SC I ENCE ADVANCES | R E S EARCH ART I C L E

HEALTH AND MED IC INE

1Department of Microbiology and Immunology, Medical University of South Carolina,Charleston, SC 29425, USA. 2CanCure LLC, Everett,WA98208, USA. 3School ofMedicine,University of Missouri, Columbia, MO 65211, USA. 4Knight Cancer Institute, OregonHealth and Science University, Portland, OR 97239, USA. 5Cancer ImmunologyProgram, Hollings Cancer Center, Charleston, SC 29425, USA.*Corresponding author. Email: wujjd@musc.edu

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

2017 © The Authors,

some rights reserved;

exclusive licensee

American Association

for the Advancement

of Science. Distributed

under a Creative

Commons Attribution

NonCommercial

License 4.0 (CC BY-NC).

Dow

nloaded from

Antibody-mediated neutralization of soluble MICsignificantly enhances CTLA4 blockade therapyJingyu Zhang,1,2 Dai Liu,3 Guangfu Li,3 Kevin F. Staveley-O’Carroll,3 Julie N. Graff,4

Zihai Li,1,5 Jennifer D. Wu1,5*

Antibody therapy targeting cytotoxic T lymphocyte–associated antigen 4 (CTLA4) elicited survival benefits in cancerpatients; however, the overall response rate is limited. In addition, anti-CTLA4 antibody therapy induces a high rate ofimmune-related adverse events. The underlying factors that may influence anti-CTLA4 antibody therapy are not welldefined. We report the impact of a cancer-derived immune modulator, the human-soluble natural killer group 2D(NKG2D) ligand sMIC (solublemajor histocompatibility complex I chain–relatedmolecule), on the therapeutic outcomeof anti-CTLA4antibodyusinganMIC transgenic spontaneousTRAMP (transgenic adenocarcinomaof themouseprostate)/MIC tumormodel. Unexpectedly, animalswith elevated serum sMIC (sMIChi) responded poorly to anti-CTLA4 antibodytherapy, with significantly shortened survival due to increased lung metastasis. These sMIChi animals also developedcolitis in response to anti-CTLA4 antibody therapy. Coadministration of an sMIC-neutralizing monoclonal antibodywith the anti-CTLA4 antibody alleviated treatment-induced colitis in sMIChi animals and generated a cooperativeantitumor therapeutic effect by synergistically augmenting innate and adoptive antitumor immune responses. Ourfindings imply that a new combination therapy could improve the clinical response to anti-CTLA4 antibody therapy.Our findings also suggest that prescreening cancer patients for serum sMIC may help in selecting candidates whowill elicit a better response to anti-CTLA4 antibody therapy.

http

on M

ay 17, 2017://advances.sciencem

ag.org/

INTRODUCTIONCytotoxic T lymphocyte–associated antigen 4 (CTLA4) is a co-inhibitoryreceptor that controls cytotoxic effector T cell activation during initiationand maintenance of healthy adaptive immune responses (1–3). The ex-pression of CTLA4 is often up-regulated during T cell activation uponinteracting with antigen-presenting cells (APCs) (1–3). CTLA4 competeswith the costimulatory T cell receptor (TCR) CD28 in binding to the B7molecules (B7.1 and B7.2) expressed on APCs and has a higher bindingaffinity thanCD28 (1–4). In contrast toCD28/B7 binding,which acts as acostimulatory signal to promote T cell activation and proliferation, thebinding of CTLA4 to B7 transmits an inhibitory signal. CTLA4 is con-sidered the “off switch” or “immune checkpoint” for effector T cellfunction. CTLA4 can also control effector T cell function through the ex-pansion of CD4+FoxP3+ regulatory T cells, in which CTLA4 is consti-tutively expressed and critical for maintaining the suppressive function(5, 6). Therapy with anti-CTLA4 antibody elicited remarkable effective-ness in controlling tumor growth in mice (7–12).

Preclinical investigations have led to the development of anti-CTLA4 antibodies for cancer immunotherapy. Ipilimumab, a full hu-man monoclonal anti-CTLA4 antibody that blocks the binding ofCTLA4 to the B7 molecule, was approved by the U.S. Food and DrugAdministration (FDA) for the treatment of advancedmelanoma (4, 13–15).Currently, ipilimumab is also undergoing clinical trials for the treatmentof non–small cell lung carcinoma, small cell lung cancer, and bladdercancer (16–18). Despite the remarkable tumor control ability of anti-CTLA4 antibody demonstrated in preclinical animal models, the overallclinical response rate of anti-CTLA4 antibody as a stand-alone agent islimited to 15% inmelanoma (4). Notably, therapywith anti-CTLA4 anti-body has been associated with a significant risk of immune-related

adverse events (IRAEs) inpatients,withmost of the severe cases restricted togastrointestinal inflammation (13, 19–21). Combination therapy of anti–programmed cell death protein 1 antibody and anti-CTLA4 antibody hasachieved greater overall response thanmonotherapy inpatientswithmel-anoma, the most immunogenic cancer; however, the response remainslimited to a subgroupof patients and is accompaniedby increased toxicity(22–24). It is thus imperative to understand the circumstances that maylimit the clinical efficacy of anti-CTLA4 antibody therapy and the risk oftoxicity to develop more effective yet safe regimens.

The differential expression of immune modulators by human androdent tumorsmay contribute to the response disparity between clinicaland preclinical anti-CTLA4 antibody therapy. The major histo-compatibility complex (MHC) I chain–related family molecules, in-cluding the molecules MICA and MICB (collectively termed MIC),are an example of such immunemodulators.MIC is frequently inducedin human tumors upon genotoxic insults but is absent in rodents (25–31).The induction ofMICon the cancer cell surface can activate natural killer(NK) cells and costimulate cytotoxic CD8 T cells, NK-like T cells, andsubsets of gd T cells through engaging the activating immune receptornatural killer group 2D (NKG2D) (32, 33). The activation of NKG2Dhas been shown to provide unique costimulation to antigen-specificCD8 T cells that is nonredundant to CD28 costimulation (32, 34–37).In cancer patients, MIC is often released as the highly immunosup-pressive soluble form sMIC through shedding (25, 28, 29). Tumor-derived sMIC at large suppresses host immune response using multiplestrategies including (but not limited to) down-regulatingNKG2Dexpres-sion and destabilizing CD3z in the TCR/CD3 complex on CD8 T cells(25, 38), perturbingNK cell function and peripheralmaintenance (39, 40),and cultivating an immunosuppressive tumor microenvironment bydriving the expansion of myeloid suppressor cells and skewing macro-phage into an alternatively activated phenotype (41). Using a clinicallyrelevant MIC-transgenic spontaneous mouse tumor model that closelyrecapitulates the onco-immune dynamic of human cancer (39), we havedemonstrated that therapy with the nonblocking anti-sMIC monoclonalantibody (mAb) B10G5 globally abrogates the immunosuppressive effect

1 of 12

SC I ENCE ADVANCES | R E S EARCH ART I C L E

of sMIC, immobilizes NK and CD8 T cells to the tumor microenvi-ronment, and overcomes immune tolerance of antigen-specific CD8T cells (42, 43). In linewith these understandings, we herein investigatedthe impact of tumor-derived human sMICon the therapeutic efficacy ofanti-CTLA4 antibody.We show that high levels of serum sMIC not on-ly negatively influenced the antitumor efficacy of anti-CTLA4 antibodybut also evoked colitis development during anti-CTLA4 antibody ther-apy. Coadministration of the sMIC-neutralizing antibody B10G5 withanti-CTLA4 antibody generated a cooperative antitumor effect and al-leviated therapy-induced colitis. These results suggest a new avenue forcombination immunotherapy of cancer.

on May 17, 2017

http://advances.sciencemag.org/

Dow

nloaded from

RESULTSAnimals with high serum levels of sMIC respond poorly toanti-CTLA4 antibody therapyTumor-derived sMIC negatively affects tumor immunity throughmultiple mechanisms (25, 39–41). To address the impact of sMIC onthe efficacy of anti-CTLA4 antibody therapy, we treated cohorts of27- to 28-week-old TRAMP (transgenic adenocarcinoma of the mouseprostate) and TRAMP/MIC littermates with anti-CTLA4 antibody(3.0 mg/kg) or control immunoglobulin G (cIgG), respectively (Fig. 1A).We previously showed that more than 40% of TRAMP/MICB miceat the indicated age had elevated serum sMIC as a result of tumorshedding (39). Consistent with other studies (44, 45), TRAMP micepresented no or marginal response to anti-CTLA4 antibodymonother-apy (Fig. 1, B and D). TRAMP/MIC mice that received anti-CTLA4antibody therapy presented reduced overall survival compared to lit-termates that received cIgG therapy (Fig. 1B). Animals that succumbedto diseases during therapy, whether receiving cIgG or anti-CTLA4 an-tibody, had significantly higher levels of serum sMIC before therapythan those that survived to the study end point (Fig. 1C). However,in animals that had high levels of sMIC before therapy, tumors pro-gressedmore aggressively in response to anti-CTLA4 antibody therapy.The disease progression was manifested through large primary tumorburdens and massive lung metastasis (Fig. 1, D to F). Animals that hadhigh serum sMIC and had received anti-CTLA4 antibody therapy alsodeveloped colitis (fig. S1). These data indicate that high levels of circu-lating sMIC adversely affect tumor response to anti-CTLA4 antibodytherapy and also contribute, at least in part, to anti-CTLA4 antibody–induced autoimmune colitis.

