Slan Monocytes and Macrophages Mediate CD20-Dependent B-Cell

Published OnlineFirst May 10, 2018; DOI: 10.1158/0008-5472.CAN-17-2344

Cancer Tumor Biology and Immunology Research

slanþ Monocytes and Macrophages Mediate CD20-Dependent B-cell Lymphoma Elimination via ADCC and ADCP William Vermi1,2, Alessandra Micheletti3, Giulia Finotti3, Cristina Tecchio4, Federica Calzetti3, Sara Costa3, Mattia Bugatti1, Stefano Calza5, Claudio Agostinelli6, Stefano Pileri7, Piera Balzarini1, Alessandra Tucci8, Giuseppe Rossi8, Lara Furlani4, Giuseppe Todeschini4, Alberto Zamo9, Fabio Facchetti1, Luisa Lorenzi1, Silvia Lonardi1, and Marco A. Cassatella3

Abstract

þ Terminal tissue differentiation and function of slan monocytes in cancer þ + + is largely unexplored. Our recent studies demonstrated that slan mono- slan monocyte slan macrophage

cytes differentiate into a distinct subset of dendritic cells (DC) in human CD16A þ tonsils and that slan cells colonize metastatic carcinoma-draining lymph CD32 nodes. Herein, we report by retrospective analysis of multi-institutional þ CD16A CD64 cohorts that slan cells inﬁltrate various types of non-Hodgkin lymphomas = RTX (NHL), particularly the diffuse large B-cell lymphoma (DLBCL) group, CD20 þ CD20 including the most aggressive, nodal and extranodal, forms. Nodal slan cells displayed features of either immature DC or macrophages, in the latter case ingesting tumor cells and apoptotic bodies. We also found in patients þ þ with DLBCL that peripheral blood slan monocytes, but not CD14 monocytes, increased in number and displayed highly efﬁcient rituxi- Lymphoma cell Lymphoma cell

mab-mediated antibody-dependent cellular cytotoxicity, almost equivalent + RTX þ to that exerted by NK cells. Notably, slan monocytes cultured in conditioned medium from nodal DLBCL (DCM) acquired a macrophage-like + RTX phenotype, retained CD16 expression, and became very efficient in rituximab-mediated antibody-dependent cellular phagocytosis (ADCP). Macro- þ phages derived from DCM-treated CD14 monocytes performed very efficient rituximab-mediated ADCP, however, using different FcgRs from þ those used by slan macrophages. Our observations shed new light on the complexity of the immune microenvironment of DLBCL and demonstrate þ plasticity of slan monocytes homing to cancer tissues. Altogether, data þ identify slan monocytes and macrophages as prominent effectors of antibody-mediated tumor cell targeting in patients with DLBCL. ADCC ADCP fi þ Signi cance: slan monocytes differentiate into macrophages that func- slan+ monocytes perform RTX-mediated antibody-dependent cell-mediated tion as prominent effectors of antibody-mediated tumor cell targeting in cytotoxicity (ADCC) via CD16A, while slan+ macrophages exert RTX-mediated antibody-dependent cellular phagocytosis (ADCP) via CD16A, CD32 and CD64. lymphoma. © 2018 American Association for Cancer Research Graphical Abstract: http://cancerres.aacrjournals.org/content/canres/ 78/13/3544/F1.large.jpg. Cancer Res; 78(13); 3544–59. 2018 AACR.

1Section of Pathology, Department of Molecular and Translational Medicine, Note: Supplementary data for this article are available at Cancer Research University of Brescia, Brescia, Italy. 2Department of Pathology and Immu- Online (http://cancerres.aacrjournals.org/). nology, Washington University, Saint Louis, Missouri. 3Section of General A. Micheletti, L. Lorenzi, and S. Lonardi contributed equally to this article. Pathology, Department of Medicine, University of Verona, Verona, Italy. 4Section of Hematology, Department of Medicine, University of Verona, Corresponding Authors: Marco A. Cassatella, Department of Medicine, Verona, Italy. 5Unit of Biostatistics, Department of Molecular and Transla- University of Verona, Strada Le Grazie 4, Verona 37134, Italy. Phone: 3904- tional Medicine, University of Brescia, Brescia, Italy. 6Haematopathology 5802-7130; Fax: 3904-5802-7127; E-mail: [email protected]; and Unit, Department of Experimental Diagnostic and Specialty Medicine, William Vermi, Department of Molecular and Translational Medicine, Section S. Orsola-Malpighi Hospital, University of Bologna, Bologna, Italy. 7Unit of of Pathology, P.le Spedali Civili 1, Brescia 25123, Italy. Phone: 3903-0399-8425; Haematopathology, European Institute of Oncology, 20141 Milan, Italy. Fax: 3903-0399-5377; E-mail: [email protected] 8Division of Haematology, ASST Spedali Civili di Brescia, Brescia, Italy. doi: 10.1158/0008-5472.CAN-17-2344 9Section of Pathology, Department of Public Health and Diagnostics, University of Verona, Verona, Italy. 2018 American Association for Cancer Research.

3544 Cancer Res; 78(13) July 1, 2018

Role of slanþ Cells in B-cell Lymphoma

Introduction Materials and Methods Blood monocytes are heterogeneous in terms of phenotype Tissues þ and function and, based on their differential expression of A tissue screening for slan -cell content was performed on CD14 and CD16, are currently subdivided into classical sections from formalin-fixed paraffin-embedded tissue blocks þþ þþ þ (CD14 CD16 ), intermediate (CD14 CD16 ), and non- and included reactive (n ¼ 67, Supplementary Table S1) and þþ classical (CD14dim/ CD16 ) monocytes. Moreover, the neoplastic (n ¼ 151, Supplementary Table S2) lymph nodes from þþ majority of nonclassical (CD14dimCD16 ) monocytes open surgical biopsies retrieved from the Department of Pathol- specifically express the "slan" antigen (1), a carbohydrate ogy, ASST Spedali Civili di Brescia (Brescia, Italy). Neoplastic modification of P-selectin glycoprotein ligand 1 (PSGL-1; lymph nodes included low- and high-grade WHO-defined non- ref. 2). This phenotype, coupled with functional dendritic cell Hodgkin lymphoma (NHL) B- and T-cell lymphomas, as reported (DC)-like features, led Sch€akel and colleagues to originally call in Supplementary Table S2. We also included normal bone them as slanDCs (2, 3). However, recent unsupervised hierar- marrow (n ¼ 10), bone marrow localization of primary nodal chical clustering has unequivocally proved that slanDCs rep- DLBCL (n ¼ 11), as well as 15 cases of primary DLBCL from þ resent a subset of monocytes (4). slan monocytes are potently various extranodal sites. The validation set was represented by proinflammatory, given their capacity to produce higher three multi-institutional cohorts of newly diagnosed and previ- amounts of IL12p70 and TNFa in response to TLR ligands ously untreated, primary DLBCL, two of which with clinical data þþ þ than classical CD14 CD16 monocytes or CD1c DCs do available (Supplementary Tables S3 and S4). The study has þ (3). slan monocytes are also potent inducers of antigen- been conducted according to the Declaration of Helsinki specific T-cell responses in vitro (2, 3) and induce a superior principles, after having obtained written informed consent þ Th1/Th17 cell polarization as compared with classical CD1c from the patients, and approved by the local ethics board DCs. Importantly, based on their highly specificreactivitytothe (NP906-WV-IMMUNOCANCERhum). þ very stable slan antigen, slan cells can be detected in tissues (including archival material) by antibodies known as DD1, Clinical and demographic features of patients with DLBCL of þ DD2, and M-DC8 (1). To date, slan cells have been detected the validation set in inflamed tissues of patients affected by Crohn disease, Cohort 1 (CH1) included tissue microarray (TMA) cores of 61 rheumatoid arthritis, psoriasis, acute intestinal GVHD samples, DLBCL (from the Haematopathology Unit, Department of Exper- as well as in reactive tonsils and Peyer patches (1, 3, 5, 6). imental, Diagnostic and Specialty Medicine - DIMES, University þ Although the type of differentiation fate of slan monocytes in of Bologna, Bologna, Italy). TMAs were prepared as reported various tissues remains to be established, we have demonstrat- previously (13). Cohort 2 (CH2) included open surgical biopsies þ ed that slan cells isolated from human tonsils, as well as of 43 patients retrieved from the archive of the Department of þ slan monocytes cultured with tonsil-derived medium, show Pathology ASST Spedali Civili di Brescia (Brescia, Italy; Supple- unequivocal terminal differentiation to DCs (5). mentary Table S3). Cohort 3 (CH3) included open surgical þ Recently, we have also reported that slan cells distinctively biopsies of 51 patients retrieved from the archive of the Depart- home to metastatic carcinoma draining lymph nodes (M-TDLNs), ment of Pathology and Diagnostic, Verona University Hospital in close proximity to transformed epithelial cells (1). Our study (Verona, Italy; Supplementary Table S4). For CH1, a panel of þ also revealed that slan cells are absent in primary carcinoma and mouse mAbs toward CD10, BCL6, and MUM1/IRF4 was applied are recruited to lymph nodes following the arrival of cancer cells to define the cell of origin, according to Hans' algorithm (14). IHC þ (1). However, the precise functional role of slan cells in M- and FISH analysis for BCL2, MYC, and BCL6 were used to TDLNs could not be defined, although our in situ analysis clearly subgroup "double expressor" and "double hit" DLBCL, according þ indicated that slan cells efficiently engulf dead cancer cells. to the literature (15, 16) In light of these observations, we pursued the hypothesis that þ slan cells could also infiltrate primary nodal lymphomas, a IHC prediction that preliminarily turned out to be true, particularly Four-micron thick tissue sections were used for IHC staining. in the case of diffuse large B-cell lymphoma (DLBCL). Thus, on the The panel of markers and the detection systems used for the IHC þ basis of the well-documented biological and clinical relevance of analysis are reported in Supplementary Table S5. slan cells were the immune microenvironment in DLBCL (7–10), we tested the identified by IHC using DD1 or DD2 mAbs, as reported previ- þ relevance of slan cells in cancer by using the DLBCL as a model. ously (1). Single immune reactions were developed using Novo- In such regard, patients with DLBCL respond to antibodies link polymer (Leica Microsystems). For double IHC, the second targeting CD20-expressing B cells, such as rituximab (11, 12), immune reaction was visualized using Mach 4 MR-AP (Biocare whose biological activity is partially mediated by CD16- Medical), followed by Ferangi Blue (Biocare Medical) as chro- dependent antibody-dependent cellular cytotoxicity (ADCC) mogen. Based on a training set of 15 cases on digitalized slides, a and/or antibody-dependent cellular phagocytosis (ADCP). three-tiered score was defined as follows: score 0 20 cell/mm2 Accordingly, we have undertaken a series of studies with the aim (mean 4.1); score 1 200 cells/mm2 (mean 145); score 2 þ þ of defining whether slan monocytes/slan cells can perform 200 cells/mm2 (mean 621.6). A semiquantitative three-tiered rituximab-mediated ADCC and ADCP. We also investigated the scoring system (S0–S2) was also used for CD56 and CD16 þ functional importance and clinical relevance of slan cells asso- expression analysis. þ þ ciated with DLBCL. Our data show that slan monocytes/slan macrophages represent hitherto neglected cellular effectors of Lymph node samples antibody-mediated tumor cell targeting in patients with DLBCL Fresh lymph node samples obtained from patients with treated with rituximab. NHL (details are reported in Supplementary Table S6) were

www.aacrjournals.org Cancer Res; 78(13) July 1, 2018 3545

Vermi et al.

