Content uploaded by Eric C Beyer
Author content
All content in this area was uploaded by Eric C Beyer
Content may be subject to copyright.
Available via license: CC BY-NC-SA 4.0
Content may be subject to copyright.
Available via license: CC BY-NC-SA 4.0
Content may be subject to copyright.
Antisera Directed against Connexin43 Peptides React with
a 43-kD Protein Localized to
Gap Junctions in Myocardium and Other Tissues
Eric C. Beyer,*¢§ Joerg Kistler, II David L. Paul,* and Daniel A. Goodenough*
* The Department of Anatomy and Cellular Biology and ¢the Department of Pediatrics, Harvard Medical School;
and §the Division of Pediatric Hematology/Oncology, the Children's Hospital and the Dana-Farber Cancer Institute,
Boston, Massachusetts 02115; and II Department of Cell Biology, University of Auckland, Auckland, New Zealand
Abstract. Rat heart and other organs contain mRNA
coding for connexin43, a polypeptide homologous to a
gap junction protein from liver (connexin32). To pro-
vide direct evidence that connexin43 is a cardiac gap
junction protein, we raised rabbit antisera directed
against synthetic oligopeptides corresponding to two
unique regions of its sequence, amino acids 119-142
and 252-271. Both antisera stained the intercalated
disc in myocardium by immunofluorescence but did
not react with frozen sections of liver. Immunocyto-
chemistry showed anti-connexin43 staining of the cyto-
plasmic surface of gap junctions in isolated rat heart
membranes but no reactivity with isolated liver gap
junctions. Both antisera reacted with a 43-kD polypep-
tide in isolated rat heart membranes but did not react
with rat liver gap junctions by Western blot analysis.
In contrast, an antiserum to the conserved, possibly
extracellular, sequence of amino acids 164-189 in
connexin32 reacted with both liver and heart gap junc-
tion proteins on Western blots. These findings support
a topological model of connexins with unique cyto-
plasmic domains but conserved transmembrane and
extracellular regions. The connexin43-specific antisera
were used by Western blots and immunofluorescence
to examine the distribution of connexin43. They dem-
onstrated reactivity consistent with gap junctions be-
tween ovarian granulosa cells, smooth muscle cells in
uterus and other tissues, fibroblasts in cornea and
other tissues, lens and corneal epithelial cells, and re-
nal tubular epithelial cells. Staining with the anti-
connexin43 antisera was never observed to colocalize
with antibodies to other gap junction proteins (con-
nexin32 or MP70) in the same junctional plaques. Be-
cause of limitations in the resolution of the immuno-
fluorescence, however, we were not able to determine
whether individual cells ever simultaneously express
more than one connexin type.
AP junctions are membrane specializations that permit
the exchange of small metabolites and ions between
neighboring cells. While gap junctions and intercel-
lular communication have been studied in a number of cell
and tissue types, the structure of gap junctions has been best
studied in rat liver where the interhepatocyte gap junctions
have been shown to be composed of collections of membrane
channels, called connexons, joined in mirror symmetry with
connexons in the membrane of the adjacent cell (5, 24). The
complete cDNA corresponding to a structural protein of the
connexon has been cloned from rat and human liver (22, 32).
We have suggested the name connexin32 for this protein,
based on its predicted molecular mass of 32 kD (2). Con-
nexin32 has been confirmed as a gap junction protein by
structural and functional criteria (6, 32).
Data from cDNA and protein sequencing suggest that
there is a family of related connexin proteins (2,-3, 21, 30).
We have used low-stringency hybridization to clone an ho-
mologous cDNA from a rat heart library, which predicts a
protein of 43 kD called connexin43 (2). The size of con-
nexin43 and its amino-terminal sequence closely match those
of the major protein in isolated preparations of rat heart gap
junctions (23, 25, 26). Comparison of the predicted con-
nexin43 polypeptide with connexin32 shows both conserved
and divergent amino acids and suggests that (a) conserved
amino acids lie in transmembrane and extracellular regions
and (b) nonconserved amino acids face the cytoplasm.
To provide a direct evidence that the protein corresponding
to connexin43 cDNA is a cardiac gap junction protein and
to test the topological model, we have raised antisera directed
against two synthetic oligopeptides from unique regions of
the predicted cormexin43 sequence, and we have demon-
strated immunoreactivity with a 43-kD protein structurally
localized to cardiac gap junctions.
Previous studies have examined the distribution of connex-
in32 expression by immunohistochemistry and have shown
its presence in a number of different tissues but not in all lo-
cations where gap junctions are known to occur (8, 19, 31).
© The Rockefeller University Press, 00"21-9525/89/02/595/11 $2.00
The Journal of Cell Biology, Volume 108, February 1989 595-605 595
Northern blot experiments have shown that the mRNAs for
connexin32 and connexin43 have different, but overlapping,
distributions in multiple organs (2, 32). In this paper, we
have used the specific antipeptide antisera to examine con-
nexin43 distribution in many locations where its mRNA has
been detected or where gap junction biology has been studied.
The availability of antisera specific for different connexin
molecules permits study of the codistribution of connexin
molecules within the same organ, tissue, and cell. Previous
studies have shown the intermixing of two different connexin
molecules in the same junctional plaques joining hepatocytes
by immunofluorescence (30). Miller and Goodenough (28)
showed that lens epithelial cells simultaneously express struc-
turally and functionally unique gap junctions with their ho-
mologous and heterologous neighbors. In studies reported
here, we take advantage of other connexin-specific antibody
reagents which have been developed. Polyclonal antipeptide
and monoclonal antibodies which specifically recognize con-
nexin32 have been raised (14). A monoclonal antibody has
been produced which specifically recognizes MP70, a pro-
tein that has been localized at lens fiber junctions and has a
related NH2-terminal sequence, suggesting that it is a mem-
ber of the connexin family (16, 20, 21). We have used these
reagents to determine the distributions of these gap junction
proteins.