We substantiated our observations in syngeneic TRAMP-C2 andsMICB-expressing TRAMP-C2-sMICB transplantable prostate tumormodels (fig. S2), in which both tumor cell lines were subcutaneouslyinoculated into MICB/B6 male transgenic mice, as we have previouslydescribed (39, 42, 46). In the absence of sMIC, TRAMP-C2 tumorspresented no response to anti-CTLA4 antibody therapy. Consistentwith our observations in transgenic TRAMP/MICB mice, TRAMP-C2-sMICB tumors grew more aggressively in response to anti-CTLA4antibody therapy (fig. S2B). In animals that received anti-CTLA4 anti-body therapy, TRAMP-C2-sMICB tumor–bearing mice developedmore severe colon inflammation thanTRAMP-C2 tumor–bearingmice(fig. S2C).

Antibody neutralizing sMIC markedly enhances therapeuticresponses to anti-CTLA4 antibodyTo confirm that sMIC contributes to subjects’ poor response to anti-CTLA4 antibody therapy and to determine whether neutralizing sMICcan enhance the therapeutic efficacy of anti-CTLA4 antibody or MIC+

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

tumors, we coadministered the anti-sMIC mAb B10G5 (3.0 mg/kg)with anti-CTLA4 antibody to a cohort of TRAMP/MICB littermatesat 26 to 27 weeks old, as depicted in Fig. 1A. In our recent publications,we described that B10G5 can effectively neutralize the immuno-suppressive effect of sMIC to induce the debulking of primary tumorsand eliminate metastasis (42). In agreement with our previous findings,all animals responded to mAb B10G5 monotherapy (Fig. 2, A to E).Remarkably, regardless of serum levels of sMIC before therapy, allanimals survived to the study end point with clearance of lung metas-tasis in response to anti-CTLA4 antibody therapy with the coadminis-tration of B10G5 (Fig. 2, A and E). The coadministration of B10G5significantly reduced the circulating level of sMIC (Fig. 2B) and gener-ated a cooperative therapeutic effect with anti-CTLA4 antibody, elicit-ing a significant reduction in primary tumors at the study end pointcompared to monotherapies (Fig. 2C). The coadministration ofB10G5 also alleviated anti-CTLA4 antibody therapy–induced colitis(fig. S3). These data demonstrate the benefit of the sMIC-neutralizingantibody B10G5 in enhancing anti-CTLA4 antibody therapy.

A melanoma patient who developed anti-MIC autoantibody duringanti-CTLA4 antibody therapy was found to elicit a superior response(47). We thus assayed the presence of anti-sMIC autoantibody duringearly response to anti-CTLA4 antibody (ipilimumab) in a phase 1/2clinical trial (NCT01498978) on a small cohort (n = 10) of metastaticprostate cancer patients who were also receiving androgen suppressiontherapy. A high titer of anti-MIC autoantibody was detected in one pa-tient (ID: OHSU 5254-8) after one cycle of ipilimumab (fig. S4). PatientOHSU 5254-8 elicited a durable response with a prostate-specific anti-gen decrease from 191 to 4.6 ng/ml after eight cycles of ipilimumab. Noautoimmune colitis has been noted in patient OHSU 5254-8 thus far.This case study reinforces the utility of sMIC-neutralizing antibody inenhancing anti-CTLA4 antibody therapy.

B10G5 neutralizing sMIC heightens CD8 T cell response toanti-CTLA4 antibody therapyThe therapeutic efficacy of anti-CTLA4 antibody is commonly deter-mined by the activation status of effector CD8 T cells. The coadminis-tration of B10G5 with anti-CTLA4 antibody to TRAMP/MICB miceresulted in a significant increase of CD8 T cells in tumor-draininglymph nodes (dLNs) and tumor infiltrates compared to the monother-apy with anti-CTLA4 antibody or B10G5 (Fig. 3A and fig. S5A).Consistent with our previous observation, B10G5 therapy alone in-creased the expression of NKG2DonCD8T cells (42). Notably, NKG2Das a nonconventional effector T cell costimulatory molecule is expressedby all humanCD8T cells but is only expressed by activatedmouse CD8T cells (48). A cocktail therapy of anti-CTLA4 antibody and B10G5 fur-ther increased the number of NKG2D+ CD8 T cells, signifying that moreCD8 T cells were activated (Fig. 3B and fig. S5B). NKG2D andCD28 candeliver nonredundant costimulatory signals to CD8T cells (35, 49). Thus,the further increased NKG2D expression in response to cocktail therapywould presumably provide a nonconventional costimulatory signal toaugment antigen-specific CD8 T cell responses to MIC+ tumors, in ad-dition to the diminution of CTLA4 co-inhibitory signals.

The coadministration of B10G5 with anti-CTLA4 antibody therapysignificantly increased the effector memory-like CD44hi CD8 T cellcompartments in the spleen, dLNs, and tumor infiltrates (Fig. 3C andfig. S5C). The combination therapy also increased the intrinsic ability ofCD8 T cells to produce interferon-g (IFN-g) in response to ex vivophorbol 12-myristate 13-acetate (PMA) and ionomycin restimulation(Fig. 3D and fig. S5D).

2 of 12

SC I ENCE ADVANCES | R E S EARCH ART I C L E

We retrospectively compared the functional potential of CD8T cellsin TRAMPandTRAMP/MICBmice that had received anti-CTLA4 an-tibodymonotherapy. CD8 T cells in all TRAMPmice uniformly exhib-ited a nominal response to anti-CTLA4 therapy (fig. S6). In TRAMP/MIC

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

mice that had received anti-CTLA4 antibody therapy, the CD8 T cellresponse to ex vivoPMA/ionomycin restimulation varied in accordanceto serum levels of sMIC; however, the percentage of CD8T cells was notprofoundly affected (fig. S7A). A severely impaired response was

on May 17, 2017

http://advances.sciencemag.org/

Dow

nloaded from

Therapy

Puberty

Age 6

TRAMP orTRAMP/MICB mice

27 35 (weeks)16 24

Tumor initiation Metastasis

Studyend point

A

D

FE

cIgG(sMIChi)

αCTLA4(sMIChi)

C

0.0

0.5

1.0

468

TRAMP

TRAMP/MICB

Ser

ums

MIC

bef

ore

ther

apy

(n

g/m

l)

P < 0.001Nonsurvivors(sMIChi)

P < 0.001

No.

of m

icro

met

asta

sis

(in lu

ng)

1015406080

100120

05

Nonsurvivors(sMIChi)

TRAMP/MICB TRAMP

ns

B

Su

rviv

al %

TRAMP + αCTLA4 TRAMP + cIgG

TRAMP/MICB + αCTLA4TRAMP/MICB + cIgG

Rx week

0

25

50

75

100

0 2 4 6 8

*

Pro

stat

e w

eigh

t (g )

(at d

eath

)

TRAMP/MICB

0.0

0.5

1.0

1.5

2.0

101418

6

TRAMP

ns2

Survivors

Nonsurvivors

Fig. 1. Mice with high circulating tumor-shed soluble MIC (sMIChi) had poor response to anti-CTLA4 antibody therapy. (A) Schematic depiction of therapy. (B) Kaplan-Meier survival curve.aCTLA4, anti-CTLA4. *P<0.05. (C) Prostate (spontaneous tumor site) weight at necropsy. Boxed animals succumbed todisease before the study endpoint.(D and E) Representative micrographs (D) and overall number (E) of lungmetastasis in TRAMP/MICmice in response to anti-CTLA4 antibody therapy. Arrows, lung nodules.(F) Serum levels of sMIC in TRAMP/MICB mice before receiving therapy. Sera from TRAMP were used as negative controls. ns, not significant.

3 of 12

SC I ENCE ADVANCES | R E S EARCH ART I C L E

observed in mice with high levels of sMIC, whereas the responses inTRAMP/MICB mice with low levels of sMIC and in TRAMP mice re-main comparable. One of the 11 TRAMP/MICB mice that developedanti-sMIC autoantibody during anti-CTLA4 antibody monotherapydemonstrated remarkable responsiveness to ex vivo restimulation (fig.S7B). Notably, the polarization of CD8 T cell functional potential was

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

correlative with the presence and function of NK cells in the dLNs(fig. S6B). Given that sMIC perturbs NK cell homeostatic maintenanceand function and thatNK cell–dendritic cell (DC) cross-talk is critical inregulating T cell priming (39, 50–53), these observations furtherhighlight a potential important role of NK cells in effectuating CTLA4blockade therapy.

on May 17, 2017

http://advances.sciencemag.org/

Dow

nloaded from

CB

0.0

0.5

1.0

468

αCTLA4

Ser

um s

MIC

at n

eucr

opos

y(n

g/m

l)

αCTLA4+ B10G5

A

cIgG B10G5

αCTLA4 αCTLA4 + B10G5

ED

0

5

10

1540

60

80

100

120

No.

of m

icro

met

asta

sis

(in lu

ng)

Su

rviv

al %

0 2 4 6 80

25

50

75

100

Rx week

αCTLA4cIgG

αCTLA4 + B10G5

B10G5

*

**

P < 0.001

Pro

stat

e w

eigh

t (g)

(at n

eucr

opos

y)

P < 0.05

0.0

0.5

1.0

1.5

2.0

6101418

Fig. 2. Antibody B10G5 cotargeting sMIC remarkably enhances response to anti-CTLA4 antibody therapy. (A) Kaplan-Meier survival curve. (B) Serum levels of sMIC atnecropsy. (C) Prostate weight at necropsy. (D and E) Number (D) and representative micrographs (E) of lung micrometastasis in animals of each treatment cohort.

