Glycosylation regulates the function and membrane localization of KCC4

Biochimica et Biophysica Acta 1833 (2013) 1133–1146

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbamcr

Glycosylation regulates the function and membrane localization of KCC4 Tzu-Yu Weng a, Wen-Tai Chiu b, Hsiao-Sheng Liu c, Hung-Chi Cheng d, Meng-Ru Shen e, David B. Mount f, Cheng-Yang Chou e,⁎ a

Institute of Basic Medical Sciences, College of Medicine, National Cheng Kung University, Tainan 704, Taiwan Department of Biomedical Engineering, College of Engineering, National Cheng Kung University, Tainan 704, Taiwan Department of Microbiology and Immunology, College of Medicine, National Cheng Kung University, Tainan 704, Taiwan d Department of Biochemistry and Molecular Biology, College of Medicine, National Cheng Kung University, Tainan 704, Taiwan e Department of Obstetrics & Gynecology, College of Medicine, National Cheng Kung University, Tainan 704, Taiwan f Renal Divisions, Brigham and Women's Hospital, VA Boston Healthcare System, Boston, MA, USA b c

a r t i c l e

i n f o

Article history: Received 6 August 2012 Received in revised form 28 December 2012 Accepted 15 January 2013 Available online 31 January 2013 Keywords: KCC4 N-linked glycosylation Glycan complexity Membrane expression Tumor formation

a b s t r a c t Glycosylation plays a role in regulating many biological activities, including protein folding and cell surface expression of biomolecules. However, the importance of glycosylation for KCC4 function has not previously been demonstrated. Site-directed mutagenesis was performed on the four putative extracellular N-linked glycosylation sites of KCC4 to determine the role of these sites in KCC4 half-life, cell surface expression, and transporter activity, as well as in KCC4-dependent tumor formation. We showed that triple (N312/ 331/344/Q) and quadruple (N312/331/344/360/Q) mutations of N-linked glycosylation sites disrupt the N-linked glycosylation of KCC4, resulting in the accumulation of KCC4, predominantly in the endoplasmic reticulum (ER) and not at the cell surface. Further investigation indicated that mutations of the central two (N331/344/Q) N-linked glycosylation sites inhibit the membrane trafficking of KCC4. Our data suggest that the glycan moieties at the N331 and N344 sites were Endo H-resistant, complex-form structures, and that the N312 and N360 sites were Endo H-sensitive, high mannose-containing structures. Under hypotonic stress conditions, the ability to adapt to changes in intracellular chloride ion concentrations and RVD (regulatory volume decrease) activities were less efficient in cells containing the deglycosylated form of KCC4 that were not expressed at the cell surface. Deglycosylated forms of KCC4 also demonstrated decreased tumor formation and lung colonization in mouse xenografts. The difference in glycan complexity may account for the differential impact of each branch on the biological effects of KCC4. We propose that glycosylation is essential for the surface expression, stabilization, and bioactivity of KCC4. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Glycosylation, which involves the enzyme-controlled addition of a sugar moiety to a protein, is the most common type of posttranslational modification. The sugar chain can link either to the amino (\NH2) group of asparagine (referred to as “N-linked” glycosylation) or to the \OH group of serine or threonine (“O-linked” glycosylation). Glycosylation is involved in many important biological processes [1,2], including cell–cell adhesion, trafficking [3], and protein stability, but the role of glycosylation in these activities is not well understood. N-linked glycosylation occurs in the endoplasmic reticulum (ER) with the addition of high-mannose sugars, and these glycan moieties are

⁎ Corresponding author at: Department of Obstetrics & Gynecology, College of Medicine, National Cheng Kung University and Hospital, Tainan 70403, Taiwan. Tel.: +886 6 2353535 x5608; fax: +886 6 2766185. E-mail address: [email protected] (C.-Y. Chou). 0167-4889/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bbamcr.2013.01.018

further processed into complex-form glycans in the Golgi [4]. This process increases the apical surface expression and stability [5,6] and alters the function [7] of the target protein. N-linked glycoproteins are also involved in the progression of certain cancers [7]. Moreover, N-linked glycosylation inhibitors have been shown to have potential in the development of novel therapeutic approaches for cancer therapy [8]. Originally described in red blood cells, K +Cl− cotransporters (KCCs) belong to the SLC12 family and act as electroneutral symporters of K + and Cl− ions across the plasma membrane [9]. K +Cl− cotransport, which can be activated by cell swelling, is necessary for the regulation of cell volume, the transepithelial transport of ions, and the maintenance of intracellular Cl− concentration [10]. Four isoforms of KCC have been identified: KCC1, KCC2, KCC3, and KCC4. In addition to being major determinants of osmotic homeostasis, KCCs are also emerging players in the field of tumor biology [11,12]. The malignant transformation of uterine cervical epithelial cells has been associated with the differential expression of volume-sensitive KCC isoforms [13]. A loss-of-function KCC mutant in cervical cancer cells was found to inhibit cell growth and decrease the activities of

1134

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

cell cycle regulators and matrix metalloproteinases [11]. Additionally, insulin-like growth factor-1 (IGF-1) upregulates KCC3 and KCC4, which are differentially required for cancer cell proliferation and invasiveness [14]. Overexpression of KCC3 downregulates E-cadherin/ β-catenin complex formation by inhibiting the transcription of the E-cadherin gene and accelerating the proteasome-dependent degradation of the β-catenin protein, promoting the epithelial–mesenchymal transition [15]. Moreover, epidermal growth factor (EGF) and IGF-1 stimulate the membrane recruitment of KCC4 at lamellipodia through the myosin Va-actin trafficking route [16]. KCC4 functions as a membrane scaffold for the assembly of signaling complexes via the association with the actin-binding protein ezrin [16]. Molecular studies of surgical specimens show that greater KCC4 expression in tumor tissues is associated with a higher risk of tumor metastasis and cancer recurrence [16]. Taken together, these findings suggest that KCCs (KCC3 and KCC4 in particular) can be used as biomarkers to predict outcomes in cancer patients. The four KCC isoforms share a common protein structure with a central core of twelve hydrophobic transmembrane domains [17,18]. Protein structure prediction analysis suggests that the 4 N-linked glycosylation sites are conserved and are found in the outer loop between the 5th and 6th transmembrane segments of KCC4. Our previous findings demonstrated that motor protein-dependent membrane trafficking of KCC4 is important for cancer cell invasion [16]. This study sought to investigate the role of N-linked glycosylation in the localization, protein stability, and activity of KCC4.