Materials and Methods
Reagents
All chemicals were obtained from Sigma Chemical Co. (St. Louis, MO)
unless otherwise specified.
Keyhole limpet hemocyanin (KLH) l was purchased from Pacific Bio-
marine Laboratories Inc. (Venice, CA). A peptide representing amino acids
119-142 of connexin43 (DGVNVEMHLKQIEIKKFKYGIEEH) was syn-
thesized by Bachem, Inc. (Torrance, CA). This peptide was demonstrated
to be >97% pure by HPLC and amino acid analysis. A peptide representing
amino acids 252-271 of connexin43 (GPLSPSKDCGSPKYAYFNGC) was
synthesized by Bioseareh, Inc. (San Rafael, CA). HPLC analysis of this
peptide demonstrated a single major peptide component but parity was not
quantitated. HPLC was conducted on a Vydac C18 column after reduction
with DTT; elution was with a gradient of 5-100% acetonitrile/0.1% trifluoro-
acetic acid over a 20-min period.
Preparation of Antisera
The peptides were coupled to KLH using the following procedure in an at-
tempt to achieve a 30:1 molar ratio of peptide to KLH. 780 nmol of peptide
were reconstituted in 400 #1 PBS. 26 amol of KLH in 250 #1 of PBS were
added. 5 #l of 25% glutaraldehyde (Sigma Chemical Co.) were added, and
the resultant solution was incubated overnight at room temperature. Some
turbidity developed. No efforts were made to further purify the conjugates.
For primary immunizations, peptide-KLH conjugate containing 200 #g
peptide was emulsified in Freund's complete adjuvant (Gibco Laboratories,
Grand Island, NY) and was injected in duplicate into preimmune bled, fe-
male New Zealand white rabbits. All injections were made in multiple
paravertebral intradermal sites. After 30 d, the rabbits were boosted with
KLH-peptide in incomplete Freund's and were bled 7-10 d later. Rabbits
were boosted thereafter at monthly intervals.
Antisera were screened for reactivity with the paptides by a modified
ELISA. Unconjugated peptide was adsorbed to small squares of polyvinyl-
idene difluoride membrane (Immobilon; Millipore Continental Water Sys-
tems, Bedford, MA). These squares were then blocked with BLOTTO (10%
nonfat dry milk in PBS) and reacted with dilutions of test antisera, peroxi-
dase-conjugated protein A (Boehringer Mannheim Biochemicals, Indianap-
1. Abbreviation used in this paper: KLH, keyhole limpet hemocyanin.
olis, IN) diluted 1:2,000 in PBS, and 0.05% 3,Y-diaminobenzidine, 0.01%
H202 in PBS.
For most experiments, crude antiserum was used. For some experiments,
afffinity-purified anti-connexin43 252-271 was prepared using peptide at-
tached to activated CH-Sepharose 4B (Pharmacia Fine Chemicals, Uppsala,
Sweden) according to manufacturer's directions.
Preparation and characterization of mouse monoclonal antibodies to con-
nexin32 (M12.13) and to the lens protein MP70 (6-4-B2-C6) have been de-
scribed previously (14, 21).
Preparation and characterization of rabbit antisera to pepfides representing
amino acids 98-124 (QQHIEKKMLRLEGHGDPLHLEEVKRHK) and
164-189 (RLVKCEAFPCPNTVDCFVSRPTEKTV) of connexin32 have
been described elsewhere (14). The antiserum to paptide 98-124 specifi-
cally reacts with the cytoplasmic face of isolal~l rat liver gap junctions. It
will be referred to as anti-connexin32 antiserum throughout this paper. The
antiserum to peptide 164-189 does not react with native rat liver gap junc-
tions unless they have been split apart using either a urea/alkali or a hyper-
tonic sucrose procedure, suggesting a possible extracellular orientation of
this epitope in the junction protein. The peptide used as immunogen for this
antiserum contains 59% identical residues to the corresponding portion of
the connexin43 sequence (residues 182-208, SAYYTKCRDPCPHQKDC-
FLSRPTEKTI) (3), suggesting that it might react with both proteins.
Western Blot Analysis
Rat liver gap junctions were isolated as described by Fallon and Good-
enough (9) with the modification of the addition of 0.5 mM diisopropyl-
fluorophosphate (Sigma Chemical Co.) to the initial homogenization buffer
and the sareosine solutions. Preparations enriched for intercalated discs
were isolated from rat hearts as described by Green and Severs (15), except
for the addition of 0.5 mM diisopropylfluorophosphate and 1 mM PMSF
(Sigma Chemical Co.) to all solutions. Urea-washed calf lens membranes
were isolated according to Paul and Goodenongh (33). For analysis of other
tissues, freshly dissected rat tissues were homogenized in 50 vol or more
of 1 mM NaHCO3/0.5 mM diisopropylfiuorophosphate/1 mM PMSF and
centrifuged at I0,000 g for 10 rain; the resulting pellet was resuspended in
boiling SDS sample buffer; insoluble material was removed by centrifuga-
tion; and the superuatant was applied to the SDS-polyacrylamide gels.
