q, Gln; T, Thr; V, Val; Y, Tyr.

RESEARCH | REPORTon M

arch 21, 2020

http://science.sciencemag.org/

Dow

nloaded from

(Fig. 2H and fig. S9, E and F). Transduction ofthe orthoIL-Rb into activated B cells renderedthem responsive to orthoIL-2 (Fig. 2H and fig. S9,E and F), butwith reduced potency and increasedspecificity relative to T cells. Specificity was dueto the lack of appreciable wild-type IL-2Rb onB cells (fig. S9E).In a host with an intact immune system, adop-

tively transferred T cells must compete with hostcells for survival signals such as IL-2 (16). How-ever, unlikewild-type IL-2, there shouldbeminimalcompetition from endogenous cells for orthoIL-2consumption. Thus, we determined the in vivoactivity of orthoIL-2 and orthoIL-2Rb T cells inmice with intact immune systems. Amixture ofwild-type and orthoIL-2Rb CD8+ T cells was adop-tively transferred into wild-type mice, and theimpact of IL-2 and orthoIL-2 administrationon transplanted T cells and the host immunesystem was quantified (Fig. 3A). OrthoIL-2 1G12significantly expanded CD8+ T cells transducedwith the orthoIL-2Rb at doses equivalent to orlower than wild-type IL-2, which acted throughthe endogenous IL-2Rb expressed in both wild-type and orthoIL-2Rb T cells (Fig. 3B and fig. S10).

The selectivity of orthoIL-2 1G12 for orthoIL-2RbT cells was dose-dependent, with increased activ-ity on wild-type cells at increased dose amountsand/or frequency of treatment (Fig. 3, B and C,and figs. S10 to S12). These results were consistentwith the in vitro selectivity of orthoIL-2 1G12,although it remained possible that orthoIL-2 1G12signaling through the orthoIL-2Rb could triggerendogenous IL-2 production by the orthoIL-2RbT cells, leading to indirect signaling through thewild-type IL-2R in cis or trans.At high doses and twice-daily administration,

orthoIL-2 3A10 resulted in the substantial expan-sion of orthoIL-2Rb T cells with high specificityand nowild-type T cell expansion (Fig. 3, B and C,and figs. S11 and S12). This finding suggests thatthe effects of high-dose orthoIL-2 1G12 treatmentwere due not to induction of endogenous IL-2by orthoIL-2Rb CD8+ T cells, but rather to low-level cross-reactivity with the wild-type IL-2Rbby this molecule. The orthoIL-2 variants also pro-moted the in vivo expansion of orthoIL-2Rb CD4+

effector T cell (Teff) (Fig. 3I and fig. S12) andorthoIL-2Rb CD4+ Treg (fig. S9, C and D) cell sub-sets with specificity similar to that in CD8+ T cells.

The two different orthoIL-2 variants exhibitedspecificities in vivo that mirrored their relativespecificities in vitro. Despite its ability to activatewild-type IL-2Rb signaling, albeit with about oneorder ofmagnitude less potency than orthoIL-2Rbsignaling, orthoIL-2 1G12 administrationwas rela-tively specific for orthoIL-2Rb T cells in vivo (Fig. 3,B to H, and figs. S10 to S12). In mice treated twicedaily with orthoIL-2 1G12 only, CD4+ Tregs wereelevated to a substantially lower degree thanobserved in IL-2–treated mice (Fig. 3F). How-ever, the orthoIL-2 3A10 variant, consistent withthe lack of wild-type IL-2Rb signaling, had nodetectable activity on host cell subset numbers(fig. S11) or expression of CD25, PD-1, and TIM-3,which are up-regulated by early or late IL-2Rsignaling (fig. S13).To improve in vivo half-life and enable more

convenient dosing, we fused IL-2 and orthoIL-2to mouse serum albumin (17) (MSA), which hasbeen shown to extend the half-life of mouse IL-2from 5 hours to 50 hours (18). Fusion toMSA hadlittle to no impact on IL-2–or orthoIL-2–dependentT cell proliferation in vitro (fig. S14); however, thein vivo activity was greatly enhanced. Fusion of