immediately minced into 4 to 5 small (3–4mm3)fragments, analysis was performed by FlowJo software Version 10.1. Cell placed in gentleMACS C Tubes (Miltenyi Biotec) and treated for viability was analyzed by Vybrant DyeCycle Violet (Life Technol- 20 minutes at 37C with 0.2 mg/mL Liberase Blendzyme 2 ogies; ref. 1). Phenotypic analysis under the various experimental (Roche) in RPMI1640 medium (Lonza). Then, 0.5 mg/mL conditions was performed on live cells, identified as Vybrant- þ DNase (Worthington Biochemical Corporation) was added to negative cells (in the case of in vitro differentiated slan mono- lymph node pieces 5 minutes prior to their processing by cytes) or PI-negative cells (in the case of lymph node suspensions; gentleMACS dissociator (Miltenyi Biotec). Lymph node suspen- ref. 5). The mean fluorescence intensity (MFI) relative to each sions were then washed, filtered through a 70-mmcellstrainer, molecule was obtained by subtracting either the MFI of the suspended in PBS containing 2% fetal low-endotoxin FBS correspondent isotype control mAb, or cell autofluorescence (<0.5 endotoxin U/mL, Sigma-Aldrich), and 2 mmol/L EDTA (fmo). To count the absolute number of each specific cell pop- (Sigma-Aldrich). Fractions of each lymph node cell suspensions ulation, 50 mL whole blood from HDs or patients with DLBCL was were immediately used for the phenotypic characterization of stained for 15 minutes at room temperature with fluorochrome- myeloid cells by flow cytometry analysis, or used to isolate conjugated mAbs listed in Supplementary Table S5. Samples were neoplastic B cells. Another aliquot of lymph node suspension then incubated for 10 minutes with the RBC Lysis Solution was instead cultured at 1 108/mL in RPMI1640 medium (Qiagen), washed and analyzed by MACSQuant Analyzer (Mil- þ þ supplemented with 10% low-endotoxin FBS. After 24 hours, tenyi Biotec). slan monocytes, CD14 monocytes, and NK cells þ þ þ cell-free DLBCL-conditioned medium (DCM) was collected were identified, respectively, as M-DC8 /CD16 /HLA-DR , þ þ þ andstoredat20C. CD14 /CD16 /HLA-DR , and CD56 /HLA-DR cells within the þ CD45 /CD3 /CD19 cell population. Neutrophils were identi- þ þ Cell isolation and culture fied as CD16 /SShi/CD45lo cells within total CD45 cells. The Peripheral blood mononuclear cells (PBMC) were isolated absolute number per mL of each blood cell population was þ under endotoxin-free conditions by density centrifugation calculated by multiplying their percentage within the CD45 - (Ficoll–Paque; GE Healthcare), of either buffy coats of healthy gated cells to the white cell counts provided by hospital laboratory donors (HD), or whole blood collected in BD Vacutainer tubes of analysis. with K2EDTA (BD Biosciences) from patients with DLBCL (whose clinical characteristics are reported in Supplementary Table S7). ADCC assay þ þ slan monocytes, CD14 monocytes, natural killer (NK) cells, ADCC of Daudi, peripheral, and neoplastic B cells was per- and neutrophils were then purified using, respectively, the slan formed and evaluated by a Calcein-Acetyoxymethyl Cytotoxicity þ (M-DC8) Monocyte Isolation Kit (Miltenyi Biotec; >96% puri- Assay, as reported previously (17). Specifically, 1–2 106 target þ ty), the CD14 MicroBeads (Miltenyi Biotec; >98% pure), the cells were initially stained with 5 mmol/L Calcein-AM (Molecular EasySep Human NK Cell Enrichment Kit (Stem Cell Technologies; Probes) in PBS plus 2% FBS for 15 minutes at 37C. After two >96% pure) and the EasySep Human Neutrophil Enrichment Kit washings in PBS plus 2% FBS, cells were adjusted to 106/mL (Stem Cell Technologies; >99.7% purity; ref. 5). CD43 B cells in complete medium and either left untreated or coated with from HD blood or lymphoma samples were purified from, 5 mg/mL aCD20 rituximab (MABTHERA from Roche), for 20 min- respectively, HD PBMCs and lymph node suspensions, by using utes at 37C. Then, 5 103 target cells (in 5 mL) were transferred in the human B Cell Isolation Kit II (Miltenyi Biotec; >98% purity). U-bottom 96-well plates and cocultured in a final volume of þ þ Neoplastic B cells were identified by their monoclonal expression 150 mL with slan monocytes, NK cells, CD14 monocytes, or of either the l- or the k-light chain, by flow cytometry. The Daudi neutrophils from either HDs or patients with DLBCL, at increasing cell line was obtained from ATCC, tested for mycoplasma nega- E:T ratios (from 1:1 to 10:1). In selected experiments, 5 103 tivity by PCR (using primers detecting at least 49 different myco- Daudi cells were cocultured with HD NK cells at 2/10:1 E:T ratios, þ plasma strains: forward 50-ACT CCT ACG GGA GGC AGC AGT with or without 5–10 103 slan monocytes. In other experi- A-30; reverse 50-TGC ACC ATC TGT CAC TCT GTT AAC CTC-30), ments, 5 103 Daudi, peripheral and neoplastic B cells were þ and cultured in RPMI1640 medium supplemented with 10% FBS. cultured with 5 104 DCM-treated slan monocytes and NK þ þ In some experiments, 0.5 106/mL HD and DLBCL slan cells, or IL34- or GM-CSF plus IL4-treated slan monocytes (at a þ monocytes, CD14 monocytes, or NK cells were cultured for 10:1 E:T ratio). After 4 hours, duplicates of 50 mL cell-free super- 5 days in RPMI1640 medium containing 10% FBS, in the presence natants from each well were transferred into flat-bottom 96-well þ þ of 20% DCM. Alternatively, HD slan monocytes and CD14 plates (Corning) to measure their fluorescence by a VICTOR3 monocytes were cultured for 5 days with either 30 ng/mL GM- multilabel reader (PerkinElmer). Percent of specific lysis was CSF (PeproTech) plus 20 ng/mL IL4 (PeproTech) to generate calculated according to the formula: [(test release spontaneous þ slan DCs (5) or 100 ng/mL IL34 (R&D Systems) to generate release)/(maximum release spontaneous release)] 100. macrophages (5). Spontaneous and maximum release represent, respectively, calcein release from target cells in medium alone or in medium plus Flow cytometry analysis 2% Triton X-100. In selected experiments, the same cell samples For antigen expression analysis, typically 2.5 105 PBMCs, were examined for phagocytosis by confocal microscopy. 5 105 cells from lymph node suspensions, and 104 in vitro þ differentiated slan monocytes cells were initially incubated for ADCP assay 10 minutes in 50 mL PBS containing 5% human serum (to prevent Daudi, normal peripheral, and neoplastic B cells were nonspecific binding), and then stained for 15 minutes at room stained with PKH26 dye (18), using the PKH26 Red Fluorescent temperature with fluorochrome-conjugated mAbs listed in Sup- Cell Linker Midi Kit for General Cell Membrane Labeling plementary Table S5. Sample fluorescence was measured by an (Sigma-Aldrich), according to the manufacturer's instructions. 8-color MACSQuant Analyzer (Miltenyi Biotec), while data After two washings in complete medium, target cells were