SDS-PAGE was performed using 12 % minigels. Immunoblots were pre-
pared essentially as described by Paul and Goodenough (33) except that
transfer buffer did not contain SDS, and transfer to nitrocellulose mem-
branes was conducted at 75 V for 1 h. Antibody binding was performed
using primary rabbit antisera diluted 1:1,000 and secondary peroxidase-con-
jugated protein A as described for the ELISA dot assay above. For some
experiments, ~251-protein A (Amersham Corp., Arlington Heights, IL) was
used, as described by Goodenough et al. (14). In some experiments, to dem-
onstrate the specificity of immunoblot reactivity, diluted antibodies were
preincubated with pure peptide at a concentration of 100 pg/ml.
Immunohistochemistry
Immunofluorescence microscopy was performed on rat tissues that were ei-
ther directly frozen unfixed or were fixed in 1% paraformaldehyde in PBS
for 1 h followed by a 1-h incubation in 0.5 M sucrose in PBS for cryoprotec-
tion before freezing. Most tissues were obtained from adult rats of either
sex. Ovarian tissue was obtained from 21-d-old female rats primed for two
successive days with intraperitoneal injections of 10 IU of pregnant mare's
serum gonadotropin (Sigma Chemical Co.) in PBS. Uteri were obtained ei-
ther from pregnant rats in labor or from 21-d-old females who had received
five consecutive daily subcutaneous injections of 500/~g of estradiol (Sigma
Chemical Co.) in sesame oil. Previous studies have shown a dramatic mor-
phological increase in gap junction structures after these hormonal treat-
ments (1, 13, 23). Small pieces of tissue were rapidly frozen in liquid freon
cooled with liquid nitrogen. 5-10-#m cryostat sections were cut and placed
on gelatin-coated slides. Sections were reacted with rabbit antipeptide an-
tisera diluted 1:200 or 1:500 in PBS or with straight culture supernatants
for the mouse monoclonal antibodies followed by rhodamine- or fluores-
cein-conjugated goat anti-rabbit or -mouse antisera (Boehringer Mannheim
Biochemicals) as appropriate. In some experiments, to demonstrate the
specificity of immunolabeling, diluted primary antisera were preincubated
with pure peptide at a concentration of 20 #g/ml.
Both anti-connexin43 antisera were used for staining of heart, liver,
ovary, and uterus and gave identical results. Anti-connexin43 252-271 was
used for all other tissues. Anti-connexin32 98-124 was used for all tissues
except the double-labeling experiments in which the anti-connexin32
mono-
The Journal of Cell Biology, Volume 108, 1989 596
Figure 1.
Immunoblot analysis of isolated liver gap junction and enriched cardiac intercalated disc preparations using anti-connexin peptide
antisera or monoclonal antibodies. (a) Coomassie Blue-stained SDS-PAGE of (lane S) molecular mass standard proteins (phosphorylase
B, 97 kD; BSA, 66 kD; ovalbumin, 43 kD; carbonic anhydrase, 31 kD; soybean trypsin inhibitor, 21.5 kD; lysozyme, 14 kD). (Lanes
L and H) Isolated rat liver gap junctions and enriched heart intercalated disc preparations. (b) Immunoblot of a gel identical to that in
a, but without standards, reacted with anti-eonnexin43 119-142. This antiserum reacts strongly with a 43-kD band in the heart preparations,
and weakly with other bands. It does not react with the liver junctions. (c) Identical immunoblot reacted with anti-connexin43 252-271.
This antiserum also reacts strongly with a 43-kD band in the heart preparations; it also recognizes a band of ,~60 kD that is not seen
by the other antiserum. This antiserum does not react with the liver junctions. (d) Immunoblot reacted with monoclonal anti-connexin32
M12.13. 5% as much liver gap junction protein was loaded as in a. This antibody specifically reacts with several forms of the rat liver
gap junction protein, including monomer, dimer, and some proteolysis fragments. It does not react with any proteins in the heart preparation.
(e) Immunoblot reacted with anti-connexin32 98-124. Lane L contained 0.5 % as much liver gap junction protein as in a. This antiserum
also reacts specifically with all forms of the liver gap junction protein, but not with any heart proteins. (f) Immunoblot reacted with anti-
connexin32 164-189. Lane L contained 5% as much liver gap junction protein as in a. This antiserum reacts with the liver gap junction
polypeptides and with a 43-kD polypeptide from the heart, suggesting common structure. (g) Immunoblot prepared as in a reacted with
preimmune serum from a rabbit which was subsequently immunized with connexin43 peptide 252-271. This serum shows no reactivity
with any polypeptides from liver or heart.
clonal M12.13 was used to permit the use of different species-specific sec-
ondary antisera.
EM immunocytochemistry was performed on a crude preparation of rat
heart membranes obtained according to the Green and Severs (15) protocol
by homogenization in bicarbonate buffer and low-speed centrifugation, but
without KCI extraction. The methods of Paul and Goodenough (33) were
followed. Gold-labeled secondary antisera were obtained from Janssen Life
Sciences Products (Piscataway, NJ) and used at full strength after exhaustive
absorption with the isolated rat heart membranes.
Results
Western Blot Analysis
Immunoblot analysis was performed to determine the spec-
ificity of the different antisera (Fig. 1). Neither of the antisera
to peptides from connexin43 reacted with the liver gap junc-
tions; both reacted specifically with a polypeptide migrating
at 43 kD in the intercalated disc preparations, exactly match-
ing the mass predicted from the cDNA for the cardiac gap
junction protein. Each of these crude antisera also reacted
less intensely with various other polypeptides in the heart
preparations, but no other common bands were seen. Anti-
connexin43 252-271 reacted with a polypeptide of ~60 kD
as well as the 43-kD band in many heart preparations (Figs.