Sockolosky et al., Science 359, 1037–1042 (2018) 2 March 2018 3 of 6

Fig. 2. OrthoIL-2 signalsthrough the orthoIL-2Rexpressed in primary mouselymphocyte subsets, resultingin specific expansion ofCD4 and CD8 Tcells in vitro.(A) Flow cytometry data ofmouse T cells transduced withthe orthoIL-2Rb and a YFPreporter (top panels) and asso-ciated cell surface levels ofCD25, IL-2Rb, and IL-2Rg. (B to F)Dose-response curves of (B)STAT5 phosphorylation after20 min of stimulation and (C)proliferation of wild-type (opencircles) and orthoIL-2Rb (solidcircles) CD8+ Tcells cultured for4 days in IL-2 or orthoIL-2; (D)table of respective pSTAT5 andproliferation EC50 from data in(B) and (C). Representative his-tograms of (E) STAT5 phospho-rylation and (F) scatterplots ofCD8+ wild-type (YFP–) andorthoIL-2Rb (YFP+) T cells ex-panded in IL-2. Data are means ±SD (n = 3 biological replicates).Dashed lines represent curves fitto a log (agonist) versus re-sponse (three parameters)model in Prism. (G) Dose-response curves of STAT5 phos-phorylation (left) and proliferation(right) of wild-type and orthoIL-2Rb CD4+ Tregs cultured in IL-2 ororthoIL-2. Data are means ± SD(n = 3 biological replicates).(H) Representative histogramsof primary mouse B cells transduced with the orthoIL-2Rb and stimulated with the indicated cytokines for quantification of intracellular pSTAT5 as in fig. S9.

-2 0 2 40

50

100

Concentration (Log10[nM])

pS

TAT

5 M

FI

(% o

f IL

-2)

-2 0 2 40

50

100

150

Concentration (Log10[nM])

Cel

l Gro

wth

(% o

f IL

-2)

-2 0 2 40

50

100

Concentration (Log10[nM])

Cel

l Gro

wth

(% o

f IL

-2)

-4 -2 0 2 40

50

100

Concentration (Log10[nM])

CD

4 T

reg G

row

th(%

of

IL-2

)

-4 -2 0 2 40

50

100

Concentration (Log10[nM])

CD

4 T

reg p

STA

T5

MF

I(%

of

IL-2

)

-2 0 2 40

50

100

Concentration (Log10[nM])

pS

TAT

5 M

FI

(% o

f IL

-2)

ortho T cell wt T cell Iso

CD25 IL-2R IL-2R

WT ortho

pSTAT5 ortho3A10

IL-2

Ctrl

ortho1G12

YFP

SS

C-A

wild-type ortho T cells

Transduced Activated

YFP

SS

C-A

wild-type ortho T cells

pSTAT5 Proliferation

WT EC50 (pM)

ORTHO EC50 (pM)

RATIO (WT/ ortho)

WT EC50 (pM)

ORTHO EC50 (pM)

RATIO (WT/ ortho)

IL-2 80 85 1 30 130 0.2 1G12 5000 930 5 3500 50 70 3A10 18000 ortho 11000 ortho

IL-2

WT ortho

pSTAT5

i i

B cells

YFP

SS

C-A

IL-2 ortho3A10

ortho1G12 ortho3A10

ortho3A10IL-2

EC50 400 pM

EC50 32 nM

EC50 70 pM

EC50 27 nM

orthoIL-2R IL-2ortho3A10

wild-type

ortho B cell wt B cell Iso

CD25 IL-2R IL-2R

YFP

CD

19

ortho3A10

IL-2

Ctrl

ortho1G12

RESEARCH | REPORTon M

arch 21, 2020

http://science.sciencemag.org/

Dow

nloaded from

orthoIL-2 1G12 to MSA substantially increasedits activity on cells that express the wild-typeIL-2R relative to native orthoIL-2 1G12, leadingto increased off-target effects and toxicity (fig.S15). However, the MSA–orthoIL-2 3A10 fusionprotein retained exquisite specificity for orthoIL-2Rb T cells (fig. S16).One of the major limitations of IL-2 in the

clinic is that IL-2 toxicity limits the use of high-dose IL-2 therapy formetastatic cancer and as anadjuvant to adoptive T cell therapy (12). IL-2

