The sialidase NEU1 directly interacts with the juxtamembranous segment of the cytoplasmic domain of mucin-1 to inhibit downstream PI3K-Akt signaling

The extracellular domain (ED) of the membrane-spanning sialoglycoprotein, mucin-1 (MUC1), is an in vivo substrate for the lysosomal sialidase, neuraminidase-1 (NEU1). Engagement of the MUC1-ED by its cognate ligand, Pseudomonas aeruginosa-expressed flagellin, increases NEU1-MUC1 association and NEU1-mediated MUC1-ED desialylation to unmask cryptic binding sites for its ligand. However, the mechanism(s) through which intracellular NEU1 might physically interact with its surface-expressed MUC1-ED substrate are unclear. Using reciprocal coimmunoprecipitation and in vitro binding assays in a human airway epithelial cell system, we show here that NEU1 associates with the MUC1-cytoplasmic domain (CD) but not with the MUC1-ED. Prior pharmacologic inhibition of the NEU1 catalytic activity using the NEU1-selective sialidase inhibitor, C9-butyl amide-2-deoxy-2,3-dehydro-N-acetylneuraminic acid, did not diminish NEU1-MUC1-CD association. In addition, glutathione-S-transferase (GST) pull-down assays using the deletion mutants of the MUC1-CD mapped the NEU1-binding site to the membrane-proximal 36 aa of the MUC1-CD. In a cell-free system, we found that the purified NEU1 interacted with the immobilized GST-MUC1-CD and the purified MUC1-CD associated with the immobilized 6XHis-NEU1, indicating that the NEU1-MUC1-CD interaction was direct and independent of its chaperone protein, protective protein/cathepsin A. However, the NEU1-MUC1-CD interaction was not required for the NEU1-mediated MUC1-ED desialylation. Finally, we demonstrated that overexpression of either WT NEU1 or a catalytically dead NEU1 G68V mutant diminished the association of the established MUC1-CD binding partner, PI3K, to MUC1-CD and reduced downstream Akt kinase phosphorylation. These results indicate that NEU1 associates with the juxtamembranous region of the MUC1-CD to inhibit PI3K-Akt signaling independent of NEU1 catalytic activity.

