Background
Breast cancer is the most common malignancy among women and the leading cause of cancer-related death in women globally, with more than 90% of mortalities associated with stage IV metastatic breast cancer [
1]. Breast cancer commonly metastasises to bone marrow, lung, liver, lymph nodes and brain tissue [
2]. Four molecular subtypes, namely luminal A, luminal B, HER2-positive and triple-negative breast cancer (TNBC), have been associated with different patterns of metastatic spread; bone metastases with luminal A and B subtypes, liver metastases with HER2-positive subtype, and brain and lung metastases with TNBC and HER2-positive subtypes [
3‐
5].
Metastases develop when malignant cells lose their connection to the primary tumour (dissemination), becoming circulating tumour cells (CTCs) which can be transported in blood to distant regions of the body [
6‐
8]. Alongside single CTCs, circulating tumour cell clusters (CTC clusters) have been detected in the blood of cancer patients [
9‐
13]. CTC clusters, which are disseminated cell aggregates of up to 50 tumour cells, have been shown to have an up to 50-fold higher chance of forming metastases than single CTCs, and are therefore associated with worse clinical outcomes [
9,
14‐
16]. This higher metastatic potential may be attributed to the increased resilience of CTC clusters in circulation, compared to single CTCs, which in turn increases the probability of dissemination into a distant organ [
12,
17]. Further, it has been shown that CTC clusters originate as cell aggregates from the original primary tumour, rather than through intravascular aggregation or proliferation of singular CTCs [
9,
18]. However, the cellular and micro-environmental factors which promote CTC cluster formation are still largely unknown.
Desmosomal proteins have been previously described to be associated with cell aggregation and, in particular, high plakoglobin expression levels have been found in CTC clusters [
9,
19,
20]. Desmosomes are cell junctions which stabilise the connection between neighbouring epithelial cells. In desmosomes, the transmembrane proteins desmocollin 1–3 and desmoglein 1–4, which belong to the cadherin superfamily, form homo- and heterophilic interactions and are intracellularly connected to the intermediate filament cytoskeleton through desmosomal plaque proteins, such as plakoglobin and plakophilin (members of the armadillo family) and desmoplakin (member of the plakin family of cytolinkers) [
21‐
23]. One theory is that an abundant expression of desmosomal proteins could lead to enhanced intercellular adhesion between disseminating tumour cells, both during CTC formation and vascular transport [
9,
18,
24]. To date, studies focusing on the expression of various desmosomal proteins in a variety of primary cancer types have shown both tumour-suppressive and tumour-promoting effects on growth and metastases [
9,
25‐
27]. Recently, the desmosomal protein desmoglein 2 (DSG2) has been associated with a poorer prognosis and higher recurrence risk in breast cancer patients. DSG2 expression is regulated by hypoxia in breast cancer cells and increases the prevalence of CTC clusters, facilitating distant metastasis [
25].
This study focused specifically on desmocollin 2 (DSC2)—a transmembrane cell anchoring protein—in primary breast cancer, for which the role in breast cancer metastases formation is still not completely understood. Several studies have found that DSC2 proteins are abnormally expressed in various types of cancer and correlate with cell proliferation and invasive behaviour [
28,
29], and showed that a high expression of DSC2 increased cell aggregation [
20,
30]. A recently published study by Li et al. accentuates that elevated DSC2 expression, in combination with the desmosomal protein Plakophilin-1 (PKP1), can activate PI3K/AKT or CDH1 to increase cluster formation to resist shear-stress-induced cell death. Furthermore, higher expression of DSC2 and PKP1 was correlated with lower overall survival and worse disease progression in patients with breast and lung cancer [
31]. The aim of this study was to investigate the potential of DSC2 at mRNA and protein level as a predisposing factor for breast cancer progression and the development of breast cancer metastases, in particular to the lung and brain.
Methods
Patient cohorts
All patients, from whom the tissue samples were derived, were treated at the University Medical Centre Hamburg-Eppendorf, Germany, Department of Gynaecology between 1991 and 2002. All patients gave written approval for the utilisation of their tissue samples and the reviewing of their medical records according to our investigational review board and ethics committee guidelines (Ethik-Kommission der Ärztekammer Hamburg, #OB/V/03). Further cohort details and patient characteristics are listed in the Additional file
2: Table S1. Microarray analyses of DSC2 mRNA levels in patients with and without distant metastases were performed on 197 mRNA extracts from primary breast cancer tissue samples. For the western blot, a total of 111 samples were collected based on tissue availability from the same patient cohort. Slides of four tissue microarrays, constructed under permission of the Ethikkommission Beider Basel (EKBB # 395/11), kindly provided within collaborative efforts in the frame of the Pathobiology study group of the EORTC by Dr. Serenella Eppenberger-Castori from the Biobank at Institute of Medical Genetics and Pathology at the University of Basel, Switzerland, were used for immunhistochemical analysis. Patient characteristics are supplied in Additional file
3: Table S2.
Microarray data
We analysed DSC2 mRNA levels using microarray data (Affymetrix HG-U133A) from the aforementioned cohort. Here, two probe sets (204750_s_at and 204751_x_at) corresponding to DSC2 were available and analysed independently. Additionally, the mean expression value of the 2 probe sets was calculated and also included in further analyses. According to the DSC2 mRNA values of each probe set and the mean value, the cohort was divided into quartiles of similar size, representing low, moderate-low, moderate-high, and high DSC2 levels. Correlations between DSC2 mRNA levels (quartiles) and clinicopathological factors, such as histological grading, stage, lymph node involvement, oestrogen, and progesterone receptor status (ER, PR) were statistically examined by χ
2-tests. Overall survival was analysed by Kaplan–Meier analysis and log-rank tests. Additionally, the correlation between DSC2 mRNA levels (continuous data) and disease-free and overall survival was calculated using Cox regression analyses. Multivariate Cox regression analyses including the clinical stage, nodal involvement and molecular subtype were performed for all probe sets and the DSC2 mean value. Here, a backwards analysis with stepwise removal of insignificant terms was used. Probability values less than 0.05 were regarded as statistically significant. All statistical analyses were conducted using SPSS software Version 26 (SPSS Inc., Chicago, IL, USA). For validation purpose we used an independent Affymetrix microarray dataset consisting of 572 breast cancer samples from Gene Expression Omnibus (GSE2603, GSE2034, GSE12276) for which detailed information on metastatic localization was available [
32].
Protein lysate preparation and western blot analysis
Tissue samples were obtained intraoperatively and immediately stored in liquid nitrogen as fresh frozen samples. The histological characteristics of each sample were evaluated on cryo-cut and haematoxylin–eosin-stained sections. The tissue was tailored, where necessary, to obtain at least 70% tumour cells in the sample used for protein extraction. Approximately 100 mg of tissue was excised and pulverised using a micro-dismembrator (Braun-Melsungen, Melsungen, Germany) for 2 min and 45 s at 200 r.p.m. Proteins were lysed in ice-cold sample buffer (50 mM Tris pH 6.8, 1% sodium dodecyl sulphate (SDS)), 10% sucrose and 10 μl/ml protease inhibitor cocktail (Sigma, Taufkirchen, Germany). For western blot analyses, volumes of tumour lysates containing 20 μg of protein were loaded per well. The following antibodies were utilised in the western blot detection process: mouse monoclonal anti-DSC2 IgG (Millipore, MABT411) dilution 1:1000, mouse monoclonal anti-β-Actin (C4) (Santa Cruz Biotechnology, sc-47778) dilution 1:2000 and goat anti-mouse IgG-HRP (Santa Cruz Biotechnology, sc-2055) dilution 1:8000. Antibodies were visualised using a chemiluminescent reagent (SuperSignal ® West Pico chemiluminescent Substrate, Thermo Scientific, Rockford, USA). Protein band intensities were quantified using a calibrated densitometer (GS-800 Imaging Densitometer, Bio-Rad, Munich, Germany). The primary breast cancer protein lysate UPA497 was used as a positive control for DSC2, with its DSC2 expression being defined as 100% for the purpose of standardisation. Protein expression values in all detected bands were also normalised using the loading control β-Actin. For the statistical analyses, these values were divided into four equal groups (quartiles), representing very low, low-moderate, moderate and high protein expression.
Breast cancer cell lines, cell culture and stable transfections
The human TNBC cell line MDA-MB231 and its brain seeking subline MDA-MB231-BR were provided by Dr Takara (University of Texas). Cells of both lines were cultivated in Dulbecco’s Modified Eagle’s Medium (DMEM, ThermoFisher Scientific, Waltham, MA, USA) supplemented with 10% fetal calf serum (FCS) under standard cell culture conditions. Cells were authenticated before usage. Two different DSC2-knockdown MDA-MB231-BR cell lines were generated by lentiviral transduction using vectors containing shRNA-sequences targeting specific regions of the DSC2 mRNA sequence (MISSION shRNA III and V, Sigma-Aldrich, GmbH). Similarly, a control cell line was established using a scramble shRNA sequence (Addgene, plasmid ID1864). The full DSC2 cDNA sequence obtained from a commercially available vector (Des476-Desmocollin 2-myc Plasmid; Addgene Plasmid ID: 32233) was cloned into LeGO-iC2-Puro + Plasmid (kindlykindly provided by AG Fehse, Center for Oncology, Department of Stem Cell Transplantation, UKE, Hamburg, Germany) using BamHI and EcoRI restriction enzymes. After lentiviral production in HEK293T cells, MDA-MB231 cells were transduced. The corresponding empty vector was taken as a negative control. After selection with puromycin (2ug/mL), the level of DSC2 mRNA and protein was detected using real time quantitative polymerase chain reaction RT-PCR and western blot analysis, respectively.
RNA isolation and real-time quantitative polymerase chain reaction
RNA was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and a QIAshredder (Qiagen, Hilden, Germany), and was subsequently reverse transcribed using qScriber cDNA Synthesis Kit (HighQu, Kraichtal, Germany) according to the manufacturer’s instructions. RT-PCR was carried out with ORA™ qPCR Green ROX H Mix (HighQu, Kraichtal, Germany) using the StepOnePlus System (Applied Biosystems, Thermo Fisher Scientific Inc.). The data analysis was performed using the ΔΔCt method. The following primers were used for DSC2: forward primer, 5’-GCCCATCTTCTTCTTGTCGTT-3’; reverse primer, 5’-CCCGTCTTGGTGAAAAAGTGT-3’. Primer sequences for the housekeeping gene were as follows: forward primer, 5’-GTCAGTGGTGGACCTGACCT- 3’; reverse primer, 5’ -TGCTGTAGCCAAATTCGTTG-3’.
Immunofluorescence
Cells (1 × 105) were seeded on coverslips, cultured for 48 h, and fixed with 3.7% formaldehyde for 20 min at room temperature. After blocking with 1% bovine serum albumin (BSA) in phosphate buffered saline (PBS) (Mg+/Ca+) for 1 h at room temperature, cells were incubated with a polyclonal DSC2 antibody (1:50 in 1% BSA/PBS; Sigma-Aldrich, Hamburg, Germany) over night at 4 °C. Subsequent to washing, cells were incubated with a second antibody solution (mouse anti-rabbit IgG Alexa Fluor® 488; Jackson ImmunoResearch, Ely, UK; 1:500 in 1% BSA/ PBS) for 1 h at room temperature. Coverslips were carefully placed on slides using mounting medium and DAPI (Vectashield). Images were acquired using a fluorescence microscope BZ9000 and the software BZII Viewer (Keyence, Germany).
Proliferation assay
For cell proliferation analyses, the Cell Proliferation Kit II (XTT) (Roche Applied Science, Mannheim, Germany) was used according to the manufacturer’s instructions. Briefly, cells were seeded in a final volume of 100 µl medium per well in a 96-well plate (1.5 × 103 MDA-MB231-BR cells per well, 2 × 103 MDA-MB231 cells per well). After 24 h, 48 h and 72 h, cell viability was determined by adding XTT labelling mixture and by measuring the absorbance after 6 h at 490 nm using a microplate reader (DIAS Max002, Dynex Technologies, Chantilly, USA). Each experiment was performed with 12 replicates (wells) per condition (n = 12). Images shown are representative of three independently performed experiments.
Cytotoxicity analysis
For the MDA-MB231-BR cell line and the DSC2-knock down sublines, an Annexin-V/PI staining was performed after cisplatin treatment in order to quantify the extent of apoptotic and necrotic cells. Briefly, cells were seeded into 6-well plates at a density of 2.5 × 105 cells per well, incubated for 24 h and treated with cisplatin (Accord Healthcare Limited, North Harrow, United Kingdom) in three different concentrations (10 µM, 25 µM, 50 µM) for 48 h using serum-reduced DMEM medium. Subsequently, cells within the supernatant, as well as adherent cells, which were carefully detached using AccuMax (eBioscience, San Diego, CA, USA), were stained with an APC-labelled Annexin-V antibody (Annexin-V-APC, AnxA100, MabTag GmbH, Oldenburg, Germany) for 30 min at 4 °C in the dark. After washing with PBS (+/+), cells were resuspended in 1% BSA in PBS and stained with PI (BD Pharmingen, San Diego, CA, USA). FACS analysis was performed using the FACS Calibur (BD Biosciences, Heidelberg, Germany) and all data were analysed using FlowJo Software.
DSC2 overexpressing and control MDA-MB231 cells display a strong fluorescence, due to the transduction with the previously mentioned LeGO-iC2-Puro + Plasmid, which includes the mCherry-coding sequence. For these cell lines, the cisplatin-induced cytotoxicity was assessed using XTT, as described above. Briefly, MDA-MB231 cells were seeded into 96-well plates with 2 × 104 cells in 100 µl per well. After incubating for 24 h, cells were treated with cisplatin in three different concentrations (10 µM, 25 µM, 50 µM) for 48 h and the Cell Proliferation Kit (Roche) was used as described in the previous section. Each experiment was performed in duplicates. Images shown are representative of three independently performed experiments.
Migration assay
To investigate cell migration, the Oris™ Universal Cell Migration Kit (Platypus Technologies, Madison, WI) was used according to the manufacturer’s protocol. Briefly, cells were seeded (5 × 104 cells in 200 µl per well) in a 96-well plate fitted with sterile silicon stoppers using serum-reduced medium (5% FCS). After 24 h incubation, stoppers were gently removed allowing cells to migrate into the central cell-free detection zone. Migration potential was assessed by analysing the cell-free area of each well at four different time points (0 h, 24 h, 48 h and 72 h post removal of stoppers) with the ImageJ Wound Healing Tool (Wayne Rasband, National Institute of Health). Each experiment was performed with 12 replicates (wells) per cell line. Images shown are representative of three independently performed experiments.
Invasion assay
Matrigel Growth Factor Reduced (BD Biosciences, Heidelberg, Germany) was diluted to a concentration of 3.5 mg/ml with serum-free medium. Afterwards, 96-well plates were coated with a 1:1 mixture of Matrigel and serum-reduced medium (5% FCS) and incubated for 30 min at 37 °C. Hereafter, Oris™ Universal Cell Migration Kit (Platypus Technologies, Madison, WI) was used according to the manufacturer’s protocol as described above. After a 24 h incubation period, the stoppers and medium were carefully removed and 40 μl of newly prepared Matrigel coating solution was added. Plates were incubated again for 30 min at 37 °C. Finally, serum-reduced medium was added to all wells. For determining cell invasion potential, analyses were performed as described above. Each experiment was performed with 12 replicates (wells) per cell line. Images shown are representative of three independently performed experiments.
Cells were seeded at 5 × 103 cells in 200 μl medium per well on 2% agarose-coated (UltraPure™ Agarose, Invitrogen, Carlsbad, CA, USA, dissolved in PBS) 96-well plates. To assess and observe spheroid formation of cells, spheroids were examined and documented every second day using light microscopy and a camera (Axiovert 40 C, Carl Zeiss AG, Leica DFC320, Wetzlar, Germany). For further investigation of compactness, spheroids were dissociated by pipetting each spheroid up and down five times and by comparing formation immediately afterwards. Each cell line was seeded in quadruplicates. Images shown are representative of three independently performed experiments.
Cytotoxicity in a 3D Model
Cisplatin-induced cell cytotoxicity on 3D-structures was assessed using immunocytochemical detection of phosphorylated gamma H2AX (γH2AX)—an established marker for DNA double-strand breaks—on tumour cell aggregates grown on polyHEMA (Sigma-Aldrich) coated flasks. Here, cell lines were seeded at a density of 2 × 106 cells in 12 ml per polyHEMA coated T75 culture flask and cultured as cell aggregates for 72 h. Subsequently, cells were incubated with cisplatin in a final concentration of 50 μM for 8 h and 24 h, and were subsequently fixed in formalin, embedded into 2% agar (Agar NOBEL, Difco Laboratories, Detroit, MI, USA) and then embedded in paraffin. Slides from FFPE-cells were pre-treated in a steamer (citrate buffer pH 6.0) at 125 °C for 4 min and S1699 (pH 6.0, DAKO) at 121° C for 10 min, respectively) and incubated with an anti-DSC2 (1:25 HPA011911, Atlas Antibodies, Sigma-Aldrich, Hamburg, Germany) or an anti-γH2A.X antibody (1:10,000; ab81299, abcam, Berlin, Germany) for 1 h at room temperature. Incubation with biotin-labelled swine anti-rabbit secondary antibody occurred for 30 min at room temperature (1:200 dilution in TBS, E0353, DAKO, Glostrup, Denmark). For detection, sections were incubated with Vectastain® ABC-AP Kit (Vector Laboratories, Burlingame, CA USA) for 30 min and stained with Permanent Red (K0640, DAKO, Glostrup, Denmark). Rabbit immunoglobulin normal fraction (X0903, Agilent, Santa Clara, CA, USA) was used as a negative control for the anti-yH2A.x primary antibody. A rabbit polyclonal IgG (ab37415, abcam, Berlin, Germany) was used as a negative control for anti-DSC2 primary antibody. All slides were slightly counterstained with haematoxylin. Stained slides were scanned using the Axio Scan.Z1 (Zeiss, Jena, Germany) and images were acquired using netScope Viewer Pro software version 1.0.7079.25167 (NetBase Software GmbH, Freiburg, Germany). For quantifying the γH2A.X staining, manual counts of positive stained cells (4 × 100 cells in 4 different areas of each slide) were performed using the assistant electronic memory counter Counter AC-15 (Karl Hecht Assistant, Altnau; Switzerland).
The intracardiac mouse model was conducted as previously described [
33]. Briefly, female 8 to 9-week-old SCID mice (CB17/lcr-Prkdcscid/lcrlcoCrl) were anesthetized and 1 × 10
6 tumour cells were injected intracardially into the left ventricle of the heart (n = 15 per group). Tumour cells were previously transduced with luciferase-bearing plasmid and bioluminescence signals were tested before injection via bioluminescence imaging (BLI). After intracardiac injection, tumour cell dissemination was monitored weekly under BLI. Assessment of subsequent metastases was monitored in vivo weekly by imaging for up to 3 weeks. Mice showing termination criteria were immediately sacrificed. At the endpoint (21 days), animals were anesthetized and blood was collected from the left ventricle by cardiac puncture immediately before the final killing was executed by cervical dislocation. Ex vivo bioluminescence imaging was conducted from the lungs and brain. Lungs and brain were equally divided and frozen down for DNA isolation and subsequent ALU-PCR or paraffin-embedded for further analysis (H&E and luciferin staining) as previously described [
33,
34]. The animal experiments were approved by the Authority for Social Affairs, Family, Health, and Consumer Protection of the Free and Hanseatic City of Hamburg through application N005/2020.
Histology and immunohistochemistry
The whole brain and the right lung of the mice were fixed in 4% buffered formalin and processed for wax histology. 4 µm sections were cut from brain for immunohistochemistry and 10 sections from the middle of the block were stained with hematoxylin and eosin (H.E.). The lungs were fixed en block and subsequently cut into 1 mm thick slices and embedded in 2% agar. Afterwards, the lung slices were paraffin-embedded and cut into 4 µm thick sections. Ten sections of each paraffin wax block were H.E. stained and metastases were counted at a 200-fold magnification using Zeiss Axiophot photomicroscope (Zeiss, Jena, Germany). Additionally, two series of serial sections out of the middle of each paraffin wax block were preserved for further immunohistochemical analyses. The immunohistochemical staining was performed on 4 µm sections. Sections were deparaffinized in descending ethanol concentrations and pre-treated with citrate buffer solution (pH 6.1) in a steamer for 4 min. After incubation for 1 h at room temperature with the primary antibody DSC2 (Atlas, HPA011911), samples were washed twice with TBS-T (TBS + 0.1% TWEEN-20) and once with TBS for 5 min. After incubation with anti-rabbit secondary antibody (LS-Bio, LS-C350860) for 30 min at room temperature, antibody binding was visualized using the Vectastain ABC-AP Kit (VectorLabs., Burlingame, CA, USA) and Permanent Red Solution (Dako) according to the manufacturer’s instruction. The nuclei were counterstained in Mayer’s hemalum solution.
Circulating tumour cell detection
Mouse blood samples (200–500 µl) were obtained via cardiac puncture and collected into EDTA KE/1.3 tubes (Sarstedt, Germany). To perform cardiac puncture, mice were deeply anaesthetized under isoflurane, and a 21-gauge needle coated with heparin was inserted into the heart. Mice were euthanized immediately following the cardiac puncture. Blood samples were processed on the label-independent, microfluidic system Parsortix® (ANGLE plc., United Kingdom), a device designed for the size-based capture of rare cells from whole blood [
35]. The isolated cells were harvested and spun onto a glass slide (190 g, 7 min). Slides were dried overnight at room temperature and stored at − 80 °C until further analysis.
Tumour cells isolated with the Parsortix® system were identified via immunocytochemistry. Briefly, dried cytospin slides were brought to room temperature and fixed with 2% PFA (Sigma Aldrich, Germany) for 10 min. The samples were washed with 0.5 mL of 1x-PBS prior to permeabilization with 0.1% Triton X 100/PBS (Sigma Aldrich, Germany) for 10 min. Following two additional wash steps, 10% AB-serum/PBS (BioRad, Germany) was applied for blocking (60 min). Standard detection of CTCs is usually achieved with epithelial antibodies [
36], however, TNBC cells lack epithelial markers and are successfully detected with CD298 [
37]. Subsequently, directly anti-human PE labelled CD298 (clone LNH-94, Biolegend, USA) and anti-mouse Alexa Fluor 488 conjugated CD45 (clone HI30, Biolegend, USA) antibodies were incubated for 60 min, followed by 5 min of DAPI-incubation (1 µg/mL). Cytospins were covered with Prolong Gold Antifade Reagent (Thermo Fisher Scientific, Dreieich, Germany), sealed with a cover slip and examined by fluorescence microscopy (Axio Observer 7, Zeiss). CD298-positive, DAPI-positive, CD45-negative cells with intact morphology were defined as tumour cells. Clusters were defined when 2 or more cells were found together.
Statistical analyses
All statistical analyses were performed using SPSS Statistics version 24 for Windows (IBM, Armonk, NY, USA). Correlations between mRNA and protein expression values were assessed using two-sided Pearson tests. Chi-square tests were used to correlate both mRNA expression (Microarray data, cohort A) and protein expression (WB data, cohort B) with the following clinical and pathological parameters; histological grading (G1/G2/G3), molecular subtype (Luminal/HER2 positive/TNBC), ER and PR Status (positive/negative) and the presence of metastases (loco-regional/bone/lung/visceral/brain). Kaplan–Meier estimates and the log-rank test were carried out to ascertain and compare disease-free and overall survival. The associated hazard ratios for the multivariate analyses were determined by Cox regression. Proliferation assays and cytotoxicity assays measured with XTT were statistically analysed using GraphPad Prism 5 (GraphPad, La Jolla, CA, USA). Each in vitro assay was performed at least three times. Statistical significance was assessed using unpaired two-tailed Student’s t-test. The assumption of homogeneity of variances was checked via Levene’s Test of Equality of Variances (p > 0.05). Results are given as mean ± s.d. or s.e. Probability values (p-value) ≤ 0.05 were considered to be statistically significant.
Discussion
Desmosomes are important structures for intercellular adhesion and are functionally present most abundantly in tissue exposed to high levels of mechanical stress, such as the epidermis and myocardium [
38]. In recent years, desmosomal proteins have become a point of interest in cancer research. Depending on protein type and primary tumour localisation, both tumour-enhancing and tumour-suppressive effects of desmosomal protein up/down-regulation have been observed [
39]. Desmoglein 2, desmocollin 2 (DSC2) and plakophilin 1 (PKP1) have been recently linked to an increased metastatic potential of breast cancer cells by promoting cell clustering and enhanced survival during the tumour cell dissemination process [
25,
31]. In the present study, we have analysed the role of DSC2 as a prognostic and predictive factor for primary breast cancer and the development of breast cancer metastases. In order to address this question, DSC2 levels at mRNA and protein levels were correlated with clinical and histopathological data. Our research has been able to show, for the first time, that higher levels of DSC2 in primary breast cancer tissue significantly influence disease progression and metastatic behaviour in HER2 positive and TNBC patients. Functionally, the extent of DSC2 expression in tumour cells directly impacts their capacity to aggregate and, in turn, influences their chemosensitivity, as shown in in vitro analyses after DSC2 up-regulation and DSC2 silencing in the TNBC cell line MDA-MB-231 and its brain-seeking subline MDA-MB-231-BR, respectively. In vivo DSC2 knock down reduces the amount of circulating tumour cells and clusters, and consequently the amount and size of established brain metastases and established metastatic lesions in lung tissue.
As mentioned, desmosomal proteins and specifically DSC2 have been previously investigated in the context of cancer research. For example, in colorectal cancer, DSC2 loss enhances tumour cell growth by altering the Akt/β-catenin signal pathway [
29]. Furthermore, knockdown of desmosomal proteins such as DSC2, DSG2 and plakoglobin was reported to impair cell aggregation and reduce anoikis resistance in lung and breast cancer cells, while their expression levels were correlated with poor overall survival in lung cancer patients and poor metastasis-free survival in breast cancer patients [
19]. A similar effect of DSC2 and DSG2 mediated cell adhesion on cell aggregation was detected in colon cancer spheroids [
20]. Aceto et al. detected an up-regulation of plakoglobin and other desmosomal components in CTC clusters of breast cancer patients and revealed its importance for the formation of CTC clusters and distant metastases [
9]. And, in a more recent study, breast and lung cancer cells resistant to shear stress revealed an up-regulation of DSC2 and PKP1, leading to more CTC cluster formation and enhanced cell survival in circulation via activation of the PI3K/AKT/Bcl-2 pathway [
31]. Here, our findings suggest that the aforementioned DSC2 mediated effect on tumour cell aggregation and survival applies for TNBC as well.
In the present study, we found that tumour DSC2 levels significantly influence the disease-free and overall survival of breast cancer patients, in particular for patients with primary tumours corresponding to the HER2 positive and TNBC molecular subgroups. Two independent microarray datasets and a western blot cohort corroborated the unfavourable prognostic role of DSC2 and, together with immunohistochemical analysis on an independent cohort, demonstrated a significantly increased expression of the desmosomal protein in the most aggressive molecular subtypes, namely the aforementioned HER2 positive and TNBC. Our results have further highlighted the potential of DSC2 as a predisposing marker for the development of breast cancer metastases to the brain and lungs, and are in line with a previous work by Landemaine et al. who identified DSC2 as a potential predictive marker for lung metastasis in breast cancer [
40]. With the rising incidence of breast cancer cerebral metastases, and the lingering difficulty in treating the disease, the ability to identify high risk patients who would benefit from increased prevention would be of great clinical value. In line with a previous study on brain metastasis samples from patients with different tumour entities, which identified DSC2 as a potential marker for brain metastasis development [
27], we found that high DSC2 mRNA expression significantly correlated with an increased risk for cerebral and lung metastases, although for the first localization, this trend could not be validated at a protein level. High DSC2 protein expression was, however, significantly associated with the development of pulmonary metastases [
41].
Our findings challenge the self-evident hypothesis that up-regulation of desmosomal proteins leads to a more mechanically cohesive primary tumour, and therefore a less aggressive one. Indeed, a possible explanation for the contrary findings is that enhanced tumour cell aggregation through DSC2 up-regulation is a factor which favours the development of CTC clusters, which have a considerably higher metastatic potency than singular CTC [
9,
18]. In various studies, apoptotic morphology was detected in single CTCs, but not within CTC clusters. This finding supports the theory that clustering of CTCs leads to a higher metastatic potential by, for example, increasing the probability of survival in the circulation [
11,
42]. Marrella et al. showed that shear stress affects CTC survival in vitro, with CTC clusters being more resistant to shear forces than single CTCs [
43]. Increasing shear stress values incrementally caused disaggregation of CTC clusters. Shear stress resistant cells were found to express more DSC2 and DSC1 [
31], and were more likely to form clusters. Additionally, a link has been drawn between increased CTC cluster density and size, and increased cluster cell survival in vitro [
44]
. Additional to an increased resilience, the formation of clusters—in particular those with a very high density—may lead to reduced chemosensitivity and, thus, a survival advantage compared to single CTCs [
12,
44,
45]. Enhanced cell–cell interactions in CTC clusters could also confer resistance to anoikis, a form of apoptosis due to deprivation of cell–cell and cell–matrix contacts [
12,
19]. Furthermore, heterogenous cluster formation with immune cells, such as macrophages or leucocytes, may provide a mechanism for immune escape [
42]. Collectively, these findings indicate that DSC2 mediated cell adhesion is probably of greater functional importance in later steps of the metastatic cascade, such as survival within the circulation and chemoresistance, rather than in the process of dissolution of future metastatic cells from the primary tumour mass.
In the present study, up-regulation of DSC2 in breast cancer cells led, as expected, to an enhanced cellular aggregation capacity and thus the formation of tight 3D cell clusters, while tumour cell aggregates after DSC2 silencing displayed a looser structure which rapidly dissociated when subjected to mechanical stress. Interestingly, higher DSC2 expressing aggregates showed lower apoptotic rates than the corresponding control clusters when treated with cisplatin and, correspondingly, reduced DSC2 expression significantly enhanced tumour cell response to cisplatin. In contrast, we did not observe any effect of DSC2 up- or down-regulation on the chemosensitivity of both TNBC cell lines cultured as a monolayer, indicating that the DSC2-mediated cohesiveness of the 3D tumour cell clusters is the main reason for the altered chemosensivity. This finding highlights the relevance of in vitro 3D culture models to accurately mimic the in vivo conditions [
46]. Tumour cell aggregation significantly influences cell response to cytotoxic drugs, as cells in a spheroid environment are more resistant to radiation and chemotherapeutic agents, a phenomenon known as multicellular resistance (MCR) that has been described for different anticancer drugs, including cisplatin [
47]. Possible mechanisms of MCR include signalling-mediated inhibition of apoptosis, an increased proportion of quiescent cells, as well as reduced permeability and, in turn, impeded drug diffusion. Li et al. recently described high DSC2 and PKP1 levels in shear stress-resistant breast and lung cancer, which facilitate cell cluster formation and also activate the PI3K/AKT/Bcl-2–mediated pathway, thereby increasing cell survival [
31]. However, the exact mechanism behind the observed DSC2-mediated chemoresistance in our model remains unclear and needs to be elucidated in the context of an ongoing project.
Under 2D culture conditions, up- or down-regulation of DSC2 was found to have only a minor effect on cell migration and invasion, although the tendency of our results is in line with recent studies on adhesion proteins (for example DSG2) and cell migration [
25,
48]. No severe impact of increased or decreased DSC2 expression on cell morphology, proliferation or apoptosis could be detected in our 2D cell culture. These findings contrast with diverse reports describing a clear effect of DSC2 down-regulation on 2D proliferation and/or apoptosis in, for example, breast, prostate or oesophageal squamous cell carcinoma in vitro [
28,
31,
49], with both a pro- and anti-tumorigenic role being postulated. Thus, the role of DSC2 seems to be entity-specific, or even subtype-specific, as shown in our study with a clear negative prognostic value of DSC2 in HER2 positive and TNBC, yet no impact on survival for patients with luminal breast cancer.
The results of the in vivo metastatic model, even though the size of the experiment did not allow a significant conclusion, clearly underline our hypothesis. Reduced DSC2 tumour expression decreases the amount of viable tumour cells in the blood circulation—CTCs as well as CTC clusters—and, as a consequence, reduces the effective formation of distant metastases. Our results are in line with those recently published by Li et al. showing the relevance of dual expression of DSC2 and PKP1 for cluster formation and survival in circulation in a lung cancer cell line in a tail vein injection model [
31].
Declarations
Competing interests
The authors declare no conflict of interest concerning the presented analysis. IW received speaker´s honoraria outside this work from Amgen, Astra Zeneca, Daiichi-Sankyo, Lilly, MSD, Novar-tis, Pierre Fabre, Pfizer, Roche, and Seagen. VM received speaker’s honoraria from Amgen, Astra Zeneca, Daiichi-Sankyo, Eisai, GSK, Pfizer, MSD, Medac, Novartis, Roche, Teva, Seagen, Onkowissen, high5 Oncology, Medscape, Gilead, Pierre Fabre; consultancy honoraria from Hexal, Roche, Pierre Fabre, Amgen, ClinSol, Novartis, MSD, Daiichi-Sankyo, Eisai, Lilly, Sanofi, Seagen, Gilead; institutional research support from Novartis, Roche, Seagen, Genentech and travel grants from Roche, Pfizer, Daiichi Sankyo, Gilead.BS received speaker´s honoraria, travel grants and consultancy honoraria as well as institutional research support outside this work from Eisai, Astra Zeneca, MSD, Pfizer, Roche, GSK, Clovis, and Ethicon. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results”.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.