1 and 4); affinity purification of this antiserum did not abol-
ish this reaction (data not shown), but it was substantially re-
duced by absorption of the serum with a crude rat liver ho-
mogenate (Fig. 4). Absorption of anti-connexin43 252-271
with liver was determined empirically and was not attempted
on anti-connexin43 119-142, because no bands other than
that at 43 kD appeared so prominent. Preimmune sera or
sera absorbed with the corresponding peptides showed no re-
activity. The antiserum to connexin32 peptide 98-124 and
the monoclonal antibody M12.13 both reacted with the 28-
kD principal polypeptide in the rat liver gap junctions as well
as its aggregate and proteolysis fragments, but did not react
with any proteins in the heart lanes, confirming their spec-
ificity for connexin32. The antiserum to connexin32 peptide
164-189 reacted with all forms of the rat liver gap junction
protein and reacted weakly with a polypeptide of 43 kD in
the heart samples that co-migrated with the bands reactive
with the anti-connexin43 antisera, suggesting that both con-
nexin43 and connexin32 shared common structure in a po-
tentially extracellular epitope. Monoclonal antibody 6-4-B2-
C6, which reacts with a 70-kD protein from lens fibers, did
not react with either the heart or liver samples (data not
shown).
Antibody Localization in Heart
Antisera were used for immunohistochemistry on frozen
sections of rat heart (Fig. 2). Both anti-connexin43 sera gave
Beyer et al.
Connexin43 lmmunolocalization
597
Figure 2.
Immunohistochemical localization of connexin43 in frozen section of rat heart. Anti-connexin43 252-271 specifically stains the
intercalated discs, which contain the cardiac gap junctions. Staining appears punetate. Bar, 10/~m.
identical staining consisting of bright staining of the interca-
lated disc structures, which was composed of discrete spots.
This staining was consistent with the known distribution of
gap junctions in the heart (10, 11, 27). The staining was com-
pletely abolished by preincubation of the antisera with 20
ttg/ml of the corresponding peptide before staining (not
shown). Reaction with a nonhomologous peptide had no
effect on staining. Preimmune sera showed no reactivity.
Neither of the connexin32 antisera nor the monoclonal anti-
bodies showed any reactivity with the heart.
EM localization of antibody binding was used to demon-
strate that gap junctions were responsible for the staining pat-
tern observed by light microscopy. The EM immunohisto-
chemistry study was performed on crude membranes isolated
from rat heart. Anti-connexin43 antibody binding was de-
tected using colloidal gold coupled to goat anti-rabbit IgG
(Fig. 3). Gold particles were observed only on the cytoplas-
mic aspects of the heart gap junctions; no specific staining
was observed on any surface of the fascia adherens or of any
nonjunctional membranes. Preimmune antisera and anti-
connexin32 showed only a low, nonspecific background
binding of gold.
Although the anti-connexin32 98-124 antiserum and the
monoclonal antibody M12.13 both brightly stained punctate
areas between hepatocytes, as published in studies reviewed
in the introduction, the anti-connexin43 antisera showed no
reaction with liver by immunofluorescence. The anti. con-
nexin43 antisera showed no immunogold staining of isolated
rat liver gap junctions (data not shown). The antiserum to the
conserved region connexin32 164-189 did not react with ei-
ther liver or heart by immunofluorescence and was not used
for any further studies.
Connexin43 in Other Tissues
Other rat organs were examined by Western blot analysis for
immunoreactive polypeptides with the connexin43 antisera
(Fig. 4). Cardiac intercalated discs, urea-washed lens mem-
branes, and whole tissue homogenates from kidney, uterus,
ovary, and cornea all contained polypeptides of'~43 kD that
reacted with both anti-connexin43 antisera. There were
some variations in the intensity of reaction of the two sera.
For instance, the calf lens membranes reacted more strongly
with anti-connexin43 119-142 than with anti-connexin43
252-271.
These rat organs were also examined by immunofluores-
cence with the anti-connexin43 sera for staining consistent
with gap junctions. Northern blot studies had shown that
pregnant mare's serum gonadotropin-primed ovary contains
large amounts of connexin43 mRNA (2). Immunofluorescent
localization with anti-connexin43 produced large, bright
spots of fluorescence along the borders of adjacent granulosa
The Journal of Cell Biology, Volume 108, 1989 598
Figure 3. Immunocytochemical localization of connexin43 in rat heart membranes. Thin section electron micrograph of crude rat heart
membranes incubated with anti-connexin43 252-271 antiserum and colloidal gold-conjugated goat anti-rabbit antibodies before fixation
and embedding. Gold particles decorate the cytoplasmic surfaces of the gap junction. No significant labeling is detectable on any other
structures, including other junctional or nonjunctional membranes. Bar, 0.1 #m.
ceils (Fig. 5). Some staining was also observed between con-
nective tissue cells of the theca externa, and small spots were
occasionally present over the oocytes in locations consistent
with the junctions between granulosa cell processes and the
oocytes (1). The anti-connexin32 sera showed no reaction
with ovary.
Smooth muscle derived from the large intestine, stomach,
and pregnant uterus in labor all showed punctate labeling at
cell borders (data not shown). Uterine smooth muscle from
nonpregnant or preterm animals showed little staining. How-
ever, treatment of juvenile female rats with pharmacologic
doses of estradiol produced an intense area of reactivity in
a circular band of smooth muscle (Fig. 6 a). These estrogen-
treated animals also showed immunofluorescent staining
consistent with gap junctions in longitudinal smooth muscle,
in the connective tissue, and on the serosal epithelium. En-
dometrium showed no anti-connexin43 reactivity in any of
the uterine samples.
