 |
INTRODUCTION |
Most cells in vertebrate tissues are coupled by gap junction
channels, which allow the transfer of low molecular mass substances (<1 kDa) between cells. The ability of cells to communicate in this
manner is thought to be important for coordinating tissue growth,
development, and physiological activities. Connexins are the sole
proteins required for the formation of gap junctions (1). Six connexin
monomers form a hemichannel (connexon) on the cell surface, which can
interact with a connexon from a neighboring cell, thus forming a
channel linking the cytoplasm of the two cells.
A number of different connexin isoforms are found in vertebrates. The
connexins characterized to date have distinct channel permeabilities
and gating properties (2, 3), as well as distinct cell and tissue
expression patterns, which is presumed to orchestrate a spatially and
temporally regulated diffusion of small molecules between cells. This
has recently been demonstrated by studies of genetically altered mice,
in which the expression of
Cx431 was replaced by
expression of Cx40 or Cx32. These mice differed functionally and
morphologically from wild type mice, demonstrating that one connexin
isoform cannot fully substitute for another (4). Regulation of the
diffusion of small molecules can be further modulated by combining
different connexin isoforms in each hemichannel (heteromeric gap
junctions) and by combining different hemichannels (heterotypic gap
junctions), thus creating channels with highly specific permeability
and gating properties.
All connexin isoforms are presumed to have a similar topology, which
has been deduced from limited proteolysis and site-directed antibodies
(5). The NH2-terminal and COOH-terminal domains are
localized in the cytoplasm, and are connected by four transmembrane domains, two extracellular loops, and a cytoplasmic loop. The major
diversity between different connexin isoforms is found in the length
and amino acid sequences of the cytoplasmic loop and COOH-terminal
domains. Two nomenclature systems are in use to designate members of
the connexin family; one is based on the theoretical molecular mass of
the proteins preceded by "Cx" (6), and the other is based on
sequence similarity analysis, in which clustering groups of connexins
are designated by Greek letters, and individual connexins in each group
by the order in which they were cloned (7). Most of the human connexins
characterized so far have orthologues in mice (the organism with the
most well characterized connexin expression), and presumably in all vertebrates.
Other proteins have been reported to interact with connexins,
e.g. v-Src and c-Src (8-10), occludin (11), tubulin (12), and zona occludens protein-1 (ZO-1) (13, 14). ZO-1 is a
membrane-associated guanylate kinase family member, which was first
found associated with tight junctions, and later shown to bind directly
to the tight junction components occludin (15) and claudins (16). Membrane-associated guanylate kinase proteins are thought to be involved in localizing cell surface proteins to specialized membrane domains (17). They contain a varying number of PDZ (PSD-95, DLG, ZO-1)
domains, a Src homology 3 domain, and an inactive guanylate kinase
domain. PDZ domains bind to other proteins, mainly to the most
COOH-terminal residues of the cell membrane proteins they recognize,
although internal binding sequences have also been identified (18). The
second PDZ domain of ZO-1 has been shown to bind the COOH terminus of
Cx43 (14). Recently, Cx45 was also shown to interact with ZO-1,
although it is not clear which domains are involved in this interaction
(19, 20).
The human genome sequencing project has revealed that 20 genes encoding
putative connexins exist in human
(21).2 However, at present it
cannot be excluded that some of these sequences represent pseudogenes.
Either the human genes themselves or their rodent homologues have been
shown to encode proteins for 17 of the 20 genes. The characterization
of these proteins, and especially their homozygous deletions in mice,
has provided valuable understanding of several human diseases caused by
mutations in connexin genes, such as congenital deafness,
Charcot-Marie-Tooth disease, erythrokeratodermia viabilis, congenital
cataracts, and congenital heart malformations (22). With the purpose of
understanding connexin function in further detail, we here report the
cloning and functional characterization of a novel human connexin,
Cx31.9, which we found expressed in vascular smooth muscle cells.
Furthermore, we present data suggesting that Cx31.9 interacts with
ZO-1.
| |
EXPERIMENTAL PROCEDURES |
Obtaining a Partial Cx31.9 cDNA and Genomic Clone
The human EST data base at NCBI (www.ncbi.nlm.nih.gov) was
searched using the BLAST algorithm (23) with published connexin sequences. An EST clone (accession number AA915885) with similarity to
connexins was obtained (Genome Systems, St. Louis, MO) and plasmid DNA
isolated (Qiagen, Hilden, Germany). The plasmid was sequenced with T3,
T7, and gene-specific primers (Invitrogen and MWG Biotech (High
Point, NC)) on both strands. The insert of clone AA915885 was excised
using EcoRI and NotI and used as a probe to
screen 1 × 106 clones from a human genomic liver
gt-10 library (Stratagene, La Jolla, CA). Four genomic clones
hybridized with the probe, and DNA was isolated from two of these.
The DNA was digested with several restriction enzymes and used for
Southern analysis with the AA915885 fragment probe. A 1949-bp
BamHI hybridization-positive fragment was subcloned in pGEM7
(Promega, Madison, WI) and sequenced on both strands.
5'-RACE
Human testis cDNA containing 5' and 3' adaptors (Marathon
cDNA, CLONTECH, Palo Alto, CA) was used as a
template in PCR with the primer PN115 (5'-TCAGCATGACCACCAGCCAG-3') from
nucleotides 91-72 of the Cx31.9 sequence and primer AP1
from the cDNA adaptors, with Expand High Fidelity PCR system (Roche
Molecular Biochemicals). PCR products were cloned into pGEM and
sequenced on both strands.
Northern Blot Analysis
A probe encompassing nucleotides 619-885 of the coding
sequence of Cx31.9 was generated by PCR using the primers
P32 (5'-ATCGAATTCGCCGAGCTGGGCCACCTGC-3') and P33
(5'-ATCCTCGAGCTAGATGGCCAGATCTCGGC-3'). Human MTN (multiple tissue
Northern) blot I containing poly(A+) RNA
(CLONTECH) was probed at 68 °C using ExpressHyb
buffer (CLONTECH) according to the instructions
from the manufacturer, and the blot was washed to a stringency of 0.1×
SSC, 65 °C. The blot was exposed to BioMax MS film (Eastman Kodak
Co.) for 48 h at 
80 °C.
Constructs
Cx31.9-COOH-terminal GST Fusion Protein--
The 1949-bp genomic
BamHI Cx31.9 fragment was used as a template in PCR
with the primers P32 (5'-ATCGAATTCGCCGAGCTGGGCCACCTGC-3') and P33
(5'-ATCCTCGAGCTAGATGGCCAGATCTCGGC-3'), using Pfu Turbo (Invitrogen). The resultant 284-bp fragment was
EcoRI/XhoI cloned into pGEX4T-1 (Amersham
Biosciences, Stockholm, Sweden). The sequence was verified by sequencing.
Cx31.9-loop GST Fusion Protein--
Using the BamHI
genomic fragment as a template for a PCR reaction with Expand
High Fidelity PCR system, with the primers P113 (5'-ATCGGATCCTACTCCATGCACCGGGCAGG-3') and P114
(5'-ATCGAATTCGTAGCAGCGGCGCGCGCG-3'), a product of 142 bp was
generated. This fragment was BamHI/EcoRI cloned
into pGEX2-TK (Amersham Biosciences), and the sequence verified.
Cx31.9 Untagged Expression Construct--
The genomic 1949-bp
BamHI fragment was cloned into BamHI-linearized
pcDNA-3 (Invitrogen). This expression vector contains a
cytomegalovirus promotor and a bovine growth hormone polyadenylation signal.
