|Home | About | Journals | Submit | Contact Us | Français|
Dystroglycan (DG) is an extracellular matrix receptor implicated in muscular dystrophies and cancers. DG belongs to the membrane-tethered mucin family and is composed of extracellular (α-DG) and transmembrane (β-DG) subunits stably coupled at the cell surface. These two subunits are generated by autoproteolysis of a monomeric precursor within a distinctive protein motif called sea urchin–enterokinase–agrin (SEA) domain, yet the purpose of this cleavage and heterodimer creation is uncertain. In this study, we identify a functional nuclear localization signal within β-DG and show that, in addition to associating with α-DG at the cell surface, the full-length and glycosylated β-DG autonomously traffics to the cytoplasm and nucleoplasm in a process that occurs independent of α-DG ligand binding. The trafficking pattern of β-DG mirrors that of MUC1-C, the transmembrane subunit of the related MUC1 oncoprotein, also a heterodimeric membrane-tethered mucin created by SEA autoproteolysis. We show that the transmembrane subunits of both MUC1 and DG transit the secretory pathway prior to nuclear targeting and that their monomeric precursors maintain the capacity for nuclear trafficking. A screen of breast carcinoma cell lines of distinct pathophysiological origins revealed considerable variability in the nuclear partitioning of β-DG, indicating that nuclear localization of β-DG is regulated, albeit independent of extracellular ligand binding. These findings point to novel intracellular functions for β-DG, with possible disease implications. They also reveal an evolutionarily conserved role for SEA autoproteolysis, serving to enable independent functions of mucin transmembrane subunits, enacted by a shared and poorly understood pathway of segregated subunit trafficking.
Dystroglycan (DG) is an integral membrane receptor linking the extracellular matrix (ECM) and cytoskeleton. Through widespread expression in a variety of cell types, including muscle, neural and epithelial cells, DG plays diverse and important roles in cell functions from basement membrane assembly to tissue morphogenesis and structural integrity (1–9). Importantly, dysfunction of DG has been implicated in several disease states from muscular dystrophies to neuronal disorders and cancer progression (1).
Structurally, DG is classified among a family of related membrane-tethered mucins that include the MUC1 oncoprotein. These proteins are each encoded by a single gene and posttranslationally cleaved into two noncovalently associated subunits by autoproteolysis within a distinctive protein motif called an sea urchin–enterokinase–agrin (SEA) domain (10,11). The resulting heterodimer is composed of a transmembrane subunit that tethers to the cell surface an extracellular subunit bearing extensive O-linked glycosylation. O-linked glycosylation of the extracellular DG subunit (α-DG) mediates binding to several ECM ligands, including laminins and perlecan. Extensive work has demonstrated the importance of α-DG glycosylation for DG functions and how altered α-DG glycosylation leads to receptor dysfunction with direct implications for human diseases (1). However, functions contained within the DG transmembrane subunit (β-DG), and the roles of this subunit in human disease, are poorly understood.
In this study, we demonstrate that β-DG not only functions at the cell surface in association with α-DG but also is selectively trafficked to the cytoplasm and nucleus as a full-length glycoprotein, independent of the α-DG subunit. We find significant differences in nuclear translocation of β-DG in a collection of breast cancer cell lines, which is independent of extracellular ligand-binding function and glycosylation of α-DG subunit. Significantly, this segregated localization is shared by the membrane-tethered SEA mucin oncoprotein, MUC1, the C-terminal subunit of which (MUC1-C) traffics to the cytoplasm and nucleus in the absence of the extracellular moiety (12,13). Our findings constitute the basis for independent functions of β-DG, mediated by its cytoplasmic and nuclear localization. They also reveal a collective trait of segregated cell surface and nuclear trafficking shared among membrane-tethered SEA mucins.
