Delineation of an immunosuppressive gradient inhepatocellular carcinoma using high-dimensionalproteomic and transcriptomic analysesValerie Chewa,1, Liyun Laia, Lu Pana, Chun Jye Lima, Juntao Lia, Raymond Onga, Camillus Chuaa, Jing Yao Leonga,Kiat Hon Limb,c, Han Chong Tohc,d, Ser Yee Leec,d,e, Chung Yip Chanc,d,e, Brian K. P. Gohc,d,e, Alexander Chungc,d,e,Pierce K. H. Chowc,d,e, and Salvatore Albania

aSingHealth Translational Immunology and Inflammation Centre, Duke University–National University of Singapore Medical School, Singapore 169856;bDepartment of Anatomical Pathology, Singapore General Hospital, Singapore 169856; cDuke University–National University of Singapore Medical School,Singapore 169857; dNational Cancer Centre, Singapore 169610; and eDepartment of Hepatopancreatobiliary and Transplant Surgery, Singapore GeneralHospital, Singapore 169856

Edited by Dennis A. Carson, University of California, San Diego, La Jolla, CA, and approved June 13, 2017 (received for review April 20, 2017)

The recent development of immunotherapy as a cancer treatmenthas proved effective over recent years, but the precise dynamicsbetween the tumor microenvironment (TME), nontumor microenvi-ronment (NTME), and the systemic immune system remain elusive.Here, we interrogated these compartments in hepatocellular carci-noma (HCC) using high-dimensional proteomic and transcriptomicanalyses. By time-of-flight mass cytometry, we found that the TMEwas enriched in regulatory T cells (Tregs), tissue resident memoryCD8+ T cells (TRMs), resident natural killer cells (NKRs), and tumor-associated macrophages (TAMs). This finding was also validatedwith immunofluorescence staining on Foxp3+CD4+ and PD-1+

CD8+ T cells. Interestingly, Tregs and TRMs isolated from the TMEexpressed multiple markers for T-cell exhaustion, including PD-1,Lag-3, and Tim-3 compared with Tregs and TRMs isolated from theNTME. We found PD-1+ TRMs were the predominant T-cell subsetresponsive to anti–PD-1 treatment and significantly reduced in num-ber with increasing HCC tumor progression. Furthermore, T-bet wasidentified as a key transcription factor, negatively correlated withPD-1 expression on memory CD8+ T cells, and the PD-1:T-bet ratioincreased upon exposure to tumor antigens. Finally, transcriptomicanalysis of tumor and adjacent nontumor tissues identified a che-motactic gradient for recruitment of TAMs and NKRs via CXCR3/CXCL10 and CCR6/CCL20 pathways, respectively. Taken together,these data confirm the existence of an immunosuppressive gradientacross the TME, NTME, and peripheral blood in primary HCC thatmanipulates the activation status of tumor-infiltrating leukocytesand renders them immunocompromised against tumor cells. By un-derstanding the immunologic composition of this gradient, moreeffective immunotherapeutics for HCC may be designed.

CyTOF | tumor microenvironment | regulatory T cells | resident memoryT cells | hepatocellular carcinoma

The immune system has a critical role in determining cancerpathogenesis and clinical fate (1, 2), and breakthroughs in the

development of cancer immunotherapeutics have revolutionizedthe way we view and treat cancers (3). Numerous clinical trialson cancer immunotherapeutics are ongoing—particularly thoseassessing the efficacy of immune-checkpoint inhibitors (4–7).Despite encouraging results from clinical trials, fundamentaldifferences in responsiveness both between different cancertypes and within the same cancer type have been reported (8).The majority of research performed to date has focused on de-finitive cellular populations or individual molecules, and al-though this approach has been informative, it has not providedus with the knowledge as to which relevant immune functions arepivotal in certain cancers compared with others.Hepatocellular carcinoma (HCC) is one of the most common

cancers among males and the second most common cause of cancer-associated deaths globally (9). Limited treatment options and a high

mortality rate pose a pressing need for the discovery of new thera-peutic interventions (10). To date, immunotherapeutics based onimmune-checkpoint inhibitors have generated promising results insome patients with HCC, but the majority do not respond well totreatment (11). Several previous studies on the tumor microenvi-ronment (TME) have shown its crucial role in tumor progressionand disease prognosis (12–15), but furthering our understanding ofthe cancer-immune landscape remains a challenge.In this study, we hypothesized that mechanistically and clinically

relevant immune signatures are profoundly shaped by the micro-environment in which immune cells reside, and that interactionsbetween the microenvironment and immune cells are dynamicalong a “cancer-immune gradient” that encompasses the TME,the nontumor microenvironment (NTME), and the peripheralblood (PB). Certain immune subsets [such as memory CD8+

T and T-regulatory (Treg) cells] may assume different phenotypesand functions depending on the microenvironment that they in-filtrate and, therefore these differences may be exploited in dif-ferent pharmacologic ways and in a tissue context-dependentmanner. By determining the cancer-immune gradient for partic-ular cancers, such as HCC, we can develop novel therapeutics that

Significance

The roles of tumor-infiltrating leukocytes in mediating cancerprogression are well recognized, but a multidimensional anal-ysis of the entire cancer immune system is lacking. Here, wedissected the cancer-immune landscape in hepatocellular car-cinoma (HCC) across tumor, nontumor, and peripheral bloodcells using time-of-flight mass cytometry, multiplex immuno-fluorescence tissue staining, and NanoString analysis. Weidentified various immune subsets that were enriched in thetumor microenvironment and their potential impact on thetumor immunity based on their detailed phenotypes. Thisstudy has validated the concept of a cancer-immune gradientand demonstrated in primary HCC that immune-cell subsetsbecome progressively suppressive as they traverse the non-tumor to tumor microenvironment. These data have openedavenues for the design of immunotherapeutics in HCC.

Author contributions: V.C. and S.A. designed research; V.C., L.L., L.P., C.J.L., R.O., C.C.,J.Y.L., K.H.L., H.C.T., S.Y.L., C.Y.C., B.K.P.G., A.C., and P.K.H.C. performed research; V.C.,L.L., L.P., J.L., C.C., and P.K.H.C. analyzed data; and V.C. and S.A. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1To whom correspondence should be addressed. Email: valerie.chew.s.p@singhealth.com.sg.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1706559114/-/DCSupplemental.

