Introduction
Mucins comprise a large family of cell surface and secreted proteins most commonly expressed by epithelial cells [
1], but they are also associated with other cell types such as the endothelial lining of vascular spaces [
2,
3]. Mucins are present on the apical surface of epithelial cells of gastro-intestinal, respiratory, breast, and reproductive tissues, and contribute to tissue lubrication, hydration, and protection. Mucins are defined by a serine/threonine-rich region within their extracellular domains that is heavily O-glycosylated, and the abundant O-linked glycans are largely responsible for the physico-chemical properties of mucins that contribute to epithelial protection [
4,
5]. It has recently become appreciated that a subset of these proteins, the membrane mucins that are physically tethered to the plasma membrane via a transmembrane domain, are capable of stimulating intracellular signaling pathways to contribute to cellular growth regulation [
6‐
8].
MUC4, a membrane mucin, is a non-covalently linked heterodimeric protein complex composed of the two subunits MUC4α and MUC4β arising from a single transcript. The enormous extracellular MUC4α subunit contains an O-glycosylation domain and a nidogen-related domain, followed by an AMOP domain towards the C-terminus. Glycans attached to repeating units within the O-glycosylation domain of the MUC4α subunit dominate the mass of MUC4, and contribute to its protective and anti-adhesive properties. The much more modest-sized MUC4β transmembrane subunit contains a von Willebrand factor D domain, and three epidermal growth factor-like domains that lie N-terminal to the transmembrane domain; these domains may be involved in protein-protein interactions that contribute to MUC4 function [
9‐
11]. A function for the short (about 20 amino acids) cytoplasmic tail of the MUC4β subunit has yet to be described [
12].
MUC4 expression has been reported in a variety of well-differentiated epithelial tissues in the adult including gastrointestinal tract, breast [
13,
14], and lung [
15,
16]. MUC4 expression has also been reported in a variety of carcinomas including ovarian [
17,
18], lung [
15,
19], pancreatic [
20,
21], gall bladder [
22], and breast [
23]. These observations are significant because MUC4 has been demonstrated to potentiate signaling by ErbB2 [
9,
11], a receptor known to contribute to the malignancy of breast and ovarian tumors, as well as other tumor types. In addition, the anti-adhesive [
24] and anti-apoptotic [
12,
25] properties of overexpressed MUC4 could provide tumor cells with a selective growth or survival advantage. Indeed, ectopic overexpression of rat MUC4 in a human melanoma model cell line increased primary tumor growth [
25] and metastasis [
26] efficiencies when introduced into nude mice.
Although work examining the impact of MUC4 on model tumor cell properties strongly supports the notion that the mucin can promote tumor progression, evidence that it might do so in human tumors has been harder to obtain. For example, while many studies document MUC4 expression in tumors, often analysis of matched normal tissue is lacking, raising questions as to the extent to which MUC4 is dysregulated in tumors. Moreover, the interpretation of expression studies has been hampered by the use of incompletely characterized antibodies that may not be entirely specific for MUC4. Here we develop a reliable reagent for the assessment of MUC4 expression in human tissues, and apply it to examine MUC4 expression in normal breast tissue, as well as in primary tumors and lymph node metastases. Unexpectedly, we find that MUC4 expression tends to be reduced in primary tumors relative to normal tissue, but is regained upon metastasis. Thus, re-expression of MUC4 by metastasizing cells could significantly augment their malignancy. Indeed, we further demonstrate that the presence of endogenous MUC4 in a cultured breast tumor line derived from a pleural metastasis promotes cell migration, proliferation and resistance to anoikis.
Materials and methods
Cell lines and cell culture
Human breast cell lines MCF10A, MCF7, MDA-MB-453, MDA-MB-435, MDA-MB-468, MDA-MB-231, MDA-MB-361, SKBR3, T47D, BT474, rat mammary tumor cell line MATB-III, and HEK293T cells were purchased from the American Type Culture Collection and cultured in their recommended media (Mediatech, Manassas, VA, USA). The JIMT-1 human breast cancer cell line [
27] and its MUC4 knockdown derivative [
12] have been previously described. A375-Rep8 and MCF7-Rep5 cells inducibly expressing rat MUC4 have been previously described [
11,
24]. Construction of the human breast cell line MCF10A-h MUC4/Y inducibly expressing the human MUC4/Y variant will be described elsewhere (Workman et al., in preparation).
