The transmembrane mucins in the enterocyte are type 1 transmembrane proteins with long and rigid mucin domains, rich in proline, threonine and serine residues that carry numerous O-glycans. Three of these mucins, MUC3, MUC12 and MUC17 are unique in harboring C-terminal class I PDZ motifs, making them suitable ligands for PDZ proteins. A screening of 123 different human PDZ domains for binding to MUC3 identified a strong interaction with the PDZ protein GOPC (Golgi-associated PDZ and coiled-coil motif-containing protein). This interaction was mediated by the C-terminal PDZ motif of MUC3, binding to the single GOPC PDZ domain. GOPC is also a binding partner for cystic fibrosis transmembrane conductance regulator (CFTR) that directs CFTR for degradation. Overexpression of GOPC downregulated the total levels of MUC3, an effect that was reversed by introducing CFTR. The results suggest that CFTR and MUC3 compete for binding to GOPC, which in turn can regulate levels of these two proteins. For the first time a direct coupling between mucins and the CFTR channel is demonstrated, a finding that will shed further light on the still poorly understood relationship between cystic fibrosis and the mucus phenotype of this disease.
The intestinal tract is a sophisticated organ responsible for the digestion and absorption of nutrients by an orchestrated action of epithelial cell functions. The intestinal tube is lined by absorptive columnar cells, termed enterocytes, which are polarized in distinct apical and basolateral membrane domains. The apical domain of the enterocyte has numerous transporters and channels involved in uptake of salts, water, sugars, peptides and amino acids. The absorption and secretion of ions are mediated by both apical and basolateral channels belonging to several families of transporters as for example the cystic fibrosis transmembrane conductance regulator (CFTR) family, a family of Na+/H+ hydrogen exchangers (NHEs), a family of chloride channels (CLCs; also known as H+/Cl− exchange transporters), epithelial sodium channels (ENaCs; also known as amiloride-sensitive sodium channels) and the down-regulated in Adenoma (DRA; also known as chloride anion exchanger) channels. These channels are important for intestinal epithelial surface homeostasis as well as proper hydration of the mucus in the lumen of the intestinal tract (Allen and Flemstrom, 2005; Field, 2003).
Mucins are the main structural components of the mucus covering the epithelium of the intestinal tract. They are highly glycosylated proteins characterized by proline–threonine–serine (PTS) domains, which comprise serine-, threonine- and proline-rich repeated sequences that act as anchoring points for O-linked glycans to generate the extended mucin domains (Gendler and Spicer, 1995). Mucins are divided into two categories; secreted and transmembrane mucins. Secreted mucins constitute the floating mucus covering epithelial surfaces, whereas the transmembrane mucins are anchored to the plasma membrane of cells through their transmembrane domains. These are probably major constituents of the dense glycocalyx on the apical membrane of enterocytes.
The genes for three membrane-associated mucins, MUC3, MUC12 and MUC17, are located on chromosome 7q22. The total size of MUC3 is still not known, but MUC12 and MUC17 consist of 5478 and 4493 amino acids, respectively. MUC3A and MUC3B are expressed in the intestine, colon and the gall bladder whereas MUC12 is expressed in the colon (Williams et al., 1999). MUC17 is found throughout the entire intestinal tract, reaching maximum levels in the duodenum (Gum et al., 2002). MUC3, MUC12 and MUC17 share common features: an N-terminal large mucin domain, a SEA (sea urchin sperm protein, enterokinase and agrin) domain, a transmembrane domain and a cytoplasmic tail (CT). The SEA domain undergoes an autocatalytic cleavage in an intramolecular strain-dependent process taking place in the endoplasmic reticulum (ER) (Ligtenberg et al., 1992; Macao et al., 2006; Parry et al., 2001). The highly conserved site of the autocatalytic cleavage is N-terminal to the serine residue in a G/SVVV motif found in the SEA domain. After cleavage, the two parts are held together with strong forces of four anti-parallel β-pleated sheets (Ligtenberg et al., 1992; Macao et al., 2006; Parry et al., 2001). The force is strong enough to anchor the N-terminal mucin domain with a mass of at least 2 MDa to the membrane.
The actual function of transmembrane mucins is still under debate. The molecular structure of the SEA domain suggests that other proteins are attached on the extracellular side of the membrane. Otherwise most known cellular signaling is mediated by their cytoplasmic tail. Most studies, related to cancer, have focused on another transmembrane mucin called MUC1 that can interact with signal transduction proteins including β-catenin, p53 and intercellular adhesion molecule 1 (ICAM1) (Kam et al., 1998; Rahn et al., 2004; Regimbald et al., 1996). It has also be demonstrated that the protein flagellin in Pseudomonas aeruginosa activates MUC1 signaling through the mitogen-activated protein kinase pathway (Lillehoj et al., 2004). Another proposed function of transmembrane mucins is their ability to act as osmosensors in yeast. In fact, two yeast mucin-like transmembrane proteins, krueppel-related zinc finger protein 1 (Hkr1) and Msb2, fulfill the criteria for an osmosensor. Besides residing at the surface of the yeast cell, they are osmosensitive because of their highly glycosylated PTS domains and transduce signals via their CT. Mutant cells with disrupted HKR1 and MSB2 genes fail to activate the high-osmolarity glycerol (HOG) signaling pathway (Tatebayashi et al., 2007). These observations suggest that transmembrane mucins can be involved in the regulation of ion transport processes.