3546 Cancer Res; 78(13) July 1, 2018 Cancer Research

Role of slanþ Cells in B-cell Lymphoma

adjusted to 106/mL and either left untreated or coated with Nemenyi–Damico–Wolfe–Dunn test for all-pairwise compari- 5 mg/mL aCD20 rituximab for 20 minutes at 37C. 104 sons (approximate with Monte Carlo resampling, B ¼ 50000) PKH26-labeled target cells were then incubated with 104 effector for multiple levels predictors. Times to event data were modeled cells in U-bottom 96-well plates, in a final volume of 100 mL. After using Cox proportional hazard models. All analyses were 4-hour incubation, plates were centrifuged, supernatants collect- performed using R (version 3.3.2), assuming a significant level ed, and cells stained by Vybrant DyeCycle Violet in PBS plus 2% of 5%. For in vitro cellular assays, data are expressed as mean FBS and 2 mmol/L EDTA for 15 minutes on ice. Immediately prior SEM of the number of experiments indicated in each figure legend. to FACS analysis, cells were incubated for additional 2 minutes Statistical analysis, including two-tailed Student t test, one-way with 400 mg/mL Crystal Violet (Merck KGaA), to quench extra- or two-way ANOVA, was performed by Prism Version 5.0 cellular fluorescence (18). By doing so, only effector cells contain- software (GraphPad). ing engulfed targets maintain PKH26 positivity, whereas free target cells or target cell externally attached to effector cells become PKH26 negative. Finally, ADCP was determined by gating Results þ þ live Vybrant negative cells and, subsequently, PKH26 cells. In slan cells infiltrate NHL, particularly nodal DLBCL selected experiments, before target cell addition, effector cells Based on our previous results (1), we hypothesized that lymph þ were pretreated for 30 minutes at 37C with or without 10 mg/mL node accumulation of slan cells might occur also in nodal þ anti-CD16 (clone 3G8 from BD Biosciences), anti-CD32 (clone lymphomas. We first obtained preliminary results of slan -cell AT10, Bio-Rad), anti-CD64 (clone 10.1, BioLegend) mAbs, either infiltration in various nonneoplastic lymph nodes by staining 67 alone or in combinations. As control, effector cells were also cases of lymphadenitis with various reactive patterns (Supple- pretreated with 10 to 30 mg/mL mouse IgG1 isotype control mentary Table S1). By using a three-tiered density score (see þ antibody (clone MG1-45, BioLegend). Materials and Methods for details), we detected slan cells in 79.1% of the cases (with score 1 or 2), even though they were Laser confocal microscopy particularly abundant (equal to score 2) in a limited fraction þ After being cultured for ADCC assay, DCM-treated slan (16.4%) of the cases only (LD in Fig. 1A; Supplementary Fig. S1A– monocytes and calcein-labeled Daudi cells were gently collected S1H; Supplementary Table S1). The slan antigen confirmed its and rapidly cytospun on Polysine Adhesion slides (Thermo specificity being slan-negative the large majority of lymphoid and Fisher Scientific) at 500 rpm for 5 minutes (Shandon CytoSpin stromal cells, sinus macrophages, Langerhans cell–derived inter- III Cytocentrifuge). Cytospins of cultured cells were then fixed digitating DCs (Supplementary Fig. S1I) and granuloma-forming with 4% (w/v) paraformaldehyde for 20 minute and then per- macrophages (Supplementary Fig. S1J–S1L). These latter findings meabilized with 0.1% Triton X-100 for 30 minutes at room confirm our previous observations revealing that the slan marker temperature. Nuclear staining was done with DAPI (Sigma- is not expressed by most of the specialized tissue-resident mono- Aldrich). Thereafter, samples were washed twice with PBS. Pre- nuclear cells (1). parations were mounted in antibleaching medium and visualized We then screened a large cohort of NHL (Fig. 1A), including 125 by confocal laser scanning microscopy (LSM 510; Zeiss) equipped cases of B-cell lymphoma (NHBCL) and 26 cases of T-cell lym- with argon and helium-neon lasers. Image analysis was per- phoma (NHTCL). In NHL, the slan antigen reacts with cells of the formed with LSM 5 Image Browser software (Zeiss). microenvironment (Supplementary Fig. S2A–S2P), but not with the B (Supplementary Fig. S2Q and S2R) and T (Supplementary þ Phenotype and function of HD and DLBCL slan monocytes Fig. S2S and S2T) lymphoid neoplastic cells. We could detect þ þ HD or DLBCL blood slan monocytes were cultured at slan cells (score 1 or 2) in 58.4% of NHBCL and in 53.9% of 0.25 106/mL in U-bottom 96-well plates, in a final volume of NHTCL (Fig. 1A and B; Supplementary Table S2). By subgroup þ þ 100 mL. slan monocytes were stimulated for 24 hours with or analysis (Supplementary Table S2), slan cells were rarely found without 100 ng/mL LPS (Ultra Pure Escherichia coli LPS; 0111:B4 in B-cell chronic lymphocytic leukemia (B-CLL)/small lympho- þ strain, from InvivoGen), 100 U/mL IFNg (R&D Systems), and cytic lymphoma, nodal marginal zone lymphoma and ALK - þ 5 mmol/L R848 (InvivoGen), alone or in combinations. Cytokine anaplastic large cell lymphoma. In contrast, slan cells were levels in cell-free supernatants were measured by specific ELISA particularly abundant in DLBCL, showing a score 2 in 33.3% of kits for human TNFa and IL6 (eBioscience, Affymetrix). Analysis the cases. We therefore extended our findings on nodal DLBCL, by of membrane markers was performed by flow cytometry on live analyzing a validation set of primary nodal WHO-defined DLBCL Vybrant-negative cells using the following mAbs: anti-human from three independent cohorts, including 155 cases in total. CD83 PE and CD14 APC (from Miltenyi Biotec), CD16 Specifically, for CH1 (n ¼ 61), the analysis was performed on TMA PerCP-Cy5.5, and HLA-DR ACP/Cy7 (from BioLegend). cores, whereas the entire nodal section was available for the analysis of CH2 (n ¼ 43) and CH3 (n ¼ 51). By staining of CH1, þ Statistical analysis we could confirm infiltration of slan cells in 49.1% (30/61) of Pairwise correlation coefficients between markers were com- the cases. For CH2 and CH3, we employed a three-tiered density puted using different statistics depending on the type of variable score and data are detailed in Supplementary Tables S3 and S4 and considered. Accordingly, "polyserial" correlation was computed summarized in Fig. 1C and Supplementary Table S8. Briefly, the between numeric and ordinal variables, whereas "polychoric" analysis on the entire nodal section confirmed a high frequency of þ correlation was used when both variables were ordinal. Associ- DLBCL cases (70.22%) containing slan cells (Fig. 1D–F), with ation between markers and qualitative variables (sex, stage, nonsignificant differences between the two cohorts (65.1% for revised International Prognostic Index; ref. 19) were performed CH2 vs. 74.5% for CH3, Supplementary Table S8). Remarkably, using nonparametric tests, namely Wilcoxon rank sum test for 26.6% of the DLBCL cases (Supplementary Table S8) were score 2 þ þ two-level variables, as well as Kruskal–Wallis test followed by for slan -cell infiltration. slan -cell distribution displayed spatial

www.aacrjournals.org Cancer Res; 78(13) July 1, 2018 3547

Vermi et al.

Figure 1. slanþ cells infiltrate nodal and extranodal DLBCL. A–C, Graphs report the frequency of the cases containing slanþ cells at the indicated score in LD, in B- (NHBCL), and T (NHTCL)-non-Hodgkin lymphoma, nodal (NDLBCL), and extranodal (EDLBCL) DLBCL. S0, S1, and S2 stand for, respectively, slan score ¼ 0, ¼ 1, and ¼ 2 (see Materials and Methods for details). Sections are from lymph nodes (D–F), bone marrow (G–L), thyroid (M–O), and skin (P–R), infiltrated by primary nodal (D–L) and primary extranodal (M–R) DLBCL, stained as labeled. CD20 and slan sections were counterstained with Meyer's hematoxylin. Original magnification, 40 (G, J,andM; scale bar, 500 mm), 100 (D and P; scale bar, 200 mm); 200 (E–F, H–I, K–L, N–O, and Q–R; scale bar, 100 mm).

þ heterogeneity (Supplementary Fig. S3A–S3D), thus explaining slan cells infiltrate aggressive forms of DLBCL the significantly lower frequency of DLBCL cases containing DLBCL is a heterogeneous form of lymphoma that includes þ þ slan cells in the TMA-based CH1. slan cells were observed in subgroups defined by clinical, histologic, and molecular features. the majority of bone marrow localizations of primary nodal By gene expression profiling, DLBCL are subdivided in germinal DLBCL (77.8%; Fig. 1G–L, Supplementary Table S2), as well as center B-cell (GBC) DLBCL and activated B-cell DLBCL, the latter in cases of primary DLBCL from various extranodal sites showing a poorer outcome (20). The distinction between GBC (80%, Fig. 1C; Supplementary Table S2) including thyroid and non-GBC DLBCL can be obtained by using a surrogate þ (Fig. 1M–O) and skin (Fig. 1P–R). IHC panel on archival material (14). In CH1, slan cells

3548 Cancer Res; 78(13) July 1, 2018 Cancer Research

Role of slanþ Cells in B-cell Lymphoma

Figure 2. slanþ cells inﬁltrate DEL and DHL. The frequency of the cases containing slanþ cells at the indicated score is reported for BCL2þ- and MYCþ- DLBCL in CH2 (A) and CH3 (B) cohorts, as well as in DEL and DHL DLBCL subgroups compared with all the DLBCLs (C). S0, S1, and S2 stand for, respectively, slan score ¼ 0, ¼ 1, and ¼ 2 (see Materials and Methods for details). Sections are from a nodal "double-expressor" DLBCL (DEL; D–F) and a "double hit" DLBCL (DHL; G–I), and stained as labeled. FISH analysis of DHL (H and I) shows split (red–green) signals. Sections were counterstained with Meyer's hematoxylin (D–G) or DAPI (H–I). Original magniﬁcation, 200 (D–G; scale bar, 100 mm); 1,000 (H and I, scale bar, 20 mm).

infiltrated both the GBC (n ¼ 13/26; 50%) and non-GBC cases (score 1 or 2) was 58% (7/12), slightly lower than those of (n ¼ 17/35; 48.5%), with no significant differences between the unselected DLBCL (70.22%; Fig. 2C). Together, these findings þ two subgroups (P > 0.5). Another subgroup of DLBCL with clini- indicate that slan -cells infiltrate also aggressive forms of DLBCL. cal relevance is the so-called "double hit" DLBCL (DHL; refs. 15, þ 21). DHL, now viewed as a new category under the umbrella of Morphology and phenotype of slan cells in DLBCL þ "high-grade B-cell lymphoma, with MYC and BCL2 and/or BCL6 By morphology, slan cells in DLBCL were large and with a translocations" (21), are characterized by a combination of trans- dendritic-like form (Supplementary Fig. S4A–S4L) or, more location activating MYC, BCL2, and, less commonly, BCL6 (21), rarely, small and monocytoid (Supplementary Fig. S4M–S4T). þ identifiable by FISH analysis. Along the same line, double-expres- Remarkably, the latter slan cell type was also observed around sor lymphomas (DEL) show overexpression of the Myc and Bcl2 HEV or within their vessel lumen, even transmigrating (Supple- proteins by IHC (16, 22). In both DHL and DEL, CHOP and mentary Fig. S4M). As previously observed in tonsils (5), and þ R-CHOP regimens are less effective treatments and patients have tumor-draining lymph nodes (1), slan cells infiltrating DLBCL shorter progression-free survival (PFS) and overall survival (21). were negative for markers of B cells (PAX5), T cells (CD3), By combining Myc and Bcl2 immunostaining, we analyzed CH2 plasmacytoid DCs (CD303), and Langerhans cells (CD1a; Sup- þ and CH3 and defined DEL as reported by Green and colleagues plementary Fig. S4A–S4D). In contrast, slan cells coexpressed, (16). In line with the literature (22), we found that a large fraction although at various frequency, a set of myeloid markers including of the cases showed Bcl2 (respectively 60.5% for CH2; 60.8% for S100 protein, CD11c, CD68, CSF1R, lysozyme, and CD16 (n ¼ 6; CH3; mean 60.6%) or Myc overexpression (respectively 46.5% Supplementary Fig. S4E–S4I). This pattern of reactivity was also þ for CH2; 47.1% for CH3; mean 46.8%), but only a minor fraction observed in the smaller slan -monocytoid cells with only slight of the cases corresponded to DEL (25.6% for CH2; 31.4% for differences (Supplementary Fig. S4M–S4T). Specifically, small þ CH3; mean 28.7%), with minor differences between CH2 and slan -monocytoid cells were negative for S100 protein (Supple- CH3 (Supplementary Table S8). Remarkably, no differences in mentary Fig. S4Q). þ the frequency of cases containing slan cells were observed in We also observed that a fraction of DLBCL displays intratu- þ þ þ þ the four (BCL2 /MYC ; BCL2 /MYC ; BCL2 /MYC ; BCL2 / moral slan cells that closely interact with neoplastic B cells þ þ MYC ) subgroups defined by the BCL2 or Myc reactivity (Fig. 3A–H). More importantly, some intratumoral slan cells (P ¼ 0.33; Fig. 2A and B; Supplementary Table S8). Furthermore, are engulfed of apoptotic bodies (Fig. 3C and D), or contain þ þ þ a sizeable fraction of cases (66.7%) containing slan cells (score 1 PAX5 (Fig. 3E and F), and MYC nuclei (Fig. 3G and H), all or 2) was observed also in DEL (Fig. 2C–F). We expanded this features suggesting efficient phagocytosis of lymphoma cells observation at the molecular level by analyzing a group of 12 DHL along with a macrophage-like differentiation. Accordingly, we þ showing BCL2 and MYC rearrangements by FISH analysis found that nodal slan cells in DLBCL also coexpress the mac- þ þ (Fig. 2C, G–I). The fraction of DHL cases containing slan cells rophage marker CD163 (Fig. 3I–K), at variance with slan cells in

