Background
Human breast carcinoma (HBC) is the most predominant cancer in females worldwide. Breast cancers can be classified histologically based upon the types and patterns of cells of which they are composed. Carcinomas can be invasive, extending into the surrounding stroma or non-invasive, confined to the epithelial cells of ducts or lobules 1. In conjunction with other protease systems such as the serine, cysteine and aspartyl proteases, members of the matrix metalloprotease (MMP) family of proteolytic enzymes degrade constituents of the extracellular matrix surrounding invasive breast carcinomas 2. Currently, 28 human MMPs have been identified and classified according to both their substrate specificities and structural similarities. There are four major subgroups: i) interstitial collagenases; ii) gelatinases; iii) stromelysins; and iv) the membrane-type (MT) -MMPs 34. Collectively, MMPs degrade all extracellular matrix proteins as well as a growing number of key regulatory genes such as cytokines, growth factors, cell surface receptors and adhesion molecules 56. Although MMPs are expressed by tissues at various stages of development, they are typically absent in normal cells of the adult organism 2, and the high frequency at which MMP transcripts or proteins are detected in invasive tumours has implicated these enzymes in the establishment, growth, invasion and/or metastasis of tumours 6. The expression of MMPs is usually tightly regulated, and a number of studies provide evidence of carefully controlled MMP involvement in developmentally regulated processes such as ovulation, embryogenic growth and differentiation, and organ development 678. In general terms, MMP activity is regulated at least at three levels: transcription/translation, proteolytic activation of the zymogen, and inhibition of the active enzyme 2 with upregulation of each of these associated with pathological events 6. Importantly in all mammalian cells except polymorphonuclear (PMN) leucocytes, most MMPs are not stored intracellularly but are rapidly secreted after biosynthesis and post-translational processing 9. As a result it is difficult to identify their cellular origin using standard immunohistochemical techniques 10.
MT1-MMP is one of the membrane bound MMPs. In vivo, MT1-MMP expression has been localised to the stroma surrounding breast tumours 1112, whilst in vitro, our recent data confirms previous studies where basal levels of MT1-MMP have been shown to be higher in the more invasive MDA-MB-231 cells as compared to the less invasive MCF-7 cells 1314. MT1-MMP has been shown to activate pro-gelatinase-A (proMMP-2) in human breast carcinoma cells 15 and can also activate proMMP-13 16. MT1-MMP degrades native type I collagen, fibronectin, laminin, fibrin, gelatin and cartilage proteoglycan core protein 2 and has also been classified as an interstitial collagenase 3. MMP-1 is the most ubiquitously expressed interstitial collagenase 3. MMP-1 cleaves fibrillar collagens including collagens types I, II and III, resulting in cleavage products that are rapidly denatured at body temperature to gelatin, the substrate preferred by the gelatinases (MMP-2 and MMP-9) 2. MMP-1 is produced by a wide variety of normal cells (eg stromal fibroblasts, macrophages and endothelial cells) 3, and is involved in tissue remodelling and repair 317. It has been implicated in matrix invasion by the MDA-MB-231 HBC cells in vitro 1819, but has a low incidence of expression in the tumour cells of breast carcinomas 211. MMP-3 is a member of the stromelysin sub-family, which show broad substrate specificity, degrading type IV collagen, laminin, fibronectin and proteoglycans. MMP-3 is expressed in areas of tissue growth 20, focally expressed around invasive cells in the stromal component of breast tumours, and expressed in both benign and malignant breast phenotypes 19.
