FASEB J.
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published as doi: 10.1096/fj.05-5309fje.
This Article
Right arrow Abstract Freely available
Right arrow Summary
Right arrow Full Text (PDF)
Right arrow Supplemental Data
Right arrow All Versions of this Article:
fj.05-5309fjev1
fj.05-5309fjev2
fj.05-5309fjev3
20/8/1221    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Locke, D.
Right arrow Articles by Harris, A. L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Locke, D.
Right arrow Articles by Harris, A. L.
(The FASEB Journal. 2006;20:1221-1223.)
© 2006 FASEB

Isoelectric points and post-translational modifications of connexin26 and connexin32

Darren Locke1,2, Irina V. Koreen2 and Andrew L. Harris

Department of Pharmacology and Physiology, New Jersey Medical School, University of Medicine and Dentistry of New Jersey, Newark, New Jersey

1Correspondence: Department of Pharmacology and Physiology, New Jersey Medical School, 185 South Orange Ave., UMDNJ, Newark, New Jersey 07103, USA. E-mail: lockeda{at}umdnj.edu

ABSTRACT

The isoelectric points of the gap junction proteins connexin26 (Cx26) and connexin32 (Cx32) were determined by isoelectric focusing in free fluids. The isoelectric points were significantly more acidic than predicted from amino acid sequences and different from each other, allowing homomeric channels to be resolved separately. The isoelectric points of the homomeric channels bracketed the isoelectric points of heteromeric Cx26/Cx32 channels. For heteromeric channels, Cx26 and Cx32 were found in overlapping, pH-focused fractions, indicating quaternary structure was retained. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry was used to identify post-translational modifications of Cx26 and Cx32 cytoplasmic domains, including the first reported post-translational modifications of Cx26. Suspected modifications were hydroxylation and/or phosphorylation near the amino terminus of both connexins, {gamma}-carboxyglutamate residues in the cytoplasmic loop of both connexins, phosphorylation in the carboxyl-terminal domain of Cx32, and palmitoylation at the carboxyl-terminus of Cx32. These modifications contribute to the measured acidic isoelectric points of Cx26 and Cx32, whereas their low molecular masses would not appreciably change connexin SDS-PAGE mobility. Most of these modifications have not previously been identified for connexins and may be instrumental in guiding and understanding novel aspects of channel trafficking and molecular mechanisms of channel regulation.—Locke, D., Koreen, I. V., Harris, A. L. Isoelectric points and post-translational modifications of connexin26 and connexin32.


Key Words: gap junction • hemichannel • IEF • mass spectrometry • Rotofor

GAP JUNCTION PLAQUES are accretions of channels providing direct intercellular signaling pathways between juxtaposed cells. Intercellular channels form by end-to-end apposition of single-membrane hemichannels composed of one (homomeric) or more (heteromeric) connexin isoforms, of which there are {approx}20. Mutation of each isoform causes unique pathologies in humans or in animal models (1) .

The functional properties of connexin channels have been regarded as primarily determined by isoform composition. Dramatic and surprising degrees of ionic and molecular selectivity of permeation, and differences in selectivity, have been observed for channels composed of different connexins (2 3 4 5 6) . However, channel function is likely to be profoundly and dynamically governed by the location, character, extent, and combination of different post-translational modifications (7 , 8) .

Modulation of connexin channel properties and cellular dynamics by covalent modification is largely unexplored, except for phosphorylation of connexin43 (Cx43; refs 9 10 11 ) and to a lesser degree of connexin32 (Cx32). Phosphorylation dynamically modulates Cx43 trafficking to junctional plaques, channel internalization, channel open probability, and unitary conductance. However, other post-translational modification(s) are also likely to be involved in channel trafficking, assembly, or function.

To date, none of the other common post-translational modifications (including acetylation, methylation, hydroxylation, deamidation, and glutamation) have been reported for any connexin, although acetylation is suspected (12) . Many of these, such as phosphorylation, have anionic charge. Others instead neutralize the charge of their target resides. Thus, post-translational modifications can easily alter the protein isoelectric point (pI), the sum of positive and negative amino acid charges. In fact, the pI of Cx43 may be significantly more acidic than expected from sequence alone (13) .

The pIs of connexin26 (Cx26) and Cx32 were determined, using homomeric and heteromeric channels. Based on sequence, Cx26 and Cx32 have the same calculated pI; however, their actual pIs were different, and more acidic than predicted, suggestive of significant post-translational processing and differences thereof. Homomeric Cx26 and homomeric Cx32 channels could be resolved separately on the basis of these differences in isoelectric charge. Heteromeric Cx26/Cx32 channels had a range of pIs between those of the corresponding homomeric channels. Therefore, it is feasible that isoelectric charge could be exploited for quantitative and/or qualitative analysis of heteromeric channel stoichiometry.

Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) was used to identify post-translational modifications of Cx26 and Cx32 that could contribute to the acidic pIs observed. For membrane proteins, such analysis can be difficult (14 , 15) . However, connexin-derived peptides from enzymatic digests were recovered and fractionated using improved reverse-phase techniques, resulting in substantial sequence coverage. Phosphorylation sites previously described for Cx32 were found, and several new phosphorylation sites and post-translational modifications of Cx32 were also suspected. Importantly, the first post-translational modifications of Cx26 are described. These findings may be instrumental in guiding and understanding novel aspects of connexin trafficking and molecular mechanisms of channel regulation.

MATERIALS AND METHODS

Components of the "Tet-On" expression system were from BD-Biosciences (Palo Alto, CA) and Life Technologies (Rockville, MD). Monoclonal anti-Cx26 antibody was from Zymed (San Francisco, CA). Agarose-conjugated and unconjugated anti-hemagglutinin mouse IgG clone HA-7 was from Sigma (St. Louis, MO), as were other reagents unless stated.

Expression and purification of tagged hemichannels from HeLa cells
Bidirectional tetracycline-responsive expression vectors (Clontech, San Diego, CA) were used to express homomeric or heteromeric connexin channels in HeLa cells, which have virtually no endogenous connexin expression. For homomeric channels, rat (r) Cx26 or rCx32 coding sequences were subcloned into one cloning site in frame with a sequence coding for a carboxyl-terminal tag. This tag was a thrombin cleavage site preceding a haemagglutinin epitope (HA, not His-Ala) and 6x(His-Asn) sequence, i.e., HA(HN)6. When both cloning sites were occupied by different connexins (heteromeric channels), only one was tagged (16) . Tet-On lines were maintained in 200 µg/ml hygromycin and 100 µg/ml G418.

For affinity purification, cells were induced with 1 µg/ml doxycycline for 48 h, during which time connexin formed functional gap junctions (16) . Cells were solubilized with 50 mM NaH2PO4, 50 mM NaCl, 5 mM EDTA, 5 mM EGTA, 80 mM n-octyl-ß-D-glucopyranoside (OG, 99.5%≤, Glycon, Germany), 1 mM ß-mercaptoethanol, 0.5 mM diisopropyl fluorophosphate (DFP; Calbiochem, La Jolla, CA), and 0.75 mg/ml azolectin, pH 7.5 (2 h, 4°C, rocking; "lysate"). Solubilization of gap junctions with OG yields hemichannels (3 , 5 , 17 18 19 20) .