E5900–E5909 | PNAS | Published online July 3, 2017 www.pnas.org/cgi/doi/10.1073/pnas.1706559114

Dow

nloa

ded

by g

uest

on

May

17,

202

0

more precisely exploit the key pathologic molecules specific tothe TME.To accurately delineate the cancer-immune gradient in HCC, we

first compared tumor-infiltrating leukocytes (TILs) to nontumorinfiltrating leukocytes (NILs) and PB mononuclear cells (PBMCs)using time-of-flight mass cytometry (CyTOF). We identified mul-tiple TIL subsets with increasingly suppressive phenotypes as theyadvanced from the PB and NTME to the TME. This was alsovalidated with multiplex immunofluorescence tissue staining ontwo important immunosuppressive subsets: Foxp3+CD4+ Treg andPD-1+ CD8+ T cells. Specifically, tumor-infiltrating memory CD8+

T cells [resident memory (TRM) and effector memory (TEM)]showed high expression of multiple exhaustion makers, particularly,programmed death 1 (PD-1), an immune-exhaustion marker that isknown to permit cancer immune evasion (16). Our in vitro studiesfound that these cells were the key subsets responsive to anti–PD-1(monoclonal antibody) treatment. Elevated expression of PD-1 byTRMs was associated with down-regulation of the transcriptionfactor T-bet upon increasing tumor-antigen exposure. Finally, weexamined the general immune gene expression profile of the TMEversus the NTME by transcriptomic analysis of 800 pan-cancerimmune genes. By this method, we found that the TME hostschemokines that likely recruit exhausted and potentially immuno-suppressive TIL subsets.Taken together, this unique in-depth understanding of the in-

terface between systemic and tumor-specific immunity in HCC basedon the combination of high-dimensional proteomic and tran-scriptomic analyses has identified axes as to how the TME mightshape the composition and activation status of TILs. These findingswill open up new avenues for future immunotherapeutic design.

MethodsPatients. Tumor and adjacent nontumor liver tissues and PBMCs were collectedfrom 54 patients with HCC who underwent curative resection at SingaporeGeneral Hospital. Patient consent was obtained according to the guidelines ofthe SingHealth Central Institutional Review Board. Fresh samples wereobtained from 28 patients where TILs or NILs were freshly isolated using en-zymatic digestion with 100 units/mL collagen IV (Thermo Fisher Scientific) for30 min at 37 °C as previously described (17) and PBMCs were isolated frompreoperative blood samples collected on the same day of surgical resection bylayering over Ficoll-Paque Plus (GE Healthcare), as previously described (12).Fresh patient samples (n = 28) were selected for subsequent analyses based onimmune-cell availability and as far as possible, equal distribution of viral statusand tumor staging. Full clinical and demographic information of the patientsand samples used for various experiments is summarized in SI Appendix, TableS1. Matched tumor and nontumor formalin-fixed paraffin-embedded (FFPE)tissue sections were obtained from an additional 26 HCC patients (SI Appendix,Table S2) for validation of CyTOF data using multiplex immunofluorescencetissue staining as described below.

CyTOF. TILs, PBMCs, and NILs were obtained from patients with HCC (n = 14, 14,and 7, respectively) as described above, and processed for CyTOF analysis. Apanel of 35 antibodies that encompassed a broad range of immune subsets wasused (SI Appendix, Table S3) together with a combination of three anti-CD45 antibody barcodes for simultaneous sample processing, as previously de-scribed (18). The antibodies were either conjugated in-house according to themanufacturer’s instructions (Fluidigm) or purchased preconjugated directly fromthe supplier (Fluidigm). The cells were thawed and rested overnight in completeRPMI medium supplemented with 10% FBS, 1% penicillin/streptomycin/gluta-mine, and 10 mM Hepes at 37 °C. The cells were then either stimulated for 6 hwith 150 ng/mL PMA (phorbol myristate acetate; Sigma) and 100 ng/mL ion-omycin (Sigma) or unstimulated and exposed to 3 μg/mL Brefeldin A (eBiosience)and 2 μM monesin (BioLegend) during the final 4 h of the incubation. Next, thecells were stained with cisplatin (Fludigm) to identify live/dead cells and in-cubated with metal-conjugated surface-membrane antibodies. The cells werethen fixed in 1.6% paraformaldehyde and permeablized in 100% methanol topermit staining with intracellular metal-conjugated antibodies. Finally, the cellswere labeled with an iridium-containing DNA intercalator before analysis on aCyTOF-II mass cytometer (Fluidigm). The signal was bead normalized using EQFour Element Calibration Beads (EQ Beads, 201078, Fluidigm) according tomanufacturer’s instructions (19). The generated files underwent auto debarcoding

and filtering for live/dead cells and DNA using R Studio (version 0.98.1073). Thefiles were then analyzed using in-house enhanced automatic classification ofcellular expression by nonlinear stochastic embedding (ACCENSE) software basedon the combination of Barnes–Hut stochastic neighbor embedding (SNE) non-linear dimension reduction algorithm and a k-means clustering algorithm (20)after down-sampling to 10,000 CD45+ immune cells from each sample for faircomparison across samples. Nodes that were significantly enriched in TILs, NILs,or PBMCs were identified using the Kruskal–Wallis rank sum test. Some of thedata were independently validated using FlowJo software (version 10.0.7, TreeStar). Both 2D and 3D heat maps were plotted based on all significant nodesusing R Studio for data visualization.

Multiplex Immunofluorescence Tissue Staining by Vectra. FFPE sections from 26HCC patients, who underwent curative resection from 1991 to 2009, wereobtained from the Department of Anatomical Pathology, Division of Pathol-ogy, Singapore General Hospital (SI Appendix, Table S2). Two cores of tumorand nontumor FFPE tissues from each patient were stained with Opal Multi-plex Immunohistochemistry Detection Kit and images were acquired using aVectra 3.0 Pathology Imaging System Microscope (Perkin-Elmer) as describedpreviously (21). Antibodies used were: anti-CD8 (DAKO, clone C8/144B), anti-CD4 (Abcam, clone EPR6855), anti–PD-1 (Abcam, clone NAT105), and anti-Foxp3 (Abcam, clone 236A/E7). Detection dye for each antibody was:Opal690 dye (CD8), Opal540 dye (CD4), Opal650 dye (PD-1), and Opal520 dye(Foxp3). DAPI were used as a nuclear counterstain. Quantification of positivelystained cells was performed on the whole 1-mm core (area = 0.785 mm2) andthe average values of two cores from each patient sample were calculated andshown as the number of cells per square millimeter.

In Vitro Cellular Coculture and Cytokine Production Assay. For detection of theT-bet:PD-1 ratio in PBMCs upon exposure to tumor antigens, autologoustumor-dissociated cells and PBMCs from eight HCC specimens were thawedand rested overnight. A total of 0.5 × 106 dissociated tumor cells (repre-senting the tumor antigen) or PBMCs (as a control) were then subjected toX-ray irradiation (RS-2000, Radsource) at 200 Gy and PBMCs (representingfeeder cells) were irradiated at 30 Gy, as previously described (22). The ir-radiated cells were mixed with 1 × 106 nonirradiated PBMCs (n = 8) andanalyzed for the T-bet:PD-1 ratio on days 0, 1, 2, 3, and 6 of coculture byflow cytometry (BD LSRFortessa X-20, BD Biosciences). The cells werestained with anti-human CD45, CD3, CD4, CD8, CD45RO, T-bet, CD103, PD-1(BioLegend), CD56 (BD Biosciences), and blue-fluorescent reactive dye (LifeTechnologies) for live/dead cell staining.