The CFTR gene, coding a protein belonging to the ATP-binding cassette (ABC) sub-family C of transporters, was first identified in 1989 (Riordan et al., 1989; Rommens et al., 1989). Mutations in the CFTR gene cause cystic fibrosis (CF), an autosomal recessive genetic disorder, characterized by decreased chloride and bicarbonate ion conductance across the apical surface of epithelial cells in organs such as the lung, pancreas, small intestine and sweat glands. CFTRΔF508 is the most common mutation resulting in incorrect folding of the CFTR protein, followed by retention in the ER (Cheng et al., 1990). The immature and retained protein is then trapped by chaperones and degraded by cytosolic proteasomes (Jensen et al., 1995).
PDZ (post synaptic density protein, Drosophila disc large tumor suppressor and zonula occludens protein 1) domains are the most abundant protein interaction modules in the metazoan (Schultz et al., 1998). A total of 1163 PDZ domains in 484 different proteins have been suggested in the human genome (Schultz et al., 2000). PDZ domains consist of 80- to 100-amino-acids long single or repeated sequences sharing sequence homology across multiple species. The key function of PDZ domains is the ability to bind ligands containing C-terminal PDZ binding motifs. The first interaction between a PDZ protein and its ligand was shown for the binding of two PDZ domains of postsynaptic density protein (PSD) −95 to a C-terminal peptide motif (–E[T/S][D/E]V) of shaker-type K+ channels (Kim et al., 1995) and the NMDA (N-methyl-D-aspartate) receptor 2 (NR2) subunit of the NMDA receptor (Kornau et al., 1995). The current understanding of PDZ domains show that they function as modular units and their C-terminal binding ability can be retained when transferred to other heterologous proteins.
Golgi-associated PDZ and coiled-coil motif-containing protein (GOPC), also designated CAL or PIST, is a 50 kDa Golgi-resident PDZ protein expressed ubiquitously in human tissues. It harbors two coiled coil motifs with a leucine zipper within the second motif that has been proved important for binding to the Rho-family GTPase TC10, known to control vesicular trafficking (Neudauer et al., 2001; Symons and Rusk, 2003). GOPC also interacts with the Q-SNARE protein syntaxin 6, which is important in protein trafficking between the trans-Golgi network and the endosomal compartments (Chao et al., 1999; Charest et al., 2001; Cheng et al., 2010; Kuliawat et al., 2004; Mallard et al., 2002; Perera et al., 2003). GOPC contains a single PDZ domain, which can bind ligands with class I type C-termini (–x[S/T]xΦ; x representing any amino acid and Φ representing a hydrophobic amino acid) where amino acids at −0 and −2 positions are vital for PDZ interactions (Hung and Sheng, 2002; Jelen et al., 2003; van Ham and Hendriks, 2003). CFTR also contains a PDZ class I motif (–DTRL) enabling the ion channel to bind to several PDZ proteins, including Na+/H+ exchange regulatory cofactor (NHERF) proteins NHERF1, NHERF2, PDZ domain-containing protein 1 (PDZK1) and GOPC (Moyer et al., 2000). The PDZ motif regulates CFTR plasma membrane half-time (Swiatecka-Urban et al., 2002). In Cos-7 cells, GOPC reduces chloride currents of CFTR by reducing the surface expression of the anion channel (Cheng et al., 2002). It has also been demonstrated that overexpression of GOPC reduces the total expression of CFTR owing to lysosomal degradation of CFTR by GOPC (Cheng et al., 2004).
We have screened PDZ domains for their ability to bind CT of the human membrane-associated mucin MUC3. Here, we present a novel PDZ-dependent interaction between the MUC3 mucin and the Golgi-associated PDZ protein, GOPC. We demonstrate that GOPC decreased overall levels of MUC3 via a direct interaction with the mucin PDZ motif and that MUC3 competed with CFTR for binding to GOPC. Overexpression of CFTR counteracted the suppressive effect of GOPC on MUC3 protein levels.
YFP–MUC3CT interacts with the PDZ protein GOPC
Analysis of the C-terminal amino acid sequence of MUC3 revealed a class I PDZ motif in the extreme C-terminus of the protein (Fig. 1B). It was therefore predicted that MUC3CT was able to act as ligand to PDZ domain-containing proteins. The cytoplasmic part of MUC3 (MUC3CT) was fused to YFP (Fig. 1B) and expressed in CHO-K1 cells. Cell lysates were used in overlay assays on a total of 123 different human recombinant PDZ domains immobilized on membranes. Bound YFP–MUC3CT was detected with an anti-GFP mAb. A strong binding of YFP–MUC3CT to the single PDZ domain of the protein GOPC (accession no. NP_065132; Fig. 2A) was observed whereas YFP–MUC3CT did not bind a negative control.