The supernatant (100,000 g, 30 min, 4°C) was incubated with agarose-immobilized anti-hemagglutinin antibody overnight at 4°C with shaking. The agarose matrix was collected by centrifugation and washed in a fritted column with 10 mM NaH2PO4, 1 M NaCl, 80 mM OG, and 1 mg/ml azolectin, pH 7.4 followed by lower salt washing (138 mM NaCl). Hemichannels were eluted with 50 mM CH3COOH.Na, 0.5 M NaCl, 10 mM KCl, 1 mM EDTA, and 80 mM OG, pH 4.0, and 0.6 ml fractions were collected into 0.05 ml 1 M NaHCO3, 10 ml KCl, and 80 mM OG, pH 9.0 (final pH{approx}7.4). Tagged connexin channels are designated as Cx26T or Cx32T if homomeric and Cx26T/Cx32 or Cx26/Cx32T if heteromeric.

Tag cleavage
Tag cleavage of purified protein with restriction-grade thrombin (Novagen, Madison, WI; 18 h at 4°C) leaves four amino acids at the carboxyl-terminus (Leu-Val-Pro-Arg), part of the cleavage site. Tag-cleaved channels are designated by Tc subscripts, e.g., Cx26Tc or Cx26Tc/Cx32.

Purification of hemichannels from rodent livers
Mouse (m) Cx26/Cx32 and rCx32 (and/or Cx26/Cx32; below) hemichannels were immunopurified from livers (5 , 17) . Briefly, a crude plasma membrane fraction was prepared from homogenate. A low speed pellet (11,000 g, 30 min) was washed by centrifugation (25,000 g, 20 min) in 50 mM NaH2PO4, 50 mM NaCl, 5 mM EDTA, 5 mM EGTA, 1 mM ß-mercaptoethanol, 0.5 mM DFP, and 0.75 mg/ml azolectin, pH 7.5, and solubilized in 80 mM OG. The supernatant (100,000 g lysate) was incubated with Sepharose-conjugated M12.13 anti-Cx32 IgG (21) . Bead washing and connexin elution were as described for tagged connexins.

Heteromeric mCx26/Cx32 channels are purified from mouse membranes (5) . From rat membranes, the bulk of the channels purified are homomeric rCx32 (17) . However, when smaller ({approx}200 µl) elution fractions were used, early fractions contain heteromeric channels with high Cx26 content that elute easily from M12.13 beads. After heteromeric channels have eluted, only homomeric rCx32 channels are in subsequent elution fractions.

Liquid-phase isoelectric focusing
Isoelectric focusing (IEF) was performed with a Rotofor using premixed ampholytes to generate pH 5–7 gradients (Bio-Rad). Focusing was performed on liver or HeLa lysates, and immunopurified hemichannels (native/untagged, tagged, or tag-cleaved). Samples were diluted to {approx}10 mM salt and run in 80 mM OG and 4.0 M urea. All solutions were made using quartz-distilled reverse-osmosis-purified water. Urea solutions were deionized overnight using 5% w/w AG501-X8 mixed-bed ion-exchange resin (Bio-Rad), as modified from Stockert et al. (13) .

The focusing core was assembled from 20 chambers, separated by liquid-permeable screens. At the ends of the core, ion-exchange membranes minimize diffusion of anodic (0.1 M H3PO4) and cathodic solution (0.1 M NaOH). The ampholine solution of connexin was injected into the core from the cathode with an ampholine solution without connexin added simultaneously from the anode. Focusing runs were {approx}18 h at 6 W constant power, until equilibrium was reached according to manufacturer. The core was kept at {approx}4°C by coolant flowing through a central finger and rotated during focusing. At equilibrium, focused fractions were collected simultaneously by vacuum through a septum below each chamber, generating 20 pI fractions. The pH of each fraction was determined.

Protein recovery for SDS-PAGE and Western blotting
Protein was recovered from each fraction (22) and analyzed by SDS-PAGE. Western blots were done with monoclonal antibodies against Cx26, Cx32, or the hemaggluttinin epitope of the tag.

pIs and grand-average hydropathy scores
pIs were calculated with "Compute pI/MW" (http://us.expasy.org/tools/pi_tool.html) using connexin sequences deposited at NCBI. Grand-average hydropathy (GRAVY) values were calculated with ProtParam (http://au.expasy.org/tools/protparam.html).

In-gel digestion and MALDI-TOF-MS
Tag-purified rCx26/Cx32T and rCx32/Cx26T channels were used for analysis of "untagged" rCx26 and rCx32, respectively. "Native" Cx26 and Cx32 were from M12.13-purified mCx26/Cx32 channels. Proteins resolved by SDS-PAGE were visualized with EZinc Stain (Pierce, Rockford, IL). Excised connexin bands were chopped into pieces, destained, and dehydrated in 100% CH3CN. Gel pieces were rehydrated in 0.1M NH4HCO3, dehydrated, and reduced with 0.1 M tris(2-carboxyethyl)phosphine in 0.1 M NH4HCO3 (30 min, 37°C). After being washed, alkylation was with 50 mM ICH2CONH2 in 0.1 M NH4HCO3 (1 h, room temperature, dark; ref 23 ).

Sequencing-grade trypsin (Stratagene, La Jolla, CA; 1 µg/µl) was activated with 50 mM NH4HCO3. Gel pieces were rehydrated in 50 mM NH4HCO3, and digestion was overnight at 37°C. Sequencing-grade Endoproteinase-GluC (Endo-GluC: Roche, Indianapolis, IN; 1 µg/µl) was diluted with 50 mM NH4HCO3. Digestion was overnight at 25°C. Approximately 2 pmol of connexin were used. Peptides from each digest were extracted with 2.5% v/v Poros-20R2 reverse-phase beads (Applied Biosystems, Foster City CA) in, 5% CH3COOH/0.2% trifluoracetic acid (TFA; v/v) on a vortex mixer overnight at 4°C. Poros beads were back-loaded onto a micro-C18 ZipTip (Millipore, Billerica, MA) and washed with 0.1% v/v TFA. Peptides were recovered by elution with increasingly concentrated organic solvent directly onto a stainless steel MALDI plate. Elution conditions were 5, 15, 25, 35, 60, and 80% CH3CN, and 50% CH3OH/20% CH3CN (all v/v in 0.1% v/v TFA). Spots were allowed to dry slightly, and matrix/standard solution was added [7 mg/ml {alpha}-cyanohydroxycinnamic acid in 50% CH3OH/20% CH3CN/0.1% TFA (v/v/v) containing 50 fmol/µl each of proteomics-grade angiotensin-II (1046 Da) and ACTH (2465 Da)].

Spectrum acquisition and analysis
Spectra were acquired in positive-ion reflector mode with an ABI Proteomics-4700 Analyzer. Data Explorer v4.5 was used for spectrum analysis (Applied Biosystems). After internal calibration with 1046 and 2465 Da standards, and trypsin autodigestion products (842 and 2211 Da), spectra were processed by standard methods including noise removal (2.0 SD), baseline correction, and deisotoping. The mass/charge (m/z) measurements of all peptide peaks (>3.0 signal-to-noise ratio) for each digestion were matched against the FindMod (http://au.expasy.org/tools/findmod/) and/or ProFound (http://prowl.rockefeller.edu) database as "unknowns" [parameters: rodentia 20–40 kDa, pI 0–14, mass tolerance 1 Da] and against the expected fragment masses of digested connexin to find m/z peaks corresponding to modified connexin peptides. The Delta Mass tables of post-translational modifications (http://prowl.rockefeller.edu/aainfo/deltamassv2.html) were also consulted. All m/z lists were filtered for artifactual modifications (including carbamidomethylcysteine, propionamide-cysteine, oxidized methionine, methylated glutamate, acetylated aspartate or carboxyl-terminus, or formylated amino terminus), human keratin contamination, and for nonspecific cleavage by trypsin and/or Endo-GluC (FindPept, http://expasy.org/tools/findpept.html).

