Breast cancer (BC) is a significant global public health issue and the leading cause of death among women around the world. Anti-cancer therapies including chemotherapy, endocrine therapy, and targeted therapy have significantly decreased BC mortality in the past 20 years. However, as BC is an extremely heterogeneous disease, resistance to treatment is a major clinical challenge for current BC management [1
]. Therefore, the development of more specific biomarkers and recognition of new therapeutic targets would really contribute to solving the problems of therapy resistance and metastasis [2
]. In recent years, an increasing number of clinical studies have suggested that an optimal vitamin D status has a protective effect against BC development and that high Vitamin D receptor (VDR) expression in breast tumors is associated with a better survival rate [5
]. As one of the nuclear receptor (NR) members, VDR is found in both normal breast tissue and breast tumors [9
]. As such, an analysis and understanding of the VDR pathway can probably provide a novel way for developing a new targeted therapy to escape resistance mechanisms. The group of NRs that are active as homodimers have been classified as type 1 NRs, whereas the NRs of the VDR group that bind as heterodimers are known as type 2 NRs. The type 1 group includes, among others, estrogen, progesterone, and androgen receptors and the type 2 group contains VDR, retinoic acid receptors (RARs), retinoid X receptors (RXRs), and thyroid hormone receptors (THRs). Ligand-bound VDR-activated vitamin D3
heterodimerizes with its cognate co-receptor RXR to control the expression of genes involved in its different functions [11
]. Besides its classic functions to maintain extracellular calcium levels by regulating calcium absorption in the gut and bone turnover, the VDR–RXR heterodimer binds to vitamin D response elements with the positive or negative transcriptional regulation of gene expression involved in various molecular pathways. This results in a wide range of calcitriol-mediated anti-cancer actions in BC [12
]. We therefore believe that VDR exploration is very relevant to evaluate its potential as a new prognostic biomarker and therapeutic target in BC.
Circulating tumor cells (CTCs) circulate in the peripheral blood of patients with solid malignancies and are shed from an existing primary tumor or from metastatic lesions into the blood stream [14
]. CTCs detected in BC patients are significantly associated with a poor outcome in both early and metastatic tumors [15
]. In metastatic patients, several tumor lesions may potentially release CTCs which therefore comprehensively reflect tumor and metastasis characteristics. CTCs can be collected via a simple venipuncture; this ‘liquid biopsy’ achieves the repeatable and real-time monitoring of tumor cell characteristics. It is a less invasive and cost-effective alternative to tissue biopsies [20
], despite the fact that technical and conceptual advances are still necessary before this ‘liquid biopsy’ can be routinely used for the diagnosis, characterization, monitoring, and treatment optimization of cancer. CTCs are a promising marker, providing important predictive and prognostic information in both early and metastatic BC. They may help to assess the response to treatment and to detect early disease recurrence [21
]. At the moment, the CellSearch®
system for CTC enumeration is the only accepted standard by the Food and Drug Administration (FDA). Only a few studies have investigated human epidermal growth factor receptor 2 (HER2) and/or estrogen receptor (ER) expression on CTCs, even though HER2 and ER are currently the only validated predictive factors used for therapy decision making in BC [22
]. In conclusion, the characterization of CTCs may be a major tool to support diagnosis, and should be included in clinical trials for the evaluation of new targeted therapies [23
]. In order to better predict disease progression and personalize treatment, new prognostic and predictive factors are needed. So far, studies on VDR status in CTCs are still lacking. Therefore, the evaluation of VDR expression on CTCs in BC patients could potentially help in individualizing BC therapy.
In this study, we describe an innovative triple fluorescence technique that we developed to simultaneously visualize the presence of cytokeratin (CK), absence of CD45, and expression of VDR. We first characterized BC cell models, before validating the preclinical data in CTCs from 23 metastatic BC patients.
