antigen-presenting intratumoral b cells affect til ...€¦ · tumoral b cells relative to b cells...

Research Article

Antigen-Presenting Intratumoral B Cells AffectCD4þ TIL Phenotypes in Non–Small Cell LungCancer PatientsTullia C. Bruno1, Peggy J. Ebner1, Brandon L. Moore1, Olivia G. Squalls1,Katherine A.Waugh1, Evgeniy B. Eruslanov2, Sunil Singhal2, John D. Mitchell3,Wilbur A. Franklin4, Daniel T. Merrick4, Martin D. McCarter3, Brent E. Palmer5,Jeffrey A. Kern6, and Jill E. Slansky1

Abstract

Effective immunotherapy options for patients with non–small cell lung cancer (NSCLC) are becoming increasinglyavailable. The immunotherapy focus has been on tumor-infiltrating T cells (TILs); however, tumor-infiltrating B cells(TIL-Bs) have also been reported to correlate with NSCLC patientsurvival. The function of TIL-Bs in human cancer has beenunderstudied, with little focus on their role as antigen-present-ing cells and their influence on CD4þ TILs. Compared with otherimmune subsets detected in freshly isolated primary tumorsfrom NSCLC patients, we observed increased numbers of intra-tumoral B cells relative to B cells from tumor-adjacent tissues.Furthermore, we demonstrated that TIL-Bs can efficiently presentantigen to CD4þ TILs and alter the CD4þ TIL phenotype using

an in vitro antigen-presentation assay. Specifically, we identifiedthree CD4þ TIL responses to TIL-Bs, which we categorized asactivated, antigen-associated, and nonresponsive. Within theactivated and antigen-associated CD4þ TIL population, activat-ed TIL-Bs (CD19þCD20þCD69þCD27þCD21þ) were associatedwith an effector T-cell response (IFNgþ CD4þ TILs). Alterna-tively, exhausted TIL-Bs (CD19þCD20þCD69þCD27�CD21�)were associated with a regulatory T-cell phenotype (FoxP3þ

CD4þ TILs). Our results demonstrate a new role for TIL-Bs inNSCLC tumors in their interplay with CD4þ TILs in the tumormicroenvironment, establishing them as a potential therapeutictarget inNSCLC immunotherapy.Cancer Immunol Res; 5(10); 898–907.�2017 AACR.

IntroductionThe goal of cancer immunotherapy is touse the immune system

to target cancer cells without harming normal cells; for example,kill cancer cells but not lung epithelial cells in non–small cell lungcancer (NSCLC) patients. Vaccination with tumor-specific anti-gens and blockade of the inhibitory PD-1:PD-L1 pathway onCD8þ and CD4þ tumor-infiltrating T cells (TILs; refs. 1–11) haveincreased survival in lung cancer patients. Despite these successes,only 20% to 40% of lung cancer patients respond to the currentimmunotherapies (7–11), suggesting that targeting of otherimmune cells in NSCLC patientsmay lead to increased responses.

Here, we address the contribution of tumor-infiltrating B cells(TIL-Bs) to the immune response in these primary tumors.

Similar to CD8þ and effector CD4þ TILs, TIL-Bs positivelycorrelate with overall survival in several solid tumors, includingNSCLC (12–19). In addition, TIL-Bs are organized in tertiarylymphoid structures (TLS) with CD4þ TILs, and these structuresalso positively correlate with survival in NSCLC (18–19). Thesecorrelations suggest an antitumor role for TIL-Bs in NSCLC;however, there is still controversy as to whether or not TIL-Bshave an anti- or protumor role in the tumor microenvironment.

In addition to correlating with survival, TIL-Bs generate anti-tumor antibodies (19–23). Germain and colleagues demonstrat-ed that germinal center somatic hypermutation and class switchrecombination machineries are activated in TIL-Bs from NSCLCpatient tumors (19). Further, approximately 50% of patientsamples tested have antibody reactivity to tumor antigens (19).It has also been suggested in murine models that TIL-Bs generateantitumor cytokines, directly kill tumor cells, and present tumorantigens (24–31). Antigen presentation is an important functionof B cells. Activated B cells from control donor peripheral bloodlymphocytes (PBL) present antigens to CD4þ and CD8þ T cells(27, 29). Autoimmunity and allograft rejection studies describethe importance of B cells in the formation of TLS and in presen-tation of auto- or allo-antigens that ultimately influence theCD4þ

T-cell phenotype (25, 29–31). TIL-Bs may also present antigensmore effectively than tumor dendritic cells (DC), due to selectivepresentation of cognate antigen through surface Igmolecules. It iscurrently hypothesized that CD4þ and CD8þ TILs are primed by

1Department of Immunology andMicrobiology, University of Colorado School ofMedicine, Aurora, Colorado. 2Division of Thoracic Surgery, University of Penn-sylvania, Philadelphia, Pennsylvania. 3Department of Surgery, University ofColorado School of Medicine, Aurora, Colorado. 4Department of Pathology,University of Colorado School ofMedicine, Aurora, Colorado. 5Division of Allergyand Clinical Immunology, University of Colorado School of Medicine, Aurora,Colorado. 6Division of Oncology, National Jewish Health, Denver, Colorado.

Current address for Tullia C. Bruno: Department of Immunology, University ofPittsburgh, Pittsburgh, Pennsylvania.

Corresponding Author: Jill E. Slansky, University of Colorado School of Med-icine, 12800 East 19th Avenue, Aurora CO 80045. Phone: 303-724-8665; Fax:303-724-8733; E-mail: [email protected]

doi: 10.1158/2326-6066.CIR-17-0075

�2017 American Association for Cancer Research.

CancerImmunologyResearch

Cancer Immunol Res; 5(10) October 2017898

on November 12, 2020. © 2017 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst August 28, 2017; DOI: 10.1158/2326-6066.CIR-17-0075

http://crossmark.crossref.org/dialog/?doi=10.1158/2326-6066.CIR-17-0075&domain=pdf&date_stamp=2017-9-21

Pennsylvania

http://cancerimmunolres.aacrjournals.org/

DCs in the lymph node and then TIL-Bs elicit a recall response inthe tumor. Thus, TIL-Bs serve as a local antigen-presenting cell(APC) by providing secondary stimulation to CD4þ TILs, ulti-mately increasing their survival and proliferation, which is similarto what occurs in chronic viral infection (32–33).