RESULTS

IEF fractionates protein mixtures with high resolution by exploiting differences in protein net charge, the sum of ionizable acidic and basic side chains of constituent amino acids, and any prosthetic groups. Several methods have been developed, using supporting matrices with ion-exchange capability or operating in free-fluids with "carrier" amphoteric electrolytes ("ampholytes") that establish a pH gradient in an applied electric field. For membrane protein focusing, solid matrices can be problematic (24 , 25) . In the present study, the Bio-Rad Rotofor was used for liquid-phase IEF of connexin channels solubilized in nonionic detergent. In the Rotofor cell, proteins focus within 20 stationary chambers separated by liquid-permeable screens. At equilibrium, their position in the pH gradient reflects their pIs.

IEF of homomeric hemichannels
Homomeric Cx26 or Cx32 hemichannels were expressed in communication-incompetent HeLa cells using tetracycline-inducible vectors (16) . These connexins have a carboxyl-terminal HA(His-Asn)6 tag (e.g., rCx26T) that may be cleaved (e.g., rCx26Tc) from purified protein, leaving four extra amino acids (LVPR) at the carboxyl-terminal that are part of a thrombin cleavage site. Calculated pIs of native/untagged rCx26 and rCx32 are identical, pH 9.20 (http://us.expasy.org/tools/pi_tool.html; Table 1 ). Charged amino acids in the tag [underlined LVPR{downarrow}GSYPYDVDYAL(HN)6, cleavage {downarrow}] add {approx}0.10 pH units calculated acidic charge to rCx26 or rCx32. After tag cleavage, rCx26Tc and rCx32Tc are calculated to be {approx}0.10 pH units more basic than rCx26 and rCx32.


View this table:
[in this window]
[in a new window]
  		Table 1. Isoelectric points of Cx26 and Cx32, predicted and determineda

The pIs of Cx26 and Cx32 were determined in the Rotofor cell using 80 mM OG (which does not disrupt hemichannel structure; refs 3 , 5 , 17 18 19 20 ), {approx}10 mM salt, 1.8% w/v ampholytes, and 4.0 M deionized urea, the latter two minimizing precipitation at pI. Under these conditions, connexin in homomeric channels resolved to distinct pIs. However, Cx26 and Cx32 pIs were significantly more acidic than predicted from amino acid sequence. The pIs of Cx26 and Cx32 were also different from each other, allowing mixed homomeric channels to be resolved separately (Fig. 1 ).


Figure 1
View larger version (17K):
[in this window]
[in a new window]
  		Figure 1. IEF of homomeric channels. pIs of Cx26 or Cx32, expressed in HeLa cells as homomeric channels, were determined by liquid-phase IEF. Connexins were HA(HN)6-tagged at CT (i.e., Cx26T or Cx32T). Tag cleavage left 4 residues (LVPR), part of a thrombin cleavage site (i.e., Cx26Tc or Cx32Tc). IEF was performed in {approx}10 mM salt, 80 mM OG, and 4.0 M deionized urea. A pH 5–7 gradient was generated in a focusing core with 20 1 ml chambers using 1.8% w/v ampholytes. Ampholine solution of connexin was injected from cathode (arrowhead). At equilibrium, fractions were collected by vacuum, and pH was determined. Not all fractions recovered are shown. For clarity, as fraction pH in adjacent chambers are unequally spaced (see Figs. 2 and 3 ), those with similar pH were combined; lanes contain protein from three focusing runs when 0.04 pH units acidic < X < 0.05 pH units basic at 0.10 pH unit intervals.

After tag cleavage, pIs of rCx26Tc were 5.82 ± 0.11n=3 and rCx32Tc were 6.33 ± 0.08n=3 (mean±SDtrials). Connexin pIs were {approx}0.70 pH units more acidic with tag uncleaved; rCx26T focused to pI 5.05 ± 0.13n=7 and rCx32T to pI 5.72 ± 0.12n=7 (Table 1) . These connexins focused at lower pIs than anticipated by sequence, irrespective of the contribution to pI of a charged HA(HN)6 or LVPR tag.

IEF of heteromeric hemichannels
We recently described an expression strategy for recombinant heteromeric hemichannels that mimics the natural structural heterogeneity of Cx26 and Cx32 channels (16) . In heteromeric channels, either Cx26 or Cx32 was tagged by HA(HN)6. Having demonstrated that tagged and tag-cleaved homomeric channels have different pIs, we speculated that heteromeric channels would show intermediate pIs, depending on stoichiometry and which isoform in the heteromer was tagged.

The pIs of Cx26 and Cx32 in heteromeric channels were determined as for homomeric channels. In striking contrast to the homomeric channels, connexins in heteromeric channels did not focus narrowly and instead focused over a broad range of individual pIs. As anticipated, these pIs were bracketed by the pIs of tagged and/or tag-cleaved homomeric channels. That is, rCx32/Cx26T channels focused between pIs 5.17 and 6.31n=8, the observed pIs of homomeric rCx32Tc and homomeric rCx26T channels, respectively (Fig. 2 A). The rCx26/Cx32T channels distributed between pIs 5.56 and 6.08n=3, the observed pIs of homomeric rCx26Tc and homomeric rCx32T channels, respectively (Fig. 2B ). After tag cleavage, rCx26/Cx32Tc and/or rCx32/Cx26Tc channels focused between pIs 5.50 and 6.30n=3, the observed pIs of homomeric rCx26Tc and homomeric rCx32Tc channels, respectively (Fig. 2C ).


Figure 2
View larger version (40K):
[in this window]
[in a new window]
  		Figure 2. IEF of heteromeric channels. pIs of heteromeric channel Cx26 and/or Cx32 were determined by liquid-phase IEF. Carboxyl-terminus of rCx26 and rCx32 was HA(HN)6-tagged but not within same channels (A: Cx26T/Cx32; B: Cx26/Cx32T). Tag cleavage leaves 4 residues (LVPR; C: Cx26Tc/Cx32, and Cx26/Cx32Tc). Measured pH of each Rotofor fraction is shown above Western blots from several different Rotofor runs. This illustrates that, from run-to-run, pH of each focusing core chamber may be slightly different, in part because gradient pH linearity can be affected by air bubbles, or salt and/or buffers transferred with connexin sample, or minor leakage within the core of anodic and cathodic solutions.

Heteromeric Cx26 and Cx32 were always found within overlapping pI fractions, indicating that quaternary structure was retained (known for OG; refs 3 , 5 , 17 18 19 20 ). Sedimentation velocity analyses of purified channels in linear sucrose gradients containing the focusing buffer components showed sedimentation values consistent with hemichannels (data not shown). Furthermore, heteromeric Cx26 and Cx32 were always distributed unevenly within the same Rotofor fractions. Therefore, it may be feasible to separate channels with different stoichiometry using isoelectric charge differences of the component isoforms.