www.aacrjournals.org Cancer Res; 78(13) July 1, 2018 3549

Vermi et al.

Figure 3. Phagocytosing intratumoral slanþ cells coexpress CD163, the macrophage marker. Sections are from human nodal DLBCL and stained as labeled. Top, intratumoral slanþ cells are enlarged and contain intracytoplasmic apoptotic bodies (A and B) expressing active caspase-3 (C and D) and containing intracytoplasmic PAX5þ/MYCþ nuclei (E–H). Bottom, representative nodal DLBCL cases containing CD163þ/slanþ macrophage-like cells (I–K) and others with þ CD163 /slan DC-like cells (L–N). Sections were counterstained with Meyer's hematoxylin. Original magniﬁcation, 100 (L–N; scale bar, 200 mm) and 200 (I–K; scale bar, 100 mm). High-power view was obtained at 600 magniﬁcation (A–H and insets in I–N).

human tonsils (5). However, we expanded this observation on nodes obtained from another cohort of patients with NHBCL (see þ DLBCL cases highly infiltrated by slan cells (score 2), document- Supplementary Table S6 for their clinical features). By doing so, þ ing two almost mutually exclusive phenotypic patterns based we calculated that the slan cells account for about 7% of the total þ þ þ of CD163 and slan reactivity. In the first pattern, slan cells HLA-DR CD11c myeloid cells in DLBCL samples (n ¼ 7). þ coexpress CD163 and were admixed to CD163 slan macro- We could also confirm the existence of two mutually exclusive þ phages (Fig. 3I–K). In the second pattern, the large majority of slan -cell phenotypes in nodal NHBCL (Fig. 4A). Specifically, we þ slan cells were negative for CD163 (Fig. 3L–N); of note, within found a CD163hi/CD14hi/CD64hi/CD16hi macrophage pheno- þ this latter subgroup, CD163 slan macrophages were either type (Fig. 4A, blue circles), which resembles that displayed by the þ þ coexisting (Fig. 3L), found in a distinct area (Fig. 3M), or totally nodal CD14 CD11b population (Fig. 4A, orange circles), or a þ absent (Fig. 3N). Finally, CD163 slan resulted substantially CD163low/CD14low/CD64low/CD16low DC phenotype (Fig. 4A, negative for CD83, CD208, and CCR7 expression, indicating a green circles), which instead resembles that exhibited by the nodal þ þ likely immature DC status (Supplementary Fig. S4J–S4L). CD1c and CD141 DC populations (Fig. 4A, gray and red circles By flow cytometry analysis of a broad panel of markers, we respectively). Examples of the macrophage and DC profiles of þ þ could analyze and compare the phenotypes of slan cells with nodal DLBCL slan cells are also shown in Fig. 4B. When com- þ those of other myeloid cell populations present in fresh lymph pared with circulating slan monocytes from either HDs (Fig. 4C)

3550 Cancer Res; 78(13) July 1, 2018 Cancer Research

Role of slanþ Cells in B-cell Lymphoma

Figure 4. Phenotypes of nodal NHL and DLBCL slanþ cells and other myeloid cells. A, Nodal slanþ cells, CD14þCD11bþ macrophages, CD1cþ DCs, and CD141þ DCs were identified by flow cytometry analysis as, respectively, M-DC8þ, CD14þ/CD11bþ, CD1cþ, and CD141þ cells within HLA-DRþ/CD11cþ/LIN myeloid cells in NHL or DLBCL lymph nodes (LN; see Materials and Methods). Each filled circle in the dot plots stands for the expression levels of CD14, CD16, CD64, and CD163 by the cell populations found in each lymph node sample, as indicated. Blue and green circles in each dot plot indicate nodal samples that, according to the phenotype, belong to the same group since display, in a mutually exclusive manner, either a macrophage (blue circles) or a DC (green circles) phenotype. Values indicate the MFI for each antigen. B, The two panels illustrate representative phenotypes displayed by nodal DLBCL slanþ macrophages (left) or slanþ DCs (right), based on their CD14, CD16, CD64, and CD163 expression. MFI values for every marker and isotype control mAbs are shown on the right side of each panel. C, Bars show the expression (mean SEM) of the indicated markers analyzed in HD slanþ monocytes (n ¼ 6–12) and in nodal DLBCL slanþ cells displaying either a macrophage (blue columns; n ¼ 3–4) or a DC (green columns; n ¼ 3–4) phenotype. Statistical significance: , P < 0.05; , P < 0.01; and , P < 0.001 by one-Way ANOVA; x, P < 0.05; xx, P < 0.01 by two-tailed Student t test and comparing slanþ monocytes with slanþ DCs.

www.aacrjournals.org Cancer Res; 78(13) July 1, 2018 3551

Vermi et al.

þ þ or unmatched patients with DLBCL, nodal DLBCL slan macro- (Fig. 5F). Importantly, HD and DLBCL slan monocytes (Fig. 5G) phages expressed significantly increased levels of CD14, CD163, and NK cells (Fig. 5H) were found to also perform a rituximab- CD32, CD64, and CD80 (Fig. 4C, blue columns), whereas nodal dependent ADCC of either primary tumor B cells directly isolated þ DLBCL slan DCs completely downmodulated CD16 (Fig. 4C, from DLBCL lymph node samples or normal autologous/heter- þ þ green columns). These findings indicate that slan monocytes ologous B cells. Altogether, data reveal that blood slan mono- homing to lymph nodes of NHBCL, and more specifically, cytes from newly diagnosed and untreated patients with DLBCL patients with DLBCL, are skewed toward two mutually exclusive are, in most cases, expanded, functionally unaltered and able to distinct differentiation programs. display efficient rituximab-mediated ADCC activities.