As MT1-MMP, MMP-1 and MMP-3 have all been implicated in the processes involved in human breast cancer, this study utilised HBC cell lines both in culture and grown as xenografts in nude mice, to investigate gene expression changes of MT1-MMP, MMP-1 and MMP-3 in vitro and in vivo using IS-RT-PCR. In order to demonstrate IS-RT-PCR detection to examine MMP expression level changes, we utilised Concanavalin A (Con A), an agent known to modulate MMP expression, and to upregulate at least one MMP 152122, in our in vitro studies. Nude mouse xenografts are heterogenous, containing neoplastic human parenchyma with mouse stroma and vasculature with many histologic features of the original tumour maintained 23. As xenograft growth varies with tumour type, the ability to mimic human tumours in vivo provides a reproducible model system that can be correlated to the human disease state. We have previously applied the IS-RT-PCR technique to examine in vitro gene expression of MT1-MMP in MDA-MB-231 HBC cells 21, while others have examined expression of MMP-2, MMP-9 and their inhibitors TIMP-1 and TIMP-2 in cervical cancer 24. Thus the objective of this study was to utilise the unique IS-RT-PCR methodology to examine the localised gene expression of members of the MMP family of proteases implicated in breast cancer. Our findings here with IS-RT-PCR demonstrate the detection of in vitro and in vivo gene expression patterns of MMPs in HBC cell lines, and suggest a contribution from the stroma to MMP expression in breast carcinomas.
Methods
Cell culture
HBC cell lines MDA-MB-231, MCF-7, ZR-75-1 and MDA-MB-453, were originally obtained from the ATCC (Virginia, USA) and maintained by the Lombardi Cancer Center Shared Cell Culture Resource. They were grown as a monolayer culture in RPMI 1640 media (Invitrogen, USA) supplemented with 10% foetal calf serum (Biowhittaker, USA) in the presence of 5% CO2. Cells (approx. 1 × 105 cells/slide) were passaged onto two eight-chambered chamber slides (Lab-Tek, Nunc, USA) and allowed to attach for 8–12 hrs. Cells were then grown for a further 16–24 hrs in RPMI 1640 with 10% foetal calf serum in the presence or absence of Con A (25 μg/ml, Sigma, USA).
Xenograft tissue
Tumour xenografts of the MDA-MB-231, MDA-MB-435, MCF-7 and Hs578T HBC cell lines were xenografted into intact female mice as described previously 25. Animal studies were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals with approval of the Georgetown University Medical Center Animal Ethics Committee. The tissue was excised, and material from the xenografted tumours fixed in 10% neutral buffered formalin (Fisher Scientific, USA) and embedded in paraffin for routine histological analysis. The Lombardi Cancer Center Tissue Resource performed sectioning of the paraffin blocks, and tissue sections of 5 μm were mounted onto Pro-Bond +/- slides (Fisher Scientific, USA).
Pre-treatment of sections
Following sectioning, slides were placed under vacuum overnight with desiccant. Our previously described RNA ISH procedure 26 was modified as follows. All slides were de-waxed in xylene (4 × 10 min), washed in ethanol (100%, 3 × 3 mins), and rehydrated through a graded ethanol series (85%, 70% and 50% in 0.9% NaCl × 2 min each). Sections were then washed in saline (0.9% NaCl, 1X × 2 min) and PBS (1X × 2 min) prior to fixation with 4% paraformaldehyde (PFA; 250 ml H2O, 1 g NaOH, 10 g PFA, 1.7 g sodium acetate pH 6.5) for 20 minutes. Following fixation, slides were rinsed in 2X PBS (2 mins), followed by Proteinase K digestion (10 mg/ml; Roche Diagnostics, USA; 25 ml 1 M Tris pH 8.0, 250 μl 10 mg/ml Proteinase K, 25 ml 0.5 M EDTA pH 8.0, 200 ml H2O). Slides were then rinsed in 2% glycine (2 mins) and re-fixed in 4% PFA (10 mins) to ensure complete deactivation of Proteinase K. Slides were rinsed in triethanolamine (TEA, Sigma, USA; 0.1 M pH 8.0 × 3 min) in preparation for further digests.
RNase digestion
Following pre-treatment all slides were rinsed in RNase buffer (10 mM HEPES, 20 mM NaCl, 1 mM EDTA) without enzyme (2 × 2 mins). Negative control sections undergoing RNase digestion were then overlayed with RNase Digest mixture containing RNase cocktail (8 mg/ml RNase A and 160 U/ml RNase T1 final volume, Bresatec, Australia) in RNase buffer, placed in a humid chamber and incubated at 37°C for 8 hrs. Untreated slides were overlayed in RNase buffer and also incubated at 37°C for 8 hrs.
