The dynamic interactions between cells and basement membranes serve as essential regulators of tissue architecture and function in metazoans, and perturbation of these interactions contributes to the progression of a wide range of human diseases, including cancers. Here, we reveal the pathway and mechanism for the endocytic trafficking of a prominent basement membrane protein, laminin-111 (referred to here as laminin), and their disruption in disease. Live-cell imaging of epithelial cells revealed pronounced internalization of laminin into endocytic vesicles. Laminin internalization was receptor mediated and dynamin dependent, and laminin proceeded to the lysosome through the late endosome. Manipulation of laminin receptor expression revealed that the dominant regulator of laminin internalization is dystroglycan, a laminin receptor that is functionally perturbed in muscular dystrophies and in many cancers. Correspondingly, laminin internalization was found to be deficient in aggressive cancer cells displaying non-functional dystroglycan, and restoration of dystroglycan function strongly enhanced the endocytosis of laminin in both breast cancer and glioblastoma cells. These results establish previously unrecognized mechanisms for the modulation of cell–basement-membrane communication in normal cells and identify a profound disruption of endocytic laminin trafficking in aggressive cancer subtypes.
Basement membranes are crucial regulators of tissue architecture and function, and, like all extracellular matrices (ECMs), are subject to dynamic assembly and turnover during development, homeostasis and tissue repair (Streuli, 1999; Yurchenco, 2011). Correspondingly, perturbation of cell–basement-membrane interactions contributes to the progression of a wide range of human diseases, including skin blistering diseases, muscular dystrophies, neurodevelopmental defects and cancers (Akhavan et al., 2012; Barresi and Campbell, 2006; Domogatskaya et al., 2012; Yurchenco and Patton, 2009). These perturbations are most often attributed to altered basement membrane receptor expression or function, altered synthesis of basement membrane proteins or degradation of basement membrane proteins by proteases (Akhavan et al., 2012; Rowe and Weiss, 2008). However, the many changes in cell–basement-membrane communication that contribute to the progression of diseases are not fully understood, and other previously unrecognized regulatory factors might also be involved (Rowe and Weiss, 2008).
The internalization and endocytic trafficking of cell membrane and extracellular components are essential and integral functions that regulate the interactions between cells and their microenvironment (Polo and Di Fiore, 2006; Scita and Di Fiore, 2010). Endocytosis orchestrates cell–microenvironment interactions through multiple mechanisms, including the turnover of extracellular ligands and receptors, their recycling to the cell surface and the spatiotemporal control of signaling events within the cell (Polo and Di Fiore, 2006; Scita and Di Fiore, 2010; Sorkin and von Zastrow, 2009). The endocytosis of some ECM components, such as collagen I and fibronectin, has been investigated and demonstrated to regulate both matrix degradation and deposition (Madsen et al., 2011; Shi and Sottile, 2008; Sottile and Chandler, 2005). However, the mechanisms driving the internalization and trafficking of basement membrane proteins have not been explored.
Laminins are major signaling and structural molecules of basement membranes and modulate a host of cellular functions, including cell polarity, survival and hormone signaling (Domogatskaya et al., 2012; Hohenester and Yurchenco, 2013; Leonoudakis et al., 2010; Streuli et al., 1995; Yurchenco and Patton, 2009). The internalization of laminin by cells was reported over two decades ago, but the mechanisms involved remain uninvestigated (Coopman et al., 1991; Liotta et al., 1987). Consequently, little is known about the pathways and mechanisms controlling the endocytic trafficking of laminins or other basement membrane proteins.
In this study, we set out to investigate the mechanisms driving endocytic trafficking of laminin-111 and to explore potential changes in these mechanisms in cancer cells. Our results identify the laminin receptor dystroglycan as the dominant regulator of laminin endocytosis. Dystroglycan is known to be functionally compromised in many cancers (Akhavan et al., 2012), suggesting that they display laminin internalization defects. Indeed, restoration of dystroglycan function dramatically enhanced the internalization and trafficking of laminin in breast cancer and glioblastoma cells. Results presented here uncover novel mechanisms regulating normal cell–basement-membrane interactions and identify these mechanisms as compromised in a broad range of cancers.
Laminin is rapidly internalized in functionally normal cells
We have previously employed direct labeling of laminin-111 (hereafter called laminin) to assay the mechanisms of receptor-facilitated laminin assembly on the surface of living cells (Akhavan et al., 2012; Leonoudakis et al., 2010; Weir et al., 2006). As observed previously, fluorescently labeled laminin assembled on the surface of functionally normal mammary epithelial cells (MECs) in the same manner as unlabeled laminin (supplementary material Fig. S1) (Leonoudakis et al., 2010; Weir et al., 2006). Endogenous laminin production was barely detectable in these cells and did not contribute significantly to the assembled laminin in our assays using exogenous laminin (supplementary material Fig. S1). Time-lapse imaging revealed the binding of rhodamine-labeled laminin (Rhod–Ln) to the surface of mammary epithelial cells (MECs), which coalesced into small patches within 10 min (supplementary material Fig. S2A). Over a 50-min time period, laminin patches were found to form larger clusters, and fibrils (supplementary material Fig. S2B,C, arrows) resembling laminin assemblies were observed after an 18-h incubation (supplementary material Fig. S1). Unexpectedly, we also observed the apparent budding of laminin-laden vesicles internally from the cell surface. These vesicles rapidly moved throughout the cytoplasm within 10 min of exposure to Rhod–Ln (supplementary material Fig. S2D, black arrowheads; Movie 1).
