Tyrosine-dependent basolateral targeting of human connexin43–eYFP in Madin–Darby canine kidney cells can be disrupted by the oculodentodigital dysplasia mutation L90V Jana Chtchetinin 1,2 , Wes D. Gifford 1,2 , Sichen Li 1,2 , William A. Paznekas 3 , Ethylin Wang Jabs 3,4 and Albert Lai 1,2 1 Department of Neurology, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA 2 Henry E. Singleton Brain Cancer Research Program, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA 3 Institute of Genetic Medicine, Johns Hopkins University, Baltimore, MD, USA 4 Department of Genetics and Genomic Sciences, Mount Sinai School of Medicine, New York, NY, USA Introduction Connexin 43 (Cx43) is a ubiquitously expressed gap junctional subunit that mediates intercellular commu- nication via the formation of gap junctions and hemi- channels [1]. In addition, Cx43 may promote normal cellular migration and development by enabling inter- cellular adhesion [2]. Cx43 is expressed in many polar- ized cell types, such as brain endothelial cells, thyroid epithelial cells, and cholangiocytes [3–6]. Cx43 is also highly expressed in astrocytes, a cell type that exhibits a polarized phenotype and participates in polarized Keywords basolateral; connexin43; oculodentodigital dysplasia; tyrosine Correspondence A. Lai, 710 Westwood Plaza, Suite 1-230, Reed Neurological Research Center, Los Angeles, CA 90095, USA Fax: +1 310 825 0644 Tel: +1 310 825 5321 E-mail: albertlai@mednet.ucla.edu (Received 10 February 2009, revised 16 September 2009, accepted 25 September 2009) doi:10.1111/j.1742-4658.2009.07407.x Polarized membrane sorting of connexin 43 (Cx43) has not been well-char- acterized. Based on the presence of a putative sorting signal, YKLV(286– 289), within its C-terminal cytoplasmic domain, we hypothesized that Cx43 is selectively expressed on the basolateral surface of Madin–Darby canine kidney (MDCK) cells in a tyrosine-dependent manner. We generated stable MDCK cell lines expressing human wild-type and mutant Cx43–eYFP, and analyzed the membrane localization of Cx43–eYFP within polarized mono- layers using confocal microscopy and selective surface biotinylation. We found that wild-type Cx43–eYFP was selectively targeted to the basolateral membrane domain of MDCK cells. Substitution of alanine for Y286 dis- rupted basolateral targeting of Cx43–eYFP. Additionally, substitution of a sequence containing the transferrin receptor internalization signal, LSYTRF, for PGYKLV(284–289) also disrupted basolateral targeting. Taken together, these results indicate that Y286 in its native amino acid sequence is necessary for targeting Cx43–eYFP to the basolateral mem- brane domain of MDCK cells. To determine whether the F52dup or L90V oculodentodigital dysplasia -associated mutations could affect polarized sorting of Cx43–eYFP, we analyzed the expression of these Cx43–eYFP mutant constructs and found that the L90V mutation disrupted basolateral expression. These findings raise the possibility that some oculodentodigitial dysplasia-associated mutations contribute to disease by altering polarized targeting of Cx43. Abbreviations Cx43, connexin 43; eYFP, enhanced yellow fluorescent protein; MDCK, Madin–Darby canine kidney; ODDD, oculodentodigital dysplasia; DPBS, dulbecco’s phosphate buffered saline. 6992 FEBS Journal 276 (2009) 6992–7005 ª 2009 The Authors Journal compilation ª 2009 FEBS functions [7–10]. So far, there have been limited studies examining the expression of Cx43 in polarized cells, and there is little information regarding characteriza- tion of the involved sorting signals. Previous studies have demonstrated basolateral expression of Cx43 and other connexins in various polarized cell types [5,11– 13]. The trafficking of a Cx43–GFP chimera expressed in Madin–Darby canine kidney (MDCK) cells has previously been examined, but not in the context of polarized monolayers [14]. MDCK cells are a well-characterized model system for the study of polarized trafficking to distinct apical and basolateral domains that are separated by tight junctions [15–17]. In MDCK cells, basolaterally expressed membrane proteins often depend on a tyro- sine or dileucine-based sorting signal located in the cytoplasmic tail. A common tyrosine-based consensus sorting motif is YXXø (where Y is tyrosine, X is any amino acid, and Ø is an amino acid with a bulky hydro- phobic side chain) [18]. These polarized sorting signals are recognized in other cell types as well [19–21]. For example, the vesicular stomatitis virus glycoprotein (VSV-G) protein, which is targeted to the basolateral membrane domain in MDCK cells, has a tyrosine-based signal directing it to the somatodendrite versus the axon in neurons and the myelin sheath versus the soma in oligodendrocytes [22,23]. The existence of such a signal in the C-terminus of Cx43 led us to hypothesize that Y286 is involved in basolateral targeting of Cx43. At least 60 mutations in Cx43 have been discovered that cause oculodentodigital dysplasia (ODDD), a rare human developmental disorder characterized by defects in the craniofacial bones, loss of tooth enamel, and abnormal soft tissue separation of two or three digits [24,25]. Depending on the particular Cx43 mutation, a wide range of other abnormalities, including neurologi- cal and cardiac abnormalities, are observed [26–37]. As yet, there is no clear understanding of the relationship between genotype and phenotype, although functional evaluations of most of the mutations that have been studied have demonstrated reduced gap junction activ- ity [38–43]. Several ODDD mutations have been found to cause altered trafficking of Cx43. For example, the C260fsX306 