However, TIL-Bs have also been reported to have a protumorfunction. TIL-Bs can suppress the antitumor immune response,and the tumormicroenvironment canpromoteTIL-Bdysfunction.The most prevalent TIL-B–mediated immunosuppression issecretion of IL10 by regulatory B cells (Breg; refs. 34–38).In NSCLC patients, the frequency and absolute numberof IL10-producing Bregs is elevated and positively associatedwith advanced clinical stage (36). Further, Bregs can secreteIL35 and TGFb, which can also suppress the immune response(39–41).

There is also evidence for B cell–mediated immunoregulationin addition to Bregs. In autoimmunity and autograft rejection,removal of B cells from TLS results in decreased T-cell function,and B cells (specifically Bregs) can skew the population toward aregulatory T cell (Treg) phenotype (25, 30–31, 38). In cancer, Bcells can convert CD4þCD25� T cells into Tregs when coculturedwith tongue squamous cell carcinoma cell lines (38). Within thissame study, Zhou and colleagues demonstrated that IL10þ TIL-Bspositively correlated with increased Tregs and decreased survival.B cell–mediated immunoregulation has also been shown using Band T cells isolated from the ascites of ovarian cancer patients(37). However, studies of TIL-Bs from NSCLC patients have notyet been published.

Based on preliminary studies of TIL-Bs in NSCLC patients, wehypothesized that TIL-Bs present antigen to CD4þ TILs andpromote a prolonged antitumor immune response. However,given the evidence for B-cell exhaustion in the context ofchronic antigen exposure (42–43), we also hypothesized thatB cells may be exhausted in some NSCLC patients. To test thesehypotheses, we compared TIL-Bs to other immune cell subsetsthat infiltrated primary NSCLC tumors. We also assessed thefunction of TIL-Bs as APCs in the tumor microenvironment. Tofurther assess TIL-B–mediated immunoregulation of an anti- orprotumor CD4þ TIL response, we analyzed the association ofan activated or exhausted TIL-B phenotype with the proportionof CD4þ Th1 cells and Tregs. Our results show that B cells areincreased at the site of the tumor relative to tumor-adjacenttissues and have variable function as APCs in NSCLC. Inaddition, we demonstrate that the activated or exhaustedTIL-B phenotype predicts the CD4þ TIL phenotype.

Materials and MethodsPatient tissues

All NSCLC tissues were acquired under a University of Color-ado Anschutz Medical Campus and National Jewish HealthInstitutional Review Board (IRB)-approved protocol with writteninformed consent obtained fromeach patient in conjunctionwiththe University of Colorado Lung Cancer SPORE. There were norestrictions on cancer subtype, smoking status, age, race, or prioradjuvant therapy; patients were segregated by tumor subtype afterfinal histological analysis. Control donor PBLs were collectedthrough a material transfer agreement with Bonfils Blood Dona-tion Center (Denver, CO). Disease-free control lung tissue wascollected through an IRB-approved protocol at National JewishHealth. Control splenocytes and untreated HIVþ PBLs were col-lected through an IRB-approval protocol at the University ofColorado. NSCLC pleural effusions and additional NSCLC pri-mary tumors were provided through collaboration at the Univer-sity of Pittsburgh and the University of Pennsylvania or throughan IRB-approved protocol with the National Disease ResearchInterchange. All patient demographics are summarized in Table 1.

Sample processing and lymphocyte selectionPatient-matched PBL samples were collected by venipuncture

into 10 mL whole blood tubes with heparin solution (BD Bios-ciences Vacutainer Systems) the day or week prior to surgery, andlymphocytes were enriched using a Ficoll gradient (GE Health-care). Tumor and tumor adjacent tissue in each patient specimenwas identified by University of Colorado pathologists. Grossidentification of the tumor was performed, and tumor adjacenttissue was procured at the farthest distance from the tumorallowed by surgical margins. Within 1 hour of resection, lungtumor and tumor-adjacent tissue were mechanically disruptedand incubated with the enzyme Liberase DL (Roche, 50 mg/mL)for 25 minutes at 37�C to facilitate the release of lymphocytesfrom tissue. After the incubation, the tissue was passed through a100-mm filter to obtain a single-cell suspension. Cells were resus-pended in complete medium (RPMI 1640 medium supplemen-ted with 1% L-glutamatine, 1% nonessential amino acids, 1%sodium pyruvate, 1% penicillin–streptomycin, and 10% FCS) forcounting. Cells were evenly distributed for cell staining and/orresuspended in 1 mL of Dynal buffer for cell separation (PBSsupplemented with 25% BSA). B cells and T cells were positivelyisolatedusing theDynalCD19,CD8, orCD4Positive IsolationKit(Invitrogen) and the manufacturer's recommended protocol,including magnetic bead detachment. Positively isolated cells

Table 1. Summary of patient data

Assay

AdenocarcinomaSquamous cellother Age Sex Pathology

Previous orcurrentsmoker Mutations� Survival

Flow cytometry 40 45–86 24 M, 34 F pT1a–pT4 52 3 KRAS mut Not known10 3 EFGRþ8 3 TTFþ

2 p53 mutAntigen presentation 9 56–71 4 M, 6 F pT1a–pT3 10 1 KRAS 2 NED, 2 DOD

10

Total 49 (72.0%) — — — 62 — —

11 (16.2%)% of Total 8 (11.8%) — — — 91.2% — —

Survival key: NED, no evident disease; DOD, dead of disease.�Specific mutations are not matched with the lung cancer subtypes.

TIL-Bs Present Antigen to CD4 TILs in NSCLC

www.aacrjournals.org Cancer Immunol Res; 5(10) October 2017 899




were counted and used for functional assays. Using directlyconjugated monoclonal antibodies against CD19, CD8, CD4,and CD3 (clones HIB19, RPA-T8, RPA-T4, and HIT3a, respective-ly, Biolegend), approximately 1,000 cells from each patient werestained and analyzed by flow cytometry to verify >90% purity ofpositively isolated B and T cells.

