Heteroligomers of type-I and type-III inositol trisphosphate receptors in WB rat liver epithelial cells.

We have previously shown that a 222-kDa polypeptide co-immunoprecipitates together with the type-I myoinositol 1,4,5-trisphosphate receptor (IP3R) in WB rat liver epithelial cell extracts, when the immunoprecipitation is carried out with a type-I isoform specific antibody (Joseph, S. K.(1994) J. Biol. Chem. 269, 5673-5679). Utilizing isoform-specific antibodies raised to unique sequences within the COOH-terminal region of IP3 receptors, we now report that the co-immunoprecipitating 222-kDa polypeptide is the type-III IP3R isoform and that type-III IP3R antibodies (Abs) can co-immunoprecipitate the type-I IP3R isoform. Co-immunoprecipitation of IP3R isoforms was not due to cross-reactivity of the antibodies for the following reasons: (a) on immunoblots the type-III antibodies did not cross-react with type-I IP3R and vice versa; (b) inclusion of the COOH-terminal type-III peptide had no effect on the ability of type-I IP3R Ab to co-immunoprecipitate the type-III IP3R but blocked the ability of type-III IP3R Ab to co-immunoprecipitate the type-I isoform; and (c) crude hepatocyte lysates contain undetectable amounts of type-III IP3R, and immunoprecipitation with type-III IP3R Ab does not co-immunoprecipitate any other isoforms. However, type-I and type-II IP3R isoforms were co-immunoprecipitated by their respective antibodies in hepatocyte lysates. Sucrose density gradient analysis of WB cell lysates indicated that the co-immunoprecipitating fraction is exclusively located at the density expected for tetrameric receptors, suggesting that co-immunoprecipitation was not a reflection of the nonspecific aggregation of IP3R isoforms. Phosphorylation of either type-I or type-III immunoprecipitates by protein kinase A indicated that only the type-I IP3R could be phosphorylated in vitro. Fractionation of WB cell membranes and immunofluorescence studies showed that the type-I and type-III isoforms have very similar sub-cellular localizations. We conclude that the WB cell contains both type-I and type-III IP3R isoforms and that a proportion of these receptors exist as heterotetramers.

The discharge of Ca 2ϩ from intracellular stores in response to the activation of cell surface receptors is mediated by the interaction of inositol 1,4,5-trisphosphate with a specific receptor that functions as a ligand-gated Ca 2ϩ channel (1)(2)(3). Molecular cloning studies have revealed the presence of three types of receptors encoded by separate genes. The type-I IP 3 R 1 is particularly enriched in the cerebellum region of the brain and has a calculated molecular mass based on cDNA sequence of 313 kDa (4,5). The purified receptor has been shown to be a homotetramer (6). The type-I receptor mRNA undergoes alternative splicing in two distinct regions of the receptor designated S1 and S2. Several studies have shown that the type-I receptor with the S2 region deleted is the predominant form in peripheral tissues (7,8). The type-II IP 3 R is 69% homologous to the type-I IP 3 R and was originally described as being expressed in brain (9). Subsequent studies have since shown the presence of type-II IP 3 R mRNA in several peripheral tissues (10). The type-III IP 3 R has 62% homology to the type-I IP 3 R and is expressed in several epithelial cells including those of the kidney, pancreas, and intestinal tract (11,12). Only partial sequences of a type-IV (13) and type-V receptor (10) have been reported, and their distribution and properties have not been characterized.
In a previous study, we investigated the biosynthesis and turnover of IP 3 R in cultured WB rat liver epithelial cells (14). We found that immunoprecipitation of 35 S-labeled WB cell extracts with type-I IP 3 R antibody resulted in the appearance of two 35 S-labeled polypeptides with molecular masses of 235 and 222 kDa. Only the 235-kDa band was immunoreactive with type-I-specific IP 3 R antibody. The 222-kDa band, which was present in lower amounts relative to the 235-kDa band, was not identified but was shown not to be a proteolytic clip of the type-I IP 3 R. Based on the cDNA the type-III IP 3 R has a calculated molecular mass of 304-kDa and would therefore be expected to run at a lower molecular mass than type-I IP 3 R on SDS-PAGE. In the present study we have utilized isoformspecific IP 3 R antibodies to show that the 222-kDa band indeed corresponds to the type-III IP 3 R. Additional data are shown demonstrating co-localization of both isoforms in WB cells, and evidence is presented to indicate that a proportion of both isoforms exist as heterotetrameric complexes.

MATERIALS AND METHODS
Antibodies-The isoform-specific antibodies used in these studies were raised against the unique COOH-terminal sequences of IP 3 receptors. The type-I Ab was a polyclonal anti-peptide antibody raised to amino acids 2731-2749 of the rat type-I IP 3 R and has previously been characterized (15,16). Two different type-III Abs were used in the present studies. The anti GST-H3CT polyclonal antibody was raised against a glutathione S-transferase fusion protein made with the 27 COOH-terminal amino acids of the human type-III IP 3 R. This antibody was affinity purified using an Affi-Gel fusion protein column (12). The second type-III antibody was a monoclonal anti-peptide antibody raised to amino acids 2657-2671 of the human type-III IP 3 R (17). This and additional monoclonal antibodies to the type-I and type-II IP 3 R isoforms were a kind gift of Dr. Mamoru Hasegawa (Kyowa Hakko Kogyo Ltd., Tokyo, Japan). The reactivity of these monoclonal antibodies in hematopoietic cells has been detailed previously (17). Antibodies to the rough ER marker protein ribophorin-I and to the nuclear marker protein lamin-B were generously provided by Dr. Christopher Nicchitta (Duke University, Durham, NC) and Dr. Gerd Maul (Wistar Institute, Philadelphia, PA), respectively. Antibody to the plasma membrane marker Na ϩ /K ϩ -ATPase (␣-subunit) was purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). All peptides used to block the activity of IP 3 R antibodies were derived from the COOH-terminal rat IP 3 R sequences and were synthesized with an additional NH 2 -terminal cysteine. Their sequences are as follows: CT-1, RIGLLGHPPHMNVNPQQPA; CT-2, FLGSNTPHENHHMPPH; and CT-3, QRLGFVDVQNCMSR.
Metabolic Labeling and Immunoprecipitation of WB Cell Extracts-WB cells, a clonal cell line derived from rat liver (18) were cultured in Richter's modified minimum Eagle's medium containing 5% fetal bovine serum. Cells were grown in 100-mm dishes to confluence and were used between passages 25 and 30. For labeling, the Richter's medium was replaced by methionine-free Dulbecco's modified Eagle's medium containing 20 Ci/ml Trans 35 S-label (ICN Radiochemicals) and incubated for 27 h. At the end of the labeling period, the medium was removed and the plates were washed twice in ice-cold phosphate-buffered saline solution. The cells were scraped into 1 ml of a buffer containing 150 mM NaCl, 50 mM Tris-HCl (pH 7.8), 1% Triton X-100 (w/v), 1 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, and 5 g/ml each of aprotinin, soy bean Trypsin inhibitor, and leupeptin (WB solubilization buffer). The cells were solubilized on ice for 30 min. Insoluble material was removed by centrifugation for 10 min at 25,000 ϫ g. All the extracts were precleared for 2 h using 100 g of pre-immune IP 3 R antibody and 25 l of a 50% (v/v) slurry of Staphylococcus aureas cell wall (Pansorbin, Calbiochem). IP 3 R was immunoprecipitated from the extracts by overnight treatment with IP 3 R isoform-specific antibodies. Immune complexes precipitated with protein A-Sepharose were washed once in WB solubilization buffer containing an additional 0.5 M NaCl and 0.1% SDS and washed twice in WB solubilization buffer alone. The labeled proteins were analyzed by 5% SDS-PAGE. The polypeptides in the gel were transferred to nitrocellulose that was autoradiographed and then immunoblotted with IP 3 R-antibody to locate the receptor. Immunoreactive bands were detected with an enhanced chemiluminescence assay (Amersham Corp.). When necessary, immunoblots were stripped using a 60°C treatment (30 min) of the blots in stripping buffer, which contained 65 mM Tris-HCl (pH 6.8), 2% SDS, and 100 mM ␤-mercaptoethanol.
Separation of Membrane Fractions on Sucrose Density Gradients-Confluent WB cells were removed from culture flasks by trypsinization and pelleted by centrifugation (5 min at 100 ϫ g). The cells were resuspended in 2 ml of TKM buffer, which contained 50 mM Tris-HCl (pH 7.5), 25 mM KCl, and 5 mM MgCl 2 supplemented with 0.25 M sucrose. The resuspended cells were disrupted using a Polytron tissue homogenizer. Trypan blue staining of cells was used to verify disruption. The 2-ml cell homogenate was loaded on to a 14-ml linear gradient of 50 -80% sucrose prepared in TKM buffer. The gradients were centrifuged for 1 h at 4°C in a SW 28.1 rotor at 37,000 ϫ g. The gradients were fractionated from the top of the tube using a Haake-Buchler gradient fractionator. Microsomal membranes were prepared from rat cerebellum homogenates by differential centrifugation as described previously (19). Isolated hepatocytes were prepared by collagenase digestion of perfused rat livers and were washed and stored on ice at 20 -30 mg of protein/ml in Ca 2ϩ /Mg 2ϩ -free Hanks' buffer as described previously (20).
Indirect Immunofluorescence Localization-Subconfluent monolayers of WB cells were grown on glass coverslips. The cells were fixed in 4% paraformaldehyde in phosphate-buffered saline for 10 min, washed, and then incubated for a further 10 min in phosphate-buffered saline containing 0.1% saponin. Nonspecific binding sites were blocked by 60-min incubation in phosphate-buffered saline containing 5% fetal bovine serum and 1% bovine serum albumin. The cells were then incubated with primary antibody (1:200) diluted in blocking buffer. Incubations with monoclonal IP 3 R antibodies (1:50) were carried out at 4°C overnight, whereas polyclonal antibody incubations were for 1 h at 37°C. The cells were incubated with 1:100 fluorescein isothiocyanateconjugated anti-mouse or anti-rabbit antibodies for 1 h at 37°C. Secondary antibodies were purchased from Upstate Biotechnology, Inc.

Co-immunoprecipitation of IP 3 R Isoforms in WB Cell
Lysates-We have previously shown that affinity purified antibody directed at the COOH terminus of the type-I IP 3 R immunoprecipitates two 35 S-labeled polypeptides from detergent extracts of 35 S-labeled WB rat liver epithelial cells (14). The major band at 235 kDa corresponded to type-I IP 3 R; the minor band at 222 kDa was not immunoreactive on Western blots against type-I IP 3 R antibody and could be shown not to be a proteolytic clip of the 235-kDa band by several criteria. Unlike the 235-kDa polypeptide, the 222-kDa polypeptide was not phosphorylated by protein kinase A and was not recognized by biotinylated concanavalin A (14). The molecular weight of the type-III IP 3 R is smaller than that of type-I IP 3 R and several epithelial cells have been reported to contain the type-III IP 3 R isoform (12,21). In order to investigate whether the unidentified-222 kDa band corresponds to the type-III IP 3 R, we utilized an affinity purified antibody directed at the 27 carboxyl-terminal amino acids of the type-III IP 3 R. Both type-I-and type-IIIspecific antibodies were used to immunoprecipitate detergent extracts of 35 S-labeled WB cells, and the immunoprecipitates were analyzed by autoradiography (Fig. 1A). The type-I immunoprecipitate contained the expected doublet of bands corresponding to the 235-and 222-kDa 35 S-labeled bands, as reported previously (14). The type-III immunoprecipitate contained a prominent labeled band whose mobility corresponded exactly to that of the lower 222-kDa band (Fig. 1A, lane 2). A faint labeled band corresponding to the upper 235-kDa band could also be observed in the type-III immunoprecipitates. In order to verify the identification of the 235-and 222-kDa 35 S-labeled polypeptides as type-I and type-III IP 3 R isoforms, the same sheet of nitrocellulose used for autoradiography was immunoblotted with type-III and type-I IP 3 R Ab. On immunoblots, the type-III Ab showed reactivity solely to the lower 222-kDa polypeptide (Fig. 1B), whereas the type-I Ab was immunoreactive only against the upper 235-kDa band (Fig. 1C). Immunoblotting with both antibodies (Fig. 1D) qualitatively reproduced the doublet pattern seen in the autoradiographs.
From these data we conclude that the previously unidentified 35 S-labeled 222-kDa band present in type-I immunoprecipitates corresponds to the type-III IP 3 R. Because traces of type-I IP 3 R are also present in type-III immunoprecipitates, the data would also suggest that a pool of both receptors are associated and are co-immunoprecipitated. However, such a 35 S-labeled WB extracts. Cells were labeled with Trans 35 S-label and immunoprecipitated with either type-I Ab (lane 1) or type-III Ab (lane 2) as described under "Materials and Methods." The immunoprecipitates were run out on a 5% SDS gel together with an unlabeled sample of WB cell extract (lane 3). The polypeptides were transferred to nitrocellulose, which was then autoradiographed (Auto-rad, A). The same nitrocellulose sheet was then sequentially immunoblotted with type-III Ab (B), stripped, immunoblotted with type-I Ab (C), stripped, and immunoblotted with both Abs (D).

FIG. 1. Co-immunoprecipitation of type-III IP 3 R with type-I IP 3 R in
conclusion relies on the antibodies being entirely specific for their respective antigens. The recognition properties of the type-III polyclonal antibody (GST-H3CT Ab; Ref. 12) is shown in Fig. 2A. Detergent extracts of cerebellum microsomes, WB cells and hepatocytes were immunoblotted against type-I and type-III antiserum. As expected, the type-I receptor that is enriched in cerebellum produces a strong signal when probed against type-I Ab ( Fig. 2A, upper panel). However, the cerebellar type-I IP 3 R shows no cross-reactivity to the type-III Ab ( Fig.  2A, lower panel). Sufficient amounts of type-I and type-III receptor are present in 50 g of WB cell extract proteins to enable a clear signal to be obtained from immunoblotting without the necessity of immunoprecipitating the extracts. This is not the case for hepatocytes where a weak signal for type-I is seen only when 200 g of extract protein is loaded into a gel lane. The amount of type-III receptor in hepatocyte extracts was below the detection limit. This is in agreement with a recent study using reverse transcriptase-polymerase chain reaction that found only 3% of the total IP 3 R mRNA in rat liver encoded the type-III isoform (10). The co-immunoprecipitation of IP 3 R isoforms seen in Fig. 1 could be duplicated using monoclonal antibodies to type-I and type-III IP 3 R (data not shown). The immunoblot in Fig. 2B (lane 2) shows that the type-III monoclonal Ab also selectively recognizes only the lower 222-kDa 35 S-labeled band in the type-I monoclonal Ab immunoprecipitates.
A further criteria of specificity in antibody recognition is that immunoprecipitation of the target antigen should be suppressed by inclusion of the peptide epitope used in immunization. In the case of type-I IP 3 R antibody, we have previously shown that the CT-1 peptide blocks the immunoprecipitation of both 35 S-labeled bands from WB extracts (14). In Fig. 3 unlabeled WB extracts were immunoprecipitated with polyclonal type-I or type-III antibody in the presence or absence of CT-3 peptide. The immunoprecipitates were immunoblotted with either type-I- (Fig. 3A) or type-III-specific antibodies (Fig. 3B). The results show that the CT-3 peptide does not influence the ability of type-I antibody to immunoprecipitate the type-I IP 3 R or to co-immunoprecipitate the type-III IP 3 R. However, the CT-3 peptide markedly inhibits the ability of the type-III antibody to immunoprecipitate the type-III IP 3 R and co-immunoprecipitate the type-I IP 3 R isoform. This selectivity of peptide inhibition suggests that co-immunoprecipitation of receptors is not due to a lack of specificity in the recognition properties of the IP 3 R antibodies.
In Vitro Phosphorylation of IP 3 R Immunoprecipitates-In our previous study we showed that the 222-kDa polypeptide co-immunoprecipitating with the type-I IP 3 R was not a substrate for protein kinase A. If the 222-kDa polypeptide is the type-III IP 3 R, it would be predicted that the type-III IP 3 R would also not be a substrate for protein kinase A in vitro. It should be noted that the primary sequence of the type-III IP 3 R does contain five consensus protein kinase A phosphorylation sites (12). Fig. 4 shows the result of protein kinase A phosphorylation of type-I and type-III immunoprecipitates prepared from WB cell extracts. As observed previously (14), the WB type-I IP 3 R is readily phosphorylated by protein kinase A (Fig.  4, lane 1). The type-III immunoprecipitates were phosphorylated to a lesser extent and only a single phosphorylated band was observed (Fig. 4, lane 2). By immunoblotting of this lane (Fig. 4, lanes 3 and 4), it could be demonstrated that the single phosphorylated band corresponded to the type-I IP 3 R and that the type-III IP 3 R itself was not an protein kinase A substrate. The data do not exclude the possibility that the type-III isoform may be fully phosphorylated under these conditions. Subcellular Localization of IP 3 R Isoforms in WB Cells-We have used subcellular fractionation and immunocytochemistry as two different approaches to compare the distribution of IP 3 R isoforms in WB cells. Fig. 5 shows the distribution of type-I and type-III IP 3 R in WB membranes fractionated on a sucrose density gradient. Both isoforms had a similar distribution profile with peak immunoreactivity being located toward the bottom of the gradient. The distribution profile of IP 3 R did not overlap exactly with any of the three membrane markers examined. In particular, only a minor proportion of IP 3 R isoforms were found in the fractions containing ribophorin-I, a marker of the rough ER. A greater degree of overlap was obtained with those fractions containing nuclear and plasma membrane markers. The similar localization of both receptor isoforms was also observed when the peak fractions of IP 3 R immunoreactivity were pooled and fractionated further on a 15% Percoll gradient (data not shown). A similar localization of both iso-  forms was also observed in immunofluorescence studies of WB cells (Fig. 6). Immunofluorescence of type-I and type-III Ab in WB cells shows that these isoforms are excluded from the nucleus and are widely distributed throughout the cell in both diffuse and punctate structures (Fig. 6, A and B). With both isoforms the intensity of staining was highest in the perinuclear region. In agreement with the fractionation experiments (Fig. 5), the pattern of labeling with IP 3 R Abs was not identical to the pattern of labeling seen with anti-ribophorin Ab, which would be expected to stain rough ER membranes (Fig. 6C). The fluorescence signal was inhibited when type-III (Fig. 6D) or type-I Ab (data not shown) was preblocked with their respective antigenic peptides.