To examine the response of TILs to anti–PD-1 treatment in vitro, similarlyirradiated TILs and PBMCs were mixed with 1 × 106 nonirradiated TILs thathad been exposed to 10 μg/mL anti-human PD-1 antibody (clone J116,eBioscience) or mouse IgG1, κ-isotype control (eBioscience) for 18 h (n = 8).The cells were exposed to Brefeldin A and monesin (BioLegend) during thefinal 4 h of the incubation. Production of TNFα and IFNγ (BioLegend), cell-surface marker expression and live/dead cell staining (as described above) innonirradiated TILs was measured by flow cytometry (BD LSRFortessa X-20,BD Biosciences). Postacquisition analysis was performed for both experi-ments using FlowJo software (version X.0.7, Tree Star).

NanoString Analysis. Unsorted tumor and adjacent nontumor liver tissueswere collected upon resection from 20 patients with HCC. Tissues were lyzedusing Tissuelyzer-II (Qiagen) and the lysates were processed using the mir-Vana miRNA Isolation Kit (Ambion) to extract total RNA. Quantification ofRNA was performed using the Quant-iT RiboGreenRNA assay kit (Invitrogen).The samples were then analyzed using the nCounter platform (NanoStringTechnologies) using a combination of the prebuilt nCounter PanCancer Im-mune Profiling code set (consisting of 770 genes) and an additional 30 cus-tomized compatible probes from the nCounter Panel-Plus code set. The rawdata were normalized against housekeeping genes using the Nano-StringNorm R package and log transformed, before analyzing by Student’spaired t test and generating heat maps for visualization.

Statistical Analyses. For CyTOF data, the nonparametric Kruskal–Wallis rank sumtest was used to identify nodes that were differentially present among threegroups and the Student’s paired t test was used to analyze NanoString data toidentify significantly expressed genes. Statistical analyses of the FlowJo data wereperformed using a paired or unpaired Student’s t test, unpairedMann–Whitney utest, and Spearson’s correlation test, as indicated (GraphPad Prism V.6.0f).

Chew et al. PNAS | Published online July 3, 2017 | E5901

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

May

17,

202

0

ResultsTME, NTME, and PB Exhibit a Unique Immune Landscape Defined bySpecific Immune Subsets. We first designed a CyTOF assay thatwould permit a deep interrogation of the TME, NTME, and PBimmune landscapes in tissues/blood extracted from patients withHCC (Fig. 1A). To this aim, we designed a panel of 35 surfacemembrane and intracellular immune markers that encompasseda broad range of immune subsets (SI Appendix, Table S3) toexamine the global phenotypes of TILs, NILs, and PBMCs frompatients (n = 7) with HCC (SI Appendix, Table S1). The cellswere prestimulated for 6 h with PMA and ionomycin, as thedifferences between the TME, NTME, and PB compartmentswere enhanced upon activation.Data files generated from the CyTOF analysis for individual

samples were subjected to a dimension reduction process based onthe Barnes–Hut SNE algorithm and then followed by cell clus-tering based on the k-means clustering algorithm using in-houseenhanced ACCENSE software (20). Following this step, the cells

formed distinct clusters (representing immune subsets) based onsimilarities in immune-marker expression. These nodes were fur-ther analyzed by Kruskal–Wallis rank sum test to identify thosethat were statistically different among the three cellular compart-ments (combined node frequency >70% in one of the compart-ments at P < 0.05) (Fig. 1A). The expressions of selective markerswere shown with 2D t-SNE plots for lineage markers, such as CD8,CD4, CD56, and CD14, as well as exhaustion markers PD-L1, PD-1, CTLA-4/CD152, Tim-3, and Lag-3 (Fig. 1B). Differences innode frequencies can be appreciated across three different com-partments (SI Appendix, Fig. S1A). With the statistical analysis, weidentified nodes that showed distinct distributions across TIL, NIL,and PBMC compartments (Fig. 1C). These data imply that theimmune landscapes of the TME, NTME, and PB in HCC areunique in terms of their cellular composition and would thereforerespond differentially to immunotherapeutic intervention.By this unbiased and unsupervised method, 15 TIL, eight NIL,

and 11 PBMC-enriched nodes were identified, and three and two

B

PBM

CN

ILTI

L

C

Bar

nes-

Hut

SN

E d

im.1

Barnes-Hut SNE dim.2

A

Blood

Tumor / Adjacent Normal

35 immune markers

CyTOF

Barnes-Hut SNE

Dimension reduction &

k-means clustering

Statistical test

Identification of unique nodes

Generation of nodes

Normalized intensity

CD8 CD56 3-miT4-ALTC1-DP4DC PD-L1CD14 Lag-3

PBMCNIL

TIL

Fig. 1. High-dimensional analysis with Barnes–Hut SNE identified differentially enriched immune subsets in TILs, NILs, and PBMCs. (A) Time-of-flight masscytometry (CyTOF) pipeline from data acquisition, dimension reduction, and clustering to node generation. The resulting nodes, clustered by similarity in theirimmune phenotypes, were subjected to statistical testing to identify significantly enriched nodes from a given group (TIL, NIL, or PBMC). (B) The 2D cellulart-SNE plots of CyTOF data from PBMCs, NILs, and TILs as gated on: CD8, CD4, CD56, CD14, PD-L1, PD-1, CTLA-4, Tim-3, and Lag-3. Each dot represents onesingle cell. Arrows showed distinct differences in TILs. (C) A 3D illustration of node percentages in either PBMC, NIL, or TIL compartments as grouped intonode ID and major immune subsets. n = 7 from each compartment.

E5902 | www.pnas.org/cgi/doi/10.1073/pnas.1706559114 Chew et al.

Dow

nloa

ded

by g

uest

on

May

17,

202

0

nodes were enriched in both TIL/NIL and NIL/PBMC, re-spectively (SI Appendix, Fig. S1B). We further analyzed the full35-immune marker expression profile of each of thestatistically significant nodes as identified above in a 2D heatmap to identify phenotypic differences or similarities betweenthe nodes (Fig. 2A). Generally, we found that the immunesubsets within the TILs and NILs were phenotypically moresimilar to each other than the immune subsets derived fromthe PBMCs. Classification of these phenotypically similarnodes identified the immune subsets that were enriched ineach compartment. TIL-enriched subsets were identified as:TRMs that were CD103+CD45RO+CD8+PD-1+CD152/CTLA-4+Lag-3+Tim-3+; TEMs that were PD-1+Lag-3+CD152+Tim-3+

CD45RO+CCR7−; Tregs that were CD4+Foxp-3+CD152+IL10+CD45RO+PD-1+; and CXCR3+ natural killer (NK) cellsthat were either resident (CD103+) NKR cells or nonresident(CD103−) NK cells that were GzmB−. The NIL-enriched subsetswere: PD-1− TEM; GzmB+ NK cells; and CD8+CD56+ NKT cells.The PBMC-enriched subsets were: naive CD8+ T cells andCD45RA+CD4+ T cells.Binary analysis of the resting or unstimulated cells confirmed

the composition of the TIL-enriched immune subsets describedabove and identified an additional TIL-enriched population:CD14+HLA-DR+PD-L1+Lag-3+ tumor-associated macrophages(TAMs). These cells were largely eliminated upon PMA andionomycin stimulation (SI Appendix, Fig. S3A). These data werevalidated by manual gating using FlowJo for these specific

immune subsets in the three compartments with additional TILand PBMC samples available (SI Appendix, Fig. S2A). Of note, theoverall lower number of NIL cases was due to the limited avail-ability of large enough surgical nontumor tissue from which im-mune cells could be isolated for CyTOF analysis. Indeed, weobserved that Tregs, NKRs, and TAMs were significantly enrichedin TILs, whereas TRMs were enriched in both TILs and NILs (Fig.2B and SI Appendix, Fig. S3B). Data validation was performedusing multiplex immunofluorescence tissue staining on an ad-ditional 26 paired tumor and nontumor FFPE samples (SIAppendix, Table S2) and demonstrated that indeed the num-ber of Foxp3+CD4+ Tregs and PD-1+CD8+ T cells per area washigher in tumor versus nontumor compartments (Fig. 2 C and D).