GST–MUC3CT binds specifically to HA–GOPC
Next, MUC3CT, MUC12CT and MUC17CT were fused with glutathione S-transferase (GST), and expressed as proteins in Escherichia coli. GST–MUC3CT was bound to glutathione–Sepharose beads and incubated with lysates of CHO-K1 cells expressing HA-tagged full-length GOPC. Material pulled down by GST–MUC3CT was separated using SDS-PAGE and developed with an anti-HA mAb, detecting HA–GOPC. GST–MUC12CT and GST–MUC17CT were used as controls in the same experiment. Results from immunoblots with anti-HA mAb showed a dense band pulled down by GST–MUC3CT, and a very faint band was revealed after pull-down with GST–MUC17CT (Fig. 2B). GST–MUC12CT and GST protein alone failed to pull down HA–GOPC. Thus, we concluded that GST–MUC3CT and GST–MUC17CT were able to bind HA–GOPC whereas GST–MUC12CT, despite having a class I PDZ motif, was not.
The interaction between GST–MUC3CT and HA–GOPC is PDZ dependent
MUC3CT is made up of 73 amino acids terminating with the class I PDZ motif –SSSV. To ensure that the binding to HA–GOPC was PDZ dependent a truncated GST fusion protein of MUC3CT was constructed by introducing a stop codon at position −5 (GST–MUC3CTΔ6; Fig. 1B). GST, GST–MUC3CT and GST–MUC3CTΔ6 were bound to glutathione–Sepharose beads and incubated with lysates of CHO-K1 cells expressing HA–GOPC. Bound products were then separated by SDS-PAGE and analyzed by immunoblotting using the anti-HA mAb. When binding of GST–MUC3CT to HA–GOPC was normalized to 100%, binding of HA–GOPC to GST–MUC3CTΔ6 was reduced to 10% (Fig. 2C). Thus, we concluded that the binding of GST–MUC3CT to GOPC was mediated by the reported class I PDZ motif, but also influenced by sequences further N-terminal in MUC3CT.
Myc–MUC17/MUC3CT and HA–GOPC colocalize in BHK-21 cells
Because GST–MUC3CT and HA–GOPC interact in vitro we investigated whether this interaction also occurs in living cells. Two plasmids encoding recombinant fusion proteins harboring native and truncated MUC3CT fused to the SEA domain and the transmembrane domain of MUC17 with a Myc tag instead of the PTS domain were constructed (Fig. 1A,B). BHK-21 cell lines stable expressing these proteins (Myc–MUC17/MUC3CT or Myc–MUC17/MUC3CTΔ6) were then established. It has previously been shown that juxtamembrane cytoplasmic amino acid residues play a vital role for proper processing of a protein in the ER, and exit from the ER and cis-Golgi (Matsuda et al., 2004; Mo et al., 2008). For that reason, correct processing of recombinant Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 proteins was confirmed by submitting them to endoglycosidases in order to test whether the proteins could exit the ER. Digestion using Endo H revealed that both Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 were resistant to digestion. Thus neither protein harbored any N-linked high-mannose glycans, proving that recombinant proteins passed through ER and cis-Golgi compartments (Fig. 3A, left panel). However, both Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 were sensitive to treatment with the endoglycosidase PNGase F, corroborating the existence of processed N-linked oligosaccharides on the protein backbones (Fig. 3A, right panel). Consequently, we concluded that the recombinant Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 proteins had migrated at least to the trans-Golgi compartment and therefore undergone proper processing in BHK-21 cells. Our conclusion was also confirmed by confocal immunofluorescence microscopy, which showed that both Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 were found on the plasma membrane of BHK-21 cells (Fig. 3B,C). To investigate a possible colocalization of recombinant Myc–MUC17/MUC3CT with HA–GOPC, HA–GOPC was transfected into BHK-21 cells expressing Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 and the cells were then examined using immunofluorescence microscopy. Overexpressed HA–GOPC in BHK-21 cells stably expressing Myc–MUC17/MUC3CT colocalized with Myc–MUC17/MUC3CT (Fig. 3B), whereas Myc–MUC17/MUC3CTΔ6 and HA–GOPC did not (Fig. 3C). Our observations were confirmed by quantifying the colocalization of HA–GOPC with Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6. Colocalization of GOPC with Myc–MUC17/MUC3CT was six times higher than with Myc–MUC17/MUC3CTΔ6 (Fig. 3D). Thus, we concluded that Myc–MUC17/MUC3CT colocalized with HA–GOPC in BHK-21 cells and that this interaction was dependent on the class I PDZ motif present on Myc–MUC17/MUC3CT.
MUC3 colocalizes with GOPC because of direct interaction
In order to ensure that the colocalization of Myc–MUC17/MUC3CT and HA–GOPC was due to a direct interaction between the two proteins, co-immunoprecipitation of Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 with HA–GOPC was studied. Material precipitated by anti-Myc mAb in HA–GOPC-transfected Myc–MUC17/MUC3CT-expressing cells was separated using SDS-PAGE and detected with anti-HA mAb. The immunoblot showed the presence of HA–GOPC (Fig. 4A). When the same experiment was performed in cells expressing Myc–MUC17/MUC3CTΔ6, HA–GOPC was not precipitated. The reciprocal experiment confirmed the interaction when HA–GOPC precipitated Myc–MUC17/MUC3CT, but not Myc–MUC17/MUC3CTΔ6 (Fig. 4B). Consequently, colocalization of Myc–MUC17/MUC3CT with HA–GOPC in BHK-21 cells was mediated by a direct protein–protein interaction.
