CXCR4 expression enhances diffuse large B-cell lymphoma dissemination and decreases patient survival María José Moreno,1,2,3 Rosa Bosch,4 Rebeca Dieguez-Gonzalez,1 Silvana Novelli,4 Ana Mozos,5 ¶
Alberto Gallardo,6 Miguel Ángel Pavón,1,3 María Virtudes Céspedes,1,3 Albert Grañena,7 Miguel Alcoceba,8 Oscar Blanco,8 Marcos Gonzalez-Díaz,8 Jorge Sierra,4 Ramon Mangues1,3* and Isolda Casanova1,3*
1
Grup d’Oncogènesi i Antitumorals, lnstitut d’Investigacions Biomèdiques Sant Pau, Barcelona,
Spain. 2Department of Biochemistry and Molecular Biology, Universitat de Barcelona, Spain. 3
CIBER en Bioingeniería, Biomateriales y Nanomecidicina (CIBER-BBN), Barcelona, Spain.
4
Department of Hematology, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain.
5
Department of Pathology, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain.
6
Department of Pathology, Clínica Girona, Girona, Spain.
7
Department of Hematology, Hospital Universitari de Bellvitge, Barcelona, Spain.
¶ Deceased. 8
Department of Hematology and Pathology, IBSAL-University Hospital, Center for Cancer Research-
IBMCC (USAL-CSIC), Salamanca, Spain. *These authors contributed equally to this work
Corresponding author: Ramon Mangues, Grup d’Oncogènesi i Antitumorals (GOA), Biomedical Research Institute Sant Pau (IIB Sant Pau), Sant Antoni Maria Claret 167, Pavelló 11, 2n pis, Barcelona, 08025, Spain. E-mail: [email protected] Conflict of interest: The authors declare no competing financial interests.
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1002/path.4446
This article is protected by copyright. All rights reserved
ABSTRACT
The chemokine receptor CXCR4 has been implicated in the migration and trafficking of malignant B cells in several hematological malignancies. Overexpression of CXCR4 has been identified in hematological tumors, but data concerning the role of this receptor in diffuse large B-cell lymphoma (DLBCL) are lacking. CXCR4 is a marker of poor prognosis in various neoplasms, correlating with metastatic disease and decreased survival of patients. We studied CXCR4 involvement in cell migration in vitro and dissemination in vivo. We also evaluated the prognostic significance of CXCR4 in 94 biopsies of DLBCL patients. We observed that the level of expression of CXCR4 in DLBCL cell lines correlated positively with in vitro migration. Expression of the receptor was also associated with increased engraftment and dissemination, and decreased survival time in NOD/SCID mice. Furthermore, administration of a specific CXCR4 antagonist, AMD3100, decreased dissemination of DLBCL cells in a xenograft mouse model. In addition, we found that CXCR4 expression is an independent prognostic factor for shorter overall survival and progression-free survival in DLBCL patients. These results show that CXCR4 mediates dissemination of DLBCL cells and define for the first time its value as an independent prognostic marker in DLBCL patients.
Keywords: Diffuse large B-cell lymphoma; CXCR4; cell migration; cell dissemination; prognostic marker.
INTRODUCTION Diffuse large B-cell lymphoma (DLBCL) is a clinically and genetically heterogeneous disease. It is the most common subtype of non-Hodgkin lymphoma, accounting for 30% of all cases [1]. The addition of rituximab to standard treatment (R-CHOP: rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) has increased disease-free survival and overall survival of DLBCL patients [2-4]. Nevertheless, overall survival remains poor, ranging from 40 to 50% over 5 years [5]. The prognosis of newly diagnosed DLBCL patients depends on clinical and biological features defined in the International Prognostic Index (IPI) developed almost 20 years ago [6]. As the prognosis of patients within the IPI groups is highly diverse, novel biomarkers with a potential impact on outcome
This article is protected by copyright. All rights reserved
are needed to refine the predictive scores [7]. Biological markers involved in cell signaling and trafficking seem to merit particular attention.
CXCR4 and its ligand SDF-1 (stromal cell-derived factor-1, also designated as CXCL12) are critically involved in the trafficking of normal lymphocytes [8]. CXCR4 functions as a classical chemokine receptor in adults and, together with its ligand, is essential for organogenesis and hematopoiesis. Its expression is detected in several tissues, predominantly in hematopoietic cells [9]. The CXCR4-SDF1 axis is essential for normal B-cell development and is involved in aspects of B-cell homeostasis such as retention of B-cell precursors in bone marrow and B-cell homing to lymph nodes [10, 11].
The receptor and its ligand have been described in the migration and trafficking of malignant B cells in hematological malignancies such as Burkitt lymphoma [12], follicular lymphoma [13] and chronic lymphocytic leukemia [14, 15]. Moreover, CXCR4 has been associated with decreased survival and considered a prognostic marker in several neoplasms, correlating with disease aggressiveness [16]. However, no data are available about its involvement in DLBCL dissemination or its relevance as a prognostic marker in this disease. We aimed to study how the alteration of CXCR4 expression affects cell migration in vitro and dissemination in vivo. In addition, we evaluated the prognostic significance of the receptor in DLBCL patients.
