Introduction

Ewing sarcoma (EWS) is the third most common primary bone sarcoma which predominantly occurs in children and adolescents [1]. It is characterized by aggressive/destructive local growth and has a high-grade malignant behavior, with (micro-) metastases at the time of presentation being common. Patients with metastases or recurrent disease have a poor outcome with 15–30 % long-term survival [2, 3].

To date, after the initial introduction of multimodal chemotherapy, no further improvement in survival of these patients has been accomplished, and besides the classical parameters such as tumor site, resectability, response to chemotherapy and size, no prognostic markers are in clinical use for decision making. EWS has a very low number of mutations compared to other tumors, which suggests that corrective apoptosis pathways are still functional, such as TNF-related apoptosis-inducing ligand (TRAIL) pathway [4–6]. The death receptor pathways and other apoptotic pathways are active in EWS and consequently the tumor is sensitive for activation of these mechanisms by natural killer (NK) cells and cytotoxic T cells [7–9]. Immunotherapy in Ewing sarcoma has been shown to have a promising potential role in vitro and is being tested in two clinical trials by administrating donor NK-cells (NCT01287104, NCT02100891) [7, 8, 10].

We previously investigated the immune microenvironment in EWS and demonstrated a relation between the number of infiltrating cytotoxic T cells and patient outcome [11]. Expression levels of pro-inflammatory chemokines [particularly chemokine (C-X-C motif) ligand 9 (CXCL9), chemokine (C-X-C motif) ligand 10 (CXCL10) and chemokine (C-C motif) ligand 5] correlated positively with the number of infiltrating CD8+ T cells [11]. Another potent T cell chemoattractant is chemokine (C-C motif) ligand 21 (CCL21), which acts via its receptor chemokine (C-C motif) receptor 7 (CCR7) as a single attractant or in combination with CXCL9 and CXCL10 [12, 13]. In addition, CCL21 may increase dendritic cell-provoked T cell responses, leading to more efficient anti-tumor immune responses [14, 15]. Successful use of CCL21 as immunotherapy has been demonstrated and a trial using dendritic cells expressing CCL21 showed better results than CCL21 used alone in nonsmall lung cancer [16]. Due to the immunogenic role of CCL21 and its immunotherapeutic potential, we studied the CCL21 expression in primary therapy-naïve Ewing sarcoma samples and EWS cell lines by analyzing the RNA expression levels of CCL21. The measured RNA expression levels were correlated with the number of infiltrating T cells and the CD4+/CD8+ T cell ratio in Ewing sarcoma samples. A reversed CD4+/CD8+ T cell ratio has been reported as predictor of improved outcome in other tumors [17, 18]. In our study, the CD4+/CD8+ T cell ratio showed inverse correlation with the CCL21 expression level, and increased CCL21 expression levels were associated with better survival. This correlation suggests that testing for CCL21 levels in therapy-naïve EWS tumor samples could be used as a prognostic marker and supports a potential role for this cytokine in anti-tumor immunity.

Materials and methods

Clinical information on patient samples

Eighteen cryopreserved primary therapy-naïve samples from 18 EWS patients, all containing more than 80 % tumor cells as assessed by light microscopy, and a validation tissue microarray (TMA) of formalin-fixed paraffin-embedded (FFPE) specimens of 16 tumors of 16 patients were obtained from the Department of Pathology, Leiden University Medical Center, and were handled in a coded fashion, according to the Dutch National Ethical Guidelines (‘Code for Proper Secondary Use of Human Tissue’). Ewing sarcoma diagnosis was established according to WHO criteria, including immunohistochemistry (IHC) and Ewing sarcoma breakpoint region 1 (EWSR1) translocation detection either by real-time quantitative reverse transcriptase PCR (RT-Q-PCR) or by interphase fluorescence in situ hybridization (FISH). A good chemotherapeutic response was defined by <10 % morphologically viable tumor cells upon histopathologic evaluation of the post-chemotherapy resection specimen [19, 20]. Median patient age at diagnosis of the cohort was 17.5 years (range of 5–35 years) (Supplementary Table S1).

Ewing sarcoma cell lines

Ewing sarcoma cell lines (n = 21) were obtained from multiple sources: L-1062 and L-872 were established in-house [21]; CHP100, RM-82, IARC-EW7, TC32 and 6647, CHP100, RM-82, IARC-EW-7, WE-68, IARC-EW-3, STA-ET-2.1, TTC-466, STA-ET-10, CADO-ES1, TC-71, VH-64, COH and STA-ET-1 were obtained from the EuroBoNeT consortium collection (Institute of Pathology, University Medical Center, Düsseldorf, Germany) [22] and SK-ES-1, SK-NM-C, A-673 and R-D-ES from the American Type Culture Collection (ATCC). All cell lines and primary culture L-4027 were cultured in a monolayer under equal conditions and in Iscove’s modified Dulbecco’s medium containing GlutaMAX supplement, supplemented with 1 % streptomycin/penicillin and 10 % heat-inactivated FCS (all from Life Technologies, Bleiswijk, The Netherlands). Authentication of cell lines using Powerplex 1.2 and CellID STR (Promega, Leiden, the Netherlands) and mycoplasma DNA Q-PCR screening were regularly performed on all cell lines.

RNA isolation

Total RNA was isolated using TRIzol reagent (Life Technologies, Bleiswijk, the Netherlands) according to the manufacturer’s instructions. RNA concentration was measured using Nanodrop, and quality of the RNA was determined using Bioanalyzer2000 RNA Nano chip (Agilent Technology, Amstelveen, The Netherlands). Samples with a RNA integrity number ≥5 were included for RT-Q-PCR analysis.

RT-Q-PCR analysis and Fluidigm

cDNA generation and RT-Q-PCR using Fluidigm BioMark system was performed according to the H format protocol of the manufacturer (QIAGEN, Venlo, the Netherlands). Samples were prepared for RT-Q-PCR using a 96 × 96 dynamic array chip and performed using BioMark HD system (Fluidigm, San Francisco, CA, USA). All primers for this array chip were obtained from QIAGEN (Venlo, The Netherlands) including nine control genes: RPL13A, BTF3, YWHAZ, UBE2D2, ATP6V1G1, IPO8, HBS1L, AHSP and TBP. Samples were measured in duplicates and analyzed using BioMark software, delivered with the HD system.

Detection of infiltrating T-lymphocytes

Number of CD4- and CD8-positive T cells were determined according to Berghuis et al. [11]. In brief, FFPE tumor sections were stained for CD3 (Dako, Heverlee, Belgium), CD4 and CD8 (Novocastra, Newcastle upon Tyne, UK) and scanned with Zeiss LSM-510 confocal microscope (Carl Zeiss AG, Göttingen, Germany). In each section 10 areas were selected, digitally photographed and lymphocytes were counted.

Immunohistochemistry

Tumor sections were stained with anti-CCL21 (clone: HPA051210) (Sigma-Aldrich, Steinheim, Germany) and CCR7 (Abcam, Cambridge, UK) antibodies. Extensive validation data for anti-CCL21 antibody (HPA051210) using IHC on various TMAs and western blots are accessible at the Human Protein Atlas portal [23]. Sections were dewaxed, rehydrated and were subjected to citrate pH6.0 (CCL21) or Tris/HCl-EDTA pH9 (CCR7) antigen retrieval. Sections stained for CCL21 expression were incubated with 5 % nonfat dry milk for 30 min at room temperature and incubated with anti-CCL21 (1:600) in 5 % ELK overnight at 4  °C. Sections stained for CCR7 expression were incubated 1.5 % BSA with anti-CCR7 (1:2000) overnight at 4  °C. Afterward sections were incubated with Immunologic Poly-HRP-GAM/R/R IgG (Leica Biosystems, Eindhoven, The Netherlands) and Dako liquid DAB+ Substrate-Chromogen System (Dako, Heverlee, Belgium). Scanning of the slides was performed by Philips Ultra Fast Scanner (Philips Healthcare, Eindhoven, Netherlands). Tonsil tissues, both regular and decalcified FFPE processed, were used as a control. All slides were evaluated by at least two experienced persons of whom one was a reference pathologist (PCWH).

Statistical analysis

Survival curves were calculated using the Kaplan–Meier method, and P values were calculated using the log-rank test using SPSS 20 (IBM Inc. Amsterdam, The Netherlands) and Prism GraphPad 6 (GraphPad Software Inc. La Jolla, CA, USA). Multivariate analysis of the parameters could not be performed due to the limited number of samples. Correlations were calculated with SPSS 20 using Pearson or Spearman correlation. High RNA expression was set as expression above the median. Student t test’s P value was calculated using Prism GraphPad 6 assuming nonparametric distribution due to limited number of samples and was corrected using Manley-Welch correction.

Results

RNA expression of CCL21 was analyzed in 18 primary therapy-naïve tumor samples, and the expression levels were correlated with the immunohistochemical staining of the CD4+- and CD8+-infiltrating T cells in eight tissue samples for which sufficient FFPE material was still available (Supplementary Table S2). In these samples, the CCL21 expression was inversely correlated to CD4+/CD8+ T cell ratio (Fig. 1). However, the absolute numbers of CD8+ or CD4+ T cells did not correlate with CCL21 expression and varied widely between the samples (data not shown). Since a high-CD8+ T cells infiltration was associated in Ewing sarcoma with a better outcome, we correlated CCL21 RNA expression levels in therapy-naïve tumor samples with development of metastases, survival and chemotherapeutic response. Kaplan–Meier survival analysis demonstrated that an increased CCL21 expression correlated significantly both with improved-event-free survival (EFS) and with overall survival (OS) (P = 0.0001; P = 0.0004) (Fig. 2a, b). Moreover, natural logarithm-transformed CCL21 expression was significantly higher in patient who did not develop a metastasis compared to patients who did (P < 0.0005) (Fig. 2c). However, no correlation with metastasis at diagnosis was observed (data not shown). The improved survival may be linked to a better chemotherapeutic response as correlation between good response and increased CCL21 expression was observed (P = 0.02). It should also be noted that good response to chemotherapy was correlated with improved outcome (P = 0.008).

Fig. 1
figure 1

Increased CCL21 RNA expression correlates with reversed CD4+/CD8+ ratio of infiltrating CD3+ T cells. CCL21 RNA expression levels of samples with available high-quality RNA and high-quality FFPE material (n = 8) were natural log-transformed and correlated with the ratio between the total counted CD3+CD4+ and CD3+CD8+-infiltrating T cells. P value of the linear regression analysis was demonstrated

Full size image
Fig. 2
figure 2

High CCL21 expression correlated to better EFS and OS. a, b CCL21 RNA expression levels of the primary therapy-naïve tumors samples were correlated to EFS and OS using Kaplan–Meier survival analysis. Median was set as threshold to determine high (dotted line) and low (straight line) CCL21 expression. c Natural log-transformed CCL21 expression levels were compared between patients who developed a metastasis (+) and patients who did not develop a metastasis (−)

Full size image

In addition, we investigated the CCL21 RNA expression in 21 cell lines and 1 primary culture. The CCL21 expression levels in the cell lines were significantly lower than the in therapy-naive tumor samples (Fig. 3), with a large variation of expression levels between tumor samples compared to cell lines.

Fig. 3
figure 3

RNA expression levels of CCL21 were significantly higher in tumor samples compared to cell lines. CCL21 expression levels of 21 cell lines and 1 primary culture were compared to expression levels of the primary therapy-naïve tumor samples

Full size image

To show that the difference in expression between tumor samples and cell lines can be accounted for by infiltrating immune cells in the tumor tissues, we studied CCL21 expression at the protein level. The eight cases for which sufficient FFPE material was available were stained for CCL21 using IHC. In addition, the tumor samples were stained for CCR7, the receptor of CCL21. In the tumor samples, EWS cells were negative for CCL21 and CCR7, while infiltrating immune cells did show expression of both CCL21 and CCR7 (Fig. 4). An additional TMA of 16 EWS cases was used for validation of the CCR7 and CCL21 expression pattern. In this TMA, similar to the other cases, EWS cells were CCR7 and CCL21 negative for all but one of the cases.

Fig. 4
figure 4

Neither CCR7 nor CCL21 expression was observed in Ewing sarcoma cells by immunohistochemical detection. Eight tumor samples included in the RNA expression analysis and a TMA with 16 samples in duplicate were stained for CCR7 and CCL21 (20× magnification). a Tumor cells showed no expression, while infiltrating immune cells showed expression of CCR7 (left inset, 40× magnification), positive control is in the right inset from tonsil. b Tumor cells showed no expression, while infiltrating immune cells showed expression of CCL21 (left inset, 40× magnification), positive control is in the right inset from tonsil. Magnification: 20×

Full size image

Discussion

Previously, we demonstrated that pro-inflammatory chemokines CXCL9 and CXCL10 were associated with an increase in tumor infiltrating CD8+ T cells [11]. CCL21 is, like CXCL9 and CXCL10, a CD8+ T cell chemoattractant, and its potency is enhanced by the interaction with CXCL9, CXCL10 and interferon gamma (IFNƴ) [13]. These findings prompted us to further investigate the role of CCL21 in EWS. We observed that an increased CCL21 RNA expression was correlated with a decreased CD4+/CD8+ ratio. It is likely that these CD3+CD8+-positive lymphocytes are T cells, but the presence of CD3+CD8+ NK T cells cannot be excluded [24]. In addition, increased CCL21 expression correlated with both better EFS and OS and inversely correlated with the development of metastasis. These observations may point toward a role of CCL21 in the anti-tumor immune response related to the proportion and type of immune cells present in or around the tumor in EWS patients; this has been observed in other tumors including breast cancer and melanoma [25–27]. Even though the presence of infiltrating immune cells in pediatric sarcomas, particularly in EWS, was found to be limited [28], the effect of these cells with regard to therapy response is significant [11].

A second factor which might have had an influence on the observed correlation with patient survival is the chemotherapeutic response of the tumor. In this study (P = 0.008) and other studies, a correlation with patient survival was reported [20, 29, 30]. Patients with a good chemotherapeutic response had a higher CCL21 expression in the tumor sample compared to patients with a poor response. Although it is generally believed that chemotherapy has an immunosuppressive effect by decreasing the number of leukocytes, by now it has become clear that certain chemotherapeutic agents can augment the tumor immunogenicity and stimulate dendritic cell maturation [31–33]. In mice, combining CCL21 immunotherapy with the chemotherapeutic agent paclitaxel had a synergistic effect [34]. CCL21 attracts dendritic cells and is suggested to improve the T cell activation of mature dendritic cells [14, 35]. Increased expression of CCL21 might be associated with an increase in the number of dendritic cells or an improved immunologic response upon tumor cell death. In patients with CCL21-expressing cells present in or around the tumor, chemotherapy could enhance the anti-tumor immunity and subsequently lead to a better chemotherapeutic response. Our study is based on a small patient cohort, and therefore, a larger study using therapy-naïve samples would be needed to validate the observed correlations.

The significantly higher CCL21 expression in primary therapy-naïve tumors compared to EWS cell lines suggests the involvement of a stromal factor in CCL21 expression. CCL21 expression can be enhanced by the interaction with CXCL9, CXCL10 and IFNƴ [13]. However, we reported previously the absence of CCL21 expression in cell lines even after IFNƴ stimulation indicating that this might be regulated by the EWSR1–friend leukemia virus integration 1 (FLI1) transcription factor [3, 11]. In this study, some cell lines expressed, at low levels, CCL21. The difference in CCL21 expression between tumor samples and cell lines might be not only caused by a stromal factor but could also be due to selective in vitro culture conditions. Therefore, the protein expression levels and localization of CCL21 were determined in EWS patient samples using IHC and demonstrated that CCL21 expression was restricted to tumor infiltrating immune cells and that it was not present in EWS cells. In addition, only in one sample CCR7 expression was detected in EWS cells. EWSR1–ETS fusion protein is known to downregulate, directly or indirectly, many chemokines and chemokine receptors, for example by altering regulatory miRNA expression levels and pattern [3]. Of these, the let-7 miRNA family is known to regulate expression of the CCL21–CCR7 [36]. The tumor suppressor let-7a is, for example, known to be directly downregulated by EWSR1–ETS, but this decrease in let-7a does not lead to increased CCR7 expression like in breast cancer cells [37, 38]. Several studies have investigated the role of the let-7 family in EWS and demonstrated a high expression of some members, mainly let-7g, in tumor samples. These studies also revealed various pathways in which these let-7 family members play a role; however, none of them could establish a direct connection between CCL21 or CCR7 and let-7 [37–41].

It is important to note that high CCL21 expression in tumor cells, for example in bladder cancer and breast cancer, is associated with an increased proliferation, number of metastases and a suppressive immune reaction. This might be as a result of paracrine or autocrine activation of a pro-tumorigenic CCL21/CCR7 axis [25, 27, 42]. As in EWS cells, no CCL21 expression was detected and CCR7 expression except in only one sample, and therefore, an active CCL21/CCR7 axis in EWS cells is unlikely. In studies which correlate CCL21 and CCR7 expression, not only the expression level but also the source, tumor cells versus infiltrating immune cells, should also be considered and recorded.

The potency of immunotherapy to treat EWS has been demonstrated by a number of studies [7, 8, 10, 43]. CCL21 is a chemoattractant for dendritic cells, cytotoxic T cells and natural killer cells and can improve the immune response. It has been tested as an immunotherapeutic agent in preclinical and clinical settings as a single agent and combination with chemotherapy [15, 16, 34]. The combination with chemotherapy had a synergistic effect [34]. This could be true for EWS as well, considering the increased expression CCL21 in patients with a good chemotherapeutic response. However, prior to administration of CCL21 immunotherapy, determination of CCR7 expression in EWS samples may be needed, as high expression of CCL21 and CCR7 expression in tumor cells was found to have negative effect and, one out of the 24 tested EWS samples showed high CCR7 expression in tumor cells. For this case, CCL21 administration might have resulted in an adverse effect, but further studies are needed to draw concrete conclusions. In addition, the potential of CCL21 treatment in not CCL21-primed tumors, meaning no CCL21 expression was present, should be further investigated.

In conclusion, in this study, we showed that patients with increased CCL21 RNA expression have a better EFS and OS. In addition, protein expression of CCL21 and its receptor CCR7 were not detected in all but one sample of EWS cells, indicating the absence of pro-tumorigenic paracrine and autocrine loops the majority of EWS cases. This tumor entity could therefore serve as a good target for an immunotherapy approach based on the use of CCL21. Furthermore, expression levels of CCL21 might be used as a potential prognostic marker for survival.