In contrast, anti-connexin32 showed no reactivity with any
of the smooth muscle preparations and little staining of con-
nective tissue. However, the endometrium of pregnant rats
showed large spots of anti-connexin32 staining along the
lateral epithelial cell membranes (Fig. 6 b). Such staining
was weak or absent in nonpregnant or estrogen-stimulated
animals.
The apparent localization of connexin43 in connective tis-
sue in a number of organs led us to examine the cornea,
which contains a connective tissue composed mainly of
stroma and fibroblasts (18). In the corneal stroma (Fig. 6 c),
anti-connexin43 stained discrete spots on the fibroblast cell
bodies in a distribution consistent with that demonstrated
previously for corneal fibroblast gap junctions (17). In addi-
tion, anti-connexin43 stained numerous contacts between
corneal epithelial cells (Fig. 6 c), showing the highest den-
sity between the basal cells and a graded reduction as the
ceils approached the free surface of this stratified epithelium.
Anti-connexin43 also stained corneal endothelial contacts
(data not shown). Anti-connexin32 showed no reactivity in
frozen sections of cornea.
Multiple Connexins in the Same Organ or 1Issue
Irnmunofluorescence was used to examine several rat organs
where previous immunological, physiological, biochemical,
or Northern blot experiments have suggested the presence of
more than one gap junction protein. In the lens, anti-con-
Beyer et al. Connexin43 lmmunolocalization 599
Figure 4. Immtmoblot of crude membrane preparations of multiple organs with anti-connexin43 sera. (a) Coomassie Blue-stained SDS-
polyacrylamide identical to those transferred to nitrocellulose. (Lane S) Molecular mass standard proteins (phosphorylase B, 97 kD; BSA,
66 kD; ovalbumin, 43 kD; carbonic anhydrase, 31 kD; soybean trypsin inhibitor, 21.5 kD; lysozyme, 14 kD). (Lanes 1-6) Preparations
of membranes from heart, kidney, ovary, lens, estrogen-stimulated uterus, and cornea, respectively. (b) Immunoblot of a gel identical to
that in a, but without standards, reacted with anti-connexin43 119-142. This antiserum reacts strongly with a 43-kD band in all the prepara-
tions, and weakly with other bands. (c) Identical immunoblot reacted with anti-connexin43 252-271 which had been absorbed with a crude
homogenate of rat liver. This antiserum also reacts strongly with a 43-kD band in all preparations. Since this antiserum does not react
with liver gap junctions, absorption with the liver homogenate was used empirically and found to reduce reactivity with polypeptides other
than that of 43 kD.
nexin43 stained spots along the most apical region of the epi-
thelial cells' lateral borders (Fig. 7, a and b) in locations con-
sistent with the distribution of epithelial-epithelial junctions
(28), and consistent with previous Northern blots of mRNA
fractions from capsulated and decapsulated lenses (2). Anti-
connexin43 showed no reaction with the fiber cells. In con-
trast, 6-4-B2-C6, an antibody to MP70, gave punctate stain-
ing between all the lens fiber cells (Fig. 7, c and d), and no
interpretable interepithelial staining.
In pancreas, anti-connexin32 gave bright staining in spots
between the acinar cells (data not shown). A few spots were
also seen in islets. In contrast, anti-connexin43 showed little
or no reactivity with the exocrine or endocrine pancreas but
produced bright spots between the cells of ductules and in
the connective tissue between lobules.
Northern blots (2) have suggested the presence of both con-
nexin32 and cormexin43 in kidney. Double immunofluores-
cence of kidney sections with the rabbit antipeptide connex-
in43 antiserum and the mouse monoclonal anti-connexin32
antibody M12.13 was performed. Connexin43 reactivity
(Fig. 8 a) localized to spots between the epithelial cells in
virtually all tubules and to some spots in the glomeruli and
connective tissue. Connexin32 reactivity (Fig. 8 b) also
localized to spots in glomeruli and between epithelial cells
in proximal tubules, but showed no reactivity with the distal
tubules. Although both connexins were localized to many of
the same proximal tubules, few or none of the immunofluo-
rescent spots coincided, suggesting that the two proteins
were present in different gap junction plaques. The complex
interdigitation of the lateral borders of these cells denied the
determination by immunofluorescence of whether an indi-
vidual cell coexpressed both connexin types.
Discussion
Gap Junction Antibodies and Connexin Topology
Antisera directed against gap junction proteins have been
raised in several laboratories and show different tissue spec-
ificities. Several antisera raised against the isolated rat liver
gap junction protein, connexin32, show immunofluorescent
staining consistent with gap junctions between hepatocytes,
pancreatic acinar cells, proximal renal tubule cells, and vari-
ous gastrointestinal epithelial cells (8, 19, 31). However,
while most antisera to this protein do not react with heart,
one stained intercalated discs (19). Another single report
suggested reactivity with lens fiber cells (34). Zervos et al.
(36) described an antiserum directed against an NH2-ter-
minal peptide that reacts on immunoblots with proteins from
liver, heart, and uterus. However, no immunohistochemistry
was performed with this antiserum.