Endocytic internalization of laminin was confirmed by multiple methods. Laminin was labeled with the pH-sensitive fluorescent label CypHer-5 (CypHer–Ln) to exclusively image internalized laminin in attached living cells. The fluorophore CypHer-5 is nonfluorescent at pH 7.4 and maximally fluorescent at pH 5.5, permitting fluorescence detection of laminin within intracellular acidic vesicles (pH 4.8–6.0). Following overnight incubation of MECs with CypHer–Ln, live-cell imaging detected bright fluorescent vesicles moving rapidly within the cytoplasm (Fig. 1A; supplementary material Movie 2). Intracellular laminin within the cytoplasm was also independently observed by removal of surface-bound laminin followed by confocal imaging. MECs that had been exposed to 10 µg/ml Rhod–Ln for 18 h were trypsinized and washed with PBS-EDTA to remove surface-bound laminin, allowed to reattach and stained with the membrane marker FITC–concanavalin-A (conA). Confocal imaging revealed undetectable surface laminin (no overlap with plasma-membrane-associated conA) and abundant laminin in internalized vesicles (Fig. 1B). This method of cell treatment permitted a quantitative flow-cytometry-based assay of laminin internalization. In this assay, cells incubated with Rhod–Ln were trypsinized to strip all surface-bound laminin (as in Fig. 1B), washed, and the remaining internal Rhod–Ln fluorescence was quantified by flow cytometry. Using this assay, we confirmed laminin internalization in diverse cell types, including in primary mammary epithelial cultures, mammary epithelial cell lines (E3D1, MEpG), murine fibroblasts (NIH 3T3 cells), primary astrocytes and human cancer cell lines (breast and glioma) (not shown and see below).
The dynamics of laminin internalization point to lysosomal degradation
To explore the dynamics of laminin internalization, we performed both pulsed timecourse experiments and steady-state timecourse experiments (measuring internalization in the continuous presence of labeled laminin). Steady-state laminin internalization assays revealed that internalized laminin was measurable within 1 h and did not plateau until after 30 h (supplementary material Fig. S2E), reflecting the continuous uptake of exogenous laminin. In pulsed timecourse experiments, the cells were exposed to exogenous Rhod–Ln at 4°C to prevent internalization, excess unbound Rhod–Ln was washed away and then the cells were returned to 37°C to allow synchronized internalization and trafficking to proceed. Following pulsed and synchronized laminin exposure, Rhod–Ln was again found to internalize within 1 h, but the amount of internalized laminin reached a maximum at 8 h, after which the levels of internal laminin declined (Fig. 2A). After 24 h, internal Rhod–Ln levels declined to ∼37% of their maximum value, suggesting that degradation or recycling to the cell exterior had occurred [mean fluorescence intensity (MFI) declined from its maximum of 315±3.5 to 119±11.5 at 24 h (mean±s.e.m.)]. To determine whether laminin was degraded using classical degradation pathways, we performed the pulsed timecourse experiments in the presence of leupeptin (a cell-permeable lysosome inhibitor) or MG-231 (a cell-permeable proteasome inhibitor) (Atsma et al., 1995). In the presence of leupeptin, we observed a large increase of 153% (105.5±3 versus 266±11.2; n = 4; P<0.001) in the amount of internal Rhod–Ln compared with that of vehicle controls (Fig. 2B); whereas the proteasome inhibitor MG-231 only increased the amount of internal Rhod–Ln by 59% relative to that of vehicle controls (105.5±3 versus 167.5±7; n = 4; P<0.001). These data indicate that the laminin internalized by epithelial cells is degraded primarily in lysosomes, and degradation is detectable at >8 h post internalization.
Laminin is trafficked through multivesicular bodies of the late endosome to the lysosome
The pathway of laminin internalization was tracked using live-cell imaging and transient expression of Rab–GFP fusion proteins to label distinct vesicles. Strong colocalization of internalized laminin was observed in conjunction with the Rab7 marker, which labels late endosomes (Fig. 3A, middle row; Fig. 3B) and within lysosomes labeled with the lysosomal-associated membrane protein 1 (LAMP1)–GFP fusion protein (Fig. 3A, lower panels; Fig. 3B). No significant colocalization was observed with Rab11-containing vesicles of the recycling endosome at time-points up to 18 h (Fig. 3A, upper row; Fig. 3B). The relatively slow movement of the laminin-laden vesicles matched the movement of Rab7 and LAMP1 markers and was clearly distinct from the rapid movement of Rab11-positive vesicles (supplementary material Movies 3–5). Laminin was clearly observed within multivesicular bodies of the late endosome, particularly when these bodies were enhanced by expression of a GTPase-deficient mutant of Rab5 (Rab5Q79L–GFP), causing the fusion of early and late endosomes (Duclos et al., 2003) (Fig. 3C). Furthermore, direct immunostaining of endogenous laminin also revealed colocalization of laminin with the Rab7 marker and endogenous LAMP1 (supplementary material Fig. S2F).