VDR has been shown to be expressed in different tissues, as well as in BC cells; however, it has not yet been investigated in BC CTCs from archived specimens. We previously used various human carcinoma cell lines to develop a simple and efficient triple fluorescence technique for CTC receptor analysis, e.g., ER, HER2 [15
-cadherin, and CD133 [26
]. We tested the VDR expression in nine BC cell lines and CTCs from 23 archived metastatic BC cases.
system, which is the only Food and Drug Administration (FDA) approved CTC enumeration method used for clinical purposes, classifies a CTC as a positive event if the nucleated cell is ≥4 µm, pan-CK positive, and CD45 negative [27
]. More than a 90% expression of CK7, CK8, CK18, and CK19 was observed in breast carcinomas of all grades, thereby confirming their usefulness for BC identification [28
]. Meanwhile, peripheral blood cells (PBCs) such as monocytes (i.e., PBMCs) only express very low mRNA levels of CK (18/19) [29
]. Therefore, using a CK antibody combined with a CD45 antibody, a recognized white blood cell (WBC) marker, allows for the characterization of CTCs by CK positive and CD45 negative staining—a technique that has been the most frequently used so far [30
]. An additional fluorescence channel is accessible for a user-defined detection of therapy relevant markers. The CellSearch®
system (but also other techniques) allows for an analysis of markers such as ER [31
], HER2 [15
], epidermal growth factor receptor (EGFR) [34
], and epithelial-mesenchymal transition (EMT) associated molecules such as N
]. Nevertheless, VDR has not yet been evaluated by this technology.
After optimization of the triple fluorescence protocol on BC cell lines, we observed that all nine tested cell lines were VDR positive, as already reported in other studies by Western blot analysis or other techniques [24
]. We then characterized the expression heterogeneity among all of the cell lines. Limited publications can be found with regard to the intensity of VDR protein expression in BC cell lines, as most of them focus on mRNA expression [8
]. According to our fluorescence analysis, we were able to divide the cell lines into three groups with distinct levels of VDR expression. Standardized identical values of exposure time for optimal pictures of VDR and CK expression were absolutely required for evaluation and comparison (namely 1000 ms for VDR and 2000 ms for CK). As already mentioned in the literature, T47D appeared to be among the high expressing cell lines (although not the highest) and MDA-MB-231 among the low expressing cell lines [36
]. Besides these differences in the average fluorescence intensity for each cell line, we observed that even within one cell line, individual cells expressed variable VDR levels, thus explaining how mRNA expression cannot always be linked to protein expression. As an example, most MCF-7 cells exhibited a substantial level of VDR fluorescence, but the expression was not identical for all cells. Some cells barely exhibited any fluorescence intensity (2.1%) and others expressed a particularly high intensity (around 4.7%). We speculate that this heterogeneity within one cell line may rely on the cell cycle–related variations of VDR expression or on a wide variety of environmental factors such as cell adhesion and cell density [42
]. Of note, given the individual heterogeneity of expression within one cell line, we characterized and graded the VDR level by semi-quantifying the average values of different cell lines as the majority of cells showed a consistently low, intermediate, or high intensity.
In order to perform VDR expression analysis on blood samples collected from a consecutive cohort of 23 metastatic BC patients, we first mimicked the CTCs analysis in blood by mixing cancer cell lines with PBMCs from healthy donors. PBMCs are identified as CD45 positive and CK negative, and in our study, only a few of them expressed VDR. This observation is in accordance with the literature stating that VDR expression is controlled by immune signals [44
]. It is noteworthy that normal human PBMCs may also express VDR and its target genes [45
]. Using our triple fluorescence method, we detected VDR expression in 42 CTCs, present in 60.8% of our patient cohort with 1 million PBMCs analyzed in each case. Our cohort only consisted of metastatic patients as they have the highest probability of exhibiting CTCs. We first observed very atypical morphologies and fluorescence patterns with cells positive for CK and negative for CD45, and secondly with cells positive for CK and for CD45, which could be CTCs because dual CK and CD45 positive cells are occasionally found in humans [48
]. Indeed, false-positive CK staining may also occur, as it is possible that antibodies against CK bind to hematopoietic cells through the Fc receptor [50
]. We thus decided to exclude both of these groups from further analysis.