The extracellular domain (ED) of the membrane-spanning sialoglycoprotein, mucin-1 (MUC1), is an in vivo substrate for the lysosomal sialidase, neuraminidase-1 (NEU1). Engagement of the MUC1-ED by its cognate ligand, Pseudomonas aeruginosa-expressed flagellin, increases NEU1-MUC1 association and NEU1-mediated MUC1-ED desialylation to unmask cryptic binding sites for its ligand. However, the mechanism(s) through which intracellular NEU1 might physically interact with its surface-expressed MUC1-ED substrate are unclear. Using reciprocal coimmunoprecipitation and in vitro binding assays in a human airway epithelial cell system, we show here that NEU1 associates with the MUC1-cytoplasmic domain (CD) but not with the MUC1-ED. Prior pharmacologic inhibition of the NEU1 catalytic activity using the NEU1-selective sialidase inhibitor, C9-butyl amide-2-deoxy-2,3-dehydro-N-acetylneuraminic acid, did not diminish NEU1-MUC1-CD association. In addition, glutathione-S-transferase (GST) pull-down assays using the deletion mutants of the MUC1-CD mapped the NEU1binding site to the membrane-proximal 36 aa of the MUC1-CD. In a cell-free system, we found that the purified NEU1 interacted with the immobilized GST-MUC1-CD and the purified MUC1-CD associated with the immobilized 6XHis-NEU1, indicating that the NEU1-MUC1-CD interaction was direct and independent of its chaperone protein, protective protein/ cathepsin A. However, the NEU1-MUC1-CD interaction was not required for the NEU1-mediated MUC1-ED desialylation. Finally, we demonstrated that overexpression of either WT NEU1 or a catalytically dead NEU1 G68V mutant diminished the association of the established MUC1-CD binding partner, PI3K, to MUC1-CD and reduced downstream Akt kinase phosphorylation. These results indicate that NEU1 associates with the juxtamembranous region of the MUC1-CD to inhibit PI3K-Akt signaling independent of NEU1 catalytic activity.
Glycoconjugates expressed on the surface of all eukaryotic cells contain oligosaccharide chains that terminate with N-acetylneuraminic acid (sialic acid) (1,2). Because of their terminal location and negative charge, sialic acid residues are strategically positioned to influence intermolecular and cellcell interactions through steric hindrance and/or electrostatic repulsion. The sialylation state of glycoproteins and glycolipids is dynamically and coordinately regulated through the opposing catalytic activities of sialyltransferases and sialidase/ neuraminidase (NEU) (3,4). NEUs hydrolyze the glycosidic linkage between terminal sialic acid and the adjacent subterminal sugar within glycan chains. Four mammalian NEUs have been identified, NEU1, 2, 3, and 4 (5)(6)(7)(8)(9). Although it lacks the conserved protein structural domains characteristic of canonical NEUs, Klotho, a protein originally associated with aging and senescence, has been reported to exhibit sialidase activity (10).
NEU1 associates with the MUC1-CD but not the MUC1s-ED in in vitro binding assays To confirm that NEU1 associates with the MUC1-CD, but not the MUC1-ED, in vitro binding assays were performed with glutathione-S-transferase (GST) NEU1 immobilized on glutathione-agarose. The GST-NEU1 recombinant protein encompassed the 217-aa Schistosoma japonicum GST protein fused to the NH 2 -terminus of the 415-aa full-length NEU1 protein and included its highly conserved Phe/Tyr-Arg-Ile-Pro motif and 5 Asp boxes ( Fig. 2A). To validate the GST-NEU1 construct, we found that GST-NEU1 immobilized on glutathione-agarose bound to the established NEU1-binding partner, PPCA, to a much greater extent than did glutathione-agarose alone (Fig. 2B, lane 2 versus lane 1). Human embryonic kidney (HEK)293T cells transfected for the expression of either the MUC1-ED or the MUC1-CD were lysed. The lysates were incubated with the immobilized GST-NEU1 and the NEU1-binding proteins processed for MUC1-ED or MUC1-CD immunoblotting. GST-NEU1 associated with the MUC1-CD (Fig. 2C, lane 7 versus 6), but not with the MUC1-ED (Fig. 2C, lane 3 versus 2).