MATERIALS AND METHODS
Cell culture SUDHL-2, RIVA and OCI-Ly10 (activated subtype, ABC), and Toledo (germinal center subtype, GC) are human DLBCL cell lines. Culture media for all cells were supplemented with 10% fetal bovine serum, 1% glutamine, 100 U/ml penicillin/streptomycin (Life Technologies), unless otherwise specified. SUDHL-2, RIVA and Oci-Ly10 cells were cultured with IMDM (Life Technologies) and Toledo cells with RPMI 1640. OCI-Ly10 were supplemented with 10% human serum (Sigma-Aldrich) and 50 µM 2-mercaptoethanol. Toledo cells were obtained from the ATCC, and SUDHL-2, RIVA and
This article is protected by copyright. All rights reserved
OCI-Ly10 cell lines were kindly provided by Dr. L. Pasqualucci (Columbia University, NY). SUDHL-2Sc, OCI-Ly10-Sc, RIVA-Sc and Toledo-Sc cells were obtained from disaggregation of subcutaneous tumors. All cell lines were cultured in a humidified atmosphere at 37ºC in 5% CO2.
RT-PCR mRNA expression from cell lines was measured on a 7900 HT Fast-Real Time PCR System (Applied Biosystems), using pre-designed Taqman® Gene Expression primer and probe assays for CXCR4 (Hs00607978_s1) and HPRT1 (Hs99999909_m1) (Applied Biosystems) as previously described [17]. Experiments were done in triplicate and repeated at least twice. Ct values were normalized against the housekeeping gene HPRT1.
FACS analysis Fluorescence-activated cell sorting (FACS) analysis was performed to detect cell surface expression of CXCR4. One million cells were washed in phosphate-buffered saline containing 0.5% bovine serum albumin (PBS-BSA) and incubated for 30 minutes at 4ºC with PE-Cy5 mouse anti-human CXCR4 monoclonal antibody or PE-Cy5 mouse IgG2a (BD Biosciences) as an isotype control. Unbound antibody was removed by two washes with PBS-BSA. Data acquisition was performed using flow cytometry (FACS Calibur, BD Biosciences) and analyzed by Cell Quest Pro software. Results are expressed as mean fluorescence intensity ± SD.
Lentivirus, infection and cell sorting Lentiviruses containing the pSIN-DUAL-Luc-GFP2 vector were used to infect DLBCL cells, as described previously [18]. GFP positive cells were sorted using the FACSAria cell sorter (BD Biosciences). The Luciferase Assay System kit (Promega) was used following manufacturer’s instructions to confirm expression of the luciferase gene. Sorted cells, expressing GFP and luciferase genes were used for all in vitro and in vivo experiments.
This article is protected by copyright. All rights reserved
Migration Assay Migration assays were performed using 8-μm pore size Transwell (Corning Costar). In brief, 5x10
5
cells were suspended in 100 μl of serum-free media and incubated at 37ºC for 2 hours in presence or absence of the CXCR4 inhibitor AMD3100 (10 µg/mL, Sigma-Aldrich). The lower compartment was filled with 600 µL of serum-free media with or without 100 ng/ml of human SDF-1α (Prospec-Tany TechnoGene). The amount of cells migrating within 24 hours to the lower compartment was quantified using a Scepter 2.0 Handheld Automated Cell Counter (EMD Millipore). Results are expressed as the total number of cells that migrated to the lower chamber ± SD.
CXCR4 internalization and apoptosis analysis Internalization of CXCR4 was evaluated by incubating cells with 100 ng/ml SDF-1α (Prospec-Tany TechnoGene) or 10 μg/ml AMD3100. Cells were then washed twice with PBS and cell surface expression of CXCR4 was detected by flow cytometry. To evaluate whether cells exposed to AMD3100 were undergoing apoptosis, we used the Annexin V-FITC Apoptosis Detection Kit (Calbiochem/EMD Millipore), following the manufacturer’s instructions.
In vivo assays and bioluminescent imaging NOD/SCID female mice (4 weeks old) were obtained from Charles River Laboratories. Mice were housed in microisolator units with sterile food and water ad libitum. Animals were monitored every other day and a loss of 15% in body weight or signs of sickness were considered as the endpoint. Procedures involving mice were approved by the Hospital de la Santa Creu i Sant Pau Animal Ethics Committee according to established guidelines. To perform the subcutaneous conditioning of cells, 10 NOD/SCID mice were inoculated in both flanks with 10x106 of SUDHL-2, RIVA, OCI-Ly10 or Toledo luminescent
cells. After subcutaneous
cell
growth, tumors
were extracted, mechanically
disaggregated and incubated at 37ºC for 15 hours. Cells were then filtered to obtain SUDHL-2-Sc, RIVA-Sc, OCI-Ly10-Sc or Toledo-Sc, and intravenously injected in new groups of mice (20x106 cells per mouse). Results were compared to direct intravenous injection of non-conditioned cells. Engraftment, dissemination, and mouse survival time were evaluated in all groups (6-10 NOD/SCID
This article is protected by copyright. All rights reserved
mice per group). To perform CXCR4 blocking experiments using AMD3100, mice were intravenously injected with 20x106 AMD3100-pretreated (10 µg/ml, 30 minutes at 37ºC) or untreated OCI-Ly10 cells (n=10). Mice injected with AMD3100-pretreated cells were administered daily with intraperitoneal AMD3100 (10 mg/kg), whereas control mice were treated with vehicle (saline solution). We monitored subcutaneous tumor growth and dissemination of intravenously-injected DLBCL cells capturing bioluminescence (BLI) once a week using the IVIS-2000 (Xenogen) imaging system. Mice were anesthetized with 3% isoflurane in oxygen and BLI was captured 10 minutes after intraperitoneal injection of firefly D-luciferin (2.25mg/mouse, Perkin Elmer). BLI pictures were quantified using Living Imaging Software (Version4.2, Xenogen), as described previously [19].