DNase pre-treatment
Prior to DNase digest, all slides were rinsed in DNase buffer (0.1 M C2H3O2Na, 5 mM MgSO4 pH 5.0, 2 × 2 min). DNase buffer (200 μl) containing 1.5 U/ml final volume of RNase-free DNase (Roche Diagnostics, USA) was added to each slide and these were incubated overnight at 37°C in a humid chamber. After incubation the slides were rinsed thoroughly in DNase buffer without enzyme, followed by washes in sterile water and dehydration through a graded ethanol series (50%, 70%, and 90% in 0.9% saline, 100% × 3 mins each).
Primers
Primers encoding MT1-MMP, MMP-1 and MMP-3 (Bresatec, Australia) were designed using the Amplimer program, product design checked by the Amplify v1.2 program, and BLASTed (NCBI, USA) for homology to the mRNA of these genes, with significant sequence homology demonstrated between the human and mouse MMPs examined. Primers encoding human β-actin were used as a positive control (Research Genetics, USA). Primer sequences are listed in Table 1. All primers were tested using traditional RT-PCR on mRNA extracted from MDA-MB-231 and MCF-7 HBC cells using the High Pure RNA Isolation Kit (Roche Diagnostics, USA). Reverse transcription was performed using 20 ng of isolated mRNA, by AMV reverse transcriptase (Promega Corporation, USA) at 50°C for 30 minutes, followed by PCR amplification using Taq DNA Polymerase (Perkin-Elmer Corporation, Applied Biosystems Division, USA). The linear phase of amplification was determined to be between cycles 20–25 using a standard two-step PCR cycle.
<p>Table 1</p>
Primer sequences encoding the human β-actin, MT1-MMP, MMP-1 and MMP-3 genes.
Gene
Primer Sequence
β-actin sense
5'- ACC CAC ACT GTG CCC ATC TA
β-actin antisense
5'- CGG AAC CGC TCA TTG CC
MT1-MMP sense
5'- AGT GGA TGG ACA CGG AGA AT
MT1-MMP antisense
5'- TCC ATC CAT CAC TTG GTT AT
MMP-1 sense
5'- AGC GTG TGA CAG TAA GCT AA
MMP-1 antisense
5'- GTT TTC CTC AGA AAG AGC AGC AT
MMP-3 sense
5'- GTC TCA AGA TGA TAT AAA TG
MMP-3 antisense
5'- AAT TGA TTT CCT TTA AAA ATG A
In situ-Reverse Transcription-Polymerase Chain Reaction (IS-RT-PCR)
Prior to the IS-RT-PCR procedure Gene-Frame gaskets (Advanced Biotechnologies, UK) were placed around the sections. IS-RT-PCR was performed on consecutive sections for each xenograft: section A) RNase/DNase negative control; B) β-actin positive control; C) MT1-MMP assay slide; D) MMP-1 assay slide; and E) MMP-3 assay slide. Reverse transcription was performed as previously reported 18 with the following modifications: dCTP concentration was 100 μM, biotin-dCTP 80 μM, MnCl2 10 mM and 7.5 U of rTth DNA Polymerase (Perkin-Elmer Corporation, Applied Biosystems Division, USA) was used in the 30 μl mix added to each slide. Linear amplification efficiency was determined using traditional RT-PCR as outlined above, to be between cycles 20 and 25. For the amplification step the initial denaturation was at 95°C for 4 minutes, the MgCl2 concentration was 15 mM, and 30 μl were added to each slide. Amplification was performed for 20 cycles.
Detection
Post IS-RT-PCR detection was performed as reported previously 21 with the following modifications: Incubation in the NBT/BCIP/ASB solution was for 15 minutes, slides were counterstained with eosin (Australian Biostain Pty Ltd, Australia) and mounted in Ultramount (Fronine, Australia). Consistency of observed gene expression in consecutive tumour sections demonstrated the reliability of the IS-RT-PCR technique and as expected, all cells were found to express the ubiquitous β-actin gene.