Cx31.9-eGFP Expression Construct--
A
SmaI/BglII genomic fragment containing the coding
region of Cx31.9 except the most 3' 11 nucleotides was
cloned in SmaI/KpnI-linearized pBluescript KS+/
(Stratagene) by the use of 5'-phosphorylated (5'P)
oligonucleotides with the sequences 5'P-GATCTGGCCATCGGGGTAC-3' and 5'P-CCCGATGGCCA-3' to link the BglII and
KpnI sites. This construct was excised from pBluescript
using SacI and KpnI and cloned into
SacI/KpnI-linearized eGFP-N1
(CLONTECH), to generate an open reading frame
encoding Cx31.9 with a COOH-terminal eGFP tag. The vector, eGFP-N1,
contains a cytomegalovirus promotor and a SV40 polyadenylation signal.
ZO-1 GST Fusion Proteins--
PDZ domains were amplified
by PCR using human ZO-1 cDNA and the following primers:
PDZ-1F (5'-GGGGGTCGACAATGGAGGAAACAGCTATATGG-3'), PDZ-1R
(5'-CCCCGAATTCCTAACGACTTACTGGTATGTT-3'), PDZ-2F
(5'-GGGGGTCGCAGTGGTCTCCAGCCAGCTT-3'), PDZ-2R
(5'-CCCCGAATTCCTAATCAAGGACATTCAATAG-3'), PDZ-3F
(5'-GGGGGTCGACAATGAAGATGGGATTTCTTCGG-3'), and PDZ-3F
(5'-CCCCGAATTCCTATTCTACAATGCGACGATA-3'). PCR products were
digested with SalI and EcoRI, subcloned and
sequenced, followed by ligation into
BamHI/EcoRI-linearized pGEX-3X (Amersham Biosciences).
Cx31.9 Xenopus Expression Construct--
The 1949-bp
BamHI fragment containing Cx31.9 was cloned into
BamHI-linearized pXLII (paw2pla). This vector was derived
from pXBG-ev1, a derivative of pBS KS containing the 5'- and 3'-UTR regions of the Xenopus laevis 
-globin gene
with an internal multiple cloning site. The correct orientation of the
insert was verified by sequencing.
Generation of mAbs against Cx31.9
GST fusion proteins were obtained using standard procedures.
Aliquots (50 µg) were used to immunize female BALB/C mice
intraperitoneally with Freund's complete adjuvant for the first
immunization, and Freund's incomplete adjuvant (Sigma) for the
following three immunizations. Three days after the final boost, mice
were sacrificed, and spleen cells fused to mouse myeloma cell line
X63Ag. Supernatants from the resultant hybridomas were screened against
the GST fusion proteins by enzyme-linked immunosorbent assay, and
positive wells were cloned and rescreened till monoclonality.
Hybridomas were expanded in Hybridoma-SFM (serum-free medium;
Invitrogen), and antibodies isolated with HiTrap Protein G columns
(Amersham Biosciences) according to the instructions from the
manufacturer. The use of mice for these experiments was approved by the
Scripps Research Institute Animal Committee.
Generation of Cx31.9-overexpressing Cell Lines
Human embryonic kidney (HEK) 293 cells were grown in Dulbecco's
modified Eagle's medium (DMEM; Invitrogen) containing 10% FCS, 2 mM L-glutamine, 100 units/ml penicillin, and
100 µg/ml streptomycin in a 37 °C incubator with a moist
atmosphere of 4% CO2. The cells were transfected with the
Cx31.9 untagged expression construct using Superfect (Qiagen) according
to the instructions from the manufacturer. After 24 h, the medium
was exchanged to selective medium (0.8 mg/ml Geneticin (Invitrogen))
for 10 days. Surviving clones were then picked, expanded, and lysates
screened by immunoblot with mAb 7G6 for expression of the connexin.
Positive clones were reseeded, repicked and rescreened three or four
times to obtain stable expression.
Immunoprecipitation Analyses
Cell lines were metabolically labeled overnight by incubation
with 100 µCi of Tran35S-label (ICN, Costa Mesa, CA) in 5 ml of DMEM without L-methionine and
L-cysteine/L-cystine (ICN), supplemented with
10% dialyzed FCS. Cells were then washed with PBS, and lysed in IP
buffer (50 mM Tris, 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, 0.25% deoxycholate, pH 8.0) or
RIPA buffer (1 M NaCl, 2 mM EDTA, 0.1% SDS,
1% Triton X-100, 100 mM Tris-HCl, pH 7.4). Following
incubation on ice for 15 min, lysates were clarified and incubated with
primary antibodies bound to Protein A/G beads (Calbiochem, La Jolla,
CA) (5 µg of antibody/IP) overnight at 4 °C. The beads were then
washed extensively in IP buffer and proteins separated by SDS-PAGE,
followed by staining with GelCode (Pierce) to visualize the amount of
antibody. Gels were dried and exposed to BioMax MS film at 80 °C
for 12 h.
SDS-PAGE and Immunoblot Analyses
Samples were separated on 10 or 12.5% SDS-PAGE gels in Hoefer
vertical gel apparatuses, followed by transfer to Protran 0.2-µm pore
size nitrocellulose membranes (Schleicher & Schuell). Membranes were
stained with 0.2% Ponceau S in 1% acetic acid, blocked with 5%
skimmed milk powder in TBST, and incubated with purified primary antibody (5-10 µg/ml) in blocking buffer. Membranes were then washed
in TBST, followed by incubation with horseradish peroxidase-conjugated secondary antibody (Bio-Rad). The antibodies were detected with chemiluminescence (SuperSignal West Pico, Pierce) by exposure to Biomax
ML film (Eastman Kodak).
Human Samples
Human tissues were obtained from the National Cancer Institute
Cooperative Human Tissue Network and from the National Disease Research
Interchange. Surgical specimens of human brain, large intestine, heart,
skeletal muscle, and aorta were homogenized in buffer containing 150 mM sucrose, 15 mM HEPES, 60 mM KCl,
5 mM EDTA, 1 mM EGTA, pH 7.9, and protease
inhibitors. Following homogenization, Triton X-100 was added to a final
concentration of 1%, and the homogenate incubated on ice for 1 h.
The lysate was then centrifuged at 18,000 × g at
4 °C, and the supernatant used for SDS-PAGE and immunoblot analysis.
The use of human material for these studies was approved by the Scripps
Clinic, Scripps Green Hospital, and the Scripps Research Institute
Human Subjects Committee.
Immunofluorescence: Cell Lines
Cells were grown on poly-L-lysine (Sigma)-treated
glass coverslips, washed in PBS, and fixed in 20 °C acetone or
methanol for 6 min. Coverslips were blocked with 5% goat serum in PBS
(blocking buffer), followed by incubations with primary antibodies
overnight in blocking buffer at 4 °C. mAbs 7G6 and 5G11 were used at
5-10 µg/ml, and a commercial anti-ZO-1 polyclonal antibody (Zymed
Laboratories, South San Francisco, CA) was used at 5 µg/ml. Following
washing with PBS, coverslips were incubated with fluorochrome-labeled secondary antibodies (Southern Biotech Associates Inc., Birmingham, AL), diluted 1:100, together with 50 nM To-Pro-3 to stain
nuclei (Molecular Probes, Eugene, OR), in blocking buffer for 1 h.
Washed slides were mounted with Fluoromount-G (Sigma), and images were collected using a Zeiss Axiovert confocal microscope. Surgical human
tissue samples were frozen and imbedded in O.C.T. Four-µm sections
were cut on a cryostat and attached to Superfrost slides at 20 °C.
Tissues were fixed in 4% paraformaldehyde in PBS for 2 h at
4 °C, followed by permeabilization with 0.1% Triton X-100 in PBS
for 20 min at room temperature. Blocking, antibody incubations, and
imaging were as described above. mAb 1E12 (IgM) against smooth muscle 
-actinin was developed by Dr. Charles D. Little and
co-workers, and mAb P2B2 (IgG1) against PECAM was developed by Drs.
Elizabeth A. Wayner and Gregory Vercellotti. These mAbs were obtained
from the Developmental Studies Hybridoma Bank developed under the
auspices of the NICHD and maintained by the Department of Biological
Sciences, University of Iowa (Iowa City, IA). For the co-localization
studies, mAb 7G6 was detected with an anti-mouse 
chain-specific
secondary antibody, which was cross-absorbed on mouse IgM, whereas mAb
1E12 was detected with a mouse IgM-specific antibody, which was
cross-absorbed on mouse IgG (Southern Biotechnology Associates,
Inc.).