4 of 12

SC I ENCE ADVANCES | R E S EARCH ART I C L E

Combination therapy of antibody neutralizing sMIC andanti-CTLA4 antibody markedly overturns antigen-specificCD8 T cell toleranceWehave previously shown that B10G5monotherapy overcomes antigen-specific CD8 T cell tolerance (42). To determine whether combinationtherapy with B10G5 and anti-CTLA4 antibody can further enhanceCD8T cell antitumor response in an antigen-specificmanner, we adop-tively transferred carboxyfluorescein succinimidyl ester (CFSE)–labeledSV40TAg-specific TCR-I CD8 T cells into cohorts of mice that had re-ceived therapies for 4 weeks, as indicated and schematically depicted inFig. 1A. Notably, these TCR-I CD8 T cells bear TCRs specific for theTRAMP/MIC oncogene SV40TAg and can be detected by the SV40TAgpeptide I–specific Db/I-tetramer.

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

on May 17, 2017

http://advances.sciencemag.org/

Dow

nloaded from

Adoptively transferred antigen-specificTCR-ICD8Tcellswouldnor-mally fail to sustain after initial expansion in TRAMP or TRAMP/MICBmice because of clonal deletion (42, 54). Whereas anti-CTLA4 antibodymonotherapy only had a marginal effect on sustaining Db/I-tetramer+

TCR-I CD8 T cells in dLNs, tumor infiltrates, or spleen in TRAMP/MICmice, B10G5 therapy achieved consistent sustainability of the adoptive-ly transferred TCR-I CD8 T cells with a great percentage in the tumorinfiltrates, as we have previously presented (Fig. 4A and fig. S8A) (42).Remarkably, the combination antibody therapy with B10G5 and anti-CTLA4 further significantly enhanced the sustainability of TCR-I CD8T cells compared to B10G5 monotherapy (Fig. 4A and fig. S8A).

Adoptively transferred naïve Db/I-tetramer+ TCR-1 CD8 T cellsexperienced initial expansion, as shown by the dilution of CFSEhi

Db/I-tetramer+ CD8 T cells in all cohorts of animals (Fig. 4B and fig.S8B). However, therapy with B10G5 or B10G5 combined with anti-CTLA4 antibody provoked the continuous expansion of Db/I-tetramer+

CD8 T cells, shown as a significantly increased percentage of CFSElo

Db/I-tetramer+ CD8 T cells compared to cIgG therapy (Fig. 4B andfig. S8B). The combination therapy resulted in a substantially higherpercentage of CFSElo Db/I-tetramer+ CD8 T cells compared to B10G5monotherapy (Fig. 4B and fig. S8B).

The continuous expansion of antigen-specific CD8 T cells in the tu-mor microenvironment and tumor-dLNs is an indication of the activa-tion of antigen-specific CD8T cells. Anti-CTLA4 antibodymonotherapygenerally did not elicit a benefit in activating Db/I-tetramer+ CD8 T cells(Fig. 4, C and D). Consistent with our previous studies (42), B10G5monotherapy evoked a significantly higher percentage of CD44hi

Db/I-tetramer+ CD8 T cells (Fig. 4C and fig. S8C) and greater intrinsiccellular response to restimulation (Fig. 4D and fig. S8D). CombinationtherapyofB10G5andanti-CTLA4antibody induced amore pronouncedincrease in the number of CD44hi Db/I-tetramer+ CD8 T cells (Fig. 4Cand fig. S8C) and in IFN-g production by Db/I-tetramer+ CD8 T cells inresponse to PMA/ionomycin restimulation (Fig. 4D and fig. S8D).Together, these data further reveal the effectormechanismwhereby com-bination therapy of B10G5 and anti-CTLA4 antibody heightens antigen-specific CD8 T cell antitumor responses.

B10G5 neutralizing sMIC and anti-CTLA4 antibodycooperatively heighten DC costimulatory potentialWe investigated the mechanisms whereby the cocktail therapy withB10G5 and anti-CTLA4 antibody would synergistically bolster antigen-specific CD8T cell responses.We previously showed that B10G5 therapyalone can enhanceDCactivation in tumor-dLNs and increase the expres-sion of the DC costimulatory molecules CD80 and CD86 (42). With theantibody cocktail therapy, a further significant increase occurs in the DCsurface costimulatory molecules CD80 and CD86, along with the DC ac-tivationmolecule CD40 (Fig. 5, A andB). Such an effectwas not observedin non–tumor-dLNs (fig. S9). Given that CD80 and CD86 can engageboth the costimulatory molecule CD28 and the co-inhibitory moleculeCTLA4, these data indicate that B10G5 neutralizing sMIC in combina-tion with CTLA4 blockade would allow more robust CD28-mediatedcostimulatory signal delivered to antigen-specific CD8 T cells that mayin part confer the synergistic effect on sustaining antigen-specific CD8T cell activity.

Cotargeting sMIC increases TCR clonality andrepertoire complexityIn general, better clinical outcome in response to immunotherapy is as-sociated with higher clonal expansion of CD8 T cells in patients (55–58).

A B

C

CD

44hi

% o

f CD

8 T

01020304050

P < 0.05P < 0.01dLN

01020304050

P < 0.05P < 0.01spln

D

IFN

γ+%

of C

D8

T

0

10

20P < 0.05

P < 0.05dLN

0

10

20

P < 0.05P < 0.01spln

010203040

P < 0.05P < 0.05TIL

0

10

20 P < 0.05

P < 0.05TIL

nsns

spln

CD

8 T

% o

f CD

45+

01020304050 P < 0.05

P < 0.01

0

10

20

dLN

01020304050 P < 0.05

P < 0.01TIL

0

10

20P < 0.05

P < 0.01

P < 0.05P < 0.05

0

10

20

NK

G2D

+%

inC

D8

T

P < 0.05

P < 0.05

0

10

20

30

spln

dLN

TIL

Fig. 3. Combination therapy of anti-sMIC and anti-CTLA4 antibody remarkablyincreases CD8 T cell population, activation, and functional potential in tumor-dLNs and tumor beds. (A to C) Percentage of CD8 T cells (A), NKG2D+ CD8 T cells(B), and activated or memory-like CD8 T cells (C) in the spleen (spln), tumor-dLNs,and tumor infiltrates (TILs). (D) CD8 T cell response to ex vivo restimulation withPMA and ionomycin, as measured by IFN-g production. All representative graphs offlow cytometry analyses are shown in fig. S5.

5 of 12

SC I ENCE ADVANCES | R E S EARCH ART I C L E

In metastatic melanoma and castration-resistant prostate cancer pa-tients, improved survival in response to anti-CTLA4 antibody therapyhas been shown to be associatedwithTCRclonal-type stability aswell asrepertoire complexity (55, 57). Surveying TCR3b sequences within theTRAMP/MIC prostate tumor tissues using in situ immunoSEQ tech-nology revealed that B10G5 cotargeting sMIC remarkably increasednot only TCR content in tumor tissues but also TCR clonality (Fig. 6,A and B). Immunohistochemistry (IHC) assessing CD8 T cell infiltra-tion in tumors further confirmed the relevance of an increase in tumor

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

httpD

ownloaded from

TCR content to B10G5 cotargeting sMIC (Fig. 6C). The data indicatethat cotargeting sMIC facilitates the recruitment of T cells into the tu-mors and the subsequent expansion, which presumably leads to theenhanced therapeutic effect of CTLA4 blockade.

B10G5 therapy stabilizes CD3z in CD8 T cellsWe sought to further understand the mechanisms whereby anti-sMICtherapy increases antigen-specific CD8 T cell maintenance during anti-CTLA4 antibody therapy. In vitro studies have shown that sMIC notonly down-modulates NKG2D expression but also impairs theTCR/CD3 complex signaling of CD8 T cells by destabilizing its keydownstream signalingmolecule CD3z (38).We thus analyzed the intra-cellular levels of CD3z in splenic CD8 T cells from the cohorts of exper-imental animals by flow cytometry. Unstimulated splenic CD8 T cellsfrom TRAMP/MIC mice with high levels of circulating sMIC (sMIChi)had diminished intracellular CD3z with anti-CTLA4 antibody therapycompared to those with low levels of circulating sMIC (sMIClo) (Fig. 7A).All TRAMP/MICB mice that received B10G5 and anti-CTLA4 anti-body therapy had an increased intracellular level of CD3z inCD8T cells(Fig. 7A). A pronounced increase in the level of CD3z was also observedin prostate tissues with B10G5 cotargeting sMIC (Fig. 7B). These data sug-gest that neutralizing sMIC to stabilize CD3z and thus sustain TCR ac-tivation would allow clonal expansion of antigen-specific T cells andincreases in TCR clonality in tumor infiltrates.

on May 17, 2017

://advances.sciencemag.org/

DISCUSSIONUsing a clinically relevant TRAMP/MICB double transgenic mousemodel that closely recapitulates the onco-immunology dynamics of hu-man cancer (39), we demonstrate that the antibody cocktail cotargetingthe soluble NKG2D ligand sMIC not only remarkably augmented thetherapeutic efficacy of immune checkpoint blockade anti-CTLA4 anti-body but also generated a cooperative therapeutic effect with anti-CTLA4therapy.We show that a cocktail therapy of antibody neutralizing the im-munosuppressive effect of sMIC and CTLA4 blockade cooperativelyprimed DC activation and up-regulated the expression of CD80/CD86costimulatorymolecule onDCs, overcameCD8Tcell tolerance, enhancedTCR/CD3 signaling capacity inCD8T cells, and increasedT cell clonalityor repertoire complexity in tumor infiltrates. Given the clinical observa-tions that melanoma patients who developed anti-sMIC autoantibodyduring anti-CTLA4 therapy displayed better prognosis (47), this is the firstproof-of-concept preclinical study to demonstrate that antibody cotar-geting soluble NKG2D sMIC and anti-CTLA4 antibody therapy cangenerate a cooperative therapeutic effect. Our study launched a newviable avenue of combination immune therapy of cancer.