The function of DG as an ECM receptor relies on its cell surface localization, yet immunostaining analysis of mammary epithelial cells (MEps) revealed that DG was also evident in the cytoplasm and nucleoplasm To examine the subcellular distribution of DG by a method independent of immunostaining, we fused the enhanced green fluorescence protein (eGFP) to the C-terminus of β-DG. Upon expression in the MEpG-C7 cell line, we confirmed the normal processing and function of the DG–eGFP fusion protein (Figure S1A,B). Fluorescence microscopy revealed that, like endogenous DG, DG–eGFP was localized at the plasma membrane and also showed cytoplasmic and nuclear localization, with a fraction of cells displaying very prominent nuclear staining. Confocal microscopy confirmed that the immunostaining and nuclear GFP signals were observed throughout the nucleoplasm, excluding some nuclear organelles (see inset in Figure 1C).
Nuclear localization of membrane receptors can be achieved by proteolysis and release of their cytoplasmic portion, which then translocates to the nucleus, as is the case for Notch receptors (14). Alternatively, the entire transmembrane protein can be transported to the nucleus, as for the fibroblast growth factor receptor (FGFR1) (15). To determine what portion of DG is present in the nucleus, we isolated nuclear and non-nuclear [Nonidet P-40 (NP-40) soluble] fractions from different mammary epithelial cell lines and from the MEpG-C7 cells expressing an empty vector (DG−/−) or the wtDG. As expected, both α-DG and the full-length 43-kDa β-DG were prominent in the non-nuclear fractions but absent from DG−/− control (Figure 2A, left panels). Immunoblots of the nuclear fractions using a C-terminal β-DG antibody showed the presence of β-DG in every tested cell line, confirming the immunolocalization studies (Figure 2A, right panels). Importantly, only the full-length 43-kDa β-DG protein was detected in the nuclear fractions; the α-DG subunit was absent, and no fragments of β-DG were evident. The nuclear fractions were enriched for the nuclear marker protein lamin B1 (68 kDa), which was absent from non-nuclear fractions. The plasma membrane protein marker β1 integrin (130 kDa) and endoplasmic reticulum (ER) protein marker calreticulin (60 kDa) were absent from nuclear fractions, demonstrating that the nuclear extracts were clean of contaminants from these compartments (Figure 2A).
β-DG contains one N-linked glycosylation site (N661XT) at its extracellular portion just seven amino acids from its N-terminus (Figure 2B, left). To confirm that nuclear β-DG is the full-length glycoprotein, we exploited this feature and tested whether the nuclear β-DG harbors N-glycans. Treatment of both non-nuclear and nuclear β-DG with endoglycosidases results in a downward shift in molecular mass, demonstrating that nuclear β-DG bears N-linked sugars and thus is full length. This result was mimicked by a mutant of the N-linked glycosylation site of β-DG (T663A; Figure 2B). Importantly, trimming of N-linked sugars can be a signal for retrotranslocation of transmembrane proteins from the ER (16). Therefore, the observed nuclear trafficking of the T663A mutant excluded the possibility of retrotranslocation from the ER through mannose trimming by demonstrating that N-glycosylation is not a prerequisite for nuclear localization.
To analyze the mobility of β-DG between the nucleus and the cytoplasm, we performed fluorescence recovery after photobleaching (FRAP) experiments. Persistent photobleaching of DG–eGFP at a discrete position within the nucleus of live cells depleted fluorescence in the entire compartment, demonstrating rapid movement of the labeled molecule within the nucleus (Figure 2C, t = 0 seconds). Photobleached areas in the nucleus recovered within 240 seconds, revealing rapid nuclear import of the DG–eGFP fusion protein from the cytoplasm (Figure 2C, t = 240 seconds). Taken together, these results show that nuclear β-DG is full length and mobile in both the cytoplasm and the nucleus.