Microinjections
Glass pipettes containing a capillary filament were pulled on a
Brown-Flaming micropipette puller (Sutter Instrument Co., San
Francisco, CA). Overnight cultures of Cx31.9-HEK cells and wt HEK cells
were microinjected (Transjector 5246, Eppendorf, Madison, WI) with 4%
Lucifer yellow, carboxyfluorescein, cascade blue (Molecular Probes,
Eugene, OR), or 5% ethidium bromide (Sigma) in 1 M LiCl,
including 2% Texas Red or FITC-labeled dextran (Molecular Probes,
Inc.) to visualize the injected cell. The spread of dyes was monitored
in a Zeiss Axiovert 100 fluorescent microscope.
Electrophysiology of Cx31.9 Expressed in Xenopus Oocytes
The Xenopus expression construct was linearized with
SacI, and cRNA was generated by in vitro
transcription with AmpliScripe T3 transcription kit in conjunction with
a monomethylated cap analog m7G[5']ppp[5']G (Epicentre
Technologies, Madison, WI). The size of the transcript was verified,
and the cRNA concentration estimated, by denaturing agarose gel
electrophoresis. Oocytes were isolated from X. laevis
(Pacific Biological, Sherman Oaks, CA) ovaries by collagenase
treatment. Stage V-VI oocytes were selected and manually
defolliculated if necessary. After a 24-h incubation in ND-96 medium to
ensure viability, a Nanoject apparatus (Drummond, Broomall, PA) was
used to inject oocytes with 46 nl of 0.05-0.5 µg/µl Cx31.9 cRNA
and 5 ng of Cx38 antisense oligonucleotide (5'-GCTTTAGTAATTCCCATCCTGCCATGTTTC-3') to block endogenous connexin synthesis (24). Injected oocytes were incubated for 48 h, followed by manual removal of the vitelline layers. Oocytes were paired at their
vegetal poles for 2-6 h, and junctional conductance between oocytes
was measured with the dual voltage clamp technique as previously
described (25), using dual Gene Clamp 500Bs driven by a D/A board under
the control of Pclamp 8.0 data acquisition software (Axon Instruments,
Foster City, CA). Each oocyte in a pair was initially clamped to a
potential equal to the average of the two resting potentials, and one
oocyte was stepped to a series of voltages to generate a series of
transjunctional potentials. Junctional currents were recorded as the
currents induced in the non-driven oocyte.
Electrophysiology of Cx31.9 Overexpressed in HEK Cells
A dual cell voltage clamp method was used (26-28).
Poly-L-lysine-coated coverslips containing cells were
mounted on the stage of an inverted Nikon Diaphot fluorescent
microscope equipped with Hoffman modulation optics. The cells were
perfused with Krebs-Ringer solution (140 mM NaCl, 4.5 mM KCl, 2 mM CaCl2, 1 mM MgCl2, pH 7.2) at room temperature
immediately prior to each experiment. Patch pipettes were pulled from
glass capillary tubes with filaments to a resistance of 1-2 megohms.
The pipettes were filled with solution containing 160 mM
potassium aspartate, 1.1 mM EGTA, 2 mM
MgCl2, 0.1 mM CaCl2, 10 mM Hepes, pH 7.2. Data were collected using an ADLab D/A
board driving two Axopatch (1B and C) patch clamps (Axon Instruments)
in voltage-clamp mode. Voltage protocols and data collection were
controlled by home-made software. For evaluation of junctional
conductance, both cells were held at 40 mV, and 50-ms pulses to 20
mV were alternately applied to each cell. Cells examined were generally
in contact only with each other, but occasionally two adjacent cells in
a chain were selected. The electrical conductance was calculated as the
junctional current divided by the potential difference between the two
cells. In some cases, IV curves were generated by stepping the voltage in one cell while holding the other cell at a fixed value.
ZO-1/Cx31.9 Co-immunoprecipitation
Cx31.9-HEK and wt HEK cells were treated as above
(immunoprecipitation analysis), without the radiolabel. The IP buffer
has been optimized by Giepmans and co-workers (14) to retain the interaction of Cx43 and ZO-1, while minimizing nonspecific
interactions. Following immunoblot transfer, ZO-1 was detected using a
commercial polyclonal antibody (Zymed Laboratories). For the detection
of immunoprecipitated Cx31.9, mAb 7G6 was biotinylated using
NHS-PEO4-Biotin (Pierce) according to the instructions from
the manufacturer. The biotinylation was done to avoid reactivity of
secondary anti-mouse antibody with the light chains of antibodies used
for the immunoprecipitation. Biotinylated 7G6 was detected using
Vectastain Elite ABC kit with peroxidase (Vector Laboratories, Inc.,
Burlingame, CA) and chemiluminescence, according to the instructions
from the manufacturer.
ZO-1-PDZ Pull-down Assay
PDZ-GST fusion proteins were prepared using standard procedures
and bound to glutathione-agarose (Sigma). The agarose beads containing
fusion proteins were then incubated with IP buffer lysates from a
confluent 85-mm dish of Cx31.9-HEK cells overnight. After washing in IP
buffer, the beads were eluted with SDS-PAGE sample buffer, separated by
SDS-PAGE, and transferred to nitrocellulose membrane. Cx31.9 was
detected with mAb 7G6.
| |
RESULTS |
Molecular Cloning of Cx31.9--
The human GenBankTM
EST data base was searched with connexin motifs using the BLAST
algorithm, and a number of clones encoding partial, putative novel
connexins were identified. One of these clones, accession number
AA915885 (derived from a human colon adenocarcinoma library), was
purchased and fully sequenced in both directions by primer walking. The
insert contained a sequence of 1499 bp, which consisted of a putative
5'-UTR (659 bp), an ATG codon flanked by G in position +4 and A in 3,
which was predicted to be a strong initiator of translation (29), and
an open reading frame (840 bp). However, the open reading frame was not
complete, but was followed by a NotI site, indicating that
this was a partial clone formed by restriction digestion of the
cDNA during library synthesis. The sequence of clone AA915885 was
also similar to human EST clone AI142991 (from a human senescent
fibroblast library).
Clone AA915885 was used to screen a human genomic library, and four
clones were identified. Restriction digests from these clones were
analyzed by Southern blots, and a 1949-bp BamHI fragment was
subcloned and fully sequenced in both directions. This genomic fragment
contained an open reading frame identical to clone AA915885 and
additional coding sequence including a stop codon (Fig.
1). The open reading frame encoded a
putative 294-amino acid protein, with a predicted molecular mass of
31,933. This protein will be referred to as Cx31.9 in accordance with the nomenclature previously proposed for connexins (6).
View larger version (48K):
[in this window]
[in a new window]
|
Fig. 1.
Nucleotide and deduced amino acid sequence of
Cx31.9. The amino acid sequence is shown in bold
single-letter code. The nucleotide sequence is
derived from a 1949-bp BamHI genomic fragment obtained from
a human genomic clone. The remaining 815 nucleotides downstream
from this fragment are not shown, but are included in the sequence
deposited in GenBankTM under accession number AY093445.
Human EST clone AA915885 terminated at the NotI site
(position 837), which is indicated by the box. Hydropathy
analysis by the method of Kyte and Doolittle (window size: 17)
indicated the presence of four putative transmembrane domains
(underlined). Cysteine residues conserved among all
connexins (except Cx31) in the putative extracellular domains are
circled.
|
|
A comparison of the Cx31.9 genomic and EST clone revealed
that no introns were present in the coding sequence of Cx31.9, as is
the case for all connexins except Cx36 (30) and
Cx40.1.3 Furthermore, no
introns were found in the 249 bp upstream of the initiating ATG codon
contained in the genomic clone. This was surprising, because all
mammalian connexins characterized to date contain one, or two (Cx45,
Ref. 31), intron(s) in the 5'-UTR in close proximity to the initiating
ATG codon. To verify that clone AA915885 had been properly spliced,
testis cDNA was used for 5'-RACE. The major PCR product contained
859 bp upstream of the initiating ATG codon of Cx31.9. This
sequence also aligned with the most 5' 659 bp of clone AA915885,
confirming that there was no intron in this region (data not shown).
