Cation-dependent mannose-6-phosphate receptor expression and distribution are influenced by estradiol in MCF-7 breast cancer cells

Cancer cells secrete procathepsin D, and its secretion is enhanced by estradiol. Although alterations in the pro-enzyme intracellular transport have been reported, the mechanism by which it is secreted remains poorly understood. In this work, we have studied the influence of estradiol on the expression and distribution of the cation-dependent mannose-6-phosphate receptor (CD-MPR), which would be a key molecule to ensure the proper localization of the enzyme to lysosomes in breast cancer cells. Immunoblotting studies demonstrated that the expression of CD-MPR is higher in MCF-7 cells, as compared to other breast cancer and non-tumorigenic cells. This expression correlated with high levels of cathepsin D (CatD) in these cells. By immunofluorescence, this receptor mostly co-localized with a Golgi marker in all cell types, exhibiting an additional peripheral labelling in MCF-7 cells. In addition, CD-MPR showed great differences regarding to cation-independent mannose-6-phosphate receptor. On the other hand, the treatment with estradiol induced an increase in CD-MPR and CatD expression and a re-distribution of both proteins towards the cell periphery. These effects were blocked by the anti-estrogen tamoxifen. Moreover, a re-distribution of CD-MPR to plasma membrane-enriched fractions, analyzed by gradient centrifugation, was observed after estradiol treatment. We conclude that, in hormone-responsive breast cancer cells, CD-MPR and CatD are distributed together, and that their expression and distribution are influenced by estradiol. These findings strongly support the involvement of the CD-MPR in the pro-enzyme transport in MCF-7 cells, suggesting the participation of this receptor in the procathepsin D secretion previously reported in breast cancer cells.


Introduction
Cathepsin D (CatD) is a soluble aspartic protease that is overexpressed and secreted in high amounts by breast cancer cells [1,2]. In primary breast carcinomas, the expression of this hormone could be correlated with the exacerbated secretion of the pro-enzyme, a phenomenon that has been reported by other authors [1,23].

Antibodies and reagents
The goat anti-CatD antiserum was purchased from Santa Cruz Biotechnology (sc-6487 Dallas, TX, USA), and used 1:1000 in PBS for immunoblotting (IB) and 1:150 for immunofluorescence (IFI). The rabbit anti-CD-MPR antiserum was gently provided by Dr. Luzio (Cambridge University, UK), and used 1:250 for IB and 1:200 for IFI. The rabbit anti-CI-MPR antiserum was gently provided by Dr. Nancy Dahms (Medical College of Wisconsin, USA) and used 1: 500 for IB and 1:100 for IFI. The rabbit anti-LAMP1 antiserum (ab-24170) and mouse anti-βtubulin monoclonal antibody (ab-56676) were obtained from Abcam (USA). The mouse anti-golgin97 monoclonal antibody was obtained from Santa Cruz Biotechnology (sc-73619). The HRP-conjugated anti-goat IgG antiserum was obtained from H&L (401515), the HRP-conjugated anti-rabbit IgG fraction was obtained from Sigma (A9169) and the HRP-conjugated anti-mouse IgG (whole molecule) was purchased from Sigma (A9044). Chemiluminescent reagents were from Pearce (Rockford, IL, USA).

Cell cultures
Three breast cell lines were used in this study; the non-tumorigenic MCF-10A, and the tumorigenic, MCF-7 and MDA-MB-231 cell lines, obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA). The MCF-10A is a human-derived mammary epithelial cell line which does not express estrogen receptor alpha (ERα), and whose characteristics are those of the normal breast epithelial cells [24]. The MCF-7 is a hormone-sensitive breast ductal adenocarcinoma-derived cell line expressing ERα. The MDA-MB-231 is also a breast ductal adenocarcinoma-derived cell line, but it presents a phenotype that is more mesenchymal than epithelial, and molecularly classified as triple negative (ER-/PR-/HER2-). All cell lines were used with 10-15 passages. MCF-7 and MDA-MB-231 cell lines were cultured in DMEM Base (Sigma) supplemented with 10% charcolized fetal bovine serum (Internegocios), 2 mM L-glutamine (Sigma), 44 mM sodium bicarbonate, 1 mM sodium pyruvate, 5.6 mM glucose, 50 IU/50 μg/ml penicillin-streptomycin (Gibco) at 37˚C under a 5% CO2 atmosphere.