The anti-CTLA4 antibody ipilimumab was approved by the FDAfor the treatment of unresectable or metastatic melanoma; however,ipilimumab was discontinued for adverse reactions in 10% of patientsin a clinical trial formelanoma (www.cancer.gov/about-cancer/treatment/drugs/fda-ipilimumab). In clinical trials with castration-resistant prostatecancer patients, there were ipilimumab toxicity–related deaths, al-though some subgroups of patients benefited from ipilimumab (18, 59–62).Of all the treatment toxicity–related clinical manifestations, the IRAE ofcolitis was the most frequent andmost severe event that had even led todeath (13, 20). Currently, there is no biomarker to predict the patientpopulation that will elicit colitis-related adverse reactions to ipilimumabtherapy. Our unexpected findings in TRAMP/MIC mice suggest thathigh levels of tumor-derived sMIC may be one of the co-founders andalso indicators for anti-CTLA4 antibody therapy–induced autoimmune

B

DCP < 0.05

P < 0.05

0

10

20

P < 0.05P < 0.05

010203040

0102030

P < 0.05P < 0.05

CD

44hi

of D

b/I-

tetra

mer

(%)

dLN

TIL

spln

02468

10

P < 0.05P < 0.05

02468

10

P < 0.05P < 0.05

IFN

γ+of

Db/I-

tetra

mer

(%)

02468

dLN

TIL

spln

A

Db /I

-tetra

mer

+(%

) of C

D8

T 02468

P < 0.05P < 0.01

02468

10P < 0.05P < 0.01

02468

P < 0.05P < 0.01

dLN

TIL

spln

CFS

Elo

%D

b /I-te

tram

er+

CD

8 T

05

1015

P < 0.05P < 0.01spln

02468

10

P < 0.05P < 0.01dLN

02468

10

P < 0.05P < 0.01TIL

P < 0.05P < 0.05

Fig. 4. Combination therapy of anti-sMIC antibody and anti-CTLA4 antibody co-operatively enhances antigen-specific CD8 T cell antitumor responses. CFSE-labeled tumor antigen SV40TAg–specific TCR-I CD8 T cells were transferred intoTRAMP/MICB mice that have received variable antibody regimens for 4 weeks, whichwere continued after transfer of TCR-I CD8 T cells. Shown are data obtained 14 daysafter the transfer of CFSE-labeled TCR-I CD8 T cells. (A) Percentage of Db/I-tetramer+

SV40TAg-specific CD8 T cells in tumor infiltrates, dLNs, and spleen. (B) Proliferationof SV40Tag-specific CD8 T cells represented by CFSElo population. (C) Activation ofDb/I-tetramer+ SV40TAg-specific CD8 T cells in response to therapy shown by theCD44hi population. (D). Response of Db/I-tetramer+ SV40TAg-specific CD8 T cells toex vivo SV40TAg peptide restimulation, as measured by IFN-g production. Representa-tive graphs of flow cytometry analyses are shown in fig. S8.

6 of 12

SC I ENCE ADVANCES | R E S EARCH ART I C L E

colitis in MIC+ cancer patients. We acknowledge that our current find-ings cannot be generalized to MIC− cancer patients. However, becauseMIC is broadly expressed by nearly all human tumors, prescreeningcancer patients for serum sMIC may be necessary to eliminate thepotential intestinal inflammation–related IRAEs of anti-CTLA4 anti-body therapy. Although not addressed here, large-scale retrospectiveclinical investigations on anti-CTLA4 antibody therapy are warrantedto support this concept.

Our data demonstrate that multiple pathways may be involved inthe cooperative therapeutic effects of antibody targeting sMIC and anti-CTLA4 antibody. We previously showed that monotherapy with thesMIC-neutralizing antibody B10G5 up-regulates CD80 and CD86 ex-pression on DCs in tumor-dLNs (42). Therefore, neutralizing sMIC in

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

combination with anti-CTLA4 antibody would provide superior avail-ability of CD80 and CD86 to CD28 compared to respective monother-apy. This mechanism was supported by our data demonstrating thatcombination therapy resulted in a marked increase in the expressionof CD80 and CD86 onDCs from tumor-dLNs and tumor parenchyma.Given that NKG2D and CD28 are known to provide nonredundantcostimulatory signals to CD8 T cells (35) and that neutralizing sMIChas been shown to restore surface NKG2D expression on CD8 T cells(42), increased dual costimulation by NKG2D and CD28 would alsocontribute to the heightened effects of combination therapy. Last, col-leagues and our current study consistently showed that sMIC destabi-lizes the critical TCR signaling molecule CD3z on CD8 T cells duringprolonged stimulation (38), which would impair antigen-specific CD8

on May 17, 2017

http://advances.sciencemag.org/

Dow

nloaded from

A

αCTLA4 + B10G5

In dLNs

αCTLA4

B10G5

mIgG

CD80 CD40CD86

In TILs

CD

11c

MHCII

B

CD80 CD40CD86

0

500

1000

1500

0

500

1000

1500

2000

MFI

incr

ease

in D

C-C

D80

(dLN

)

MFI

incr

ease

in D

C- C

D86

(dLN

)

MFI

incr

ease

in D

C- C

D40

(dLN

)

0

500

1000

1500

0

500

1000

1500

0200400600800

10001200

MFI

incr

ease

in D

C-C

D80

(TIL

s)

MFI

incr

ease

in D

C- C

D86

(TIL

s)

MFI

incr

ease

in D

C-C

D40

(TIL

s)

0

500

1000

1500

2000

P < 0.05 P < 0.05 P < 0.05

P < 0.05P < 0.05

P < 0.05

Fig. 5. B10G5andanti-CTLA4antibody therapy cooperatively increasesDCactivation (CD40) and the expressionof costimulatorymolecules CD80andCD86 in tumorsites. (A) Representative histograms from flow cytometry analyses of CD40, CD80, and CD86 expression on DCs from tumor-dLNs and tumor beds. Gray-filled profiles, controlisotype staining; open dark profiles, antibody to specific DC surface molecules. (B) Summary data of mean fluorescence intensity (MFI) of CD40, CD80, and CD86 on DCs.

7 of 12

SC I ENCE ADVANCES | R E S EARCH ART I C L E

hD

ownloaded from

T cell function and, in turn, restrict its clonal expansion. Thus, stabiliz-ing the TCR/CD3 complex and restoring antigen-specific TCR sig-naling in CD8 T cells by B10G5 also contribute to the enhancedresponse to anti-CTLA4 antibody therapy. Our data of increasedTCR repertoire clonality and CD3z expression in the tumor tissue inresponse to B10G5 therapy corroborated this underlying mechanism.

Notably, animals with high levels of circulating sMIC developed se-vere colon inflammation or colitis in response to anti-CTLA4 antibodytherapy; an antibody neutralizing sMIC alleviated this adverse auto-immune effect. This biology was never observed or reported in otherpreclinical mouse models with anti-CTLA4 antibody therapy. TheNKG2D signaling pathway has been associated with inflammatorybowel disease in humans (63–65); however, this finding is not conclu-sive in mice because of the differences in the models studied, some ofwhich do not naturally express NKG2D ligands (66, 67).

We also observed progressive lung metastasis in sMIChi TRAMP/MICB mice in response to anti-CTLA4 antibody monotherapy. Onelimitation to the current study is that we were not be able to determinewhether increased lung metastasis and development of colitis were re-lated as a cause and an effect or whether both were the collateral or con-current events of a common cause. Studies have shown that anti-CTLA4 antibody therapy selectively propagates specific species of gutmicrobiota (68). Studies have also shown that gutmicrobiota could reg-

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

ulateNKG2Dexpression in the gut and thatNKG2D signaling in IFN-g–producing intestinal CD4 T cells plays a role in the pathogenesis of colitis(66, 69, 70). Together, it is conceivable to propose that anti-CTLA4 anti-body therapy–induced colitis in sMIChi mice may result from the inter-play of microbiota composition and NKG2D signaling in pathogenicCD4 T cells in the gut. Moreover, gut microbiota has also been shownto regulate the therapeutic efficacy ofCTLA4blockade (68). Thus, wheth-er the development of colitis and increased lungmetastasis are related andhow they interplay could be complicated biological events. Thesequestions are under active investigation.