Immunohistochemical staining Immunohistochemical analysis was performed using paraffin-embedded tissue samples, cell lines and sections of infiltrated organs to detect CD10, MUM1, BCL6 (DAKO) or CXCR4 (Abcam) expression. Staining was performed in a DAKO Autostainer Link48 following the manufacturer’s instructions. We quantified CXCR4 expression in cell lines and infiltrated organs by assigning an H-score (0-300), resulting from the product of the intensity (0-3) and the percentage of staining (0-100). We also evaluated if membrane CXCR4 was present or absent in tumor samples of DLBCL patients and used dichotomized values for statistical analysis. DLBCL biopsies with more than 5% of tumor cells showing higher CXCR4 expression than normal B cells from lymphoid tissues were considered positive (Supplementary Figure S1). Two independent observers evaluated all samples using an Olympus BX51 microscope. There was inter-observer agreement in 95% of the samples; the remaining slides were re-evaluated and consensus decisions were made. Images were acquired using an Olympus DP72 digital camera and processed with the Olympus CellD Imaging 3.3 software (Olympus Corporation).
Patients and tissue samples Ninety-four biopsies from patients with primary DLBCL were diagnosed in the Hospital de la Santa Creu i Sant Pau (HSCSP) or Hospital Universitario de Salamanca (HUS) between 2001 and 2012, based on the WHO criteria [1]. All cases were evaluated for CXCR4 expression by
This article is protected by copyright. All rights reserved
immunohistological analysis. Patients with follicular lymphoma or another type of indolent lymphoma transformed into DLBCL were excluded. All patients received rituximab in their treatment; 92.5% were treated with R-CHOP chemotherapy and 7.5% received other chemotherapeutic agents in combination with rituximab. The median clinical follow-up time was 49 months. Table 2 shows the main features of the series. The Institutional Review Boards at HSCSP and HUS approved the study and informed consent was provided according to the declaration of Helsinki and obtained from patients.
Statistical analysis Results are reported as mean ± SD for experiments performed in triplicates and were analyzed using the non-parametric Mann–Whitney U test. Survival rates were estimated by the Kaplan-Meier method and differences between groups were compared using the log-rank test. Overall survival (OS) was measured in mice from the day of cell injection to the end-point. Progression-free survival (PFS) of DLBCL patients was calculated from the onset of treatment until relapse or death. OS was calculated as the time between the onset of treatment and death or date of the last follow-up. Univariate and multivariate analyses were done using the COX proportional hazard model. Correlations between clinico-pathological variables and CXCR4 expression were tested using Fisher’s exact test. Results were considered significantly different at P ≤ 0.05. Statistical calculations were performed using SPSS software (version 21).
This article is protected by copyright. All rights reserved
RESULTS
CXCR4 expression in DLBCL cell lines is associated with SDF-1α-driven migration We evaluated CXCR4 expression in DLBCL cell lines and found that CXCR4 mRNA expression was the highest in OCI-Ly10 cells, followed by Toledo and RIVA cells (Figure 1A). CXCR4 expression in SUDHL-2 cells was undetectable. Expression of the CXCR4 protein was also evaluated by immunohistochemistry and FACS analysis (Supplementary Figure S2) and correlated with the results obtained by RT-PCR. Next, we evaluated the migration induced by SDF-1α in DLBCL cell lines. OCILy10, Toledo and RIVA cells increased their migration capacity in the presence of SDF-1α (Figure 1B). In contrast, SUDHL-2 cell line, which does not express CXCR4, migrated equally in presence or absence of SDF-1α (data not shown). To assess if SDF-1α-driven migration was dependent on CXCR4 expression, we evaluated whether exposure to AMD3100, a CXCR4 antagonist, inhibited this process. We found that AMD3100 significantly inhibited the migration stimulated by SDF-1α in OCILy10, Toledo and RIVA cell lines (Figure 1B).
SDF-1α induces CXCR4 internalization in DLBCL cell lines We investigated the regulation of CXCR4 by the SDF-1α chemokine. We used the DLBCL cell lines OCI-Ly10, Toledo and RIVA, which expressed CXCR4 receptor in their membrane (Figure 1C, Control). When cells were exposed to SDF-1α, the receptor was not detected in membrane (Figure 1C, SDF-1α). After exposure to SDF-1α, CXCR4 was internalized from the membrane showing a dotlike staining (punctuated) in the cell cytosol (Figure 1D). Dot-like immunostaining of CXCR4 was also detected in RIVA and OCI-Ly10 cells after exposure to SDF-1α (data not shown).
CXCR4 expression in DLBCL cell lines correlates with engraftment, dissemination and survival time in xenograft mice We studied the aggressiveness of DLBCL cell lines displaying different levels of CXCR4 expression in xenograft mice. The intravenously injected OCI-Ly10 cells, with a high expression of CXCR4, showed 100% of engraftment, rapid dissemination to lymph nodes (LN), bone marrow (BM) and central nervous system (CNS), and mouse survival time of less than 30 days (Table 1). In contrast, SUDHL-2 cells, which do not express CXCR4, showed 50% of engraftment, and longer mouse survival time
This article is protected by copyright. All rights reserved
(67.3 ± 21.5 days). Moreover, no infiltration of LN was detected and only 33% of the animals presented BM infiltration. Bioluminescent images (BLI) of OCI-Ly10 or SUDHL-2 injected mice were acquired from the first week after cell injection (Figure 2A). In OCI-Ly10 injected mice, BLI was first observed 2 weeks after injection. In SUDHL-2 injected mice, however, the BLI signal was not detected until week 6 post-injection. The infiltration of lymphoma cells observed in BLI images was confirmed by histological analysis (Figure 2B-C, H&E). Immunohistochemical analysis showed that CXCR4 expression of OCI-Ly10 cells was maintained in infiltrated LN, BM and CNS (Figure 2B, CXCR4). Moreover, in LN and BM, CXCR4 showed dot-like cytoplasmic immunostaining indicating that the receptor was internalized from the membrane. In agreement with the lack of CXCR4 expression in SUDHL-2 cells, the receptor was almost undetectable in the infiltrated BM (Figure 2C, CXCR4). Both OCI-Ly10 and SUDHL-2 cells maintained their DLBCL immunophenotype in the infiltrated tissues (Supplementary Figure S3).