mutation (leading to truncation of most of the Cx43 cytoplasmic tail) and the G60S mutation have been found to cause impaired cell membrane expression of Cx43 and hence impaired intercellular communication [44,45]. Interestingly, the G138R and G143S ODDD mutations have been found to cause enhanced hemichannel function with absent gap junc- tional signaling in HeLa cells, and these mutations were associated with decreased Cx43 degradation [46]. These studies provide evidence that abnormal mem- brane protein trafficking may be responsible for the altered function of Cx43 associated with disease. In this study, we determined that a fusion between Cx43 and enhanced yellow fluorescent protein (Cx43– eYFP) is targeted to the basolateral membrane domain of MDCK cells. We also found that Y286 in the sequence PGYKLV(284–289) is necessary for basolat- eral targeting of Cx43–eYFP. In addition, we found that the L90V ODDD mutation disrupts the selective delivery of Cx43–eYFP to the basolateral membrane domain. These results imply that aberrant polarized sorting of Cx43 may be associated with particular ODDD phenotypes. Results Stable expression of wild-type and mutant Cx43–eYFP constructs in MDCK cells To examine the localization of human Cx43–eYFP constructs in MDCK cells, we stably expressed eYFP- tagged wild-type (WT) and mutant Cx43 cDNA fusion constructs in MDCK (strain II) cells. As indicated in Fig. 1, eYFP-tagged Cx43 constructs comprise the Cx43 sequence joined at its C-terminus to eYFP by an eight- amino-acid ‘linker’ segment (SRDPPVAT). The loca- tions of the four mutations analyzed (Y286A, LSYTRF, L90V and F52dup) are also indicated. We confirmed by western blot analysis that Cx43–eYFP constructs were properly translated in MDCK cells (Fig. 2A). Using a Cx43 antibody, a prominent band migrating at approxi- mately 70 kDa, the predicted molecular weight of full- length Cx43–eYFP, was detected in total and surface lysates for WT and each mutant cell line, but not in uninfected control cells. Similarly, using the polyclonal GFP antibody, a prominent band migrating at approxi- mately 70 kDa was detected in total and surface lysates for WT and each mutant cell line, but not in uninfected control cells. For reference, a band at approximately 29 kDa was detected in the total cell lysate of cells expressing only eGFP. No GFP expression was detected after surface biotinylation of cells expressing only eGFP, demonstrating the selectivity of surface biotiny- lation for cell surface proteins. As the amounts of pro- tein loaded for surface and total Cx43–eYFP western blots were not normalized to each other, no conclusions can be drawn from these experiments regarding the relative level of surface expression of the constructs. Confocal images in the XY plane show that all Cx43– eYFP constructs are predominantly expressed at the cell membrane. Gap junction aggregates, or plaques, can also be seen at areas of cell–cell contact (arrow) for all constructs except the F52dup mutant (Fig. 2B). J. Chtchetinin et al. Basolateral sorting of Cx43 in MDCK cells FEBS Journal 276 (2009) 6992–7005 ª 2009 The Authors Journal compilation ª 2009 FEBS 6993 WT Cx43–eYFP is selectively expressed on the basolateral domain of MDCK cells To determine the steady-state polarized membrane distribution of WT Cx43–eYFP in MDCK cells, polarized monolayers expressing WT Cx43–eYFP were cultured on filter inserts and examined using confocal microscopy. Reconstructed Z sections (XZ plane) show that WT Cx43–eYFP is selectively expressed on the basolateral surface of MDCK cells (Fig. 3A–C; three M G D W S A L G K L D L K V Q A Y S T A G G K L W V S V L F I F R I L L L G T A V E S A W G D E Q S A F R R C N T Q Q P G C E N V C Y D K S F P I S H V F W V L Q I I F V S V P T L L Y A H V F Y V V M R K E E K L N K K E E E E E L K V A Q T D G N V D M H L K Q I K K I F K Y G I E E H G K M R V K G G L L R T Y I I I S L F K S I F E V A F L L I Q W Y I Y G F S L S A V Y C K R D P C P H Q V D C F L S R P T E K T I F I I F M L L S V V V S L A L N I I L F Y V F F K G V K D R V K G K S D P Y H A T T S G A L S P A K D C G S Q K Y Y A F N G C S S P T A P L S P M S P P G Y KLV T G D R N N S S C R N Y N K Q A S E Q N W A N Y S A E Q N R M G Q A S G T I S N S H A Q P F D F P D D N Q N S K K L A A G H E L Q P L A I V D Q R P S S R A S S R A S S R P R P D D L E I SRDPPVAT eYFP NH 2 L Extracellular L90V F52dup Y286A Lipid bilayer Cytoplasm LSYTRF (284–289) Linker Fig. 1. Schematic diagram of the Cx43– eYFP amino acid sequence indicating the predicted topology of Cx43 and the position of eYFP and its fusion to the C-terminus via an eight amino acid linker (SRDPPVAT). Cx43 is predicted to span the plasma membrane four times, and has cytoplasmi- cally located N- and C-termini. The locations of the amino acid mutations examined in this study are indicated. F52dup is located in the extracellular domain, L90V is located in the second transmembrane domain, and Y286A and LSYTRF are located in the cyto- plasmic tail. This figure has been modified from one that has been published previously [43]. Uninfected eGFP Wild-type Y286A LSYTRF F52dup L90V 72 - 56 - 34 - 26 - Total Surface Anti-Cx43 Anti-GFP 43 - Cx43- eYFP eGFP Cx43- eYFP 10 µm Wild-type Y286A LSYTRF 10 µm 10 µm 10 µm 10 µm F52dup L90V Uninfected 10 µm Uninfected eGFP Wild-type Y286A LSYTRF F52dup L90V 72 - 56 - 34 - 26 - 43 - A B Fig. 2. Expression of Cx43–eYFP constructs in MDCK cells. (A) Surface and total protein were isolated by biotinylation and analyzed by western blot. Using Cx43 and GFP antibodies to confirm proper translation of Cx43–eYFP constructs, bands of approxi- mately 70 kDa representing the full-length