2. Materials and methods 2.1. Cell culture and reagents The AD-293 cell line is a more adherent derivative of HEK293 cells (Stratagene, La Jolla, CA), and the AS-2 cell line is a lung adenocarcinoma cell line [19]. AS-2 stable clones were selected with blasticidin following transfection with various KCC4 constructs. Cell growth was determined by thiazolyl blue tetrazolium bromide (MTT) (Sigma, St Louis, MO) assay in 96-well plates. To analyze protein stability, cycloheximide (20 μg/ml) was added to AD-293 cells that had been transfected with different KCC4 constructs and cell lysates were collected at the indicated time points. For the analysis of KCC4 degradation, the proteasome inhibitor MG132 (Sigma), the lysosome inhibitors NH4Cl and leupeptin (Sigma), and a protease inhibitor cocktail tablet (Roche, Grenzacherstrasse, Basel) were incubated with cells for 6 h, and the lysates were then collected.

2.2. Plasmid constructs Full-length mouse KCC4 harboring an exogenous V5 (referred to here as KCC4wt) was constructed by cloning pXT7–mKCC4 into the pCDNA6/V5-His vector (Invitrogen, Grand Island, NY). Three mouse KCC4 N-linked glycosylation mutant constructs (kindly provided by Dr. David B. Mount) were obtained, in which asparagine (N) was replaced with the amino acid glutamine (Q) to block N-linked glycosylation of mouse KCC4. These mutant constructs included the following: pGEMHE-double (N312/331/Q), pGEMHE-triple (N312/331/344/Q), and pGEMHE-quadruple (N312/331/344/360/Q). These 3 constructs were subcloned from the pGEMHE oocyte system into a mammalian expression system (pCDNA6/V5-His) with a V5 tag to differentiate endogenous KCC4 from exogenous KCC4 mutants (referred to here as M II, M III, and M IV). All sequences of newly constructed clones used in this study were confirmed by sequencing. Arrest-In reagent (Thermo Scientific, Rockford, IL) or Lipofectamine 2000 reagent (Invitrogen) was used for plasmid DNA transfections, which were carried out according to the manufacturers' protocols.

2.3. Site-directed mutagenesis A QuikChange®Site-Directed Mutagenesis kit (Stratagene) was used to generate the N344/Q and N360/Q point-mutant constructs, using the KCC4wt construct containing the V5 tag as a template. The primers used were as follows (only forward primers are listed): 331 mutant primer: 5′'-GATGCAGGTTGTCAGCCAGGGTACAGTGACCACTG-3′; 344 mutant primer, 5′-GGCGCCTCTTCTGCCAGGGCTCCAGCTTGGG-3′; and 360 mutant primer, 5′-TGAGTACTTTGCACAGAACCAGGTTACTGAGATACAGGG CA-3′. The N344/360/Q mutation was generated using the 360 mutant primer and the N344/Q mutant construct as a template. N312/331/ 360/Q was generated using the 360 mutant primer and the M II construct as a template. N331/344/Q was generated using the 331 mutant primer and the N344/Q construct as a template. Procedures were conducted according to the manufacturer's protocol. In brief, the template, primers, and dNTPs were mixed with PfuTurboDNA polymerase and subjected to PCR using the following protocol: 95 °C for 30 s (1 cycle), 95 °C for 30 s, 55 °C for 1 min, 68 °C for 9.5 min (18 cycles), and finally, 68 °C for 9.5 min (1 cycle). DpnI was added, and the reactions were incubated in a 37 °C water bath for 1.5 h. Positive clones were verified by sequencing. 2.4. Glycosidase assay The cells were lysed in radioimmunoprecipitation (RIPA) buffer, and 20 μg of protein was used in each reaction. The protein was then mixed with glycoprotein denaturing buffer (New England Biolabs, Ipswich, MA) and incubated at 100 °C for 10 min. For endoglycosidase H and peptide N-glycosidase F (New England Biolabs) digestions, the appropriate buffer was added to the whole cell lysates according to the manufacturer's instructions, and the reactions were incubated in a 37 °C water bath for 1.5 h. 2.5. Western blotting The cells were lysed with either RIPA or sample buffer and then harvested using a cell lifter (Costar, New York, NY). Protein concentrations were determined using the Bio-Rad protein assay. Proteins were then separated by 6% or 12% SDS-PAGE and transferred to polyvinylidene difluoride membranes. Antibodies used in this study included the following: anti-V5 antibody (Invitrogen), anti-mKCC4 antibody (kindly provided by Dr. David B. Mount), and anti-beta actin antibody (Sigma). 2.6. Immunofluorescence The cells were seeded and transfected in chamber slides (Nalge Nunc, Rochester, NY) for 48 h. The cells were then fixed in 3.7% paraformaldehyde, washed, and incubated in 0.05% Triton-X100 for 15 min. Next, the cells were blocked with SuperBlock solution (Thermo Scientific) for 1 h at room temperature. Primary antibodies were incubated with cells at 4 °C overnight followed by TBST wash, and the appropriate secondary antibody (anti-mouse Alexa-488, anti-mouse Alexa-594 or anti-rabbit Alexa-594 (Molecular Probes, Eugene, OR)) and Hoechst 33342 dye were incubated with the cells for 1 h in the dark. Slides were then washed and fixed with GG1 gel (Sigma). Confocal imaging was conducted using a FV1000 microscope (Olympus, Tokyo, Japan). Antibodies used for the immunofluorescence imaging included the following: anti-V5 antibody (Invitrogen), anti-calnexin antibody (Abcam, San Francisco, CA), anti-mannosidase II antibody (Chemicon, Darmstadt, Germany), and anti-NKCC1 antibody (Santa Cruz, Delaware Avenue, CA). 2.7. Cell surface botinylation Cell membrane proteins were isolated by cell surface protein isolation kit (Pierce Biotechnology, Rockford, IL). Briefly, cells were grown

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

1135

of labeled proteins, cell lysates were treated with NeutrAvidin Agarose. Proteins were then washed and eluted in the column by centrifugation according to the manufacturer's instructions.

to 90% confluence and washed with PBS then incubated with SulfoNHS-SS-Biotin at 4 °C for 30 min. Biotinylation was inhibited by treating cells with Quenching solution prior to cell lysis. For isolation

331 344 360 312

A

Plasma membrane NH2

COOH 331 344 360 312

KCC4wt

344 360

N312/331/Q (M II)

360

N312/331/344/Q (M III) N312/331/344/360/Q (M IV)

B

KCC4wt M II

M III

M IV

KDa 130

95

C

KCC4 wt PNGase F

-

+

M II

Endo H

-

PNGase F

+

-

+

Endo H

-

+

KDa 130

130

95

95 M III Endo H PNGase F

-

+

-

M IV Endo H PNGase F

+

-

130

130

95

95

+

-

+

Fig. 1. Serial mutations at KCC4 N-linked glycosylation sites alter the molecular pattern of KCC4. (A) Topology model of KCC4. Four putative N-linked glycosylation sites in the outer, largest loop between the 5th and 6th transmembrane segments are boxed. Schematic representations of the different N-linked mutations, including M II, M III, and M IV. (B) AD-293 cells were transiently transfected with the constructs listed in (A), and whole cell lysates were analyzed by SDS-PAGE and immunoblotting with an anti-V5 antibody. (C) AD-293 cells were transiently transfected with the constructs listed in (A). Whole cell lysates were incubated in the presence (+) or absence (−) of PNGase F or Endo H and then immunoblotted with anti-V5 antibody. All molecular forms are sensitive to PNGase F, except for the M IV mutant, which has no N-linked glycosylation sites. The 130-kDa molecular form is insensitive to Endo H, as shown in KCC4wt and M II.