administered as aMSA fusion resulted in a num-ber of dose-dependent and dose-accumulatingtoxicities that led toweight loss, restrictedmobil-ity, hypothermia, ruffled fur, hunched posture,splenomegaly, lymphomegaly, and death (Fig. 3,J to L, and figs. S15 to S18). In contrast, MSA–orthoIL-2 3A10was nontoxic at all doses evaluated.MSA–orthoIL-2 3A10 activity was negligible on allIL-2–responsive host cell subsets evaluated.In addition to its role as a proliferative cyto-

kine, IL-2 is a potent effector cytokine capable of

activating cytotoxic T cell functions and T cell in-flammatory pathways (19). We determined thecapacity of adoptively transferred orthoIL-2RbCD8+ T cells to produce interferon-g (IFN-g)and cell surface levels of the immune inhibitoryreceptors PD-1 and TIM-3 after expansion in vivowith orthoIL-2. TIM-3 expression correlates witha highly dysfunctional CD8+ T cell state, whereasPD-1 expression is associated with both T cell ac-tivation and exhaustion (20).OrthoIL-2Rb T cellsexpanded in orthoIL-2 produced significantlymore

Sockolosky et al., Science 359, 1037–1042 (2018) 2 March 2018 4 of 6

Fig. 3. OrthoIL-2 promotesthe specific expansion oforthoIL-2Rb–modifiedT cells in mice with negligi-ble toxicity. (A) Schematicof the adoptive CD8+ T celltransplant mouse model.(B) Quantification of donorwild-type and ortho CD8+

T cells in the spleen of recip-ient mice treated twicedaily with phosphate-buffered saline (PBS), IL-2(250,000 IU/dose), orthoIL-21G12 (250,000 IU/dose),or orthoIL-2 3A10(2,500,000 IU/dose).(C) Representative flowcytometry data quantified in(B) depicting donor (Thy1.1+)wild-type (YFP–) andorthoIL-2Rb (YFP+) CD8+

T cells in the spleen of recip-ient mice. (D) Spleen weightof mice treated in (B) nor-malized to total bodyweight on day of killing.(E to G) Quantification ofexogenous cytokine admin-istration on host (E) CD8+

memory phenotype T cell(MP, CD44+CD62L+),(F) CD4+ Treg (CD25+Foxp3+),and (G) natural killer (NK)cell (CD3-NK1.1+CD49b+)numbers in the spleen ofmice treated in (A).(H) Representative flowcytometry data as quantifiedin (F) and (G). Data in(B) to (H) are means± SD (n = 5 mice per group).*P < 0.05, ****P < 0.0001[analysis of variance(ANOVA)]; ns, not significant. (I) Quantification of donor wild-type andorthoIL-2Rb CD4+ Teff in the spleen of recipient mice treated once dailywith PBS, IL-2 (250,000 IU/dose), or orthoIL-2 1G12 (1,000,000 IU/dose).Data are means ± SD and are representative of two independentexperiments (n = 4 mice per group). *P < 0.05, ***P < 0.001 (ANOVA).(J) Survival of mice that received a mixture of wild-type and orthoIL-2RbCD8+ T cells followed by daily administration of IL-2 or orthoIL-2 fusedto MSA. All mice received a total of 250,000 IU/day of the respectiveMSA fusion protein on an IL-2 basis for 5 days. (K) Mouse body weightover time normalized to the group average on day 0 as treated in (J).

(L) Platelet counts in peripheral blood on day 4 as treated in (J). Data in(J) to (L) are means ± SD (n = 5 mice per group). ****P < 0.0001(ANOVA). (M to O) Quantification of cytokine administration on host (M)CD8+ and (N) CD4+ T cell production of IFN-g upon ex vivo restim-ulation with phorbol 12-myristate 13-acetate (PMA) and ionomycin.(O) Representative flow cytometry data as quantified in (M) and (N).(P and Q) Serum (P) IFN-g and (Q) IL-5 concentrations on day 7 in micetreated daily with PBS or with MSA–IL-2, MSA-1G12, or MSA-3A10 (each25,000 IU/dose) for 7 days. Data are means ± SD (n = 5 mice per group).****P < 0.0001 (ANOVA).