fusion proteins were obtained from cells expressing WT and mutant Cx43–eYFP. (B) Confocal XY images showing the XY plane from above the apical surface of cell monolayers. Uninfected control MDCK cells showed minimal background fluorescence. WT and mutant Cx43–eYFP constructs were expressed on the cell membrane. Gap junction plaques (indicated by the arrow in the WT image) formed at points of cell–cell contact in all mutants with the same frequency and morphology as WT except for F52dup, which formed plaques much less frequently. Scale bar = 10 lm. Basolateral sorting of Cx43 in MDCK cells J. Chtchetinin et al. 6994 FEBS Journal 276 (2009) 6992–7005 ª 2009 The Authors Journal compilation ª 2009 FEBS individual cell lines are shown). For comparison, lim- ited signal was detected in uninfected control cells (Fig. 3D), and cells infected with eGFP only showed a diffuse cytoplasmic signal but no specific surface locali- zation (Fig. 3E). Localization of plaques in the XZ plane is shown in Fig. 7 (see below). To confirm these results, we performed selective sur- face biotinylation on the apical and basolateral surfaces and western blot analysis using the Cx43 antibody in order to visualize the apical ⁄ basolateral distribution of Cx43–eYFP. This analysis confirmed that the majority of the signal is found on the basolateral surface of cell lines expressing WT Cx43–eYFP, although there appeared to be a small amount of apically expressed WT Cx43–eYFP (Fig. 6A). Five independent WT clones were tested, all yielding similar results. The results for three are shown. No signal could be detected for the apical or basolateral surfaces of uninfected con- trol cells at either 43 or approximately 70 kDa, and, similarly, no signal could be detected at approximately 29 kDa for the apical or basolateral surface of cells expressing only eGFP (Fig. 6B). Additionally, quantita- tive analysis of the distribution between the two domains using a fluorescence plate reader revealed that 86% of the surface WT Cx43–eYFP was located on the basolateral surface (Fig. 6G). As described in Experimental procedures, all experi- ments were performed after sodium butyrate pre-incu- bation to increase expression of the Cx43–eYFP constructs. We confirmed that sodium butyrate treat- ment did not affect results by performing confocal microscopy in the absence of sodium butyrate, dye transfer experiments with and without butyrate, and transepithelial resistance measurements with and without butyrate on a selected cell line (Fig. S1). Y286 in the context of its native sequence PGYKLV(284–289) is necessary for selective basolateral expression of Cx43–eYFP in MDCK cells Y286 is contained within a putative tyrosine-based sorting signal, YKLV(286–289), and within a PPXY motif, PPGY(283–286), which is a ubiquitin ligase (NEDD4) binding site that is involved in internaliza- tion and degradation of Cx43 [47]. To determine whether Y286 is involved in the basolateral targeting of WT Cx43–eYFP, we substituted an alanine for Y286 of the Cx43–eYFP sequence and examined the surface distribution of the resulting Y286A mutant construct in polarized MDCK cell monolayers. Confo- cal analysis of XZ sections revealed that the Y286A mutation causes Cx43–eYFP to be expressed predomi- nantly on the apical surface, indicating that Y286 is necessary for the proper basolateral distribution of Cx43–eYFP in MDCK cells (Fig. 4A,B). Western blot analysis of apical/basolateral surface biotinylation frac- tions confirmed this predominantly apical distribution 10 µm 10 µm 10 µm X Y X Y X Y X Z X Z X Z 10 µm X Y X Z 10 µm X Y X Z A C B DE Fig. 3. WT Cx43–eYFP is selectively targeted to the basolateral domain of MDCK cells. (A–C) Three independent MDCK clones expressing WT Cx43–eYFP were cultured on filter inserts. Z sections of cell monolayers are shown in the top panels. Lines drawn through the XY plane in the bottom panels indicate the location of the Z sections. WT Cx43–eYFP was expressed on the basolateral membrane domain of MDCK cells. (D) Z section of a monolayer of uninfected MDCK cells showing background fluorescence. (E) Z section of a monolayer of MDCK cells expressing only eGFP showing a diffuse cytoplasmic pattern. Scale bar = 10 lm. J. Chtchetinin et al. Basolateral sorting of Cx43 in MDCK cells FEBS Journal 276 (2009) 6992–7005 ª 2009 The Authors Journal compilation ª 2009 FEBS 6995 (Fig. 6C), and quantitative analysis showed that 78% of the surface signal for this mutant construct was located on the apical surface (Fig. 6G). Data from two individual cell lines are shown. To determine the effect of substitution of a sequence containing the internalization signal of the transferrin receptor in place of PGYKLV(284–289), we expressed a mutant construct in which PGYKLV(284–289) was replaced by LSYTRF [48]. Confocal Z sections of MDCK cells expressing the LSYTRF mutation showed that the Cx43–eYFP sig- nal was apparently equally distributed on the apical and basolateral surfaces (Fig. 4C,D). Western blot analysis of selective surface biotinylation fractions confirmed that cells expressing the LSYTRF mutant construct express Cx43–eYFP at similar levels on the apical and basolateral surfaces (Fig. 6E). Interest- ingly, quantification of the surface protein using a fluorescence plate reader indicated that, similar to Y286A, 74% of the LSYTRF surface signal resides on the apical surface (Fig. 6G). As Y286 is intact in this mutant construct but its surrounding sequence is altered, these findings strongly suggest that the context in which Y286 exists is important for main- tenance of basolateral targeting. The L90V but not the F52dup ODDD mutation disrupts basolateral