Isolation of CD11cþ myeloid cellsCD11cþ myeloid cells were isolated by culturing single-cell

suspensions for 2 hours at 37�C for adherence to the culture plate.Subsequently, adherent cells were stimulated for 24 hours withLPS (100 ng/mL, Sigma) and harvested the next day for an in vitroAP assay. Prior to the assay, cells were analyzed with monoclonalantibodies to CD11c, HLA-DR, and CD86 (clones 3.9, L243, andIT2.2, respectively, Biolegend) to verify mature CD11cþ myeloidcells.

Antibody staining and flow cytometrySingle-cell suspensions of tumor and tumor-adjacent tissue

were stained with surface antibodies for CD19, CD20, CD3, CD8,CD4, CD56, and CD11c (clones HIB19, 2H7, HIT3a, RPA-T8,RPA-T4, HCD56, and 3.9, respectively, Biolegend) and an intra-cellular stain for the transcription factor FoxP3 (clone PCH101,Ebioscence) for Figure 1 and for CD19, CD20, CD21, CD69,CD27, and HLA-DR (clones HIB19, 2H7, HB5, FN50, O323, andL243, respectively, Biolegend and Ebioscience) for Figs. 2 and 3.Positive and negative controls for staining were PBLs frompatients or control donors, lymphocytes from control donorspleens, lymphocytes from control (disease-free) lungs, lympho-

cytes from NSCLC pleural metastases, and PBL from untreated,HIVþ patients. Flow cytometrywas conducted using an LSR II (BDBiosciences), and data were analyzed using FlowJo software (TreeStar, Inc.). For Fig. 2B, total lymphocyte infiltrationwas calculatedusing the following equation: [Total cell number from tissue�%lymphocyte gate�%lymphocyte population, i.e., CD19þ]/[massof tissue in grams].

Antigen presentation (AP) assayCD4þ T cells, B cells, and DCs were selected from the indicated

tissues. CD4þ T cells were labeled with CFSE and cocultured at anequivalent ratio with B cells or DCs � the indicated protein �costimulation with anti-CD40 (clone 5C3, 4 mg/mL, Biolegend)and anti-CD28 (cloneCD28.2, 1mg/mL, Ebioscience). Anti–HLA-DR, DP, DQ was utilized to block MHC class II antigen presen-tation (clone Tu39, 1 mg/mL, BD Biosciences). CD4þ T cells wereleft unstimulated for a negative control and were activated withplate bound anti-CD3 (clone OKT3, 0.5 mg/mL, Ebioscience) andsoluble anti-CD28 (1 mg/mL) as a positive control for prolifera-tion and intracellular protein staining. Control PBL donors withincreased exhausted B cells (solid triangles) were rested for 2din vitro without stimulation to decrease CD21 and CD27 expres-sion. Control PBL donors with increased activated B cells (opentriangles) were used ex vivo. The cells were cocultured for 5 daysand were then harvested and restimulated for 4 to 6 hours in vitrowith PMA (0.1 mmol/L, Sigma) and ionomycin (0.5 mmol/L,Sigma) plus BD Golgistop (Monensin, BD recommended con-centration) for intracellular cytokine analysis. Cell division wasanalyzed by CFSE dilution and Ki67 nuclear protein. Ki67, IFNg ,

Figure 1.

Frequency and total number of TIL-Bs increase in NSCLC primary tumors. A, Single-cell suspensions of lymphocytes from the indicated tissues were surfacestained with anti-CD19 and anti-CD20. B, B-cell frequency of paired tumor and tumor-adjacent tissue is shown: adenocarcinomas (black, n ¼ 37, P < 0.0001),squamous cell carcinomas (blue, n ¼ 9), large cell carcinomas (orange, n ¼ 2), and carcinoid tumors (green, n ¼ 4), P ¼ 0.0596 (non-adenocarcinomas).Relative B-cell frequency in other control and patient tissues is shown for comparison.

Bruno et al.

Cancer Immunol Res; 5(10) October 2017 Cancer Immunology Research900




and FoxP3 expression (clones Ki-67, 4S.B3, and PCH101, respec-tively, Biolegend and Ebioscience) was detected using a nuclearintracellular staining kit (FoxP3 staining kit, Ebioscience). Flowcytometry was performed as above.

Generation of cell lysates and purified XAGE-1b proteinNSCLC patient tumor lysates were generated by two freeze/

thaw cycles followed by sonication. EBV lysates were prepared by0.45-mm filtration of B95-8 EBV cell line supernatant and freeze/thaw lysate. XAGE-1b protein was purified utilizing a previouslydescribed method (44).

Statistical analysisAll statistical analyses were performed using GraphPad Prism

6.0. In vitro experiments were repeated at least three times forstatistical significance. P � 0.05 was considered statistically sig-nificant, and statistical differences were measured as follows: Fig.1B, Fig. 2B, Fig. 4B–E, and Fig. 5B used the paired, one-sidedStudent t test, and Fig. 3A, Fig. 4A and Fig. 6C used the unpaired,one-sided Student t test.

ResultsB cells are increased in the tumors of NSCLC patients

To define the role of TIL-Bs in human NSCLC, we first per-formed a comprehensive quantification of intratumoral TILscompared with lymphocytes in tumor-adjacent tissue using flowcytometric analyses. TIL-Bs were increased in frequency and totalnumber compared with tumor-adjacent tissue in all subtypes ofNSCLC, but most prominently in adenocarcinomas (Fig. 1A andB). To understand how the B-cell density in tissues from NSCLCpatients compares with other tissues, we also examined thefrequency of B cells from PBL of NSCLC patients and lung, PBLand spleen of control donors (Fig. 1B). Next, we used otherlymphocyte markers to compare the frequency of B cells in thetumor and tumor-adjacent tissues to other lymphocyte subsets.The total number of TIL-Bs in the tumor was increased relative toother TILs, butwere similar in numbers to intratumoral Tregs (Fig.2A and B). Whereas Tregs have been well studied in NSCLCpatients (45), TIL-Bs have been understudied. Thus, because weobserved an increase of TIL-Bs in NSCLC tumors, we assayed theirfunction as an APC.

Figure 2.

Total number of TIL-Bs increase in NSCLC tumors compared with other immune subsets. A, Single-cell suspensions were stained with the indicated antibodies;representative flow cytometry plots are shown. B, Using the gating strategy shown in A, the fold increase of total tumor infiltrating lymphocytes at thesite of the tumor comparedwith tumor-adjacent tissue is shown (bar indicates median value, n¼ 12–26), P values: TIL-B vs. CD8, P¼ 0.010, TIL-B vs. CD4, P¼ 0.014,TIL-B vs. 11cþ, P ¼ 0.008, TIL-B vs. Treg, n.s. 11cþ represents myeloid CD11cþ cells.