1136

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

V5

NKCC1

Hoechst

Merge

A

KCC4 wt 10 µm

M II 10 µm

M III 10 µm

M IV 10 µm

V5

B

Calnexin

Hoechst

Merge 10 µm

KCC4 wt

10 µm

M II

M III 10 µm

10 µm

M IV

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

V5

Man II

1137

Merge

Hoechst

C KCC4 wt 10 µm

M II 10 µm

M III 10 µm

M IV 10 µm

D Pull-down

KCC4 wt

M III

M IV

- +

- +

- -+

KDa V5

130

130

130

95

95

95

actin Fig. 2. KCC4wt and M II colocalize with the cell membrane, whereas M III and M IV are retained in the ER. Representative confocal images of AD-293 cells transiently transfected with wild-type and mutated KCC4 proteins, detected by immunostaining for the V5 tag. To verify cell membrane or ER localization, constructs were (A) costained with an endogenous membrane marker NKCC1 and (B) the ER marker calnexin, or (C) Golgi marker mannosidase II. Staining of KCC4wt and M II extended to the cell membrane border, whereas M III and M IV exhibited perinuclear staining and colocalization with calnexin. The mannosidase II stains were confined in the high mannose-containing mutants and deglycosylated M IV. (Man II: Mannosidase II) (D) AD-293 cells were stably cloned with KCC4wt, M III, and M IV. Cells were treated with biotin to label surface protein. Total cell lysate and NeutrAvidin Agarose pull-down lysate were immunoblotted with an anti-V5 antibody. Antibody against actin was used as the control.

2.8. Chloride measurement Intracellular chloride concentration was measured under hypotonic stress by cotransfecting the fluorescence resonance energy transfer (FRET) plasmid Clomeleon with mock, KCC4wt or M IV plasmids into AD-293 cells. Clomeleon (kindly provided by Dr. Thomas Kuner) encodes a CFP–YFP fusion protein, for which the fluorescent emission peak of the CFP overlaps the excitation peak of the YFP. The emission of YFP is quenched by chloride ions; therefore, the efficiency of FRET between these chromophores inversely correlates with chloride concentration [20]. Forty-eight hours following the cotransfections, the cells

were treated with an isotonic buffer containing chloride and a chloride-free hypotonic buffer for 10 min. Cells treated with chloridefree isotonic buffer and chloride-free hypotonic buffer represented the maximal emission FRET ratio. The cells were then fixed with 3.7% paraformaldehyde and mounted with GG1 gel. The images were captured, and FRET ratios were determined using a FV1000 confocal microscope (Olympus). Isotonic buffer with Cl− contained 150 mM NaCl, 5 mM KCl, 1 mM CaCl2, 1 mM MgCl2, and 10 mM HEPES (pH = 7.4, titrated by Tris). Isotonic buffer without Cl− contained 150 mM Na-gluconate, 5 mM K-gluconate, 5 mM Ca-gluconate, 1 mM Mg-gluconate, and 10 mM HEPES (pH =7.4, titrated by Tris). Finally, hypotonic buffer

1138

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

A

331 312

360

N344/Q 331 344 312

N360/Q

331 312

N344/360/Q 360

312

N331/344/Q

344

N312/331/360/Q

B

KDa 130

95

C

N344/Q PNGase F Endo H

N360/Q Endo H PNGase F

- + - +

- + - +

KDa 130

130

95

95 N344/360/Q PNGase F Endo H

-

+ -

N331/344/Q Endo H PNGase F

+

-

130

130

95

95 N312/331/360/Q Endo H PNGase F

-

130

95

+ -

+

+ -

+

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

without Cl− contained 80 mM Na-gluconate, 5 mM K-gluconate, 5 mM Ca-gluconate, 1 mM Mg-gluconate, and 10 mM HEPES (pH =7.4, titrated by Tris). 2.9. Regulatory volume decrease Isotonic buffer (osmolarity = 300 mOsmol/l) contained 100 mM NaCl, 5 mM KCl, 1 mM MgCl2, 1.5 mM CaCl2, 10 mM Glucose, 10 mM HEPES, 70 mM Mannitol, pH= 7.4. Hypotonic buffer (osmolarity= 230 mOsmol/l) contained 100 mM NaCl, 5 mM KCl, 1 mM MgCl2, 1.5 mM CaCl2, 10 mM Glucose, 10 mM HEPES, pH= 7.4. The cells were resuspended in the isotonic buffer, and FSC of the cells was determined by FACS using the FACSCalibur (BD, Franklin Lakes, NJ). To monitor volume change during osmotic stress, cells that had been resuspended in the hypotonic buffer were incubated in a separate tube and analyzed by FACS to determine the FSC of the cells at the indicated time point. A total of 104 cells were analyzed using identical laser intensities. The data were analyzed using histograms, with gates set at 50% of the population of each isotonic solution to determine the standard condition. The median intensity shift for each population compared with its standard condition was used to determine the changes in cell volume (hypotonic condition minus isotonic condition).

1139

(Fig. 1B). The lower band of KCC4wt, M II and M III may represent the core-glycosylated form of KCC4, located in the ER, whereas the upper band likely represents the complex glycosylated form of KCC4, located in the plasma membrane (Fig. 1B) [21]. Glycosidase was then used to further distinguish the different glycan structures of these mutants. Peptide N-glycosidase F (PNGase F) cleaves all N-linked structures from asparagine, whereas endoglycosidase H (Endo H) cleaves only the high-mannose structure or hybrid structures of N-linked branches (herein referred to as high-mannose containing structures). We found that KCC4wt, M II and M III were sensitive to PNGase F, as evidenced by the downward shift of the upper band following PNGase F treatment. Under Endo H digestion, a slight downward shift of the lower band of KCC4wt and M II was detected, whereas the upper bands for each of these mutants remained unchanged. Endo H treatment also caused the cleavage of M III. M IV to be insensitive to both PNGase F and Endo H, suggesting that M IV did not contain N-glycans (Fig. 1C). Taken together, these results suggest that KCC4wt and M II contain complex-form and high-mannose N-linked structures, whereas M III contains only high mannose-containing structures.