MUC3 expression is reduced when GOPC is overexpressed
Previous studies showed that GOPC can suppress CFTR in cells (Cheng et al., 2004). A relevant question is whether the level of MUC3 is also regulated by GOPC and if there is a competition between MUC3 and CFTR. First we set out to confirm that total protein levels of CFTR are reduced upon overexpression of GOPC. We transiently coexpressed CFTR and HA–GOPC in BHK-21 cells and analyzed total CFTR levels in whole cell lysates by SDS-PAGE. Immunoblots revealed a decrease of CFTR expression after introduction of HA–GOPC (Fig. 5A). When GOPC was overexpressed in BHK-21 cells expressing Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 the total protein level of Myc–MUC17/MUC3CT was reduced upon introduction of the pKH3–GOPC plasmid (Fig. 5B). However, the identical experiment in the Myc–MUC17/MUC3CTΔ6-expressing cells did not alter the levels of Myc–MUC17/MUC3CTΔ6. Thus, given the colocalization and co-immunoprecipitation of Myc–MUC17/MUC3CT with HA–GOPC, we concluded that GOPC suppressed Myc–MUC17/MUC3CT protein via a PDZ-dependent interaction.
Overexpression of GOPC does not affect the cell surface expressed Myc–MUC17/MUC3CT
The total amount of Myc–MUC17/MUC3CT was affected by the expression of GOPC, but it was not clear whether this was due to alterations of the plasma membrane or intracellular membrane pools (ER, Golgi and endosomes). Therefore, the cell-surface-expressed MUC3 was studied using FACS analysis in BHK-21 cells expressing Myc–MUC17/MUC3CT or Myc–MUC17/MUC3CTΔ6 transfected with mock or the pKH3–GOPC plasmid. Unpermeabilized BHK-21 cells transfected with pKH3–GOPC showed a plasma membrane expression of MUC3 equivalent to that of mock-transfected cells expressing Myc–MUC17/MUC3CT. Tansfected BHK-21 cells stably expressing Myc–MUC17/MUC3CTΔ6 also showed no alterations in cell surface levels of Myc–MUC17/MUC3CTΔ6 (Fig. 5C). A similar experiment analyzing permeabilized cells for total protein expression revealed that total Myc–MUC17/MUC3CT was 60% lower upon HA–GOPC overexpression compared with mock-transfected cells (Fig. 5D). A weaker effect was observed in BHK-21 cells expressing Myc–MUC17/MUC3CTΔ6 cells as the reduction was only 30% compared with mock-transfected cells. Thus, overexpression of HA–GOPC resulted in a reduction in total Myc–MUC17/MUC3CT, because the plasma membrane pool of the protein was unaffected.
CFTR and MUC3 compete for the suppressive role of GOPC
GOPC suppresses CFTR levels by a direct interaction between mature CFTR and GOPC in a process that involves post-Golgi vesicle formation and trafficking to the lysosome (Cheng et al., 2002). Because both MUC3 and CFTR bound to GOPC, we postulated that MUC3 is targeted for degradation through its interaction with GOPC, and furthermore that MUC3 and CFTR compete for binding to GOPC. This was tested in BHK-21 cells stably expressing Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 by transfecting the cells with mock and increasing levels of pcDNA–CFTR followed by analysis of total Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6. FACS analysis revealed that overexpression of pcDNA–CFTR in BHK-21 cells expressing Myc–MUC17/MUC3CT increased the total level of Myc–MUC17/MUC3CT in a dose-dependent manner, with the largest effect at 8 μg/ml pcDNA3–CFTR (Fig. 6A). This dose dependency demonstrated that CFTR had a specific effect on the overall levels of Myc–MUC17/MUC3CT. Repeating the experiment in cells stably expressing Myc–MUC17/MUC3CTΔ6 showed no effect on the total levels of Myc–MUC17/MUC3CTΔ6 when the CFTR protein was expressed (Fig. 6B). Thus, the CFTR can counteract the suppressive role of GOPC on MUC3, probably by simply competing with Myc–MUC17/MUC3CT for binding to GOPC. Moreover, the PDZ motif on Myc–MUC17/MUC3CT was vital for CFTR-mediated regulation of MUC3 levels mediated through GOPC.