Hormone treatments
MCF-7 and MCF-10A cells (60% confluence) were cultured as described above in T25 flasks with the corresponding culture media in the absence or in the presence of either 20 nM estradiol (Sigma) or 20 nM estradiol plus 2 μM tamoxifen (Sigma), for 12, 24 or 48 h. After incubation, cells were harvested after treatment with 0.1% trypsin (Gibco) for 5 min and processed for immunoblotting. and boiled for 5 min. For CI-MPR detection, homogenate proteins were resuspended in the Laemmli's buffer without SDS and not boiled. Proteins were analyzed by electrophoresis in 6-10% polyacrylamide gels. The immunoblotting was carried out following the protocol of Romano et al. [26]. Briefly, after electrophoresis, proteins were electrotransferred onto nitrocellulose membranes (GE Healthcare, Amersham, Germany), for 4 h at 250 mA for detection of CI-MPR or for 1 h, at 250 mA for the other proteins under study. Membranes were blocked with 5% skimmed milk in buffer A (0.2% Tween 20 in PBS) for 1 h and incubated with the corresponding primary antibodies overnight at 4˚C. After three washings with buffer A, membranes were incubated with the corresponding HRP-conjugated secondary antibodies. Specific bands were revealed by chemiluminescence and the signal was detected with a LAS 4000 imaging system (Fujifilm Lifescience, USA). Band intensities were quantified by densitometry using the Image J software (Image Processing and Analysis in Java; National Institutes of Health, Bethesda, MD, USA).

Indirect immunofluorescence
Cells were grown on 1 cm diameter round coverslips seated at the bottom of culture wells under the conditions described above. Once 50% confluence was reached, cells were washed once with PBS and fixed with 3.7% paraformaldehyde for 20 min. Subsequently, cells were permeabilized with 0.1% saponin for 15 min, washed three times with PBS and blocked with 5% horse serum for 30 min. Afterwards, cells were incubated with primary antibodies, at the dilution indicated above, overnight at 4˚C, then washed three times with PBS, and incubated with the corresponding fluorochrome-conjugated secondary antibodies diluted in PBS-horse serum for 90 min. Cell nuclei were stained with Hoescht, and coverslips were mounted on slides with Mowiol mounting solution. Samples were analysed with an Olympus FV 1000 confocal microscope and images were acquired using the FV 10-ASW 1.7 software (Olympus, Japan).

Quantitative co-localization analysis
The co-localization analysis was carried out with the JACoP plugin of the Image J software (NIH [http://rsb.info.nih.gov/ij/plugins/track/jacop.html]). Pearson and Mander correlation coefficients (PCC and MCC, respectively) were calculated. For PCC, the dependency of pixels in dual-channel images (green/red channel for detection of CD-MPR/CatD; LAMP-1 /CatD and CD-MPR/Golgin) was assessed by plotting the pixel grey values of two images against each other. These values were displayed in a pixel-distribution diagram (scatter plot), and a linear equation describing the relationship between the intensities of the two images was determined by linear regression. A cross-correlation function (CCF) was obtained by plotting the corresponding PCC for each pixel shift (δx) of the green image in the x direction relative to the red image, or viceversa. The PCC value varied from 1 to -1, where values of 1 denote complete correlation, while values of -1 suggest a negative correlation. Among the MCC coefficients, the MCC-M1 and MCC-M2 were useful to describe the proportion of each protein that co-localized with the other, since this coefficient is independent of the fluorophore fluorescence intensity. The MCC values varied from 0 to 1, indicating no co-localization or complete colocalization, respectively.