However, putative strong splice acceptor sequences were located at
position 5, and potential lariat forming branch sites were found at
positions 25 and 42. Furthermore, a weaker splice acceptor was
found at position 53, with a potential branch site at position 70.
This raises the possibility that the 5' region may be spliced under certain circumstances.
During the course of this study, a human genomic sequence from a BAC
clone containing the Cx31.9 gene was deposited in the GenBankTM data base under accession number AC090426. This
clone has been localized to chromosome 17q21.1, a locus that does not
contain any of the other connexin genes present in the human genome.
The 859-bp 5'-UTR obtained by 5'-RACE aligns with the genomic sequence of this clone, consistent with the lack of introns in this region (data
not shown).
Interestingly, another transcript appears to initiate within the 5'-UTR
of Cx31.9. When the genomic sequence upstream of the Cx31.9 coding
region was searched against the EST data base, a number of EST clones,
e.g. BG027595, were found to have their most 5' end 459 bp
upstream of the initiating ATG codon of Cx31.9, thus within
the Cx31.9 5'-UTR. This mRNA was transcribed in the opposite direction of Cx31.9.
A human EST clone from an esophagus squamous cell carcinoma library
(GenBankTM accession number AW08188) was identified
corresponding to nucleotides 1081-1632 of the Cx31.9
genomic BamHI fragment, suggesting that Cx31.9
was expressed in several tissues.
Analyses of the Cx31.9 Protein Sequence--
Hydrophobicity
analysis of Cx31.9 suggested the presence of four putative
transmembrane spanning domains (Fig. 1) similar to other connexins. The
spacing of cysteine residues in the two extracellular domains,
C(X6)C(X3)C (residues
54-65), and C(X4)C(X5)C (residues 168-179) are conserved in all connexins (except Cx31). Several consensus motifs for potential posttranslational modifications were identified by searching the Cx31.9 sequence against the PROSITE data base. Four potential PKC ((S/T)X(R/K)) phosphorylation
sites were identified in the COOH terminus at residues 252, 272, 278, and 287, as well as a casein kinase II (SXXE)
phosphorylation site at residue 272. To compare the sequence of Cx31.9
to other connexins, we first searched the GenBankTM human
genomic data base for all connexin genes present in the human genome.
In total, 20 putative human connexins were identified, several of which
have not been previously characterized.2 These 20 genes
appear to be similar to those recently reported by Willecke et
al. (21). The overall amino acid sequence similarity between
Cx31.9 and the other human connexins ranged from 48 to 60%, with
32-41% identical residues. The connexins most similar to Cx31.9 were
members of the -group of connexins. To elucidate the sequence
relationship of Cx31.9 with its human paralogues, we aligned the
connexins using ClustalW, and compared the conserved regions by
phylogenetic analysis. Using this method, Cx31.9 was also found to
group with the -group of connexins (Fig.
2) and was designated connexin 11.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 2.
Phylogenetic tree of human connexins.
Figure shows phylogenetic sequence analysis of 20 putative human
connexins identified from the GenBankTM genomic data base.
Cx31.9 is underlined, and was found to cluster with the
-group of connexins. The tree was obtained by analysis of the
conserved regions between connexins. Sequences were aligned using
ClustalW, and the average similarity between aligned sequences was
calculated for each sequence segment (using a previously described
algorithm; see saier-144-37.ucsd.edu/biotools/avehas.html). Sequence
segments having less than 25% similarity were excluded from the
analysis (most of the intracellular loops and COOH-terminal
domains).
|
|
Northern Analysis--
A probe encompassing the 3' end of the
Cx31.9 coding region, which showed no homology to other connexins, was
used for Northern blot analysis of human poly(A)+ RNA. The
major signal detected was a 4.4-kb band, which was seen in all tissues
examined, although with varying intensities (Fig. 3, arrowhead). The highest
expression levels were found in heart and pancreas. Lower intensity
bands were also detected at 1, 0.5, and 0.4 kb. Although we cannot
exclude that these bands represent nonspecific hybridization of the
probe, this seems unlikely because of their correlation with the
intensity of the 4.4-kb transcript. Although most connexin transcripts
migrate as single bands in Northern analysis, more than one transcript
size has also been detected for human Cx45 (32) and mouse Cx31 (33).
However, because of the small size of the Cx31.9 bands, they
were more likely generated by degradation of the RNA than by
alternative splicing. Reprobing of the blot with a probe against
-actin showed that comparable amounts of poly(A)+ RNA
were loaded in all lanes (Fig. 3, lower panel).
View larger version (54K):
[in this window]
[in a new window]
|
Fig. 3.
Northern blot analysis of Cx31.9
expression. A CLONTECH MTN I blot was probed
with the 3' end of the Cx31.9 coding sequence. The major signal
detected was a 4.4-bp band (arrowhead), which was seen in
all tissues after long exposure of the blot (data not shown). Reprobing
for -actin confirmed that comparable amounts of mRNA were loaded
(lower panel). The faster migrating forms of -actin
observed in heart and skeletal muscle are splice variants previously
described (34).
|
|
Generation of mAbs against Cx31.9--
The intracellular loop
(residues 95-136; Cx31.9 L-GST), and COOH-terminal domain (residues
207-294; Cx31.9 Ct-GST) of Cx31.9 were expressed as GST fusion
proteins. The fusion proteins showed a decrease in SDS-PAGE mobility
compared with GST alone, which corresponded approximately to the
predicted mass of the domains (Fig.
4A, lower panel).
In Cx31.9 L-GST, a second band migrating faster than GST alone was also
observed. This band likely represented a cleaved form of the fusion
protein, although other possibilities, such as a co-purified
interacting protein, cannot be excluded.
View larger version (44K):
[in this window]
[in a new window]
|
Fig. 4.
Characterization of mAbs 7G6 and 5G11 against
Cx31.9. A, mAb 7G6 and 5G11 immunoblot analysis of
Cx31.9-GST fusion proteins. The proteins analyzed were as follows:
lanes 1 and 4, fusion proteins containing the
COOH terminus of Cx31.9 (Cx31.9 Ct-GST); lanes 2 and
5, GST; lanes 3 and 6, the loop region
of Cx31.9 (Cx31.9 L-GST). mAb 7G6 only reacted with Cx31.9 Ct-GST,
whereas mAb 5G11 only reacted with Cx31.9 L-GST (upper
panels). Ponceau staining of the blots showed that the Cx31.9
domains changed the SDS-PAGE mobility of GST fusion proteins as
expected (lower panels). B, immunoprecipitation
of 35S-labeled Cx31.9 from overexpressing HEK cells. Cells
were metabolically labeled and lysed in buffer containing 1% Nonidet
P-40, 0.25% deoxycholate. mAb 7G6, 5G11, and an irrelevant mAb (5 µg
each) were used for the immunoprecipitations. Lysates were analyzed to
verify uniform labeling (lanes 1 and 5). Cx31.9
was immunoprecipitated by both mAb 7G6 and mAb 5G11 from Cx31.9-HEK
cells, but not from wt cells (black arrowhead). A 25-kDa
protein was also present in mAb 5G11 immunoprecipitates from Cx31.9-HEK
cells (white arrowhead), possibly representing a proteolytic
fragment of Cx31.9 lacking the COOH-terminal domain.
|
|
mAbs were generated against these fusion proteins. Six
hybridomas that produced antibodies reactive with Cx31.9
Ct-GST were identified and found to react similarly in all assays. Only
one, mAb 7G6 (IgG), was further characterized. mAb 7G6 reacted with Cx31.9 Ct-GST (Fig. 4A, upper panel, lane
1) but not with GST (Fig. 4A, upper panel,
lane 2) or Cx31.9 L-GST (Fig. 4A, upper panel, lane 3) in immunoblot analysis. From the fusion
of the mouse immunized with Cx31.9 L-GST, only one hybridoma that
secreted an antibody reactive with Cx31.9 L-GST was identified. This
mAb, 5G11 (IgG), reacted with Cx31.9 L-GST (Fig. 4A,
upper panel, lane 6), but not with GST or Cx31.9
Ct-GST alone (Fig. 4A, upper panel, lanes
4 and 5) in immunoblot analysis.