Laminin internalization is receptor mediated
Laminin internalization could be mediated by either receptor-dependent or receptor-independent mechanisms (e.g. pinocytosis). Steady-state laminin internalization was measured using flow cytometry in the presence of potential inhibitors and compared with the internalization of 500S FITC–dextran, a molecule that is of similar molecular size to laminin and is known to be endocytosed by receptor-independent mechanisms. The specific inhibitor of dynamin, dynasore (Macia et al., 2006), inhibited the internalization of laminin by >75% (Fig. 4A). Under hypertonic sucrose, a condition known to inhibit receptor-mediated endocytosis (Heuser and Anderson, 1989), laminin internalization was reduced by 69% (Fig. 4A). The addition of heparin, a molecule known to bind to the laminin LG4-5 domain (Harrison et al., 2007), decreased laminin internalization by 65% (Fig. 4A). By contrast, internalization of FITC–dextran was not significantly changed by any of these reagents (Fig. 4A). Additionally, a 24 h timecourse of Rhod–Ln and FITC–dextran internalization revealed clearly different internalization dynamics (Fig. 4B). Specifically, under steady-state, laminin internalization was nearly linear throughout the 24 h timecourse, whereas dextran internalization reached a plateau after 16 h. Taken together, these results indicate that laminin internalization is receptor-mediated and regulated by the GTPase dynamin.
Dystroglycan is the predominant mediator of laminin internalization
Genetic manipulation of laminin receptor expression was employed in order to identify the specific receptor(s) mediating laminin internalization. The observed inhibition of laminin internalization by heparin (Fig. 4A) suggested that dystroglycan could act as a mediator, because dystroglycan binding to the laminin LG4-5 domain is blocked by heparin (Harrison et al., 2007). To test the role of dystroglycan in laminin internalization, we used a MEC cell line containing an engineered knockout of dystroglycan (MEpG) (Weir et al., 2006). The MEpG cell line was infected with retroviral empty vector (creating control DG−/− cells) or retrovirus expressing dystroglycan (creating DG+ cells). Immunoblotting confirmed the expected presence or absence of dystroglycan expression in these cell types (supplementary material Fig. S3A). Confocal microscopy and flow cytometry both demonstrated a strong reduction in laminin internalization upon deletion of the dystroglycan gene, which was restored in DG+ cells (Fig. 5A,B). The compiled MFI data of cells expressing dystroglycan showed that laminin internalization by these cells was nearly fourfold higher than that of cells lacking dystroglycan expression [DG+, 18.04±4.6; DG−/−, 4.84±1.43 (mean±s.e.m.); n = 6, Fig. 5C]. Therefore, the majority of laminin internalization observed in functionally normal mammary epithelial cells appeared to depend on dystroglycan function.
The integrin family of ECM receptors is expressed in the DG−/− cell population (MepG-vec and MepL-vec cell lines; supplementary material Fig. S3A), but these receptors appear to be unable to mediate significant laminin internalization alone (Fig. 5B). To directly test the role of integrins in laminin internalization, we used a MEC cell line containing an engineered deletion of β1 integrin, E3D1cre19 (β1 integrin−/−) (cre19-vec; supplementary material Fig. S3A; see Materials and Methods). These β1 integrin−/− cells were infected with the empty vector retrovirus or with a retrovirus expressing wild-type β1 integrin. The re-expression of wild-type β1 integrin in β1 integrin−/− cells produced a modest but statistically insignificant change (a 20% increase) in laminin internalization (The MFI of β1 integrin−/− cells was 21.4±3; that of β1 integrin+ cells was 26.8±1.2; n = 6, P = 0.17) (Fig. 5D). A pulsed timecourse assay showed that the kinetics of laminin internalization were also not altered by β1 integrin expression, although, in this assay, the magnitude of internalization was moderately lower in β1 integrin−/− cells (supplementary material Fig. S3B). In addition, a β1-integrin-blocking antibody also inhibited laminin internalization in a pulsed internalization assay, whereas an α6-integrin-blocking antibody showed no effect (supplementary material Fig. S3C). Therefore, although the β1 integrins are not essential for the majority of laminin internalization, they can enhance it, possibly by acting as co-receptors with dystroglycan (Leonoudakis et al., 2010; Weir et al., 2006). Taken together, these data identify dystroglycan as the dominant regulator of laminin internalization in functionally normal epithelial cells.