TIL-Bs presented antigen to CD4þ TILsWe first established an in vitro assay to test B cells for antigen

presentation. Utilizing healthy donor splenocytes, B cells werecocultured with CFSE-labeled T cells and T-cell proliferationmeasured as a marker of successful antigen presentation. Asexpected, CD4þ T cells from control PBL divided in the presenceof B cells with EBV lysates, but not in the presence of EBV lysatesand an MHC class II blocking antibody (Fig. 3A–B). In addition,antigen presentation by B cells was similar to CD11cþ myeloidcells in this assay (Fig. 3A).

Next, we examined the expression of antigen presenting mole-cules, MHC class II (HLA-DR) on TIL-Bs. Most TIL-Bs expressedMHC class II, although at lower levels than in B cells from control,healthy donor spleen (Fig. 4A). We compared TIL-Bs to otherintratumoral CD11cþ myeloid cells, which are potent APCs.Although the density of B cells in the tumor relative to thetumor-adjacent tissue increasedmore than the density of CD11cþ

myeloid cells (Fig. 2B), there were similar total numbers ofintratumoral TIL-Bs and CD11cþ myeloid cells in NSCLC (Fig.4B). Because these findings suggested an APC function for TIL-Bs,we utilized the AP assay to determine if TIL-Bs present antigen toCD4þ TILs in NSCLC patients. As with cells from healthy donorsplenocytes, CD4þ T cells divided in the presence of B cells withEBV lysate (Fig. 4C). Further, we tested if patient-derived autol-ogous tumor lysate and a purified cancer–testis protein, XAGE-1b,were presented by TIL-Bs to CD4þ TILs. We identified three typesof CD4þ TIL responses: activated responses, antigen-associatedresponses, and no responses. Activation of some CD4þ TILresponses did not require exogenous antigen, just the additionof TIL-Bs (Fig. 4D, purple). Antigen-associated CD4þ TIL

responses occurred when exogenous antigen was added to theTIL-B coculture (Fig. 4D–E, green). In this group, CD4þ TILsresponded to addition of EBV lysate, autologous tumor lysate,and purified XAGE-1b protein. Lastly, there were patients with noCD4þ TIL responses regardless of addition of TIL-Bs alone or TIL-Bs with exogenous antigen (Fig. 4D–E, blue).

Exhausted and activated TIL-B phenotypes found in NSCLCpatients

We analyzed B-cell exhaustion as one mechanism to explainCD4þ TIL nonresponsiveness to TIL-B antigen presentation. B-cellexhaustion was first described in the tonsils of HIVþ patients as amemory population of B cells with downregulation of CD21 andCD27 (42–43). In some studies, there is evidence of decreasedfunction and increased inhibitory receptors on exhausted B cells;however, the phenotype is still being defined. We used untreatedHIVþ PBL as a positive control for downregulation of CD27 andCD21 on CD69þ B cells and NSCLC pleural metastases as apositive control for upregulation of CD27 and CD21 on CD69þ

B cells. We also utilized control PBLs to demonstrate CD21 andCD27 expression onB cells ex vivo fromadult donors (Fig. 5A).Weidentified both activated (CD69þHLA�DRþCD27þCD21þ) andexhausted (CD69þHLA�DRþCD27�CD21�) TIL-B phenotypesin NSCLC tumors; however, NSCLC patient tumors had moreactivated than exhausted TIL-Bs (Fig. 5B).

TIL-B phenotype associates with CD4þ TIL phenotypeBecause we observed both activated and exhausted TIL-Bs in

NSCLCand apreferential expansionofCD4þTILs over CD8þ TILsinAP assays (Fig. 6A),we queriedwhether TIL-Bs associatedwith a

Figure 3.

B cells from healthy donors present antigen toCD4þ T cells in vitro. A, In representative APassays, autologous CD4þ T cells and B cells or DCsfrom control PBL were cocultured withcostimulation and EBV lysate or a blockingantibody to HLA class II. The "TCR stim" samplecontained plate-bound anti-CD3 and solubleanti-CD28.B,Compiled results (right) from the APassay (P ¼ 0.0235) using control splenocytes(open symbols, n ¼ 5) and control PBL (closedsymbols, n ¼ 2) are shown. 11cþ representsmyeloid CD11cþ cells.

Bruno et al.





particular CD4þ TIL phenotype. To determine whether the TIL-B phenotype correlated with the CD4þ TIL phenotype, the TIL-Bexhaustion profile prior to the AP assay was analyzed with thefinal CD4þ TIL phenotype at the end of our AP assay. Tregs wereidentified as CD4þFoxP3þIFNg� and Th1 cells were identifiedas CD4þFoxP3�IFNgþ (Fig. 6B). B cells from control donorsthat had induced B-cell exhaustion and exhausted TIL-Bs fromNSCLC patients correlated with increased Tregs at the end of theAP assay (Fig. 6C). Conversely, B cells from control donors thathad activated B cells and activated TIL-Bs from NSCLC patientscorrelated with increased Th1 CD4þ TILs at the end of the APassay (Fig. 6C). Finally, we examined the disease status of thesmall cohort of patients that had follow-up data from the startof the study. Patients with TIL-Bs that did not efficiently presentantigen either had relapsed disease or were deceased (n ¼4, Table 1). In contrast, the patients that had either activatedor antigen-associated CD4þ TIL responses remained tumor-free. These results suggest the hypothesis that some NSCLC

patients have a more activated tumor microenvironment andTIL-B antigen presentation, which may contribute to the overallquality of the immune response in NSCLC patients. However,further investigation and more patient samples are required toconfirm these findings.

DiscussionConclusions regarding the function of TIL-Bs in human cancer

are usually either anti- and protumor, depending on the study.However, in NSCLC patients, evidence suggests an antitumor rolefor TIL-Bs. TIL-Bs generate antibodies to tumors, and detection ofTIL-Bs and TLS correlates with better prognosis. Despite thesereports, comprehensive studies on the association of TIL-Bs withother immune subsets had not been completed and TIL-Bs hadnot been thoroughly studied as APCs in NSCLC patients. Further,many studies on the immune contexture in primary NSCLCtumors are done in histological sections, not in fresh, ex vivo

Figure 4.