3.2. M III and M IV mutants are not expressed at the cell surface 2.10. Colony formation and animal experiments Lower agar (0.6%) was cooled and solidified in 6-well plates at room temperature. For upper agar, 0.3% agar containing 5 × 104 cells was added to the base agar and incubated in a 37 °C incubator. Colonies >2 mm were counted under a microscope. AS-2 cells were injected into NOD/SCID mice via subcutaneous or intravenous route. Mice were maintained in the NCKU Laboratory Animal Center in a day/ night cycle facility and were sacrificed at indicated time points. All animal studies were approved by the Institutional Animal Care and Use Committee at National Cheng Kung University. 2.11. Statistical analysis Values in this study are represented as the means ± standard deviation (s.d.). Student's t test was used for statistical analyses. 3. Results 3.1. Mutations at 4 N-linked glycosylation sites change the glycoprotein structure of KCC4 KCC4 is a glycoprotein with 12 transmembrane domains and 4 putative N-linked glycosylation sites in its largest outer loop, at amino acid residues 312, 331, 344, and 360 (Fig. 1A). To explore how glycosylation modulates KCC4 function, serial mutant constructs of KCC4's N-linked glycosylation sites were generated by replacing each glycosylation site Asn (N) with Gln (Q). The mutant constructs of glycosylation included double (N312/331/Q), triple (N312/331/344/Q), and quadruple (N312/331/344/360/Q) mutants (Fig. 1A). For clarity, the double, triple and quadruple mutants are herein referred to as M II, M III, and M IV, respectively. Following transfection of AD-293 cells with the KCC4 mutant constructs, immunoblot analysis on whole cell lysates was performed. In cells transfected with KCC4wt, or the M II mutant, two protein bands were visible: a lower band, near 100 kDa, and an upper band, near 130 kDa. In contrast, only a lower band was observed in cells transfected with the M III or M IV mutant

Glycosylation is known to be important for protein transport. The differences in the molecular patterns detected in the immunoblots and the glycan structures among these KCC4 mutants led us to investigate whether altered N-linked glycosylation impairs cell surface expression of KCC4. AD-293 cells transfected with the mutant constructs were first analyzed using confocal immunofluorescence microscopy to examine KCC4 localization. The majority of the KCC4wt- and M II-transfected cells demonstrated KCC4 immunostaining at the cell membrane, whereas immunostaining of KCC4 in the M III- and M IV-transfected cells occurred largely in the juxta-nuclear area. To confirm the localization of KCC4 in these cell lines, we used an endogenous membrane marker NKCC1 to verify membrane expression. As demonstrated in Fig. 2A, V5 staining in KCC4wt and M II colocalized with NKCC1, whereas V5 staining in M III and M IV did not. To further confirm these results, pAcGFP1-Mem (which encodes the N-terminus of neuromodulin fused to GFP and demonstrates strong labeling at the cell membrane) was transfected into AD-293 cells, and similar results were obtained (Supplementary Fig. 1). pAcGFP1-Endo, which encodes a RhoB gene fused to GFP, was transfected into the AD-293 cells; RhoB is largely expressed in late endosomes, the cell membrane, and nuclear margins. Colocalization of KCC4-V5 and RhoB-GFP was observed in KCC4wt and M II at the cell surface, whereas cells transfected with M III and M IV displayed a perinuclear, ER-like staining pattern (Supplementary Fig. 2). Because a perinuclear pattern is a hallmark of ER-resident proteins, we next tested whether KCC4 in M III and M IV colocalized with the ER marker calnexin. KCC4 and calnexin colocalized (Fig. 2B), indicating that KCC4 in M III and M IV failed to exit in the ER and in turn was not expressed at the cell surface. A Golgi marker, mannosidase II, was used and further confirmed that M III and M IV were arrested in the ER and Golgi (Fig. 2C). In a surface biotinylation experiment, only the upper band of KCC4wt was labeled with biotin, as shown in the NeutrAvidin Agarose pull-down lane, but not the lower band (Fig. 2D). This result further suggested that the upper band is the complex-form structure (Fig. 1C) that can reach the cell membrane. No biotin-labeled protein of M III and M IV was observed, which indicated that M III and M IV were not expressed in the cell membrane.

Fig. 3. Combinations of KCC4 N-linked glycosylation mutations. (A) Schematic model of different N-linked glycosylation mutations generated by site-directed mutagenesis, including N344/Q, N360/Q, N344/360/Q, N331/344/Q, and N312/331/360/Q. (B) AD-293 cells were transiently transfected with constructs listed in (A), and whole cell lysates were analyzed by SDS-PAGE and immunoblotting with an anti-V5 antibody or by (C) treating the cells with glycosidase, followed by immunoblotting with an anti-V5 antibody. Total cell lysates were incubated with (+) or without (−) PNGase F or Endo H. All molecular forms are sensitive to PNGase F, whereas the N331/344/Q mutant is uniquely sensitive to Endo H.

1140

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

NKCC1

Hoechst

Merge

KCC4wt

V5

N344/Q

10 µm

N360/Q

10 µm

N344/360/Q

10 µm

N331/344/Q

10 µm

N312/331/360/Q

10 µm

10 µm

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

A KDa h 0 V5

KCC4 wt 4 8 12

0

M II 4 8 12

M III 4 8 12

0

M IV 4 8 12

1141

130

actin

h0 V5

130

actin

B Stability (/Control)

1.2 WT Complex WT High-mann M II Complex M II High-mann M III High-mann M IV

1.0 0.8 0.6 0.4 0.2

0

4

8

12

Time (h)

C

KCC4wt

MG132 NH4Cl PI KDa

M II

M III

-+- - -+- -- -- +- +- -- -- +- +-

M IV

-+- - -+- -- -- +- +- - - + - - - +

130

130

95

95

V5

actin Fig. 5. Glycosylation is important for KCC4 protein stability. (A) AD-293 cells were transiently transfected with KCC4wt, M II, M III, or M IV for 24 h. Cells were then treated with cycloheximide (20 mg/ml). At indicated time points post-treatment, cell lysates were harvested and immunoblotted with an anti-V5 antibody. (B) Densitometric analysis of protein stability in AD-293 cells transiently transfected with KCC4wt, M II, M III, or M IV and treated with cycloheximide (20 mg/ml). Complex-form glycan has a longer half-life than the high mannose-containing form. Data are expressed as percentage of complex-form or high mannose-containing form relative to untreated controls. (Complex: complex-form; High-mann: high-mannose form) (C) Following transfection with KCC4wt, M II, M III or M IV, AD-293 cells were treated with (+) or without (−) MG132 (20 μM), NH4Cl (10 μM) or protease inhibitors (1 mg/ml) for 6 h prior to cell lysis. The cell lysates were then immunoblotted with an anti-V5 antibody. KCC4 degradation occurred primarily through the proteasome pathway. (PI: Protease inhibitors).