IEF of rodent hemichannels
Previous biochemical and functional studies have shown that purified hemichannels from mouse liver are heteromeric mCx26/Cx32 and from rat liver are homomeric rCx32 (with small amounts of heteromeric channels; see Materials and Methods). The predicted pI of mCx26 is 9.22 rather than 9.20 (rCx26); two amino acids are different, one (ratY68->mouseH68) replacing negative with positive charge. The predicted pI of rCx32 (9.20) is identical to mCx32 (Table 1) .

By liquid-phase IEF, mCx26/Cx32 channels focused between pI 5.73 and 6.49n=8 (Fig. 3 A). Heteromeric mCx26/Cx32 channels were slightly more basic than HeLa rCx26/Cx32Tc and/or rCx32/Cx26Tc channels (Fig. 2C ), or rCx26/Cx32 channels in rat liver lysates (between pI 5.52 and 6.04n=8; Fig. 3B ), consistent with mCx26 being slightly more basic than rCx26. Rat liver homomeric rCx32 channels resolved narrowly at pI 6.22 ± 0.18n=6 (Fig. 3C ) and were more basic than rCx26/Cx32 channels because they contained no Cx26, which has a more acidic pI (Fig. 1) . Homomeric rodent rCx32 channels were also slightly more acidic than HeLa rCx32Tc homomeric channels (Fig. 1) ; the LVPR sequence remaining after tag cleavage adding {approx}0.10 pH units calculated basic charge (Table 1) .


Figure 3
View larger version (47K):
[in this window]
[in a new window]
  		Figure 3. IEF of connexin channels from rodent livers. pIs of Cx26 and/or Cx32 in mouse (A: heteromeric mCx26/Cx32) and rat liver lysates were determined (B: homomeric Cx32 and heteromeric rCx26/Cx32 channels in lysates). Homomeric rCx32 channels with no detectable rCx26 were also immunopurified with an anti-Cx32 monoclonal (C). Measured pH of each fraction is shown above Western blots from several experiments. See Fig. 2 legend for further details.

MALDI-TOF-MS of Cx26 and Cx32
To identify post-translational processing that may contribute to the differences between the observed acidic and expected basic pIs, Cx26 and Cx32 were immunopurified from mouse liver and HeLa cells and peptide fragments prepared by enzymatic digestion for analysis by MALDI-TOF-MS.

In-gel trypsin digestion and MALDI-TOF-MS of soluble proteins generally yields 40–60% sequence coverage. Typically, coverage of membrane proteins is lower due to proteolytic resistance, the low ionization ability of hydrophobic transmembrane domains, and therefore the difficulty of detecting low abundance peptides in complex spectra or their sequencing by tandem MS/MS approaches (14 , 15) . Enzymatic digestions of Cx26 and Cx32 with trypsin and/or Endo-GluC were carried out in parallel to maximize sequence coverage. Peptides from each digest were fractionated by stepwise elution from reverse-phase beads with increasing concentrations of organic solvent (see Materials and Methods). Spectra were acquired from each elution, and m/z peaks from all elution steps of a single digest were combined for sequence analysis. In this way, coverage values obtained were above the typical sequence coverage for membrane proteins.

The matched m/z lists of Cx26 trypsin digests are shown in Table 2A . Submission of m/z values to the ProFound fingerprint database led to the unambiguous verification of Cx26 (as an unknown set of peptides, the z-score for rCx26 trypsin digest matching Cx26 in the database was 2.43; values>1 are significant). Sequence coverage was 43% for mCx26 and 70% for HeLa rCx26. Connexin domains detected in both trypsin digests were the extracellular loops (E1/E2), cytoplasmic amino-terminal domain (NT), and the middle and carboxyl-tail portions of the cytoplasmic loop (CL).


View this table:
[in this window]
[in a new window]
  		Table 2. Sequence coverage of Cx26a A.

The carboxyl-terminal domain (CT) of Cx26 is impossible to detect in complete trypsin digests due to its preponderance of acidic residues, leading to sequence tracts too short to be detected by MALDI-TOF-MS. Endo-GluC digestion of HeLa rCx26 provided complete coverage of CT and additional coverage in CL (Table 2B) . Sequence coverage by Endo-GluC digestion (17%) was too low to match significantly with Cx26 in the ProFound database. The combined coverage for Cx26 was 82%.

The matched m/z lists of Cx32 trypsin digests are shown in Table 3A . Submission of m/z values to the ProFound fingerprint database led to the unambiguous verification of Cx32 (z-scores for rCx32 trypsin digest and rCx32 Endo-GluC digest matching Cx32 as unknowns were 2.43 and 1.37, respectively). Sequence coverage was 61% for mCx32 and 47% for HeLa rCx32 with trypsin and 18% when using Endo-GluC. The combined coverage for Cx32 was 66%. Domains covered in both digests were the carboxyl-tail halves of E1 and E2, the entire NT, the middle portion of the first transmembrane domain (TM1), the middle of CL, and two fragments from the middle of CT. Digestion of mCx32 with Endo-GluC provided additional coverage of CL, the end of CT, and the middle of HeLa rCx32 CT (Table 3B) .


View this table:
[in this window]
[in a new window]
  		Table 3. Sequence coverage of Cx32a A.

Efforts to extend coverage by limited proteolysis were unsuccessful. The reasons for different sequence coverage of the same isoform purified from different sources are unclear.

Post-translational modification of Cx26 and Cx32
Residual m/z peaks not accounted for by specific, or potential nonspecific, cleavage of Cx26 and Cx32 (see Materials and Methods) were analyzed for matches to connexin peptides with post-translational modifications. Position(s) of any modified amino acid(s) in these peptide fragments could not be confirmed because of low peptide abundance and/or suppression and therefore the difficulty of peptide gating for MS/MS sequencing (14 , 15) .

The credibility of our assignments is based primarily on the recurrence of observation in different digests and the mass consistency of expected peptide fragments (error <20 ppm). We acknowledge the limitations of using single m/z peaks rather than MS/MS sequencing to identify post-translational modifications. However, false positives were very much minimized by the following strict criteria: Modifications were not considered if the modified peptides were not also found unmodified in the same digest, if residues carrying the modifications were at the ends of the peptides, or if the first amino acid or all amino acids involved were cysteine residues that were carboxyaminomethyl (CAM)- or PAM-modified (see Supplemental Materials). The post-translational modifications identified here are therefore highly likely and will serve as starting points for detailed further investigation. Also, as strict criteria were applied, those described here should not necessarily be regarded as complete.

Post-translational modifications have not been reported for Cx26, unlike for other connexins. Our analysis suggests several (Table 4A ). A NT peptide fragment (residues 1–15: MDWGTLQSILGGVNK) was present in mCx26 digests in hydroxylated form (positions D2 or N14), whereas the same peptide was phosphorylated in HeLa rCx26 (D2, T5, or S8). Furthermore, a fragment of the mCx26 CL (residues 113–122: NEFKDIEEIK) contained a {gamma}-carboxyglutamate (Gla) residue (one of E114, E119, or E120), which was absent in rCx26.


View this table:
[in this window]
[in a new window]
  		Table 4. Post-translational modifications of Cx26 and Cx32a A.