Tregs Assume a More Immunosuppressive Phenotype When in Contactwith the TME. We next aimed to address whether the cellular sub-sets enriched in the TILs exhibited divergent phenotypes when ex-posed to different microenvironments. First, we focused on Tregs,given their well-established correlation with poor prognosis in vari-ous cancers, including HCC (23–25). Recent studies have alsodemonstrated the expression of exhaustion markers on Tregs, inparticular PD-1, and their effect on Treg function (26, 27). We thenexamined whether the Tregs expressed different immune exhaustionmarkers between the TIL, NIL, and PBMC groups. Interestingly, weobserved that a higher percentage of Tregs expressed multiple ex-haustion markers in those derived from the TIL compared with theNIL and PBMC compartments: PD-1 (TIL, 40.7 ± 16.1%; NIL,

Fig. 2. High-dimensional Barnes–Hut SNE analysis defines the distribution of specific immune subsets in TILs, NILs, and PBMCs. (A) A 2D heat map showingdifferential expression of 35 immune markers by TIL-enriched (red bar), NIL-enriched (blue bar), PBMC-enriched (green bar), TIL/NIL-enriched (purple bar), orNIL/PBMC-enriched (orange bar) nodes. Immune subsets were categorized based on their marker expression. (B) Percentage of Tregs, TRMs, NKRs, and TAMs inPBMCs (n = 12), TILs (n = 12), or NILs (n = 7). Data represent the means ± SD and were analyzed by paired Student’s t test, **P < 0.01 and *P < 0.05.(C) Representative images from multiplex immunofluorescence tissue staining for CD8 (green), PD-1 (red), CD4 (magenta), Foxp3 (white), and DAPI (blue) ontumor and nontumor FFPE tissues. (Scale bar, 20 μm.) (D) Quantification of the number of Foxp3+CD4+ Treg and PD-1+CD8+ T cells per square millimeter intumor versus nontumor tissues from n = 26 paired HCC samples. Paired Student’s t test. **P < 0.01 and ***P < 0.001.

Chew et al. PNAS | Published online July 3, 2017 | E5903

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

May

17,

202

0

22.5 ± 6.9%; and PBMC, 20.0 ± 9.8%), Lag-3 (TIL, 23.7 ± 6.6%;NIL, 12.7± 3.9%; and PBMC, 10.6± 6.5%), and Tim-3 (TIL, 27.1 ±12.7; NIL, 11.3 ± 3.1%; and PBMC, 9.5 ± 6.7%) (Fig. 3A). In ad-dition, we found that upon stimulation with PMA and ionomycin,

Tregs from the TILs expressed significantly more IL-10 than Tregsfrom NILs or PBMCs (Fig. 3 B and C). Representative plots of IL-10 expression in Treg populations of the TILs, NILs, and PBMCsclearly demonstrated a higher percentage of IL-10+ Tregs derived

Fig. 3. The TME is enriched with more exhausted Tregs compared with the NTME. (A) Percentage of PD-1–, Lag-3–, or Tim-3–expressing Treg cells from TILsversus NILs or PBMCs. ***P < 0.001, **P < 0.01; *P < 0.05, paired Student’s t test. (B) Percentage of IL10+ Tregs from TILs, NILs, or PBMCs with or without 6-hPMA and ionomycin stimulation. **P < 0.01, paired Student’s t test. (C) Representative plots showing expression of IL-10 pregated on CD4+Foxp3+CTLA-4+

Tregs in TILs versus NILs and PBMCs from one HCC patient HEP178. (A–C) TILs (n = 12) versus NILs (n = 7–8) and PBMCs (n = 12). Data represent the means ± SD.(D) Correlation between expression of exhaustion markers: PD-1, Lag-3, and Tim-3 with IL-10 on Tregs from TILs and NILs. n = 21. P values and correlationcoefficients (r) were calculated with Spearman’s correlation test. ***P < 0.001 and **P < 0.01.

Fig. 4. Exhaustion marker and cytokine expression by memory CD8+ from TILs, NILs, and PBMCs. (A) Percentage of PD-1–, CTLA-4–, Lag-3–, or Tim-3–expressingCD8+CD45RO+CD103− TEMs and CD8+CD45RO+CD103+ TRMs in TILs (n = 12) versus NILs (n = 7) or PBMCs (n = 12). Data represent the means ± SD and wereanalyzed by paired Student’s t test, ***P < 0.001, **P < 0.01, and *P < 0.05. (B) A 2D heat map representing the expression of exhaustion markers by TEM (red line)or TRM (blue line) nodes in TILs or NILs. (C) Percentage of TNFα and IFNγ-expressing TEMs and TRMs in TILs (n = 12) versus NILs (n = 5–7) or PBMCs (n = 12) with orwithout 6-h PMA and ionomycin stimulation. Data represent the means ± SD and were analyzed by paired Student’s t test, *P < 0.05. (D) Correlation betweenpercentage of PD-1+, CTLA-4+, Lag-3+, or Tim-3+ TEMs and TRMs versus TNFα+IFNγ+ TEMs and TRMs in TILs and NILs upon 6-h PMA and ionomycin stimulation. n =19 each. P values and correlation coefficients (r) were calculated with Spearman’s correlation test. ***P < 0.001, **P < 0.01, and *P < 0.05.

E5904 | www.pnas.org/cgi/doi/10.1073/pnas.1706559114 Chew et al.

Dow

nloa

ded

by g

uest

on

May

17,

202

0

from the TILs upon stimulation (Fig. 3C). The suppressive pheno-type of Tregs from TILs was also demonstrated by its ability tosuppress T-cell proliferation (SI Appendix, Fig. S4).Finally, we examined whether high IL-10 expression (indicative of

a suppressive phenotype) was associated with exhaustion-markerexpression. Indeed, we observed that the expression of the exhaus-tion markers PD-1, Lag-3, and Tim-3 correlated with the expressionof IL-10 in Tregs (Fig. 3D). These data support the concept of animmune-cancer gradient, whereby the same cell type (i.e., Tregs) canassume different phenotypes and functions, depending on their re-siding microenvironment. Specifically, we found that Tregs exhibitedan increased immunosuppressive behavior in the TME.