Recent amino acid sequence data from Edman degradation
and cDNA cloning experiments have helped explain these
discrepancies (2, 3, 21, 29, 32). Connexin32, connexin43,
connexin46, and MPT0 all have regions at their amino ter-
mini containing many identical amino acids. The connexins
also have a second internal region of high homology (3). But
they have other regions that are unique, containing few iden-
tical residues. Topological models of the connexin proteins
(2, 14, 37) suggest that the unique regions may be located
on the cytoplasmic face of the junctional membrane, and the
conserved sequences may lie in transmembrane and extracel-
lular areas. Thus, the connexins potentially contain many
shared antigenic sites; but most of the shared epitopes result-
ing from conserved amino acid sequences may be inaccessi-
The Journal of Cell Biology, Volume 108, 1989 600
Figure 5.
Immunohistochemical lo-
calization of connexin43 in a frozen
section of rat ovary. Anti-connex-
in43 (252-271) staining is observed
in many large bright spots between
granulosa cells and in rare small
spots over the oocyte and in the
theca externa. Bar, 10 #m.
ble to antibodies applied to the cytoplasmic face of the native
gap junction, if they are buried in the hydrophobic bilayer
or sequestered within the minute extracellular space. Anti-
sera may also recognize different connexin molecules in cy-
toplasmic areas without amino acid sequence homology if
they recognize secondary or tertiary structures (conforma-
tional determinants). We have no evidence for such antibodies
within our reagents; however, the broad species and tissue
specificity reported for some anti-connexin32 antibodies sug-
gests the existence of such antibodies (12, 35).
We have used the sequence data to choose peptides to elicit
specific anti-connexin antisera. Localization of the reactivity
of these antisera supports the model of connexin topology.
The antiserum to connexin32 peptide 164-189, which may
contain extracellular epitopes and which is predicted to be
60% conserved in connexin43, reacts on immunoblots with
proteins of the correct size in both liver and heart. However,
because the epitopes recognized by this antiserum are only
accessible after experimental splitting of the junctional mem-
branes (14), this antiserum shows no reactivity with the na-
tive gap junction structure in either liver or heart. Antisera
directed against unique cytoplasmically located peptide se-
Beyer et al.
Connexin43 lrnmunolocalization
601
Figure 6. Immunohistochemical localization of connexins in uterus and cornea. (a) In uterine tissue from a juvenile female rat treated
with estradiol, intense anti-connexin43 (252-271) staining is observed in a circular band of smooth muscle. Punctate staining with this
antiserum is also present in longitudinal smooth muscle, in the connective tissue, and in the serosal epithelium. The endometrial epithelium
shows no reactivity with the cormexin43 antisera. Bar, 25/~m. (b) Endometrial epithelium from a pregnant rat near term stained with anti-
connexin32 shows large bright spots between epithelial cells. Bar, 25/zm. (c) In rat cornea, intercellular borders between corneal epithelial
cells and fibroblasts show punctate anti-connexin43 (252-271) reactivity. Bar, 10 #m.
quences from connexin32 and connexin43 are connexin spe-
cific and have been used in this paper to examine the distribu-
tions of these gap junction proteins.
Connexin43 Is a Cardiac Gap Junction Protein
The demonstration that a protein is a gap junction protein re-
quires the satisfaction of rigorous morphological and func-
tional criteria, as we have discussed previously (2). In this
paper, we have prepared two antisera that are specific for pep-
tide sequences unique to connexin43 and that recognize a 43-
kD protein in intercalated disc preparations. These antisera
do not react with liver gap junctions either on Western blots
or by immunofluorescence. These antisera specifically label
the cardiac gap junction structure as examined by immuno-
histo- and immunocytochemistry, providing a morphological
demonstration that connexin43 is a gap junction protein. A
functional demonstration might be provided by showing that
that connexin43-specific antisera can block intercellular com-
munication in pairs of cardiac myocytes; however, the per-
formance and interpretation of such experiments is very diffi-
cult, as discussed previously (2). We have recently shown
that connexin43 expressed in pairs of Xenopus oocytes from
in vitro-synthesized mRNA will form functional cell-to-cell
channels (Swenson, K. I., Beyer, E. C., Paul, D. L., Good-
enough, D. A., manuscript in preparation).
Connexin43 Is a Gap Junction Protein in
Other 7Issues
Previous Northern blot experiments had suggested that con-
nexin43 mRNA was expressed in many organs besides heart.
In this paper, immunofluorescence studies have shown stain-
ing consistent with gap junctions in many tissues using the
anti-connexin43 sera. The antipeptide antisera used in this
paper have revealed intercellular, macular staining between
many different cell types, consistent in size and distribution
with gap junctions as visualized by thin section and freeze-
fracture methodologies. We will refer to this fluorescence
The Journal of Cell Biology, Volume 108, 1989 602
Figure
7. Immunohistochemical localization of connexins in frozen sections of rat lens. (a) Connexin43 (252-271) localizes to bright spots
in the apical regions of the lateral contacts between adjacent lens epithelial cells in a distribution consistent with that of epithelial-epithelial
gap junctions. (b) Phase-contrast micrograph of same field as a. (c) Immunohistochemistry using anti-MP70 antibody 6-4-B2-C6 produces
staining between lens fiber cells in a distribution consistent with that of fiber-fiber gap junctions. (d) Phase-contrast micrograph of same
field as c. Bar, 10 #rn.
staining as gap junctional, mindful that electron microscopic
localization has not yet been done.
Connexin43 appears to be associated with gap junctions
between smooth muscle cells. It is found between cells in
connective tissues in many organs, and since connexin43 can
be localized between corneal stromal fibroblasts, part of this
connective tissue staining probably arises from interfibroblast
gap junctions. Connexin43 localizes in the gap junctions be-
Beyer et al.