To further establish the NEU1-MUC1-CD interaction, we constructed a GST-fusion protein encompassing the entire 72-aa MUC1-CD (aa1-72) (Fig. 2D). To validate the GST-MUC1-CD (aa1-72) construct, we found that in GST pulldown assays, it associated with 3 established MUC1-CD binding partners, c-Src (30)  To further support that NEU1 did not associate with the MUC1-ED, binding assays were performed using the MUC1-ED protein containing an NH 2 -terminal 6XHis epitope tag ( Fig. 2G) immobilized on nickel-nitrilotriacetic acid (Ni-NTA)-agarose. To validate the 6XHis-MUC1-ED construct, in Ni-NTA-agarose pull-down assays, MUC1-ED associated with its established ligand, Pa-expressed flagellin, dramatically more than did the Ni-NTA-agarose bead control (Fig. 2H, lane 2 versus lane 1). The lysates of human airway cells infected with recombinant adenovirus encoding FLAGtagged NEU1 (Ad-NEU1-FLAG) were similarly incubated with the validated 6XHis-MUC1-ED immobilized on Ni-NTAagarose or Ni-NTA-agarose alone and the MUC1-ED binding proteins processed for FLAG (NEU1) immunoblotting. NEU1 binding to the immobilized MUC1-ED was not detected (Fig. 2I, lane 2). Therefore, in a tightly controlled in vitro system, again, NEU1 directly/indirectly interacts with the MUC1-CD but not with the MUC1-ED.

NEU1 fails to bind to the MUC1-ED deletion mutants
We next asked whether the inability of NEU1 to bind to the MUC1-ED might be explained through intramolecular interference of NEU1 binding to one or more site(s) within the MUC1-ED by other portions of the same MUC1-ED molecule. Six 6XHis epitope-tagged deletion mutants (6XHis-MUC1-ED-Δ1, -Δ2, -Δ3, -Δ4, -Δ5, and -Δ6) were constructed (Fig. S2A). Lysates of A549 cells infected with Ad-NEU1-FLAG were incubated with the 6XHis-MUC1-ED mutant constructs immobilized on Ni-NTA-agarose, and the MUC1-ED bound proteins processed for FLAG (NEU1) immunoblotting (Fig. S2B, lanes 1-7). None of the immobilized MUC1-ED mutant proteins exhibited NEU1 binding. As a loading and transfer control, A549 cell proteins that were not retained by the immobilized MUC1-ED-WT or its deletion mutants were similarly processed for FLAG (NEU1) immunoblotting, revealing NEU1 immunoreactive bands at the expected gel mobility corresponding to 45.5 kDa (11) (Fig. S2C, lanes 1-8). In the reciprocal approach, the MUC1-ED-WT and MUC1-ED mutant proteins expressed in the HEK293T cells were each adsorbed to Ni-NTA-agarose beads and eluted with imidazole. The purified, unbound MUC1-ED-WT and deletion mutants were incubated with the validated GST-NEU1 immobilized on glutathione-agarose beads, and the NEU1-binding proteins processed for 6XHis (MUC1-ED) immunoblotting (Fig. S2D, lanes 1-4, 6-8). Again, MUC1-ED binding to the immobilized NEU1 was not detected. As a positive control, MUC1-CD expressed in the HEK293T cells associated with the immobilized GST-NEU1 (Fig. S2D, lanes 5,9). As an added control, the MUC1-ED-WT and mutant proteins that did not bind to the immobilized GST-NEU1 were processed for 6XHis (MUC1-ED) immunoblotting. Here, immunostaining revealed protein bands with the anticipated gel mobilities of the respective MUC1-ED-WT and mutant proteins (Fig. S2E, lanes 1-7). These combined results suggest that intramolecular interference of any NEU1-MUC1-ED interaction is unlikely.