Characterization of mAbs 7G6 and 5G11 by
Immunoprecipitation--
To further determine the specificity of mAbs
7G6 and 5G11, these mAbs were used in immunoprecipitation experiments
with HEK 293 cells stably overexpressing Cx31.9, and wt HEK cell
lysates. A small aliquot of radiolabeled lysate from each cell line was analyzed by SDS-PAGE and autoradiography. The lysates were found to
contain equal amounts of radiolabel and showed uniform labeling of
proteins (Fig. 4B, lanes 1 and 5).
Both mAb 7G6 and mAb 5G11 were able to immunoprecipitate a protein
migrating at 30-33 kDa from Cx31.9-HEK cells (Fig. 4B,
lanes 2 and 3, black arrowhead). No
proteins with this mass were immunoprecipitated from wt HEK cells (Fig.
4B, lanes 6 and 7), suggesting that
the immunoprecipitated protein from Cx31.9-HEK cells was indeed Cx31.9.
Furthermore, Cx31.9 was not immunoprecipitated using the same amount of
an irrelevant mouse antibody, indicating that mAbs 7G6 and 5G11 were both specific for Cx31.9. This experiment was also carried out using
RIPA buffer. mAb 7G6 immunoprecipitated Cx31.9 equally efficient using
both buffers, whereas very little Cx31.9 was immunoprecipitated with
mAb 5G11 in RIPA buffer (data not shown).
A protein migrating at 25 kDa was also immunoprecipitated with mAb 5G11
from Cx31.9-HEK cells, but not from wt HEK cells, and not by mAb 7G6
(Fig. 4B, white arrowhead). This protein possibly represented a cleaved form of Cx31.9, which lacked the COOH-terminal domain, and therefore was not recognized by mAb 7G6.
Immunoblot Analyses of Cx31.9 in Overexpressing Cells and Human
Tissues--
A fusion protein of Cx31.9 with eGFP as a COOH-terminal
tag (Cx31.9-eGFP), and untagged Cx31.9 were transiently expressed in
HeLa cells. Lysates from these cells, and lysates from Cx31.9-HEK and
wt HEK cells, were analyzed by immunoblot with mAb 7G6. A protein
migrating at 30-33 kDa was detected in Cx31.9-transfected cells (Fig.
5A, lanes 2 and
5, arrowhead), but not in wt HeLa or wt HEK cells
(Fig. 5A, lanes 1 and 4). wt HEK cells
transiently expressing Cx43 from the same expression vector also
contained no protein in this range reactive with mAb 7G6 (data not
shown). In Cx31.9-eGFP-expressing cells, a protein migrating at 53 kDa (Fig. 5A, lane 3, arrow) was detected.
The migration of this protein was within the range expected from the
composite weight of Cx31.9 and eGFP (58.9 kDa). The amount of protein
loaded was visualized by reprobing the blot with an antibody against
-actin, which showed that comparable amounts of protein were loaded
in all lanes (Fig. 5 A, lower panel).
View larger version (60K):
[in this window]
[in a new window]
|
Fig. 5.
Immunoblot detection of Cx31.9 from
overexpressing cells and human colon. A, mAb 7G6 immunoblot
of cell line lysates (upper panel). wt HeLa cells
(lane 1), HeLa transiently overexpressing Cx31.9
(Cx31.9eGFP) (lane 2), HeLa cells transiently overexpressing
Cx31.9 with a COOH-terminal eGFP tag (lane 3), wt HEK cells
(lane 4), and HEK cells stably overexpressing Cx31.9
(lane 5) were analyzed. Cx31.9 was observed migrating at
30-33 kDa (arrowhead). Cx31.9eGFP was also detected, but
with slower mobility because of the eGFP tag (arrow).
Reprobing of the blot for actin showed comparable amounts of protein in
all lanes (lower panel). B, immunoblot with mAb
7G6 from human tissues. Lysates from heart (lane 1), colon
(lane 2), artery (lane 3), brain (lane
4), and skeletal muscle (lane 5) were analyzed. Cx31.9
was detected in heart, colon, and artery, but not in brain and skeletal
muscle (arrowhead). Reprobing of the blot for actin showed
comparable amounts of protein in all lanes (lower
panel)
|
|
When mAb 5G11 was applied to similar immunoblots, no proteins were
detected despite applying large amounts of antibody (data not shown).
This suggested that the mAb 5G11 epitope was
conformation-dependent and was destroyed by SDS when
present in intact Cx31.9, but not when expressed as the Cx31.9 L-GST
fusion protein. The low immunoprecipitation efficiency of mAb 5G11 in
RIPA buffer was consistent with this hypothesis.
To investigate the expression of Cx31.9 in human tissues, immunoblots
containing lysates from human tissues were analyzed using mAb 7G6.
Because of the difficulties encountered in obtaining human tissue
samples of adequate quality, only a limited number of tissues were
available for this study. A protein migrating at 30-33 kDa was
detected in human heart, colon, and artery, but could not be detected
in human brain and skeletal muscle (Fig. 5B, upper
panel, arrowhead). The blot was reprobed with an
antibody against actin, which demonstrated that similar amounts of
lysates had been analyzed (Fig. 5B, lower panel). A similar
blot probed with an irrelevant mouse IgG antibody revealed no bands
migrating in this range (data not shown).
Immunofluorescence of Cx31.9-overexpressing HEK Cells--
wt
HEK cells transiently expressing Cx31.9-eGFP, or eGFP alone, were
analyzed by fluorescent confocal microscopy. In a low proportion of
cell pairs expressing Cx31.9-eGFP, structures resembling gap junction
plaques were observed (Fig.
6A, arrow). In
contrast, cells expressing eGFP alone showed diffuse fluorescence in
intracellular compartments (Fig. 6C).
View larger version (86K):
[in this window]
[in a new window]
|
Fig. 6.
Cx31.9 form structures resembling gap
junctions in overexpressing HEK cells. These confocal images show
either eGFP fluorescence (panels A-D) or FITC
immunofluorescence (panels E-L). Each image in the first
row (panels A, E, and I) is combined
with a nuclear stain in the row beneath to identify cells
(panels B, F, and J). The images in
the third row (panels C, G,
and K) are combined with a nuclear stain in the
fourth row (panels D, H,
and L). A and B, HEK cells transiently
overexpressing Cx31.9-eGFP. C and D, HEK cells
transiently overexpressing eGFP. E and F,
Cx31.9-overexpressing HEK cells immunostained with mAb 7G6.
G and H, wt HEK cells immunostained with mAb 7G6.
I and J, Cx31.9-expressing HEK cells
immunostained with mAb 5G11. K and L, wt HEK
cells immunostained with mAb 5G11. The scale bar shown in
panel A is 20 µm, and all panels are
the same magnification. White arrows indicate structures
resembling gap junctions between cells.
|
|
Cx31.9-HEK cells were analyzed by confocal immunofluorescence with mAbs
7G6 and 5G11. mAb 7G6 was found to stain mainly large gap junction-like
structures (Fig. 6E, arrow), and intracellular puncta. No significant staining was seen in wt HEK cells with mAb 7G6
(Fig. 6G), or in either cell line with an irrelevant mouse IgG mAb (data not shown). mAb 5G11 was also found to stain mainly gap
junction-like structures in Cx31.9-HEK cells (Fig. 6I,
arrow), and showed no significant staining in wt HEK cells
(Fig. 6K). The staining of mAb 7G6 was similar whether the
cells had been fixed in 4% paraformaldehyde, methanol, or acetone,
whereas mAb 5G11 only worked well with acetone fixation.