Discontinuous sucrose gradients
Subcellular fractions from MCF-7 cells subjected or not to hormone treatment, were obtained in discontinuous sucrose gradients, according to other authors [27]. Briefly, MCF-7 cells were harvested and homogenized with a teflon Dounce tissue homogenizer in buffer B (10 mM Tris-HCl pH 7.4, containing 0.25 M sucrose, 1 mM EDTA, and 0.02% PMSF). Homogenates were centrifuged at 3,000 g for 10 min and the resulting supernatants were then centrifuged at 30,000 g for 20 min. Pellets were then resuspended in buffer B, and the protein concentration was determined according to Lowry [28]. Each membrane sample was loaded on top of a 20-50% (w/w) discontinuous sucrose gradient (prepared in buffer C (10 mM Tris-HCl (pH 7.4), containing 1 mM EDTA, 0.02% PMSF) and centrifuged at 100,000 g for 60 min at 4˚C. Fractions of 1 mL were collected from the bottom and weighed to estimate the fraction density (W/ V). Subsequently, 2 ml of buffer C were added to each fraction, and centrifuged at 30,000 g for 30 min at 4˚C to remove sucrose. Final pellets were processed for immunoblotting.

Statistics
Data were analysed by the Tukey-Kramer multiple comparisons test. The level of significance was set at p 0.05. At least three independent experiments were performed in each case.

Tumorigenic and non-tumorigenic breast cell lines express different levels of CatD, CD-MPR and CI-MPR
The expression levels of the proteins under study were evaluated by immunoblotting. As observed in Fig 1A, the mature form of CatD (33 kDa) is highly expressed in MCF-7 tumorigenic cells and, at a lesser extent in non-tumorigenic (MCF-10A) and in tumorigenic triple negative MDA-MB-231 cells. Moreover, the levels of the immature form of the enzyme (52 kDa) were also significantly higher in MCF-7 cells, indicating that the total expression of the enzyme is increased in these cells. A similar trend was observed for the CD-MPR, since the expression of this receptor was higher in the MCF-7 cells than in the MCF-10A or MDA-MB-231 cells (Fig 1B). In contrast, the CI-MPR expression levels were significantly higher in MCF-10A and MDA-MB-231 than in MCF-7 cells (Fig 1C).

The subcellular distributions of CatD and CD-MPR differ between tumorigenic and non-tumorigenic breast cell lines
By IFI, a perinuclear location and a granular cytoplasmic distribution of CatD were observed in the three breast cell lines. However, an additional peripheral punctuated distribution of the enzyme, neighbouring the plasma membrane, was also observed in MCF-7 and MDA-MB-231 cell lines, but not in MCF-10A cells (Fig 2A, arrows). From these observations we evaluated the degree of co-localization of the enzyme with the lysosomal associated membrane protein 1 (LAMP1) and compared it between MCF-7 and MCF-10A cells, since among these cells the major difference in CatD distribution was observed. Although CatD showed partial co-localization with LAMP1 in MCF-7 and MCF-10A cell lines (Fig 2B), this co-localization was significantly lower in MCF-7 cells (Fig 2C), indicating the occurrence of higher location of the enzyme in non-lysosomal/endosomal compartments, when compared to MCF-10A cells.
On the other hand, the CD-MPR localization was found to be mostly perinuclear in the three cell lines (Fig 3A). Such location would correspond to the Golgi apparatus, as evidenced by the simultaneous reactivity with golgin-97. However, in MCF-7 cells, an additional punctuate CD-MPR labelling was observed dispersed in the cytoplasm. Such reactivity did not colocalize with golgin-97 ( Fig 3B). It is worth mentioning that by IFI, an apparent higher CD-MPR signal was observed in MCF-7 cells. The latter finding is in line with the higher expression observed by immunoblotting (Fig 1B).

CD-MPR and CI-MPR display a differential distribution in MCF-7 cells
The CD-MPR and CI-MPR co-exist in most mammalian cells; however, the biological significance for such co-existence is still unknown. Although the CI-MPR has already been studied in MCF-7 cells, no comparative studies have been performed between both MPRs in this cell line. Therefore, we analysed the distribution of the CI-MPR and compared it with that of CD-MPR in the MCF-7 cell line. As observed in Fig 4B, CI-MPR showed a dispersed cytoplasmic distribution with a signal appearing in the cell periphery, suggesting its presence in the plasma membrane, in contrast with the perinuclear CD-MPR distribution (Figs 3A and 4A). Moreover, only 10% of CI-MPR co-localized with golgin-97, indicating a major location outside the Golgi stacks for this receptor.
On the other hand, unlike MCF-7 cells, in the non-tumorigenic MCF-10A cells, both CD-MPR and CI-MPR showed a more concentrated perinuclear localization with high colocalization with golgin-97 (Fig 5).