Subcutaneous-conditioning increased CXCR4 expression and aggressiveness in RIVA and Toledo cells In previous studies we developed disseminated lymphoma mouse models by performing subcutaneous-conditioning of cells to increase their aggressiveness prior to their intravenous injection [19, 20]. Here, we used this approach to evaluate the effect of subcutaneous-conditioning of DLBCL cells on CXCR4 expression and its impact on their take rate and dissemination in vivo. We observed that CXCR4 expression was increased in subcutaneously-conditioned RIVA and Toledo cells (Supplementary Table S1). Moreover, mice injected with conditioned cells showed a shorter survival time than mice injected with non-conditioned cells (Table 1, Figure 3B and Supplementary Figure S4B). BLI images were acquired from the first week after cell injection and showed that in RIVAinjected mice, the BLI signal first appeared at week 8, whereas in RIVA-Sc-injected mice it appeared at week 5 (Figure 3A). Macroscopic and microscopic infiltrations of LN were detected in RIVA-Sc injected mice (Figure 3C and 3D). BM and CNS were also infiltrated in some mice (Figure 3D and Table 1). In contrast, in the RIVA-injected group the CNS was infiltrated but the BM of vertebral column and femur were not (Table 1). In agreement with the higher aggressiveness of RIVA-Sc cells, CXCR4 expression was increased compared to the non-conditioned counterpart (Figure 3E). Membrane expression of the receptor increased in infiltrated LN and was maintained in cells
This article is protected by copyright. All rights reserved
infiltrating the CNS. Interestingly, the membrane expression of CXCR4 increased in lymphoma cells surrounding the bone but decreased in cells in the BM, in which a dot-like staining was observed (Figure 3E and Supplementary Table S1). The results of subcutaneous conditioning of Toledo cells were similar to results obtained in RIVA cells. When Toledo cells were intravenously injected, 100% of mice presented BM and CNS infiltration, whereas 88.8% of the animals presented LN infiltration (Table 1). Although the infiltration pattern of Toledo-Sc cells was similar to that in Toledo cells, survival time of mice was shorter correlating with enhanced CXCR4 expression and cell aggressiveness (Supplementary Table S1 and Supplementary Figure S4).
In contrast, subcutaneous conditioning of OCI-Ly10 cells decreased CXCR4 expression (Supplementary Table S1 and Supplementary Figure S5A) associated with lower dissemination of cells and longer survival time of mice (Table 1 and Supplementary Figure S5B). Thus, expression levels of CXCR4 also maintained a positive correlation with dissemination in this model.
SUDHL-2 cells showed low engraftment and dissemination when they were injected in mice (Table 1) and no statistical differences were observed when compared to SUDHL-2-Sc mice. CXCR4 expression was almost undetectable in SUDHL-2 and SUDHL-2-Sc cells, and in the infiltrated tissues (Supplementary Figure S5C and Supplementary Table S1). Thus, in this cell line, subcutaneous conditioning did not enhance take rate or dissemination, and did not increase CXCR4 expression. The absence of changes in CXCR4 expression correlated with the absence of changes in cell dissemination.
Blockage of CXCR4 decreases dissemination of DLBCL cells in xenograft mice We performed in vivo experiments to evaluate whether the antagonist AMD3100 could decrease the dissemination of lymphoma cells by blocking CXCR4. Mice were intravenously injected with AMD3100-pretreated OCI-Ly10 or untreated OCI-Ly10 cells and administered with AMD3100 or vehicle, respectively. The BLI signal was observed in all mice from day 14 post-injection of cells. The time course of BLI signal in the AMD3100-treated group showed that DLBCL cells took longer time to disseminate than cells from the vehicle group (Figure 4A). Significant differences were observed in
This article is protected by copyright. All rights reserved
the BLI curves from day 14 after cell injection, and photon counts (PHCs) were higher in the vehicle group than in the AMD3100-treated group (Figure 4B). Furthermore, cell migration to LN was significantly decreased in the AMD3100-treated group. No infiltration of axillary or cervical LN was detected in AMD3100-treated mice, whereas 60% of mice presented infiltration of these LN in the vehicle group (Supplementary Table S2). In contrast, cells infiltrating BM and CNS were detected equally in both groups. To ensure that the effect of AMD3100 on cell infiltration was due to the inhibition of cell dissemination rather than to the induction of cell death, we evaluated whether the pretreatment of OCI-Ly10 cells with AMD3100 affected cell viability. We observed that after 30 minutes’ incubation, AMD3100 did not induce cell apoptosis (Figure 4C). More than 95% of the cells were viable in AMD3100-treated and non-treated cells. Moreover, AMD3100 induced the internalization of CXCR4 from the cell membrane to the cytosol (Figure 4D). Thus, we confirmed that AMD3100 reduced the dissemination of DLBCL cells to LN by down-regulating CXCR4 without inducing cell death.