MIC is a human molecule that is not expressed by rodents. Becauseof this limitation, preclinical studies onCTLA4blockade therapy to datehave not had the opportunity to consider the immune modulatoryimpact of sMIC. Using an “MIC humanized” double transgenic model,we demonstrated that antibody neutralizing sMIC not only enhancesthe therapeutic efficacy of anti-CTLA4 antibody but also generates acooperative therapeutic effect with anti-CTLA4 antibody. Given thatsMIC was prevalent in a broad range of malignancies and that anti-CTLA4 antibody is currently in active clinical trial for cancers other thanmelanoma, our study revealed a new avenue to enhance the therapeuticeffect of anti-CTLA4 antibody and concurrently eliminate therapy-induced autoimmune colitis. This concept was supported by a pub-lished case report in a melanoma patient (47) and our current case

on May 17, 2017

ttp://advances.sciencemag.org/

CD8

C

A BP < 0.05

TCR

abu

ndan

ce

0.00

0.02

0.04

0.06

0.08

αCTLA4 + B10G5αCTLA4

TCR

clo

nalit

y

0

1000

2000

3000P < 0.05

Fig. 6. B10G5 cotargeting sMIC during CTLA4 blockade therapy increases T cell content and TCR clonality in tumors.DNAwere isolated from paraffin-embedded tumorsections and subjected to immunoSEQ assay, as previously described (Adaptive Biotechnologies). Tumor in situ TCR abundance (A) and clonality (B) were assayed by surveyingTCR3b expression and analyzed with the Shannon diversity index using immunoSEQ Analyzer software. (C) Representative IHC confirming increased CD8 T cell infiltration intumors with B10G5 cotargeting sMIC.

8 of 12

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on May 17, 2017

http://advances.sciencemag.org/

Dow

nloaded from

study in a metastatic prostate cancer patient that reveals the associ-ation of better clinical prognosis with the development of anti-sMICautoantibody during early response to ipilimumab. The impact of sMICon colon inflammation during anti-CTLA4 antibody therapy is an un-expected aspect of our findings. Although the underlying mechanismsmay be complex and are not currently understood, our findingsnevertheless also suggest a potential clinical biomarker to predict re-sponsiveness or toxicity of anti-CTLA4 antibody therapy in cancer pa-tients. Large-scale clinical investigations are warranted.

MATERIALS AND METHODSAntibodies and flow cytometrySingle-cell suspensions from spleens, dLNs, non-dLNs, or tumor tissueswere prepared as previously described (42). Combinations of thefollowing fluorochrome-conjugated antibody were used for cell surfaceor intracellular staining to define populations of NK, CD8, and subsetsof CD4 T cells: CD3e (clone 145-2c11), CD8a (clone 53-6.7), CD4(clone GK1.5), NK1.1 (clone PK136), NKG2D (clone CX5), CD44

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

(clone eBio4B10), CD11c (clone N418), MHCII (clone M5/114.15.2),CD80 (clone 16-10A1), CD86 (clone PO3), CD40 (clone 1C10), andCD3z (clone 6B10.2, BioLegend). For ex vivo restimulation, freshlyisolated single-cell suspension was cultured in complete RPMI 1640medium containing PMA (50 ng/ml) and ionomycin (500 ng/ml) for6 hours before it was analyzed for IFN-g production by intracellularstaining with an antibody specific to IFN-g (XMG1.2). Multicoloredflow cytometry analyses were performed on LSR II (BD). Data wereanalyzed with FlowJo software (Tree Star).

Animals and antibody therapyAll animal experimental procedures were approved by the InstitutionalAnimal Care and Use Committee at the Medical University of SouthCarolina (MUSC).Miceweremaintained at theMUSCHollingsCancerCenter animal facility under specific pathogen–free conditions. MaleTRAMP mice at the age of 27 to 28 weeks were randomized into twotreatment cohorts: (i) cIgG or (ii) anti-CTLA4 antibody (clone 9H10,BioXCell). In parallel, age-matched TRAMP/MICB double transgenicmice were randomized into four treatment cohorts: (i) anti-sMICmAbB10G5 [whichwas previously described (39)], (ii) anti-CTLA4 antibody(clone 9H10), (iii) a cocktail of B10G5 and anti-CTLA4 antibody, and(iv) a cocktail of cIgGs. All antibodies were administered at a dose of3 mg/kg body weight through intraperitoneal injection twice per week.All animals received the treatment for 8 weeks before they were eutha-nized at the study end point, unless they succumbed to diseases or ad-verse effects at an earlier time point.

Antigen-specific T cell tolerance experimentThe experimental procedure was described previously (42). Briefly,CD8 T cells from TCR-I transgenic mice were labeled with CFSE andinjected intravenously into animals (2 × 106 cells per mouse) in the re-spective experimental groups. Animals were sacrificed at the indicatedtimepoints to assess TCR-I T cell in vivo frequencywithTCR-I–specificH-2Db/TAg epitope I–tetramer (Db/I-tetramer) (71). To assay antigen-specific CD8 T cell responses, a bulked single-cell suspension preparedfrom the spleen, lymph nodes, and tumor digests was stimulatedovernight with 0.5 mMTAg epitope I peptide (SAINNYAQKL) and as-sayed for intracellular IFN-g expressionofCD8+orDb/I-tetramer+Tcellsby flow cytometry.

Tissue collectionBlood was collected via tail bleeding during therapy and via cardiacpuncture after euthanasia. Splenocytes, tumor-dLNs, and non–tumor-dLNs were collected for immunological analyses. Prostates, lungs, livers,kidneys, pancreata, and colons were collected and fixed in 10% neutralfixation buffer, followed by paraffin or Tissue-Tek OCT (optimumcutting temperature) compound embedding for pathological and histo-logical examinations. In some experiments, portions of prostate tumorswere used for parathion in single-cell suspensions.

Histology and IHC stainingAll collected tissues were sectioned and stained with hematoxylin andeosin for histological evaluation. For IHC staining to detect specific anti-gens, the following antibodies were used: anti-Ki67 (Neomarkers), anti–cleaved caspase-3 (clone5A1E,Cell Signaling), anti-CD8 (BDBiosciences),anti-NK1.1 (PK136, eBioscience), anti-SV40Tag (Santa Cruz), anti–arginase I (Santa Cruz), and anti-p63 (Neomarkers). The IHC stainingprotocol has been previously described (39, 42). All sections werecounterstained with hematoxylin for nucleus visualization.

αCTLA4 αCTLA4 + B10G5

A

C

αCTLA4 +

B10G5

αCTLA4

CD3ζCD8

CD

3C

D3ζ

sMIChisMIClo

0

50

100

150

200

αCTLA4 +

B10G5αCTLA4

sMIChisMIClo

P < 0.05P < 0.05

MFI

of C

D3ζ

inC

D8T

(spl

een)

B

MFI: 151

MFI: 96.5

MFI: 128

Fig. 7. Neutralizing sMIC stabilizes the TCR signaling molecule TCR3z. (A and B)Splenocytes from TRAMP/MIC mice that received anti-CTLA4 monotherapy and acocktail therapy of B10G5 and anti-CTLA4were intracellularly stained for CD3z. (A) Rep-resentative fluorescence-activated cell sorting graphs demonstrating CD3z expressionin CD8 T cells with different regimens. (B) Summary data of theMFI of CD3z expressionin CD8T cells. (C) Representative IHC stainingdemonstrating the level of CD3z in tumorparenchyma.

9 of 12

SC I ENCE ADVANCES | R E S EARCH ART I C L E

http://aD

ownloaded from

Serum sMIC and cytokine detectionSerum levels of sMICB were assessed using respective sandwichenzyme-linked immunosorbent assay kits (R&D Systems). Serumlevels of cytokines were assayed by the Eve Technologies Corpo-ration using the Luminex technology.

High-throughput TCR sequencing in tumor tissuesA high-throughput TCR sequence survey from paraffin-embeddedtissue was performed according to protocol described elsewhere (56).Briefly, sections from paraffin blocks of the prostate were collected forDNA extraction using the QIAamp DNA FFPE Tissue Kit (QIAGENInc.), per the manufacturer’s instructions. The TCRb CD3 (CDR3b)region was amplified, sequenced, and quantified from a standardized1200 ng of DNA using the immunoSEQ assay as previously described(Adaptive Biotechnologies) (56, 72). TCR sequence and clonality weredetermined by the Shannon diversity index using immunoSEQAnalyzersoftware (Adaptive Biotechnologies) (56).

Statistical analysisAll results are means ± SEM, unless specified otherwise. Mouse andsample group sizes were n > 5, unless indicated otherwise. Data wereanalyzed using the analysis of variance unpaired t test. Differences be-tweenmeans were considered significant at P < 0.05. Kaplan-Meier sur-vival curves were generated and analyzed using GraphPad Prismsoftware.

on May 17, 2017

dvances.sciencemag.org/

SUPPLEMENTARY MATERIALSSupplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/3/5/e1602133/DC1fig. S1. Induction of subclinical colitis in sMIChi TRAMP/MIC mice in response to anti-CTLA4therapy.fig. S2. Recapitulation of the negative effect of sMIC on anti-CTLA4 therapy in a transplantabletumor model.fig. S3. Antibody neutralizing sMIC eliminates colitis in TRAMP/MICB mice that received anti-CTLA4 therapy.fig. S4. Detection of anti-sMIC autoantibody in the sera of a small cohort of prostate cancerpatients who have metastatic disease and enrolled in a clinical trial (NCT01498978) ofipilimumab in combination with hormone suppression at the Knight Cancer Institute.fig. S5. Representative graphs of flow cytometry analyses demonstrate that combinationtherapy of anti-sMIC and anti-CTLA4 antibody remarkably increases CD8 T cell population,activation, and functional potential in tumor-draining lymph nodes and tumor beds.fig. S6. Marginal response of CD8 T cells to anti-CTLA4 therapy in TRAMP mice.fig. S7. Circulating sMIC or anti-sMIC autoantibody affects response to anti-CTLA4 therapy inTRAMP/MICB mice compared to MIC-negative TRAMP mice.fig. S8. Representative graphs of flow cytometry analyses demonstrate that combinationtherapy of anti-sMIC antibody and anti-CTLA4 antibody cooperatively enhances antigen-specific CD8 T cell anti-tumor responses.fig. S9. Therapy has no effect on the activation or costimulatory molecule on DCs in the spleenor non–tumor-dLNs.