Nuclear shuttling of proteins can be mediated by intrinsic signal sequences. Indeed, sequence analysis of β-DG revealed a potential nuclear localization signal (NLS) in the juxtamembrane region of the cytoplasmic domain. This putative NLS contains a stretch of basic amino acids (RKKRKGK 776–782) and two lysines (KK 793–794) located 10 amino acids downstream of the first basic stretch (called NLS1 and NLS2, respectively) following the rule of a bipartite NLS (17). To test for NLS activity within β-DG, we expressed the cytoplasmic domain of β-DG fused to eGFP (cyto-DG–eGFP; Figure 3A). Strikingly, cells expressing this construct displayed predominantly nuclear accumulation of the GFP signal (Figure 3B), demonstrating that the cytoplasmic tail of β-DG can be very efficiently trafficked to the nucleus. To identify the amino acids critical for nuclear localization, we mutated putative NLS and non-NLS sequences in the cyto-DG–eGFP (Figure 3A). Whereas non-NLS mutants exhibited nuclear localization, the NLS mutants cyto-nls1–eGFP and cyto-nls2–eGFP showed predominant cytoplasmic localization (Figure 3B). Mutating NLS1 and NLS2 in the same construct produced no greater effect in abolishing nuclear localization (data not shown). We also verified NLS activity by nuclear fractionation of full-length DG. Relative to the wtDG, NLS mutants exhibited a markedly reduced β-DG signal in the nuclear extracts (Figure 3C). These results demonstrate that β-DG contains a functional bipartite NLS that efficiently traffics this subunit to the nucleus.
DG has recently been recognized as a member of the membrane-tethered SEA mucin family, which includes the MUC1 oncoprotein (10). Like DG, members of this family exist as cell surface heterodimers created by precursor autoproteolysis within a SEA domain (11), yet the purpose of this cleavage and heterodimer creation is uncertain. Strikingly, the transmembrane subunit of MUC1, MUC1-C, has been reported to localize to the cytoplasm and nucleus in the absence of the N-terminal subunit (12,13), closely mimicking our observations for β-DG. Similar to β-DG, the extracellular domain of MUC1-C is N-glycosylated (18). Moreover, we find that the nuclear MUC1-C is a glycoprotein, like nuclear β-DG, as demonstrated by sensitivity to peptide-N-glycosydase F (PNGase F) (Figure 4A). Interestingly, both nuclear and cell surface MUC1-C are resistant to Endo H, indicating that MUC1-C has passed through the Golgi compartment prior to nuclear translocation. Altogether, these data indicate that cytoplasmic and nuclear trafficking of the transmembrane subunits is a shared trait of membrane-tethered SEA mucins, perhaps dependent on SEA autoproteolysis.
To investigate potential roles for SEA autoproteolysis in DG and MUC1, we asked for each whether cleavage and heterodimer creation is a prerequisite for nuclear translocation of their C-terminal subunits or utilized to permit selective translocation of the transmembrane and cytoplasmic portion in the absence of the N-terminal sequences and mucin domain. To directly assess the significance of autoproteolysis, we assayed the nuclear localization of monomeric DG and MUC1, created by a serine-to-alanine mutation at their cleavage sites that abolishes SEA autoproteolysis. The DG mutant (S654A) was previously shown to reach the cell membrane and to be functional in MEpG-C7 cells (10). Significantly, fractionation of cells expressing the S654A mutant indicated the presence of the monomeric DG at the cell surface and in the nucleus. The 180 kDa monomer was clearly detected by an antibody for the α-DG subunit in both the non-nuclear and the nuclear fractions of cells expressing the S654A mutant (Figure 4B). In contrast, in the wtDG-expressing cells, the α-DG subunit was detected only in the non-nuclear fraction (Figure 4B), as observed previously. Moreover, a monomeric DG–eGFP fusion protein was also detected in the nucleus by fluorescence microscopy (data not shown). A similar result was obtained for a monomeric MUC1, created by the S1098A mutation. Nuclear fractionation of cells expressing the wtMUC1 and S1098A revealed that both the wt 22-kDa MUC1-C fragment and the 200-kDa monomeric MUC1 were present and prominent in the nucleus (Figure 4C). Altogether, these results signify that SEA cleavage of the DG and MUC1 precursors is required to allow autonomous localization and function of their transmembrane subunit independent of the extracellular moiety.