CFTR and MUC3 display competition in Caco-2 cells
To investigate the competitive interplay between CFTR and MUC3 in a cell system where CFTR and MUC3 were endogenously expressed, we turned to Caco-2 cells, a human epithelial colorectal adenocarcinoma cell line. In order to achieve overexpression of CFTR and MUC3, we overexpressed YFP-fused cytoplasmic tails of CFTR and MUC3 and then assessed total levels of the corresponding competitive partner. In the first set up, YFP–MUC3CT was overexpressed in Caco-2 cells and total protein levels of endogenous CFTR were quantified using FACS. Because overexpression of YFP–MUC3CT resulted in a 25±4% increase in total CFTR levels in comparison with non- and mock-transfected cells, no corresponding increase of CFTR levels could be observed upon overexpression of YFP–MUC3CTD6 (Fig. 7A). Thus we concluded that the YFP–MUC3CT caused an increase in total protein levels of endogenous CFTR in Caco-2 cells and that this effect was PDZ dependent. A reciprocal experiment was performed in which the C-terminus of CFTR fused to an YFP protein was overexpressed in Caco-2 cells, and total protein levels of endogenous MUC3 were measured using FACS. Similar to the experiment in Fig. 6A, we overexpressed increasing amounts of pYFP–CFTR1440-80 plasmid in Caco-2 cells and stained total MUC3 using a polyclonal antibody raised against the cytoplasmic tail of MUC3. Compared with mock transfectants, overexpression of pYFP–CFTR1440-80 resulted in increasing total levels of MUC3, reaching a maximum of 41±6% increase in cells transfected with 500 ng pYFP–CFTR1440-80 (Fig. 7B).
In a previous study, PDZK1, well-known to interact with CFTR, was shown to be essential for the apical surface expression of the transmembrane mucin MUC17 on the intestinal enterocyte. This effect was dependent on the class I PDZ motif (–TTSF) on the extreme C-terminus of MUC17 (Malmberg et al., 2008). Here we demonstrate that another transmembrane mucin, MUC3, interacts with a well-known CFTR regulator, the Golgi-associated PDZ protein GOPC. This novel interaction is PDZ dependent as MUC3 lacking the most C-terminal six amino acids encoding the class I PDZ motif failed to bind GOPC and mediate the GOPC-dependent effects on MUC3 levels. Moreover, GOPC reduced the total levels of MUC3 because plasma membrane levels were largely unaffected. Lastly, we used a heterologous expression system as well as an intestinal cell line endogenously expressing both MUC3 and CFTR to demonstrate that CFTR counteracts the suppressive effect of GOPC on total levels of MUC3 as a result of competition with MUC3 for binding to GOPC.
Alterations in mucus composition and mucin expression levels have been coupled to the disease cystic fibrosis. Previous studies have demonstrated a tenfold increase in levels of soluble mucins in Cftr knock-out (CF) mice (Malmberg et al., 2006). Also, increased amounts of soluble Muc3(17) are observed in CF–Muc1 double knock-out and CF–MUC1 transgenic mice. However, these and other more descriptive studies has so far not been able to reveal a direct coupling between the mucus phenotype typical for cystic fibrosis and specific mucins and their function.
The interaction of GOPC with CFTR has been well studied. The expression of CFTR and the presentation of the anion channel on apical surfaces of epithelial cells is a highly dynamic process, involving several different PDZ proteins. The PDZ protein GOPC binds CFTR in a PDZ-dependent manner, in turn reducing chloride currents as a result of reduced surface expression and degradation of CFTR in lysosomes (Cheng et al., 2002; Cheng et al., 2004; Gentzsch et al., 2003). It has also been postulated that the negative effect of GOPC on CFTR levels is counteracted by at least one member of the NHERF family of PDZ proteins, namely NHERF1. CFTR levels are restored by NHERF1 in the presence of GOPC (Cheng et al., 2002).
We have identified GOPC as a regulator of MUC3 levels. Firstly, GOPC bound MUC3 at its C-terminal PDZ motif because the binding of the MUC3 cytoplasmic tail to GOPC decreased to 10% upon deletion of the C-terminal amino acid residues –SSSV. This sequence conforms to a class I PDZ binding motif. When two more amino acids were removed the binding was further decreased to 6%. This suggests that also amino acids residues N-terminal of the class I PDZ motif are important for the interaction. It has been shown in other studies that amino acids outside the typical PDZ-binding motif can influence the PDZ domain–ligand interactions (Kozlov et al., 2000; Niethammer et al., 1998; Songyang et al., 1997). Secondly, MUC3 and GOPC colocalized when transfected into cells. Thirdly, MUC3 coprecipitated with GOPC, further suggesting a direct interaction between the two proteins. Fourthly, overexpression of GOPC reduced the overall levels of MUC3 in BHK-21 cells. Finally, the suppressive effect of GOPC on MUC3 was reversed upon introduction of CFTR, suggesting that CFTR and MUC3 compete for binding GOPC.
Previous studies of CFTR indicate a complex and dynamic regulatory system where GOPC acts as a coordinator at the Golgi level for targeting proteins to the degradation pathway instead of their plasma membrane destination. Our studies suggested that GOPC has a similar effect on the MUC3 mucin. In fact, the observations suggest that MUC3 and CFTR both bind and compete for the single PDZ domain present on GOPC. According to this model, alterations in the expression levels of partner A of GOPC affect the number of unoccupied GOPC proteins presented to partner B of GOPC. In the first case, the levels of partner A are increased, thus resulting in a lower number of free GOPC PDZ domains presented to partner B. This, in turn, generates lower numbers of partner B that can be targeted to the degradation pathway via GOPC and so the total levels of partner B are increased. This model is outlined in Fig. 8 and suggests that an increased concentration of CFTR will give higher total levels of MUC3.