3.3. The central two N-linked glycans of KCC4 are important for KCC4 membrane expression Because KCC4 in M III (but not M II) failed to localize to the cell membrane, we hypothesized that the N344 glycosylation site is important for KCC4 membrane expression. To investigate this hypothesis, we

performed site-directed mutagenesis to generate additional mutant constructs, as shown in Fig. 3A. Immunoblot analysis revealed that, as with the immunoblot patterns for KCC4wt and M II, the N344/Q and N360/Q mutants exhibited two protein bands (Fig. 3B). Glycosidase analysis further demonstrated that the major band of N344/Q and N360/Q were sensitive to PNGase F and insensitive to Endo H treatment

Fig. 4. Mutations at the central two N-linked glycosylation sites of KCC4 alter KCC4 cell membrane expression. Representative confocal images of AD-293 cells transiently transfected with wild-type and mutated KCC4 proteins, detected by immunostaining for the V5 tag. KCC4wt or mutant constructs were costained with an anti-V5 antibody and the membrane marker NKCC1 to verify cell membrane expression. Immunostaining of N344/Q and N360/Q was observed at the cell membrane border. N344/360/Q and N312/ 331/360/Q partially localized to the cell membrane marker, whereas N331/344/Q largely localized to the intracellular compartment. Colocalization of proteins is denoted by arrowheads.

1142

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

(Fig. 3C), indicating that N344/Q and N360/Q contain complex-form glycan structures. Immunofluorescence analysis showed that KCC4 in the N344/Q and N360/Q mutants was expressed at the cell membrane and colocalized with NKCC1 (Fig. 4). This finding suggests that it is not sufficient for a single N344/Q or N360/Q mutations to impair membrane expression of KCC4. To further investigate this hypothesis, N344/ 360/Q, N331/344/Q, and N312/331/360/Q mutants were constructed. Immunoblot analysis revealed only a lower protein band in N331/ 344/Q. Intriguingly, the N344/360/Q and N312/331/360/Q mutants displayed two protein bands, whose positions in the immunoblots were substantially lower and more variable than those observed for the KCC4wt and M II mutants (Fig. 3B). Glycosidase treatment demonstrated that N331/344/Q (harboring the central two mutations) was sensitive to PNGase F and Endo H, suggesting the presence of a high

A

mannose-containing structure (Fig. 3B). N344/360/Q and N312/331/ 360/Q mutants were sensitive to both PNGase F and Endo H, with the upper band being insensitive to Endo H, indicating that KCC4 in these two mutants contained a complex-form structure, albeit a structure that was different from that of KCC4wt (Fig. 3C). Immunofluorescence microscopy demonstrated that V5 staining in N331/344/Q did not colocalize with NKCC1, indicating that the cell surface expression of this mutant was impaired (Fig. 4). Further analysis with calnexin further indicated that N331/344/Q was retained in the ER (Supplementary Fig. 5B). No biotin-labeled protein of N331/344/Q was observed, which indicated that N331/344/Q was not expressed in the cell membrane (Supplementary Fig. 5C). In N344/360/Q and N312/331/360/Q mutants, only a few scattered colocalizations of KCC4 with NKCC1 were observed (Fig. 4). Similar results were obtained using pAcGFP1-Mem for cell

B 50

FRET Ratio

2.4

*** ***

**

2.2 2.0 1.8 1.6

Iso Hypo

Volume Change (%)

2.6

AS-2 Mock AS-2 M IV-1 AS-2 M IV-2

40 30 20

** *

10 0 0

0.5 1

2

4

6

8

10

Time (Min)

10

0.6

8

0.5

MTT (OD 590)

Cl-Change Ratio (%)

C

6 4 2

AS-2 Mock AS-2 M IV-1 AS-2 M IV-2

0.4

*** ***

0.3 0.2 0.1

0

0.0 1

2

3

4

Day

120

Colony Number (%)

Relative

D

100 80

**

60 40

***

20 0

Fig. 6. M IV exhibited a lower capacity for fluctuations in intracellular chloride ion concentration and cell volume regulation. (A) Clomeleon was transiently cotransfected with mock, KCC4wt, or M IV plasmids. FRET ratios were higher following hypotonic stress in each group (upper panel). The degree of intracellular Cl− change was less in M IV cells (lower panel). Gluconate, a chloride-free buffer, represented the maximal FRET ratio. (B) RVD ability under hypotonic stress is less efficient in AS-2 cells transfected with the M IV construct (n = 7–9). (C) AS-2 cells transfected with the M IV construct demonstrated reduced MTT activities and (D) were less efficient at colony formation compared to mock-transfected cells.

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

membrane imaging (Supplementary Fig. 3). Collectively, our data suggest that the central two N-linked glycosylation sites are critical for KCC4 membrane expression.

3.4. The high-mannose form of N-glycan is degraded through the proteasome pathway Glycan structures play an important role in protein stability. We were interested in determining whether KCC4 protein stability is affected by the presence of its N-linked glycan motifs. Cycloheximide, an inhibitor of protein synthesis, was used to test the rate of KCC4 protein degradation. We found that the complex-forms of KCC4wt and M II (upper band of M II, Fig. 5A) have a longer half-life than the high mannose-containing forms in KCC4wt, M II and M III, indicating that complex-form glycans augment the stability of KCC4 (Fig. 5A, B). Because protein stability is reduced in the high-mannose form, we next investigated the major route of KCC4 degradation, namely, the proteasome, lysosome, and protease pathways of protein degradation, using inhibitors specific for each degradation pathway. Fig. 5C

A

Mock

1143

shows that, following treatment with MG132 (a specific inhibitor of the proteasome pathway), more high-mannose forms were present, accompanied by the attenuation of the complex-form. No significant change was observed using the lysosome inhibitors (NH4Cl, Fig. 5C; Leupeptin, Supplementary Fig. 4) or protease inhibitor (Fig. 5C). Therefore, the proteasome pathway is the major protein degradation pathway involved in the degradation of KCC4, particularly in regard to the high-mannose form.

3.5. M IV and N331/344/Q mutants suppress KCC4-mediated RVD and tumor formation Our previous studies demonstrated that membrane trafficking and cell surface expression of KCC4 are important for cancer cell invasion [16]. Because M III and M IV were not expressed at the cell surface, we postulated that KCC4 function could also be affected by these mutations. First, M IV was used for the functional study. To test whether glycosylation is important for KCC4 cotransporter function, Clomeleon and mock, KCC4wt or M IV plasmids were transiently transfected into

M IV-1

M IV-2

Tumor Volume (mm3)

700 600 AS-2 Mock AS-2 M IV-1 AS-2 M IV-2

500 400 300 200

** ***

100 0 10

12

14

16

17

Day

B Mock

M IV-2

Nodule Number

250 200 150

1 cm

100 50 0

*

*

M IV-1

Fig. 7. Tumor growth and lung colonization in SCID mice. (A) Tumor growth curves of SCID mice subcutaneously inoculated with AS-2 mock, AS-2 M IV-1 or AS-2 M IV-2 cells (n = 6). Mice were sacrificed 17 days following tumor transplantation. (B) Lung nodules in SCID mice intravenously inoculated with the AS-2 mock, AS-2 M IV-1 or AS-2 M IV-2 cells (n = 4). Mice were sacrificed at 21 days following tumor injection. Representative tumor (A) and lung (B) xenografts are shown. Lung samples were immersed in Bouin's solution. Arrowheads indicate white lung nodules. (C) Immunohistochemistry of subcutaneous tumor sections. Arrowheads indicate perinuclear patterns in M IV tumors, a finding that is in contrast to membrane-localized immunostaining in KCC4WT tumors (400×). (D) Immunoblot analysis of the V5 tag in subcutaneous tumors. Mock was used as negative control.