Receptor signaling is most often initiated by ligand binding. In this context, it was important to determine whether ligand binding to α-DG regulated the nuclear translocation of the β-DG subunit. Additionally, altered DG glycosylation or cellular signaling might lead to variable trafficking of β-DG in diseased cells. Alterations in the O-linked glycosylation of α-DG arise in some congenital muscular dystrophies and is prominent among breast carcinoma cell lines, leading to a loss of ligand binding (1,19). To address these issues, we tested a panel of 14 human breast epithelial and carcinoma cell lines to determine whether loss of ligand binding and/or hypoglycosylation of α-DG could be associated with changes in the nuclear trafficking of β-DG. Carcinoma cell lines were chosen that displayed diverse properties based on factors such as estrogen receptor status, invasiveness and luminal or basal phenotypes (19). Interestingly, the amount of nuclear β-DG varied widely among these different cancer cells (Figure 5A). Observing the relatively constant signal for lamin B1 in the same nuclear extracts, we conclude that the differences in β-DG detection were not caused by variable protein loading. The disparity in nuclear β-DG levels was evident also when normalized to the total DG expression levels in these carcinoma cells (Figure 5B). The varied levels did not segregate with the known cancer cell traits, such as luminal and basal phenotypes, but other potentially associated traits are under investigation.
Among the 14 cell lines tested, only BT474, T47D, SKBR-3 and ZR-75 cells displayed a functionally glycosylated α-DG, as measured by detection with the carbohydrate-sensitive IIH6 antibody (20) (Figure 5A). Interestingly, the cells expressing the form of DG detected by the IIH6 antibody all clustered within the luminal cell type, but IIH6 antibody binding did not correlate with those cells displaying high levels of nuclear β-DG. Therefore, the nuclear translocation of β-DG seemed to be independent of the glycosylation or ligand-binding status of α-DG.
To directly test the role of ligand binding, we compared the nuclear localization of β-DG in cells expressing wtDG and a DG mutant lacking the entire mucin domain of α-DG, named ΔM. We have previously shown that ΔM is cleaved into α and β subunits, transported to the cell surface but unable to bind laminin (8). As shown in Figure 5C, β-DG was translocated to the nucleus equally in wtDG and ΔM-expressing cells. Consistently, the addition of the exogenous DG ligand, laminin-111, to DG–eGFP-expressing cells did not appear to alter β-DG localization to the nucleus (data not shown). These results indicate that the nuclear translocation of β-DG is a regulated process that is highly variable among cancer cell models but is enacted independent of variable glycosylation or functions within the α-DG subunit.
Data presented in this study reveal that β-DG is classified among a small and distinctive set of transmembrane molecules that localize to the plasma membrane and also to the cytoplasm and nucleoplasm. These findings expand the possible functional properties of DG beyond its known role as a linker between the ECM and the cytoskeleton. β-DG resembles proteins such as MUC1-C and FGFR1, which are transported to the nucleus as full-length molecules, including the transmembrane domain (12,13,15). Importantly, a direct evolutionary relationship has recently been established between DG and membrane-tethered mucins, including MUC1, MUC3, MUC12 and MUC17, by virtue of their shared SEA domain structures and precursor cleavage (10). In light of this relationship and the discovery that both β-DG and MUC1-C are diverted from the membrane into the cytoplasm and nucleus, we speculate that the cleavage and segregated compartmentalization of these molecules share a common origin, mechanism and purpose. Proposed functions for SEA cleavage have included shedding of the extracellular subunits, the sensing of extracellular cues and the modulation of ligand-binding properties (10,11,21). Our data suggest that an important function for SEA cleavage is to permit the functional diversification of the transmembrane subunits independent of the extracellular moiety.