CD8+ TRMs and TEMs Exhibit an Enhanced T-Cell–Exhausted State in theTME. We next explored the cancer-immune gradient between theTME and the NTME and focused on analyzing their commonT-cell subsets, TEM and TRM. We examined the expression ofmultiple exhaustion markers on these immune subsets by manualgating for PD-1, CTLA-4, Lag-3, and Tim-3 and observed a sig-nificantly higher percentage of PD-1+CTLA-4+ and Lag-3+ CD8+

TEM and TRM cells in the TILs compared with the NILs (Fig. 4A).Conversely, a significantly higher percentage of Tim-3+ cells wasonly observed in TEMs from the TILs versus NILs (Fig. 4A). A 2Dheat map illustration of the expression level of exhaustion markerson TEM or TRM cells also showed a generally higher expression ofthese markers from those infiltrating TME versus NTME (Fig. 4B).We next examined the responsiveness of the TEM and TRM

cells expressing exhaustion markers to ex vivo immune activationwith PMA and ionomycin. Upon activation, a lower percentageof TNFα+IFNγ++ TEM and TRM cells were identified in the TILs

compared with the NILs (Fig. 4C). We then examined the asso-ciation between the expression of exhaustion markers with theexpression of TNFα and IFNγ on these cells. Indeed, we observedan inverse correlation between the expressions of PD-1, CTLA-4,and Lag-3 with the expression of TNFα+IFNγ++ on TEM and TRMcells upon PMA and ionomycin stimulation (Fig. 4D), indicating alink between the expression of these exhaustion markers and thefunctional competency of these cells. The correlation between theexpression of Tim-3 and the cytokines produced by either TEM orTRM cells was, however, not significant (Fig. 4D), perhaps sug-gesting a less critical role for Tim-3 on the functional competencyof these cells. These data further strengthen the hypothesis of acancer-immune gradient and demonstrate the generation of anincreasingly immunosuppressive phenotype from the PB to theTME in HCC. Further, these data demonstrate that exhaustionmarkers are widely expressed by multiple T-cell subsets enrichedin the TME, including Tregs, CD8+ TRMs, and TEMs.

PD-1–Expressing CD45RO+CD8+ TILs in the TME Are Responsive toAnti–PD-1 Treatment and Mediate Tumor Progression. An immedi-ate question arising from these data is how two distinct andspecific adjacent microenvironments (TME and NTME) canshape T-cell function. To address this question, we decided toanalyze the possible link between the expression of the tran-scription factor T-bet and PD-1 on TEMs and TRMs from theTILs and NILs, as previous data had found that T-bet can sup-press PD-1 expression in mice (28, 29). In our human HCCsamples, we observed a higher percentage of T-bet−-expressingTEMs and TRMs isolated from the NILs compared with the TILs(Fig. 5A). An inverse correlation between T-bet and PD-1 was

Fig. 5. T-bet is the critical transcription factor that correlates with down-regulation of PD-1 upon tumor antigen exposure and tumor progression.(A) Percentage of T-bet–expressing TEMs and in TILs (n = 12) versus NILs (n = 7). Data represent the means ± SD and were analyzed by paired Student’s t test.*P < 0.05. (B) Correlation between percentage of PD-1+ TEMs and TRMs versus T-bet

+ TEMs and TRMs in TILs and NILs upon 6-h PMA and ionomycin stimulation.n = 19 each, P values and correlation coefficients (r) were calculated with Spearman’s correlation test. **P < 0.01. (C) Ratios of percentages of PD-1+ versusT-bet+ on CD8+CD45RO+ TEM cells from PBMCs upon coculture with irradiated autologous tumor (n = 8) or PBMC (n = 6) and feeder cells. Data represent themeans ± SD and were analyzed by paired Student’s t test with reference to day 0. **P < 0.01 and *P < 0.05. (D) Percentage of PD-1+ or T-bet+ TEMs and TRMs instage 1 versus stages 2–4. Data represent the means ± SD and were analyzed by unpaired Student’s t test. n = 10 each. ***P < 0.001 and **P < 0.01.(E) Percentage of TNFα-expressing TEMs and TRMs from TILs after 18 h cocultured with autologous irradiated tumor cells with anti–PD-1 antibody (α–PD-1) orwith isotype control mouse IgG1,κ-antibody (Ctl). n = 8. Data represent the means ± SD and were analyzed by paired Student’s t test with reference to control.**P < 0.01 and *P < 0.05. (F) Percentage of PD-1+ or T-bet+ TEMs and TRMs in HBV-infected (n = 12) versus nonvirally infected (n = 10) HCC patients. Datarepresent the means ± SD. P values were calculated using unpaired Student’s t test. *P < 0.05.

Chew et al. PNAS | Published online July 3, 2017 | E5905

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

May

17,

202

0

also found in the same cells from both the TILs and NILs (Fig.5B). These data suggest that T-bet may control PD-1 expressionwhen exposed to the TME.To test this hypothesis, we cocultured PBMCs from patients

with HCC that exhibited low PD-1:T-bet ratio (at day 0, Fig. 5C)with irradiated autologous tumor samples (as a source of tumorantigens) and feeder cells, as previously described (22). After48-h exposure to the irradiated tumor sample, the ratio ofPD-1 versus T-bet expression on TEM cells from the PBMCs wasenhanced compared with day 0, and this ratio was maintaineduntil day 6 (Fig. 5C). Of note, the same experiment performedusing irradiated autologous PBMCs showed no increase in

PD-1/T-bet ratio, demonstrating the specificity of this up-regulation of PD-1 when exposed to tumor antigen (Fig. 5C).We hypothesized that this phenomenon may have a clinicallyrelevant implication on tumor progression and thus analyzed thepercentage of PD-1+ and T-bet+ CD8+CD45RO+ cells in earlystage (stage 1) versus advanced stage (stage ≥2) tumors. Strik-ingly, we observed that the percentages of PD-1+ memory CD8+

T cells were significantly increased in the later-staged HCC tu-mors compared with the early-staged tumors, whereas the per-centages of T-bet+ memory CD8+ T cells were significantlydecreased in the later-staged tumors (Fig. 5D). Importantly,these cellular subsets from the TILs were also responsive to

Fig. 6. The tumor microenvironment shapes and dictates TIL infiltration in HCC. (A) Heat map showing NanoString analysis data from unsorted tumor versusadjacent nontumor tissues. n = 20 paired tissues. Box shows enriched genes in TME versus NTME with the magnified image on the Right. (B) Expression ofCCR6 and CXCR3 on Tregs (CD4+Foxp3+), TEMs (CD8

+CD45RO+), NKs (CD56+), or monocytes (CD14+HLA-DR+) from PBMCs of HCC patients. n = 6–10. Data rep-resent the means ± SD. (C) Correlation of RNA expression of CCL20 to percentage of TAMs, and CXCL10 to percentage of NKRs in TILs. n = 12. P values andcorrelation coefficients (r) were calculated with Spearman’s correlation test. *P < 0.05 and **P < 0.01. (D) Percentage of IL-10+ TAMs and granzyme B (GB)+ NKsfrom TILs (n = 12) versus NILs (n = 5–7) or PBMCs (n = 12) with or without 6-h PMA and ionomycin stimulation. Data represent the means ± SD and were analyzedby paired Student’s t test. *P < 0.05 and **P < 0.01. (E) Working hypothesis model showing tumor-enriched chemokines: CCL20 attracting CCR6+ TAMs, whichproduce high levels of IL-10; whereas CXCL10 attracts CXCR3+ NKRs that express low levels of granzyme B (GB) within the tumor site. Lower expression level ofT-bet, and a reciprocal higher expression level of PD-1, and lower production of TNFα and IFNγ was observed on TEMs and TRMs infiltrating the tumor.