Immunoblotting experiments
Primary antibodies were from the following sources: anti-MUC4 mouse monoclonal antibody 1G8s was used as conditioned media from hybridoma line #2D10, clone HL1718, and 1G8c was purchased from Zymed (Carlsbad, CA, USA); anti-MUC4 antibodies 8G7, P-20 and H-300 were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA); anti-actin and anti-α-tubulin were from Sigma (St. Louis, MO, USA). Horseradish peroxidase-conjugated secondary antibodies were from Invitrogen (Carlsbad, CA, USA), and SuperSignal West developing chemicals were from Pierce (Rockford, IL, USA). An Alpha Innotech (San Leandro, CA, USA) imaging station with FluorChem software was used to capture and quantify images.
Human breast tissue analysis
Fresh frozen human tissues from clinical samples were provided by the National Cancer Institute Cooperative Human Tissue Network and the National Cancer Institute-funded UC Davis Cancer Center Biorepository, and were used in western blotting experiments. All of the samples were approved for laboratory use by the Institutional Review Board of the UC Davis School of Medicine. Samples were homogenized in 10 μl T-Per (Pierce, Rockford, IL, USA) per mg of tissue in the presence of 4 μg/ml leupeptin, 4 μg/ml pepstatin, 4 μg/ml aprotinin, and 100 nmol 4-(2-aminoethyl) benzenesulfonyl fluoride, and then centrifuged to remove insoluble products. Cleared lysates were added to sample buffer and analyzed by immunoblotting.
Specificity of immunohistochemical immunoreactivity
Immunoreactivity was compared in cell lines expressing and not expressing MUC4. MCF10A-hMUC4/Y and A375-rRep8 cells stably expressing inducible MUC4 were treated with 100 ng/ml and 2 μg/ml tetracycline (Sigma, St. Louis, MO, USA) to induce and repress expression, respectively. JIMT-1-pSuper-shRNAi-hMUC4 and JIMT-1-pSuper-shRNAi-scramble stably transduced cell lines [
12] were compared with assess endogenous MUC4 expression. In each case, cells grown to about 70% confluency were scraped, pelleted by centifugation, fixed in 10% buffered formalin for one hour, stored in 70% ethanol and ultimately paraffin embedded.
Tissue micro-array staining
Unstained human tissue micro-array (TMA) slides BR451, BR480, BR481, BR701, BR721, BR722, BR801, BR1001, and BR1003 (with no overlapping cases) were purchased from US Biomax (Rockville, MD, USA). TMA samples had a core size of 1 to 2 mm and each core had a thickness of 5 μm. Matched assays containing normal tissues typically consisted of adjacent uninvolved tissue taken approximately 1.5 cm from primary tumor. Unmatched normal breast tissue came from patients of good health. TMAs were prepared as suggested by the manufacturer using the UltraVision LP Detection System (Thermo Scientific, Pittsburgh, PA, USA). Slides were deparaffinized with xylene and rehydrated with alcohol. Slides were incubated in 3% H2O2 (in deionized water) for 10 minutes to suppress endogenous peroxidase activity. Antigen retrieval was carried out by incubating slides in 1 mM ethylenediaminetetraacetic acid (EDTA), pH 8.0, for 15 minutes at 98°C. Slides were incubated for 1.5 hours with 1:25 1G8s, and counterstaining was performed using ImmunoMaster Hematoxylin (American MasterTech, Lodi, CA, USA). Immunoreactivity levels were assigned a value on a 0 to 3 scale (see below). Images were captured on an Olympus BX-40 using DP2-BSW software (Center Valley, PA, USA).
Expression analysis
Samples were examined by the primary author (HCW) and two additional pathologists/authors (ADB and RDC). Results were compiled and statistics were provided by an author statistician (LAB). Paired samples (normal vs. primary tumor or primary tumor vs. metastasis) were analyzed in two complementary ways. First, the fraction of times one sample within a pair stained more intensely exceeded 50% was tested; McNemar's test was used, excluding cases where both stained equally, and calculated the exact two-sided binomial probability for a disparity as extreme or more extreme if the true proportion were 0.5. Next, the difference in the mean staining score was tested, and the staining levels 0, 1+, 2+ and 3+ were treated as scale values and a paired t-test to test whether the mean difference was zero was used. All tests were two-sided at level 0.025, to ensure experiment-wise error rate below 0.05 that allowed for two comparisons (normal to primary tumor; primary tumor to metastasis). Individual data samples for immunoreactivity intensity from patients with tumor, normal tissue, or metastatic tumor were also compared. For these groups of patients, we carried out independent samples t-tests to compare the mean immunoreactivity levels between groups (patients with normal vs. those with primary tumor; patients with primary tumor vs. those with metastatic tumor). Again, all tests were two-sided at level 0.025.