Dystroglycan is the dominant regulator of both laminin assembly and internalization
Dystroglycan has been shown in prior studies to be the dominant regulator of cell-surface laminin assembly (Akhavan et al., 2012; Leonoudakis et al., 2010; Weir et al., 2006), and this is shown again in supplementary material Fig. S4A. It is intriguing that this same receptor should also dominantly regulate laminin internalization. To validate that dystroglycan simultaneously and dominantly regulates both laminin assembly and endocytosis in a cell-autonomous manner, we next assessed the dynamics of laminin assembly and internalization by live imaging in co-cultured DG+ and DG−/− cells. Both assembly and internalization of laminin was easily observed in DG+ cells, with the internalized laminin visibly associated with rapidly moving vesicles within the cytoplasm (Fig. 6A, arrows; supplementary material Movie 6). By contrast, both internalization and assembly were undetected in DG−/− cells during the entire 20 h timecourse of the experiment (Fig. 6A, arrowheads; supplementary material Movie 6).
The binding of dystroglycan to laminin is a high affinity protein–carbohydrate interaction that might persist following laminin internalization, and dystroglycan might accompany laminin through the protein degradation pathway. Alternatively, their intracellular trafficking patterns might diverge. To assess dystroglycan trafficking, a dystroglycan–RFP-encoding fusion construct was co-transfected with the GFP-labeled Rab7 endocytic marker, and these cells were treated with Alexa-Fluor-647-labeled laminin to permit simultaneous tracking of dystroglycan and laminin. Live-cell imaging of all three proteins showed clear and prominent colocalization of dystroglycan and laminin within the Rab7-positive late endosomes (Fig. 6B; supplementary material Movie 7). Therefore, although a direct association of dystroglycan and laminin has not been demonstrated within the endocytic compartments, dystroglycan appears to co-traffic with laminin through the protein degradation pathway.
Laminin assembly is not required for laminin endocytosis
Because dystroglycan mediates both assembly and internalization of laminin, we tested whether assembly was required for internalization, employing the E1′ or E4 fragments of laminin. The laminin E1′ and E4 fragments are known to perturb laminin assembly without blocking laminin receptor functions (Schittny and Yurchenco, 1990), and they block laminin-111 assembly on myotubes and in MECs (Colognato et al., 1999; Weir et al., 2006). The laminin E1′ and E4 fragments both blocked the assembly of Rhod–Ln on the surface of E3D1 MECs (supplementary material Fig. S4B); however, this blockade of laminin assembly had no effect on the levels of laminin internalization (supplementary material Fig. S4C). Therefore, laminin assembly is not a prerequisite for laminin internalization, and laminin assembly does not impede internalization, despite both being mediated by the same laminin receptor.
Matrix degradation by the action of proteases could modulate laminin internalization. Matrix metalloproteinase (MMP) activity has been shown to modulate the internalization of fibronectin (Shi and Sottile, 2011). Two different broad-spectrum MMP inhibitors, GM6001 [50 µM GM6001, 18.62±1; vehicle, 20.2±1.82; (mean±s.e.m.); n = 4] and BB2516 (marimostat; vehicle, 20.85; 5 µM BB2516, 20.46; n = 2) showed no significant effect on steady-state laminin endocytosis in E3D1 MEC cells (supplementary material Fig. S4D), indicating that MMP activity does not regulate laminin endocytosis.
Loss of dystroglycan function perturbs laminin internalization in cancer cells of diverse tissue origin
Loss of the laminin binding function of dystroglycan is a cause of some congenital muscular dystrophies (CMDs) and is a frequent defect in cancers, including those of the breast, prostate, colon and brain (Akhavan et al., 2012; Beltrán-Valero de Bernabé et al., 2009; Singh et al., 2004). This loss of function arises from altered glycosylation of dystroglycan and can be restored in many carcinoma and glioblastoma cells by expression of the enzyme LARGE, a glycosyltransferase that confers the laminin-binding properties to dystroglycan (Akhavan et al., 2012; Beltrán-Valero de Bernabé et al., 2009). On the basis of these facts and our current findings, we reasoned that laminin internalization would be severely disrupted in cancer cells lacking dystroglycan function and enhanced by its restoration. This hypothesis was tested in MDA-MB-231 human breast cancer cells and LN18 human glioma cells, both of which lack dystroglycan glycosylation and function. Expression of an empty retroviral vector in these cells created the control cells, which exhibited the hypoglycosylated dystroglycan (Fig. 7A, vector) and lack of laminin assembly at the cell surface (Fig. 7B, vector). The expression of LARGE restored normal glycosylation of dystroglycan as determined by western blot analysis with the IIH6 antibody against glycosylated dystroglycan (Fig. 7A, LARGE) and functional interaction of dystroglycan in the laminin assembly assay (Fig. 7B, LARGE). These cells were subsequently assayed for laminin internalization by flow cytometry. Control cells showed almost no measurable internalization of laminin over background, despite the expression of multiple laminin-binding integrin receptor subunits; MDA-MB-231 cells express the α1, α2, α3, α6, β1 and β4 integrin subunits, but not α7, α8 or α9 (Daemen et al., 2013). By contrast, LARGE-expressing cells showed robust laminin internalization (Fig. 7C,D). Compiled flow cytometry data demonstrate the increase in the MFI of internalized laminin in both LARGE-expressing MDA-MB-231 (68-fold) and LN18 (10-fold) cells. In addition, restoration of laminin internalization by LARGE expression in cancer cells also restored the trafficking of laminin to Rab7-containing vesicles of the late endosome (Fig. 7E).