TIL-Bspresent antigen toCD4þTILs in someNSCLCprimary tumors.A,HLA-DRexpressionon the indicatedB cells is shown (n¼7–15).B,Total numbers of TIL-Bs andDCs in NSCLC tumor samples is shown (n ¼ 12). C, Representative AP assays using NSCLC tumors are shown. D, Comparison of CD4þ T-cell activatedresponses (purple), antigen-associated responses (green), and no responses (blue) in autologous AP assay in the absence of costimulation or exogenous antigen isshown (n ¼ 5). E, AP assay was performed in the presence of EBV lysate, autologous tumor lysate or XAGE-1b protein; each symbol represents an NSCLCpatient sample (n ¼ 10). P values are no antigen versus tumor P ¼ 0.0709, no antigen versus EBV P ¼ 0.0018, and no antigen versus XAGE-1b P ¼ 0.0134.






tissue. Finally, most functional studies that describe TIL-B–medi-ated antigen presentation or immunoregulation are not per-formed utilizing TIL-Bs from the patient primary tumor, butrather, murine or in vitro culture models.

Our data fill the gaps described above and provide two obser-vations about TIL-Bs in NSCLC patients. First, we demonstratedthat TIL-Bs can present antigen to CD4þ TILs in some NSCLCpatients. Second, we determined that the TIL-B phenotype, acti-vated or exhausted, associates with the CD4þ TIL phenotype.Consequently, we suggest a two-pronged model for TIL-B func-tion in NSCLC patients. In the antigen presentation part of themodel, TIL-Bs can present endogenous tumor antigens to CD4þ

TILs spontaneously (activated responses) or after re-stimulation

with antigen (antigen-associated responses) or not at all (noresponses). In the immunoregulation part of themodel, theCD4þ

TIL phenotype can be predicted by the prevalence of TIL-Bexhaustion or activation markers. Thus, the TIL-B phenotypeinfluences the CD4þ TIL phenotype, which may determine ifdownstream cytolytic CD8þ TIL responses are further activatedor suppressed in NSCLC patients.

To understand this two-pronged model of TIL-B function, wefirst considered the twomodalities for TIL-B function. The "B cell–intrinsic model" suggests recruitment of TIL-Bs to the tumor byIFNg produced by T cells. TIL-Bs in this model do not form TLSwithCD4þ TILs and suppress antitumor function via the secretionof the suppressive cytokines IL10 and IL35 (30, 34–41). Thisintrinsicmodel does not support our data on antigenpresentationand immunoregulation as both of these functions require cell-to-cell contact. In contrast, the "B cell-extrinsic model" suggests thatTIL-Bs are recruited to the tumor site and formTLSwithCD4þ TILsto influence the overall immune response (30). Our hypothesisthat TIL-Bs present antigen to CD4þ TILs and promote an anti-tumor immune response supports the extrinsic model of TIL-Bfunction. Our hypothesis was supported by the observation ofTIL-Bs presenting antigen to CD4þ TILs in NSCLC patients.Support for antigen-specific TIL-B responses has been describedin two ways in NSCLC patients: (i) tumor-specific antibodyproduction and (ii) oligoclonality of TIL-Bs in TLS. Further, instudies with mice that receive an antigenic vaccination, B cellsenhance the response to vaccine (30). Our data combinedwith these other studies provide evidence for a B-cell extrinsicrole in NSCLC patients. It remains unclear why TIL-Bs fromsome patients do not present antigen to CD4þ TILs. Perhaps theTIL-Bs pick up but do not properly process antigen. Or, perhapsTIL-Bs are anergized by other components of the tumormicroenvironment.

The second portion of our hypothesis that is supported by the Bcell–extrinsic model relates to TIL-B–mediated immunoregula-tion of CD4þ TILs. In line with our hypothesis, we observed anexhausted TIL-B phenotype in our NSCLC patients. However, thefunction of these exhausted TIL-Bs does not fit with the definitionof lymphocyte exhaustion as described in models of chronicinfection or cancer. In these models, T cells and B cells that areexhausted are described as less functional. Such T cells proliferateless and produce less cytokine (46–48); B cells produce lessantibody (42–43). The TIL-Bs in our NSCLC patient sampleswith an exhausted signature still present antigen. However, theyalso associate with CD4þ T cells of a Treg phenotype. Oneconsideration in understanding the TIL-B exhaustion profile inNSCLC patients is control of host damage during chronic antigenexposure. In chronic infection and autoimmunity, B cells mayattenuate T-cell responses in an attempt to decrease host damage.It is possible that in the context of chronic self-antigen, such as intumors, TIL-Bs may obtain an exhausted phenotype to skew theCD4þ TILs to a Treg population for this purpose, ultimatelydampening the overall antitumor response in the tumor. B cell-mediated immunoregulation has been implicated in variousmalignancies. However, the phenotypic markers of activated andexhausted TIL-Bs must be enhanced to better understand howthey differ from Breg populations that dampen the immuneresponse by secretion of suppressive cytokines. Location maycontribute to the difference between Bregs and exhausted TIL-Bs:IL10 producing Bregs may be scattered throughout the tumorwhereas exhausted TIL-Bs may be found in TLS.

Figure 5.

TIL-Bs fromNSCLC tumors have both an activated and exhausted phenotype.A,Single-cell suspensions of NSCLC tumors were gated on live lymphocytes andstained with anti-CD19 and anti-CD20 to identify the TIL-B population and anti-CD69, anti-HLA-DR, anti-CD21, and anti-CD27 to quantitate the frequencyof CD69þHLA�DRþCD21þCD27þ (activated) andCD69þHLA�DRþCD21�CD27�

(exhausted) TIL-Bs. Control PBL, untreatedHIVþPBL, and effusions fromNSCLCpleural metastases were used for comparison. B, Data were compiled forfrequency of activated and exhausted TIL-Bs in NSCLC patient tumors(n ¼ 3–9). P value for exhausted versus activated TIL-Bs ¼ 0.010.

Bruno et al.