Migration assay
JIMT-1 scramble and knockdown cells were seeded in triplicate at 4 × 104 cells per well in 24-well Boyden chambers with 8 μm pore polycarbonate membranes (Corning, Corning, NY, USA), using complete media in the upper chamber and serum starve media in the lower chamber. After 16 hours, the lower chamber was replaced with complete media and the upper chamber replaced with serum starve media. Cells were allowed to migrate for 18 hours, filters were fixed, stained using the Diff-quik system (Dade Behring, Newark, USA), and photographed using an Olympus DP70 200× objective and DP Controller software (Center Valley, PA, USA). Three fields of view for each well were quantified by counting all the cells in each field and averaging the results for each condition.
Anoikis and cell cycle analyses
JIMT-1 derivatives were plated in ultra-low attachment 60 mm or 100 mm flat-bottom plates (Corning, Corning, NY, USA) in complete media containing 1% methyl cellulose (Sigma, St. Louis, MO, USA), and grown for 96 hours. Suspended cells were then collected by centrifugation, washed in PBS, and fixed in 4% paraformaldehyde for one hour followed by 70% ethanol for one hour to overnight at 4°C. Cells were then rinsed in PBS and incubated for 30 minutes in propidium iodide solution (Sigma, St. Louis, MO, USA), EDTA, RNAse, and spermine (Sigma, St. Louis, MO, USA) SubG1 and cell cycle analysis of 15,000 to 30,000 cells per sample was carried out using a Becton-Dickinson fluorescent-activated cell sorting scanner using Cellquest software by Becton Dickinson (Oakville, Ontario, Canada) and ModFit software vy Veity (Topsham, ME, USA). Anoikis data is presented as a Forest meta-analysis plot [
12]. Odds ratios were calculated using the sub-G1 positive and non-positive populations for cells harboring scrambled and MUC4 knockdown shRNAs and plotted using Graphpad Prism software (La Jolla, CA, USA).
Discussion
A number of studies underscore the notion that MUC4 may be capable of contributing to the malignant properties of tumor cells. Inducible expression of rat MUC4 in human A375 melanoma cells has been demonstrated to augment primary tumor growth [
25] and metastasis [
26] in a nude mouse xenograft model.
In vitro studies using the inducible rat MUC4 expression model and other cell lines indicate that ErbB2-dependent and ErbB2-independent signaling activities of MUC4 contribute to its proliferative and anti-apoptotic functions [
9,
11,
12,
35]. Moreover, the abundant O-linked glycosylation of MUC4 contributes to its anti-adhesive properties [
24], masks antigens on tumor cell surfaces and inhibits cell killing by cytotoxic lymphocytes [
36]. Collectively, these observations raise the possibility that dysregulation of MUC4 in patient tumors can confer properties to tumor cells that promote tumor progression.
Although ample
in vitro and nude mouse model evidence exists that dysregulated MUC4 can potentially play a role in human tumors, evidence that it does so has been harder to obtain. The most convincing data come from pancreatic carcinomas, where normal tissue lacks MUC4 and expression increases with the progression of the disease [
37]. MUC4 is commonly expressed in pancreatic tumor cell lines, and knockdown has been demonstrated to suppress pancreatic tumor cell proliferation, survival, and invasive properties [
38]. However, involvement of MUC4 in the progression of other tumor types has been more difficult to assess because normal tissues express abundant MUC4 and because cell lines often rarely express the protein. In these cases MUC4 overexpression relative to normal tissue, MUC4 mislocalization in cells that have lost their polarity, or re-expression of lost MUC4 in more advanced tumor stages, can all markedly impact disease progression but can be difficult to detect and characterize.
Another significant challenge in discerning MUC4 involvement in the progression of many tumor types concerns its detection. Early studies employed
in situ hybridization methods to detect the MUC4 message in normal and tumor tissue [
39,
40]. Although such studies provide information on where MUC4 protein can possibly be expressed, observations that rat MUC4 may be post-transcriptionally and post-translationally regulated by factors such as transforming growth factor-β and other basement membrane components [
41] raise questions as to the extent to which this method can be used to accurately assess MUC4 protein expression by tissues. More recently immunohistochemical methods have been employed to assess MUC4 expression. The antibodies most commonly employed are 8G7, raised to a peptide within the repeating units of the O-glycosylation domain, and 1G8, originally raised to rat MUC4β but more recently demonstrated to react with human protein. In our hands, both of these antibodies obtained from commercial sources recognized bands that could not be knocked down with MUC4-specific RNA interference in immunoblotting experiments, raising the possibility that staining observed in immunohistochemical studies with these antibodies includes unrelated proteins. Moreover, as 8G7 is raised to an epitope whose post-translational modification could interfere with immunoreactivity, this antibody may preferentially recognize an underglycosylated subset of MUC4 in tissues.
In our study we develop a preparation of the 1G8 antibody that specifically recognizes MUC4β by immunoblotting, and whose immunoreactivity with cultured MUC4-positive breast cancer cells by immunohistochemical staining is markedly suppressed when MUC4 expression is knocked down. The reason underlying the difference in specificities between the commercial and hybridoma sources is unknown, but may be related to differences in antibody production. This reagent allowed us to localize MUC4 to the luminal surface of normal breast epithelium, observations that recapitulate those made with rat mammary glands [
42]. Unexpectedly, we observed that MUC4 expression levels tended to be suppressed in primary tumors relative to normal tissue, whether examining patient-matched sample pairs or individual patient samples. The simplest explanation for these observations is that MUC4 expression is a marker for fully differentiated breast epithelium, and dedifferentiated breast tumor cells are impaired in their ability to support MUC4 expression. MUC4 expression is regained in many lymph node metastases relative to primary breast tumor, raising the possibility that the presence of MUC4 confers an advantage to metastasizing tumor cells. Consistent with these observations, our MUC4 knockdown experiments reveal that its expression contributes to the aggressive properties of breast tumor cells. Interestingly, a recent study found that MUC4 expression levels in primary prostate tumors is lower on average than in normal or benign hyperplastic tissue [
43], although patient-matched tissues were not employed. In light of our findings it would be interesting to determine whether prostate metastases similarly regain MUC4 expression and contribute to prostate tumor cell aggressiveness. Such observations would lend support to the broader notion that MUC4 presence in metastasizing carcinomas contributes to tumor malignancy.
Collectively, the accumulated data point to a scenario where re-expression of MUC4 by a subset of primary breast tumor cells promotes their metastasis via several mechanisms. Overexpression or mislocalization of heavily glycosylated MUC4 by a subset of cells within a primary tumor population can contribute to the disruption of cell-cell and cell-matrix interactions, which in turn facilitates the migration of tumor cells away from the primary tumor and into the circulatory or lymphatic systems. Moreover, the anti-apoptotic signaling properties of MUC4 [
12] can minimize the chances that primary tumor cells that have lost adhesion and are undergoing metastasis will undergo anoikis. Based on these arguments, we would predict that MUC4 overexpression might be particularly prevalent in actively metastasizing cells such as circulating tumor cells, or in metastatic cells that accumulate in abdominal or pleural effusions. Indeed, given the potent anti-adhesive properties of MUC4, these tumor cell populations could express very high levels of MUC4 protein, which may again be suppressed to some degree upon metastatic seeding of a solid target tissue.
Finally, if expression facilitates metastasis then MUC4 could ultimately serve as an independent prognostic marker of the most aggressive tumors. Patients whose primary tumors exhibit elevated MUC4 expression could be at higher risk of developing metastases than those whose MUC4 remains suppressed. However, it should be noted that MUC4 expression by a small subset of cells within the primary tumor mass may be sufficient to facilitate metastasis, and this population could easily be overlooked by immunohistochemical analysis. Moreover, as we have observed that pre-malignant atypical hyperplasias and DCIS have not yet suppressed their MUC4 expression, such studies would necessarily need to couple MUC4 expression analysis with careful pathological evaluation.
Acknowledgements
This research was supported by NIH grants GM068994 (KLCIII), CA118384 (CS), CA52498 (KLC), T32 RR07038 (HCW), and by a grant UL1 RR024146 (LAB) from the National Center for Research Resources (NCRR), a component of the NIH. The research was also supported by US Army Medical Research and Materiel Command fellowship W81XWH-06-1-0402 (JKM) and by Fort Dodge Animal Health Fellowship (HCW). The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the funding agencies.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
The study was conceived and designed by HCW, CS, and KLC III, and the majority of the experiments were carried out by HCW. Pathology analysis was carried out by HCW, ADB, and RDC. Western blotting analysis of human tumors was carried out by JKM and EQI, and RPK contributed to tumor cell growth analysis. DIY, LJTY, and KLC contributed critical reagents and advice, and LAB carried out statistical analysis of the data. All authors read and approved of the final manuscript.