Previous results suggest that it is not only the total amounts of CFTR that is reduced by increased levels of GOPC, but also the cell surface expression (Cheng et al., 2002). When similar experiments were performed with MUC3, it was only the total intracellular pool that was reduced, as the cell surface levels were not significantly affected. This suggests that mechanisms other than the total level of MUC3 regulate the surface expression of this mucin. Our previous study of the role of Pdzk1 protein in Muc3(17) (the mouse ortholog of MUC17) targeting to apical membranes showed that although the lack of Pdzk1 results in a larger pool of intracellular Muc3(17), the pool of Muc3(17) present on the apical surface is not altered (Malmberg et al., 2008). This is in line with our observation in BHK-21 cells stably expressing MUC3, where overexpression of GOPC only exerts its effect on the total pool of MUC3.
The results emphasize the dynamic precision by which MUC3 and CFTR interact with GOPC. Although CFTR asserted a dose-dependent effect on MUC3, where even low concentrations (50 ng/ml plasmid DNA in BHK-21 and 100 ng/ml in Caco-2 cells) provoked an increase in MUC3 total levels, a considerably higher concentration of MUC3 (2.5 μg/ml plasmid DNA) was required to suppress total CFTR expression levels. Although differences in DNA delivery and introduction into the cell have to be taken into account, these findings suggest that CFTR has a higher affinity for GOPC than MUC3.
The understanding of the dynamic interplay involving GOPC, MUC3 and CFTR in cystic fibrosis is of great importance. A previous study by Malmberg et al. demonstrated increased mucin levels in Cftr−/− mice (Malmberg et al., 2006). At first glance these results contradict our hypothesis that MUC3 levels are decreased in absence of CFTR, but one has to take into account that semiquantitative analysis of mucins performed by Malmberg et al. involved the murine Muc1 and Muc3(17) mucins, and not the Muc3 mucin studied here. In addition, reports in the literature regarding the mutant CFTRΔF508 are indecisive. It has been established that misfolded CFTRΔF508 is retained in the ER and degraded (Gelman and Kopito, 2002). According to these conclusions, a state where CFTRΔF508 is present and trapped in ER, resembles the situation in which CFTR is absent from the trans-Golgi network because of a knockout, thus resulting in decreased MUC3 levels. One report, however, states that surface expression of CFTRΔF508 is in fact increased upon suppression of GOPC levels using siRNA (Wolde et al., 2007). These inconclusive reports call for a future evaluation of the effects of GOPC–MUC3 interaction on CFTRΔF508 levels.
The other two transmembrane mucins, MUC12 and MUC17, are similar to MUC3 and also abundant in the intestine. Interestingly, MUC12 did not bind to GOPC and MUC17 bound only weakly. MUC17, but not MUC3 and MUC12, has been shown to interact with PDZK1, another well-known regulator of CFTR trafficking (Malmberg et al., 2008; Wang et al., 2000). This suggests a special function of MUC3 in relation to CFTR. However, the significance of this is not understood as there is little known about the normal function of transmembrane mucins. MUC1 has been studied in the context of cancers, but no function has been assigned to the human MUC3, MUC12 and MUC17 mucins. Most transmembrane mucins (MUC1, 3, 12, 13, 16 and 17) have one or several SEA domains close to the membrane on the extracellular side of the cell. The SEA domain is cleaved, but still held together, and has a structure that suggest interacting partners (Macao et al., 2006). Thus the SEA domain was suggested to be involved in the transmembrane mucins role as sensors for extracellular phenomena. One hypothesis is that physical stress, such as stress exerted by adhering bacteria, applied on the extended mucin domain of MUC3 causes dissociation of the mucin at the SEA domain. Dissociation of the SEA domain in turn generates intracellular signals that promote expression of MUC3 in order to compensate for loss of MUC3 mucin domains on the apical surfaces of the cell. According to our model, higher expression of MUC3 enables CFTR to bypass lysosomal degradation mediated by GOPC, thus yielding increased apical expression of CFTR and higher efflux of ions and liquid which in turn results in clearance of bacteria.
Interestingly, two transmembrane mucin-like molecules in yeast, Hkr1 and Msb2, contain a serine/threonine-rich (STR) domain, which can be glycosylated to become a mucin-like domain. Such a domain mimics an organic gel with the ability to undergo volume changes in response to alteration in solvent composition as suggested in previous studies (Kesimer and Sheehan, 2008; Tanaka et al., 1980). Hkr1 and Msb1 have been shown to act as sensors for extracellular osmotic pressure and transduce information to the cell interior using signaling pathways (Tatebayashi et al., 2007).