Post-translational modifications of Cx32 are listed in Table 4B . Those different in the mouse and HeLa digests are shaded. A NT peptide fragment (residues 1–15: MNWTGLYTLLSGVNR) was present in phosphorylated form in mCx32 (T4, Y7, T8, or S11), whereas the same fragment was either phosphorylated or hydroxylated (one of N2 or N14) in HeLa rCx32. No MALDI-TOF-MS peak was detected that would account for this rCx32 NT peptide carrying both modifications. Phosphorylation (H237, S233, or S240) was suspected in two peptides corresponding to residues 231-(238 and 232-)244 of the HeLa rCx32 CT (KGSGFGHRLSPEYK). A fragment of mCx32 CL (residues 108–122: LEGHGDPLHLEEVKR) contained two Gla-modifications (two of E109, E118, or E119). Furthermore, a palmitoylated CT peptide was detected in mCx32 (either C280 or C283 of residues 277–283: SDRCSAC).

These data may be consistent with each connexin being differently processed, with the modifications also depending on the cellular background. For all the modifications listed, except palmitoylation of mCx32, the peptides were also detected unmodified in the same digest, indicating that the modifications were only partial or were labile and lost during sample preparation or MALDI-TOF-MS. However, no estimate of the relative proportions of modified and unmodified forms can be obtained with the methods used in this study. The unmodified Cx32 CT peptide could not be detected because its molecular weight was below the m/z range set for spectrum acquisition.

DISCUSSION

pIs of Cx26 and Cx32 are more acidic than expected
Protein pIs can be calculated from sequence if each charged residue is treated independently, assuming that the ionization is not influenced by neighboring segments. Different algorithms predict slightly different pIs based on modestly different assumptions of amino acid pK values. For example, Compute pI/MW (us.expasy.org/tools/pi_tool.htm) predicts Cx26 and Cx32 pI at 9.20 (Table 1) , whereas ProFound (http://prowl.rockefeller.edu) predicts pIs 9.73 and 9.88, respectively. However, we determined the pIs of Cx26 and Cx32 to be significantly more acidic ({approx}3 pH units) than expected by computation (Fig. 1) . This suggests significant post-translational charge modifications of both connexins, although even minor charge addition, or charge elimination of target residues, might dramatically affect pI (26) .

Cx32 is a phosphoprotein (27 , 28) . Addition of 12 phosphate groups to rCx32 would shift its pI from 9.20 to the observed rCx32Tc pI of {approx}6.33, increasing molecular weight by only 935 Da (3%; http://ccr-224.mit.edu/calc_mw_pi.html). However, this extent of phosphorylation is unreasonable. Similarly, hyperphosphorylation of Cx26 is unreasonable, particularly as 13 phosphate groups would account for rCx26Tc pI (mass increase 1.10 kDa or 3.8%). Cx26 is not detectably phosphorylated by the same radioactivity or immunobased techniques applied to Cx32 (28 , 29) .

MALDI-TOF-MS suggested several novel post-translational modifications of Cx26 and Cx32 cytoplasmic domains. Suspected modifications included hydroxylation and/or phosphorylation in the NT of both connexins, {gamma}-carboxyglutamate residues in the CL of both isoforms, phosphorylation in the CT of Cx32, and palmitoylation of the carboxyl-terminus of Cx32 (Table 4) . Because of their small mass and anionic nature, most could contribute to the measured pIs of Cx26 and Cx32, without appreciable molecular weight changes.

Unfortunately, pKs for these modified residues are unavailable. However, the pIs of the processed proteins were calculated using protein sequences modified to reflect the charge changes arising from the suspected post-translational modifications. To estimate in silico the effect of hydroxylation of aspartic acid or asparagine (Cx26, D2 or N14; Cx32, N2, or N14) on expected pI, these residues were substituted in the primary sequences with two aspartic acids or aspartic acid, respectively, to simulate a deprotonable group. A similar approach was taken for hydroxylation of T4, Y7, T8, or S11 in Cx32. For {gamma}-carboxylation of glutamic acid (Cx26, one of E114, E119, or E120; Cx32, two of E109, E118, or E119), one additional glutamic acid residue was inserted to simulate the second acid group. For palmitoylation of Cx32, cysteine (C280 or C283) was replaced with alanine.

Recalculated pIs for rCx26 and rCx32 were 8.32 and 7.15, respectively, meaning only three additional phosphorylations could account for the observed pI of Cx32Tc. Even so, we cannot fully account for the acidic connexin pIs observed (especially of Cx26) solely using the MALDI-TOF-MS data. However, calculating expected pIs on sequence only has limitations. For example, net-charge algorithms do not take into account disulfide bond formation and protein-protein or electrostatic interactions within the protein that perturb ionization. The pIs of membrane proteins may differ to some extent from expected because charges in hydrophobic domains remain unexposed after detergent-binding (or are masked by associated lipid), which is typically required to maintain proteins such as Cx26 and Cx32 in "soluble" form (30 , 31) .

Therefore, some of the observed differences from expected in connexin pI may arise from these considerations. Remaining discrepancies could be attributed to post-translational modifications that failed to meet our stringent criteria (see Supplemental Materials) that were lost during sample preparation or MALDI-TOF-MS, or that were within sequence tracts not covered by MALDI-TOF-MS analysis (Tables 2 3) .

Progression and prospects toward purification of heteromeric channels with defined stoichiometry
Heteromeric Cx26/Cx32 channels do not have the same molecular selectivity as homomeric channels toward cAMP, cGMP (3 , 5) , and inositol phosphates (3 , 32) . They are functionally heterogeneous, presumably due to heterogeneities of isoform stoichiometry and/or arrangement. Fit of permeability data to statistical occurrences of heteromeric stoichiometry indicates cGMP permeation correlates with stoichiometry (3) , whereas isoform arrangement within a channel appears more important for cAMP and inositol phosphates (in preparation). More specific and mechanistic conclusions require working with defined isoform stoichiometry or arrangements.

Connexin channels have been purified from native tissues using immunopurification (3 , 5 , 17) and alkaline extraction of junctional plaques (33 , 34) or of plaques of hemichannels (35) . Homomeric channels are difficult to obtain since most native cells express more than one connexin and many coexpressed isoforms form heteromeric channels. Control over isoform ratios of heteromeric channels is limited with immunopurification (3 , 5 , 32) and nonexistent with alkaline extraction.

By exploiting the different pIs of tagged and/or tag-cleaved connexins, it was hoped that heteromeric channels of single stoichiometries could be obtained. Such fractionation has been achieved for polymers of ferritins using liquid-phase IEF (36) ; pIs of acidic and basic ferritin homopolymers bracketed the pIs of acidic-basic heteropolymers—directly corresponding to the separations shown here for homomeric and heteromeric connexin channels (Figs. 1 2 3) . The HA(HN)6-tag added {approx}0.7 pH units anionic charge to connexin pI. This enabled a greater pH range for separation of heteromeric stoichiometry than was achieved without the tag (Fig. 2) . However, the resolution of separation is limited by the Rotofor focusing core, because it lacks true separation barriers between chambers (37 , 38) . Additionally, salt and/or buffers transferred with connexin sample, bubbles, or minor leakage within the core of anodic and cathodic solutions affected gradient linearity. Therefore, from run-to-run the pH of the same focusing core chamber was also slightly different (Fig. 2 or 3 ).