REFERENCES AND NOTES1. H. Bour-Jordan, J. A. Bluestone, Regulating the regulators: Costimulatory signals control

the homeostasis and function of regulatory T cells. Immunol. Rev. 229, 41–66 (2009).2. J. G. Egen, M. S. Kuhns, J. P. Allison, CTLA-4: New insights into its biological function and

use in tumor immunotherapy. Nat. Immunol. 3, 611–618 (2002).3. K. S. Peggs, S. A. Quezada, A. J. Korman, J. P. Allison, Principles and use of anti-CTLA4

antibody in human cancer immunotherapy. Curr. Opin. Immunol. 18, 206–213 (2006).4. J. D. Wolchok, F. S. Hodi, J. S. Weber, J. P. Allison, W. J. Urba, C. Robert, S. J. O’Day, A. Hoos,

R. Humphrey, D. M. Berman, N. Lonberg, A. J. Korman, Development of ipilimumab: Anovel immunotherapeutic approach for the treatment of advanced melanoma. Ann. N. Y.Acad. Sci. 1291, 1–13 (2013).

5. K. Wing, Y. Onishi, P. Prieto-Martin, T. Yamaguchi, M. Miyara, Z. Fehervari, T. Nomura,S. Sakaguchi, CTLA-4 control over Foxp3+ regulatory T cell function. Science 322, 271–275(2008).

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

6. S. Read, V. Malmström, F. Powrie, Cytotoxic T lymphocyte–associated antigen 4 plays anessential role in the function of CD25+CD4+ regulatory cells that control intestinalinflammation. J. Exp. Med. 192, 295–302 (2000).

7. D. R. Leach, M. F. Krummel, J. P. Allison, Enhancement of antitumor immunity by CTLA-4blockade. Science 271, 1734–1736 (1996).

8. B. Kavanagh, S. O’Brien, D. Lee, Y. Hou, V. Weinberg, B. Rini, J. P. Allison, E. J. Small,L. Fong, CTLA4 blockade expands FoxP3+ regulatory and activated effector CD4+ T cellsin a dose-dependent fashion. Blood 112, 1175–1183 (2008).

9. K. S. Peggs, S. A. Quezada, C. A. Chambers, A. J. Korman, J. P. Allison, Blockade of CTLA-4on both effector and regulatory T cell compartments contributes to the antitumoractivity of anti–CTLA-4 antibodies. J. Exp. Med. 206, 1717–1725 (2009).

10. S. A. Quezada, K. S. Peggs, M. A. Curran, J. P. Allison, CTLA4 blockade and GM-CSFcombination immunotherapy alters the intratumor balance of effector and regulatoryT cells. J. Clin. Invest. 116, 1935–1945 (2006).

11. M. J. Selby, J. J. Engelhardt, M. Quigley, K. A. Henning, T. Chen, M. Srinivasan, A. J. Korman,Anti-CTLA-4 antibodies of IgG2a isotype enhance antitumor activity through reduction ofintratumoral regulatory T cells. Cancer Immunol. Res. 1, 32–42 (2013).

12. T. R. Simpson, F. Li, W. Montalvo-Ortiz, M. A. Sepulveda, K. Bergerhoff, F. Arce,C. Roddie, J. Y. Henry, H. Yagita, J. D. Wolchok, K. S. Peggs, J. V. Ravetch, J. P. Allison,S. A. Quezada, Fc-dependent depletion of tumor-infiltrating regulatory T cells co-definesthe efficacy of anti–CTLA-4 therapy against melanoma. J. Exp. Med. 210, 1695–1710(2013).

13. F. Stephen Hodi, S. J. O’Day, D. F. McDermott, R. W. Weber, J. A. Sosman, J. B. Haanen,R. Gonzalez, C. Robert, D. Schadendorf, J. C. Hassel, W. Akerley, A. J. M. van den Eertwegh,J. Lutzky, P. Lorigan, J. M. Vaubel, G. P. Linette, D. Hogg, C. H. Ottensmeier, C. Lebbé,C. Peschel, I. Quirt, J. I. Clark, J. D. Wolchok, J. S. Weber, J. Tian, M. J. Yellin, G. M. Nichol,A. Hoos, W. J. Urba, Improved survival with ipilimumab in patients with metastaticmelanoma. N. Engl. J. Med. 363, 711–723 (2010).

14. C. Robert, L. Thomas, I. Bondarenko, S. O’Day, J. Weber, C. Garbe, C. Lebbe, J.-F. Baurain,A. Testori, J.-J. Grob, N. Davidson, J. Richards, M. Maio, A. Hauschild, W. H. Miller Jr.,P. Gascon, M. Lotem, K. Harmankaya, R. Ibrahim, S. Francis, T.-T. Chen, R. Humphrey,A. Hoos, J. D. Wolchok, Ipilimumab plus dacarbazine for previously untreated metastaticmelanoma. N. Engl. J. Med. 364, 2517–2526 (2011).

15. J. D. Wolchok, L. Thomas, I. N. Bondarenko, S. O’Day, J. S. Weber, C. Garbe, S. Francis,R. A. Ibrahim, A. Hoos, C. Robert, Phase III randomized study of ipilimumab (IPI) plusdacarbazine (DTIC) versus DTIC alone as first-line treatment in patients with unresectablestage III or IV melanoma. J. Clin. Oncol. 29, LBA5 (2011).

16. C. Rolfo, G. Sortino, E. Smits, F. Passiglia, G. Bronte, M. Castiglia, A. Russo, E. S. Santos,A. Janssens, P. Pauwels, L. Raez, Immunotherapy: Is a minor god yet in the pantheon oftreatments for lung cancer? Expert Rev. Anticancer Ther. 14, 1173–1187 (2014).

17. B. Xia, R. S. Herbst, Immune checkpoint therapy for non-small-cell lung cancer: Anupdate. Immunotherapy 8, 279–298 (2016).

18. S. Bracarda, A. Altavilla, A. Hamzaj, M. Sisani, F. Marrocolo, S. Del Buono, R. Danielli,Immunologic checkpoints blockade in renal cell, prostate, and urothelial malignancies.Semin. Oncol. 42, 495–505 (2015).

19. K. Dubin, M. K. Callahan, B. Ren, R. Khanin, A. Viale, L. Ling, D. No, A. Gobourne,E. Littmann, C. Huttenhower, E. G. Pamer, J. D. Wolchok, Intestinal microbiome analysesidentify melanoma patients at risk for checkpoint-blockade-induced colitis.Nat. Commun. 7, 10391 (2016).

20. A. Bertrand, M. Kostine, T. Barnetche, M. E. Truchetet, T. Schaeverbeke, Immune relatedadverse events associated with anti-CTLA-4 antibodies: Systematic review andmeta-analysis. BMC Med. 13, 211 (2015).

21. J. M. Michot, C. Bigenwald, S. Champiat, M. Collins, F. Carbonnel, S. Postel-Vinay,A. Berdelou, A. Varga, R. Bahleda, A. Hollebecque, C. Massard, A. Fuerea, V. Ribrag,A. Gazzah, J. P. Armand, N. Amellal, E. Angevin, N. Noel, C. Boutros, C. Mateus, C. Robert,J. C. Soria, A. Marabelle, O. Lambotte, Immune-related adverse events with immunecheckpoint blockade: A comprehensive review. Eur. J. Cancer 54, 139–148 (2016).

22. J. Larkin, V. Chiarion-Sileni, R. Gonzalez, J. J. Grob, C. L. Cowey, C. D. Lao, D. Schadendorf,R. Dummer, M. Smylie, P. Rutkowski, P. F. Ferrucci, A. Hill, J. Wagstaff, M. S. Carlino,J. B. Haanen, M. Maio, I. Marquez-Rodas, G. A. McArthur, P. A. Ascierto, G. V. Long,M. K. Callahan, M. A. Postow, K. Grossmann, M. Sznol, B. Dreno, L. Bastholt, A. Yang,L. M. Rollin, C. Horak, F. S. Hodi, J. D. Wolchok, Combined nivolumab and ipilimumab ormonotherapy in untreated melanoma. N. Engl. J. Med. 373, 23–34 (2015).

23. K. K. Tsai, A. I. Daud, Nivolumab plus ipilimumab in the treatment of advancedmelanoma. J. Hematol. Oncol. 8, 123 (2015).

24. J. D. Wolchok, H. Kluger, M. K. Callahan, M. A. Postow, N. A. Rizvi, A. M. Lesokhin,N. H. Segal, C. E. Ariyan, R.-A. Gordon, K. Reed, M. M. Burke, A. Caldwell, S. A. Kronenberg,B. U. Agunwamba, X. Zhang, I. Lowy, H. D. Inzunza, W. Feely, C. E. Horak, Q. Hong,A. J. Korman, J. M. Wigginton, A. Gupta, M. Sznol, Nivolumab plus ipilimumabin advanced melanoma. N. Engl. J. Med. 369, 122–133 (2013).

25. V. Groh, J. Wu, C. Yee, T. Spies, Tumour-derived soluble MIC ligands impair expression ofNKG2D and T-cell activation. Nature 419, 734–738 (2002).

10 of 12

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on May 17, 2017

http://advances.sciencemag.org/

Dow

nloaded from

26. A. Märten, M. von Lilienfeld-Toal, M. W. Buchler, J. Schmidt, Soluble MIC is elevated in theserum of patients with pancreatic carcinoma diminishing gd T cell cytotoxicity.Int. J. Cancer 119, 2359–2365 (2006).