sorting of Cx43–eYFP in MDCK cells To determine whether ODDD-associated mutations affect the basolateral targeting of Cx43–eYFP, we examined the localization of F52dup and L90V mutant Cx43–eYFP constructs in polarized MDCK cell monolayers. These mutants were selected for analysis based on our previous study of ODDD mutants in C6 rat glioma cells [43]. We chose the F52dup mutant because it failed to form gap junction plaques and the L90V mutant because it appeared to have an increased amount of plaques in the glial cell processes compared to WT (unpublished observation). Confocal Z sections of cells expressing the F52dup mutant construct showed predominantly basolateral expression, indicating that this mutation does not alter polarized targeting of Cx43–eYFP (Fig. 5A,B). These results were confirmed by western blot analysis of the selective surface biotinylation fractions, with nearly all signal (97%) found on the basolateral surface (Fig. 6D,G). In contrast, confocal analysis of the distribution of the L90V mutant construct in polarized monolayers revealed that this mutation causes Cx43–eYFP to be distributed on both the api- cal and basolateral surfaces (Fig. 5C,D). Western blot analysis of selective surface biotinylation fractions confirmed that surface expression of the L90V mutant construct is not restricted to the basolateral surface, with 57% of the signal being found on the apical sur- face (Fig. 6F,G). The L90V-1 clone appeared to be exclusively expressed on the apical surface (Fig. 6F). The results for two independent clones for each mutant construct are shown. Gap junction plaques reside predominantly on the lateral surface of cells expressing WT and apically distributed mutant constructs With the exception of the F52dup mutant, all mutant constructs formed plaques with the same frequency 10 µm X Y 10 µm X Y X Z 10 µm X Y X Z X Z 10 µm X Y X Z AB AB CD Fig. 4. Y286 in its native context is neces- sary for selective basolateral expression of Cx43–eYFP in MDCK cells. (A,B) Z sections of MDCK monolayers expressing the Y286A mutant construct showing predominantly apical distribution of Cx43–eYFP (two inde- pendent clones). (C,D) Z sections of MDCK monolayers expressing the LSYTRF mutant construct showing signal on both the apical and basolateral membranes (two independent clones). Scale bar = 10 lm. Basolateral sorting of Cx43 in MDCK cells J. Chtchetinin et al. 6996 FEBS Journal 276 (2009) 6992–7005 ª 2009 The Authors Journal compilation ª 2009 FEBS 10 µm X Y 10 µm X Y 10 µm X Y 10 µm X Y X Z X Z X Z X Z AB CD Fig. 5. L90V but not the F52dup ODDD mutation disrupts basolateral targeting of Cx43–eYFP in MDCK cells. (A,B) Z sections of MDCK monolayers expressing the F52dup mutant construct showing that this mutation does not affect basolateral targeting of Cx43–eYFP (two independent clones). (C,D) Z sections of MDCK monolay- ers expressing the L90V mutant construct showing that basolateral targeting of Cx43–eYFP was disrupted (two independent clones). Scale bar = 10 lm. A WT-1 WT-2 WT-3 UninfectedUninfected eGFP Y286A-1 Y286A-2 LSYTRF-1 LSYTRF-2 F52dup-1 F52dup-2 72 -72 - 72 - 72 - 72 - 72 - 29 -43 - L90V-2L90V-1 abab B C abab ab ab ab ab ab ab D ab ab ab ab F E G WT Y286A LSYTRF F52dup L90V Apical 14% Basolateral 86% 78% 74% 26% 97% 57% 43% 3% 22% 20% 40% 60% 80% 100% WT Y286A LSYTRF F52dup L90V Surface expression (%) Cell line Apical Basolateral * * * 0% Fig. 6. Apical ⁄ basolateral cell surface distribution of Cx43–eYFP. Selective surface biotinylation followed by western blot analysis using the Cx43 antibody showed that WT and the F52dup mutant construct were expressed predominantly on the basolateral surface (A,D), but the Y286A, LSYTRF and L90V mutant constructs were not exclusively distributed on the basolateral surface (C,E,F). (B) Uninfected MDCK cells showed no bands at 43 or 70 kDa. The absence of bands at approximately 29 kDa in control cells expressing only eGFP indicates that non-specific biotinylation of cytoplasmic protein did not occur. Two or three individual cell lines are shown for each construct. (G) Surface expression as quantified by a fluorescence plate reader following selective surface biotinylation. The percentage of signal found on the basolateral surface of Y286A, LSYTRF and L90V was significantly different from that of WT (*). Error bars indicate SEM. n = 3–6 (varies between cell lines). J. Chtchetinin et al. Basolateral sorting of Cx43 in MDCK cells FEBS Journal 276 (2009) 6992–7005 ª 2009 The Authors Journal compilation ª 2009 FEBS 6997 and morphology as WT (Fig. 2B). The F52Dup mutant formed large gap junction plaques far less fre- quently, consistent with our previous results when the F52dup mutant was expressed in C6 rat glioma cells [43]. The various plaques formed in cells expressing the WT construct were representative of plaques seen in all mutant cell lines (Fig. 7A–E). Plaques formed most frequently on the lateral surface of the cell membrane and spanned either the entire area of cell– cell contact (Fig. 7A) or a smaller area closer to the basal (Fig. 7B) or apical (Fig. 7C) membrane. Plaques were also sometimes seen apparently unbound to the cell membrane near the basal surface (Fig. 7D) or near the apical membrane (Fig. 7E); however, these examples occurred less frequently. Mutants with api- cal expression formed plaques predominantly on the lateral surface; a representative plaque formed by the Y286A mutant is shown (Fig. 7F). We found no dif- ference in the relative frequencies of these types of plaques between WT and the various mutants (data not shown). Degradation from the cell membrane is impaired for Y286A Cx43–eYFP The tyrosine-based