PBS

MSA-IL-2

MSA-1G12

MSA-3A10

05

10152025

IFN

(pg

/mL

)

****

ns ns

PBS

MSA-IL-2

MSA-1G12

MSA-3A10

0

5

10100020003000

IL-5

(pg

/mL

)

****

nsns

0 1 2 3 4 5

90

100

110

Days post ACTW

eig

ht (

%)

****

0 1 2 3 4 50

50

100

Days post ACT

Su

rviv

al (%

)

PBSMSA-IL-2MSA-1G12MSA-3A10

PBSIL

-2

MSA-1G12

MSA-3A10

0.0

0.5

1.0

1.5

Pla

tele

t co

un

t(F

old

ove

r P

BS

)

****

nsns

ortho

wild-ty

pe0

5

10

15

Do

no

r C

D8

T c

ell #

(Fo

ld o

ver

PB

S) PBS

IL-2ortho1G12ortho3A10

****

ns *

****

nsYFP

PBSIL

-2

ortho1G

12

ortho3A

100

5

10

15

20

% IF

N+

CD

8 T

cel

ls

****

ns

Transplant

mixture

Transduce

CD8 enrich

PBSIL

-2

ortho1G

12

ortho3A

100

5

10

Sp

leen

wei

gh

t(m

g/g

)

ns

****

PBSIL

-2

ortho1G

12

ortho3A

100

5

10

15

Ho

st C

D4+

Tre

g #

(ove

r P

BS

) ****

ns

ns

YFP

Th

y1.1

Donor BL6 (Thy1.1)

Recipient BL6 (Thy1.2)

Cytokine:

Day: 0 4 5

T cells:

Spleen Blood

PBS IL-2

1G12 3A10

PBS IL-2 3A10

CD49b

NK

1.1

CD25

Fo

xp3

PBSIL

-2

ortho1G

12

ortho3A

100

1

2

3

4

Ho

st N

K c

ell #

(Fo

ld o

ver

PB

S)

****

ns

PBSIL

-2

ortho1G

12

ortho3A

100

2

4

6

Ho

st C

D8

MP

(Fo

ld o

ver

PB

S)

nsns

****

PBS IL-2 3A10 1G12

IL-2

IFN

Host CD4

Host CD8

CD

8

NK cells CD4 Treg Spleen weight CD8 MP

PBSIL

-2

ortho1G

12

ortho3A

100.0

0.5

1.0

1.5

2.0

% IF

N+

CD

4 T

cel

ls

****

ns

CD

8

ortho

wild-ty

pe0

5

10

15

Do

no

r C

D4

Tef

f #

(Fo

ld o

ver

PB

S)

***

* **

PBS IL-2 ortho1G12

RESEARCH | REPORTon M

arch 21, 2020

http://science.sciencemag.org/

Dow

nloaded from

IFN-g than IL-2–expanded cells (Fig. 4A). PD-1 levelswere similar on orthoIL-2Rb T cells from both IL-2–and orthoIL-2–treatedmice (Fig. 4B). Interestingly,TIM-3 levels were significantly lower on orthoIL-2Rb T cells from mice treated with orthoIL-2 rela-tive to those treated with IL-2 (Fig. 4B).The differential activity of orthoIL-2 on both

T cell expansion and function may be due to in-creased bioavailability of orthoIL-2 for orthoIL-2RbT cells as the result of a reduced antigen sink oralternative host factors influenced by IL-2 but notorthoIL-2, which in turn may influence the func-tion of transplanted T cells. For instance, IL-2 butnot orthoIL-2 treatment increased host CD4+ andCD8+ T cell IFN-g production upon ex vivo restim-ulation (Fig. 3, M to O) and increased the serumconcentration of numerous inflammatory cyto-kines, including IFN-g, IL-4, IL-5, IL-6, and IL-13(Fig. 3, P and Q, and fig. S17). The ability to decou-ple direct IL-2 activity on transplanted T cells fromindirect host bystander effects using orthoIL-2/IL-2R pairs may have important therapeuticimplications.To investigate prospective clinical applications

of orthogonal IL-2/IL-2R pairs, we determinedthe efficacy of tumor-specific orthoIL-2Rb T cellsin the B16-F10mousemodel ofmelanoma. Trans-genic pmel-1 T cell receptor (TCR) cells (pmel-1T cells) express a high-affinity TCR that recognizes