CXCR4 expression independently predicts outcome in DLBCL patients We evaluated CXCR4 expression in 94 biopsies from patients with primary DLBCL. Representative CXCR4 staining of DLBCL tissue sections are shown in Supplementary Figure S1. Kaplan-Meier analysis showed a significant decrease in progression-free survival (PFS) and in overall survival (OS) in patients with CXCR4 expression (log-rank P = 0.001 and P = 0.021, respectively) (Figures 4E and 4F). Stratification of CXCR4 expression according to the clinical characteristics of the patients (Table 2) also showed a correlation between recurrent patients and CXCR4-expressing tumors (P = 0.005). By applying univariate COX analysis, the Eastern Cooperative Oncology Group performance status (ECOG,
Alberto Gallardo,6 Miguel Ángel Pavón,1,3 María Virtudes Céspedes,1,3 Albert Grañena,7 Miguel Alcoceba,8 Oscar Blanco,8 Marcos Gonzalez-Díaz,8 Jorge Sierra,4 Ramon Mangues1,3* and Isolda Casanova1,3*
1
Grup d’Oncogènesi i Antitumorals, lnstitut d’Investigacions Biomèdiques Sant Pau, Barcelona,
Spain. 2Department of Biochemistry and Molecular Biology, Universitat de Barcelona, Spain. 3
CIBER en Bioingeniería, Biomateriales y Nanomecidicina (CIBER-BBN), Barcelona, Spain.
4
Department of Hematology, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain.
5
Department of Pathology, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain.
6
Department of Pathology, Clínica Girona, Girona, Spain.
7
Department of Hematology, Hospital Universitari de Bellvitge, Barcelona, Spain.
¶ Deceased. 8
Department of Hematology and Pathology, IBSAL-University Hospital, Center for Cancer Research-
IBMCC (USAL-CSIC), Salamanca, Spain. *These authors contributed equally to this work
Corresponding author: Ramon Mangues, Grup d’Oncogènesi i Antitumorals (GOA), Biomedical Research Institute Sant Pau (IIB Sant Pau), Sant Antoni Maria Claret 167, Pavelló 11, 2n pis, Barcelona, 08025, Spain. E-mail: [email protected] Conflict of interest: The authors declare no competing financial interests.
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1002/path.4446
This article is protected by copyright. All rights reserved
ABSTRACT
The chemokine receptor CXCR4 has been implicated in the migration and trafficking of malignant B cells in several hematological malignancies. Overexpression of CXCR4 has been identified in hematological tumors, but data concerning the role of this receptor in diffuse large B-cell lymphoma (DLBCL) are lacking. CXCR4 is a marker of poor prognosis in various neoplasms, correlating with metastatic disease and decreased survival of patients. We studied CXCR4 involvement in cell migration in vitro and dissemination in vivo. We also evaluated the prognostic significance of CXCR4 in 94 biopsies of DLBCL patients. We observed that the level of expression of CXCR4 in DLBCL cell lines correlated positively with in vitro migration. Expression of the receptor was also associated with increased engraftment and dissemination, and decreased survival time in NOD/SCID mice. Furthermore, administration of a specific CXCR4 antagonist, AMD3100, decreased dissemination of DLBCL cells in a xenograft mouse model. In addition, we found that CXCR4 expression is an independent prognostic factor for shorter overall survival and progression-free survival in DLBCL patients. These results show that CXCR4 mediates dissemination of DLBCL cells and define for the first time its value as an independent prognostic marker in DLBCL patients.
Keywords: Diffuse large B-cell lymphoma; CXCR4; cell migration; cell dissemination; prognostic marker.
INTRODUCTION Diffuse large B-cell lymphoma (DLBCL) is a clinically and genetically heterogeneous disease. It is the most common subtype of non-Hodgkin lymphoma, accounting for 30% of all cases [1]. The addition of rituximab to standard treatment (R-CHOP: rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) has increased disease-free survival and overall survival of DLBCL patients [2-4]. Nevertheless, overall survival remains poor, ranging from 40 to 50% over 5 years [5]. The prognosis of newly diagnosed DLBCL patients depends on clinical and biological features defined in the International Prognostic Index (IPI) developed almost 20 years ago [6]. As the prognosis of patients within the IPI groups is highly diverse, novel biomarkers with a potential impact on outcome
This article is protected by copyright. All rights reserved
are needed to refine the predictive scores [7]. Biological markers involved in cell signaling and trafficking seem to merit particular attention.
CXCR4 and its ligand SDF-1 (stromal cell-derived factor-1, also designated as CXCL12) are critically involved in the trafficking of normal lymphocytes [8]. CXCR4 functions as a classical chemokine receptor in adults and, together with its ligand, is essential for organogenesis and hematopoiesis. Its expression is detected in several tissues, predominantly in hematopoietic cells [9]. The CXCR4-SDF1 axis is essential for normal B-cell development and is involved in aspects of B-cell homeostasis such as retention of B-cell precursors in bone marrow and B-cell homing to lymph nodes [10, 11].
The receptor and its ligand have been described in the migration and trafficking of malignant B cells in hematological malignancies such as Burkitt lymphoma [12], follicular lymphoma [13] and chronic lymphocytic leukemia [14, 15]. Moreover, CXCR4 has been associated with decreased survival and considered a prognostic marker in several neoplasms, correlating with disease aggressiveness [16]. However, no data are available about its involvement in DLBCL dissemination or its relevance as a prognostic marker in this disease. We aimed to study how the alteration of CXCR4 expression affects cell migration in vitro and dissemination in vivo. In addition, we evaluated the prognostic significance of the receptor in DLBCL patients.
MATERIALS AND METHODS
Cell culture SUDHL-2, RIVA and OCI-Ly10 (activated subtype, ABC), and Toledo (germinal center subtype, GC) are human DLBCL cell lines. Culture media for all cells were supplemented with 10% fetal bovine serum, 1% glutamine, 100 U/ml penicillin/streptomycin (Life Technologies), unless otherwise specified. SUDHL-2, RIVA and Oci-Ly10 cells were cultured with IMDM (Life Technologies) and Toledo cells with RPMI 1640. OCI-Ly10 were supplemented with 10% human serum (Sigma-Aldrich) and 50 µM 2-mercaptoethanol. Toledo cells were obtained from the ATCC, and SUDHL-2, RIVA and
This article is protected by copyright. All rights reserved
OCI-Ly10 cell lines were kindly provided by Dr. L. Pasqualucci (Columbia University, NY). SUDHL-2Sc, OCI-Ly10-Sc, RIVA-Sc and Toledo-Sc cells were obtained from disaggregation of subcutaneous tumors. All cell lines were cultured in a humidified atmosphere at 37ºC in 5% CO2.