þ þ slan monocytes from patients with DLBCL are increased in DCM induces slan monocytes to differentiate into frequency and efficiently perform a rituximab-mediated ADCC macrophages displaying an effective rituximab-mediated ADCP þ We subsequently analyzed the frequency and functional status As in vitro surrogate of nodal DLBCL-associated slan cells þ of blood slan monocytes in newly diagnosed patients with (that were impossible to purify in working numbers), we exam- þ DLBCL (see Supplementary Table S7 for their clinical features) ined DLBCL-conditioned slan monocytes. The latter cells were þ and age-matched HDs. For comparison, we also analyzed autol- obtained by incubating HD slan monocytes in DCM for 5 days, þ ogous NK cells, CD14 monocytes, and neutrophils. We found similarly to the approach previously used to generate tonsil-like þ þ that, within similar WBC counts (Supplementary Fig. S5A), blood slan DCs (5). By flow cytometry, DCM-treated slan monocytes þ slan monocytes are significantly increased in patients with increase 2/3-fold in size and reproducibly display a macrophage- DLBCL as compared with age-matched HDs, in terms of either like phenotype, at least in terms of CD11b, CD14, CD16, CD32, absolute number (Fig. 5A, top) or percentage within the PBMC CD40, CD64, CD80, CD86, and CD163 expression (Supplemen- þ fraction (Fig. 5A, bottom). In contrast, NK cells of patients with tary Fig. S7A). Also slan monocytes from patients with DLBCL DLBCL were significantly decreased, at least in terms of absolute treated with DCM (n ¼ 3) displayed a macrophage phenotype, þ number (Fig. 5B, top), whereas CD14 monocytes followed a based on CD16, CD32, CD64, HLA-DR, CD86, and CD163 þ þ trend similar to slan monocytes in terms of percentage within marker expression. DCM-conditioned slan cells significantly PBMCs (Fig. 5C, bottom). DLBCL neutrophils were significantly decreased their ability to perform rituximab-mediated ADCC of increased as compared with their age-matched HD counterpart Daudi cells, lymph node tumor B cells, or normal blood B cells (Supplementary Fig. S5B), whereas total PBMCs were diminished (Fig. 6A), as opposed to matched DCM-conditioned NK cells, (Supplementary Fig. S5C). In addition to their selective increased which maintained a potent rituximab-mediated ADCC capacity þ þ frequency, DLBCL slan monocytes normally responded to IFNg toward any type of B-cell target (Fig. 6A). When slan cells and NK plus LPS or R848, in terms of either modulation of CD16, CD14, cells obtained from tonsils, or IL34- and IL4 plus GM-CSF–treated þ CD83, and HLA-DR expression (Supplementary Fig. S5D–S5G), slan monocytes (differentiated to, respectively, macrophage-like or TNFa and IL6 production (Supplementary Fig. S5H and S5I). or DC-like cells) were tested for rituximab-mediated ADCC, only Furthermore, no expression of surface PD-L1 was detected in tonsil NK cells were effective. þ þ both CD14 and slan monocytes, when freshly isolated from As mentioned above, rituximab is known to also mediate þ either HDs or patients with DLBCL. ADCP (24). Accordingly, we found that HD slan monocytes are Patients with DLBCL are usually treated with R-CHOP (ritux- efficient in performing rituximab-mediated ADCP, but only of imab-cyclophosphamide, doxorubicin, vincristine, and predni- normal (small) B cells from HDs, and not of B cells of larger size sone) or R-CHOP–like immunochemotherapy regimens (11). such as Daudi or tumor B cells from nodal DLBCL (Fig. 6B). In þ Rituximab is a humanized mAb targeting CD20 expressed contrast, DCM-treated slan cells were highly efficient in per- by B cells. Rituximab-dependent clinical activity is based on forming rituximab-mediated ADCP of B cells of any size (Fig. 6B), various mechanisms, including ADCC and ADCP (23, 24). It is as also confirmed by confocal microscopy experiments showing þ commonly accepted that NK cells, via the engagement of their that DCM-treated slan cells engulf calcein-labeled Daudi cells in FcgRIIIA/CD16, represent the most relevant effector cells of ritux- the presence of rituximab (Fig. 6C). Consistent with their mac- þ imab-mediated ADCC of target B cells (25, 26). Nonetheless, rophage-like differentiation (5), also IL34-treated slan mono- þ þ given that also CD16 monocytes/slan monocytes have been cytes were found very efficient in performing rituximab-mediated reported to perform ADCC activities toward tumor cells via CD16 ADCP of normal/tumor B cells, as well as Daudi cells (Fig. 6D), as þ (27, 28), we tested their ability to exert a rituximab-dependent opposed to the GM-CSF plus IL4-treated slan monocytes ADCC of CD20-expressing cells, as compared with other effector (Fig. 6D). As expected, no rituximab-dependent ADCP was per- immune cells (Supplementary Fig. S6A–S6D). We found that formed by NK cells, either if freshly isolated from the blood, þ slan monocytes purified from HDs (Fig. 5D; Supplementary or if treated with DCM. Altogether, data suggest that, although þ Fig. S6A, left) and patients with DLBCL (Fig. 5E; Supplementary slan monocytes exert a potent rituximab-mediated ADCC, þ Fig. S6A, right) performed a very efficient and comparable ritux- DCM- or IL34-derived slan macrophages display instead an imab-dependent ADCC of Daudi cells, being much more potent efficient RTX-ADCP of lymphoma cells, but completely lack þ than CD14 monocytes (Fig. 5D and E; Supplementary Fig. S6C). ADCC competence. As expected, NK cells from either HDs (Fig. 5D; Supplementary Fig. S6B, left) or patients with DLBCL (Fig. 5E; Supplementary Fig. Expression of FcgRIII/CD16, FcgRII/CD32, and FcgRI/CD64 þ S6B, right) exerted the strongest rituximab-mediated ADCC, by slan cells whereas neutrophils were completely inefficient (Fig. 5D and E; Because FcgRs play an essential role in mediating both ADCC Supplementary Fig. S6D). In addition, rituximab-mediated and ADCP (29, 30), we investigated CD16, CD32, and CD64 þ þ ADCC was more efficient when suboptimal numbers of slan surface expression in slan -cells and, as comparison, in NK cells, monocytes were cocultured with optimal numbers of NK cells obtained under our various experimental conditions. We found

3552 Cancer Res; 78(13) July 1, 2018 Cancer Research

Role of slanþ Cells in B-cell Lymphoma

Figure 5. slanþ monocytes are significantly augmented in the blood of patients with DLBCL and, ex vivo, perform an efficient rituximab-dependent ADCC of B cells. A–C, Dot plots show the absolute number (top), and the percentage within PBMCs (bottom), of blood slanþ monocytes (A), NK cells (B), and CD14þ monocytes (C) in HDs (white circles) and in newly diagnosed patients with DLBCL (red circles). , P < 0.01 and , P < 0.001 by two-tailed Student t test. D and E, Blood NK cells, slanþ monocytes, CD14þ monocytes, and neutrophils isolated from HDs (D) or patients with DLBCL (E) were cultured with Daudi cells, previously treated with or without 5 mg/mL aCD20 mAbs at a 10:1 E:T ratio. After 4 hours, cytotoxicity by effector cells was investigated and calculated as þ specified in Materials and Methods. P < 0.05; , P < 0.01; and , P < 0.001 by one-way ANOVA. For HDs in D, n ¼ 23 for NK cells, n ¼ 22 for slan monocytes, n ¼ 12 for CD14þ monocytes, and n ¼ 4 for neutrophils. For patients with DLBCL in E, n ¼ 15 for NK cells, n ¼ 13 for slanþ monocytes, n ¼ 9 for CD14þ monocytes, and n ¼ 4 for neutrophils. F, A total of 104–5 104 NK cells from HDs were cocultured for 4 hours with 5 103 Daudi, alone (white columns) or in the presence of 5 103 (light gray columns) or 104 (dark gray columns) HD slanþ monocytes. Bar graph reports the percentage (mean SEM; n ¼ 3) of the specific lysis of aCD20-coated tumor cells after subtraction of the percentage of lysis of untreated tumor cells (net lysis). , P < 0.05 and , P < 0.001 by one-way ANOVA. G and H, A total of 5 104 slanþ monocytes (G) and NK cells (H) from HDs (gray columns) or patients with DLBCL (red columns) were cultured with 5 103 healthy or neoplastic B cells isolated from NHL samples, previously treated with or without 5 mg/mL aCD20 mAbs. ADCC was then determined after 4 hours of coculture. n ¼ 4 for normal blood B cells and n ¼ 17 for nodal neoplastic B cells as target cells for HD slanþ monocytes (G) or NK cells (H); n ¼ 3 for nodal neoplastic B cells as target cells for DLBCL slanþ monocytes (G) or NK cells (H). n.d., not done.

www.aacrjournals.org Cancer Res; 78(13) July 1, 2018 3553

Vermi et al.

Figure 6. þ þ ADCC, ADCP, and expression of FcgRs by the various slan cell types under investigation. A, A total of 5 104 slan monocytes (left part of the graph) or 5 104 NK cells (right part of the graph) treated with DCM were cultured with 5 103 calcein-labeled healthy (n ¼ 3), neoplastic B cells (n ¼ 4–7), or Daudi cells (n ¼ 6–8), in the absence (white columns) or the presence (black columns) of aCD20 to perform ADCC. B, A total of 5 104 slanþ monocytes from HDs untreated (left part of the graph; n ¼ 2) or treated with DCM (right part of the graph) were cultured with 5 103 PKH26-labeled normal (n ¼ 3–4), neoplastic B cells (n ¼ 3–4), or Daudi cells (n ¼ 5–11) in the absence (white columns) or in the presence (black columns) of aCD20. ADCP was then determined after 4 hours of coculture as detailed in Materials and Methods. C, Confocal images (from two independent experiments) showing ADCP of calcein-labeled Daudi cells by DCM-treated slanþ cells. Light blue nuclei (by DAPI staining) of slanþ macrophages and green Daudi cells (by calcein staining) are included as either single or merged fluorescent images. D, A total of 5 104 slanþ monocytes from HDs treated with 100 ng/mL IL34 (left part of the graph) or 30 ng/mL GM-CSF plus 20 ng/mL IL4 (right part of the graph) were cultured with 5 103 PKH26-labeled normal, neoplastic B cells or Daudi cells in the absence (white columns) or in the presence (gray columns) of aCD20 for ADCP performance. E and F, CD16, CD32, and CD64 expression were analyzed by flow cytometry either in slanþ monocytes (E; white columns) and NK cells (E; black columns) gated within freshly isolated PBMCs from HDs (n ¼ 5–9), or in slanþ monocytes from HDs treated with DCM (F; striped columns; n ¼ 4–5), IL34 (F; black columns; n ¼ 3), and GM-CSF plus IL4 (F; gray columns; n ¼ 3). Values for each FcgR are reported as MFI after subtracting the fluorescence given by the respective isotype control mAb. G, Bar graph shows the effect of 10 mg/mL anti FcgR-blocking antibodies (clone 3G8 for CD16; clone AT10 for CD32; clone 10.1 for CD64; and clone MG1–45 for irrelevant isotype control antibodies) on ADCP of PKH26-labeled B-cell targets by DCM-treated slanþ monocytes (n ¼ 4–6). Data are expressed as percentage inhibition of ADCP.

3554 Cancer Res; 78(13) July 1, 2018 Cancer Research

Role of slanþ Cells in B-cell Lymphoma

Figure 7. Characterization of CD16þ and CD56þ cells in DLBCL. Frequency of the cases containing slanþ cells, CD56þ NK cells and CD16þ cells in CH2 (A), CH3 (B), and CH2/CH3 (C) DLBCL cohorts (S0, S1, and S2 stand for, respectively, score ¼ 0, ¼ 1, and ¼ 2, as indicated in the Materials and Methods section). Sections are from three human nodal DLBCL cases (respectively, D–F; G–I;andJ–L) representing recurrent patterns of expression of CD16 (D, G, and J), CD56 (E, H,andK), slan (F, I,andL), CD68 (O), and CSF1R (O), as labeled. Double staining for CD16 and the slan marker (M and N) illustrates DLBCL cases containing a dominant component of slanþ/CD16 (M)or slanþ/CD16þ (N). slanþ/CD16þ also coexpress CD163, as shown by slan/CD163 double stain (inset in N). Most of CD68þ-macrophages in DLBCL coexpress CSF1R, as shown in a representative case (O). Sections are counterstained with Meyer's hematoxylin. Original magniﬁcation, 100 (D–L; scale bar, 200 mm), 200 (M and N; scale bar, 100 mm), 400 (O; scale bar, 50 mm), 600 (insets).