1144

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

C Mock

M IV-1

M IV-2

KCC4wt

D KDa 130 V5 95 actin

Fig. 7 (continued).

AD-293 cells, and the transfected cells were subsequently exposed to hypotonic stress for 10 min. The gluconate solution, which is chloridefree and therefore represents the maximal emission FRET ratio, served as the positive control. Under hypotonic stress conditions, the degree of intracellular chloride change was significantly reduced in cells transfected with M IV compared with that of the mock-transfected or KCC4wt-transfected cells (Fig. 6A). To examine the role of KCC4 glycosylation in cell volume regulation, we compared the regulatory volume decrease (RVD) process in mock and M IV mutant stably transfected lung cancer AS-2 cells, M IV-1 and M IV-2. Forward scatter (FSC) represents the cell volume under specific laser intensity, and the percent volume change was calculated using equivalent laser intensities. As shown in Fig. 6B, the typical volume response of AS-2 cells to hypotonic solution can be divided into three phases: (1) an initial, rapid cell swelling to reach a peak cell volume; (2) subsequent cell shrinkage; (3) and finally, a more gradual decrease in cell volume that eventually plateaus. AS-2 cells expressing the M IV mutation exhibited an increase in initial cell swelling, attenuated cell shrinkage, and an inhibition in the characteristic gradual decrease in cell volume, indicating that cells harboring

the M IV mutation have a lower capacity for cell volume regulation compared to mock-transfected cells. We next examined whether deglycosylated KCC4 could affect cell biology. MTT assays and colony formation experiments demonstrated that M IV-expressing AS-2 cells were less viable and less able to form colonies compared with mock-transfected cells (Fig. 6C, D). To test whether manipulation of KCC4 glycosylation alters tumor growth and lung colonization, we inoculated NOD/SCID mice subcutaneously or intravenously with KCC4 mock or M IV-expressing AS-2 cells. Rapid tumor growth in terms of subcutaneous tumor (Fig. 7A) or lung tumor nodules (Fig. 7B) was obvious in the groups inoculated with the KCC4 mock-expressing AS-2 cancer cells. The M IV glycosylation mutation significantly reduced the rate of tumor formation and decreased the tumor size in the dermis and the number of tumor nodules in the lungs. Consistent with the in vitro immunofluorescence study, the histology of subcutaneous tumors resulting from inoculation of M IV mutants exhibited predominantly cytoplasmic or perinuclear staining of KCC4-V5, which was in contrast to the membrane staining in the KCC4wt (Fig. 7C). Immunoblots further confirmed the presence of the

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

A Volume Change (%)

50

AS-2 Mock AS-2 N360/Q AS-2 CM AS-2 M IV-1

40 30 20

**

10 0 0 0.5

1

2

4

6

8 10

Time (Min)

B

Mock

N360/Q

M IV -1

CM

1 cm

Tumor Volume (mm3)

1200 AS-2 Mock AS-2 N360/Q AS-2 CM AS-2 M IV-1

1000 800 600 400 200 0

** 10

11

12

13

14

Day Fig. 8. Mutation of the central two glycan branches reduced the RVD ability and tumor formation. (A) AS-2 cells transfected with N360/Q construct did not abrogate the RVD activities of KCC4. Similar to AS-2 M IV-1 cells, AS-2 cells transfected with central mutation (central mutation: CM, N331/344/Q) construct demonstrated reduced RVD activities under hypotonic stress compared with AS-2 mock cells (n = 4). (B) Tumor growth curves of SCID mice subcutaneously injected with AS-2 mock, AS-2 N360/Q, AS-2 CM, and AS-2 M IV-1 cells (n = 6–8). Mice were sacrificed 15 days following tumor transplantation. Representative tumor xenografts are shown.

M IV mutant in the tumors (Fig. 7D). These results suggest that glycosylation plays an important role in tumor formation and lung colonization. To further investigate whether delivery of KCC4 at the cell surface affects its bioactivities, N360/Q and N331/344/Q (central mutation: CM) mutants were used in this study. These two constructs were stably transfected in the AS-2 cells. The abilities to adapt to osmotic challenge and tumor formation in N360/Q-expressing AS-2 cells were not affected, as compared with the mock-containing AS-2 cells. In contrast, AS-2 cells expressing the central mutation exhibited significantly less efficient RVD activities and showed a reduced rate and size of tumor formation, which resembled the impaired KCC4 bioactivities in AS-2 M IV-1 cells (Fig. 8A, B). 4. Discussion We investigated the importance of the four N-linked glycosylation sites in the large extracellular loop between the 5th and 6th transmembrane domains of KCC4 in cell membrane expression and protein stability as well as in various functions of KCC4, including cell volume regulation, cell proliferation, tumor growth, and tumor progression. The principal finding of our study was that the localization of KCC4 to the cell surface depends on the central two N-linked glycosylation sites located at amino acid residues 331 and 344 of KCC4. Our results also suggest that glycan structure (i.e., the presence of complex-form glycans) is the major determinant of the cell surface expression of KCC4, and that surface expression of KCC4 plays an important role in its bioactivities. Additionally, we found that N-linked glycosylation