Recent data have indicated independent roles for the two subunits of DG (5,7,8,22) and we support this model in this study, showing nuclear localization of β-DG that appears not to be regulated by ligand binding through α-DG. Clues to β-DG cytoplasmic and nuclear functions may be found in the many important roles of MUC1-C in the cytoplasm and nucleus, modulating signaling and transcription through binding to factors like the IκB kinase complex (23) or to transcription-activating proteins like β-catenin, p53 and estrogen receptor (24). One publication has previously reported components of dystrophin-associated protein complex, including a high-molecular-mass isoform of β-DG, in the nuclear matrix of HeLa cells and suggested that these components may work as a nuclear scaffolding complex (25); however, our results are inconsistent with these observations. Distinct from the observations in HeLa cells, our localization experiments, performed in functionally normal epithelial cells lines, reveal prominent cytoplasmic and nuclear detection of β-DG, and immunostaining and FRAP experiments show evidence of highly mobile β-DG that is not restricted to the nuclear matrix or immobilized at the nuclear envelope. Moreover, we do not observe a distinct nuclear β-DG isoform with a higher molecular mass than that displayed at the cell surface, either in functionally normal MEps or in breast carcinoma cells.
The identification of a functional NLS in β-DG reveals a novel functional moiety within the DG cytoplasmic tail and further supports putative nuclear β-DG functions. Sequence comparison shows that the basic residues comprising the bipartite NLS, and many surrounding residues, are evident in all DG homologs from humans to Caenorhabditis elegans (Figure 6), suggesting that a role for DG in the nucleus is conserved through evolution. Interestingly, the same basic residues of β-DG were shown to mediate interaction with the cytoskeletal adaptor ezrin and the agrin-induced acetylcholine receptor clustering in muscle cells (22,26). These interactions may ultimately be critical in modulation of the NLS signal and, thereby, nuclear translocation of β-DG. The presence of a classical NLS in β-DG makes the transmembrane subunit a candidate for importin α/β-dependent nuclear import through the nuclear pore complex (27). MUC1-C, however, does not contain a classical NLS and was shown to be imported by direct binding to importin β and to the nuclear pore protein Nup62 (28). Thus, β-DG and MUC1-C may use different nuclear import pathways.
The precise route by which these proteins traffic from the cell membrane to the nucleus remains to be determined. We have demonstrated that both β-DG and MUC1-C reside in the ER membrane prior to nuclear translocation, as evidenced by N-linked glycosylation of the nuclear pool. Retrotranslocation from the ER induced by mannose trimming (16) is ruled out because the T663A mutant of DG, lacking N-glycosylation, still localized to the nucleus. Moreover, Endo H resistance indicates that MUC1-C transits the Golgi en route to the nucleus. Therefore, it is conceivable that nuclear β-DG and MUC1-C are derived from their cell surface pools. Nuclear localization of transmembrane proteins like the FGFR1 and the heparin-binding EGF-like growth factor can be regulated by endocytic trafficking (29,30). MUC1 itself is known to be internalized from the cell surface by clathrin-mediated endocytosis (31,32). For DG, it was shown that the arenavirus Lassa virus, which utilizes α-DG as a receptor for cellular entry, enters the host through a clathrin-mediated endocytic pathway, suggesting a role for DG in the endocytic process (33). We therefore speculate that endocytosis may be the pathway used to initiate β-DG internalization. However, how and when β-DG and MUC1-C are released from their membrane-bound state to enter the cytoplasm is uncertain and represents an important and poorly understood pathway of protein trafficking.