þ that CD16 is highly expressed in slan monocytes and NK cells particularly CD16, yet in cooperation with both CD64 and CD32, from either HDs or patients with DLBCL (Fig. 6E), consistent with is involved in rituximab-dependent ADCP of B-cell targets by þ þ their strong ADCC capacity. CD32 was expressed by slan mono- DCM- and IL34-treated slan monocytes (Fig. 6G). Altogether, þ cytes only, whereas CD64 was absent in both populations (Fig. data indicate that DCM- and IL34-treated slan monocytes switch þ 6E). Compared with slan monocytes, DCM- and IL34-treated into ADCP-effector cells dependent on the combined action of þ slan monocytes were found to upregulate the expression of both CD16 with CD64 and CD32. CD16 and CD32 and to de novo express CD64 (Fig. 6F), thus þ þ displaying an FcgR expression pattern mirroring the DLBCL slan DCM-treated CD14 monocytes exert a rituximab-mediated þ macrophages (Fig. 4C, blue columns). In contrast, GM-CSF plus ADCP equivalent to that by DCM-treated slan monocytes, but þ IL4-treated slan monocytes were found to maintain the expres- using different FcgRs þ þ sion of CD32, but to lose CD16 (Fig. 6F), similarly to the DLBCL In comparative experiments, CD14 and slan monocytes þ slan DCs (Fig. 4C, green columns). Remarkably, experiments incubated with either DCM or IL34 were found to perform using speciﬁcFcgR-neutralizing mAbs demonstrated that equivalent levels of rituximab-mediated ADCP of Daudi cells

www.aacrjournals.org Cancer Res; 78(13) July 1, 2018 3555

Vermi et al.

þ (Supplementary Fig. S7B), but no significant rituximab-mediated Of note, the two slan populations were found in DLBCL cases, þ ADCC (Supplementary Fig. S7C). Moreover, CD14 monocytes irrespectively of their clinical and molecular features (Supple- incubated with DCM were found to express higher CD32 and mentary Table S10). Together, these findings suggest that cofac- þ CD64, but lower CD16, levels than DCM-treated slan mono- tors other than the density of cells involved in ADCC/ADCP cytes (Supplementary Fig. S7D). Accordingly, neutralization interfere with the rituximab-mediated clinical response. þ assays performed in DCM-treated CD14 monocytes uncovered that CD32 and CD64 represent the FcgRs mostly responsible Discussion of rituximab-dependent ADCP of B-cell targets (Supplementary þ Fig. S7E). In this study, we report that slan cells are detectable in nodal and extranodal localizations of several types of NHL, especially þ Heterogeneity and clinical significance of CD16-expressing DLBCL. We were stunned by the high density of slan cells in cells in DLBCL a remarkable fraction of DLBCL cases, in which intratumoral þ DLBCL is an aggressive, heterogeneous B-cell malignancy that slan cells closely interact with lymphoma cells and, with our includes different biological subtypes. Accordingly, patients with surprise, displayed recurrent patterns of lymphoma cell phago- DLBCL might be cured, relapse, or be resistant to standard cytosis. Thus, we investigated the DLBCL as a model to get insights þ R-CHOP (31, 32). We could not detect significant differences in on the role of slan monocytes, and their tissue-differentiated þ slan -cell density by subgrouping CH2 and CH3 for major clinical counterparts, in cancer. By using in vitro systems, as well as and demographic variables (Supplementary Table S8). In addi- analyzing peripheral blood and lymph node tissues from patients tion, no differences were observed in term of PFS, based on the with DLBCL, we uncovered a dichotomous, mutually exclusive þ þ slan -cell content, in patients homogeneously treated with fate of slan monocytes in lymphoma tissues, differentiating into R-CHOP o R-CHOP–like regimens (Supplementary Table S9). either DCs or macrophages. Finally, we could substantiate their We expanded our IHC analysis to cells potentially involved relevant ADCC and ADCP capacity. þ in RTX-ADCC/ADCP by including all CD16 cells, as well as Based on morphology, clinical presentation, and molecular þ CD56 -NK cells. Based on a three-tiered density score, we uncov- features, DLBCL encompasses a highly heterogeneous disease þ ered that the large majority of DLBCL samples contained CD16 spectrum, as clearly indicated by the recent literature (31–33). þ cells (respectively 95.6% for CH2; 76.5% for CH3; mean 85.1%) Our data demonstrate that slan cells are observed across any þ with a large fraction of cases showing score 2 (respectively 72.1% DLBCL subtype (14, 16, 21, 22). The high slan -cells' density for CH2; 41.2% for CH3; mean 55.3%) (Fig. 7A–N; Supplemen- observed in a substantial fraction of DLBCL supports the notion of þ þ tary Table S8). Significantly, a higher density of CD16 cells was a steady recruitment of slan monocytes from the circulation. associated with a more advanced disease stage (P ¼ 0.0096; Based on in situ analysis of lymph node sections of DLBCL, we þ Supplementary Table S8). In contrast, DLBCL cases containing could detect a population of smaller slan cells, with monocytoid þ þ CD56 NK cells represented a minor fraction of DLBCL (respec- morphology. This slan population was localized within and tively 25.6% for CH2; 58.8% for CH3; mean 43.7%), with only around the high endothelial venules, suggesting recent cell immi- 7.5% (respectively 0% for CH2; 13.7% for CH3) of the cases gration from the blood, fueled by an increased marrow output þ showing score 2 (Fig. 7A–C; Supplementary Table S8). None of (6). Accordingly, slan monocytes in patients with DLBCL are the biomarkers tested (e.g., slan, CD16 and CD56) was able, alone significantly increased as compared with HDs. It is conceivable or in combination, to predict a better outcome to rituximab- that their increase in the circulation, followed by nodal and containing regimens (Supplementary Table S9). extranodal tissue infiltration, contribute to sustain the high proin- þ A significant fraction of DLBCL cases contained CD16 cells flammatory stromal-derived signatures documented in DLBCL þ not coexpressing slan (Fig. 7; Supplementary Table S5; Supple- (7, 8). Consistently, we have reported that localization of slan mentary Fig. S8A–S8T), as also proved by serial sections of two cells in metastatic lymph nodes of patients with colon cancer is þ DLBCL cases (Supplementary Fig. S8A–S8H and S8K–S8R, respec- paralleled by a slight increase in the frequency of slan monocytes tively) and double staining (Fig. 7M–N; Supplementary Fig. S8I, in the patients' blood, as well as the maintenance of their proin- þ S8J, S8S and S8T). We further characterized these CD16 cells and flammatory functions (1). found that most of them colocalize with macrophages on serial Most of the patients with DLBCL are treated with rituximab- sections (Supplementary Fig. S8F–S8H and S8P–S8R) and coex- containing regimens (11, 12). Rituximab is an mAb targeting press the macrophage markers CD68 (Supplementary Fig. S8I and CD20-expressing B cells. Rituximab activity is dependent on the S8S), CD163 (Supplementary Fig. S8J and S8T), and CSF1R blockade of the CD20 signaling on tumor cells but also on þ (Fig. 7O). A minor CD16 component was represented by round immune-dependent elimination of lymphoma cells by ADCC to oval cells (Fig. 7J and inset) and polymorphs (Fig. 7G and and ADCP. NK cells are the main effectors of rituximab-mediated inset). Significantly, as verified in nine DLBCL cases, the observed ADCC via their CD16 (25, 26). Although data are still limited, the macrophage-like phenotype of slan-expressing cells, such as the involvement of nonclassical monocytes as effector cells in ADCC, þ þ coexpression of CD163, was restricted to CD16 /slan cells by virtue of their CD16 expression, has started to be recognized in þ (n ¼ 9; Fig. 7N and inset), whereas CD16 /slan cells were largely solid tumors (34). Moreover, data indicate that circulating þ negative for CD163, likely corresponding to the DC-like coun- CD16 monocytes from HDs and patients with B-CLL exert terpart. By double IHC for the slan antigen and CD163, we ADCC activities toward Burkitt lymphoma (Raji), lung (A549), extended this analysis subgrouping DLBCL samples based on the and breast (SKBR3) adenocarcinoma cell lines (28). Remarkably, þ þ þ dominant polarization of slan cells, respectively, in CD163 also slan monocytes have been shown to perform some level þ þ slan macrophages (n ¼ 21) and CD163 slan DCs (n ¼ 45). of ADCC toward cancer cell lines via CD16 and CD32, We could not detect significant differences in terms of patient although this observation was limited to a single study (27). survival between the two subgroups (Supplementary Table S9). Here, we have significantly extended this finding by showing that