Connexin43 Immunolocalization 603
Figure 8. Double-label immunohistochemical localization of connexins in a frozen section of rat kidney. (a) Reactivity with anti-connexin43
252-271 antiserum. (b) Reactivity with anti-connexin32 monoclonal antibody M12.13. Both antibodies stain spots between renal tubular
epithelial cells. However, few, if any, of the spots from the two antibodies coincide, suggesting that they are reacting with different gap
junction plaques; the two connexins are not mixed in the same plaque. Distal tubules (.) only show cormexin43 reactivity. Bar, 10 ~m.
tween granulosa cells and possibly between the oocyte and
granulosa cells. Connexin43 is contained in gap junctions
between epithelial cells in the lens, cornea, and renal tubules.
In smooth muscle, as in the heart, gap junctions permit
electrical coupling (7) but common properties of the other
tissues that contain connexin43 are unclear. However, we
now have developed appropriate connexin-specific antibody
probes that will facilitate the study of gap junction function
in these diverse locations.
Connexin43 Makes Different Junctions from Other
Gap Junction Proteins
Nicholson et al. (30) showed intermixing of two different
connexin molecules within the same gap junction plaques in
mouse liver. While we have seen no evidence yet for inter-
mixing of connexin43 with other connexins, we have found
a few cases where connexin43 is localized in the same organ
as another gap junction protein. In uterus and pancreas, for
example, both connexin43 and connexin32 are present but in
different cell types. In kidney, both connexin43 and connex-
in32 are present in the proximal convoluted tubule, but they
are present in different plaques and likely in different cells.
This suggests that adjacent renal tubular cells can differ in
connexin expression. Interestingly, Brown et al. (4) have
found similar local heterogeneity of H+-ATPase expression
in kidney tubules.
Finally, in the lens, connexin43 localizes to epithelial cells
(probably epithelial-epithelial gap junctions) and the con-
nexin-related protein MP70 localizes to fiber-fiber junctions.
We do not yet know how the heterocellular junctions are
made between epithelial and fiber cells.
We are grateful to Dr. A. Sytkowski (New England Deaconess Hospital,
Boston) for his helpful advice regarding the preparation of antipeptide an-
tisera. J. Jordan provided invaluable technical assistance in the isolation of
liver and lens gap junctions.
E. C. Beyer is the recipient of a Clinician-Scientist award from the
American Heart Association. This work was also supported by grants
GM18974, EY02430 (to D. A. Goodenough), and GM37751 (to D. L.
Paul) from the National Institutes of Health.
Received for publication 18 May 1988 and in revised form 19 October
1988.
The Journal of Cell Biology, Volume 108, 1989 604
References
1. Albertini, D. F., and E. Anderson. 1974. The appearance and structure of
intercellular connections during the ontogeny of the rabbit ovarian follicle
with particular reference to gap junctions.
J. Cell Biol.
63:234-250.
2. Beyer, E. C., D. L. Paul, and D. A. Goodenough. 1987. Connexin43: a
protein from rat heart homologous to a gap junction protein from liver.
J.
Cell Biol.
105:2621-2629.
3. Beyer, E. C., D. A. Goodenough, and D. L. Paul. 1988. The connexins:
a family of related gap junction proteins.
In
Gap Junctions. E. L. Hertz-
berg and R. Johnson, editors. Alan R. Liss Inc., New York. 167-175.
4. Brown, D., S. Hirsch, and S. Gluck. 1988. An H+-ATPase in opposite
plasma membrane domains in kidney epithelial subpopulations.
Nature
(Lond.). 331:622-624.
5. Caspar, D. L. D., D. A. Goodenough, L. Makowski, and W. C. Phillips.
1977. Gap junction structures. I. Correlated electron microscopy and
x-ray diffraction.
J. Cell Biol.
74:605-628.
6. Dahl, G., T. Miller, D. Paul, R. Voellmy, and R. Werner. 1987. Expres-
sion of functional cell-cell channels from cloned rat liver gap junction
complementary DNA.
Science (Wash. DC).
236:1290-1293.
7. Daniel, E. E. 1987. Gap junctions in smooth muscle.
In
Cell-to-Cell Com-
munication. W. C. DeMello, editor. Plenum Publishing Corp., New
York. 149-186.
8. Dermietzel, R., A. Leibstein, U. Frixen, U. Janssen-Timmen, O. Traub,
and K. Willecke. 1984. Gap junctions in several tissues share antigenic
determinants with liver gap junctions.
EMBO (Eur. Mol. Biol. Organ.)
J. 3:2261-2270.
9. Fallon, R. F., and D. A. Goodenough. 1981. Five-hour half-life of mouse
liver gap junction protein.
J. Cell Biol.
90:521-526.
10. Fawcett, D. W., and N. S. McNutt. 1969. The ultrastructure of the cat myo-
cardium. I. Ventricular papillary muscle.
J. Cell Biol.
42:1-45.
1 !. Forbes, M. S., and N. Sperelakis. 1985. Intercalated discs of mammalian
heart: a review of structure and function.
Tissue & Cell.
17:605-648.
12. Fraser, S. E., C. R. Green, H. R. Bode, and N. B. Gilula. 1987. Selective
disruption of gap junctional communication interferes with a patterning
process in Hydra.
Science (Wash. DC).
237:49-55.
13. Garfield, R. E., E. E. Daniel, M. Dukes, and J. D. Fitzgerald. 1982.
Changes of gap junctions in myometrium of guinea pig at parturition and
abortion.