Endocytosis orchestrates cellular communication with the extracellular environment by multiple mechanisms that include the regulated distribution and turnover of receptors and ligands (Polo and Di Fiore, 2006; Scita and Di Fiore, 2010; Sorkin and von Zastrow, 2009). Despite the recognized importance of signaling control by endocytosis, surprisingly little attention has been given to the endocytic trafficking of ECM molecules in the regulation of cell behavior and tissue organization. The endocytosis and trafficking of ECM receptors, particularly that of the integrins, have been studied extensively and are known to play important roles in cytokinesis and cell migration (Caswell et al., 2009). Among ECM molecules, fibronectin, collagen-I and vitronectin have been reported to undergo endocytosis, with internalization mediated, in part, by the integrins (Madsen et al., 2011; Memmo and McKeown-Longo, 1998; Shi et al., 2010; Shi and Sottile, 2008; Sottile and Chandler, 2005). However, the endocytosis and trafficking of laminins and other basement membrane proteins is not understood at the level of mediators, pathways or biological significance, despite these being crucial structural and signaling molecules that contact the majority of cells in metazoans (Yurchenco, 2011).
Here, we have elucidated the key mechanisms and pathways of laminin-111 endocytosis in functionally normal cells. The endocytosis of laminin was observed to proceed predominantly through late endosomes for degradation; however, our results do not exclude the possibility that some laminin is recycled to the cell surface and/or trafficked through other endocytic compartments. Unexpectedly, we find that laminin endocytosis is controlled predominantly by the laminin receptor dystroglycan. Genetic manipulations in functionally normal cells revealed that dystroglycan mediated the vast majority of detectable laminin internalization, with a lesser, but still significant, influence contributed by β1 integrins. The dominant role of dystroglycan in laminin internalization was validated in multiple independent cell models, including functionally normal epithelial cells, breast carcinoma cells and glioblastoma cells, where the restoration of dystroglycan function was shown to dramatically enhance (10–68-fold) the levels of laminin internalization.
Surprisingly, dystroglycan dominantly regulates laminin internalization while simultaneously controlling cell-surface laminin assembly. An important remaining question is whether dystroglycan mediates the endocytosis of soluble laminin only or whether it also mediates the endocytosis of assembled laminin. We show that laminin assembly is not a prerequisite or an inhibitor of laminin internalization, suggesting that this endocytic mechanism could be indifferent to the state of laminin assembly. However, although our results clearly show that soluble laminin can be internalized by dystroglycan, they do not resolve whether assembled laminin is also internalized by this mechanism, or whether basement membrane degradation might also be required. The endocytic mechanisms for soluble versus assembled fibronectin appear to be distinct (Shi and Sottile, 2008), and this remains a possibility for laminin internalization as well. Our assays rely primarily on measures of exogenous laminin internalization, and laminin internalization in vivo remains to be demonstrated; however, the internalization of endogenous laminin was observed in cultured cells.
Our discovery that dystroglycan is a potent mediator of laminin internalization is consistent with discoveries from the study of infectious diseases, where dystroglycan has been identified as the mediator of cell entry for multiple pathogens; dystroglycan mediates cell internalization and infection by Mycobacterium leprae (the leprosy vector) and old world arenaviruses, including the Lassa virus (LASV) and the lymphocytic choriomeningitis virus (LCMV) (Oldstone and Campbell, 2011; Rambukkana et al., 1998). This places dystroglycan amongst other important pathogen receptors, including the transferrin receptor, noted for efficient internalization of extracellular ligands (Choe et al., 2011). Interestingly, LCMV and LASV have also been shown to traffic to the late endosomes, multivesicular bodies and lysosomes, mirroring our results for laminin and dystroglycan trafficking (Jae et al., 2014; Pasqual et al., 2011). Our observations of laminin trafficking to the late endosome and lysosome are supported by previous electron microscopic imaging of gold-labeled laminin-111, which revealed laminin accumulation in non-coated pits at the cell surface and in multivesicular bodies (Coopman et al., 1991).