E5906 | www.pnas.org/cgi/doi/10.1073/pnas.1706559114 Chew et al.

Dow

nloa

ded

by g

uest

on

May

17,

202

0

anti–PD-1 treatment, as determined by the production of TNFαupon anti–PD-1 treatment in vitro (Fig. 5E).We also explored the balance between T-bet and PD-1 expres-

sion in patients with HCC and concomitant chronic hepatitis B virus(HBV) infection. Higher expression of PD-1 and correspondinglylower expression of T-bet was detected in patients infected withHBV compared with uninfected patients (Fig. 5F), indicating anenhanced level of TIL exhaustion in chronic HBV infection.Overall, these in vitro data directly indicate that these describedTIL subsets (PD-1+CD8+ TRMs and TEMs) are the primary tar-get for anti–PD-1 immunotherapy, which has important impli-cations for the design of anti–PD-1 treatments in HCC.

TME Shapes the Immune Landscape and Determines the Immune SubsetsThat Accumulate in the Tumor. Our data so far have shown that theTME is infiltrated with phenotypically more immunosuppressive/exhausted T cells, compared with the NTME and PB, thus in-dicating the critical role of the TME in shaping the phenotypes ofTILs. To further support the crucial role of TME in shaping theimmune landscape of HCC, we analyzed the RNA expression of800 cancer-immune genes from unsorted tumor and nontumorsamples using NanoString analysis. We found overall an higherexpression of cancer-associated genes (including BIRC5, CDK1,TTK, ITGA6, PBK, and SPA17, and melanoma-associated antigengenes) in the TME compared with the NTME, which validated thepurity of our samples (Fig. 6A). We also found that the TMEexpressed lower levels of multiple inflammatory cytokines andchemokines (including IL1B, IL6, IL8, TNF, GZMH, GZMK,CXCL12, CXCL14, CXCL1, CXCL16, CCL21, and CCL19) (Fig.6A), which is indicative of a more immunosuppressive TME.Several chemokines (including CCL20, CXCL10, and CXCL11)

were highly expressed in the TME compared with the NTME (Fig.6A, arrows). CCL20 is of particular interest as it is a ligand forCCR6 that is known to be expressed by Treg (30), CD8+ TEM (31,32), Th-17 cells (33), B cells, immature dendritic cells (DCs) (34),and tumor-promoting macrophages (35). We also observed highCD200 expression levels in the TME (Fig. 6A, arrow), which hasknown roles in tumor immune evasion via interactions with TAMs(36) and in suppressing antitumor immunity (37).Given the enrichment of CCL20, CXCL10, and CXCL11

chemokines in the TME, we examined the protein expression oftheir respective receptors—CCR6 for CCL20, CXCR3 forCXCL10, and CXCL11 on the TILs. First, we examined by flowcytometry the protein expression of CCR6 and CXCR3 onPBMCs from patients with HCC, based on the hypothesis thatthese PBMC subsets were the potential target for chemotaxisalong the chemokine-gradient toward the TME. A higherproportion of CCR6+ macrophages/monocytes (defined asCD14+HLA-DR+ cells) and CXCR3+CD56+ NK cells were ob-served compared with other immune subsets (Fig. 6B). By com-bining the chemokine expression level with the RNA expressionanalysis and CyTOF data, we identified a correlation between theRNA expression level of CCL20 with the percentage of intra-tumoral TAMs and the expression level of CXCL10 with thepercentage of intratumoral NKRs (Fig. 6C). In addition, TIL-residing NK cells expressed a lower level of granzyme B and theTAMs expressed higher levels of IL-10 (Fig. 6D). These data againindicate that tumor-enriched chemokines attract more suppressiveimmune subsets to the TME.Taking all these data together, we propose that the primary HCC

immune landscape might hinge on the following: two key chemo-kines, CCL20 and CXCL10, that may induce the accumulation oftwo immune-cell subsets, TAM and NKR, which express high levelsof IL-10 and low levels of granzyme B in the TME (Fig. 6E); andthe transcription factor T-bet, which may be the coordinating factorcontrolling the expression of PD-1 and the corresponding pro-duction of TNFα and IFNγ by TEM and TRM cells along an im-munosuppressive gradient from NILs to TILs (Fig. 6E).

DiscussionThe vital contribution of the immune surveillance system to cancerprogression has been swiftly translated to the clinical setting with theintroduction of immunotherapeutics to treat various cancers (3, 4).However, the marked variability in patient (and cancer-type) re-sponsiveness to such treatments, including in HCC, renders the ur-gent need for an advanced understanding of the interaction betweenthe host immune system and tumor-cell survival (5). In this study, wedissected the immune compartments within the TME, NTME, andPB in patients with HCC to define the cancer-immune gradient anddelineate the phenotypes and functions of the infiltrating immunesubsets. This holistic approach permitted a broad understanding ofHCC as a systemic disease whereby the immune response is differ-entially manipulated according to various factors (such as exposureto tumor antigens and differential expression of intratumoral che-mokines) present in the cancer and noncancer microenvironments.This study uses CyTOF for deep phenotyping of the immune-

cell subsets that accumulate in the TME, NTME, and PB inpatients with HCC. We found that the cellular subsets that repre-sented the tumor tissue exhibited an exhausted and immunosup-pressive phenotype, as evidenced by the enhanced expression ofmultiple exhaustion markers (PD-1, CD152, and Lag-3), accumu-lation of IL-10–expressing immunosuppressive Treg and TAMsubsets, and low expression of inflammatory cytokines (TNFα,IFNγ, and granzyme B) by tumor-infiltrating T and NK cells. Therole of exhaustion markers (particularly PD-1) on Treg functionremains controversial, as it was demonstrated in malignant gliomasthat PD-1hi Treg was dysfunctional and produced IFNγ (26). Othershave described that PD-1 expression is correlated with more sup-pressive Treg phenotypes in chronic viral infection (27). Our data,however, have clearly demonstrated a correlation between Tregsuppression and exhaustion-marker expression.Despite their phenotypically exhausted and suppressive nature,

some of the TIL-enriched immune subsets remained functionallycompetent upon ex vivo stimulation. For instance, CD8+ TRMs andTEMs from the TILs expressed TNFα and IFNγ upon activation withPMA and ionomycin, and their overall percentage was reduced inadvanced-stage tumors. These data imply that these cell subtypes arecritical for tumor progression. This observation is consistent withprevious reports that high abundance of CD45RO+ memory-CD8+