27. H. Wang, D. Yang, W. Xu, Y. Wang, Z. Ruan, T. Zhao, J. Han, Y. Wu, Tumor-derived solubleMICs impair CD3+CD56+ NKT-like cell cytotoxicity in cancer patients. Immunol. Lett.120, 65–71 (2008).

28. S. Holdenrieder, P. Stieber, A. Peterfi, D. Nagel, A. Steinle, H. Rainer Salih, Soluble MICA inmalignant diseases. Int. J. Cancer 118, 684–687 (2006).

29. S. Holdenrieder, P. Stieber, A. Peterfi, D. Nagel, A. Steinle, H. Rainer Salih, Soluble MICB inmalignant diseases: Analysis of diagnostic significance and correlation with solubleMICA. Cancer Immunol. Immunother. 55, 1584–1589 (2006).

30. H. R. Salih, S. Holdenrieder, A. Steinle, Soluble NKG2D ligands: Prevalence, release, andfunctional impact. Front. Biosci. 13, 3448–3456 (2008).

31. H. Rainer Salih, H. Antropius, F. Gieseke, S. Z. Lutz, L. Kanz, H.-G. Rammensee, A. Steinle,Functional expression and release of ligands for the activating immunoreceptorNKG2D in leukemia. Blood 102, 1389–1396 (2003).

32. S. Bauer, V. Groh, J. Wu, A. Steinle, J. H. Phillips, L. L. Lanier, T. Spies, Activation ofNK cells and T cells by NKG2D, a receptor for stress-inducible MICA. Science 285, 727–729(1999).

33. J. Wu, V. Groh, T. Spies, T cell antigen receptor engagement and specificity in therecognition of stress-inducible MHC class I-related chains by human epithelial gd T cells.J. Immunol. 169, 1236–1240 (2002).

34. V. Groh, R. Rhinehart, J. Randolph-Habecker, M. S. Topp, S. R. Riddell, T. Spies,Costimulation of CD8ab T cells by NKG2D via engagement by MIC induced on virus-infected cells. Nat. Immunol. 2, 255–260 (2001).

35. M. A. Markiewicz, L. N. Carayannopoulos, O. V. Naidenko, K. Matsui, W. R. Burack, E. L. Wise,D. H. Fremont, P. M. Allen, W. M. Yokoyama, M. Colonna, A. S. Shaw, Costimulationthrough NKG2D enhances murine CD8+ CTL function: Similarities and differences betweenNKG2D and CD28 costimulation. J. Immunol. 175, 2825–2833 (2005).

36. E. Serrano-Pertierra, E. Cernuda-Morollón, C. Lopez-Larrea, NKG2D- and CD28-mediatedcostimulation regulate CD8+ T cell chemotaxis through different mechanisms: Therole of Cdc42/N-WASp. J. Leukoc. Biol. 95, 487–495 (2014).

37. A. I. Roberts, L. Lee, E. Schwarz, V. Groh, T. Spies, E. C. Ebert, B. Jabri, NKG2D receptorsinduced by IL-15 costimulate CD28-negative effector CTL in the tissuemicroenvironment. J. Immunol. 167, 5527–5530 (2001).

38. N. Hanaoka, B. Jabri, Z. Dai, C. Ciszewski, A. M. Stevens, C. Yee, H. Nakakuma, T. Spies,V. Groh, NKG2D initiates caspase-mediated CD3z degradation and lymphocyte receptorimpairments associated with human cancer and autoimmune disease. J. Immunol.185, 5732–5742 (2010).

39. G. Liu, S. Lu, X. Wang, S. T. Page, C. S. Higano, S. R. Plymate, N. M. Greenberg, S. Sun,Z. Li, J. D. Wu, Perturbation of NK cell peripheral homeostasis accelerates prostatecarcinoma metastasis. J. Clin. Invest. 123, 4410–4422 (2013).

40. J. D. Wu, L. M. Higgins, A. Steinle, D. Cosman, K. Haugk, S. R. Plymate, Prevalent expressionof the immunostimulatory MHC class I chain-related molecule is counteracted byshedding in prostate cancer. J. Clin. Invest. 114, 560–568 (2004).

41. G. Xiao, X. Wang, J. Sheng, S. Lu, X. Yu, J. D. Wu, Soluble NKG2D ligand promotesMDSC expansion and skews macrophage to the alternatively activated phenotype.J. Hematol. Oncol. 8, 13 (2015).

42. S. Lu, J. Zhang, D. Liu, G. Li, K. F. Staveley-O’Carroll, Z. Li, J. D. Wu, Non-blockingmonoclonal antibody targeting soluble MIC revamps endogenous innate and adaptiveanti-tumor responses and eliminates primary and metastatic tumors. Clin. Cancer Res.21, 4819–4830 (2015).

43. J. Wu, Antibody targeting soluble NKG2D ligand sMIC refuels and invigorates theendogenous immune system to fight cancer. Oncoimmunology 5, e1095434 (2016).

44. A. A. Hurwitz, B. A. Foster, E. D. Kwon, T. Truong, E. M. Choi, N. M. Greenberg, M. B. Burg,J. P. Allison, Combination immunotherapy of primary prostate cancer in a transgenicmouse model using CTLA-4 blockade. Cancer Res. 60, 2444–2448 (2000).

45. S. Wada, C. M. Jackson, K. Yoshimura, H.-R. Yen, D. Getnet, T. J. Harris, M. V. Goldberg,T. C. Bruno, J. F. Grosso, N. Durham, G. J. Netto, D. M. Pardoll, C. G. Drake,Sequencing CTLA-4 blockade with cell-based immunotherapy for prostate cancer.J. Transl. Med. 11, 89 (2013).

46. F. Basher, E. K. Jeng, H. Wong, J. Wu, Cooperative therapeutic anti-tumor effect of IL-15agonist ALT-803 and co-targeting soluble NKG2D ligand sMIC. Oncotarget 7, 814–830(2016).

47. M. Jinushi, F. S. Hodi, G. Dranoff, Therapy-induced antibodies to MHC class I chain-relatedprotein A antagonize immune suppression and stimulate antitumor cytotoxicity.Proc. Natl. Acad. Sci. U.S.A. 103, 9190–9195 (2006).

48. D. H. Raulet, Roles of the NKG2D immunoreceptor and its ligands. Nat. Rev. Immunol. 3,781–790 (2003).

49. K. Rajasekaran, V. Xiong, L. Fong, J. Gorski, S. Malarkannan, Functional dichotomybetween NKG2D and CD28-mediated co-stimulation in human CD8+ T cells. PLOS ONE 5,e12635 (2010).

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

50. M. Della Chiesa, S. Sivori, R. Castriconi, E. Marcenaro, A. Moretta, Pathogen-inducedprivate conversations between natural killer and dendritic cells. Trends Microbiol.13, 128–136 (2005).

51. C. Consentius, L. Akyüz, J. A. Schmidt-Lucke, C. Tschöpe, L. Pinzur, R. Ofir, P. Reinke,H.-D. Volk, K. Juelke, Mesenchymal stromal cells prevent allostimulation in vivo andcontrol checkpoints of Th1 priming: Migration of human DC to lymph nodes and NK cellactivation. Stem Cells 33, 3087–3099 (2015).

52. S. Mandaric, S. M. Walton, T. Rülicke, K. Richter, M. J. H. Girard-Madoux, B. E. Clausen,A. Zurunic, M. Kamanaka, R. A. Flavell, S. Jonjic, A. Oxenius, IL-10 suppression of NK/DCcrosstalk leads to poor priming of MCMV-specific CD4 T cells and prolonged MCMVpersistence. PLOS Pathog. 8, e1002846 (2012).

53. T. Walzer, M. Dalod, E. Vivier, L. Zitvogel, Natural killer cell–dendritic cell crosstalk in theinitiation of immune responses. Expert Opin. Biol. Ther. 5 (suppl. 1), S49–S59 (2005).

54. A. Bai, E. Higham, H. N. Eisen, K. D. Wittrup, J. Chen, Rapid tolerization of virus-activatedtumor-specific CD8+ T cells in prostate tumors of TRAMP mice. Proc. Natl. Acad. Sci.U.S.A. 105, 13003–13008 (2008).

55. E. Cha, M. Klinger, Y. Hou, C. Cummings, A. Ribas, M. Faham, L. Fong, Improved survivalwith T cell clonotype stability after anti–CTLA-4 treatment in cancer patients.Sci. Transl. Med. 6, 238ra270 (2014).

56. N. Sheikh, J. Cham, L. Zhang, T. DeVries, S. Letarte, J. Pufnock, D. Hamm, J. Trager, L. Fong,Clonotypic diversification of intratumoral T cells following sipuleucel-T treatment inprostate cancer subjects. Cancer Res. 76, 3711–3718 (2016).

57. L. Robert, C. Harview, R. Emerson, X. Wang, S. Mok, B. Homet, B. Comin-Anduix, R. C. Koya,H. Robins, P. C. Tumeh, A. Ribas, Distinct immunological mechanisms of CTLA-4 andPD-1 blockade revealed by analyzing TCR usage in blood lymphocytes. Oncoimmunology3, e29244 (2014).

58. P. C. Tumeh, C. L. Harview, J. H. Yearley, I. P. Shintaku, E. J. M. Taylor, L. Robert,B. Chmielowski, M. Spasic, G. Henry, V. Ciobanu, A. N. West, M. Carmona, C. Kivork,E. Seja, G. Cherry, A. J. Gutierrez, T. R. Grogan, C. Mateus, G. Tomasic, J. A. Glaspy,R. O. Emerson, H. Robins, R. H. Pierce, D. A. Elashoff, C. Robert, A. Ribas, PD-1 blockadeinduces responses by inhibiting adaptive immune resistance. Nature 515, 568–571 (2014).