theB16-F10 specific ortholog of humangp100 (19),a self antigen overexpressed in humanmelanoma(Fig. 4, C andD). Adoptive transfer of pmel-1 T cellsin combination with lymphocyte depletion andIL-2 administration can model ACT approachesto treat human cancer. Adoptive transfer of pmel-1T cells accompanied by five daily injections of IL-2significantly delayed tumor growth in mice andincreased survival relative to mice treated onlywith T cells and saline (Fig. 4, E to G). Transferof orthoIL-2Rb pmel-1 T cells followed by treat-ment with native orthoIL-2 1G12 at a dose thathad minimal activity on wild-type IL-2R cells(fig. S10) produced a significant tumor growthdelay and survival advantage that mirrored theIL-2 treatment group (Fig. 4, E and F). Similarantitumor responseswere observed inmice treatedwith orthoIL-2Rb pmel-1 T cells andMSA–orthoIL-23A10 (Fig. 4, G and H). There was no therapeuticbenefit of orthoIL-2 in mice that received wild-type pmel-1 T cells, indicating that orthoIL-2activity is dependent on expression of the orthoIL-2Rb in pmel-1 T cells.Our results constitute an approach to redirect

the specificity of IL-2 toward engineered T cellsusing orthogonal IL-2 cytokine-receptor pairs,which enables the selective expansion of de-sired T cell subsets in settings of adoptive celltherapy, but with limited off-target activity and

negligible toxicity. Engineering orthogonal mo-lecular recognition at a protein–small moleculeor protein-protein interface has resulted insynthetic enzymes, kinases, transcription fac-tors, and receptors with controllable biologicalfunctions, but here we apply this concept toprotein interactions with cell surface receptorsto control signaling specificity and downstreamcellular functions (21–28). Orthogonal IL-2/IL-2R pairs may be useful not only as a researchtool but in the clinic to specifically enrich trans-duced T cells that express a target gene of inter-est, such as a CAR or engineered TCR, whencoupled with expression of the orthoIL-2Rb. Ourapproach, and variations of this orthogonaliza-tion strategy,may be applicable to other cytokines,growth factors, hormones, and ligand-receptorinteractions to decipher and manipulate other-wise complex biological systems.

REFERENCES AND NOTES

1. M. Kalos, C. H. June, Immunity 39, 49–60 (2013).2. S. A. Rosenberg, N. P. Restifo, Science 348, 62–68 (2015).3. C. Yee et al., Proc. Natl. Acad. Sci. U.S.A. 99, 16168–16173

(2002).4. R. Andersen et al., Clin. Cancer Res. 22, 3734–3745 (2016).5. S. A. Rosenberg, J. Immunol. 192, 5451–5458 (2014).6. X. Wang, M. Rickert, K. C. Garcia, Science 310, 1159–1163

(2005).7. A. M. Levin et al., Nature 484, 529–533 (2012).

Sockolosky et al., Science 359, 1037–1042 (2018) 2 March 2018 5 of 6

Fig. 4. OrthoIL-2–expandedTcells retain effector functionand promote an antitumorresponse against syngeneicB16-F10 tumors in mice.(A) Quantification of total numberof IFN-g–positive wild-type ororthoIL-2Rb CD8+ T cells recov-ered from the spleen as treated inFig. 3 (left) and representativeflow cytometry data (right).(B) Cell surface expression levelsof PD-1 (left) and TIM-3 (right)on wild-type and orthoIL-2RbCD8+ T cells in the spleen afteradministration of the indicatedcytokines. Data are means ± SD(n = 5 mice per group). *P < 0.05,****P < 0.0001 (ANOVA).(C) gp100 pMHC tetramer stain-ing of orthoIL-2Rb–transducedpmel-1 transgenic CD8+ T cells.(D) In vitro cytotoxicity of orthoIL-2Rb pmel-1 transgenic T cellsagainst antigen-positive (B16-F10)but not antigen-negative (MC38)tumor cells at a 20:1 (E:T) ratio.Data are means ± SD (n = 3biological replicates). **P < 0.01(Student t test). (E and F) Tumorgrowth (E) and survival (F) of C57BL/6J mice bearing subcutaneousB16-F10 tumors treated with wild-type (wt T) or orthoIL-2Rb pmel-1 transgenic CD8+ T cells (ortho T) and IL-2 or orthoIL-2 1G12. Data aremeans ± SEM (n = 5 mice per group). ****P < 0.0001 (two-way ANOVA)(E); **P < 0.01 (log-rank test) (F). (G and H) Tumor growth (G) and survival