The potent role of dystroglycan in the control of laminin internalization implicates dystroglycan as a central coordinator of the trafficking and turnover of soluble basement membrane proteins. Dystroglycan has many other extracellular basement membrane binding partners – it binds to the majority of laminin isoforms (containing α1, α2, α4 and α5 subunits), as well as perlecan, agrin, pikachurin and neurexin (Barresi and Campbell, 2006; Sato et al., 2008). Based on the ability of dystroglycan to internalize a wide variety of binding partners, from viruses to bacteria and now laminin-111, we speculate that dystroglycan is likely to play a key role in the endocytic trafficking of many extracellular ligands. Additionally, laminin itself is capable of interactions with a wide variety of ECM proteins (Yurchenco, 2011); therefore, the turnover of many other laminin-binding proteins might also be linked to laminin internalization through dystroglycan.
Our findings might have important clinical implications, as alterations in the functions of dystroglycan functions are involved in the progression of many human diseases. In cancers, suppressed expression of the glycosyltransferase LARGE leads to loss of dystroglycan function in ∼20–30% of all solid tumors (Akhavan et al., 2012; Beltrán-Valero de Bernabé et al., 2009). Loss of dystroglycan function in cancer cells modulates tumor growth and invasion and is clearly associated with aggressive subtypes and poor outcomes in breast cancers and glioblastomas (Akhavan et al., 2012). Alterations in dystroglycan function are also linked to the majority of muscular dystrophies. A number of germ-line mutations lead to direct loss of functional dystroglycan glycosylation and produce a range of muscular dystrophies, from milder limb-girdle to severe congenital muscular dystrophies with cardiac hypertrophy and neurodevelopmental defects (Barresi and Campbell, 2006; Mercuri and Muntoni, 2012). Dystroglycan is a central component of the dystrophin-associated glycoprotein complex (DGC), and alterations in DGC composition and function are implicated in not only Duchenne muscular dystrophy but also in a broader array of muscular dystrophies (Durbeej and Campbell, 2002).
Our findings demonstrate a functional complexity for dystroglycan that has not been previously described and that prompts the re-thinking of the mechanisms of action of dystroglycan in normal cell and tissue regulation, as well as in human disease. The data presented here demonstrate that dystroglycan controls both cell-surface laminin assembly and laminin internalization. The precise signaling compartment of trafficked cell surface receptors can be challenging to define, and raises the question as to whether functions previously attributed to dystroglycan as a cell surface receptor indeed originate from its roles at the cell surface or from its roles in protein internalization and trafficking. Models of dystroglycan function have thus far focused entirely on the actions of dystroglycan at the cell surface. In epithelial cells, dystroglycan-mediated laminin assembly facilitates signals of polarity and STAT5 activation (Leonoudakis et al., 2010; Weir et al., 2006; Xu et al., 2009). In skeletal muscle cells, dystroglycan is necessary for sarcolemmal integrity (Barresi and Campbell, 2006) and neuromuscular junction development (Singhal and Martin, 2011). In tissues of the central nervous system, dystroglycan functions as a plasma membrane targeting, anchoring and clustering protein in both neurons and glia, and participates in neuronal migration in the developing cortex and cerebellum (Waite et al., 2012). In cancer cells, dystroglycan modulates growth and invasion (Akhavan et al., 2012; Beltrán-Valero de Bernabé et al., 2009). To this list of dystroglycan functions we can add that, through modulation of laminin endocytosis, dystroglycan controls endocytic laminin trafficking. Our results ultimately suggest that altered basement membrane protein endocytosis is a contributing factor in the etiology of CMDs and cancers.
Many alterations in endocytosis and intracellular trafficking have been implicated in cancer progression. These alterations include global changes in trafficking networks, and altered trafficking of receptor tyrosine kinases and integrins (Mellman and Yarden, 2013). Intuitively, it might be imagined that such trafficking changes will also include an increased turnover and destruction of ECM components by endocytosis, which will, in turn, favor invasion and metastasis. In this light, our results are unique and surprising in that they point to a reduction in the endocytic turnover and degradation of basement membrane ligands in aggressive cancer subtypes; the loss of dystroglycan function is most evident in more aggressive cancer subtypes (Akhavan et al., 2012), and we clearly demonstrate that loss of dystroglycan function profoundly disrupts laminin endocytosis. There are many potential mechanisms by which altered laminin trafficking could contribute to cancer progression; disrupted endocytosis of laminins might result in a disordered and functionally compromised basement membrane, it might perturb signaling that can originate from the endocytic compartment or it might alter the trafficking of additional associated proteins that are crucial to normal cell regulation. The demonstration of altered laminin trafficking adds greater breadth to the recognized changes in protein trafficking that arise in disease and might point to unexpected roles for laminin endocytosis in the maintenance of normal tissue architecture and cellular homeostasis.