signal containing Y286 also con- forms to a putative lysosomal degradation signal [47]. To determine whether any of the mutations inhibit degradation of surface Cx43–eYFP, we performed pulse–chase surface biotinylation experiments on all of the cell lines (grown as monolayers on 60 mm cell culture dishes). The amount of Cx43–eYFP present after 2, 4 and 8 h chase intervals was measured by a fluorescence plate reader and normalized to the amount present at time 0. We found that surface WT Cx43–eYFP was rapidly degraded, with a half-life of under 2 h. We performed western blots probed with the GFP and Cx43 antibodies to confirm that we were properly measuring the disappearance of intact Cx43– eYFP without tracking degradation products contain- ing eYFP (data not shown). The Y286A mutation resulted in a slightly longer half-life of surface Cx43– eYFP compared to WT, indicating that Y286A has a role in mediating degradation from the cell surface (Fig. 8). The L90V, LSYTRF and F52dup mutations had minimal effect on degradation of Cx43–eYFP from the surface compared to WT (Fig. S2). Discussion We sought to determine the localization of surface expression of Cx43–eYFP in polarized MDCK cells and whether two ODDD mutations could alter this distribution. We hypothesized that Cx43–eYFP would have tyrosine-dependent basolateral expression in MDCK cells based on the presence of YKLV(286– 289) within the amino acid sequence of the cyto- plasmic tail. Using confocal microscopy and selective surface biotinylation, we have shown that WT Cx43–eYFP is targeted to the basolateral membrane domain of MDCK cells (Figs 3A and 6A). The selec- tive expression of Cx43–eYFP on the basolateral 10 µm X Y X Z 10 µm X Y 10 µm X Y 10 µm X Y 10 µm X Y X Z X Z X Z X Z 10 µm X Y X Z ABC DEF Fig. 7. Gap junction plaques are located predominantly on the lateral surface of cell monolayers expressing WT and mutant constructs. Z sections generated from areas containing the various types of plaques expressed by the WT construct are shown, and are representative of plaques seen in all mutants. Plaques were found to span the entire membrane (A) or part of the membrane (B,C). Less frequently, plaques were found to be apparently suspended near the basal membrane (D) or the apical membrane (E), rather than at cell–cell junctions. (F) Z section of a representative plaque formed by the Y286A mutant construct. Arrows indicate plaques. Scale bar = 10 lm. Basolateral sorting of Cx43 in MDCK cells J. Chtchetinin et al. 6998 FEBS Journal 276 (2009) 6992–7005 ª 2009 The Authors Journal compilation ª 2009 FEBS domain of MDCK cells is consistent with other stud- ies showing that Cx43 and other connexins are typi- cally distributed on the basolateral surface of polarized cells [5,11–13]. As all experiments were per- formed in the presence of sodium butyrate to increase expression of the Cx43–eYFP constructs, as previ- ously shown using this model system [49], we used a variety of approaches to confirm that sodium butyrate incubation did not alter the distribution of Cx43– eYFP or disrupt the integrity of the monolayer (Fig. S1). First, we performed confocal experiments in the absence of sodium butyrate pre-incubation and found similar distributions between the apical and ba- solateral surface for all constructs. We then performed dye transfer assays on monolayers expressing each construct and found no consistently significant changes in dye transfer after sodium butyrate pre- incubation compared to untreated monolayers. Lastly, we performed transepithelial resistance measurements on a representative cell line and found that the sodium butyrate did not alter the transepithelial resis- tance of the filter-grown monolayer. We demonstrated that Y286 in the cytoplasmic domain of Cx43 is necessary for basolateral sorting of Cx43–eYFP in MDCK cells, as evidenced by the absence of selective basolateral targeting of the Y286A mutant construct (Fig. 4A,B). Therefore, Y286 of the tetrapeptide sequence YKLV(286–289) represents a crit- ical tyrosine residue of the common basolateral sorting motif YXXø [18]. To further characterize this signal, we then substituted PGYKLV(284–289) by the LSYTRF sequence containing the transferrin receptor internaliza- tion signal (YTRF), and found that selective basolateral targeting of Cx43–eYFP was not preserved (Fig. 4C,D). This finding was not unexpected given that the transfer- rin receptor internalization signal does not contain basolateral targeting information [48]. Interestingly, gap junction plaques were found predominantly on the lat- eral membrane domain, even in cell lines expressing con- structs that have apical expression (Fig. 7). This raises the possibility that the biotinylation assay does not cap- ture this population completely, possibly due to poor accessibility of fully assembled gap junctions to sulfo- NHS-LC-biotin (see methods). Alternatively, this pla- que population may be a component of small basolat- eral fraction of Cx43-eYFP detected for Y286A and the other apically expressed mutant Cx43–eYFP constructs. Despite our inability to determine which assembly states are efficiently captured by the biotinylation assay, the combination of confocal microscopy with the biotinyla- tion data strongly suggest that trafficking of Cx43, pre- sumably as undocked Cx43 connexons (hemichannels), is directed to the basolateral surface. From our data, it remains unclear by what mechanism gap junctional plaques are retained at the basolateral surface. We found that the ODDD-associated L90V mutant disrupts basolateral expression of Cx43–eYFP without affecting the rate of surface degradation, whereas F52dup does not affect either basolateral expression or degradation (Figs 5 and 6). The finding of altered basolateral expression of the L90V mutant Cx43–eYFP construct may indicate the presence of another basolat- eral sorting determinant located in the second trans- membrane domain of Cx43. Although not as common as cytoplasmic sorting signals, transmembrane sorting signals have been identified. For example, the gastric H,K-ATPase has an apical sorting signal in its 4th transmembrane domain, although the exact amino acids responsible have not been identified [50]. Studies have shown that Cx43 oligomerizes into connexons in the ER or Golgi prior to delivery to the cell membrane [1]. Therefore, an alternative hypothesis is that the L90V mutation may affect oligomerization of Cx43 subunits, which impairs recognition of the Y286-based sorting signal. Overall, these findings may provide an explanation for the additional phenotypic features of neurodegeneration and hearing loss observed in ODDD patients with the L90V and not the F52dup Time (h) Fluorescence remaining (%) 0 246 8 WT Y286A 20 10 70 40 30 50 60 80 90 100 * * * Fig. 8. The Y286A mutant construct shows impaired degradation compared to WT. Surface protein degradation assays on tissue cul- ture dishes were performed by labeling surface protein with mem- brane-impermeable sulfo-NHS-LC-biotin and lysing cells at 0, 2, 4 and 8 h. eYFP fluorescence was quantified using a fluorescence plate reader. Fluorescence remaining (%) was calculated by normal- izing to the reading at time 0 for each cell line. The Y286A mutation slightly impaired degradation of Cx43–eYFP from the surface compared to WT. For each cell type, at least three independent experiments were performed on two clones. Values shown are means ± SEM. J. Chtchetinin et al. Basolateral sorting of Cx43 in MDCK cells FEBS Journal 276 (2009) 6992–7005 ª 2009 The Authors Journal compilation ª 2009 FEBS 6999 mutation [25]. Characterization of polarized trafficking of other ODDD-associated mutants in MDCK cells will be necessary to correlate aberrant polarized traf- ficking with a particular phenotype. Consistent with other studies, we found that Cx43– eYFP is rapidly degraded from the surface, with a half-life of about 2 h (Fig. 7) [47,51]. We detected a slight but significant decrease in surface protein degra- dation between WT and the Y286A mutant but not between WT and any of the other mutants. We expected to see a greater difference between WT and Y286A given that assay of the Y286A mutant in SKHep1 cells demonstrated that the mutation increased the half-life of total cellular Cx43 from 2 to 6 h [51]. The disparity between these results may be due to the difference in cell lines used or to the fact that our assay examined degradation of surface protein as opposed to total protein. However, the lack of an appreciable effect on degradation of the construct containing the substituted transferrin internalization signal suggests that another signal may be involved in the degradation of Cx43 other than the PGYKLV(284–289) sequence. Our findings imply that targeting of Cx43 to specific domains of polarized cells may be crucial for its func- tional regulation, by concentrating or restricting inter- cellular interactions to a specific plasma membrane domain. For example, astrocytes have a polarized mor- phology with formation of specialized endfeet that make contacts with endothelial cells [9,10]. Cx43 has been found to be abundantly expressed at the connec- tion of blood vessels and astrocytic endfeet [52]. Although there are no known studies correlating tar- geting to the basolateral domain of MDCK cells with targeting to astrocytic endfeet, we predict that Cx43 is selectively targeted to the astrocytic endfeet, based on the finding that the VSV-G protein is targeted to the processes that form the myelin sheath in oligodendro- cytes, another glial cell type [22]. We also predict that Cx43 is targeted to the basolateral domain of endothe- lial cells, based on findings that other basolateral sort- ing signals active in MDCK cells are recognized for basolateral targeting in endothelial cells [53]. By simi- lar mechanisms, Cx43-dependent neuronal migration along glial fibers via gap junctional adhesion during development may require polarized targeting of Cx43 [17]. Alteration of polarized expression may explain the central nervous system developmental abnormali- ties found in ODDD. Lastly, processes such as glioma migration along white matter or endothelial basement membrane paths may also utilize Cx43-dependent mechanisms that rely on proper targeting of Cx43 in polarized cells [54,55]. Experimental procedures Cell culture MDCK (strain II) cells expressing the RSV(A) receptor [obtained from Dr G. Odorizzi, Department of Molecular, Cellular, and Developmental Biology (MCDB), University of Colorado, Boulder, CO, USA] and DF-1 cells (purchased from the American Type Culture Collection, Manassas, VA, USA) were maintained in DMEM ⁄ F12 (Mediatech, Hern- don, VA, USA) supplemented with 10% fetal bovine serum (Lonza, Walkersville, MD, USA), 100 UÆmL )1 penicillin and 100 lgÆmL )1 streptomycin (Lonza, Walkersville, MD, USA). 293T cells (obtained from Dr P. Mischel, Department of Pathology, University of California, Los Angeles, CA, USA) were maintained in Iscove’s Modified Dulbecco’s Medium (IMDM) (Hyclone, Logan, UT, USA) supplemented with 10% fetal bovine serum, 100 UÆmL )1 penicillin and 100 lgÆmL )1 streptomycin. All cells were grown in a 5% CO 2 humidified atmosphere. Generation of wild-type and mutant Cx43–eYFP fusion constructs Generation of the WT, L90V and F52dup Cx43–eYFP fusion constructs in the pEYFP-N1 vector has been described previously [43]. To introduce the Y286A and LSYTRF mutations into the Cx43 sequence, two-stage mutagenesis was performed using the WT plasmid as the template. An upstream forward primer and a mutagenic reverse primer were used to amplify a 5¢ product carrying the mutation, and an overlapping 3¢ product was amplified using a forward mutagenic primer (complement of the mutagenic reverse primer) and a downstream reverse pri- mer. The 5¢ and 3¢ Cx43 amplification products were com- bined and amplified using HindIII forward and XmaI reverse adapter primers, and the resultant altered Cx43 sequences were cloned into pEYFP-N1. The following mutagenic forward primers were used: Y286A, 5¢-GATCA TGAATTGTTTCTGTCGCCAGTAACCAGCTTGGCCC CAGGAGGAGACATAGGCG-3¢; LSYTRF, 5¢-GCAAG AAGAATTGTTTCTGTCGCCAGTGAACCGGGTATAT GACAAAGGAGACATAGGCGAGAGGGGAGC-3¢. The complementary sequences were used as reverse primers. Subcloning of Cx43–eYFP constructs into BH-RCAS and pLPCX retroviral expression vectors Mutant and WT Cx43–eYFP fusion constructs were ampli- fied using the following adapter primers containing ClaI sites (underlined): 5¢-GATCAT ATCGATACAGCAGCGGAG TTT-3¢ (forward) and 5¢-GATCAT ATCGATGCCGCT TTACTTGTA-3¢ (reverse). PCR products were digested with Basolateral sorting of Cx43 in MDCK cells J. Chtchetinin et al. 7000 FEBS Journal 276 (2009) 6992–7005 ª 2009 The Authors Journal compilation ª 2009 FEBS ClaI and ligated into ClaI-linearized BH-RCAS, a repli- cation-competent retroviral vector derived from the Rous sarcoma virus [56]. In addition, the insert encoding Y286A– eYFP was excised from the BH-RCAS vector using ClaI and inserted into the ClaI-linearized pLPCX vector. Therefore, one Y286A cell line was made using pLPCX and one was made using BH-RCAS. Cx43-coding sequences were verified at the UCLA Sequencing Core Facility. Retroviral expression of Cx43–eYFP constructs in MDCK cells These procedures have been described previously [43,56,57]. Briefly, transfection of DF-1 cells with BH-RCAS constructs encoding wild-type and mutant Cx43–eYFP was performed using Superfect (Qiagen, Valencia, CA, USA) in 60 mm tissue culture plates according to the manufacturer’s instruc- tions. Prior to infection, the RSV(A) receptor had been expressed in the MDCK cells, rendering them susceptible to infection. MDCK cells containing the RSV(A) receptor were selected using 0.5 mgÆmL )1 G418 (Sigma, St Louis, MO, USA). For transfection with pLPCX vectors, 293T cells were co-transfected with 10 lg of the designated pPLCX con- struct, 5 lg Hit-60 (a plasmid expressing MLV gag-pol) and 5 lg VSV-G using Hepes-buffered saline and 150 mm CaCl 2 . For infection of MDCK cells, conditioned medium was collected from transfected DF-1 or 293T cells (containing recombinant virus particles), filtered using a 0.45 lm filter (Whatman, Florham Park, NJ, USA), supplemented with 5 lgÆmL )1 polybrene (Sigma) and added to target MDCK cells. For each construct, multiple clonal cell lines were derived from the selected populations by limited dilution. Isolation of surface protein by biotinylation on tissue culture plates To extract protein for western blot analysis (Fig. 2A), MDCK cells expressing wild-type and mutant Cx43–eYFP constructs were cultured on 60 mm tissue culture plates. Sodium butyrate (10 mm; Alfa Aesar, Ward Hill, MA, USA) dissolved in complete cell culture medium was added to cells 24 h prior to the experiment to boost protein expression. Confocal microscopy, paracellular dye flux assays and trans- epithelial resistance measurements were used to confirm that addition of 10 mm sodium butyrate does not alter the polar- ized membrane properties of MDCK cells (Fig. S1). Cells were kept on ice for the duration of the experiment. Cells were rinsed (all washes were brief – about 1 minute) three times with cold Dulbecco’s Phosphate Buffered Saline (DPBS). Membrane impermeable sulfo-NHS-LC-biotin (2 mL; Pierce, Rockford, IL, USA) dissolved in DPBS (at a concentration of 0.5 mgÆmL )1 ) was applied to each plate for 30 min. The labeling reaction was quenched by three rinses with 100 mm glycine (Fisher, Fair Lawn, NJ, USA) dissolved in DPBS, and cells were washed once more with DPBS. Cells were lysed using lysing buffer containing 0.5% SDS (Tekno- va, Hollister, CA, USA), 1% nonidet P-40 (United States Biological, Swampscott, MA, USA) and 0.25% sodium deoxycholate (Sigma), supplemented with Complete Mini protease inhibitors (Roche Diagnostics, Indianapolis, IN, USA), 1 lm sodium vanadate (Fisher), 1 lm sodium fluoride (Fisher) and 1 lm phenylmethanesulfonyl fluoride (Sigma), for 30 min. Lysates were passed through a 25 5 8 G syringe three times (Becton-Dickinson, Franklin Lakes, NJ, USA). Lysates were centrifuged for 10 min at high speed at 4 °C, then 750 lL of the total lysate was combined with 75 lL streptavidin–agarose beads (Novagen, Gibbstown, NJ, USA) and incubated overnight on a rotating shaker at 4 °C. On the following day, beads were rinsed four times with DPBS. After the fourth rinse, the beads (which remained suspended in approximately 175 lL of DPBS) were transferred to a 96-well plate, and fluorescence was quantified using a Wallac Victor 2 plate reader (Perkin-Elmer, Waltham, MA, USA) with 485 nm excitation and 535 nm emission filters. These beads were prepared for western blot as indicated below. Selective isolation of surface protein from the apical and basolateral domains by biotinylation For polarized protein distribution studies, cells were seeded at a high density (0.5–1 · 10 6 cells per filter, depending on cell line), and cultured on Corning PET Transwell perme- able filter supports for 5 days (Corning