(H) of C57BL/6J mice bearing subcutaneous B16-F10 tumors treatedwith wild-type (wt T) or orthoIL-2Rb pmel-1 transgenic CD8+ T cells(ortho T) and IL-2 or orthoIL-2 3A10 fused to MSA. Data are means± SEM (n = 4 mice per group). ****P < 0.0001 (two-way ANOVA) (G);**P < 0.01 (log-rank test) (H).

0 5 10 15 200

500

1000

1500

2000

Days

Tum

or

volu

me

(mm

3 )****

****

0 5 10 15 20 250

50

100

Days post ACT

% S

urv

ival

**

0 5 10 15 20 250

50

100

Days post ACT

% S

urv

ival

**

ortho

wild-ty

pe0.0

5.0 105

1.0 106

1.5 106

2.0 106

Tota

l # o

f IF

N+

cells

ns

*ns

PBSIL-2

ortho1G12ortho3A10

********

ortho

wild-ty

pe0

2

4

6

8

TIM

-3 M

FI

(Fo

ld o

ver

PB

S)

ns

****

**

ns

****

ns

ortho

wild-ty

pe0.5

1.0

1.5

2.0

PD

-1 M

FI

(Fo

ld o

ver

PB

S)

***

******** ****

PBS IL-2 ortho1G12 ortho3A10

IFN

PBS IL-2

1G12 3A10

IL-2

B16-F

10

MC38

0

50

100

% A

live

No T cellsortho pmel

**

gp

100

pM

HC

YFP

wt T + PBS

ortho T + MSA-3A10

wt T + IL-2 wt T + MSA-3A10

ortho T + ortho1G12

wt T + PBSwt T + IL-2 wt T + ortho1G12

0 5 10 15 200

500

1000

1500

2000

Days Tu

mo

r vo

lum

e(m

m3 )

********

RESEARCH | REPORTon M

arch 21, 2020

http://science.sciencemag.org/

Dow

nloaded from

8. A. B. Shanafelt et al., Nat. Biotechnol. 18, 1197–1202(2000).

9. M. Rickert, M. J. Boulanger, N. Goriatcheva, K. C. Garcia,J. Mol. Biol. 339, 1115–1128 (2004).

10. E. Roessler et al., Proc. Natl. Acad. Sci. U.S.A. 91, 3344–3347(1994).

11. R. Moriggl et al., Immunity 10, 249–259 (1999).12. T. R. Malek, Annu. Rev. Immunol. 26, 453–479 (2008).13. D. Klatzmann, A. K. Abbas, Nat. Rev. Immunol. 15, 283–294

(2015).14. S. Amu, I. Gjertsson, A. Tarkowski, M. Brisslert, Scand. J.

Immunol. 64, 482–492 (2006).15. O. Boyman, J. Sprent, Nat. Rev. Immunol. 12, 180–190

(2012).16. L. Gattinoni et al., J. Exp. Med. 202, 907–912 (2005).17. J. T. Sockolosky, F. C. Szoka, Adv. Drug Deliv. Rev. 91, 109–124

(2015).18. E. F. F. Zhu et al., Cancer Cell 27, 489–501 (2015).19. W. W. Overwijk et al., J. Exp. Med. 198, 569–580 (2003).20. K. Sakuishi et al., J. Exp. Med. 207, 2187–2194 (2010).21. M. G. J. Baud et al., Science 346, 638–641 (2014).

22. S. Atwell, M. Ultsch, A. M. De Vos, J. A. Wells, Science 278,1125–1128 (1997).

23. D. M. Spencer, T. J. Wandless, S. L. Schreiber, G. R. Crabtree,Science 262, 1019–1024 (1993).

24. D. J. Mandell, T. Kortemme, Nat. Chem. Biol. 5, 797–807 (2009).25. G. T. Kapp et al., Proc. Natl. Acad. Sci. U.S.A. 109, 5277–5282

(2012).26. J. S. Park et al., Proc. Natl. Acad. Sci. U.S.A. 111, 5896–5901

(2014).27. S. M. Lewis et al., Nat. Biotechnol. 32, 191–198 (2014).28. C. Y. Wu, K. T. Roybal, E. M. Puchner, J. Onuffer, W. A. Lim,

Science 350, aab4077 (2015).