3556 Cancer Res; 78(13) July 1, 2018 Cancer Research

Role of slanþ Cells in B-cell Lymphoma

þ slan monocytes from both HDs and patients with DLBCL are to perform rituximab-dependent ADCC. This functional skewing þ highly efficient in rituximab-mediated ADCC toward a variety of is consistent with our observation that nodal slan cells can þ CD20-expressing B-cell targets, including freshly purified lym- express CD163 and contain apoptotic bodies (or even PAX-5 þ þ phoma cells. Furthermore, by comparative analysis, slan mono- MYC nuclei from lymphoma cells). The latter finding strongly þ cytes are significantly more efficient than CD14 monocytes or supports their intrinsic propensity to phagocytose lymphoma þ neutrophils in rituximab-mediated ADCC. In addition, slan cells (even in the absence of rituximab, as revealed by IHC), monocytes and NK cells cooperate in performing rituximab- when macrophage differentiation is favored by the DLBCL micro- þ mediated ADCC, in agreement with their reciprocal activation, environment. Moreover, our data showing that slan DC-like cells þ in terms of IFNg production or tumoricidal activities in the case (generated by culturing slan monocytes with GM-CSF plus IL4; þ of NK cells (35), or IL12p70 production in the case of slan ref. 5) are unable to perform both ADCC and ADCP reinforce the monocytes (36). Together with these previous observations view that only the monocyte stage and their terminal differenti- þ (35, 36), our new findings expand the role of slan monocytes ation into macrophage-like cells plays a role in target cell clearance as effector cells in various clinical settings in which ADCC might via rituximab in DLBCL. Although definitive data on ADCC þ play a significant role. We would speculate that, in addition to versus ADCP efficiency by human macrophages from CD16 perform ADCC of circulating CD20-expressing cells, recently monocytes are missing, our results suggest their poor efficiency þ lymphoma-immigrating slan monocytes might also contribute as ADCC effectors, as herein and previously (39) observed for þ to ADCC of lymphoma cells under rituximab therapy. macrophages derived from CD14 monocytes. Accordingly, Based on the pattern of expression of the slan marker, our ADCC activity observed in human macrophages is low (40), and analysis on reactive lymph nodes and NHL confirms previous often measured at high E:T ratios (i.e., 50:1; ref. 41), or at long þ observation that slan cells are distinct from other specialized time points (i.e., 18 hours or even longer; refs. 40, 41). In contrast, þ macrophages and DCs (1). This study provides an additional level we found that ADCP by slan macrophages is already maximal þ of knowledge on the tissue fate of slan monocytes. By combin- within few hours of coincubation with target cells. Interestingly, þ þ þ ing IHC and flow cytometry analysis, we found that slan cells we found that slan and CD14 monocyte–derived macrophages within the lymphoma tissue displayed either macrophage or perform substantially similar levels of rituximab-mediated DC features. These two phenotypes likely reflect the dominant ADCP. However, we could demonstrate that they use a different cytokine milieu of the lymphoma tissues. Interestingly, nodal set of FcgRs, namely CD16 in combination with CD32 and þ þ þ CD163 /CD14hi/CD64hi/CD16hi slan macrophages are CD64 in the case of DCM-treated slan monocytes, and CD64/ þ þ þ similar to slan cells obtained upon incubation of slan mono- CD32 in the case of DCM-treated CD14 monocytes. It is of note þ cytes with IL34 for 5 days (5) and also resemble the slan cells that ADCP, including phagocytosis of rituximab-coated target residing in the lamina propria of colon tissue from patients with tumor B cells, represents an essential killing mechanism adopted Crohn disease (37). However, neither did we detect sizeable by macrophages with a potential clinical relevance (42–44). Of amounts of IL34 mRNA and protein in nodal DLBCL contain- note, multiple evidence of this study indicates that CD16 expres- þ ing slan macrophages, nor could we block the prosurvival sion is retained during macrophage differentiation, but is down- þ effect of DCM on slan monocytes by using GW2580 (a CD115 regulated following a DC program. tyrosine kinase inhibitor; ref. 38), suggesting that macrophage Gene expression profiling on DLBCL has shown that "stromal- þ polarization of slan monocytes within infiltrated lymph nodes 1" (7) or "host–response" (8) signatures, partially expressed by is driven by cytokines and growth factors other than IL34 and monocytes/macrophages and including the TNF pathway, are M-CSF. Likewise, nodal CD163 /CD14low/CD64 /CD16low predictors of good outcome to R-CHOP (7). In this study, how- þ þ þ slan DCs, similarly to slan monocytes incubated in the pres- ever, the density of slan cells could not predict the outcome of ence of GM-CSF plus IL4 for 5 days, resembled instead the patients with DLBCL to rituximab-containing regimens, even slanDCs from human tonsils (5). However, according to their when considering their dominant polarization toward macro- negativity for CD83, CD208, and CCR7, they appear largely in phages or DCs. We found that CD16, which is known to be the þ an immature state. Clarifying whether such an immature DC critical FcgR mediating ADCC by both NK cells (25) and slan phenotype is coupled with immunosuppressive functions monocytes (27), is also involved, in cooperation with CD64 and þ þ requires further studies on DLBCL purified slan cells. Interest- CD32, in ADCP by DCM-generated slan macrophages. It should þ ingly, when slan monocytes were cultured in conditioned medi- be noted that the molecular mechanisms dictating the switch of um from nodal DLBCL tissue (DCM), they regularly acquired CD16 from an ADCC to an ADCP receptor are presently þ a macrophage-like phenotype. Therefore, we propose that soluble unknown. Only a fraction of slan cells in DLBCL express CD16, factors are responsible for the macrophage terminal differentia- and they are certainly macrophages. Although still neglected by þ tion of slan monocytes (at least in vitro), whereas complex cell- the literature, our data indicate that infiltration of high amounts þ to-cell interactions are required to mediate a DC fate within of CD16 cells is recurrent in DLBCL cases. However, CD16 the heterogeneous DLBCL microenvironments. These data, expression (alone or combined with slan and CD56 markers) once again, testify for the extreme developmental plasticity of was unable to predict a better response to rituximab-containing þ slan monocytes based on the tissue microenvironment in regimens in terms of PFS; on the contrary, it was associated with þ which they locate. advanced disease stage. We found that CD16 cells in DLBCL As a limitation of this study, we could not purify sufficient largely overlap with CD68 and CD163, marking the pool of þ slan cells from lymphoma tissue to perform functional assays. tumor-associated macrophages (TAM). The latter population is þ To overcome such a limitation, we tested slan macrophages more frequently associated with worse outcome in patients with generated in vitro using DCM as differentiation medium. We cancer (45) and represents a relevant component also of the þ found that slan macrophages were very effective in rituximab- DLBCL microenvironment (46, 47). Of note, most of the TAMs þ mediated ADCP, but, unlike slan monocytes, totally incapable in DLBCL, including those expressing the slan marker, coexpress

www.aacrjournals.org Cancer Res; 78(13) July 1, 2018 3557

Vermi et al.

CSF1R. This finding suggests that their fine modulation, by Analysis and interpretation of data (e.g., statistical analysis, biostatistics, combining CSF1R blockade (48–50) with compounds mediat- computational analysis): W. Vermi, A. Micheletti, S. Costa, S. Calza, S. Pileri, ing ADCC/ADCP effector functions, can become a clinically L. Lorenzi, S. Lonardi, M.A. Cassatella Writing, review, and/or revision of the manuscript: W. Vermi, A. Micheletti, relevant option in DLBCL and, by extension, to other cancer C. Tecchio, L. Lorenzi, S. Lonardi, M.A. Cassatella types. Overall, our results highlight a potentially relevant role Administrative, technical, or material support (i.e., reporting or organizing þ of slan cells in the NHL microenvironment, particularly in data, constructing databases): A. Micheletti, S. Pileri, A. Zamo, F. Facchetti, DLBCL. Of note, within DLBCL tissue, numerous CD16-expres- M.A. Cassatella sing macrophages are found, whose function remains largely Study supervision: W. Vermi, M.A. Cassatella þ unknown. Here, we could substantiate a new role of slan þ Acknowledgments monocytes and DLBCL slan macrophages as effectors of anti- This work was supported by grants from Associazione Italiana per la Ricerca body-mediated tumor targeting in human cancers. Prospective sul Cancro – Italy to M.A. Cassatella (AIRC, IG-15454) and W. Vermi (AIRC, clinical studies should incorporate the slan marker to further IG-15378), as well as from University of Verona (Ricerca di Base) to M.A. þ extend the knowledge on the role slan cells as effector in DLBCL. Cassatella. A. Micheletti and G. Finotti are recipients of fellowships from, respectively, "Fondazione Umberto Veronesi" and "Fondazione Italiana per la Disclosure of Potential Conflicts of Interest Ricerca sul Cancro," whereas S. Lonardi. is supported by "Fondazione Beretta" S. Pileri has received speakers bureau honoraria from Takeda. No potential (Brescia, Italy). We thank Luisa Benerini Gatta for analysis of IL34 expression in conflicts of interest were disclosed by the other authors. human DLBCL.

Authors' Contributions The costs of publication of this article were defrayed in part by the Conception and design: W. Vermi, A. Micheletti, S. Lonardi, M.A. Cassatella payment of page charges. This article must therefore be hereby marked Development of methodology: A. Micheletti, G. Finotti, F. Calzetti, M. Bugatti, advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate C. Agostinelli, P. Balzarini, G. Todeschini, S. Lonardi this fact. Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Finotti, C. Tecchio, F. Calzetti, S. Costa, M. Bugatti, Received August 5, 2017; revised March 2, 2018; accepted May 3, 2018; S. Pileri, A. Tucci, G. Rossi, L. Furlani, A. Zamo, L. Lorenzi published ﬁrst May 10, 2018.

References 1. Vermi W, Micheletti A, Lonardi S, Costantini C, Calzetti F, Nascimbeni R, 12. Zelenetz AD, Gordon LI, Wierda WG, Abramson JS, Advani RH, Andreadis et al. slanDCs selectively accumulate in carcinoma-draining lymph nodes CB, et al. Diffuse large B-cell lymphoma version 1.2016. J Natl Compr and marginate metastatic cells. Nat Commun 2014;5:3029. Cancer Netw 2016;14:196–231. 2. Sch€akel K, Kannagi R, Kniep B, Goto Y, Mitsuoka C, Zwirner J, et al. 13. Zinzani PL, Dirnhofer S, Sabattini E, Alinari L, Piccaluga PP, Stefoni V, et al. 6-Sulfo LacNAc, a novel carbohydrate modification of PSGL-1, defines Identification of outcome predictors in diffuse large B-cell lymphoma. an inflammatory type of human dendritic cells. Immunity 2002;17: Immunohistochemical profiling of homogeneously treated de novo 289–301. tumors with nodal presentation on tissue micro-arrays. Haematologica 3. Sch€akel K. Dendritic cells–why can they help and hurt us. Exp Dermatol 2005;90:341–7. 2009;18:264–73. 14. Hans CP, Weisenburger DD, Greiner TC, Gascoyne RD, Delabie J, Ott G, 4. Hofer TP, Zawada AM, Frankenberger M, Skokann K, Satzl AA, Gesierich W, et al. Confirmation of the molecular classification of diffuse large B-cell et al. slan-defined subsets of CD16-positive monocytes: impact of granu- lymphoma by immunohistochemistry using a tissue microarray. Blood lomatous inflammation and M-CSF receptor mutation. Blood 2015;126: 2004;103:275–82. 2601–10. 15. Johnson NA, Savage KJ, Ludkovski O, Ben-Neriah S, Woods R, Steidl C, 5. Micheletti A, Finotti G, Calzetti F, Lonardi S, Zoratti E, Bugatti M, et al. et al. Lymphomas with concurrent BCL2 and MYC translocations: the Correction: slanDCs/M-DC8þ cells constitute a distinct subset of dendritic critical factors associated with survival. Blood 2009;114:2273–9. cells in human tonsils. Oncotarget 2016;7:66323. 16. Green TM, Young KH, Visco C, Xu-Monette ZY, Orazi A, Go RS, et al. 6. Mimiola E, Marini O, Perbellini O, Micheletti A, Vermi W, Lonardi S, et al. Immunohistochemical double-hit score is a strong predictor of outcome in Rapid reconstitution of functionally active 6-sulfoLacNAc(þ) dendritic patients with diffuse large B-cell lymphoma treated with rituximab plus cells (slanDCs) of donor origin following allogeneic haematopoietic stem cyclophosphamide, doxorubicin, vincristine, and prednisone. J Clin Oncol cell transplant. Clin Exp Immunol 2014;178:129–41. 2012;30:3460–7. 7. Lenz G, Wright G, Dave SS, Xiao W, Powell J, Zhao H, et al. Stromal 17. Neri S, Mariani E, Meneghetti A, Cattini L, Facchini A. Calcein-acetyox- gene signatures in large-B-cell lymphomas. N Engl J Med 2008;359: ymethyl cytotoxicity assay: standardization of a method allowing addi- 2313–23. tional analyses on recovered effector cells and supernatants. Clin Diagn Lab 8. Monti S, Savage KJ, Kutok JL, Feuerhake F, Kurtin P, Mihm M, et al. Immunol 2001;8:1131–5. Molecular profiling of diffuse large B-cell lymphoma identifies robust 18. Van Amersfoort ES, Van Strijp JA. Evaluation of a flow cytometric fluores- subtypes including one characterized by host inflammatory response. cence quenching assay of phagocytosis of sensitized sheep erythrocytes by Blood 2005;105:1851–61. polymorphonuclear leukocytes. Cytometry 1994;17:294–301. 9. Gomez-Gelvez JC, Salama ME, Perkins SL, Leavitt M, Inamdar KV. 19. Sehn LH, Berry B, Chhanabhai M, Fitzgerald C, Gill K, Hoskins P, et al. The Prognostic impact of tumor microenvironment in diffuse large B-cell revised International Prognostic Index (R-IPI) is a better predictor of lymphoma uniformly treated with R-CHOP chemotherapy. Am J Clin outcome than the standard IPI for patients with diffuse large B-cell Pathol 2016;145:514–23. lymphoma treated with R-CHOP. Blood 2007;109:1857–61. 10. Riihijarvi S, Fiskvik I, Taskinen M, Vajavaara H, Tikkala M, Yri O, et al. 20. Rosenwald A, Wright G, Chan WC, Connors JM, Campo E, Fisher RI, et al. Prognostic influence of macrophages in patients with diffuse large B-cell The use of molecular profiling to predict survival after chemotherapy for lymphoma: a correlative study from a Nordic phase II trial. Haematologica diffuse large-B-cell lymphoma. N Engl J Med 2002;346:1937–47. 2015;100:238–45. 21. Swerdlow SH, Campo E, Pileri SA, Harris NL, Stein H, Siebert R, et al. The 11. Coiffier B, Lepage E, Briere J, Herbrecht R, Tilly H, Bouabdallah R, et al. 2016 revision of the World Health Organization classification of lymphoid CHOP chemotherapy plus rituximab compared with CHOP alone in neoplasms. Blood 2016;127:2375–90. elderly patients with diffuse large-B-cell lymphoma. N Engl J Med 2002; 22. Hu S, Xu-Monette ZY, Tzankov A, Green T, Wu L, Balasubramanyam A, et al. 346:235–42. MYC/BCL2 protein coexpression contributes to the inferior survival of