Incorporated, Corn- ing, NY, USA). Experiments were performed as described above with a few adjustments. Membrane impermeable sulfo-NHS-LC-biotin was added to either the apical or basolateral side, and DPBS was added to the side not receiving sulfo-NHS-LC-biotin. Prior to lysing, filters were cut out, and placed into new six-well plates. Isolation of total protein by biotinylation on tissue culture plates This procedure was performed as described above with a few changes. Membrane permeable NHS-LC-biotin (Pierce, Rockford, IL, USA) was used instead of sulfo-NHS-LC- biotin. A 40 mm solution of NHS-LC-biotin was prepared in dimethylsulfoxide, and then diluted 10-fold in NaCl ⁄ P i . NHS-LC-biotin (2 mL) was applied to each plate for 4 h on a shaker at 4 °C. All rinses were performed as described above in ‘Isolation of surface protein by biotinylation on tissue culture plates’ using NaCl ⁄ P i or 100 mm glycine dis- solved in NaCl ⁄ P i instead of DPBS. Paracellular permeability of MDCK monolayers MDCK cells expressing WT and mutant Cx43–eYFP were plated at high density on filter inserts and cultured for J. Chtchetinin et al. Basolateral sorting of Cx43 in MDCK cells FEBS Journal 276 (2009) 6992–7005 ª 2009 The Authors Journal compilation ª 2009 FEBS 7001 [...]... mL of DPBS were added to the basolateral side Because sodium fluorescein is membrane-impermeable, it can only cross the monolayer via the paracellular pathway, thus the accumulation of dye on the basolateral side represents paracellular flux Accumulation of the dye was determined by taking 25 lL aliquots every 30 mins for 3 h from the basolateral chamber, dissolving them in 1.5 mL DPBS, and quantifying... (1997) Structural requirements for basolateral sorting of the human transferrin receptor in the biosynthetic and endocytic pathways of Madin–Darby canine kidney cells J Cell Biol 137, 1255–1264 49 Odorizzi G, Pearse A, Domingo D, Trowbridge IS & Hopkins CR (1996) Apical and basolateral endosomes of MDCK cells are interconnected and contain a polarized sorting mechanism J Cell Biol 135, 139–152 50 Dunbar... FL, USA) to the membrane and measuring the induced voltage drop Resistance was calculated using Ohm’s law (V = IR) The current was delivered using silver chloride electrodes in both the apical and basolateral compartments of the culture dishes The MDCK cell monolayer provides a barrier between the apical and basolateral compartments The voltage drop over the membrane was measured using an instrumentation... centrifuged for 10 min at high speed at 4 °C All of the supernatant (containing the protein) was then loaded into pre-cast Tris ⁄ Hepes ⁄ SDS gels (Pierce) Protein was transferred to nitrocellulose paper using a Trans-Blot SD semi-dry transfer cell apparatus (Bio-Rad, Hercules, CA, USA) Immunoblotting was performed by incubation in primary antibodies diluted in 1% milk in Tris-buffered saline using a 1 : 500... cognate basolateral trafficking in myelin biogenesis J Neurosci Res 86, 1007–1016 FEBS Journal 276 (2009) 6992–7005 ª 2009 The Authors Journal compilation ª 2009 FEBS 7003 Basolateral sorting of Cx43 in MDCK cells J Chtchetinin et al 23 Winckler B & Mellman I (1999) Neuronal polarity: controlling the sorting and diffusion of membrane components Neuron 23, 637–640 24 Laird DW (2008) Closing the gap on... made in DPBS, and, as a control, in NaCl ⁄ Pi, which creates calcium-free conditions that disrupt tight junctions in MDCK cells and therefore cause transepithelial resistance to be drastically decreased Western blot analysis of Cx43–eYFP fusion proteins Streptavidin–agarose beads were boiled in sample buffer containing b-mercaptoethanol and dithiothreitol for 10 min 7002 to strip off protein Then the. .. varies between cell lines) The results are normalized to time 0 for each cell line Values given are means ± SEM Significance was determined using Student’s t-test P values £ 0.05 were considered to be significant Acknowledgements This work was made possible through funding from the American Brain Tumor Association ⁄ Michael Reiss Fellowship (A.L.), Art of the Brain (A.L.), unrestricted funds from the Cancer... Center of Santa Barbara (A.L.), and National Institutes of Health grant R01 DE13849 (E.W.J.) We thank Seema Tiwari-Woodruff PhD (Department of Neurology, UCLA School of Medicine) for the use of her microscope, Guido Faas PhD (Department of Neurology, UCLA School of Medicine) for performing all of the transepithelial resistance measurements, and Olga Vagin PhD (Department of Physiology, UCLA School of Medicine)... not affect the polarized distribution of Cx43–eYFP constructs or the integrity of MDCK cell monolayers Fig S2 LSYTRF, F52dup and L90V mutant constructs show the same rate of degradation as wild-type This supplementary material can be found in the online version of this article Please note: As a service to our authors and readers, this journal provides supporting information supplied by the authors... C, Kiessling F & Kubler W (1998) Polarized expression of heterologous membrane proteins transfected in a human endothelial-derived cell line Eur J Cell Biol 75, 353–361 54 Oliveira R, Christov C, Guillamo JS, de Bouard S, Palfi S, Venance L, Tardy M & Peschanski M (2005) Contribution of gap junctional communication between tumor cells and astroglia to the invasion of the brain parenchyma by human glioblastomas . Tyrosine-dependent basolateral targeting of human connexin43–eYFP in Madin–Darby canine kidney cells can be disrupted by the oculodentodigital dysplasia mutation. require- ments for basolateral sorting of the human transferrin receptor in the biosynthetic and endocytic pathways of Madin–Darby canine kidney cells. J Cell