RT-PCR mRNA expression from cell lines was measured on a 7900 HT Fast-Real Time PCR System (Applied Biosystems), using pre-designed Taqman® Gene Expression primer and probe assays for CXCR4 (Hs00607978_s1) and HPRT1 (Hs99999909_m1) (Applied Biosystems) as previously described [17]. Experiments were done in triplicate and repeated at least twice. Ct values were normalized against the housekeeping gene HPRT1.
FACS analysis Fluorescence-activated cell sorting (FACS) analysis was performed to detect cell surface expression of CXCR4. One million cells were washed in phosphate-buffered saline containing 0.5% bovine serum albumin (PBS-BSA) and incubated for 30 minutes at 4ºC with PE-Cy5 mouse anti-human CXCR4 monoclonal antibody or PE-Cy5 mouse IgG2a (BD Biosciences) as an isotype control. Unbound antibody was removed by two washes with PBS-BSA. Data acquisition was performed using flow cytometry (FACS Calibur, BD Biosciences) and analyzed by Cell Quest Pro software. Results are expressed as mean fluorescence intensity ± SD.
Lentivirus, infection and cell sorting Lentiviruses containing the pSIN-DUAL-Luc-GFP2 vector were used to infect DLBCL cells, as described previously [18]. GFP positive cells were sorted using the FACSAria cell sorter (BD Biosciences). The Luciferase Assay System kit (Promega) was used following manufacturer’s instructions to confirm expression of the luciferase gene. Sorted cells, expressing GFP and luciferase genes were used for all in vitro and in vivo experiments.
This article is protected by copyright. All rights reserved
Migration Assay Migration assays were performed using 8-μm pore size Transwell (Corning Costar). In brief, 5x10
5
cells were suspended in 100 μl of serum-free media and incubated at 37ºC for 2 hours in presence or absence of the CXCR4 inhibitor AMD3100 (10 µg/mL, Sigma-Aldrich). The lower compartment was filled with 600 µL of serum-free media with or without 100 ng/ml of human SDF-1α (Prospec-Tany TechnoGene). The amount of cells migrating within 24 hours to the lower compartment was quantified using a Scepter 2.0 Handheld Automated Cell Counter (EMD Millipore). Results are expressed as the total number of cells that migrated to the lower chamber ± SD.
CXCR4 internalization and apoptosis analysis Internalization of CXCR4 was evaluated by incubating cells with 100 ng/ml SDF-1α (Prospec-Tany TechnoGene) or 10 μg/ml AMD3100. Cells were then washed twice with PBS and cell surface expression of CXCR4 was detected by flow cytometry. To evaluate whether cells exposed to AMD3100 were undergoing apoptosis, we used the Annexin V-FITC Apoptosis Detection Kit (Calbiochem/EMD Millipore), following the manufacturer’s instructions.
In vivo assays and bioluminescent imaging NOD/SCID female mice (4 weeks old) were obtained from Charles River Laboratories. Mice were housed in microisolator units with sterile food and water ad libitum. Animals were monitored every other day and a loss of 15% in body weight or signs of sickness were considered as the endpoint. Procedures involving mice were approved by the Hospital de la Santa Creu i Sant Pau Animal Ethics Committee according to established guidelines. To perform the subcutaneous conditioning of cells, 10 NOD/SCID mice were inoculated in both flanks with 10x106 of SUDHL-2, RIVA, OCI-Ly10 or Toledo luminescent
cells. After subcutaneous
cell
growth, tumors
were extracted, mechanically
disaggregated and incubated at 37ºC for 15 hours. Cells were then filtered to obtain SUDHL-2-Sc, RIVA-Sc, OCI-Ly10-Sc or Toledo-Sc, and intravenously injected in new groups of mice (20x106 cells per mouse). Results were compared to direct intravenous injection of non-conditioned cells. Engraftment, dissemination, and mouse survival time were evaluated in all groups (6-10 NOD/SCID
This article is protected by copyright. All rights reserved
mice per group). To perform CXCR4 blocking experiments using AMD3100, mice were intravenously injected with 20x106 AMD3100-pretreated (10 µg/ml, 30 minutes at 37ºC) or untreated OCI-Ly10 cells (n=10). Mice injected with AMD3100-pretreated cells were administered daily with intraperitoneal AMD3100 (10 mg/kg), whereas control mice were treated with vehicle (saline solution). We monitored subcutaneous tumor growth and dissemination of intravenously-injected DLBCL cells capturing bioluminescence (BLI) once a week using the IVIS-2000 (Xenogen) imaging system. Mice were anesthetized with 3% isoflurane in oxygen and BLI was captured 10 minutes after intraperitoneal injection of firefly D-luciferin (2.25mg/mouse, Perkin Elmer). BLI pictures were quantified using Living Imaging Software (Version4.2, Xenogen), as described previously [19].