Signal quantitation
The level of signal intensity was assigned a value from zero to four (0+ to 4+). Slides showing no purple precipitation and only eosin counterstain were assigned a value of zero (0+), whilst cells stained with precipitate were assigned a value between one and four (1+ to 4+). The observed signal intensity was assessed based on the observed level of signal intensity and tissue morphology using light microscopy. Visual characteristics including the overall level of precipitation, direct comparison of the entire tissue area to compensate for background, and the number and intensity of cells exhibiting purple precipitation were also considered. In addition, a coded labelling system was used to ensure that slide examination was unbiased. In vitro signal intensity following exposure to Con A and cytokines was compared to the untreated or basal level allowing an estimate of up/down-regulation to be determined. The in vivo xenograft material was histologically assessed and scored from 0+ to 4+ (with 0 = no expression; 1+ = low expression; 2+ = moderate expression; and 3+ = strong expression). We compared the relative levels of gene expression of each MMP within the xenografts analysed to controls. Following independent examination of slides and sections by three individuals the signal level intensity was averaged. The photographs presented in Figures 1, 2 and 3 are of the same area within consecutive sections and are therefore a representation of the overall observed gene expression levels.
<p>Figure 1</p>
Representative photomicrographs of IS-RT-PCR analysis of Con A effects in MDA-MB-231 cells counterstained with Eosin
Representative photomicrographs of IS-RT-PCR analysis of Con A effects in MDA-MB-231 cells counterstained with Eosin. Negative controls include, RNase/DNase treated sections (A) and No RT (B). β-actin staining provided a positive Control without (C) or with (D) Con A treatment. MT1-MMP levels before (E) and after (F) Con A treatment are also shown. Photographs were on an Nikon Eclipse TE300 Microscope with a Nikon F-600 camera attachment and are × 40 (except for A, which is × 20).
![]()
<p>Figure 2</p>
Representative photomicrographs of IS-RT-PCR results of MCF-7 xenograft showing the RNase/DNase negative control (A), positive control β-actin (B), MT1-MMP gene expression (C), MMP-1 gene expression (D) and MMP-3 gene expression (E)
Representative photomicrographs of IS-RT-PCR results of MCF-7 xenograft showing the RNase/DNase negative control (A), positive control β-actin (B), MT1-MMP gene expression (C), MMP-1 gene expression (D) and MMP-3 gene expression (E). All photographs ×40 taken on an Olympus Microscope with a Nikon F-600 camera attachment.
![]()
<p>Figure 3</p>
Representative photomicrographs of IS-RT-PCR results of MDA-MB-435 xenograft showing the No RT control (A), RNase/DNase negative control (B), positive control β-actin (C), MT1-MMP gene expression (D), MMP-1 gene expression (E) and MMP-3 gene expression (F)
Representative photomicrographs of IS-RT-PCR results of MDA-MB-435 xenograft showing the No RT control (A), RNase/DNase negative control (B), positive control β-actin (C), MT1-MMP gene expression (D), MMP-1 gene expression (E) and MMP-3 gene expression (F). All photographs ×40 taken on an Olympus Microscope with a Nikon F-600 camera attachment.
![]()
Discussion
The central aim of investigations into molecular carcinogenesis is the identification of gene products involved in cancer progression during which tumour cells acquire the capacity for invasion with resulting metastasis 27. Metastasis is an active process involving the altered attachment of the tumour cell to the basement membrane, localised degradation of connective tissue, and migration through stromal tissue 2829. Such matrix degradation is mediated by members of several proteinase families (the serine, cysteine, aspartate proteinases and MMPs), with substantial tissue destruction being carried out by members of the MMP family 3. The expression of members of the MMP family and their inhibitors, the TIMPs, have been examined in both physiological and pathological conditions by many methods both in vitro and in vivo, including the use of total RNA from tissues and cell lines, Northern analysis, RT-PCR in situ and standard in situ hybridisation (ISH) 13212430313233. Xenografts of cell lines representing increased progression of breast cancer were examined in the present study using the IS-RT-PCR technique. The samples ranged from the poorly invasive, estrogen-receptor positive cells (MCF-7), through to examples of more aggressive, estrogen-receptor negative human breast carcinoma (MDA-MB-231, MDA-MB-435, Hs578T) 25. We analysed and compared the localised expression of MT1-MMP, MMP-1, MMP-3 and β-actin, in both the tumour parenchyma and the surrounding stroma. Both the MDA-MB-231 splenic and MDA-MB-435 xenografts demonstrate no detectable MMP gene expression in the stromal compartment. This highlights that although significant sequence homology exists between the human and mouse MMPs examined, minor differences may influence the detection of gene expression. Localised gene expression in both the epithelial tumour cells and the stroma compartments in the remaining xenografts however, demonstrate the detectable homology between human and mouse for the IS-RT-PCR primer sequences used.