NEU1 binds to the juxtamembranous region of the MUC1-CD
To begin to define the MUC1-CD structural requirements for its association with NEU1, we constructed GST fusion proteins comprising its 36-aa membrane proximal NH 2 -

NEU1 binds directly to the MUC1-CD
To determine whether the NEU1-MUC1-CD interaction was direct, binding assays were performed using NEU1 that , each immobilized on glutathione-agarose beads, or were directly loaded on the gel. C, the lysates were incubated with GST-MUC1-CD (aa1-18) or GST-MUC1-CD (aa19-36), each immobilized on glutathione-agarose beads, or were directly loaded on the gel. D, the lysates were incubated with GST or GST-MUC1-CD (aa1-72) each immobilized on glutathione-agarose beads, after which the beads were incubated with Factor Xa to proteolytically release the bound proteins, and the released proteins purified on a GST trap column. B-F, the proteins bound to beads, eluted proteins, and total lysates were all processed for FLAG (NEU1) immunoblotting. E and F, GST (lane 1) and GST-MUC1-CD (lane 2), each immobilized on glutathione-agarose beads, were eluted with free glutathione and incubated with 6XHis-NEU1 coupled to Ni-NTA-agarose beads. The 6XHis-NEU1-binding proteins and purified GST-MUC1-CD (lane 3) were processed for (E) MUC1-CD or (F) GST immunoblotting. The results are representative of two to three independent experiments. B-F, The molecular weights in kDa are indicated on the left. Ad-NEU1-FLAG, adenovirus encoding FLAG-tagged NEU1; EGFR, epidermal growth factor receptor; IB, immunoblot; MUC1-CD, mucin-1 cytoplasmic domain; NEU1, neuraminidase-1; Ni-NTA, nickel-nitrilotriacetic acid; PD, pull-down. NEU1 inhibits MUC1-CD-dependent PI3K-Akt signaling had been proteolytically cleaved from its GST epitope tag by Factor Xa and purified on a GST trap column. The purified NEU1 associated with GST-MUC1-CD (aa1-72) immobilized on glutathione-agarose, but not with immobilized GST alone (Fig. 3D, lane 2 versus 1). In the reciprocal approach, 6XHis-NEU1 was immobilized on Ni-NTA-agarose and incubated with either GST-MUC1-CD (aa1-72) or GST alone, each of which was eluted from glutathione-agarose with free glutathione. The purified GST-MUC1-CD (aa1-72) associated with immobilized 6XHis-NEU1, whereas GST alone did not, as revealed by both MUC1-CD (Fig. 3E, lane 2 versus 1) and GST (Fig. 3F, lane 2 versus 1) immunoblotting. Taken together, these results indicate that NEU1 specifically and directly interacts with the juxtamembranous portion of the MUC1-CD.
We previously reported that the MUC1-CD associates with both NEU1 and PPCA in reciprocal coimmunoprecipitation assays, using cultured airway epithelial cells, in vitro (12), and murine lungs, in vivo (13). Furthermore, the NEU1-MUC1-CD and PPCA-MUC1-CD protein-protein interactions were increased after either Pa or flagellin stimulation in cell cultures (12) or intrapulmonary Pa or flagellin challenge in mice (13). The current data indicate that both the MUC1-CD and PPCA associate with the NH 2 -terminal NEU1 (aa1-139) region.
Previous studies with human erythrocytes demonstrated NEU1 as a peripheral membrane protein in the absence of PPCA (38). Here, NEU1 was present on the outer leaflet of isolated erythrocyte plasma membranes prepared by hypotonic lysis and released from the erythrocyte ghosts by alkaline treatment. In another study using an in silico approach, two putative transmembrane regions of the NEU1 sialidase were identified, NEU1 (aa139-159) and NEU1 (aa316-333) (39). In these same studies, catalytically active NEU1, in the absence of PPCA, was demonstrated on the surface of both COS7 cells ectopically overexpressing NEU1 and human macrophages endogenously expressing NEU1. Immunofluorescence studies with NEU1 recombinant proteins containing NH 2 -or COOHterminal epitope tags and fluoroprobe-labeled antibodies against the epitope tags suggested that the sialidase was oriented on the cell surface as a double-pass integral membrane protein with its termini oriented toward the intracellular compartment and its central loop region facing the extracellular space. This predicted topology places the NEU1 (aa1-139) region, which binds to both PPCA and MUC1-CD, within the intracellular compartment, immediately proximal to the putative NEU1 (aa139-159) transmembrane segment. However, in the current cell-free assays, we observed that the NEU1-MUC1-CD interaction is direct and does not require accessory proteins, including the established NEU1-binding partners, PPCA, and β-galactosidase.
The current data support a direct protein-protein interaction between NEU1 and the MUC1-CD, whereas the MUC1-ED is a proven in vivo NEU1 substrate (11)(12)(13). To our surprise, we found that MUC1-CD was not required for Pa flagellin-induced, NEU1-mediated MUC1-ED desialylation. Therefore, a direct link between NEU1-MUC1-CD association and MUC1-ED desialylation could not be established. NEU1 bound to the MUC1-CD might constitute a distinct intracellular sialidase pool uninvolved in MUC1-ED desialylation but strategically positioned to influence multiple intracellular signaling events. More specifically, our demonstration that NEU1 binds to the MUC1-CD and inhibits PI3K-Akt signaling in a noncatalytic manner provides evidence to support a role for NEU1 in regulating PI3K-related events, including cell proliferation and migration, apoptosis, glucose metabolism, and angiogenesis (64) (Fig. 7). In fact, forced NEU1 overexpression in human airway epithelia dampens EGFstimulated EGFR Tyr 1068 autophosphorylation (11) and restrains EGFR-driven epithelial cell migration in an in vitro wounding assay (65). In human lung microvascular endothelia, NEU1 overexpression reduces cell migration in wounding assays (37,66) and disrupts capillary-like tube formation on a Matrigel substrate, i.e., in vitro angiogenesis, via CD31 desialylation (65)(66)(67). In a mouse model of atherosclerosis, administration of elastin-derived peptides increased atherosclerotic plaque size In previous studies, we identified a Pa-expressed flagellin-NEU1-MUC1-ED axis that releases a shed, hyperadhesive MUC1-ED decoy receptor from the airway epithelial cell surface as a protective component of the host response to Pa lung infection (11)(12)(13). In this model, Pa-derived flagellin engages the MUC1-ED to recruit the NEU1 sialidase from the lysosome to the membrane-spanning MUC1 sialoglycoprotein. NEU1 desialylates the MUC1-ED to unmask a cryptic Gly-Ser protease recognition site, thereby permitting the shedding of soluble MUC1-ED that functions as a decoy receptor to competitively inhibit Pa from establishing an invasive infection. Pa invasion of epithelial cells is mediated, in part, through the PI3K-Akt pathway (70,71). Internalization of Pa strain PAK into Madin-Darby canine kidney or HeLa cells required activation of PI3K and Akt (70). Furthermore, Pa type IV pili mediated Pa adhesion to polarized human airway epithelial cell glycans on the apical surface, leading to activation of the PI3K-Akt signaling and bacterial invasion. However, Pa-derived flagellin engagement of heparan sulfate proteoglycans on the basolateral cell surface also activated PI3K-Akt signaling to stimulate Pa internalization (71). Together with our current data, these results suggest that NEU1, through its ability to noncatalytically disrupt the MUC1-CD-PI3K-Akt pathway, provides a host-protective mechanism to limit Pa entry into host-airway epithelia. Combined with its catalytic role in generating a hyperadhesive, shed MUC1-ED decoy receptor, NEU1 offers a bipartite level of regulation over the airway epithelial response to Pa colonization and/or infection.

Human bronchoalveolar lavage fluid
Bronchoalveolar lavage was performed on mechanically ventilated patients undergoing diagnostic bronchoscopy, as described (12). The study was conducted in accordance with the Declaration of Helsinki, protecting patients in biomedical research and was approved by the University of Maryland Institutional Review Board (protocol number HP-00059183) with informed consent obtained from all the participants. The BALF was filtered through sterile gauze, centrifuged at 450×g to remove cells, and the supernatants concentrated 25fold by passage through membrane filters (pore size, 100 kDa) in Centricon tubes (Sigma-Aldrich).
Reciprocal NEU1-MUC1-ED and NEU1-MUC1-CD coimmunoprecipitation assays To determine whether NEU1 binds to the MUC1-ED and/ or the MUC1-CD, A549 cells were stimulated for 10 min with 10 ng/ml of Pa flagellin, washed, and lysed in PBS containing 0.5% Triton X-100. In other experiments, A549 cells were stimulated with flagellin in the presence of 100 μM of the NEU1-selective sialidase inhibitor, C9-butyl amide-2-deoxy-2,3-dehydro-N-acetylneuraminic acid (67), and lysed. The cell lysates were treated with 0.1% SDS for 20 min at room temperature to separate the noncovalently associated MUC1-ED and MUC1-CD subunits and subsequently diluted 10-fold in PBS, 1.0% Triton X-100 to allow for immunoprecipitation with anti-NEU1, anti-MUC1-ED, or anti-MUC1-CD antibodies (72). The resulting immune complexes were immobilized on protein G-agarose for 2 h at 4 C, centrifuged, washed, boiled in SDS-PAGE sample buffer, and again centrifuged. The supernatants from NEU1 immunoprecipitates were processed for MUC1-ED or MUC1-CD immunoblotting, and the supernatants from MUC1-ED and MUC1-CD immunoprecipitates processed for NEU1 immunoblotting. To control for loading and transfer, the blots were stripped and reprobed with the immunoprecipitating antibody.

Recombinant plasmids and proteins
The recombinant plasmids and encoded proteins used in this study are listed in Table 1. The GST fusion proteins of  Table 1, as described (73,74). Because of the variable number of tandem repeats in the MUC1-ED, the 72-aa MUC1-CD is numbered beginning the first amino acid of MUC1-CD. The full-length NEU1 cDNA in the pBluescript II KS plasmid (6)

GST-MUC1-CD protein binding assays
To identify the aa sequence(s) within the MUC1-CD that were responsible for binding to NEU1, equal protein aliquots of Ad-NEU1-FLAG-infected cell lysates were incubated for 3 h at 4 C with GST-MUC1-CD (aa1-72), GST-MUC1-CD To establish whether the NEU1-MUC1-CD interaction was direct, the GST-NEU1-binding assay was repeated, but here GST alone or GST-MUC1-CD (aa1-72), each immobilized on glutathione agarose was incubated with NEU1 that was proteolytically cleaved from GST by Factor Xa and purified in a GST trap column, as described (73,74), and the MUC1-CDbinding proteins processed for NEU1 immunoblotting. In other experiments, GST-MUC1-CD and GST alone, each immobilized glutathione-agarose were eluted with free glutathione and incubated with 6XHis-NEU1 coupled to Ni-NTA-agarose, and the NEU1-binding proteins processed for MUC1-CD or GST immunoblotting.

Effect of NEU1 binding to MUC1-CD on PI3K-Akt signaling
To assess whether NEU1 overexpression might influence the interaction between the MUC1-CD and four of its established binding partners, PI3K, p53, c-Met, and PDFGRβ, A549 cells were infected with Ad-NEU1, Ad expressing a catalytically inactive NEU1 containing a Gly 68 -to-Val substitution (Ad-NEU1-G68V), or Ad expressing GFP (Ad-GFP), as described (11,12). Equal protein aliquots of lysates of Ad-NEU1-, Ad-NEU1-G68V-, or Ad-GFP-infected cells were incubated for 3 h at 4 C with GST-MUC1-CD (aa1-72), or GST alone, each immobilized on glutathione-agarose. The agarose-bound proteins were resolved by SDS-PAGE, transferred to PVDF membranes, and processed for PI3K (p85), p53, c-Met, or PDGFRβ immunoblotting. In still other experiments, equal protein aliquots of lysates of Ad-NEU1-, Ad-NEU1-G68V-, or Ad-GFP-infected cells were processed for pAkt immunoblotting. To control for protein loading and transfer, the immunoblots were stripped and reprobed for total Akt. The pAkt signal was quantified by densitometry and normalized to the total Akt signal in the same lane on the same stripped and reprobed blot. In all immunoblot assays described above, unmanipulated total cell lysates were included as expression and gel mobility controls for each protein of interest.

Statistical analysis
All values were expressed as means ± S.E. Differences between means were compared using the Student's t test and considered significant at p < 0.05.

Data availability
All data are contained within the manuscript.
Supporting information-This article contains supporting information (67,78). Funding and additional information-This work was supported by grant AI144497 from the National Institutes of Health (E. P. L.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflict of interest-The authors declare that they have no conflicts of interest with the contents of this article.