Immunohistology of Human Tissue Sections Using mAbs 7G6 and
5G11--
Frozen human tissue sections of normal testis, brain, and
tonsil were examined with mAb 7G6 and mAb 5G11 by confocal
immunofluorescence. In all three tissues mAb 7G6 reacted with blood
vessels (Fig. 7, A,
D, and G). No staining of these structures was
observed with an irrelevant IgG mAb (data not shown). To identify the
cell types recognized by mAb 7G6, co-labeling was done with mAb 1E12, which reacts with smooth muscle -actinin (35). Extensive
co-localization was observed between mAb 7G6 and 1E12 (Fig. 7,
A, B, D, E, G, and H), indicating that mAb 7G6 reactivity was localized to
smooth muscle cells. Controls included staining with primary antibodies separately, and with secondary antibodies separately (data not shown).
Staining with mAb P2B1, which recognizes the endothelial cell marker
PECAM (36), labeled capillaries or lymphatic vessels, which do not
contain smooth muscle cells (Fig. 7J, arrows), as well as
the endothelial cells in the larger vessels that contained smooth
muscle cells. Staining of adjacent sections with PECAM antibodies (mAb
P2B1) showed more extensive staining than mAb 7G6 (indicated by
arrows in Fig. 7J when compared with Fig.
7G), because of the labeling of capillaries or lymphatic
vessels by mAb P2B1. These results indicated that mAb 7G6 stained
smooth muscle cells, but not endothelial cells.
View larger version (54K):
[in this window]
[in a new window]
|
Fig. 7.
Detection of Cx31.9 in smooth muscle
(SM) by immunoconfocal microscopy of normal human
tissue sections. Three sections shown in panels
A-I were co-labeled with mAb 7G6 (FITC), and mAb 1E12
against smooth muscle (TRITC). Only the FITC channel (A,
D, and G) and the TRITC channel (B,
E, and H) are shown. Extensive co-localization
was observed by merging the two channels (data not shown). Cells were
visualized by a nuclear stain shown in C, F, and
I. A-C, human testis. D-F, human
brain. G and H, human tonsil. J,
section adjacent to sections G-I, stained with a marker for
endothelial cells. Endothelial cells (arrows) were detected
in areas where no staining was observed with mAb 7G6 (G),
suggesting that mAb 7G6 does not stain endothelial cells. L,
human testis stained with mAb 5G11. Staining of smooth muscle was also
observed. Broken lines in panels B and
L indicate the borders of seminiferous tubules
(asterisks). The scale bar in A is 40 µm, and all panels are the same magnification.
|
|
Upon acetone fixation of these sections, mAb 5G11 stained vascular
smooth muscle cells in testis (Fig. 7L) as well as in brain and tonsil (data not shown), in accordance with the staining pattern of
mAb 7G6. This suggests that Cx31.9 is mainly expressed by vascular smooth muscle cells.
Functional Characterization of Cx31.9 Gap Junctions: Dye
Transfer--
Confluent cultures of Cx31.9-HEK cells and wt HEK cells
were microinjected with the fluorescent tracers Lucifer yellow
(Mr 457, net charge 2), carboxyfluorescein
(Mr 376, net charge 1), cascade blue
(Mr 596, net charge 3), and ethidium bromide
(Mr 394, net charge +1), and dye spread was
monitored in a fluorescent microscope. These dyes have previously been
shown to permeate some gap junctions. Fluorochrome-labeled dextran,
which does not permeate gap junctions because of its size, was included
in the injected solution, to identify the injected cell. No dye
transfer beyond the spread of dextran was observed in wt HEK cells, or in 10/10 injections for each dye in Cx31.9-HEK cells (data not shown).
Expression of Cx31.9 in Xenopus Oocytes--
To examine the
functionality of Cx31.9 channels, we expressed Cx31.9 in
Xenopus oocytes, and assayed for the presence and properties
of gap junction-mediated electrical coupling between oocyte pairs. RNA
containing the Cx31.9 coding sequence, flanked by 5'- and
3'-UTR sequences from the X. laevis -globin gene, was
transcribed in vitro, and injected in oocytes together with a Cx38 antisense oligonucleotide, to reduce the coupling effect of
endogenous oocyte connexin. Significant cell-to-cell conductances were
detected in oocyte pairs injected with Cx31.9 cRNA, whereas no
conductances were detected in pairs of uninjected oocytes. Oocyte pairs
injected with 0.05 µg/µl Cx31.9 cRNA showed cell-to-cell conductances ranging between 0.1 and 0.9 µS, whereas oocyte pairs injected with 0.5 µg/µl Cx31.9 cRNA showed conductances between 0.3 and 11.3 µS. The average conductances for oocyte pairs increased with
the concentration of injected Cx31.9 cRNA (Table
I), further indicating that the observed
gap junctional coupling resulted from Cx31.9 and not from any
endogenously expressed component. The voltage-dependent
properties of Cx31.9 channels were then examined and compared with the
properties of previously reported gap junction channels. Interestingly,
Cx31.9 gap junctions showed very little sensitivity to voltage as
demonstrated by the family of junctional currents induced by imposing a
series of transjunctional potentials in oocyte pairs ranging from 180
to +120 mV (Fig. 8A). Note
that any reduction in junctional current was only seen at very large
transjunctional potentials (greater than ±120 mV). The absence of
strong voltage-dependent gating was confirmed by a scatter
plot (Fig. 8B) of the normalized initial
(squares) and near steady-state (circles)
junctional conductances from six different Cx31.9-injected oocyte
pairs. Because the voltage dependence was weak and never resulted in
reduction of the conductance to more than about 80% of its initial
value, we did not attempt the traditional Boltzmann fit of the data.
These results were similar for both concentrations of Cx31.9 cRNA
injected, and for all levels of macroscopic conductance, suggesting
that the lack of voltage dependence is not an artifact arising from
increased access resistance that may be associated with larger gap
junction plaques and larger total whole cell junctional
conductances.
View this table:
[in this window]
[in a new window]
|
Table I
Junctional conductances of paired Xenopus oocytes expressing Cx31.9
Two different concentrations of Cx31.9 cRNA (0.05 and 0.5 µg/µl)
were injected in oocytes together with a Cx38 antisense
oligonucleotide. Increased junctional conductances were observed with
increased amounts of Cx31.9 cRNA injected.
|
|
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 8.
Electrophysiological analyses of
Xenopus oocytes expressing Cx31.9.
Voltage-dependent properties of Cx31.9 channels formed
between paired X. laevis oocytes. A, an example
of junctional currents elicited in the non-driven oocyte by voltage
pulses applied to the driven oocyte ranging from 180 mV to +120 mV in
30-mV increments from a 40 mV holding potential. The transjunctional
potential across gap junction channels ranged from +140 mV to 160 mV
relative to the cytoplasmic side of the non-driven oocyte.
B, the Gj-Vj
relationship for Cx31.9 gap junction channels. Initial and approximate
steady-state conductances were obtained by dividing the junctional
current at 30 ms and 5 s, respectively, by the transjunctional
potential (Vj) as determined by the difference
between the measured voltages of the two oocytes. Initial and
steady-state conductances were then normalized to their respective
values at ±10 mV and plotted against Vj.