1145

plays a role in maintaining protein stability and in regulating KCC4-mediated functions, such as the regulation of cell volume, anchorage-independent cell growth, tumor formation, and lung colonization by tumors. It would be interesting to investigate whether N-linked glycosylation plays similar roles in other KCCs that share a common protein structure with KCC4. Many previous studies have demonstrated that complex-form glycans are mature and functional proteins, whereas the high-mannose form is core-glycosylated and is retained in the ER [3,5,21]. The functional differences between complex and high-mannose structures are rooted in the structural differences (such as the accessibility of amphiphilic or hydrophobic domains) that facilitate or inhibit interactions with other proteins; for example, high-mannose N-glycans can promote protein folding, whereas complex-form N-glycans contribute to protein stability [22]. Based on our results, which were obtained using a combination of approaches, we propose that glycosylation sites N312 and N360 are high mannose-containing structures, and that N331 and N344 are complex-form structures (Supplementary Fig. 5A). Moreover, we suggest that N344 contains a larger or more complex glycan branch than does N331, because the position of the N312/331/360/ Q protein band in the immunoblot analysis was higher than the band from N344/360/Q (Fig. 3B). In both N344/360/Q and N312/331/360/Q mutants, small scattered expression of KCC4 can be detected in the cell membrane by confocal imaging. As a result, we suggest that N344 and N331 residues are critical for cell surface expression, and mutation of either residue alone is not sufficient to completely abrogate the trafficking of KCC4 to the cell membrane. We also found that M III is sensitive to Endo H treatment, suggesting that residue 360 harbors a high mannose-containing structure. Mutation at the central two sites (N331/344/Q) causes KCC4 to be retained in the ER and renders KCC4 sensitive to both PNGase F and Endo H, suggesting the existence of the high-mannose form structures. This finding also suggests that the glycosylation site at residue 312 of KCC4 contains the high-mannose form structure. M II contains both complex-forms and high-mannose forms, which were identified by glycosidase treatment and immunoblotting (Fig. 1C). These findings, along with the results from the N344/Q, N360/Q, N344/360/Q, and N312/331/360/Q mutants, further suggest that N344/Q and N360/Q contain complex-form glycan structures. Collectively, our results demonstrate that the complex-form N-linked glycan is the fully glycosylated form of KCC4, which is expressed at the cell surface. In contrast, the high-mannose form is likely the less mature glycan that is retained in ER and may be more quickly degraded by the proteasome. N-linked glycosylation is involved in cell surface expression of potassium channels [6,23]. Disruption of N-linked glycosylation reduces the apical expression of proteins [6,24,25]. Our findings are consistent with these observations, and further demonstrate that the central two N-linked glycosylation sites of KCC4 are the major determinants of KCC4 cell membrane expression. The importance of the central two N-linked sites for KCC4 expression might result from structural differences in the protein near these residues or from differences in the accessibility of the N-glycans at their distinct locations in the protein. The process of converting a high-mannose form structure into a complexform glycan in the Golgi has been reported to improve the apical surface expression of glycoproteins [3,5,6]. Consistent with these findings, our results demonstrate that complex-form N-glycans (such as those contained in the N331 and N344 sites of KCC4) can reach the cell surface; indeed, the N331 and N344 mutants in this study were more likely to be retained in the cytoplasm, possibly due to being targeted by the ER quality control system. In contrast, high mannose-containing structures, such as the N312 and N360 sites, play less important roles in cell surface expression of KCC4. The KCC1-ΔN117 mutant construct, which harbors a truncation of the first 117 amino acids in the N-terminus of KCC1, but retains an intact N-linked glycan, is a loss-of-function variant of pan-KCCs [26]. KCC1-ΔN117 can be expressed at the cell membrane in the oocyte

1146

T.-Y. Weng et al. / Biochimica et Biophysica Acta 1833 (2013) 1133–1146

system [26], which indicates that the first 117 amino acids of KCC1 N-terminus are not required for its cell surface expression. However, mutation of the cysteine residues in the large outer loop does not impair membrane expression of KCC4 or a KCC2–KCC4 chimera [27]. This indicates that different regions of the KCC4 protein may be involved in different processes. Our previous studies have highlighted the importance of KCC4 in tumor development and progression. KCC4 functions as a membrane scaffold which facilitates the cytoskeletal reorganization required for cellular invasiveness [16]. KCC4-mediated cancer cell invasion and metastasis may be due to greater cell motility, adhesion and migration, as well as increased MMP activation, which is important for cancer invasiveness [28]. Furthermore, KCC4 plays a role as a plasma membrane scaffold, which may regulate the integrity of the cytoskeleton. The present study demonstrates the role of N-linked glycosylation in KCC4-regulated cell volume regulation and tumor biology, and provides further evidence that N-linked glycosylation is required for KCC4-regulated tumor behavior (such as tumor formation and lung colonization by cancer cells). Disruption of N-linked glycosylation in proteins, such as KCC4, disrupts proper cell volume regulation, and the activity of cancer cells. Additionally, deglycosylation may induce apoptosis in glioma and prostate cancer cells [29,30]. Additional studies are required to better understand the precise mechanisms of N-linked glycosylation in tumor biology with a view to possibly develope novel therapies using N-linked glycan inhibitors. In conclusion, our data reveal an important role for the central two N-linked glycosylation sites in cell membrane expression and activity of KCC4. Furthermore, our study demonstrates strong relationships of N-linked glycosylation with protein stability and osmotic homeostasis, and it highlights the important role of N-linked glycosylation in KCC4-mediated tumor behavior. These findings offer additional insight into the glycan structure and function of KCCs, and potentially suggest that future cancer therapies involving inhibition of KCC4 recruitment to the cell membrane or inhibition of N-linked glycosylation in KCC4-abundant cancer cells could be developed. Abbreviations The following abbreviations are used: KCC, K +Cl − cotransporter; RVD, regulatory volume decrease; KCC4wt, wide-type KCC4; peptide N-glycosidase F, PNGase F; endoglycosidase H, Endo H. Acknowledgements We thank Dr. David B. Mount for providing the mouse KCC4 mutant constructs (pGMEHE) and mouse KCC4 antibody. The FRET plasmid Clomeleon was kindly provided by Professor Thomas Kuner (Institute for Anatomy and Cell Biology, University of Heidelberg, Germany). This research was supported by the established Center of Excellence for Cancer Research in Taiwan (DOH101-TD-C-111-003), and the Department of Health, Executive Yuan, Taiwan. Appendix A. Supplementary data Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.bbamcr.2013.01.018. References [1] R.S. Haltiwanger, J.B. Lowe, Role of glycosylation in development, Annu. Rev. Biochem. 73 (2004) 491–537. [2] M.M. Fuster, J.D. Esko, The sweet and sour of cancer: glycans as novel therapeutic targets, Nat. Rev. Cancer 5 (2005) 526–542.