Liquid-phase IEF, developed for nonintegral membrane proteins such as ferritins (37 , 38) , appears not to achieve the level of resolution of matrix-based techniques that permit the separations of protein that differ by single charge alterations (24 , 25) . However, liquid-phase IEF offers certain advantages with hydrophobic proteins such as Cx26 and Cx32, or protein-protein complexes. For example, high-quality resolution of membrane proteins by 2D-electrophoresis is difficult. Proteins with GRAVY scores >0.10, 0.15, and 0.30 from Saccharomyces cerevisiae, Bacillus subtilis, and Escherichia coli have never been detected by 2D-electrophoresis (39) . In contrast, these resolve easily by SDS-PAGE. Accordingly, the limit of separation of hydrophobic proteins in 2D-electrophoresis comes from the resolution provided by IEF in the first dimension.

Connexin GRAVY scores are 0.263 (mCx26), 0.273 (rCx26), and 0.194 (Cx32; http://au.expasy.org/tools/protparam.html). Attempts at 2D-electrophoresis of Cx26 and Cx32 were less than successful (13) , although superior reagents have enabled some progress (Hertzberg, E., unpublished observations). However, 2D-electrophoretic systems are almost always designed to provide the pIs of dissociated and denatured proteins, not of oligomers such as hemichannels.

Implications of acidic pI for the functions of Cx26 and/or Cx32
For most membrane proteins, acidic pI is usually attributed to glycosylation. However, connexins are not modified by carbohydrate. The discrepancy between the anticipated and actual pIs of Cx26 and Cx32 is therefore striking and must arise by other mechanisms. Other connexins may also have pIs unanticipated from amino acid sequence (Table 5 ). Post-translational modification(s), reflected by different pIs, might be involved in channel trafficking and assembly, or functional modulation. We discuss these possibilities based on our MALDI-TOF-MS findings, which infer several new post-translational modifications of Cx26 and Cx32.


View this table:
[in this window]
[in a new window]
  		Table 5. Predicted connexin isoelectric pointsa

Phosphorylation of Cx32 CT
Our studies suggest phosphorylation of the CT of Cx32 but not that of Cx26. We propose S233 as a phosphorylation site in Cx32 CT; in vitro phosphorylation of S233 was demonstrated by cAMP-dependent protein kinase and by PKC (27) . We also suggest that phosphorylation in the Cx32 CT can occur on S240 or H237. However, histidine phosphorylation is generally not detected by MALDI-TOF-MS due to the acid lability of the modified residue, making serine phosphorylation likely. NetPhos v2.0 (http://www.cbs.dtu.dk/services/NetPhos) also identifies S240 as a phosphorylation site.

NT modifications
The data also suggest phosphorylation in the NT of Cx26 and Cx32 (D2, T5, or S8 of Cx26, and T4, Y7, T8, or S11 of Cx32). No modifications have been reported in the NT of any connexin. The short NT of Cx26 and Cx32 ({approx}10 aa length) lack consensus phosphorylation sites. However, the sensitivity of MS-based approaches for detecting connexin phosphorylation is greater (fmol) than radioactivity or immunobased methods used previously (40) .

The data also suggest hydroxylation in the same NT peptide of Cx26 and Cx32 (D2 or N14, and N2 or N14, respectively). No m/z peak was detected that would account for the rCx32 NT peptide containing both hydroxylated and phosphorylated residues. Aspartate or asparagine (Asx) hydroxylation has been identified in a small group of proteins containing a NT epidermal growth factor-homology domain (41 , 42) .

Charged residues in the NT of Cx26 and Cx32 form part, if not all, of the transjunctional voltage sensor of the channels, which also directly affects ion permeation (43) . Charcot-Marie-Tooth disease mutations are found at Y7, T8, S11, and N14 of Cx32 (http://www.molgen.ua.ac.be/CMTMutations/default.cfm). There is a deafness mutation of Cx26 at S8 (http://davinci.crg.es/deafness/). In general, NT post-translational modifications may be novel regulatory mechanisms for fine-tuning of ionic or molecular permeation, and important new pharmacological targets for therapeutic intervention in connexin diseases.

Gla residues in the CL of Cx26 and Cx32
Peptide fragments were identified suggesting {gamma}-carboxyglutamate (Gla) modifications in the CL of Cx26 and Cx32 (one of E114, E119, or E120, and two of E109, E118, or E119). E120 is also a deafness mutation in Cx26.

Gla-residues co-ordinate with Ca2+ (Ca2+>Gla-residues) to enable domain-level membrane interaction, causing large conformational alterations in the entire protein (44 , 45) . Proteins possess 9–12 Gla residues within {approx}45 amino acids; a Gla-domain (46) . A cluster of hydrophobic amino acids is able to insert into the hydrocarbon region of the membrane, to a depth allowing Ca2+>Gla residues to interact with phospholipid head groups (44) . There are five glutamates between CL residues 99–139 of rCx26 and four between residues 96–130 in the CL of Cx32. However, there are nine glutamates within residues 99–153 of the rCx43 CL. If {gamma}-carboxylated, as we suggest for Cx26 and Cx32, this putative "Gla domain" could modulate Cx43 channel activity.

In fact, Ca2+-sensitive conformational changes are seen in individual Cx26 and Cx43 hemichannels (47 , 48) . It is worthwhile to speculate that the Ca2+-dependent open/closed pore conformations of connexin channels involve Ca2+>Gla residues. Additionally, Ca2+>Gla residues could play a role in the association of Cx32 with Ca2+-binding proteins such as calmodulin (49 , 50) .

Palmitoylation of Cx32 CT
A m/z peak was identified suggesting palmitoylation of the end of the CT of mCx32 (residues 277–283) but not Cx26 or HeLa rCx32. The CT of Cx32 has a carboxyl-terminal CSAC sequence, in which one or both cysteines could be palmitoylated (51) . Consistent with our MALDI-TOF-MS data, metabolic labeling of mCx32 with [3H]palmitic acid has been shown in cultured mouse hepatocytes (52) . Geranyl-geranylation of HeLa rCx32 was suspected, although excluded by our rigorous selection criteria (see Supplemental Materials). However, geranylation is consistent with the "unspecified prenylation" of Cx32 C280SAC283 in transfected COS cells (53) .

Lipid modifications may alter the affinity of membrane proteins for "lipid rafts." Rafts have functional relevance with respect to protein sorting and intracellular signaling interactions (54) . We observed subtle differences between Cx26 and Cx32 raft associations (55) . Mutations C280 and S281{Delta} are implicated in CMTX. Trafficking of these mutants is "normal"; channel permeability or its modulation is impaired. Raft localization may ensure proximity to protein and lipid modulators, permitting dynamic regulation of channel function(s) (56 , 57) . It is tempting to attribute the pathology of these mutants to membrane mis-localizations of pore regulatory sites within the CT.

CONCLUSION

pI differences of Cx26 and Cx32, which allowed homomeric and heteromeric channels to be separated by isoform content, pointed to significant post-translational modification of connexin. Several new post-translational modifications of Cx26 and Cx32 were identified, and differences in post-translational processing of the same isoform in liver and HeLa cells were also observed. These modifications may be instrumental in guiding and understanding novel molecular mechanisms of connexin regulation and disease.

ACKNOWLEDGMENTS

This study was supported by GM36044 and GM61406 to Andrew Harris. The Authors thank Jade Liu and Hong Li (UMDNJ) and especially Martine Cadene and Brian Chait (Rockefeller University). We also thank Elliot Hertzberg (Albert Einstein College of Medicine, NY) and Vincent Chen (University of Manitoba, Winnipeg, Canada) for critical comment on the manuscript.