Estradiol regulates the expression and distribution of CatD and CD-MPR in MCF-7 cells
Taking into account that CatD and CD-MPR are highly expressed in the tumorigenic MCF-7 cells, and that an estrogen response element is present in the CatD gene, we evaluated the effect of 17-β-estradiol on CatD and CD-MPR expression and distribution on this estrogen-responsive cell line.
It was observed that 17-β-estradiol induces an increase in CatD and CD-MPR expression (Fig 6A and 6B) at 12, 24 and 48 h of incubation with the hormone, and this effect was blocked by the antiestrogen drug tamoxifen. The increment of both proteins was also observed by IFI (Fig 7B and 7D). In addition, 17-β-estradiol induced a redistribution of CatD and CD-MPR to the cell periphery (Fig 7A and 7C, respectively). The latter effect was also blocked by tamoxifen.
A similar experiment was carried out with the non-tumorigenic MCF-10A cells, which lack the ERα. No significant changes in CatD and CD-MPR expression were observed after treatment with the hormone (Fig 8). These findings support the idea that the changes observed in MCF-7 cells were mediated by ERα.

Estradiol induces the redistribution of CatD and CD-MPR towards common compartments
By IFI, we observed that 80% of CatD co-localizes with CD-MPR (MCC-M1: 0.81 ± 0.06) and that 95% of the CD-MPR co-localizes with CatD (MCC-M2: 0.95 ± 0.05) in control MCF-7 cells. After treatment with estradiol for 24 h, both proteins redistributed partially from a mostly perinuclear location to a more peripheral granular location, maintaining the high colocalization rates (Fig 9B). This result suggests that both proteins could share intracellular compartments, either as free or complexed forms, and that they relocate together under estradiol stimulation. This effect was blocked by tamoxifen ( Fig 9B). To further confirm that the re-distribution of CD-MPR to the plasma membrane was driven by estradiol, we performed a subcellular fractionation by a discontinuous sucrose gradient (Fig 10). We observed that estradiol induced the appearance of CD-MPR in low density fractions enriched in plasma membrane markers [29]. Again, the stimulation with estradiol induced an increase in the CD-MPR expression. Both effects were also blocked by tamoxifen.