MT1-MMP analysis of the HBC xenografts demonstrated mRNA in the tumour cells of all of the samples examined. In vitro, we have previously demonstrated MT1-MMP expression in MCF-7 cells by IS-RT-PCR 21, however Northern analysis of MCF-7 cells failed to find MT1-MMP in MCF-7 cells, whereas more invasive HBC cell lines showed MT1-MMP expression in concordance with their ability to be induced by Con A to activate MMP-2 123435. MT1-MMP has also been shown to correlate with increased invasion and to enhance migration of MCF-10A epithelial cells 61236, and transfection of MT1-MMP into MCF-7 cells stimulates higher migration and invasiveness 3738, and also stimulates VEGF production, angiogenic stimulation, and xenograft take rate in nude mice 3940. Presumably the IS-RT-PCR method is more sensitive than Northern analysis, as may be expected. Indeed, we detected lower levels of MT1-MMP in the MCF-7 xenografts than in the majority of other xenografts derived from more invasive HBC cells. More recently, using isolated RNA for quantitative analyses, we were unable to detect MT1-MMP in MCF-7 derived xenografts, but consistent with our observations here, increasing levels of MT1-MMP was detected in MDA-MB-231 derived xenografts as compared to the parental cells 13. The reasons for these differences are not clear and require further experimentation, but it is important to note that the cultured MCF-7 cells would have received estrogenic stimuli from both the phenol red and foetal calf serum in the medium 41.
MT1-MMP overexpression in the mammary gland results in abnormalities including hyperplasia, fibrosis, lymphocytic infiltration and adenocarcinoma 42, suggesting a pivotal role for MT1-MMP in carcinogenesis. However, ISH analyses of human breast carcinomas have primarily localised MT1-MMP mRNA expression to the stromal cells 1129434445, although one study found localised MT1-MMP in the tumour 46. An immunocytochemical analysis showed expression of MT1-MMP in both tumour and stromal cells 47. More recently, one study by Bisson et al, combining ISH and IHC data has co-localised MT1-MMP to the α-smooth muscle positive myofibroblast cells in close contact with tumour cells 48. Our studies certainly show the potential for the cell lines examined to make MT1-MMP, even the relatively well-conserved cell lines such as MCF-7. Although it is acknowledged that in vitro propagation of cell lines may alter some genetic pathways, it is also apparent that cell lines and tumour samples have distinctive gene expression patterns in common 49. Being ER-positive, and predominantly epithelial, MCF-7 cells far better represent what one may find clinically in breast cancers than the other cell lines examined here 50. The invasive HBC cell lines show a mesenchymal-like phenotype that may have resulted from epithelial-mesenchymal transition (EMT) in breast carcinoma. EMT is thought to have occurred in the MDA-MB-231, MDA-MB-435 and Hs578T cell lines 51. The demonstration of MT1-MMP expression observed in the stromal component is not totally unexpected. MT1-MMPs ability to stimulate invasion and metastasis in in vitro systems along with the estrogen receptor status of the cells is well documented as described above. Also, the basal levels observed between the MCF-7 and more invasive HBC cell lines may also be of impact in respect to the higher invasive potential of the cells.