In accordance with the studies in yeast and mammals, the mucin domain of transmembrane mucins such as MUC3 can be hypothesized to play a role in monitoring extracellular ion gradients, pH and the amount of liquid in the vicinity of the epithelial cell surface. Because CFTR is the major ion channel controlling the amount of liquid on epithelia, as studied in human airway epithelial cell cultures, a coupling between CFTR and the transmembrane mucin is not unrealistic (Matsui et al., 1998). A deeper understanding of the relationship of the transmembrane mucins and the CFTR channel might shed further light on the still poorly understood relationship between cystic fibrosis and the mucus phenotype observed in this disease.
In summary, a novel PDZ-dependent interaction between MUC3 and GOPC was demonstrated, and that binding to CFTR can modulate MUC3 expression levels by competing for binding to GOPC. These observations are the first demonstration of a direct coupling between a mucin and the CFTR channel, something that will further increase the understanding of mucins in the disease cystic fibrosis.
Materials and Methods
Plasmids encoding glutathione S-transferase (GST) and yellow fluorescent protein (YFP) fusion proteins of MUC3, MUC12 and MUC17 were described previously (Malmberg et al., 2008). pSM-MUC17/MUC3CT and pSM-MUC17/MUC3CTΔ6 plasmids were constructed by fusing the CT of MUC3 (amino acids 2467–2540) to a pSM-MUC17ΔPTS plasmid using a PstI restriction site at positions 4412–4413. The empty pSM plasmid was used as mock treatment where stated. pYFP–CFTR1440-80 was designed by fusing amino acid residues 1140–1480 to an YFP protein. Primers 5′-gtgcaggtaccgccatcagcccctccgacaggg-3′ and 5′-tgacgggatccctaaagccttgtatcttgcacc-3′ were used to clone CFTR CT into the pYFP plasmid using KpnI and BamHI restriction sites. The plasmids encoding HA-tagged GOPC and pcDNA3–CFTR have been described previously (Gentzsch et al., 2003) and were kindly provided by John R. Riordan (UNC Chapel Hill, NC).
Cell culture and transfection
Syrian hamster kidney (BHK) 21 and Chinese hamster ovary (CHO) K1 cell lines (ATCC) were cultured at 37°C in 5% CO2 in Iscove's modified Dulbecco's medium (Invitrogen) containing 10% (v/v) fetal calf serum and supplemented with sodium pyruvate (110 mg/l), L-arginine (116 mg/l), L-glutamine (290 mg/l), L-asparagine (36 mg/l), folic acid (10 mg/l) and β-mercaptoethanol (3.49 μl/l) at 37°C in 5% CO2. The BHK-21 cells were transiently transfected using Lipofectamine™ 2000 reagent (Invitrogen) according to manufacturer's recommendations, whereas CHO-K1 cells were transfected using polyethylenimine (PEI; Sigma-Aldrich). CHO-K1 cells were seeded on a 10 cm Petri dish and at 90% confluency, cells were transfected with 36 μg DNA mixed with 20% glucose (1 μl/μg DNA) and 22 μl 0.1 M PEI. DNA–PEI complexes were allowed to form at room temperature for 10 minutes before adding to 9 ml culture medium.
Stable tranfectants were produced by transfecting BHK-21 cells seeded on 9.6 cm2 plates with 4 μg pSM-MUC17/MUC3CT or pSM-MUC17/MUC3CTΔ6 plasmids using Lipofectamine™ 2000 reagent (Invitrogen). Cells were harvested 48 hours post-transfection, and sparsely seeded in 10 cm Petri dishes and selected with 700 μg/ml Genetecin (Invitrogen) for 14 days. Upon selection, individual clones were picked and screened for expression of Myc–MUC17/MUC3CT or Myc–MUC17/MUC3CTΔ6. Positive clones were recloned.
The hemagglutinin (HA) tag was detected using an anti-HA mAb (Sigma) and the YFP tag was detected using an anti-GFP mAb (Sigma). Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 were detected using anti-Myc mAb obtained directly from the culture medium of the 9E10.2 hybridoma (CRL-1729; ATCC). Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 were also detected using an anti-MUC17S2 polyclonal antiserum raised in rabbit against the sequence KYTPEYKTVLDNATEVVKEKITKVC. Endogenous human MUC3 was detected using an anti-MUC3C1 polyclonal antiserum raised in rabbit against a C-terminal sequence CDTTMKVHIKRPEMT. Briefly, 2 mg of peptide was conjugated to KLH using its terminal cysteine and maleimide crosslinker, purified and injected to two rabbits. The mouse mAb 450, raised against the R domain of purified CFTR protein, has been described previously (Cui et al., 2007). A monoclonal antibody against β-actin was purchased from BD Biosciences.
Immunoblotting, pull-down and co-immunoprecipitation
Cells were washed in ice-cold PBS, harvested and solubilized in radioimmunoprecipitation assay buffer (RIPA: 25 mM Tris-HCl pH 7.6, 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS) or 0.09% NP-40 lysis buffer (Nonidet P40, 150 mM NaCl, 50 mM Tris-HCl, pH 7.4, and 20 mM NaMoO4) supplemented with Complete™ protease inhibitor cocktail (Roche), N-ethylmaleimide (Sigma) and EDTA (Merck). Cells lysates were centrifuged at 14,000 g for 5 minutes at 4°C to pellet insoluble material. The supernatants were incubated with reducing sample buffer containing 1,4-dithiothreitol (DTT) for 30 minutes, and subjected to SDS-PAGE.