All of the studied BC cell lines expressed VDR from low to high levels. Based on the MCF-7 positive controls, we therefore grouped CTCs into two VDR status groups. We could demonstrate that 45.2% (n
= 19) of the CTCs were VDR positive. In terms of VDR intensity, VDR staining in CTCs was relatively low compared to the high VDR expression observed in some BC cell lines. Optimal vitamin D status has a protective effect against BC development, but epidemiological and early clinical studies are inconsistent. Resistance to vitamin D develops or exists in many BC patients [51
]. The anticancer role of vitamin D is mainly mediated by the VDR. Our hypothesis is that VDR may be abnormally (poorly) expressed in BC tissue and/or CTCs. It is known that VDR is lost during carcinogenesis and this may be the reason why tumors become insensitive to vitamin D [25
]. Therefore, it would be very helpful to compare the VDR expression of CTCs with that of the corresponding primary tumor. Unfortunately, primary tumor tissue was not available for our cohort. In a previous study [7
], Nina Ditch et al. analyzed the relationship between VDR expression in primary tumor tissue and survival in 82 BC patients. Patients with high VDR expression showed significantly better progression-free (PFS) and overall survival (OS) results than patients with moderate/negative VDR expression [7
]. In the 13 CTC positive patients of our study, six (46.1%) had at least one VDR positive CTC, with three (23%) patients only having VDR positive CTCs. In contrast, 10 (76.9%) had at least one VDR negative CTC, including seven (53.8%) patients with only VDR negative CTCs. Therefore, we believe that VDR expression may be of clinical significance and that the CTC results need to be correlated to PFS or OS in a larger cohort of patients. Similar conclusions on the role of VDR expression as a prognostic marker have already been addressed in pancreatic cancer and gastric cancer [52
]. In a large patient population, VDR expression on primary tumor tissue is inversely associated with more aggressive BC including a large tumor size, HR negativity, and triple-negative subtype (p
< 0.05) [54
]. A preclinical study suggested that calcitriol and inecalcitol, an epi-analogue of calcitriol, can inhibit BC cell line growth, especially in cells expressing ER and VDR [55
]. This suggests that the VDR-mediated inhibition of ER-positive BC cells may be at least partly affected by the downregulation of ER [56
]. Besides, in contrast to ER-positive cells, treatment with calcitriol was reported to induce the expression of ER in the ER-negative cell line. If confirmed in patients, this ability of calcitriol would have major implications for BC treatment [59
]. In order to see whether it could be a potential biomarker, we correlated the VDR expression results observed in CTCs with the related clinicopathological parameters. Most likely due to the small number of patients and only one cytospin analyzed per patient, no significant association was found between VDR expression in CTCs and tumor subtype according to ER, PR, and HER2 status. In further larger studies, it will be essential to correlate VDR expression in CTCs (with at least duplicate cytospins for each patient) with ER, PR, and HER2 status, as well as the VDR itself of the primary tumors. Moreover, repeat analysis of VDR expression on CTCs during the course of disease and after treatments may give very relevant information. Another parameter to consider in parallel will be the serum level of the partially activated 25-hydroxyvitamin D, and consequently 1,25di-hydroxyvitamin D that are expected to be relatively low in BC patients. A large number of studies have concluded that low blood levels of vitamin D are associated with an increased BC incidence and decreased survival in BC [60
]. Similar to postmenopausal patients with ER-positive tumors and extremely low serum estrogen levels, it could be speculated that moderate or low levels of 25-hydroxyvitamin D may be sufficient to activate VDR.