MATERIALS AND METHODS
Primary mammary epithelial cells from control or ΔDGK14-Cre mid-pregnant mice were obtained as described previously (Weir et al., 2006). Dystroglycan-knockout (MEpG and MEpL cells) and wild-type E3D1 mammary epithelial cells were established as described previously (Weir et al., 2006) from floxed-dystroglycan mice. β1-integrin-knockout (E3D1 cre19) MECs were established from floxed-β1-integrin primary MECS, as above. MECs were grown in DMEM/F12, 2% FBS, 10 µg/ml insulin and 5 ng/ml EGF (BD Biosciences, San Jose, CA). Human dystroglycan or β1 integrin genes (Luo et al., 2005) were cloned into the retroviral expression vector pBMN-IRES-PURO, as described previously (Weir et al., 2006) and verified by sequencing. Retrovirus was generated using Phoenix-ECO packaging cells grown in DME/H21 (UCSF Cell Culture Facility, San Francisco, CA) containing 10% FBS and transfected using calcium phosphate (Sambrook et al., 1989). Clones were seeded into 100-mm dishes, infected with 2 ml of retroviral supernatant in 6 ml of complete medium with 8 µg/ml polybrene, and selected in complete medium with 5 µg/ml puromycin (Sigma-Aldrich, St Louis, MO). Primary mammary epithelial cells from wild-type mid-pregnant mice were obtained and cultured as described previously (Weir et al., 2006). Co-culture experiments utilized DG−/− MEpG cells infected with retrovirus to express either control GFP or a full-length dystroglycan–GFP fusion protein, and were performed as described previously (Oppizzi et al., 2008). To produce astrocyte cultures, postnatal day 3 mouse cortex was dissociated with papain and plated in DMEM containing 10% FBS. After 1 week in culture, flasks were shaken on a rotator to remove microglia and split into 10-cm cell culture dishes, grown to 90% confluency and split into 24-well dishes for experiments. These cultures produced >95% astrocytes as determined by staining with the GFAP astrocyte marker antibody.
Laminin-111 (1 mg) (Sigma-Aldrich, St Louis, MO) was dialyzed twice overnight against 500 ml of PBS with 10 µM CaCl2. The dialyzed laminin was then reacted with 10 µg NHS–Rhodamine, or a 50-fold molar excess of NHS–CypHer5 (GE) for 2 h on ice, followed by dialysis twice overnight against 500 ml of PBS with 10 µM CaCl2.
Live imaging of laminin assembly and internalization
MECs were plated on 35-mm cell culture dishes with cover glass bottoms pre-coated with poly-D-lysine. 10 µg/ml Rhod–Ln was added for 10 min and excess unbound laminin was washed out. The temperature was controlled at 37°C using a thermoelectric stage and objective warmer (Bioscience Tools, San Diego, CA). Images were acquired using Nikon Elements software running a Cascade II, QuantEM 512C camera (Photometrics, Tucson, AZ) at a rate of 1 frame per 30 s. The co-culture experiment was captured using a Zeiss Axiovert 200 microscope with a Yokogawa spinning disk (Stanford Photonics XR/Mega-10 ICCD and QED InVivo version 3.1.1 software, Palo Alto, CA).
Labeled laminin-111 was prepared as described above. Cells were grown overnight on Nunc Lab-Tek II glass chamber slides (ThermoScientific, Rochester, NY). Labeled laminin was added at 10 µg/ml, and the cells were incubated overnight and fixed with paraformaldehyde. For staining of exogenous unlabeled laminin, cells were blocked with PBS containing 3% BSA and 2% goat serum. Cells were then incubated with anti-laminin primary antibodies (Sigma) followed by anti-rabbit-Cy3 secondary antibodies. Light and fluorescence microscopy was performed on a TE2000 Nikon inverted microscope (Melville, NY) with a Photometrics Coolsnap HQ CCD camera (Tucson, AZ) controlled with Nikon Elements software.
GFP-labeled vesicle expression
cDNA expression constructs for GFP-tagged Rab proteins – Rab5a, Rab5Q79L, Rab7 and Rab11a –were generous gifts from Nigel Bunnett (UCSF, San Francisco, CA). Cell lines exhibiting laminin trafficking were transiently transfected with GFP–Rab expression constructs using Lipofectamine (Invitrogen), incubated for 2 days to allow for transgene expression and exposed to labeled laminin for 4–24 h prior to imaging. LAMP1–GFP expression was performed using the CellLight Lysosomes-GFP BacMam 2.0 expression system (Life Technologies, Grand Island, NY). Cells were imaged at 18 h after transduction using an Olympus IX71 inverted microscope equipped with environmental chamber.
Cells were plated on 12-well plates at 200,000 cells/well. The following day, the medium was changed to serum-free medium with or without 10 µg/ml Rhod–laminin or 40 µg/ml FITC–dextran (500S-Sigma-Aldrich, St Louis, MO). Unless otherwise indicated, cells were incubated for 18–24 h, washed once with PBS and trypsinized to remove all surface-bound laminin. Cells were washed in 5 ml of cold PBS containing 1 mM EDTA, pelleted and resuspended in 1 ml of PBS-1 mM EDTA. Using a BD FACScan flow cytometer (BD Biosciences, San Jose, CA), 10,000 cells/well were counted, background fluorescence from the cell counts with no added laminin was subtracted and the MFI values were recorded. The data presented here are compiled from duplicate wells from each experiment with a minimum of three separate experiments, unless indicated otherwise.