59. J. N. Graff, S. Puri, C. B. Bifulco, B. A. Fox, T. M. Beer, Sustained complete response toCTLA-4 blockade in a patient with metastatic, castration-resistant prostate cancer.Cancer Immunol. Res. 2, 399–403 (2014).

60. E. J. Small, N. S. Tchekmedyian, B. I. Rini, L. Fong, I. Lowy, J. P. Allison, A pilot trial ofCTLA-4 blockade with human anti-CTLA-4 in patients with hormone-refractory prostatecancer. Clin. Cancer Res. 13, 1810–1815 (2007).

61. T. M. Beer, E. D. Kwon, C. G. Drake, K. Fizazi, C. Logothetis, G. Gravis, V. Ganju,J. Polikoff, F. Saad, P. Humanski, J. M. Piulats, P. G. Mella, S. S. Ng, D. Jaeger, F. X. Parnis,F. A. Franke, J. Puente, R. Carvajal, L. Sengeløv, M. B. McHenry, A. Varma,A. J. van den Eertwegh, W. Gerritsen, Randomized, double-blind, phase III trial ofipilimumab versus placebo in asymptomatic or minimally symptomatic patients with metastaticchemotherapy-naive castration-resistant prostate cancer. J. Clin. Oncol. 35, 40–47 (2017).

62. X. X. Wei, L. Fong, E. J. Small, Prospects for the use of ipilimumab in treating advancedprostate cancer. Expert Opin. Biol. Ther. 16, 421–432 (2016).

63. B. Pariente, I. Mocan, M. Camus, C.-A. Dutertre, J. Ettersperger, P. Cattan, J.-M. Gornet,N. Dulphy, D. Charron, M. Lémann, A. Toubert, M. Allez, Activation of the receptor NKG2Dleads to production of Th17 cytokines in CD4+ T cells of patients with Crohn’s disease.Gastroenterology 141, 217–226.e2 (2011).

64. R. La Scaleia, A. Stoppacciaro, S. Oliva, S. Morrone, G. Di Nardo, A. Santoni, S. Cucchiara,G. Palmieri, NKG2D/ligand dysregulation and functional alteration of innate immunitycell populations in pediatric IBD. Inflamm. Bowel Dis. 18, 1910–1922 (2012).

65. M. Allez, V. Tieng, A. Nakazawa, X. Treton, V. Pacault, N. Dulphy, S. Caillat-Zucman, P. Paul,J.-M. Gornet, C. Douay, S. Ravet, R. Tamouza, D. Charron, M. Lémann, L. Mayer, A. Toubert,CD4+NKG2D+ T cells in Crohn’s disease mediate inflammatory and cytotoxic responsesthrough MICA interactions. Gastroenterology 132, 2346–2358 (2007).

66. Y. Ito, T. Kanai, T. Totsuka, R. Okamoto, K. Tsuchiya, Y. Nemoto, A. Yoshioka, T. Tomita,T. Nagaishi, N. Sakamoto, T. Sakanishi, K. Okumura, H. Yagita, M. Watanabe, Blockadeof NKG2D signaling prevents the development of murine CD4+ T cell-mediated colitis.Am. J. Physiol. Gastrointest. Liver Physiol. 294, G199–G207 (2008).

67. N. Guerra, K. Pestal, T. Juarez, J. Beck, K. Tkach, L. Wang, D. H. Raulet, A selective role ofNKG2D in inflammatory and autoimmune diseases. Clin. Immunol. 149, 432–439 (2013).

68. M. Vétizou, J. M. Pitt, R. Daillère, P. Lepage, N. Waldschmitt, C. Flament, S. Rusakiewicz,B. Routy, M. P. Roberti, C. P. M. Duong, V. Poirier-Colame, A. Roux, S. Becharef, S. Formenti,E. Golden, S. Cording, G. Eberl, A. Schlitzer, F. Ginhoux, S. Mani, T. Yamazaki, N. Jacquelot,D. P. Enot, M. Bérard, J. Nigou, P. Opolon, A. Eggermont, P.-L. Woerther, E. Chachaty,N. Chaput, C. Robert, C. Mateus, G. Kroemer, D. Raoult, I. G. Boneca, F. Carbonnel,M. Chamaillard, L. Zitvogel, Anticancer immunotherapy by CTLA-4 blockade relies on the gutmicrobiota. Science 350, 1079–1084 (2015).

69. C. H. F. Hansen, T. L. Holm, Ł. Krych, L. Andresen, D. S. Nielsen, I. Rune, A. K. Hansen,S. Skov, Gut microbiota regulates NKG2D ligand expression on intestinal epithelial cells.Eur. J. Immunol. 43, 447–457 (2013).

11 of 12

SC I ENCE ADVANCES | R E S EARCH ART I C L E

70. S. Kjellev, C. Haase, D. Lundsgaard, B. Ursø, D. Tornehave, H. Markholst, Inhibition ofNKG2D receptor function by antibody therapy attenuates transfer-induced colitis in SCIDmice. Eur. J. Immunol. 37, 1397–1406 (2007).

71. K. Staveley-O’Carroll, T. D. Schell, M. Jimenez, L. M. Mylin, M. J. Tevethia,S. P. Schoenberger, S. S. Tevethia, In vivo ligation of CD40 enhances priming againstthe endogenous tumor antigen and promotes CD8+ T cell effector function in SV40T antigen transgenic mice. J. Immunol. 171, 697–707 (2003).

72. H. S. Robins, P. V. Campregher, S. K. Srivastava, A. Wacher, C. J. Turtle, O. Kahsai,S. R. Riddell, E. H. Warren, C. S. Carlson, Comprehensive assessment of T-cell receptorb-chain diversity in ab T cells. Blood 114, 4099–4107 (2009).

AcknowledgmentsFunding: This work was supported by the NIH–National Cancer Institute (NCI) grants 1R01CA149405,1R01CA208246, and 1R01CA204021; by the Department of Defense–Prostate Cancer ResearchProgram award W81XWH-15-1-0406 (to J.D.W.); partly by the Flow Cytometry Core Facility SharedResource, Hollings Cancer Center, Medical University of South Carolina (P30 CA138313); and byCanCure LLC (1R41CA206688). Z.L. was funded by NIH–NCI (P01CA186866) (funding period: 01September 2015 to 31 August 2020). D.L., G.L., and K.F.S.-O. were supported by NIH–NCI grant

Zhang et al., Sci. Adv. 2017;3 : e1602133 17 May 2017

1R01CA164335-01A1. Author contributions: J.D.W. conceived the concept, directed all the studies,and prepared the manuscript. J.Z. performed all animal experiments and data analyses. D.L.,G.L., and K.F.S.-O. prepared the TCR-I CD8 T cells and assisted with the experiments. J.N.G.provided patient serum samples and associated clinical information. J.N.G. also contributedto manuscript preparation. Z.L. participated in the manuscript preparation. Competing interests:J.D.W. is on the scientific advisory board of CanCure LLC. All other authors declare that theyhave no competing interests. Data and materials availability: All data needed to evaluatethe conclusions in the paper are present in the paper and/or the Supplementary Materials.Additional data related to this paper may be requested from the authors.

Submitted 6 September 2016Accepted 9 March 2017Published 17 May 201710.1126/sciadv.1602133

Citation: J. Zhang, D. Liu, G. Li, K. F. Staveley-O’Carroll, J. N. Graff, Z. Li, J. D. Wu, Antibody-mediated neutralization of soluble MIC significantly enhances CTLA4 blockade therapy. Sci.Adv. 3, e1602133 (2017).

D

12 of 12

on May 17, 2017

http://advances.sciencemag.org/

ownloaded from

doi: 10.1126/sciadv.16021332017, 3:.Sci Adv 

Julie N. Graff, Zihai Li and Jennifer D. Wu (May 17, 2017)Jingyu Zhang, Dai Liu, Guangfu Li, Kevin F. Staveley-O'Carroll,enhances CTLA4 blockade therapyAntibody-mediated neutralization of soluble MIC significantly

this article is published is noted on the first page. This article is publisher under a Creative Commons license. The specific license under which

article, including for commercial purposes, provided you give proper attribution.licenses, you may freely distribute, adapt, or reuse theCC BY For articles published under

. hereAssociation for the Advancement of Science (AAAS). You may request permission by clicking for non-commerical purposes. Commercial use requires prior permission from the American

licenses, you may distribute, adapt, or reuse the articleCC BY-NC For articles published under

http://advances.sciencemag.org. (This information is current as of May 17, 2017):The following resources related to this article are available online at

http://advances.sciencemag.org/content/3/5/e1602133.fullonline version of this article at:

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

http://advances.sciencemag.org/content/suppl/2017/05/15/3.5.e1602133.DC1 can be found at: Supporting Online Material

http://advances.sciencemag.org/content/3/5/e1602133#BIBL 25 of which you can access for free at: cites 72 articles,This article

trademark of AAAS otherwise. AAAS is the exclusive licensee. The title Science Advances is a registeredYork Avenue NW, Washington, DC 20005. Copyright is held by the Authors unless stated

Newpublished by the American Association for the Advancement of Science (AAAS), 1200 (ISSN 2375-2548) publishes new articles weekly. The journal isScience Advances

on May 17, 2017

http://advances.sciencemag.org/

Dow

nloaded from