Immunohistochemical staining Immunohistochemical analysis was performed using paraffin-embedded tissue samples, cell lines and sections of infiltrated organs to detect CD10, MUM1, BCL6 (DAKO) or CXCR4 (Abcam) expression. Staining was performed in a DAKO Autostainer Link48 following the manufacturer’s instructions. We quantified CXCR4 expression in cell lines and infiltrated organs by assigning an H-score (0-300), resulting from the product of the intensity (0-3) and the percentage of staining (0-100). We also evaluated if membrane CXCR4 was present or absent in tumor samples of DLBCL patients and used dichotomized values for statistical analysis. DLBCL biopsies with more than 5% of tumor cells showing higher CXCR4 expression than normal B cells from lymphoid tissues were considered positive (Supplementary Figure S1). Two independent observers evaluated all samples using an Olympus BX51 microscope. There was inter-observer agreement in 95% of the samples; the remaining slides were re-evaluated and consensus decisions were made. Images were acquired using an Olympus DP72 digital camera and processed with the Olympus CellD Imaging 3.3 software (Olympus Corporation).
Patients and tissue samples Ninety-four biopsies from patients with primary DLBCL were diagnosed in the Hospital de la Santa Creu i Sant Pau (HSCSP) or Hospital Universitario de Salamanca (HUS) between 2001 and 2012, based on the WHO criteria [1]. All cases were evaluated for CXCR4 expression by
This article is protected by copyright. All rights reserved
immunohistological analysis. Patients with follicular lymphoma or another type of indolent lymphoma transformed into DLBCL were excluded. All patients received rituximab in their treatment; 92.5% were treated with R-CHOP chemotherapy and 7.5% received other chemotherapeutic agents in combination with rituximab. The median clinical follow-up time was 49 months. Table 2 shows the main features of the series. The Institutional Review Boards at HSCSP and HUS approved the study and informed consent was provided according to the declaration of Helsinki and obtained from patients.
Statistical analysis Results are reported as mean ± SD for experiments performed in triplicates and were analyzed using the non-parametric Mann–Whitney U test. Survival rates were estimated by the Kaplan-Meier method and differences between groups were compared using the log-rank test. Overall survival (OS) was measured in mice from the day of cell injection to the end-point. Progression-free survival (PFS) of DLBCL patients was calculated from the onset of treatment until relapse or death. OS was calculated as the time between the onset of treatment and death or date of the last follow-up. Univariate and multivariate analyses were done using the COX proportional hazard model. Correlations between clinico-pathological variables and CXCR4 expression were tested using Fisher’s exact test. Results were considered significantly different at P ≤ 0.05. Statistical calculations were performed using SPSS software (version 21).
This article is protected by copyright. All rights reserved
RESULTS
CXCR4 expression in DLBCL cell lines is associated with SDF-1α-driven migration We evaluated CXCR4 expression in DLBCL cell lines and found that CXCR4 mRNA expression was the highest in OCI-Ly10 cells, followed by Toledo and RIVA cells (Figure 1A). CXCR4 expression in SUDHL-2 cells was undetectable. Expression of the CXCR4 protein was also evaluated by immunohistochemistry and FACS analysis (Supplementary Figure S2) and correlated with the results obtained by RT-PCR. Next, we evaluated the migration induced by SDF-1α in DLBCL cell lines. OCILy10, Toledo and RIVA cells increased their migration capacity in the presence of SDF-1α (Figure 1B). In contrast, SUDHL-2 cell line, which does not express CXCR4, migrated equally in presence or absence of SDF-1α (data not shown). To assess if SDF-1α-driven migration was dependent on CXCR4 expression, we evaluated whether exposure to AMD3100, a CXCR4 antagonist, inhibited this process. We found that AMD3100 significantly inhibited the migration stimulated by SDF-1α in OCILy10, Toledo and RIVA cell lines (Figure 1B).
SDF-1α induces CXCR4 internalization in DLBCL cell lines We investigated the regulation of CXCR4 by the SDF-1α chemokine. We used the DLBCL cell lines OCI-Ly10, Toledo and RIVA, which expressed CXCR4 receptor in their membrane (Figure 1C, Control). When cells were exposed to SDF-1α, the receptor was not detected in membrane (Figure 1C, SDF-1α). After exposure to SDF-1α, CXCR4 was internalized from the membrane showing a dotlike staining (punctuated) in the cell cytosol (Figure 1D). Dot-like immunostaining of CXCR4 was also detected in RIVA and OCI-Ly10 cells after exposure to SDF-1α (data not shown).
CXCR4 expression in DLBCL cell lines correlates with engraftment, dissemination and survival time in xenograft mice We studied the aggressiveness of DLBCL cell lines displaying different levels of CXCR4 expression in xenograft mice. The intravenously injected OCI-Ly10 cells, with a high expression of CXCR4, showed 100% of engraftment, rapid dissemination to lymph nodes (LN), bone marrow (BM) and central nervous system (CNS), and mouse survival time of less than 30 days (Table 1). In contrast, SUDHL-2 cells, which do not express CXCR4, showed 50% of engraftment, and longer mouse survival time
This article is protected by copyright. All rights reserved
(67.3 ± 21.5 days). Moreover, no infiltration of LN was detected and only 33% of the animals presented BM infiltration. Bioluminescent images (BLI) of OCI-Ly10 or SUDHL-2 injected mice were acquired from the first week after cell injection (Figure 2A). In OCI-Ly10 injected mice, BLI was first observed 2 weeks after injection. In SUDHL-2 injected mice, however, the BLI signal was not detected until week 6 post-injection. The infiltration of lymphoma cells observed in BLI images was confirmed by histological analysis (Figure 2B-C, H&E). Immunohistochemical analysis showed that CXCR4 expression of OCI-Ly10 cells was maintained in infiltrated LN, BM and CNS (Figure 2B, CXCR4). Moreover, in LN and BM, CXCR4 showed dot-like cytoplasmic immunostaining indicating that the receptor was internalized from the membrane. In agreement with the lack of CXCR4 expression in SUDHL-2 cells, the receptor was almost undetectable in the infiltrated BM (Figure 2C, CXCR4). Both OCI-Ly10 and SUDHL-2 cells maintained their DLBCL immunophenotype in the infiltrated tissues (Supplementary Figure S3).