Voltage dependence with slow kinetics is evident only at large,
non-physiological voltages.
|
|
It should be noted that we consistently observed asymmetry in the
voltage dependence of the instantaneous conductance, but it is
difficult to ascertain exactly what this asymmetry represents. One
possibility is that Cx31.9 gap junction channels are sensitive to the
membrane potential in addition to the transjunctional potential and
undergo Vm gating. However, the initial
conductance is not monotonically dependent on Vm
as is the case for other connexins that show Vm
sensitivity. Rather, we found that the normalized initial conductance
increased only for negative transjunctional potentials and was most
prominent only for the largest voltages. Under these specific
conditions, which arise from depolarizing the driven oocyte to voltages
greater than +60 mV, we also observed the activation of large outward
currents in the plasma membrane of the driven oocyte originating from
endogenous channels (data not shown). It is therefore possible that the
asymmetry we observed in the instantaneous conductance is an artifact
introduced by the activation of large endogenous currents.
Double Whole Cell Patch Clamping--
To verify that Cx31.9 also
formed gap junction channels in mammalian cells, and to confirm the
lack of voltage dependence of Cx31.9 channels observed in the oocyte
system, we examined the electrophysiological properties of Cx31.9-HEK
versus wt HEK cells, using double whole cell patch clamping.
We used HEK293 cells for this purpose, because these cells have
previously been reported to express very low levels of Cx43, and to
exhibit minimal coupling (13, 37). We recorded data on two types of
cell pairs, either isolated pairs, or pairs of cells located in a
string of cells, such that neither of the two cells contacted any cell
that contacted both members of the pair. Using these pairs allowed a
reasonable accurate estimate of junctional conductance, but could give
rise to small series resistance errors as a result of lowered effective
input impedance of the driven cell. Such errors would result in
underestimates of the junctional conductance.
The voltage in both cells was initially clamped to the same
holding potential (in the range from 0 to 60 mV, but usually 40
mV). Junctional current, Ij, was measured during
periodic positive and/or negative pulses of transjunctional voltage,
Vj. The junctional conductances,
Gj, of Cx31.9-HEK cell pairs varied between 5 and 125 nS, with a mean value of 29.6 nS (n = 15) and a
standard deviation of 30 nS. The Gj of wt HEK
cells varied from 0 nS to 50 nS, with a mean value of 13.2 nS
(n = 7) and a standard deviation of 18 nS (Fig.
9A). Thus, the highest
junctional conductances and a higher average conductance were seen in
Cx31.9-transfected cells, indicating that Cx31.9 channel expression was
associated with increased junctional conductance. The standard
deviation of the junctional conductances of transfected cell pairs was
also higher than of wt cells, suggesting that transfection resulted in
variable, significantly higher levels of junctional conductance than
those seen in wt cells. However, we cannot exclude that the highest
junctional conductances from both Cx31.9-expressing and wt cells were
caused by cytoplasmic bridges. The residual background junctional
conductance of wt HEK cells was probably contributed by endogenous
Cx43.
View larger version (12K):
[in this window]
[in a new window]
|
Fig. 9.
Double cell voltage clamp of
Cx31.9-overexpressing HEK cells and wt HEK cells. A,
junctional conductances between HEK cells stably overexpressing Cx31.9
(n = 15), and wt HEK cells (n = 7).
Each triangle represents one measurement.
Horizontal lines indicate mean conductances.
B, junctional current of a pair of Cx31.9-overexpressing HEK
cells. The non-driven cell was held at 0 mV and the voltage of the
driven cell was pulsed to +160 to 120 mV in 40-mV steps. Note that
very little voltage dependence is evident. These results are similar to
those obtained from the paired oocyte expression system (Fig. 8).
|
|
Using Cx31.9-HEK cell pairs, we confirmed the lack of voltage
dependence of the current seen in the paired oocyte experiments. Fig.
9B shows the current response of the non-driven cell, held at 0 mV, whereas the driven cell was subjected to a series of voltage
pulses from 160 mV to 120 mV in 40-mV steps. None of the IV curves,
generated by similar, but not identical protocols, from nine different
Cx31.9-transfected HEK cell pairs showed significant voltage
dependence. In some cases we were not able to measure complete current
voltage curves, but were able to estimate junctional conductance from a
voltage protocol in which each cell was held at 40 mV while the other
was pulsed to 20 mV. Thus, the electrophysiological results from two
different expression systems confirm that Cx31.9 induces formation of
voltage-independent cell-to-cell coupling within a physiologically
relevant range.
Cx31.9 Binding to ZO-1--
The most COOH-terminal sequence of
Cx31.9 contained a motif (RDLAI) that resembled the COOH-terminal
sequence of Cx43 (DDLEI). The most COOH-terminal isoleucine residue of
Cx43 has been shown to be critical for binding to the second PDZ domain
of ZO-1 (14). Based on this, we addressed whether Cx31.9 also
interacted with ZO-1. Double immunoconfocal staining for ZO-1 and
Cx31.9 in Cx31.9-HEK cells showed that most ZO-1 staining co-localized
to membrane domains with cell-cell contact and Cx31.9 gap junctions
(Fig. 10A). However, cell
membrane staining of ZO-1 was also occasionally found where no Cx31.9
gap junctions were detected (data not shown), and no ZO-1
co-localization was observed with Cx31.9 in intercellular compartments
(Fig. 10A). This showed that Cx31.9 was partially co-localized with ZO-1. We then examined whether ZO-1
co-immunoprecipitated with Cx31.9. Immunoblot analysis showed that ZO-1
was present in equal amounts in lysates from both Cx31.9-HEK and wt HEK
cells (Fig. 10B, upper panel, lanes 1 and 5). ZO-1 was found to co-precipitate with Cx31.9 which
was pulled down using both mAb 7G6 and 5G11 (Fig. 10B,
upper panel, lanes 3 and 4), whereas
ZO-1 was not immunoprecipitated from wt HEK cells (Fig. 10B,
upper panel, lanes 6 and 7). An
irrelevant antibody did not immunoprecipitate ZO-1 (Fig.
10B, upper panel, lane 2). Ponceau
staining of the blot showed that similar amounts of antibodies had been
used (Fig. 10B, middle panel). Cx31.9 was detected in the immunoprecipitations of mAbs 7G6 and 5G11 from Cx31.9-HEK cells with biotinylated mAb 7G6, as expected (Fig. 10B, lower panel). Interestingly, although
severalfold more Cx31.9 was precipitated with mAb 7G6 than with mAb
5G11, the amount of co-precipitated ZO-1 was nearly identical in the
two samples. This could be the result of a partial steric hindrance of
mAb 7G6 binding to Cx31.9 when complexed with ZO-1.
View larger version (39K):
[in this window]
[in a new window]
|
Fig. 10.
Cx31.9 binds to the second PDZ domain of
ZO-1. A, immunoconfocal images of Cx31.9-overexpressing HEK
cells double-stained for ZO-1 (left panel) and Cx31.9 with
mAb 7G6 (right panel). Nuclei are shown in each image. When
the two stains were superimposed (data not shown), extensive
co-localization of ZO-1 and Cx31.9 was observed at gap junctions
between cells (arrows). B, co-immunoprecipitation
of ZO-1 with Cx31.9. Cx31.9-HEK and wt HEK cell lysates were used in
immunoprecipitation analyses with mAbs 7G6, 5G11, and an irrelevant
mAb. ZO-1 was immunoprecipitated with mAb 7G6 and mAb 5G11 (lanes
3 and 4), but not with an irrelevant mAb (lane
2, upper panel). Ponceau staining of the blot showed
similar amounts of antibodies in all samples (middle panel).
The membrane was reprobed with biotinylated mAb 7G6 to monitor the
amount of Cx31.9 in the immunoprecipitates (lower panel).