Fluorescence microscopy of internalized laminin
10 µg/ml Rhod–laminin or unlabeled laminin (Sigma-Aldrich, St Louis, MO) was added to cells in serum-free medium overnight. Cells were then prepared as for flow cytometry. Cells were replated on Lab-Tek II glass chamber slides and allowed to adhere for 2–3 h followed by fixation with 4% paraformaldehyde (PFA). Cells were stained with the membrane marker FITC–concanavalin-A for 1 h, washed three times with PBS and mounted with Fluoromount G (Electron Microscopy Sciences, Hatfield, PA). Confocal images were acquired with a Nikon C1 laser scanning confocal attached to a Nikon TE2000 inverted microscope (Melville, NY). Immunostaining for endogenous laminin and LAMP1 was performed on E3D1 cells cultured in the presence of leupeptin, fixed in 4% PFA and blocked in PBS containing 3% BSA, 2% horse serum, 0.3% Tween-20 and 0.2% Triton X-100. Immunostaining was achieved with a polyclonal anti-laminin antibody (Sigma-Aldrich) and monoclonal anti-LAMP1 antibody (clone 1D4B, EMD Millipore, Billerica, MA), with Alexa-Fluor-568-labeled anti-rabbit-IgG and Alexa-Fluor-647-labeled anti-rat-IgG secondary antibodies. Confocal imaging of endogenous laminin and LAMP1 staining was achieved using a Zeiss LSM710 confocal microscope and Imaris 7.7.1 software.
Biochemistry and SDS-PAGE
Cells were lysed in RIPA lysis buffer [50 mM Tris pH 8.0, 1% NP-40, 0.5% deoxycholate, 0.1% SDS, 1 mM EDTA, 1 mM EGTA, 1 mM PMSF, 50 mM NaF, 100 mM Na4P2O7, 10 mM sodium β-glycerophosphate, 1 mM Na3VO4, 1× protease inhibitor cocktail (EMD Chemicals, Philadelphia, PA)], and the protein concentration was quantified by using the DC protein assay (Bio-Rad). A total of 10 µg of extracted proteins were resolved on SDS-PAGE gels, transferred to PVDF membranes (Immobilon-P; EMD Millipore) and immunoblotted as described previously (Weir et al., 2006). The following primary antibodies were used for immunoblotting: 1∶5000 rabbit anti-actin (Sigma-Aldrich, St Louis, MO); 1∶2000 mouse anti-E-cadherin; 1∶1000 mouse anti-β1-integrin (BD Biosciences, San Jose, CA); 1∶2000 mouse anti-β-dystroglycan (MANDAG-2, Developmental Studies Hybridoma Bank, Iowa); and 1∶1000 IIH6 mouse anti-α-dystroglycan IgM (EMD Millipore). HRP-conjugated secondary antibodies specific for rabbit and mouse IgG were used at 1∶10,000; anti-IgM–HRP was used at 1∶1000 (Jackson Immunoresearch, West Grove, PA). Immunoblot signals were visualized by enhanced chemiluminescence (Super Signal West Femto–ThermoScientific, Rockford, IL) and digitally imaged with an Alpha Innotech imager (San Leandro, CA). Figures are inverted images processed with Adobe Photoshop.
Populations are described as the mean±s.e.m., and statistical significance was determined by using the paired Student's t-test (two populations).
The authors wish to thank Nigel Bunnet for gifts of the Rab-GFP constructs and Peter Yurchenco (Robert Woods Johnson Medical School, Piscataway, NJ) for the generous gift of laminin E1 and E4 fragments. We wish to thank Aurelie Snyder and Stephanie Kaech Petrie of the OHSU microscopy core for imaging assistance, and Mina Bissell (Lawrence Berkeley National Laboratory, Berkeley, CA) for use of the spinning disk confocal microscope.
D.L., G.H., A.A., J.E.F., M.S. and J.L.M. performed experiments. Experiments were conceived of and designed by J.L.M. and D.L. Data were analyzed and interpreted and the paper was written by D.L., A.A., J.W.G. and J.L.M.
This work was supported by grants from the National Institutes of Health [grant numbers CA109579 to J.L.M. and CA058207 and U54CA112970 to J.W.G.]; the Department of Defense Breast Cancer Research Program [grant number W81XWH-07-1-0416] to J.L.M.; the Stand Up to Cancer American Association for Cancer Research Dream Team Translational Cancer Research [grant number SU2C-AACR-DT0409]; and the OHSU Knight Cancer Institute [grant number 5P30CA069533-16] to J.W.G.; and by a Komen for the Cure Postdoctoral fellowship to A.A. This work was also supported by Oregon Clinical and Translational Research Institute (OCTRI) [grant number UL1TR000128] from the National Center for Advancing Translational Sciences at the National Institutes of Health. Deposited in PMC for release after 12 months.
The authors declare no competing interests.