Discussion
In cancer cells, CatD trafficking is altered, and this phenomenon leads to an increase in the pro-enzyme secretion [30,31,32]. Several mechanisms can be proposed to explain the  magnitude of such secretion; e.g., the excessive expression would cause the enzyme to be secreted to the medium, a detour from the route to lysosomes or a selective transport to the extracellular medium, among others. In most cell types, acid hydrolases, including CatD, are delivered to lysosomes by mannose-6-phosphate receptors (MPRs). The present study shows that MCF-7 cells have higher levels of CD-MPR than other breast-derived cell lines (MCF-10A and MDA-MB-231), while showing the lowest levels of CI-MPR. This phenomenon would indicate that both receptors are alternative for the trafficking of mannose-6-phosphate bearing enzymes [33], and that, in these cell lines, their expression is cross-regulated. Although this hypothesis is in agreement with that proposed by other authors, it does not fully explain the fact that both receptors recognize different enzyme sets [34]. Even though in this study we do  not present direct supporting evidence, the increased simultaneous expression of CD-MPR and CatD in MCF-7 cells would suggest that CD-MPR is a receptor for this enzyme. However, the role of sortilin in cathepsin D transport in breast cancer cells should not be ruled out, since this receptor in known to participate in the sorting of lysosomal proteins including cathepsin D in other cell types [12,13]. It has been documented that sortilin can mediate cathepsin D delivery to lysosomes when the MPR pathway is impaired [12]. Exploring the role of sortilin in intracellular transport in breast cancer cells could prove to be of substantial interest for future studies.
By IFI, we observed that CatD is mostly located perinuclearly, and also dispersed in the cytoplasm. However, it is noteworthy that the enzyme is also located at the periphery of the tumorigenic (MCF-7), but not in the non-tumorigenic cells. In tumorigenic cells, the presence of CatD in the cell periphery could be attributed to vesicle accumulation for eventual secretion. In fact, about 36% of the total intracellular CatD is present in LAMP1-negative compartments in MCF-7 cells.
Apart from the differences observed in the expression of the MPRs, we have also observed that both, CD-MPR and CI-MPR, are distributed differently in MCF-7 cells; while CD-MPR is mostly concentrated in the Golgi stacks and likely in endosomal compartments, the CI-MPR is more dispersed throughout the cytoplasm and also at the cell periphery. In addition, the high degree of co-localization coefficient of the CD-MPR with golgin-97 confirms that this receptor is mostly located in the Golgi stacks, as compared to the CI-MPR. The atypical golgin-97 dispersed distribution in MCF-7 cells was consistent with the presence of a fragmented Golgi apparatus already described in these cells [35]. The perinuclear CD-MPR distribution is common in most mammalian cell types [36], while the additional dispersed signal appears to be a feature of tumorigenic MCF-7 cells. The different distribution of the two MPRs in MCF-7 cells is consistent with the idea that both proteins are not redundant but complementary [26]. On the other hand, the CI-MPR has been proposed as a tumor suppressor molecule whose expression is downregulated by estradiol [21,37,38]. Unlike CI-MPR, in this work we have observed that the expression of CD-MPR is significantly increased by the action of estrogen in MCF-7 cells, and that this effect is blocked by the antiestrogenic drug tamoxifen. Moreover, this effect was accompanied by an increase of CatD levels. It is noteworthy that this is the first study suggesting an estradiol-driven regulation of CD-MPR. The fact that estradiol did not induce changes in MCF-10A cells indicates that the hormonal effect on MCF-7 cells is mediated by ERα. Moreover, this hypothesis is supported by the blocking effect of tamoxifen observed in MCF-7 cells. The results obtained in other models indicate that CD-MPR expression and distribution can be modulated by testosterone [39,40]. Whether the high levels of CD-MPR found upon estradiol stimulation are due to an increased synthesis or a diminished degradation of the receptor remains to be determined. Since the CD-MPR-encoding gene does not have hormone responsive elements, it may be possible that the levels of this receptor were regulated indirectly through the CatD response to the hormone. In fact, it has been reported that CatD can localize to the nucleus of cancer cells, where it could participate in transcription regulation by cleaving and/or interacting with nuclear proteins, thus modulating their activity [41,42]. As in other physiological models [26], the possibility of a cross-regulation of the expression of CD-MPR and CI-MPR cannot be ruled out.
Taking into account the results obtained herein, it is of interest to elucidate the role of CD-MPR in tumor cells, and whether there exists a direct link between CD-MPR and CatD. Although CD-MPR has scarcely been explored in cancer models, some authors have proposed a pro-tumorigenic role for this receptor [43]. It is hypothesized that the CD-MPR is a receptor for CatD, which participates in the enzyme secretion by cancer cells. In this regard, we observed high co-localization levels of the two proteins, both in the perinuclear region and adjacent to the plasma membrane. In addition, estradiol induced a redistribution of both proteins to the adjacency of the cell membrane, maintaining that high co-localization. The redistribution of the enzyme induced by the hormone is consistent with the estradiol-inducedsecretion reported in the literature [6,2]. Moreover, we confirmed the estradiol-induced CD-MPR redistribution when the receptor localized to low density fractions of the discontinuous sucrose gradient. These observations are supported by a previous work performed in another model in which the redistribution of CD-MPR towards the plasma membrane was found to correlate with high CatD secretion rates [39]. Moreover, Chao et al. have suggested a role for CD-MPR in selective enzyme secretion [20].
To sum up, we suggest that the CD-MPR would be selectively re-routed together with mannose-6-phosphate ligands (e.g. cathepsins) towards the plasma membrane by some mechanism that involves recognition signals in the CD-MPR cytoplasmic domain, which are different from those motifs that are known to participate in other intracellular transport routes [44]. Subsequently, the acidic pH of the tumor microenvironment [45] would favor the ligandreceptor complex dissociation and release of the enzyme to the extracellular medium.
In conclusion, our results provide new insights to clarify the mechanism by which human breast tumor cells distribute and secrete high amounts of proteases. This process would involve a receptor-mediated selective transport regulated by estradiol. Interfering with these processes would be a new strategy for future therapies against breast cancer.