Cx31.9 was detected in mAb 7G6 and mAb 5G11 immunoprecipitates
(lanes 3 and 4). C, pull-down
experiment with Cx31.9 and PDZ domains from ZO-1. PDZ domains were
expressed in bacteria as GST fusion proteins, and incubated with
Cx31.9-HEK cell lysates. Proteins bound to fusion proteins were
separated by SDS-PAGE, and Cx31.9 was detected with mAb 7G6 in
immunoblot analysis. The second PDZ domain bound to Cx31.9 (lane
3, upper panel). Fusion proteins were visualized by
Ponceau staining of the blot (lower panel).
|
|
These experiments suggested that ZO-1 interacted directly or indirectly
with Cx31.9. Reprobing of these blots with an antibody against Cx43
revealed small amounts of Cx43 in the lysates, but this was not
co-immunoprecipitated with Cx31.9 (results not shown), excluding that
ZO-1 was co-immunoprecipitated because of the presence of heteromeric
Cx31.9/Cx43 connexons. To determine which domains of ZO-1 interact with
Cx31.9, GST fusion protein pull-down assays were performed with the
first, second, and third PDZ domains of ZO-1. These experiments
demonstrated that the second PDZ domain interacts with Cx31.9 (Fig.
10C, upper panel, lane 3), whereas the
first and third PDZ-domain did not (Fig. 10C, upper
panel, lanes 2 and 4). Ponceau staining of
the blot showed that similar amounts of fusion proteins had been used
(Fig. 10C, lower panel). Interestingly, in
similar experiments using lysates of HEK cells overexpressing
Cx31.9-eGFP, Cx31.9-eGFP could be detected in the lysate as shown
earlier (Fig. 5A), but was not pulled down by any of the PDZ
domains from ZO-1 (data not shown). These experiments suggested that
the most COOH-terminal residues of Cx31.9 interacted with the second
PDZ domain of ZO-1, and that these residues were made unavailable for
binding by the eGFP tag. Similarly, addition of a COOH-terminal Myc tag
and green fluorescent protein tag to Cx43 has been reported to disrupt
binding to ZO-1 (14).
| |
DISCUSSION |
In the present study a novel human connexin gene, designated
Cx31.9, was cloned and characterized. A mAb against Cx31.9
detected a protein with the expected SDS-PAGE mobility from human
tissue lysates, suggesting that Cx31.9 is a bona
fide gene. We furthermore showed that Cx31.9 forms functional gap
junctions and is expressed in vascular smooth muscle cells. One unusual
feature of the Cx31.9 gene is the probable lack introns in
the 5'-UTR. We cannot exclude the possibility that the 5'-UTR could be
spliced in some instances, because potential splice acceptor sites were
present close to the initiating ATG codon. However, a spliced form is
unlikely to represent a large proportion of Cx31.9 transcripts, because only one major transcript size was detected by Northern analysis (Fig.
3).
The existence of a separate transcript, originating within the 5'-UTR,
and being transcribed in the opposite direction of Cx31.9,
was also unusual, and has to our knowledge not been described for other
connexin genes. Transcription of this gene would be expected to be
incompatible with simultaneous transcription of Cx31.9.
Given the abundance of the Cx31.9 transcript by Northern
blot analysis, it was surprising that we only identified three
Cx31.9 EST clones in GenBankTM. This may be a
result of the following. 1) The transcript was larger than most
published connexins, which reduces the number of (partially sequenced)
ESTs containing the coding sequence. 2) The GC content of the coding
region of Cx31.9 was 76%, which is higher than any other
connexin gene. The high GC content could reduce the efficiency of
cDNA synthesis, thereby under representing the sequence in cDNA
libraries. Furthermore, the high GC content could make sequencing
difficult, which may have contributed to sequences from
Cx31.9 EST clones being discarded because of poor sequence
quality. 3) It is possible that more EST clones containing the most 3'
end of the Cx31.9 transcript exist in
GenBankTM.
Analysis of the Cx31.9 protein sequence showed a number of similarities
to other connexins. The presence of four transmembrane domains,
conserved spacing of cysteine residues in the extracellular domains,
and a number of potential phosphorylation sites in the COOH-terminal
domain. Cx31.9 also contained a stretch of 10 proline residues in the
COOH-terminal domain (residues 238-248). This domain is possibly
involved in protein-protein interactions, because proline-rich
sequences are known to bind to a number of specific protein interaction
modules, such as Src homology 3 domains (38).
Analysis of amino acid similarities in domains conserved between
connexins placed Cx31.9 in the -group of connexins (Fig. 2).
However, when the full sequences were compared, Cx31.9 had a shorter
cytoplasmic loop and COOH-terminal domain than the other members of the
-group. The cytoplasmic loop of Cx31.9 consisted of 37 amino acid
residues, but contained 50-58 amino acid residues for the other
members of the -group. The members of the -group of connexins had
cytoplasmic loops containing 30-38 residues, and were thus more
similar to Cx31.9 in this respect. The significance of these
differences is not clear.
Immunostaining of human tissue sections with mAb 7G6 indicated that
Cx31.9 was mainly expressed in vascular smooth muscle, and this was
further corroborated by immunostaining with mAb 5G11. The ubiquitous
expression pattern of the Cx31.9 transcript was consistent
with expression in vascular smooth muscle (and cardiac myocytes).
Furthermore, there was a correlation of Cx31.9 expression levels, as
detected by Northern and immunoblot analysis of human tissues. mAb 7G6
detected Cx31.9 in heart lysates, but not in lysates from brain and
skeletal muscle, consistent with the expression levels of
Cx31.9 transcripts observed by Northern analysis (Fig. 3).
Notably, the organ expression pattern of Cx31.9 as determined by
Northern analysis differed from the organ expression patterns of other
connexins also found in smooth muscle cells, such as Cx43, Cx40, and
Cx45 (39, 40). For example, although the expression of Cx31.9 is low in
lung (Fig. 3), Cx40 is highly expressed in this organ by Northern
analysis (33, 41). Because several different cell types are present in
the lung, one possible explanation for this difference is that
endothelial cells are the main cells expressing Cx40 in lung, whereas
these cells do not express Cx31.9. To further investigate whether
Cx31.9 was expressed by endothelial cells, adjacent human tissue
sections were stained for an endothelial cell marker (PECAM), and with
mAb 7G6. No co-localization of PECAM with mAb 7G6 was observed in
capillaries or lymph endothelial cells (Fig. 7, G and
J). Furthermore, we could not detect Cx31.9 expression in
second passage human umbilical vein endothelial cells with mAb
7G6. These endothelial cells have been reported to express Cx37 and
Cx40 (42). Taken together, this suggested that Cx31.9 is not highly
expressed, if at all, by endothelial cells. Interestingly, the
expression levels of mouse Cx45 in lung is relatively high by Northern
analysis (33), and is confined to tracheo-bronchial smooth muscle (40).
Whether Cx31.9 is expressed in tracheo-bronchial smooth muscle remains
to be determined, but a comparison of the Cx31.9 and Cx45 expression
levels in lung by Northern analyses suggests that Cx31.9 may not be
expressed by this type of smooth muscle, barring possible species and
regional differences of smooth muscle abundance in lung.
Cx31.9 forms functional gap junction channels, as shown by
electrophysiological analyses in paired oocytes expressing Cx31.9, and
in HEK cell pairs overexpressing Cx31.9. We found the average junctional conductance of Cx31.9-HEK cell pairs to be a slightly more
than twice that of wt HEK cell pairs (29.6 versus 13.2 nS). The large variance in junctional conductance of Cx31.9-HEK pairs suggests that the efficiency of junctional coupling induced by transfection is quite variable. It is possible that some of the conductance observed between transfected cell pairs is not caused by
Cx31.9, but to cytoplasmic bridges or Cx43. However, this cannot be the
case for all of the pairs, because of the close correspondence in the
voltage dependence observed for Cx31.9-HEK cell pairs and oocyte pairs
expressing Cx31.9, suggesting that a majority of the currents seen
between Cx31.9-HEK cell pairs are most likely caused by Cx31.9.
In contrast to other connexin channels, Cx31.9 channels showed no
gating dependence of transjunctional voltage within a physiological relevant range, but only at extreme transjunctional voltages (>125 mV)
(Figs. 8 (A and B) and 9B). The
voltage independence of Cx31.9 channels was observed by expressing two
different c
§,
,
TOP
ABSTRACT
INTRODUCTION