[3] O. Vagin, J.A. Kraut, G. Sachs, Role of N-glycosylation in trafficking of apical membrane proteins in epithelia, Am. J. Physiol. Renal Physiol. 296 (2009) F459–F469. [4] O. Vagin, S. Turdikulova, G. Sachs, The H,K-ATPase beta subunit as a model to study the role of N-glycosylation in membrane trafficking and apical sorting, J. Biol. Chem. 279 (2004) 39026–39034. [5] D. Boassa, C. Ambrosi, F. Qiu, G. Dahl, G. Gaietta, G. Sosinsky, Pannexin1 channels contain a glycosylation site that targets the hexamer to the plasma membrane, J. Biol. Chem. 282 (2007) 31733–31743. [6] R. Khanna, M.P. Myers, M. Laine, D.M. Papazian, Glycosylation increases potassium channel stability and surface expression in mammalian cells, J. Biol. Chem. 276 (2001) 34028–34034. [7] M.K. Hall, T.A. Cartwright, C.M. Fleming, R.A. Schwalbe, Importance of glycosylation on function of a potassium channel in neuroblastoma cells, PLoS One 6 (2011) e19317. [8] J.N. Contessa, M.S. Bhojani, H.H. Freeze, B.D. Ross, A. Rehemtulla, T.S. Lawrence, Molecular imaging of N-linked glycosylation suggests glycan biosynthesis is a novel target for cancer therapy, Clin. Cancer Res. 16 (2010) 3205–3214. [9] P.B. Dunham, G.W. Stewart, J.C. Ellory, Chloride-activated passive potassium transport in human erythrocytes, Proc. Natl. Acad. Sci. U. S. A. 77 (1980) 1711–1715. [10] P.K. Lauf, N.C. Adragna, K–Cl cotransport: properties and molecular mechanism, Cell. Physiol. Biochem. 10 (2000) 341–354. [11] M.R. Shen, C.Y. Chou, K.F. Hsu, Y.M. Hsu, W.T. Chiu, M.J. Tang, S.L. Alper, J.C. Ellory, KCl cotransport is an important modulator of human cervical cancer growth and invasion, J. Biol. Chem. 278 (2003) 39941–39950. [12] M.R. Shen, A.C. Lin, Y.M. Hsu, T.J. Chang, M.J. Tang, S.L. Alper, J.C. Ellory, C.Y. Chou, Insulin-like growth factor 1 stimulates KCl cotransport, which is necessary for invasion and proliferation of cervical cancer and ovarian cancer cells, J. Biol. Chem. 279 (2004) 40017–40025. [13] M.R. Shen, C.Y. Chou, J.C. Ellory, Volume-sensitive KCI cotransport associated with human cervical carcinogenesis, Pflugers Arch. 440 (2000) 751–760. [14] Y.M. Hsu, C.Y. Chou, H.H. Chen, W.Y. Lee, Y.F. Chen, P.W. Lin, S.L. Alper, J.C. Ellory, M.R. Shen, IGF-1 upregulates electroneutral K–Cl cotransporter KCC3 and KCC4 which are differentially required for breast cancer cell proliferation and invasiveness, J. Cell. Physiol. 210 (2007) 626–636. [15] Y.M. Hsu, Y.F. Chen, C.Y. Chou, M.J. Tang, J.H. Chen, R.J. Wilkins, J.C. Ellory, M.R. Shen, KCl cotransporter-3 down-regulates E-cadherin/beta-catenin complex to promote epithelial–mesenchymal transition, Cancer Res. 67 (2007) 11064–11073. [16] Y.F. Chen, C.Y. Chou, R.J. Wilkins, J.C. Ellory, D.B. Mount, M.R. Shen, Motor protein-dependent membrane trafficking of KCl cotransporter-4 is important for cancer cell invasion, Cancer Res. 69 (2009) 8585–8593. [17] D.B. Mount, A. Mercado, L. Song, J. Xu, A.L. George Jr., E. Delpire, G. Gamba, Cloning and characterization of KCC3 and KCC4, new members of the cation-chloride cotransporter gene family, J. Biol. Chem. 274 (1999) 16355–16362. [18] K. Hiki, R.J. D'Andrea, J. Furze, J. Crawford, E. Woollatt, G.R. Sutherland, M.A. Vadas, J.R. Gamble, Cloning, characterization, and chromosomal location of a novel human K+–Cl− cotransporter, J. Biol. Chem. 274 (1999) 10661–10667. [19] H.H. Yeh, W.W. Lai, H.H. Chen, H.S. Liu, W.C. Su, Autocrine IL-6-induced Stat3 activation contributes to the pathogenesis of lung adenocarcinoma and malignant pleural effusion, Oncogene 25 (2006) 4300–4309. [20] K. Berglund, W. Schleich, P. Krieger, L.S. Loo, D. Wang, N.B. Cant, G. Feng, G.J. Augustine, T. Kuner, Imaging synaptic inhibition in transgenic mice expressing the chloride indicator, Clomeleon Brain Cell Biol. 35 (2006) 207–228. [21] N. Zaarour, S. Demaretz, N. Defontaine, D. Mordasini, K. Laghmani, A highly conserved motif at the COOH terminus dictates endoplasmic reticulum exit and cell surface expression of NKCC2, J. Biol. Chem. 284 (2009) 21752–21764. [22] Y. Jitsuhara, T. Toyoda, T. Itai, H. Yamaguchi, Chaperone-like functions of highmannose type and complex-type N-glycans and their molecular basis, J. Biochem. 132 (2002) 803–811. [23] J. Zhu, J. Yan, W.B. Thornhill, N-Glycosylation promotes the cell surface expression of Kv1.3 potassium channels, FEBS J. 279 (15) (2012) 2632–2644. [24] G. Hendriks, M. Koudijs, B.W. van Balkom, V. Oorschot, J. Klumperman, P.M. Deen, P. van der Sluijs, Glycosylation is important for cell surface expression of the water channel aquaporin-2 but is not essential for tetramerization in the endoplasmic reticulum, J. Biol. Chem. 279 (2004) 2975–2983. [25] R. Glozman, T. Okiyoneda, C.M. Mulvihill, J.M. Rini, H. Barriere, G.L. Lukacs, Nglycans are direct determinants of CFTR folding and stability in secretory and endocytic membrane traffic, J. Cell Biol. 184 (2009) 847–862. [26] S. Casula, B.E. Shmukler, S. Wilhelm, A.K. Stuart-Tilley, W. Su, M.N. Chernova, C. Brugnara, S.L. Alper, A dominant negative mutant of the KCC1 K–Cl cotransporter: both N- and C-terminal cytoplasmic domains are required for K–Cl cotransport activity, J. Biol. Chem. 276 (2001) 41870–41878. [27] A.M. Hartmann, M. Wenz, A. Mercado, C. Storger, D.B. Mount, E. Friauf, H.G. Nothwang, Differences in the large extracellular loop between the K(+)–Cl(−) cotransporters KCC2 and KCC4, J. Biol. Chem. 285 (2010) 23994–24002. [28] D.A. Fishman, Y. Liu, S.M. Ellerbroek, M.S. Stack, Lysophosphatidic acid promotes matrix metalloproteinase (MMP) activation and MMP-dependent invasion in ovarian cancer cells, Cancer Res. 61 (2001) 3194–3199. [29] J.Y. Sun, H. Yang, S. Miao, J.P. Li, S.W. Wang, M.Z. Zhu, Y.H. Xie, J.B. Wang, Z. Liu, Q. Yang, Suppressive effects of swainsonine on C6 glioma cell in vitro and in vivo, Phytomedicine 16 (2009) 1070–1074. [30] T. Shiraishi, T. Yoshida, S. Nakata, M. Horinaka, M. Wakada, Y. Mizutani, T. Miki, T. Sakai, Tunicamycin enhances tumor necrosis factor-related apoptosis-inducing ligand-induced apoptosis in human prostate cancer cells, Cancer Res. 65 (2005) 6364–6370.