FOOTNOTES

2 These authors contributed equally to this work.

Received for publication November 8, 2005. Accepted for publication February 2, 2006.

REFERENCES

  1. Wei, C. J., Xu, X., Lo, C. W. (2004) Connexins and cell signaling in development and disease. Annu. Rev. Cell Dev. Biol. 20,811-838[CrossRef][Medline]
  2. Cottrell, G. T., Burt, J. M. (2005) Functional consequences of heterogeneous gap junction channel formation and its influence in health and disease. Biochim. Biophys. Acta 1711,126-141[Medline]
  3. Locke, D., Stein, T., Davies, C., Morris, J., Harris, A. L., Evans, W. H., Monaghan, P., Gusterson, B. (2004) Altered permeability and modulatory character of connexin channels during mammary gland development. Exp. Cell Res. 298,643-660[CrossRef][Medline]
  4. Goldberg, G. S., Moreno, A. P., Lampe, P. D. (2002) Gap junctions between cells expressing connexin43 or 32 show inverse permselectivity to adenosine and ATP. J. Biol. Chem. 277,36725-36730[Abstract/Free Full Text]
  5. Bevans, C. G., Kordel, M., Rhee, S. K., Harris, A. L. (1998) Isoform composition of connexin channels determines selectivity among second messengers and uncharged molecules. J. Biol. Chem. 273,2808-2816[Abstract/Free Full Text]
  6. Veenstra, R. D. (1996) Size and selectivity of gap junction channels formed from different connexins. J. Bioenerg. Biomembr. 28,327-337[CrossRef][Medline]
  7. Dolphin, A. C. (2003) Beta-subunits of voltage-gated calcium channels. J. Bioenerg. Biomembr. 35,599-620[CrossRef][Medline]
  8. Korovkina, V. P., England, S. K. (2002) Molecular diversity of vascular potassium channel isoforms. Clin. Exp. Pharmacol. Physiol. 29,317-323[CrossRef][Medline]
  9. Solan, J. L., Lampe, P. D. (2005) Connexin phosphorylation as a regulatory event linked to gap junction channel assembly. Biochim. Biophys. Acta 1711,154-163[Medline]
  10. Laird, D. W. (2005) Connexin phosphorylation as a regulatory event linked to gap junction internalization and degradation. Biochim. Biophys. Acta 1711,172-182[Medline]
  11. Moreno, A. P. (2005) Connexin phosphorylation as a regulatory event linked to channel gating. Biochim. Biophys. Acta 1711,164-171[Medline]
  12. Hertzberg, E. L. (2005) Acetylation: A novel and important connexin modification with connexin-specific effects. Mol. Biol. Cell 15,185a
  13. Stockert, R. J., Spray, D. C., Gao, Y., Suadicani, S. O., Ripley, C. R., Novikoff, P. M., Wolkoff, A. W., Hertzberg, E. L. (1999) Deficient assembly and function of gap junctions in Trf1, a trafficking mutant of the human liver-derived cell line HuH-7. Hepatology 30,740-747[CrossRef][Medline]
  14. Zhang, N., Li, N., Li, L. (2004) Liquid chromatography MALDI-MS/MS for membrane proteome analysis. J. Proteome Res. 3,719-727[CrossRef][Medline]
  15. Amini, A., Dormady, S. J., Riggs, L., Regnier, F. E. (2000) The impact of buffers and surfactants from micellar electrokinetic chromatography on matrix-assisted-laser-desorption-ionization (MALDI) mass spectrometry of peptides. Effect of buffer type and concentration on mass determination by MALDI-time-of-flight mass spectrometry. J. Chromatogr. A 894,345-355[Medline]
  16. Koreen, I. V., Elsayed, W. A., Liu, Y. J., Harris, A. L. (2004) Tetracycline-regulated expression enables purification and functional analysis of recombinant connexin channels from mammalian cells. Biochem. J. 383,111-119[CrossRef][Medline]
  17. Rhee, S. K., Bevans, C. G., Harris, A. L. (1996) Channel-forming activity of immunoaffinity-purified connexin32 in single phospholipid membranes. Biochemistry 35,9212-9223[CrossRef][Medline]
  18. Sikerwar, S. S., Malhotra, S. K. (1983) A structural characterization of gap junctions isolated from mouse liver. Cell Biol. Int. Rep. 7,897-903[CrossRef][Medline]
  19. Mazet, F., Mazet, J. L. (1990) Restoration of gap junction-like structure after detergent solubilization of the proteins from liver gap junctions. Exp. Cell Res. 188,312-315[CrossRef][Medline]
  20. Konig, N., Zampighi, G. A. (1995) Purification of bovine lens cell-to-cell channels composed of connexin44 and connexin50. J. Cell Sci. 108,3091-3098[Abstract]
  21. Goodenough, D. A., Paul, D. L., Jesaitis, L. (1988) Topological distribution of two connexin32 antigenic sites in intact and split rodent hepatocyte gap junctions. J. Cell Biol. 107,1817-1824[Abstract/Free Full Text]
  22. Wessel, D., Flugge, U. I. (1984) A method for the quantitative recovery of protein in dilute solution in the presence of detergents and lipids. Anal. Biochem. 138,141-143[CrossRef][Medline]
  23. Krutchinsky, A. N., Kalkum, M., Chait, B. T. (2001) Automatic identification of proteins with a MALDI-quadrupole ion trap mass spectrometer. Anal. Chem. 73,5066-5077[Medline]
  24. Righetti, P. G., Castagna, A., Herbert, B., Reymond, F., Rossier, J. S. (2003) Prefractionation techniques in proteome analysis. Proteomics 3,1397-1407[CrossRef][Medline]
  25. Issaq, H. J., Conrads, T. P., Janini, G. M., Veenstra, T. D. (2002) Methods for fractionation, separation and profiling of proteins and peptides. Electrophoresis 23,3048-3061[CrossRef][Medline]
  26. Kumar, Y., Khachane, A., Belwal, M., Das, S., Somsundaram, K., Tatu, U. (2004) ProteoMod: A new tool to quantitate protein post-translational modifications. Proteomics 4,1672-1683[CrossRef][Medline]
  27. Saez, J. C., Nairn, A. C., Czernik, A. J., Spray, D. C., Hertzberg, E. L., Greengard, P., Bennett, M. V. (1990) Phosphorylation of connexin32, a hepatocyte gap-junction protein, by cAMP-dependent protein kinase, protein kinase C and Ca2+/calmodulin-dependent protein kinase II. Eur. J. Biochem. 192,263-273[Medline]
  28. Traub, O., Look, J., Dermietzel, R., Brummer, F., Hulser, D., Willecke, K. (1989) Comparative characterization of the 21-kD and 26-kD gap junction proteins in murine liver and cultured hepatocytes. J. Cell Biol. 108,1039-1051[Abstract/Free Full Text]
  29. Lampe, P. D., TenBroek, E. M., Burt, J. M., Kurata, W. E., Johnson, R. G., Lau, A. F. (2000) Phosphorylation of connexin43 on serine368 by protein kinase C regulates gap junctional communication. J. Cell Biol. 149,1503-1512[Abstract/Free Full Text]
  30. Kashino, Y. (2003) Separation methods in the analysis of protein membrane complexes. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 797,191-216[Medline]