Our data suggest strategies to target TIL-Bs in NSCLC patientsto improve patient responses to immunotherapy. Total B-celldepletion has little clinical benefit in patients with solid tumors(49–50); here, we suggest that targeting a specific subtype of TIL-Bis necessary for effective immunotherapies. Focusing ablation onB cells with an exhausted phenotypemay reduce promotionof theTreg phenotype, resulting in reduced tumor growth. In addition,the link between tumor-specific antibodies and the tumor anti-gens that are processed and presented by TIL-Bs could be targetedTIL-B–based immunotherapies. TIL-Bs inNSCLC are often in TLS,which suggests an antigen-specific role for TIL-Bs. We observe TLSin our patient samples, and they have been reported in NSCLCpatients by other groups (18–19). However, whether the tumorantigens that are targeted by TIL-B antibodies can also be pro-cessed and presented to CD4þ TILs is unknown. If so, this strategywould link cellular and humoral immunity in solid tumors.Finally, an increased understanding of the activated andexhausted TIL-B phenotypes may be reflective of an immunolog-ical biomarker to measure the quality of the immune response inNSCLC patients who are being considered for immunotherapy.

These studies may improve patient responses to immunother-apy in two ways: (i) increased antigen presentation by TIL-Bs toCD4þ TILs in the tumor microenvironment in combination withanti–PD-1 therapy may further increase tumor-specific immuneresponses, and (ii) methods to skew exhausted B cells to activatedB cells or to deplete exhausted TIL-Bs may promote an activated,antitumor microenvironment in NSCLC patients. An activatedmicroenvironment offers an advantage in obtaining a productiveresponse to immunotherapy. Thus, by targeting TIL-Bs in com-bination with current immunotherapies, we can further improveNSCLC patient survival.

Disclosure of Potential Conflicts of InterestJ.E. Slansky is a consultant/advisory boardmember for Agenus. No potential

conflicts of interest were disclosed by the other authors.

Authors' ContributionsConception and design: T.C. Bruno, K.A. Waugh, J.E. SlanskyDevelopment of methodology: T.C. Bruno, E.B. Eruslanov, S. Singhal,W.A. Franklin, D.T. MerrickAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): T.C. Bruno, P.J. Ebner, E.B. Eruslanov, S. Singhal,J.D. Mitchell, W.A. Franklin, D.T. Merrick, M.D. McCarter, B.E. Palmer, J.A. KernAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): T.C. Bruno, O.G. Squalls, J.E. SlanskyWriting, review, and/or revision of the manuscript: T.C. Bruno, P.J. Ebner,O.G. Squalls, J.D. Mitchell, W.A. Franklin, D.T. Merrick, M.D. McCarter,J.A. Kern, J.E. SlanskyAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): T.C. Bruno, P.J. Ebner, O.G. SquallsStudy supervision: T.C. Bruno, J.E. SlanskyOther (assisted in conducting experiments): B.L. Moore

AcknowledgmentsWe thank the University of Colorado Lung Cancer SPORE group for their

helpwith patient consent and tissue collection, especiallyMary Jackson,MelissaLemar, HeatherMalinowski, and Robert Tsay. We thank RobertMason, ClaudiaJakubzick, and William Jansen for help with accruing normal lung specimens,and Steve Albeda (University of Pennsylvania) and Vera Donnenberg (Univer-sity of Pittsburgh) for help with accruing NSCLC patient samples. Further, wethank Colt Egleston and Peter Lee for providing EBV lysates. Lastly, we thankBiolegend for providing multiple human antibodies.

Grant SupportT32 AI007505 (T.C. Bruno, trainee); American Cancer Society RSG

LIB-114645 (PI J.E. Slansky); the Cancer League of Colorado (PI T.C. Bruno);

Figure 6.

TIL-B phenotype predicts CD4þ TIL phenotype in NSCLC tumors. A, CD4þ and CD8þ TIL frequencies were analyzed in AP assay; TIL without stimulation (T only),stimulation with plate-bound anti-CD3 and soluble anti-CD28 (TCR stim), and autologous TIL-Bs and antigen are shown (n ¼ 3-4). B, Representative IFNgand FoxP3 intracellular staining after AP assay. D, Treg:Th1 populations were compared after AP assay in donors that had activated and exhausted B cells. Assaywas performed as in Fig. 3, but with analysis of IFNg and FoxP3 proteins. Closed symbols represent patients with exhausted TIL-B phenotype greater than50% and dominant Treg phenotype at the end of the AP assay (n ¼ 4). Open symbols represent patients with activated TIL-B phenotype greater than 50% anddominant Th1 phenotype at the end of the AP assay (n ¼ 4). Triangle, control donor; square/circle, NSCLC patient. P ¼ 0.0022 for Treg:Th1 log ratio.






University of Colorado Lung Cancer SPORE Career Development Award5P50CA058187-20 (PI T.C. Bruno); CCSG CA 046934, CCTSI UL1 TR001082, RO1 CA187392-01A1 (PI E.B. Eruslanov); and DoD W81XWH-15-1-0717 (PI E.B. Eruslanov).

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby marked

advertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received February 6, 2017; revised July 10, 2017; accepted August 21, 2017;published OnlineFirst August 28, 2017.

References1. Hanibuchi M, Yano S, Nishioka Y, Yanagawa H, Miki T, Sone S. Immu-

nological circumvention of multiple organ metastases of multidrug resis-tant human small cell lung cancer cells by mouse-human chimeric anti-ganglioside GM2 antibody KM966. Clin Exp Metastasis 2000;18:353–60.

2. Lara PN Jr, Laptalo L, Longmate J, Lau DH, Gandour-Edwards R, Gumer-lock PH, et al. Trastuzumab plus docetaxel in HER2/neu-positive non-small-cell lung cancer: a California Cancer Consortium screening andphase II trial. Clin Lung Cancer 2004;5:231–6.

3. Vuillez JP, Kraeber-Bod�er�e F,MoroD, Bardi�esM,Douillard JY,Gautherot E,et al. Radioimmunotherapy of small cell lung carcinoma with the two-stepmethod using a bispecific anti-carcinoembryonic antigen/anti-diethylene-triaminepentaacetic acid (DTPA) antibody and iodine-131 Di-DTPA hap-ten: results of a phase I/II trial. Clin Cancer Res 1999;5:3259s–67s.