T cells and TRMs are associated with good prognosis in several typesof cancer (1, 38). We found that the PD-1+ TRMs and TEMs couldalso be activated with anti–PD-1 treatment ex vivo and hence may becandidate immune cells that respond to immune-checkpoint block-ade. These data have important clinical relevance as the relativepresentation of these critical immune subsets could be considered asa biomarker for potential clinical response against anti–PD-1 therapy.The majority of cases of HCC reportedly arise from chronic

viral hepatitis infection (39, 40). Furthermore, chronic HBVinfection has been linked to a marked increase in exhaustion-marker expression, including PD-1 on T cells (41). Due to thisincrease, concerns with regards to possible toxicity of immuno-therapeutics in HCC have been raised, as the application ofcheck-point blockage will very likely illicit an off-target immuneresponse to the remaining noncancerous liver tissues infectedwith HBV. Interestingly, our data suggest that such toxicity willnot be a major concern, due to the identification of preferentiallyhigher expression of PD-1 by TILs than NILs. Even though wevalidated that T cells from patients infected with HBV exhibitedhigher PD-1 expression, the percentage of PD-1+ T cells remainshigher in TILs versus NILs, even in this group of HCC patients.This concept is again consistent with the hypothesis of a cancer-immune gradient, particularly regarding the TRMs and TEMsfrom the TILs that showed an increasingly exhaustive profilefrom the adjacent NTME to the TME. This increase in ex-haustion was induced upon exposure to tumor antigens, thusindicating an important role for the TME in shaping and

Chew et al. PNAS | Published online July 3, 2017 | E5907

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

May

17,

202

0

maintaining an exhausted and immunosuppressive microenvi-ronment. The approach used in this study has important im-plications for the design of novel immunotherapeutics in HCC,as the expression profile of various exhaustion markers couldbe analyzed on multiple TIL or NIL subsets simultaneously toselect the most effective therapeutics with the least potentialfor adverse effects. For instance, our detailed phenotypic datahave provided insight into the potential for combined immu-notherapeutic treatments in HCC, such as the combination ofPD-1 and CTLA-4 or Lag-3 blockade based on their differentialexpression levels in TILs versus NILs.This study also demonstrated an inverse relationship between

T-bet and PD-1 expression in TILs and NILs in the cancer set-ting, suggesting a role for T-bet in modulating PD-1 expression.Indeed, exposure of patient PBMCs to tumor antigens resultedin an up-regulation of PD-1 and a corresponding down-regulationof T-bet in TEM cells. This paper reports the relationship betweenT-bet and PD-1 in HCC, although a similar association hasbeen previously described during chronic virus infection (28),including chronic HBV infection (42). Consistently, we alsoobserved higher PD-1 and lower T-bet expression in TILs frompatients with concurrent HCC and HBV infections comparedwith those without HBV infection. Our previous characteriza-tion of HCC immune gene signatures demonstrated that TBX21(T-bet) was an immune gene associated with a good prognosisin HCC (12). Taking this previous observation and our currentdata together, we propose that the inverse relationship betweenT-bet and PD-1 is one of the mechanisms for the evolutionfrom chronic infection and inflammation to tumor developmentand cancer progression.Finally, we explored how the TME might shape the relative

composition and function of infiltrating immune subsets. Nano-String data comparing the TME versus NTME revealed enhancedexpression of CCL20, CXCL10, and CXCL11 in the TME. Nextwe detected the expression of the respective chemokine receptorsCCR6 on TAM and CXCR3 on NK cells. Production of thesechemokines correlated with an increased percentage of TAMs andNKRs in tumors. Most importantly, both of these immune subsetsexhibited an immunosuppressive phenotype in the TME, charac-terized by high IL-10 and low granzyme B expression. Thesefindings are consistent with previous reports that the accumulationof TAMs is associated with poor prognosis in cancer (43).

One of the biggest challenges in this study design was thelimited number of cases from adjacent nontumor liver tissue,which was typically smaller in size and thus provided a limitednumber of immune cells that could be isolated for CyTOF anal-ysis. Therefore, in the majority of the analyses, the NILs exhibitedthe lowest number of cases compared with TILs or PBMCs. De-spite this caveat, most of the data were independently validated byother in vitro assays or by conventional flow cytometry, and in allcases, paired statistical tests were performed to validate the find-ings from individual patients rather than pooling. Most impor-tantly, the unbiased and nonsupervised nature of our CyTOFanalysis using statistical tests to screen and guide subsequent dis-covery is a method of data analysis showcased by our study, whichto the best of our knowledge has not been previously shown inother reported CyTOF data analysis pipelines.In summary, our data provide evidence of a dynamic immune

gradient that is dominated by immune subsets that are pro-gressively more exhausted and immunosuppressive. Our holisticapproach has emphasized two main points: (i) immune sup-pression is microenvironment specific, as immune competency isgenerally maintained in the periphery or in the adjacent NTMEwith chronic inflammation; and (ii) the relevant immune subsetsfollow a chemokine gradient that is expressed in the tumor, andthe cellular phenotypes are modified upon exposure to the TME.This chemokine gradient preferentially recruits and/or retainsTAMs and NKRs in the tumor. These data may be readilytranslated into the context of immune monitoring and accuratecancer staging, and may be applied to the design of novel ther-apeutic targets toward the cancer itself.

ACKNOWLEDGMENTS. The authors thank Win Htut Oo, Fiona Ni Ni Moe, andPhang Su Ting (National Cancer Centre, Singapore) for coordination of thepatient sample collection and assistance in obtaining patient consent;Dr. Davide Lucchesi, Dr. Lakshmi Ramakrishna, and Dr. Chin Teck Ng (SingHealthTranslational Immunology and Inflammation Centre, STIIC) for their contribu-tions in scientific discussion; and Insight Editing London for language editingof this manuscript prior to submission. This work was supported by the Na-tional Medical Research Council (NMRC), Singapore (MOHIAFCAT2001,TCR15Jun006, CIRG16may048, NMRC/CG RIE2015, NMRC/STaR/020/2013,NMRC/MOHIAFCAT2/005/2015, CIRg13nov032, and NMRC/MOHIAFCAT1-6003), Duke University–National University of Singapore Medical School,STIIC, and Biomedical Research Council (BMRC) (BMRC-EDB IAF: IAF311020and SPF2014/005).

1. Fridman WH, Pagès F, Sautès-Fridman C, Galon J (2012) The immune contexture inhuman tumours: Impact on clinical outcome. Nat Rev Cancer 12:298–306.

2. Makarova-Rusher OV, Medina-Echeverz J, Duffy AG, Greten TF (2015) The yin andyang of evasion and immune activation in HCC. J Hepatol 62:1420–1429.

3. Couzin-Frankel J (2013) Breakthrough of the year 2013. Cancer immunotherapy.Science 342:1432–1433.

4. Errico A (2015) Immunotherapy: PD-1-PD-L1 axis: Efficient checkpoint blockadeagainst cancer. Nat Rev Clin Oncol 12:63.

5. Greten TF, Wang XW, Korangy F (2015) Current concepts of immune based treat-ments for patients with HCC: From basic science to novel treatment approaches. Gut64:842–848.

6. Hodi FS, et al. (2010) Improved survival with ipilimumab in patients with metastaticmelanoma. N Engl J Med 363:711–723.