The physiopathological significance of the observed CTC size heterogeneity is also a crucial point that may lead to further analyses [61
]. The Cell Search®
system classifies a CTC as a positive event if the nucleated cell is ≥4 μm [27
]. We observed 24 (57.2%) CTCs that were “tiny” CTCs (around 5 μm, but higher than 4 μm) and 18 (42.8%) that had a normal size (≥5 μm). Prior publications by us and other research groups have already described and discussed the character of these “tiny” CTCs [15
]. There are diverse explanations for “tiny” CTCs: Compared to the size of CTCs in patients with metastases or primary tumors, the size in dormancy candidates is smaller [64
]. Furthermore, Marrinucci et al. found by DNA disruption analysis and microscopic images that the early apoptosis category of cells contains many CTCs that seem surprisingly small for carcinoma cells, suggesting that these small CTCs are undergoing cell death through apoptosis [66
]. Stem cell-like CTCs which are smaller and more aggressive than other CTCs could be another possibility [67
]. Similar findings were reported for disseminated tumor cells (DTCs) in bone marrow, where tumor cells with a stem cell-like phenotype were demonstrated [68
]. CTCs that are in the process of EMT may be as deformable as WBCs to become more WBC-like and better adapt to the blood flow based on this size issue [69
]. These findings likely explain some of the size heterogeneity we observed. However, further studies regarding the exact significance of these “tiny” CTCs in cancer research have not been able to give clear answers yet [70
]. Of note, the technical issue of CTC enrichment methods also has to be considered when the methods rely on a size filtration of CTCs from PBMCs [71
Some specific atypical subtypes of cells were also observed in our study. For example, patient M1’s had more than 500 CTCs in the 1 million PBCMs analyzed, and CK and VDR staining was faint and often superposed. For some CTCs in patients M9 and M25, the fluorescence labeling of CK and VDR performed a superposition with a very high expression of VDR and the phase contrast image of some CTCs shows a specific morphology with a very faint aspect. We speculate that these specific subtypes derive from some stressed, evolving cells that may be going through the apoptosis process: When CTCs are shed from solid tumors, half of these CTCs perish within 2.4 h in vivo [65
]. CTCs impaired through apoptotic events exhibit membrane perforation that triggers the leakage of intracellular components, including not only electrolytes or small molecules, but also DNA and chromatin. The technical steps of blood drawing and sample processing can induce the additional stress or degradation of CTCs by various factors. These parameters include the selected purification and analysis techniques, such as temperature shock, fluidic turbulence, shear force, or surface tension. As a result of the various harsh conditions, it is expected that fragmented cells, cellular debris, microparticles, and clump-like aggregates next to “normal” CTCs will be observed [72
]. Although we did not include these CTCs in our analysis, they can clearly play a role in the dissemination and metastasis process [26
4. Materials and Methods
4.1. Cell Culture and Cytospin Preparation
The human adenocarcinoma cell lines MCF-7, T47D, ZR-75-1, and MDA-MB-231, and an endometrial cancer cell line Ishikawa (Heraklio) 02 ER- (Ishikawa ERneg), were obtained from the European Collection of Cell Cultures (ECACC, Salisbury, UK) and the Cama-1 (HTB-21), SK-BR-3 (HTB-30), HCC1937 (CRL-2336), and MDA-MB-468 (HTB-132) cell lines from the American Type Culture Collection (ATCC, Rockville, MD, USA). The HCC 3153 cell line was kindly provided by Adi F. Gazdar (Hamon Center for Therapeutic Oncology Research and Department of Pathology, University of Texas Southwestern Medical Center at Dallas, Dallas, TX, USA). Cryopreservation of cell cultures ranged from passages 1 to 10. Cells were used during up to 20 passages. Cells were grown routinely in Dulbecco’s modified Eagle’s medium (Biochrom, Berlin, Germany), supplemented with 10% FBS (PAA, Pasching, Austria).
For cytospin preparation, trypsinized cells were centrifuged (700 g, 10 min, 4 °C) and resuspended in phosphate-buffered saline (PBS; Biochrom, Berlin, Germany). Then, 1 million cells were spread on each cytospin and centrifuged (45 g, 5 min, room temperature). Cytospins were allowed to dry overnight at room temperature and then stored at −80 °C. They were prepared with either 1 million adenocarcinoma cells or mixed with PBMCs from healthy volunteer donors, as indicated in the legends.
4.2. Triple Fluorescence Labeling of CK, VDR, and CD45 with Parallel Phase Analysis
According to the optimized procedure (see Supplementary Material
), cytospins were thawed and immediately fixed in 3.7% neutral buffered formalin (Fischar, Saarbrücken, Germany) in PBS for 15 min at room temperature and permeabilized in cold (−20 °C) methanol (Sigma-Aldrich, Steinheim, Germany) for 2 min. After washing in PBS, Ultra V Blocking medium (Thermo Scientific, Fremont, CA, USA) was used for 15 min. This blocking step and all of the following steps were performed in a humidified chamber at room temperature. All antibodies were diluted in Dako Antibody Diluent with Background Reducing Components (Dako, Carpinteria, CA, USA).
As previously described [77
], we selected a two-step protocol. Cells were first incubated for 45 min with a monoclonal mouse anti-human VDR antibody (clone 2F4, mouse IgG2a, MCA3543Z, Serotec, Puchheim, Germany) efficiently used in other studies on BC [7
], washed in PBS, incubated for 30 min with a goat anti-mouse IgG-Fab fragment labeled with Cy3 (Jackson ImmunoResearch, Suffolk, UK), and washed in PBS.
Cells were then incubated for 45 min with a monoclonal rabbit anti-human CD45 antibody (D9M8I, 13917, Cell signaling, Leiden, The Netherlands) and a monoclonal mouse anti-human cytokeratin antibody (Igg1 A45 B/B3, Glycotope, Berlin, Germany), washed in PBS, incubated for 30 min with a goat anti-rabbit IgG labeled with Coumarin-AMCA (Jackson ImmunoResearch) and a goat anti-mouse IgG labeled with DyLight488 (Jackson ImmunoResearch), and washed in PBS.
After drying (30 min, at room temperature), the slides were mounted with Kaiser’s glycerol gelatin (Merck, Darmstadt, Germany) before manual analysis with a computerized fluorescence microscope Axioskop (Carl Zeiss Micro Imaging GmbH, Göttingen, Germany) for phase and fluorescence, with 40× magnification. An AxioCam MR camera and AxioVision software (version AxioVision LE 4.8, Göttingen, Germany) were used to capture, analyze, and save high-resolution images for the three fluorescence channels, considered independently or in combination. Criteria for CK and CD45 positivity were the ring-like appearance (cytoplasm staining in periphery) and we considered that the VDR positivity was always high, average, and low for specific punctuated staining of the nucleus with a low background and no cytoplasmic or peripheral staining. We followed the criteria already described for the identification of CK and CD45 positive CTCs by immunofluorescence [78
] and the consensus recommendations for standardized tumor cell detection [79
Definite threshold values of exposure time for VDR, CD45, and CK fluorochromes were determined, on the basis of the analysis of the cancer cell lines, and were systematically used for the patient analysis described below.
4.3. Fluorescence Labeling of VDR and Cytokeratin (CK) with Parallel 4′-6-Diamidino-2-Phenylindole (DAPI) Analysis
Cytospins were thawed, immediately fixed, permeabilized, and blocked as described above.
Cells were incubated as described above, first with a monoclonal mouse anti-human VDR and related secondary anti-mouse IgG-Fab fragment labeled with Cy3, and then with the monoclonal mouse anti-human cytokeratin antibody and the related goat anti-mouse IgG labeled with DyLight488, without the anti-CD45.