4. DuN, Li X, Li F, ZhaoH, FanZ,Ma J, et al. Intrapleural combination therapywith bevacizumab and cisplatin for non-small cell lung cancermediatedmalignant pleural effusion. Oncol Rep 2013;29:2332–40.

5. Wang Y, Deng G, Liu X, Cho WC. Monoclonal antibodies in lung cancer.Expert Opin Biol Ther 2013;13:209–26.

6. Brahmer JR, DrakeCG,Wollner I, Powderly JD, Picus J, SharfmanWH, et al.Phase I study of single-agent anti-programmed death-1 (MDX-1106) inrefractory solid tumors: safety, clinical activity, pharmacodynamics, andimmunologic correlates. J Clin Oncol 2010;28:3167–75.

7. Giri A, Walia SS, Gajra A. Clinical trials investigating immune checkpointinhibitors in non-small-cell lung cancer. Rev Recent Clin Trials 2016;11:297–305.

8. Qin A, Coffey DG, Warren EH, Ramnath N. Mechanisms of immuneevasion and current status of checkpoint inhibitors in non-small cell lungcancer. Cancer Med 2016;5:2567–78.

9. Morgensztern D, Herbst RS. Nivolumab and pembrolizumab for non-small cell lung cancer. Clin Cancer Res 2016;22:3713–7.

10. Brahmer J, Reckamp KL, Baas P, Crin�o L, Eberhardt WE, Poddubskaya E,et al. Nivolumab versus docetaxel in advanced squamous-cell non-small-cell lung cancer. N Engl J Med 2015;373:123–35.

11. Garon EB, Rizvi NA, Hui R, Leighl N, Balmanoukian AS, Eder JP, et al.Pembrolizumab for the treatment of non-small-cell lung cancer. N Engl JMed 2015;372:2018–28.

12. Al-Shibli KI, Donnem T, Al-Saad S, Persson M, Bremnes RM, BusundLTPrognostic effect of epithelial and stromal lymphocyte infiltration innon-small cell lung cancer. Clin Cancer Res 2008;14:5220–7.

13. Milne K, K€obel M, Kalloger SE, Barnes RO, Gao D, Gilks CB, et al.Systematic analysis of immune infiltrates in high-grade serous ovariancancer reveals CD20, FoxP3 and TIA-1 as positive prognostic factors. PLoSOne 2009;4:e6412.

14. Nedergaard BS, Ladekarl M, Nyengaard JR, Nielsen K. A comparative studyof the cellular immune response inpatientswith stage IB cervical squamouscell carcinoma. Low numbers of several immune cell subtypes are stronglyassociated with relapse of disease within 5 years. Gynecol Oncol2008;108:106–11.

15. Woo JR, Liss MA, Muldong MT, Palazzi K, Strasner A, Ammirante M, et al.Tumor infiltrating B-cells are increased in prostate cancer tissue. J TranslMed 2014;12:30.

16. Shi JY,GaoQ,WangZC, Zhou J,WangXY,MinZH, et al.Margin-infiltratingCD20(þ) B cells display an atypicalmemory phenotype and correlate withfavorable prognosis in hepatocellular carcinoma. Clin Cancer Res2013;19:5994–6005.

17. Shimabukuro-Vornhagen A, Schl€oßer HA, Gryschok L, Malcher J, Wenn-hold K, Garcia-Marquez M, et al. Characterization of tumor-associated B-cell subsets in patientswith colorectal cancer.Oncotarget 2014;5:4651–64.

18. Dieu-Nosjean MC, Antoine M, Danel C, Heudes D, Wislez M, Poulot V,et al. Long-term survival for patients with non-small-cell lung cancer withintratumoral lymphoid structures. J Clin Oncol 2008;26:4410–7.

19. Germain C, Gnjatic S, Tamzalit F, Knockaert S, Remark R, Goc J, et al.Presence of B cells in tertiary lymphoid structures is associated with aprotective immunity in patients with lung cancer. Am J Respir Crit CareMed 2014;7:832–44.

20. Mizukami M, Hanagiri T, Shigematsu Y, Baba T, Fukuyama T, Nagata Y,et al. Effect of IgG produced by tumor-infiltrating B lymphocytes on lungtumor growth. Anticancer Res 2006;26:1827–31.

21. Yasuda M, Mizukami M, Hanagiri T, Shigematsu Y, Fukuyama T,Nagata Y, et al. Antigens recognized by IgG derived from tumor-infiltrating B lymphocytes in human lung cancer. Anticancer Res 2006;26:3607–11.

22. Carmi Y, Spitzer MH, Linde IL, Burt BM, Prestwood TR, Perlman N, et al.Allogeneic IgG combined with dendritic cell stimuli induce antitumourT-cell immunity. Nature 2015;521:99–104.

23. Kotlan B, Simsa P, Teillaud JL, Fridman WH, Toth J, McKnight M, et al.Novel ganglioside antigen identified by B cells in human medullary breastcarcinomas: the proof of principle concerning the tumor-infiltratingB lymphocytes. J Immunol 2005;175:2278–2285.

24. Rivera A, Chen CC, Ron N, Dougherty JP, Ron Y. Role of B cells as antigen-presenting cells in vivo revisited: antigen-specific B cells are essential for Tcell expansion in lymph nodes and for systemic T cell responses to lowantigen concentrations. Int Immunol 2001;13:1583–1593.

25. Nelson BH. CD20þ B cells: the other tumor-infiltrating lymphocytes.J Immunol 2010;185:4977–4982.

26. Candolfi M, Curtin JF, Yagiz K, Assi H, Wibowo MK, Alzadeh GE, et al.B cells are critical to T-cell-mediated antitumor immunity induced bya combined immune-stimulatory/conditionally cytotoxic therapy forglioblastoma. Neoplasia 2011;13:947–960.

27. Rodríguez-Pinto D. B cells as antigen presenting cells. Cell Immunol 2005;238:67–75.

28. Haynes NM, Hawkins ED, Li M, McLaughlin NM, H€ammerling GJ,Schwendener R, et al. CD11cþ dendritic cells and B cells contribute tothe tumoricidal activity of anti-DR5 antibody therapy in establishedtumors. J Immunol 2010;185:532–541.

29. Yanaba K, Bouaziz JD, Matsushita T, Magro CM, St Clair EW, Tedder TF. B-lymphocyte contributions to human autoimmune disease. Immunol Rev2008;223:284–299.