7. Hong YP, Li ZD, Prasoon P, Zhang Q (2015) Immunotherapy for hepatocellular carci-noma: From basic research to clinical use. World J Hepatol 7:980–992.

8. Callahan MK, Wolchok JD (2013) At the bedside: CTLA-4- and PD-1-blocking anti-bodies in cancer immunotherapy. J Leukoc Biol 94:41–53.

9. Ferlay J, et al. (2015) Cancer incidence and mortality worldwide: Sources, methodsand major patterns in GLOBOCAN 2012. Int J Cancer 136:E359–E386.

10. El-Serag HB (2011) Hepatocellular carcinoma. N Engl J Med 365:1118–1127.11. Kudo M (2015) Immune checkpoint blockade in hepatocellular carcinoma. Liver

Cancer 4:201–207.12. Chew V, et al. (2012) Chemokine-driven lymphocyte infiltration: An early intra-

tumoural event determining long-term survival in resectable hepatocellular carci-noma. Gut 61:427–438.

13. Chew V, et al. (2010) Inflammatory tumour microenvironment is associated withsuperior survival in hepatocellular carcinoma patients. J Hepatol 52:370–379.

14. Hernandez-Gea V, Toffanin S, Friedman SL, Llovet JM (2013) Role of the microenvironmentin the pathogenesis and treatment of hepatocellular carcinoma. Gastroenterology 144:512–527.

15. Schmidt N, Flecken T, Thimme R (2014) Tumor-associated antigen specific CD8(+)T cells in hepatocellular carcinoma: A promising target for immunotherapy.OncoImmunology

3:e954919.16. Blank C, Gajewski TF, Mackensen A (2005) Interaction of PD-L1 on tumor cells with

PD-1 on tumor-specific T cells as a mechanism of immune evasion: Implications for

tumor immunotherapy. Cancer Immunol Immunother 54:307–314.17. Han A, Glanville J, Hansmann L, Davis MM (2014) Linking T-cell receptor sequence to

functional phenotype at the single-cell level. Nat Biotechnol 32:684–692.18. Lai L, Ong R, Li J, Albani S (2015) A CD45-based barcoding approach to multiplex

mass-cytometry (CyTOF). Cytometry A 87:369–374.19. Finck R, et al. (2013) Normalization of mass cytometry data with bead standards.

Cytometry A 83:483–494.20. Shekhar K, Brodin P, Davis MM, Chakraborty AK (2014) Automatic Classification of

Cellular Expression by Nonlinear Stochastic Embedding (ACCENSE). Proc Natl Acad Sci

USA 111:202–207.21. Garnelo M, et al. (2017) Interaction between tumour-infiltrating B cells and T cells

controls the progression of hepatocellular carcinoma. Gut 66:342–351.22. Carluccio S, et al. (2015) Generation of tumor-specific cytotoxic T-lymphocytes from

the peripheral blood of colorectal cancer patients for adoptive T-cell transfer. J Cell

Physiol 230:1457–1465.23. Chen KJ, et al. (2012) Intratumoral regulatory T cells alone or in combination with

cytotoxic T cells predict prognosis of hepatocellular carcinoma after resection. Med

Oncol 29:1817–1826.24. Wilke CM, Wu K, Zhao E, Wang G, Zou W (2010) Prognostic significance of regulatory

T cells in tumor. Int J Cancer 127:748–758.25. Shin JI, Ha SJ (2014) Regulatory T cells-an important target for cancer immunother-

apy. Nat Rev Clin Oncol 11:307.26. Lowther DE, et al. (2016) PD-1 marks dysfunctional regulatory T cells in malignant

gliomas. JCI Insight 1(5).

E5908 | www.pnas.org/cgi/doi/10.1073/pnas.1706559114 Chew et al.

Dow

nloa

ded

by g

uest

on

May

17,

202

0

27. Park HJ, et al. (2015) PD-1 upregulated on regulatory T cells during chronic virus in-

fection enhances the suppression of CD8+ T cell immune response via the interaction

with PD-L1 expressed on CD8+ T cells. J Immunol 194:5801–5811.28. Kao C, et al. (2011) Transcription factor T-bet represses expression of the inhibitory

receptor PD-1 and sustains virus-specific CD8+ T cell responses during chronic in-

fection. Nat Immunol 12:663–671.29. Taylor A, et al. (2016) Glycogen synthase kinase 3 inactivation drives T-bet-mediated

downregulation of co-receptor PD-1 to enhance CD8(+) cytolytic T cell responses.

Immunity 44:274–286.30. Yamazaki T, et al. (2008) CCR6 regulates the migration of inflammatory and regu-

latory T cells. J Immunol 181:8391–8401.31. Kondo T, Takata H, Takiguchi M (2007) Functional expression of chemokine receptor

CCR6 on human effector memory CD8+ T cells. Eur J Immunol 37:54–65.32. Rivino L, et al. (2010) CCR6 is expressed on an IL-10-producing, autoreactive memory

T cell population with context-dependent regulatory function. J Exp Med 207:

565–577.33. Acosta-Rodriguez EV, et al. (2007) Surface phenotype and antigenic specificity of

human interleukin 17-producing T helper memory cells. Nat Immunol 8:639–646.34. Schutyser E, Struyf S, Van Damme J (2003) The CC chemokine CCL20 and its receptor

CCR6. Cytokine Growth Factor Rev 14:409–426.

35. Boyle ST, Faulkner JW, McColl SR, Kochetkova M (2015) The chemokine receptorCCR6 facilitates the onset of mammary neoplasia in the MMTV-PyMT mouse modelvia recruitment of tumor-promoting macrophages. Mol Cancer 14:115.

36. Liao KL, Bai XF, Friedman A (2013) The role of CD200-CD200R in tumor immuneevasion. J Theor Biol 328:65–76.

37. Kretz-Rommel A, et al. (2007) CD200 expression on tumor cells suppresses antitumorimmunity: New approaches to cancer immunotherapy. J Immunol 178:5595–5605.

38. Djenidi F, et al. (2015) CD8+CD103+ tumor-infiltrating lymphocytes are tumor-specifictissue-resident memory T cells and a prognostic factor for survival in lung cancerpatients. J Immunol 194:3475–3486.

39. Kremsdorf D, Soussan P, Paterlini-Brechot P, Brechot C (2006) Hepatitis B virus-relatedhepatocellular carcinoma: Paradigms for viral-related human carcinogenesis.Oncogene 25:3823–3833.

40. Nakamoto Y, Guidotti LG, Kuhlen CV, Fowler P, Chisari FV (1998) Immune patho-genesis of hepatocellular carcinoma. J Exp Med 188:341–350.

41. Ye B, et al. (2015) T-cell exhaustion in chronic hepatitis B infection: Current knowl-edge and clinical significance. Cell Death Dis 6:e1694.

42. Fan R, et al. (2016) T-bet expression in CD8+ T cells associated with chronic hepatitis Bvirus infection. Virol J 13:14.

43. Zhang QW, et al. (2012) Prognostic significance of tumor-associated macrophages insolid tumor: A meta-analysis of the literature. PLoS One 7:e50946.

Chew et al. PNAS | Published online July 3, 2017 | E5909

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

May

17,

202

0