ACKNOWLEDGMENTS

We thank M. McCracken for expertise in T cell immunotherapy;N. Saligrama, A. Cravens, M. Hollander, F. Zhao, Y. Rosenberg-Hasson, and the Stanford Human Immune Monitoring Core (HIMC)for technical assistance; and R. Fernandes for helpful discussion.The data presented in this paper are tabulated in the main text andsupplementary materials. Supported by NIH grants R37 AI051321and HHMI (K.C.G.); a 2016 Stanford Cancer Institute translational

research grant (J.T.S., A.R., and K.C.G.); NIH grant R35 CA197633,the Ressler Family Fund, and the Parker Institute for CancerImmunotherapy (G.P. and A.R.); the Sean N. Parker AutoimmunityResearch Laboratory (J.A.B.); and fellowship support from StanfordMolecular and Cellular Immunobiology NIH training grant 5T32AI072905 and a PhRMA Foundation Translational Medicine andTherapeutics postdoctoral award (J.T.S.). J.A.B. and A.R. aremembers of the Parker Institute for Cancer Immunotherapy. K.C.G.,J.T.S., I.C.K., and D.B. are inventors on patent applications 62/217,364and 62/375,089 submitted by Stanford University that cover theuse of orthogonal cytokine-receptor pairs for use in cellularimmunotherapy.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/359/6379/1037/suppl/DC1Materials and MethodsFigs. S1 to S18References (29–35)

30 October 2017; accepted 11 January 201810.1126/science.aar3246

Sockolosky et al., Science 359, 1037–1042 (2018) 2 March 2018 6 of 6

RESEARCH | REPORTon M

arch 21, 2020

http://science.sciencemag.org/

Dow

nloaded from

Selective targeting of engineered T cells using orthogonal IL-2 cytokine-receptor complexes

Antoni Ribas, Jeffrey A. Bluestone and K. Christopher GarciaSilveria, Benson M. George, Indigo C. King, Matthew R. Tiffany, Kevin Jude, Leah V. Sibener, David Baker, Judith A. Shizuru, Jonathan T. Sockolosky, Eleonora Trotta, Giulia Parisi, Lora Picton, Leon L. Su, Alan C. Le, Akanksha Chhabra, Stephanie L.

DOI: 10.1126/science.aar3246 (6379), 1037-1042.359Science 

, this issue p. 1037; see also p. 990Sciencetoxicities associated with some cytokine-based immunotherapies.engineered T cells to reject tumors with no obvious side effects. This type of approach may provide a way to mitigate only interacted with each other and not with endogenous IL-2/IL-2R. Treatment of mice with IL-2 improved the ability of(i.e., IL-2 and its receptor, IL-2R) (see the Perspective by Mackall). Engineered complexes transmitted IL-2 signals but

receptor pairs−based immunotherapy by engineering synthetic IL-2− set out to improve IL-2et al.responses. Sockolosky great therapeutic promise but is limited by toxic side effects and its capacity to both activate and repress immune

Interleukin-2 (IL-2) is an important cytokine that helps T cells destroy tumors and virus-infected cells. IL-2 hasEngineering cytokine-receptor pairs

ARTICLE TOOLS http://science.sciencemag.org/content/359/6379/1037

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2018/02/28/359.6379.1037.DC1

CONTENTRELATED

http://stm.sciencemag.org/content/scitransmed/7/280/280ps7.fullhttp://stm.sciencemag.org/content/scitransmed/4/149/149ra120.fullhttp://stm.sciencemag.org/content/scitransmed/8/367/367ra166.fullhttp://stm.sciencemag.org/content/scitransmed/7/311/311ra170.fullhttp://science.sciencemag.org/content/sci/359/6379/990.full

REFERENCES

http://science.sciencemag.org/content/359/6379/1037#BIBLThis article cites 35 articles, 17 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 21, 2020

http://science.sciencem

ag.org/D

ownloaded from