After drying (30 min, at room temperature), the slides could be mounted with Vectashield Mounting Medium with DAPI (Vector Laboratories, Burlingame, CA, USA) before manual analysis with a computerized fluorescence microscope Axioskop (Carl Zeiss Micro Imaging GmbH, Göttingen, Germany) with 40× magnification.
4.4. Patient Cohort
This analysis was performed at the Department of Obstetrics and Gynecology, Ludwig Maximilian University (Munich, Germany). Twenty-nine metastatic BC patients were recruited between May 2010 to July 2012; with six exclusions (patient M5 with a cancer of unknown primary origin (CUP) syndrome and patient M3, M5, M14, M21, and M23 with an insufficient blood sample). Table 2
describes the characteristics of the final cohort of 23 patients and their primary tumors. Written consent forms were collected from the patients using protocols approved by the institutional ethics committee (approval number 148-12; 12.05.2012, Ethikkommission bei der Ludwig-Maximilians-Universität, Munich). The phenotype of the primary tumor was routinely assessed at the time of diagnosis by immunohistochemical staining and potentially by FISH, Fluorescence In Situ Hybridization (for most HER2 ++ patients) in the original Department of Pathology and collected from the patient files. ERα status was classified by an evaluation of the percentage of tumor-stained cells and staining intensity, allowing for an assessment of an Immunoreactive Score (% score × intensity score). HER2-negative assessment include 0 or + staining and ++ staining with FISH-negative amplification and HER2-positive assessments include ++ with FISH amplification or +++ staining. The ERα and HER2 status were indicated in Table 2
, if available. Because the HER2 status of the primary tumor of patient P7 was not determined in 1999, at the time of diagnosis, we considered the HER2-negative status of the local recurrence assessed in 2009, with ERα positivity and PR negativity recorded for the primary tumor.
4.5. Blood Sampling, Ficoll and Cytospin Preparation
Fifteen milliliters of blood from each patient was collected by needle aspiration and placed in EDTA tubes. The blood was processed by a modified Ficoll protocol, with Ficoll-Hypaque (Pharmacia, Erlangten, Germany) density gradient centrifugation (density 1.007 g/mol) at 900× g
for 30 min [80
]; the mononucleated cells or PBMCs, were counted and centrifuged (700 g, 10 min at 4 °C), and then 1 million cells were spread out on each cytospin and centrifuged (45× g
, 5 min, room temperature), before being processed as described above.
4.6. CTC Analysis by Triple Fluorescence
The triple fluorescence labeling of VDR, CK, and CD45 with parallel phase analysis was performed on the cytospins prepared from patient blood, as described above. The preparation of BC cell lines MCF-7 was mixed with PBMCs from healthy donors, in which MCF-7 cells served as a positive control for CK and VDR stainings and as a negative control for CD45; Mixed PBMCs served as a positive control for CD45 staining and as a negative control for CK. In each batch of patient samples we analyzed, one MCF-7 mixed with the PBMCs control slide was systematically performed. The screening of CTCs through CK staining was performed using a 20× magnification to get an optimal sensitivity, and the VDR and CD45 expressions were then assessed using a 40× magnification. For each patient, one cytospin was analyzed (1 million cells per patient) and each slide was evaluated by two independent investigators, and three in doubtful cases (X. Zhang, S. Sixou and U. Jeschke). For one patient (4.3%), the evaluation of the two observers differed for either the CTC detection or VDR positivity. These cases were re-evaluated by the three observers together. After the re-evaluation, the observers came to the same result. The concordance before the re-evaluation was 95.7%. The analysis was always performed within 72 h after the labeling procedure. Each observed CTC was recorded with at least one picture for each channel of analysis.
4.7. Statistical Analysis
A Fisher exact probability test was used to evaluate the relationship between the receptor status of the primary tumor and the VDR expression of the CTC for the 14 CTC-positive patients. p < 0.05 was considered statistically significant.