30. ShenM, SunQ,Wang J, PanW, Ren X. Positive and negative functions of Blymphocytes in tumors. Oncotarget 2016;7:55828–39.

31. Guy TV, Terry AM, Bolton HA, Hancock DG, Shklovskaya E, Fazekas de St.Groth B. Pro- and antitumour effects of B cells and antibodies in cancer: acomparison of clinical studies and preclinical models. Cancer ImmunolImmunother 2016;65:885–96.

32. Wakim LM,Waithman J, van Rooijen N, HeathWR, Carbone FR. Dendriticcell-induced memory T cell activation in nonlymphoid tissues. Science2008;319:198–202.

33. McGill J, Van Rooijen N, Legge KL. Protective influenza-specific CD8 T cellresponses require interactions with dendritic cells in the lungs. J Exp Med2008;205:1635–1646.

34. Tedder TF.B10 cells: a functionally defined regulatory B cell subset.J Immunolo 2015;194:1395–1401.

35. Chen T, Song D, Min Z, Wang X, Gu Y, Wei B, et al. Perioperative dynamicalterations in peripheral regulatory T and B cells in patients with hepato-cellular carcinoma. J Translat Med 2012;10:14.

36. Liu J, Wang H, Yu Q, Zheng S, Jiang Y, Liu Y, et al. Aberrant frequency ofIL-10-producing B cells and its association with Treg and MDSC cellsin non-small cell lung carcinoma patients. Human Immunol 2016;77:84–89.

37. Wei X, Jin Y, Tian Y, ZhangH,Wu J, LuW, et al. Regulatory B cells contributeto the impaired antitumor immunity in ovarian cancer patients. TumourBiol 2016;37:6581–6588.

Bruno et al.





38. Zhou X, Su YX, Lao XM, Liang YJ, Liao GQ. CD19(þ)IL-10(þ) regulatory Bcells affect survival of tongue squamous cell carcinomapatients and induceresting CD4(þ) T cells to CD4(þ)Foxp3(þ) regulatory T cells. Oral Oncol2016;53:27–35.

39. Zhang Y, Sun H, Wu H, Tan Q, Xiang K. Interleukin 35 is an independentprognostic factor and a therapeutic target for nasopharyngeal carcinoma.Contemporary Oncol 2015;19:120–124.

40. Fan YG, Zhai JM, Wang W, Feng B, Yao GL, An YH, et al. IL-35 over-expression is associatedwith genesis of gastric cancer. Asian Pacific J CancerPrev 2015;16:2845–2849.

41. Pylayeva-Gupta Y,Das S,Handler JS,HajduCH,CoffreM,Koralov SB, et al.IL35-producing B cells promote the development of pancreatic neoplasia.Cancer Discov 2016;6:247–255.

42. Moir S,Ho J,MalaspinaA,WangW,DiPotoAC,O'SheaMA, et al. Evidence forHIV-associated B cell exhaustion in a dysfunctional memory B cell compart-ment in HIV-infected viremic individuals. J Exp Med 2008;205:1797–805.

43. Sciaranghella G, Tong N, Mahan AE, Suscovich TJ, Alter G. Decouplingactivation and exhaustion of B cells in spontaneous controllers of HIVinfection. AIDS 2013;27:175–80.

44. Jordan KR, Buhrman JD, Sprague J, Moore BL, Gao D, Kappler JW,et al. TCR hypervariable regions expressed by T cells that respond to

effective tumor vaccines. Cancer Immunol Immunother 2012;61:1627–38.

45. Zhao S, Jiang T, Zhang L, YangH, Liu X, Jia Y, et al. Clinicopathological andprognostic significance of regulatory T cells in patients with non-small celllung cancer: A systematic review with meta-analysis. Oncotarget 2016;7:36065–36073.

46. Wherry EJ. T cell exhaustion. Nat Immunol 2011;12:492–99.47. Youngblood B, Wherry EJ, Ahmed R. Acquired transcriptional program-

ming in functional and exhausted virus-specific CD8 T cells. Curr OpinHIV.AIDS 2012;7:50–57.

48. Wherry EJ, Ha SJ, Kaech SM, Haining WN, Sarkar S, Kalia V, SubramaniamS, et al. Molecular signature of CD8þ T cell exhaustion during chronic viralinfection. Immunity 2007;27:670–684.

49. Aklilu M, Stadler WM, Markiewicz M, Vogelzang NJ, Mahowald M, John-sonM, et al. Depletion of normal B cells with rituximab as an adjunct to IL-2 therapy for renal cell carcinoma and melanoma. Ann Oncol 2004;15:1109–1114.

50. Barbera-Guillem E, NelsonMB, Barr B, Nyhus JK, May KF Jr, Feng L, et al. Blymphocyte pathology in human colorectal cancer. Experimental andclinical therapeutic effects of partial B cell depletion. Cancer ImmunolImmunother 2000;48:541–549.






2017;5:898-907. Published OnlineFirst August 28, 2017.Cancer Immunol Res Tullia C. Bruno, Peggy J. Ebner, Brandon L. Moore, et al.

Small Cell Lung Cancer Patients−Phenotypes in Non TIL +Antigen-Presenting Intratumoral B Cells Affect CD4

Updated version

10.1158/2326-6066.CIR-17-0075doi:

Access the most recent version of this article at:

Cited articles

http://cancerimmunolres.aacrjournals.org/content/5/10/898.full#ref-list-1

This article cites 47 articles, 15 of which you can access for free at:

Citing articles

http://cancerimmunolres.aacrjournals.org/content/5/10/898.full#related-urls

This article has been cited by 3 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerimmunolres.aacrjournals.org/content/5/10/898To request permission to re-use all or part of this article, use this link



http://cancerimmunolres.aacrjournals.org/lookup/doi/10.1158/2326-6066.CIR-17-0075

http://cancerimmunolres.aacrjournals.org/content/5/10/898.full#ref-list-1

http://cancerimmunolres.aacrjournals.org/content/5/10/898.full#related-urls

http://cancerimmunolres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerimmunolres.aacrjournals.org/content/5/10/898


antigen-presenting intratumoral b cells affect til ...€¦ · tumoral b cells relative to b cells...

Documents