  31. Tanford, C., Nozaki, Y., Reynolds, J. A., Makino, S. (1974) Molecular characterization of proteins in detergent solutions. Biochemistry 13,2369-2376[CrossRef][Medline]
  32. Ayad, W. A., Locke, D., Koreen, I. V., Harris, A. L. (2003) Heteromeric, but not homomeric, connexin channels are selectively permeable to inositol phosphates. J. Biol. Chem. doi: 10.1074/jbc.M600136200
  33. Stauffer, K. A., Kumar, N. M., Gilula, N. B., Unwin, N. (1991) Isolation and purification of gap junction channels. J. Cell Biol. 115,141-150[Abstract/Free Full Text]
  34. Buehler, L. K., Stauffer, K. A., Gilula, N. B., Kumar, N. M. (1995) Single channel behavior of recombinant beta-2-gap junction connexons reconstituted into planar lipid bilayers. Biophys. J. 68,1767-1775[Medline]
  35. Sosinsky, G. E., Perkins, G. A. (2001) Purification of gap junctions. Meth. Mol. Biol. 154,57-75[Medline]
  36. Rucker, P., Torti, F. M., Torti, S. V. (1997) Recombinant ferritin: Modulation of subunit stoichiometry in bacterial expression systems. Prot. Eng. 10,967-973[Abstract/Free Full Text]
  37. Bier, M. (1998) Recycling isoelectric focusing and isotachophoresis. Electrophoresis 19,1057-1063[CrossRef][Medline]
  38. Ayala, A., Parrado, J., Machado, A. (1998) Use of Rotofor preparative isoelectrofocusing cell in protein purification procedure. Appl. Biochem. Biotechnol. 69,11-16[Medline]
  39. Herbert, B. (1999) Advances in protein solubilisation for two-dimensional electrophoresis. Electrophoresis 20,660-663[CrossRef][Medline]
  40. Thaler, F., Valsasina, B., Baldi, R., Xie, J., Stewart, A., Isacchi, A., Kalisz, H. M., Rusconi, L. (2003) A new approach to phosphoserine and phosphothreonine analysis in peptides and proteins: Chemical modification, enrichment via solid-phase reversible binding, and analysis by mass spectrometry. Anal. Bioanal. Chem. 376,366-373[Medline]
  41. Lando, D., Peet, D. J., Whelan, D. A., Gorman, J. J., Whitelaw, M. L. (2002) Asparagine hydroxylation of the HIF transactivation domain: A hypoxic switch. Science 295,858-861[Abstract/Free Full Text]
  42. Linke, S., Stojkoski, C., Kewley, R. J., Booker, G. W., Whitelaw, M. L., Peet, D. J. (2004) Substrate requirements of the oxygen-sensing asparaginyl hydroxylase factor-inhibiting hypoxia-inducible factor. J. Biol. Chem. 279,14391-14397[Abstract/Free Full Text]
  43. Purnick, P. E., Benjamin, D. C., Verselis, V. K., Bargiello, T. A., Dowd, T. L. (2000) Structure of the amino terminus of a gap junction protein. Archiv. Biochem. Biophys. 381,181-190[CrossRef][Medline]
  44. Nelsestuen, G. L., Shah, A. M., Harvey, S. B. (2000) Vitamin K-dependent proteins. Vitam. Horm. 58,355-389[CrossRef][Medline]
  45. Bandyopadhyay, P. K., Garrett, J. E., Shetty, R. P., Keate, T., Walker, C. S., Olivera, B. M. (2002) gamma-Glutamyl carboxylation: An extracellular post-translational modification that antedates the divergence of molluscs, arthropods, and chordates. Proc. Natl. Acad. Sci. U. S. A. 99,1264-1269[Abstract/Free Full Text]
  46. Presnell, S. R., Tripathy, A., Lentz, B. R., Jin, D. Y., Stafford, D. W. (2001) A novel fluorescence assay to study propeptide interaction with gamma-glutamyl carboxylase. Biochemistry 40,11723-11733[CrossRef][Medline]
  47. Muller, D. J., Hand, G. M., Engel, A., Sosinsky, G. E. (2002) Conformational changes in surface structures of isolated connexin26 gap junctions. EMBO J. 21,3598-3607[CrossRef][Medline]
  48. Thimm, J., Mechler, A., Lin, H., Rhee, S., Lal, R. (2005) Calcium-dependent open/closed conformations and interfacial energy maps of reconstituted hemichannels. J. Biol. Chem. 280,10646-10654[Abstract/Free Full Text]
  49. Ahmad, S., Martin, P. E., Evans, W. H. (2001) Assembly of gap junction channels: Mechanism, effects of calmodulin antagonists and identification of connexin oligomerization determinants. Eur. J. Biochem. 268,4544-4552[Medline]
  50. Peracchia, C., Sotkis, A., Wang, X. G., Peracchia, L. L., Persechini, A. (2000) Calmodulin directly gates gap junction channels. J. Biol. Chem. 275,26220-26224[Abstract/Free Full Text]
  51. Linder, M. E., Deschenes, R. J. (2003) New insights into the mechanisms of protein palmitoylation. Biochemistry 42,4311-4320[CrossRef][Medline]
  52. Willecke, K., Traub, O., Look, J., Stutenkemper, R., Dermietzel, R. (1988) Different protein components contribute to the stucture and function of hepatic gap junctions. Hertberg, E. L. Johnson, R. G. eds. Gap Junctions 7,41-52 Alan R. Liss, Inc. New York.
  53. Huang, Y., Sirkowski, E. E., Stickney, J. T., Scherer, S. S. (2005) Prenylation-defective human connexin32 mutants are normally localized and function equivalently to wild-type connexin32 in myelinating Schwann cells. J. Neurosci. 25,7111-7120[Abstract/Free Full Text]
  54. Simons, K., Vaz, W. L. (2004) Model systems, lipid rafts, and cell membranes. Ann. Rev. Biophys. Biomolec. Struct. 33,269-295
  55. Locke, D., Liu, J., Harris, A. L. (2005) Lipid rafts prepared by different methods contain different connexin channels, but gap junctions are not lipid rafts. Biochemistry 44,13027-13042[CrossRef][Medline]
  56. O’Connell, K. M., Martens, J. R., Tamkun, M. M. (2004) Localization of ion channels to lipid raft domains within the cardiovascular system. Trend. Cardiovasc. Med. 14,37-42[CrossRef][Medline]
  57. Tillman, T. S., Cascio, M. (2003) Effects of membrane lipids on ion channel structure and function. Cell Biochem. Biophy. 38,161-190



This article has been cited by other articles:


Home page
 Toxicol SciHome page
P. Marx-Stoelting, J. Mahr, T. Knorpp, S. Schreiber, M. F. Templin, T. Ott, A. Buchmann, and M. Schwarz
Tumor Promotion in Liver of Mice with a Conditional Cx26 Knockout
Toxicol. Sci., June 1, 2008; 103(2): 260 - 267.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Summary
Right arrow Full Text (PDF)
Right arrow Supplemental Data
Right arrow All Versions of this Article:
fj.05-5309fjev1
fj.05-5309fjev2
fj.05-5309fjev3
20/8/1221    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Locke, D.
Right arrow Articles by Harris, A. L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Locke, D.
Right arrow Articles by Harris, A. L.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS