Introduction
Treg are a subset of CD4+ T cells that maintain self tolerance by exerting a plethora of immune suppressive activities on various immune cells—most prominently autoreactive T effector cells. Treg also play a detrimental role in anti-tumor T-cell responses in cancer patients. Infiltration of Treg into tumors is associated with enhanced tumor growth and poor prognosis [
1‐
3] and their therapeutic depletion through antibody-dependent cell-mediated cytotoxicity (ADCC)-proficient antibodies against a Treg-associated molecule CTLA-4 or low-dose cyclophosphamide treatment can improve anti-tumor T-cell responses and overall survival of patients [
4,
5]. Conversely, in tumors promoted by chronic inflammation, Treg infiltration is beneficial as it dampens inflammation [
6]. Together, data available until date clearly demonstrate the prognostic relevance of tumor infiltrating Treg [
7]. This is also the case in breast cancer. Many breast cancer patients not only develop tumor antigen-specific effector T-cell responses [
8‐
12], but also tumor specific Treg [
13]. Treg accumulation in breast tumors is associated with poor overall and relapse free survival and used as an independent prognostic factor in predicting patients at high risk of relapse [
14].
T-cell entry into peripheral tissues requires their exit from lymphoid organs, where they reside until their mobilization into the blood. Such an exit can be triggered by antigen-specific activation [
15] or by changes in chemokine gradients and related signals [
16]. The BM is an important lymphoid organ for Treg accumulation and recirculation and thus an important source of tumor infiltrating Treg [
17]. However, the events triggering Treg exit from BM into PB remain largely elusive.
Here we addressed a potential role for S1P1 in regulating Treg exit from BM into PB in breast cancer patients. S1P1 mediates immune cell exit [
18], particularly T cell exit from thymus and lymph nodes [
19,
20]. Sawicka et al. provided evidence that treatment of mice with an S1P1 agonist resulted in an accumulation of Treg in PB and spleen but not in lymph nodes, while the same treatment resulted in increased lymph node accumulation of naïve CD4+ T cells suggesting that S1P1 could be involved in differential sequestration of Treg [
21]. Cell surface residence of S1P1 is crucial to determine cell egress and is tightly regulated by S1P concentrations in the microenvironment thereby modulating trafficking patterns. It was reported that activation induces migration of lymph node Treg towards S1P [
22] and that Treg tend to lose S1P1 more slowly from the surface upon activation in contrast to Tcon [
23]. Hence, we here wondered whether in breast cancer patients S1P1 might be involved in the emigration of Treg populations from the BM.
Using PB and BM samples from breast cancer patients, we studied the expression, induction, and functional relevance of S1P1 and its ligand S1P in bulk and tumor antigen-specific Treg. Our findings suggest that upregulation of S1P1 on tumor antigen-specific Treg mediates their preferential exit from the BM.
Materials and methods
Patient samples
Collection of PB and BM samples from cancer patients and healthy donors was performed in compliance with the norms of the ethics committee [“HLA compatible allogeneic cell therapy of human tumors in mouse model’’ (approval reference number 70/99)] of the University of Heidelberg upon signed consent. Samples were withdrawn from patients (29–73-year-old Caucasians) who did not receive neo adjuvant chemotherapy and healthy donors without a history of breast cancer (21–63 years old Caucasians) after informed consent. Age range of patients and healthy donors was matched. Formalin-fixed paraffin-embedded (FFPE) tissue sections and cryosections of mammary tumors were provided by the tissue bank of the National Center for Tumor Diseases (NCT, Heidelberg, Germany) in accordance with the regulations of the tissue bank and the approval of the ethics committee of the University of Heidelberg.
Cell isolation
PB mononuclear cells (PBMCs) and BM mononuclear cells (BMMCs) were isolated using established protocols by Ficoll (Biocoll) gradient centrifugation. Cells from the interface were collected, washed twice with X-vivo 20 medium (Lonza), and used directly for phenotyping using flow cytometry or purified and taken into culture for in vitro assays.
Cell purification
HLADR+ primary APC were isolated using HLA-DR Microbeads Human (Miltenyi Biotec, 130-046-101) as per the manufacturer’s instructions. Alternatively, DC were differentiated in vitro as per standard protocols described earlier [
24]. CD4+ Tcon and Treg were isolated using Human CD4+ CD25+ Regulatory T-cell isolation kit (Miltenyi Biotec, 130-091-301) according to the manufacturer’s instructions.
Antibodies and flow cytometry
Tcon and Treg were phenotyped by flow cytometry (FACS Canto, BD Biosciences). Live cells were distinguished using Live/Dead Fixable Yellow Dead Cell Stain (Life Technologies, L34959) for 15 min at 4 °C followed by blocking with Kiovig (human plasma derived immunoglobulin, Baxter, PZN-4668835) at a concentration of 100 µg/ml in PBS for 15 min at room temperature. Monoclonal anti-human S1P1-APC (R&D systems, FAB2016A) and corresponding mouse IgG2B isotype control (R&D systems, IC0041A) were used at concentrations recommended by the manufacturer in PBS for 30 min at 37 °C. This was followed by surface staining for CD3-Alexa Fluor 700 (Invitrogen, CD0329), CD4-Brilliant violet 421 (BD Biosciences, 562424) and CD25-PerCp Efluor 710 (eBioscience, 46-0257-42) in PBS for 20 min at 4 °C. For some experiments, CD127-PECy7 (Biolegend, 351320) was used in conjunction to distinguish the CD127− Treg subsets. Cells were fixed and permeabilized using the fixation and permeabilization concentrate (eBioscience, 00-5123-43) and diluent (eBioscience, 00-5223-56) for 20 min at 4 °C followed by intracellular staining for FoxP3-FITC (Biolegend, 320106) in 1× permeabilization solution (eBioscience, 00-8333-56) for 20 min at 4 °C.
To detect antigen-specific T cells, DRB1*04:01 and DRB1*07:01 tetramers (PE conjugated) presenting mammaglobin as well as a negative control tetramer presenting a Class II-associated invariant chain (CLIP) peptide were obtained from NIH tetramer core facility (Emory University, Atlanta). Cells blocked with Kiovig were stained with tetramers (1 µg/100 µl PBS) for 1 h at 37 °C followed by staining with other antibodies as described above. HLA typing for the DRB1 locus was initially performed by FACS using mouse monoclonal HLADR-04 (Abcam, ab33903) or HLADR-07 (Abcam, ab34054-50) antibody for 15 min at 4 °C and detected using secondary rat anti-mouse IgM-APC (BD Pharmingen, 50676) and later reconfirmed by PCR. Data were collected using FACS Canto (BD Biosciences) with the FACS Diva software. Data analysis was done with FlowJo 8.8.6. Total number of events collected per sample varies between 0.05 mill and 0.8 mill cells. For all individual patients, Treg subsets were gated according to isotype control stainings (as shown in supplementary Fig. 1). CD25+ FoxP3+ Treg were the core population of Treg that were analyzed. In addition, CD25+ subset, CD25+ CD127− subset, and CD25+ FoxP3+ CD127− subset were also analyzed. GraphPad Prism 6 was used for graphical representations and associated statistics.
ELISA
Plasma from PB and BM of patients were collected by centrifugation at 3000 rpm for 5 min at room temperature. The clear supernatant was used for quantifying S1P concentrations using S1P ELISA kit (Echelon Boiosciences Inc, K-1900) as per the manufacturer’s instructions.
In vitro T-cell activation assays
T-cell activation assays were performed by coculture of purified Treg as described above with either MACS purified fresh HLADR+ primary BM APC or by differentiating DC in vitro with the standard protocols using GMCSF (560 units/ml) and IL4 (500 units/ml) for 7 days. Treg were maintained in X-vivo medium with 300 units/ml IL2 (for Tcon 100 units/ml IL2 was used) and 60 units/ml IL4. BM APC were cocultured with Treg in the ratio 1:10 along with polyclonal stimulation using the Staphylococcus aureus enterotoxin B (SEB) (1 µg/ml) or cocultured with APC alone or left unstimulated. On day 3, cells were analyzed by flow cytometry.
Migration assay
MACS purified primary APC were cocultured with Treg/Tcon in the ratio 1:10 along with polyclonal stimulation using SEB (1 µg/ml). Unstimulated cells were used as control. On day 3 after activation, calcein staining and migration assay were performed as described by Frevert et al. [
25]. Cells and chemoattractants were prepared in RPMI with 20 mM HEPES. 96 well-disposable chemotaxis plates (Neuroprobe) were coated with 10 µl of collagen (10 µg/ml in 10 mM acetic acid). Chemoattractants were added to the lower chamber, and 50,000 calcein labeled Treg were added to the upper chamber. After incubation for 1 h at 37 °C, cells that migrated to the lower well were read using a calcein reader (Perkin Elmer 2030 multi-label reader). Results obtained as fluorescence intensity units were normalized to cells that migrated to medium alone.
Epigenetic Treg assay
Tumor areas on FFPE breast tumor sections were demarcated by Hematoxylin Eosin (H&E) stained reference slides and subsequently scraped into eppendorf tubes manually. DNA isolation and epigenetic quantification of Treg based on amplifying the Treg Specific Demethylated Region (TSDR) by qPCR were performed at Epiontis, Berlin as previously described [
26]. Total Treg and CD3 T-cell numbers that infiltrated the tumor were quantified based on the number of GAPDH plasmid units obtained in each assay. Cell numbers were normalized to 1 mm
3 tumor volume.
Multicolour immunofluorescence staining and data acquisition using TissueFAXS
Intratumoral T-cell subpopulations were detected by a combination of primary antibodies: anti-CD3 (Dako, A0452, Host-rabbit), anti-CD8 (Clone YTC182.20, Abcam, Ab60076, Host-rat), and anti-FOXP3 (Clone 236 A/E7, Abcam, Ab20034, Host-mouse) on acetone-fixed breast tumor cryosections as described earlier [
27]. Primary specific secondary antibodies (anti-rabbit Alexafluor 647, A21245; anti-rat Alexa Fluor 488, A11006; anti-mouse Alexa Fluor 555, A31570) were purchased from Life Technologies. DAPI (Invitrogen, D1306) was utilized to stain cell nuclei.
Total tissue slides were scanned on Olympus IX51 microscope equipped with a F-View II camera (both Olympus) and analyzed by the TissueQuest Cell Analysis Software package (version 4.0.1.0137, TissueGnostics GmbH). For automated analysis with TissueQuest, DAPI staining was used as a master marker for cell identification on the basis of nuclei detection. Based on H&E stained reference slides, regions of interest (ROI) were defined to distinguish between tumor and surrounding non-tumor area. All tissues were analyzed with identical parameters for detection of T cells based on nuclear size, mean staining intensity, and background threshold. Cells were visualized in scattergrams, while the cutoff between positive and negative gated cells was validated manually by backward gating on the original image.
Statistical analyses
Distributions of data were described by mean with the standard error of the mean (SEM) or median with interquartile range as appropriate considering the form of the distribution. Consequently, the unpaired or paired t test or, respectively, the Wilcoxon matched pairs signed rank test were used to compare two different distributions test. Non-parametric Spearman Correlation was used to analyze the association between the frequency of Treg in BM and tumor. Two-way analysis of ANOVA with repeated measurements in both factors (concentration and stimulation status) and paired t tests as post hoc test was used for analyzing migration potential of Treg to S1P. *p value <0.05; **p value <0.01; ***p value <0.001; ****p value <0.0001; Differences were considered statistically significant when p < 0.05.
Discussion
While increased infiltration of Treg into solid tumors has often been described, factors driving their migration from lymphoid organs into cancerous tissues largely remain elusive. Previous work published by Peters et al. [
35] suggested that Treg trafficking patterns might be strongly perturbed in solid tumor patients. A potential role of the BM for Treg recirculation in cancer patients has been already suggested by Zhao et al. [
36]. Here, we show for the first time in breast cancer patients, major changes in Treg distribution between BM, PB, and tumor. This was characterized by an overall increase in PB, markedly decreased Treg frequencies in BM and an associated Treg accumulation in tumor tissue, suggesting that Treg might be mobilized from the BM into the blood and tumor. Because of high variations in overall Treg abundance between the breast cancer patients which might reflect different dynamics in Treg generation and because of potentially very short recirculation periods of activated, mobilized Treg in the blood, we could not directly identify an inverse correlation between their frequencies in the blood and bone marrow. However, we could show that in breast cancer patients, Treg distribution is inversely correlated between bone marrow and tumor tissue.
An exploratory assessment of CD25+ FoxP3+ Treg in 50 patients revealed that this characteristic Treg distribution pattern was observed in patients with different grades (Grades I, II, and III) with no significant differences with varying disease grades (data not shown). Interestingly, tumor antigen-specific Treg were significantly under-represented in the BM compared to the blood—pointing to the possibility of a link between TCR stimulation and Treg depletion in the BM. This was corroborated by our finding of selective S1P1 expression on tumor antigen-specific BM Treg. However, our in vitro activation experiments showed that TCR stimulation by CD3 CD28-coated beads alone did not induce S1P1 expression in Treg but that S1P1 surface expression on Treg required additional signals provided by APC emphasizing their importance in regulating S1P1 expression. In this regard, Shannon et al. have attributed a role for CCL19 in inducing S1P1 expression on murine T lymphocytes [
37]. As CCL19 can be secreted by APC subsets [
38], it would be interesting to unravel the nature of putative CCL19+ BM APC that could potentially present breast tumor antigens to Treg.
Since extracellular S1P is required for signaling via S1P1 on the cell surface, the exploitation of such S1P1 mediated mobilization of antigen specific Treg seems to be based on two conditions; first -a Treg selective induction of S1P1 through the presentation of cognate antigen by APC in the BM, and second—increased concentrations of S1P in PB plasma that have been previously reported in patients with solid tumors [
39]. Here, we show that in breast cancer patients, BM S1P levels are not elevated despite increased PB plasma S1P levels creating an S1P gradient between the BM and PB. Multiple cell types that can secrete S1P [
40] and produce sphingosine kinases that regulate the phosphorylation and release of S1P [
41] have been described and might contribute to the observed S1P gradient between BM and PB in cancer patients.
A putative role for S1P1 in regulating Treg trafficking from human BM to PB has not been studied so far, although S1P1 surface expression has been reported on human CD4+ T cells subjected to serum starvation [
42]. In our in vitro migration experiments, only TCR stimulated Treg migrated to S1P, whereas resting Treg did not. This result is in accordance with the work of Ishimaru et al. who previously reported that murine lymph node Treg activated with anti CD3 antibody migrated to S1P [
22], thus emphasizing the importance of Treg activation to respond to S1P signals.
Our finding that S1P1 was scarcely expressed by tumor antigen-specific Tcon but strongly and specifically induced in tumor antigen-specific Treg hints towards a role for BM APC in the regulation and maintenance of peripheral tolerance—e.g., through mobilizing populations of self reactive Treg for dampening chronic inflammatory processes at peripheral sites. The reduced migratory potential of activated Tcon to S1P could be explained by work of Liu et al. who demonstrated that after TCR stimulation, Treg maintain S1P1 surface expression, while Tcon rapidly loose S1P1 expression [
23]. Taken together, our study suggests that BM may be an important site for the activation of tumor antigen-specific Treg that get preferentially mobilized from the BM into PB via S1P1. Increased PB plasma S1P levels as well as the capacity of APC to induce S1P1 expression on Treg through TCR stimulation may contribute to such peripheral tolerance mechanism. In cancer patients, this preferential mobilization of antigen-specific Treg mediated by S1P1 from the BM might act as a preceding step in the Treg trafficking process resulting in their accumulation in tumor thus contributing to poor clinical responses. Therefore, disrupting increased S1P levels prevailing in PB plasma of cancer patients may represent a future therapeutic strategy to contain Treg within the BM thereby reducing Treg infiltration into tumors.
Acknowledgements
We thank the tetramer core facility at Emory University, Atlanta for providing mammaglobin specific HLA class II and control CLIP tetramers. We thank Dr. Maria Pritsch, documentation and data management, University Hospital Heidelberg for suggestions with statistical analysis and Epiontis GmBH for performing epigenetic studies.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (
http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.