GST pull-down experiments were performed as previously described (Malmberg et al., 2004). Briefly, binding was performed in 0.09% NP-40 lysis buffer containing Complete™ protease inhibitor cocktail (Roche) overnight at 4°C. Glutathione–Sepharose™ 4B beads (GE Healthcare) were washed three times with 0.09% NP-40 lysis buffer. Bound proteins were eluted for 30 minutes with SDS-PAGE sample buffer containing DTT, and separated using SDS-PAGE.
Co-immunoprecipitations were carried out by adding anti-Myc mAb or anti-HA mAb to cell lysates, followed by binding overnight at 4°C. Resuspended protein G PLUS-agarose beads (Santa Cruz) were added to lysates and were isolated after 2 hours. Beads were washed three times with 1 ml 0.09% NP-40 lysis buffer, coprecipitated material was eluted with reducing sample buffer containing DTT for 30 minutes, and separated on 10% and 12% acrylamide gels using SDS-PAGE. A LAS-4000 mini chemiluminescence system (Fujifilm) was used to visualize the samples.
Lysates from BHK-21 cells stably expressing Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 were subjected to endoglycosidase (Endo) H and N-glycosidase (PNGase) F treatment (Roche). Briefly, protein was reduced in SDS-PAGE sample buffer containing DTT for 5 minutes at 95°C and treated with 25 mIU Endo H in 20 mM sodium phosphate, pH 5, overnight at 37°C. Treatment with 1.0 IU PNGase F was performed in NP-40 lysis buffer overnight at 37°C. Samples without enzymes, incubated at room temperature or stored at −20°C, served as controls. Samples were reheated to 95°C before separation using SDS-PAGE.
Cells were seeded on glass coverslips at 90% confluency and transfected with 4 μg of pKH3–GOPC plasmid. 24 hours after transfection, cells were fixed in 3.7% paraformaldehyde for 10 minutes, permeabilized in 0.1% Triton X-100 (in PBS) for 5 minutes and blocked with 5% bovine serum albumin in PBS for 20 minutes. Myc–MUC17/MUC3CT and Myc–MUC17/MUC3CTΔ6 were detected with MUC17S2 pAb, and HA–GOPC was stained with a anti-HA mAb. Secondary antibodies were either goat-anti rabbit Alexa Fluor 488 or goat-anti mouse Alexa Fluor 546 (Invitrogen). Cells were examined using a Bio-Rad Radiance 2000 confocal microscope.
PDZ domain array
YFP-tagged MUC3CT was expressed in CHO-K1 cells. TranSignal™ PDZ Domain Array IV (Panomics, Redwood City, CA) was prepared according to the manufacturer's instructions. Bound YFP–MUC3CT was probed and detected with anti-GFP mAb (Sigma).
Fluorescence-activated cell sorting (FACS)
Seeded and transfected cells were harvested with 13 Puc A solution (4.0g/l KCl, 80.0 g/l NaCl, 3.5 g/l NaHCO3, 10.0 g/l D-glucose and 2.0 g/l EDTA) and passed through a 70 μm nylon cell strainer (BD Falcon). Cells were centrifuged at 200 g for 5 minutes and plated into a V-shaped 96 MicroWell™ Plate (Nunc, Thermo Fisher Scientific). Cells were kept either unpermeabilized on ice in FACS solution (2% BSA in PBS) or permeabilized in FACS solution containing 0.1% saponin (Sigma) for 20 minutes. Cells were stained for 30 minutes with Myc mAb (1:50), anti-HA mAb (1:200) or mAb 450 (1:100) diluted in FACS solution and washed once with FACS solution. Finally, cells were stained with goat anti-mouse conjugated to allophycocyanin (APC; 1:100) and 7AAD (1:100; both from BD Biosciences) for 30 minutes and 5 minutes, respectively. Samples were analyzed on a FACSCalibur (BD Biosciences).
Where stated, results are expressed as means ± s.e.m. Differences were regarded as significant at P<0.05, P<0.01 or P<0.001.
We are indebted to J. R. Riordan for the pKH3–GOPC and pcDNA3–CFTR plasmids. This work was supported by the Swedish Research Council (no. 7461, 21027 and 342-2004-4434), the Swedish Cystic Fibrosis Foundation, the Swedish Cancer Foundation, the Knut and Alice Wallenberg Foundation (KAW2007.0118), IngaBritt and Arne Lundberg Foundation, Sahlgren's University Hospital (LUA-ALF), European Union-FP7 IBDase (no. 200931), Wilhelm and Martina Lundgren's Foundation, Torsten och Ragnar Söderbergs Stiftelser, and the Swedish Foundation for Strategic Research, The Mucosal Immunobiology and Vaccine Center (MIVAC) and the Mucus-Bacteria-Colitis Center (MBC) of the Innate Immunity Program (2010–2014). We acknowledge the Centre for Cellular Imaging at the University of Gothenburg for technical help.