Subcutaneous-conditioning increased CXCR4 expression and aggressiveness in RIVA and Toledo cells In previous studies we developed disseminated lymphoma mouse models by performing subcutaneous-conditioning of cells to increase their aggressiveness prior to their intravenous injection [19, 20]. Here, we used this approach to evaluate the effect of subcutaneous-conditioning of DLBCL cells on CXCR4 expression and its impact on their take rate and dissemination in vivo. We observed that CXCR4 expression was increased in subcutaneously-conditioned RIVA and Toledo cells (Supplementary Table S1). Moreover, mice injected with conditioned cells showed a shorter survival time than mice injected with non-conditioned cells (Table 1, Figure 3B and Supplementary Figure S4B). BLI images were acquired from the first week after cell injection and showed that in RIVAinjected mice, the BLI signal first appeared at week 8, whereas in RIVA-Sc-injected mice it appeared at week 5 (Figure 3A). Macroscopic and microscopic infiltrations of LN were detected in RIVA-Sc injected mice (Figure 3C and 3D). BM and CNS were also infiltrated in some mice (Figure 3D and Table 1). In contrast, in the RIVA-injected group the CNS was infiltrated but the BM of vertebral column and femur were not (Table 1). In agreement with the higher aggressiveness of RIVA-Sc cells, CXCR4 expression was increased compared to the non-conditioned counterpart (Figure 3E). Membrane expression of the receptor increased in infiltrated LN and was maintained in cells
This article is protected by copyright. All rights reserved
infiltrating the CNS. Interestingly, the membrane expression of CXCR4 increased in lymphoma cells surrounding the bone but decreased in cells in the BM, in which a dot-like staining was observed (Figure 3E and Supplementary Table S1). The results of subcutaneous conditioning of Toledo cells were similar to results obtained in RIVA cells. When Toledo cells were intravenously injected, 100% of mice presented BM and CNS infiltration, whereas 88.8% of the animals presented LN infiltration (Table 1). Although the infiltration pattern of Toledo-Sc cells was similar to that in Toledo cells, survival time of mice was shorter correlating with enhanced CXCR4 expression and cell aggressiveness (Supplementary Table S1 and Supplementary Figure S4).
In contrast, subcutaneous conditioning of OCI-Ly10 cells decreased CXCR4 expression (Supplementary Table S1 and Supplementary Figure S5A) associated with lower dissemination of cells and longer survival time of mice (Table 1 and Supplementary Figure S5B). Thus, expression levels of CXCR4 also maintained a positive correlation with dissemination in this model.
SUDHL-2 cells showed low engraftment and dissemination when they were injected in mice (Table 1) and no statistical differences were observed when compared to SUDHL-2-Sc mice. CXCR4 expression was almost undetectable in SUDHL-2 and SUDHL-2-Sc cells, and in the infiltrated tissues (Supplementary Figure S5C and Supplementary Table S1). Thus, in this cell line, subcutaneous conditioning did not enhance take rate or dissemination, and did not increase CXCR4 expression. The absence of changes in CXCR4 expression correlated with the absence of changes in cell dissemination.
Blockage of CXCR4 decreases dissemination of DLBCL cells in xenograft mice We performed in vivo experiments to evaluate whether the antagonist AMD3100 could decrease the dissemination of lymphoma cells by blocking CXCR4. Mice were intravenously injected with AMD3100-pretreated OCI-Ly10 or untreated OCI-Ly10 cells and administered with AMD3100 or vehicle, respectively. The BLI signal was observed in all mice from day 14 post-injection of cells. The time course of BLI signal in the AMD3100-treated group showed that DLBCL cells took longer time to disseminate than cells from the vehicle group (Figure 4A). Significant differences were observed in
This article is protected by copyright. All rights reserved
the BLI curves from day 14 after cell injection, and photon counts (PHCs) were higher in the vehicle group than in the AMD3100-treated group (Figure 4B). Furthermore, cell migration to LN was significantly decreased in the AMD3100-treated group. No infiltration of axillary or cervical LN was detected in AMD3100-treated mice, whereas 60% of mice presented infiltration of these LN in the vehicle group (Supplementary Table S2). In contrast, cells infiltrating BM and CNS were detected equally in both groups. To ensure that the effect of AMD3100 on cell infiltration was due to the inhibition of cell dissemination rather than to the induction of cell death, we evaluated whether the pretreatment of OCI-Ly10 cells with AMD3100 affected cell viability. We observed that after 30 minutes’ incubation, AMD3100 did not induce cell apoptosis (Figure 4C). More than 95% of the cells were viable in AMD3100-treated and non-treated cells. Moreover, AMD3100 induced the internalization of CXCR4 from the cell membrane to the cytosol (Figure 4D). Thus, we confirmed that AMD3100 reduced the dissemination of DLBCL cells to LN by down-regulating CXCR4 without inducing cell death.
CXCR4 expression independently predicts outcome in DLBCL patients We evaluated CXCR4 expression in 94 biopsies from patients with primary DLBCL. Representative CXCR4 staining of DLBCL tissue sections are shown in Supplementary Figure S1. Kaplan-Meier analysis showed a significant decrease in progression-free survival (PFS) and in overall survival (OS) in patients with CXCR4 expression (log-rank P = 0.001 and P = 0.021, respectively) (Figures 4E and 4F). Stratification of CXCR4 expression according to the clinical characteristics of the patients (Table 2) also showed a correlation between recurrent patients and CXCR4-expressing tumors (P = 0.005). By applying univariate COX analysis, the Eastern Cooperative Oncology Group performance status (ECOG,