Can. J. Physiol. Pharmacol.
60:335-341.
14. Goodenough, D. A., D. L. Paul, and L. A. Jesaitis. 1988. Topological dis-
tribution of two connexin32 antigenic sites in intact and split rodent hepa-
tocyte gap junctions.
J. Cell Biol.
107:1817-1824.
15. Green, C. R., and N. J. Severs. 1983. A simplified method for the rapid
isolation of cardiac intercalated discs.
Tissue & Cell.
15:17-26.
16. Gruitjers, W. T. M., J. Kistler, S. Bullivant, and D. A. Goodenough. 1987.
Immunolocalization of MP70 in lens fiber 16-17-rim intercellular junc-
tions.
J. Cell Biol.
104:565-572.
17. Hasty, D. L., and E. D. Hay. 1977. Freeze-fracture studies of the develop-
ing cell surface. I. The plasmalemma of the corneal fibrnblast.
J. Cell
Biol.
72:667-686.
18. Hay, E. D., and J. P. Revel. 1969. Fine structure of the developing avian
cornea.
In
Monographs in Developmental Biology. Vol. 1. A. Wolsky
and P. S. Chert, editors. S. Karger, Basel. 1-144.
19. Hertzberg, E. L., and R. V. Skibbens. 1984. A protein homologous to the
27,000 dalton liver gap junction protein is present in a wide variety of
species and tissues.
Cell.
39:61-69.
20. Kistler, J., B. Kirkland, and S. Bullivant. 1985. Identification ofa 70,000-D
protein in lens membrane junctional domains.
J. Cell Biol.
101:28-35.
21. Kistler, J., D. Christie, and S. Bullivant. 1988. Homologies between gap
junction proteins in lens, heart, and liver.
Nature (Lond.).
331:721-723.
22. Kumar, N. M., and N. B. Gilula. 1986. Cloning and characterization of
human and rat liver cDNAs coding for a gap junction protein.
J. CellBiol.
103:767-776.
23. MacKenzie, L. W., and R. E. Garfield. 1985. Hormonal control of gap
junctions in the myometrium.
Am. J. Physiol.
248:C296-C308.
24. Makowski, L., D. L. D. Caspar, W. C. Phillips, and D. A. Goodenough.
1977. Gap junction structures. II. Analysis of the x-ray diffraction data.
J. Cell Biol.
74:629-645.
25. Manjunath, C., and E. Page. 1985. Cell biology and protein composition
of cardiac gap junctions.
Am. J. Physiol.
248:H783-H791.
26. Manjunath, C. K., B. J. Nicholson, D. Teplow, L. Hood, E. Page, and J.-P.
Revel. 1987. The cardiac gap junction protein (Mr 47,000) has a tissue-
specific cytoplasmic domain of Mr 17,000 at its carboxy-terminus.
Bio-
chem. Biophys. Res. Commun.
142:228-234.
27. McNutt, N. S., and D. W. Fawcett. 1969. The ultrastructure of the cat myo-
cardium. II. Atrial muscle.
J. Cell Biol.
42:46-67.
28. Miller, T. M., and D. A. Goodenough. 1986. Evidence for two physiologi-
cally distinct gap junctions expressed by the chick lens epithelial cell. J.
Cell Biol.
102:194-199.
29. Nicholson, B. J., D. B. Gros, S. B. H. Kent, L. E. Hood, and J.-P. Revel.
1985. The Mr 28,000 gap junction proteins from rat heart and liver are
different but related.
J. Biol. Chem.
260:6514-6517.
30. Nicholson, B. J., R. Dermietzel, D. Teplow, O. Traub, K. Willecke, and
J.-P. Revel. 1987. Two homologous protein components of hepatic gap
junctions.
Nature (Lond.).
329:732-734.
31. Paul, D. L. 1985. Antibody against liver gap junction 27-kD protein is tis-
sue specific and cross-reacts with a 54-kD protein.
In
Gap Junctions.
M. V. L. Bennett and D. C. Spray, editors. Cold Spring Harbor Labo-
ratory, Cold Spring Harbor, NY. 107-122.
32. Paul, D. L. 1986. Molecular cloning of cDNA for rat liver gap junction
protein.
J. Cell Biol.
103:123-134.
33. Paul, D. L., and D. A. Gcx~enough. 1983. Preparation, characterization,
and localization of antisera against bovine MP26, an integral protein from
lens fiber plasma membrane. J.
Cell Biol.
96:625-632.
34. Traub, O., and K. Willecke. 1982. Cross reaction of antibodies against
liver gap junction protein (26K) with lens fiber junction protein (MIP)
suggests structural homology between these tissue specific gene products.
Biochem. Biophys. Res. Commun.
109:895-901.
35. Warner, A. E., S. C. Guthrie, and N. B. Gilula. 1984. Antibodies to gap-
junctional protein selectively disrupt junctional communication in the
early amphibian embryo.
Nature (Lond.).
311:127-131.
36. Zervos, A. S., J. Hope, and W. H. Evans. 1985. Preparation of a gap junc-
tion fraction from uteri of pregnant rats: the 28-kD polypeptides of
uterus, liver, and heart gap junctions are homologous.
J. Cell Biol.
101 : 1363-1370.
37. Zimmer, D. B., C. R. Green, W. H. Evans, and N. B. Gilula. 1987. Topo-
logical analysis of the major protein in isolated intact rat liver gap junc-
tions and gap junction-derived single membrane structures. J.
Biol.
Chem.
262:7751-7763.
Beyer et aL
Connexin43 lmmunolocqlization 605