MMP-1 mRNA was detected in the tumour cells of five of the six xenografts examined, the exception, surprisingly, being the MDA-MB-231 splenic metastasis. This is unexpected since MDA-MB-231 cell overexpress MMP-1 in culture 52, and these cells have a mutation/polymorphism in the promotor region which drives strong MMP-1 expression 18. This transcriptional repression has not been found to occur extensively in breast cancer 18. The reasons for suppression of the MMP-1 expression in splenic metastasis are not clear, but could include a strong repression, possibly due to altered signalling from the stroma. The splenic metastasis stroma was notably also lacking all three MMPs examined, but showed a strong β-actin signal in both compartments. One cannot rule out the possibility of selective aberrant loss of MMP mRNA rather than the β-actin, however, this is unprecedented. Further analysis of splenic metastases of the MDA-MB-231 cells, and primary human tumours, would be warranted. MMP-1 has been previously demonstrated to be equally expressed in vitro in both MCF-7 and MDA-MB-231 cells 53, although in a similar study using Northern analysis we found selective expression in MDA-MB-231 cells but not MCF-7 51. Increased production and secretion of MMP-1 has been correlated with increased metastatic potential 5254. More recently, Bachmeir et al have demonstrated a cell density-dependent regulation of gene expression of several MMPs in HBC cell lines 31. Increased cell density resulted in decreasing levels of both MMP-1 and MMP-3 expression levels, with a resultant decreased invasion concurrent with invasive potential 313233. Increased expression of MMP-1 has been reported on contact with Matrigel 55, and also in response to fibronectin fragments 56, attesting to the potential regulation of this MMP by the microenvironment. In vivo, MMP-1 expression has been demonstrated in the stroma of nine of thirty-four breast carcinomas 11 when examined by ISH, localised to stromal constituents during tumour formation 57, and to be upregulated in the stromal component of ductal carcinomas when examined by ISH and IHC 575859. Thus, again, our observation that MMP-1 expression was absent in the stroma of three out of six HBC xenografts appears to contrast the clinical situation. This may reflect the increased sensitivity of IS-RT-PCR and may also reflect the stage in tumour formation of the MCF-7 and MDA-MB-231 mesenteric xenografts that we examined.
MMP-3 mRNA was localised to the tumour component in all six xenograft samples examined. In vitro, MMP-3 mRNA has been demonstrated in the MDA-MB-231 HBC cell line 55, where it is stimulated by fibroblasts via the extracellular matrix metalloproteinase inducer (EMMPRIN) 60, and can enhance tumourigenicity and migration 54. In vivo, MMP-3 is centrally involved in mammary gland development 61, and has been demonstrated to promote tumour initiation and formation in the tetracycline-regulated mouse mammary model 6263. IHC and ISH in vivo analysis has demonstrated MMP-3 in both the tumour and stroma cellular compartments of both invasive and non-invasive tumours, with the level of stromal expression increasing with tumourigenicity 596465, and in the extracellular matrix adjacent to breast tumours 66. Our observation of MMP-3 gene expression in the stroma of four out of six of the xenografts examined, and in particular in the Hs578T xenograft, further support a role for this MMP in breast cancer.
Although HBC cell lines have been demonstrated to express certain MMPs, this is less evident in vivo where, as detailed above, the surrounding stromal cells contribute to much of the MMP activity. Important differences in MMP expression between the stromal cells recruited by each cell line and the tumour cells themselves, were detected in the current study. The MCF-7 xenograft stroma demonstrated low stromal MT1-MMP, MMP-1 and MMP-3 expression. MT1-MMP expression was observed at a lower level in the stroma associated with the MDA-MB-231 mesenteric metastasis. In the Hs578T xenograft stroma, MMP-1 was absent while MMP-3 was observed at a higher level in the stroma as compared to the tumour cells. No stromal expression of the three MMPs examined was observed in either the MDA-MB-231 spleen metastasis or the MDA-MB-435 xenografts. The level of stromal gene expression will depend on signals from the different HBC lines in the primary site, and also on the different responsivities in the different host sites for the splenic and mesenteric metastases. In regard to the primary site, overall recruitment of the stroma will differ among the HBC lines; however, β-actin mRNA detection was used to normalise the data. We did detect lower levels of β-actin in the Hs578T xenograft, and indeed higher levels of β-actin in the tumour cell component of the MCF-7 xenograft. However, this does not appear to have influenced the detection of the MMPs as indicated by the moderate levels of all MMPs examined in both compartments in the MCF-7 xenograft, and the moderate level of MMP-3 in the Hs578T tumour compartment.