3558 Cancer Res; 78(13) July 1, 2018 Cancer Research

Role of slanþ Cells in B-cell Lymphoma

activated B-cell subtype of diffuse large B-cell lymphoma and demonstrates 37. Ogino T, Nishimura J, Barman S, Kayama H, Uematsu S, Okuzaki D, et al. high-risk gene expression signatures: a report from The International Increased Th17-inducing activity of CD14þ CD163 low myeloid cells in DLBCL Rituximab-CHOP Consortium Program. Blood 2013;121: intestinal lamina propria of patients with Crohn's disease. Gastroenterol- 4021–31. ogy 2013;145:1380–91. 23. Weiner GJ. Rituximab: mechanism of action. Semin Hematol 2010;47: 38. Lin H, Lee E, Hestir K, Leo C, Huang M, Bosch E, et al. Discovery of a 115–23. cytokine and its receptor by functional screening of the extracellular 24. Gul N, van Egmond M. Antibody-dependent phagocytosis of tumor cells proteome. Science 2008;320:807–11. by macrophages: a potent effector mechanism of monoclonal antibody 39. Leidi M, Gotti E, Bologna L, Miranda E, Rimoldi M, Sica A, et al. M2 therapy of cancer. Cancer Res 2015;75:5008–13. macrophages phagocytose rituximab-opsonized leukemic targets more 25. Dall'Ozzo S, Tartas S, Paintaud G, Cartron G, Colombat P, Bardos P, et al. efficiently than m1 cells in vitro. J Immunol 2009;182:4415–22. Rituximab-dependent cytotoxicity by natural killer cells: influence of 40. Herter S, Birk MC, Klein C, Gerdes C, Umana P, Bacac M. Glycoengineering FCGR3A polymorphism on the concentration-effect relationship. Cancer of therapeutic antibodies enhances monocyte/macrophage-mediated Res 2004;64:4664–9. phagocytosis and cytotoxicity. J Immunol 2014;192:2252–60. 26. Beum PV, Lindorfer MA, Taylor RP. Within peripheral blood mononuclear 41. Liesveld JL, Frediani KE, Winslow JM, Duerst RE, Abboud CN. Cytokine cells, antibody-dependent cellular cytotoxicity of rituximab-opsonized effects and role of adhesive proteins and Fc receptors in human macro- Daudi cells is promoted by NK cells and inhibited by monocytes due to phage-mediated antibody dependent cellular cytotoxicity. J Cell Biochem shaving. J Immunol 2008;181:2916–24. 1991;45:381–90. 27. Schmitz M, Zhao S, Sch€akel K, Bornhauser M, Ockert D, Rieber EP. Native 42. Chao MP, Alizadeh AA, Tang C, Myklebust JH, Varghese B, Gill S, et al. Anti- human blood dendritic cells as potent effectors in antibody-dependent CD47 antibody synergizes with rituximab to promote phagocytosis and cellular cytotoxicity. Blood 2002;100:1502–4. eradicate non-Hodgkin lymphoma. Cell 2010;142:699–713. 28. Yeap WH, Wong KL, Shimasaki N, Teo EC, Quek JK, Yong HX, et al. CD16 is 43. Church AK, VanDerMeid KR, Baig NA, Baran AM, Witzig TE, Nowakowski indispensable for antibody-dependent cellular cytotoxicity by human GS, et al. Anti-CD20 monoclonal antibody-dependent phagocytosis of monocytes. Sci Rep 2016;6:34310. chronic lymphocytic leukaemia cells by autologous macrophages. Clin Exp 29. Nimmerjahn F, Gordan S, Lux A. FcgammaR dependent mechanisms of Immunol 2016;183:90–101. cytotoxic, agonistic, and neutralizing antibody activities. Trends Immunol 44. Uchida J, Hamaguchi Y, Oliver JA, Ravetch JV, Poe JC, Haas KM, et al. The 2015;36:325–36. innate mononuclear phagocyte network depletes B lymphocytes through 30. Hogarth PM, Pietersz GA. Fc receptor-targeted therapies for the treatment of Fc receptor-dependent mechanisms during anti-CD20 antibody immuno- inflammation, cancer and beyond. Nat Rev Drug Discov 2012;11:311–31. therapy. J Exp Med 2004;199:1659–69. 31. Sehn LH, Gascoyne RD. Diffuse large B-cell lymphoma: optimizing out- 45. Biswas SK, Allavena P, Mantovani A. Tumor-associated macrophages: come in the context of clinical and biologic heterogeneity. Blood functional diversity, clinical significance, and open questions. Semin 2015;125:22–32. Immunopathol 2013;35:585–600. 32. Moffitt AB, Dave SS. Clinical applications of the genomic landscape of 46. Kridel R, Steidl C, Gascoyne RD. Tumor-associated macrophages in diffuse aggressive non-hodgkin lymphoma. J Clin Oncol 2017;35:955–62. large B-cell lymphoma. Haematologica 2015;100:143–5. 33. Schneider C, Pasqualucci L, Dalla-Favera R. Molecular pathogenesis of 47. Scott DW, Gascoyne RD. The tumour microenvironment in B cell lym- diffuse large B-cell lymphoma. Semin Diagn Pathol 2011;28:167–77. phomas. Nat Rev Cancer 2014;14:517–34. 34. Romano E, Kusio-Kobialka M, Foukas PG, Baumgaertner P, Meyer C, 48. Galletti G, Scielzo C, Barbaglio F, Rodriguez TV, Riba M, Lazarevic D, Ballabeni P, et al. Ipilimumab-dependent cell-mediated cytotoxicity of et al. Targeting macrophages sensitizes chronic lymphocytic leukemia regulatory T cells ex vivo by nonclassical monocytes in melanoma patients. to apoptosis and inhibits disease progression. Cell Rep 2016;14: Proc Natl Acad Sci U S A 2015;112:6140–5. 1748–60. 35. Schmitz M, Zhao S, Deuse Y, Sch€akel K, Wehner R, Wohner H, et al. 49. Cassier PA, Italiano A, Gomez-Roca CA, Le Tourneau C, Toulmonde M, Tumoricidal potential of native blood dendritic cells: direct tumor cell Cannarile MA, et al. CSF1R inhibition with emactuzumab in locally killing and activation of NK cell-mediated cytotoxicity. J Immunol advanced diffuse-type tenosynovial giant cell tumours of the soft tissue: 2005;174:4127–34. a dose-escalation and dose-expansion phase 1 study. Lancet Oncol 36. Costantini C, Calzetti F, Perbellini O, Micheletti A, Scarponi C, Lonardi S, 2015;16:949–56. et al. Human neutrophils interact with both 6-sulfo LacNAcþ DC and NK 50. Ries CH, Hoves S, Cannarile MA, Ruttinger D. CSF-1/CSF-1R targeting cells to amplify NK-derived IFN{gamma}: role of CD18, ICAM-1, and agents in clinical development for cancer therapy. Curr Opin Pharmacol ICAM-3. Blood 2011;117:1677–86. 2015;23:45–51.

www.aacrjournals.org Cancer Res; 78(13) July 1, 2018 3559

slan+ Monocytes and Macrophages Mediate CD20-Dependent B-cell Lymphoma Elimination via ADCC and ADCP

William Vermi, Alessandra Micheletti, Giulia Finotti, et al.

Cancer Res 2018;78:3544-3559. Published OnlineFirst May 10, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-17-2344

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2018/05/10/0008-5472.CAN-17-2344.DC1

Visual A diagrammatic summary of the major findings and biological implications: Overview http://cancerres.aacrjournals.org/content/78/13/3544/F1.large.jpg

Cited articles This article cites 50 articles, 23 of which you can access for free at: http://cancerres.aacrjournals.org/content/78/13/3544.full#ref-list-1

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/78/13/3544. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.