Importantly, defects in the function of DG or DG-associated molecules are linked to muscular dystrophies and cancers, which could be influenced by modulating the nuclear trafficking of β-DG. Dystrophin, the protein defective in Duchenne muscular dystrophy, interacts directly with β-DG and could modulate β-DG localization. Intriguingly, some forms of muscular dystrophies are generated by mutations in nuclear proteins, namely the lamin-binding protein emerin and lamins A and C (34). In addition, expression of a monomeric (uncleaved) DG in transgenic mice produced a genetically dominant form of muscular dystrophy of uncertain etiology (35). Given our observation that a monomeric DG can be trafficked to the nucleus, this disease phenotype could be explained by disruption of nuclear β-DG functions. In the realm of cancer, the MUC1 oncoprotein has known roles, mediated by its cytoplasmic and nuclear localization (24). The many shared properties of the MUC1 and DG revealed in this study, including cytoplasmic and nuclear trafficking, advocate for potential roles of a nuclear β-DG in cancer pathophysiology that are independent of the known DG hypoglycosylation in carcinoma cells and loss of ligand-binding properties. Deregulation of nuclear transport pathways has been detected in many different types of cancer cells and can arise from modifications of the cargo molecule, the transport receptors (importins/exportins) or the nuclear pore (36,37). Indeed, we find that the levels of nuclear β-DG are highly variable among cancer cell models, indicating either that nuclear transport of DG is differentially regulated or that it is modulated by the differential transport capacities of each cancer cell line. Therefore, in addition to revealing cytoplasmic and nuclear localization as a shared pathway for the segregated trafficking of SEA mucin subunits, the results presented in this study also highlight the potential importance of this pathway as a regulator of cell behavior, and as a possible therapeutic target, through controlling the segregated trafficking of multiple disease-related proteins.
eGFP coding sequence from the EGFP-N2 plasmid (Clontech) was inserted as an EagI–NotI cassette at the C-terminus of human DG in the retroviral expression vector, pBMN-IRES-PURO-DG (8). The DG sequence from amino acid 775 to 895 was used to obtain pBMN-IRES-PURO-cyto-DG–eGFP. This was used to generate nls1: KKRK777NPGE, nls2: KK793TA, K823A, PP828AA and Y831A by site-directed mutagenesis (QuikChange XL Site-Directed Mutagenesis Kit; Stratagene). Nls1, nls2 and nls1 + 2 were also generated on the full-length DG using the same approach. The human MUC1 complementary DNA was inserted as a BamHI cassette into pBMN-IRES-PURO. All constructs were verified by sequencing. Retroviral infections were performed as previously described (8).
Mammary epithelial cell lines MEpG-C7, MEpG and MEpL were previously described (8). MEpE cells were created by methods described in Weir et al. (8). Phoenix-ECO packaging cells were a kind gift from the Nolan Lab, Stanford University. The cancer cell lines MCF10A, MDA-MB-231, BT549, MDA-MB-436, MDA-MB-157, MDA-MB-435, MCF7, BT474, T47D, SKBR3, CAMA1, ZR75-1 and MDA-MB-134 were obtained from American Type Culture Collection and cultured as described in Neve et al. (19). 184A1 cells were a gift from Joe Gray's laboratory (LBNL, Berkeley, CA, USA).
MEpG-C7 cells grown on Lab-Tek II CC2 glass chamber slide (Nalge Nunc) were washed twice in PBS. Samples were fixed in methanol/acetone 1:1 for 5 min at −20°C and air dried. After blocking in PBS, 10% goat serum (Sigma) and 0.1% Tween-20 for 1 h at room temperature, samples were incubated in blocking solution for 1 h at room temperature with primary antibody, followed by 1 h at room temperature with fluorescent secondary antibody. Nuclei were counterstained with 10 μg/mL propidium iodide (Sigma), and washes were carried out for 3 × 10 min in PBS. Samples were mounted in Vectashield mounting media (Vector Laboratories) with glass coverslips.
Confocal images were obtained as previously described (8). Images were cropped and adjusted for contrast using Adobe Photoshop 7. Photobleaching of nuclear DG–eGFP was achieved by persistent laser scanning within the nucleus for 20 seconds, using a 7-micron field of view and 50-microsecond pixel dwell.
Cells were chilled on ice, rinsed in cold PBS and harvested in cold buffer I [0.32 m sucrose, 10 mm Tris–HCl (pH 8.0), 3 mm CaCl2, 2 mm magnesium acetate, 0.1 mm ethylenediaminetetraacetic acid (EDTA), 0.5% NP-40, 1 mm DTT, 0.5 mm phenylmethylsulphonyl fluoride (PMSF) and complete protease inhibitor mixture). Cells were dounce homogenized and spun at 600 × gfor 10 min at 4°C. The supernatant was spun again at 9300 × gfor 10 min at 4°C, and this supernatant was saved as non-nuclear fraction. The first pellet was resuspended in buffer I mixed with buffer II (2 m sucrose, 10 mm Tris–HCl pH 8.0, 5 mm magnesium acetate, 0.1 mm EDTA, 1 mm DTT, 0.5 mm PMSF and complete protease inhibitor mixture) to adjust sucrose concentration to 1 m and centrifuge through a 1.8 m sucrose cushion at 30 000 × gfor 50 min. The supernatant was removed, and the pellet was resuspended in lysis buffer (3% Triton-X-100, 3% sodium dodecyl sulfate and complete protease inhibitor mixture). Nuclei and DNA were broken by centrifugation over a QIA Shredder column at 15 700 × gfor 10 min at 4°C.
Protein extracts concentration was measured using the Lowry protein assay (38). SDS–PAGE was performed under reducing conditions using equal amounts of protein and 4–12% polyacrylamide Bis–Tris gradient gels (Invitrogen). Proteins were electrophoretically transferred to Immobilon-P membranes (Millipore). Blots for α-DG were blocked in 5% non-fat dry milk in TBS-T low salt (50 mm Tris–HCl, pH 7.4; 100 mm NaCl and 0.1% Tween-20) for 1 h at room temperature, followed by incubation in blocking buffer overnight at 4°C with primary antibody and then for 1 h at room temperature with horseradish peroxidase (HRP)-conjugated secondary antibody. Blots were washed in TBS-T low salt after antibody incubations, and bands were visualized with the SuperSignal West Femto Max Sensitivity Substrate (Pierce). The remaining blots were developed using 5% non-fat dry milk in TBS-T (25 mm Tris–HCl, pH 7.4; 150 mm NaCl and 0.1% Tween-20).
Antibodies specific for C-terminal β-DG (NCL-b-DG; Novocastra), N-terminal β-DG (BD Biosciences), MUC1-C (CT2; Thermo Scientific), β1 integrin (BD Transduction Laboratories), lamin B1 (Santa Cruz), calreticulin (Stressgen Bioreagents), mouse immunoglobulin M (IgM) monoclonal antibody (mAb) IIH6 specific for α-DG (Upstate, Inc.) plus HRP-conjugated antibodies specific for mouse immunoglobulin G (IgG) (Jackson Laboratories), mouse IgM (Sigma) and rabbit (Santa Cruz) were used for western blots. C-terminal β-DG antibody (Mandag2) and Alexa Fluor-488-conjugated antibody specific for mouse IgG (Invitrogen) were used for immunocytochemistry. The Mandag2 antibody, developed by G.E. Morris (39), was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development (NICHD) and maintained by Department of Biological Sciences, The University of Iowa, Iowa City, IA, USA.
Figure S1: Normal processing and function of DG–eGFP fusion protein. A) Immunoblots of proteins extracted from cells expressing either the wtDG or the DG–eGFP fusion show that the fusion protein produced an α-DG subunit of expected molecular mass, reflecting normal levels of glycosylation, and a β-DG subunit of higher mass, confirming eGFP fusion. B) Assays of laminin binding and assembly at the cell surface show the fusion protein to function like the wild-type protein.
(Figure 1A). This staining was specific as it could be blocked by preincubation with the competing peptide (Figure 1A, +peptide). To further confirm the specificity of immunostaining, we tested the MEpG-C7 mammary epithelial cell line, which we previously engineered to lack DG expression (8). As predicted, the DG−/− control cells (vector infected) did not show nuclear or cell membrane immunostaining for DG (Figure 1C, vec). Re-expression of the wild-type DG (wtDG) restored both staining patterns, with subsets of cells displaying bright nuclear staining (Figure 1B, DG).
We thank Thomas Vaccari, Pierre-Yves Desprez and Dmitri Leonoudakis for helpful comments, Sarah Cohen and Terry Chang for technical assistance and Joyce Schroeder for providing the MUC1 complementary DNA. This study was supported by National Institutes of Health grant R01 CA109579.