Heterotetramer Formation or Nonspecific Aggregation of Receptor Complexes-Nonspecific association of the two homotetrameric isoforms would be expected to generate large multimeric complexes with molecular masses that greatly exceed 10 6 daltons. However, complexes that are heterotetramers of two isoforms would still have a molecular mass of approximately 10 6 daltons. To identify the molecular mass of the complexes, we subjected WB cell lysates to sucrose gradient centrifugation (Fig. 7A). Sufficient lysate protein was loaded to enable the profile of both IP 3 R isoforms in the gradient to be determined by immunoblotting alone. These results showed that both isoforms had a similar distribution in the gradient. The predominant peak of immunoreactive protein was located at the migration position of tetrameric cerebellar type-I IP 3 R (Fig. 7B). In addition, a minor peak was seen at the position corresponding to IP 3 R monomer, and immunoreactive protein was also observed in intermediate fractions. When the fractions were immunoprecipitated with type-I Ab and then immunoblotted with type-I Ab, the overall profile of type-I IP 3 R was identical to that observed by direct immunoblotting of the fractions. This is the anticipated result if the antibody shows no preference for any particular oligomeric structure. Immunoblotting of the type-I immunoprecipitates with type-III anti-body indicated that the co-immunoprecipitating type-III IP 3 R was restricted to just three fractions that coincide with the expected position of tetrameric receptor (Fig. 7A). The same fractions also contained the type-I IP 3 R that was co-immunoprecipitated by type-III Ab (data not shown). There was no evidence from these experiments for the presence of large multimeric heterocomplexes migrating at the bottom of the gradient. We conclude that a fraction of type-I and type-III IP 3 R isoforms in WB cells exist as heterotetramers.
Co-immunoprecipitation of Receptor Isoforms in Hepatocyte Lysates-In order to determine if co-immunoprecipitation of isoforms is a unique property of WB cell lysates or to type-I and -III isoforms, we carried out experiments with hepatocyte lysates. Hepatocytes are reported to contain predominantly type-I and type-II mRNA (10), and this has been verified at the protein level by immunoblotting. 2 Fig. 8A shows a type-I IP 3 R Ab immunoblot of type-I, -II, and -III immunoprecipitates. It is clear that the type-II antibody can co-immunoprecipitate the type-I IP 3 R (Fig. 8A, lane 2). Similarly, type-I antibody coimmunoprecipitates some type-II IP 3 R (Fig. 8B, lane 1). In contrast to WB cells, the hepatocytes contain little type-III IP 3 R, and under the conditions of the experiment, the type-III antibody does not bring down detectable amounts of the type-III isoform (data not shown). The type-III antibody also does not co-immunoprecipitate any type-I or type-II IP 3 R (lanes 3 in Fig. 8, A and B). This attests to the specificity of the type-III antibody. As a further test of specificity, the COOH-terminal peptides of the types I and II IP 3 R were included during the immunoprecipitation protocol. The results show that the CT-2 peptide blocks the ability of the type-II antibody to bring down its target antigen (Fig. 8B, lane 7) and the co-immunoprecipitating type-I IP 3 R (Fig. 8A, lane 7). The CT-1 peptide did not interfere with the activity of type-II IP 3 R Ab (Fig. 8, A and B,  lane 6). However, the CT-1 peptide blocked the ability of the type-I antibody to bring down its target antigen (Fig. 8A, lane  4) and the co-immunoprecipitating type-II IP 3 R (Fig. 8B, lane  4). The CT-2 peptide did not interfere with the activity of type-I IP 3 R Ab (Fig. 8, A and B, lane 5). These data show that coimmunoprecipitation of IP 3 Rs extends to other combinations of isoforms and to other cell types besides WB cells. DISCUSSION The results of the present study establish that the lower component of the 35 S-labeled doublet of polypeptides immunoprecipitated by the type-I IP 3 R Ab from WB cell lysates is the type-III IP 3 R isoform. Although the presence of multiple IP 3 R isoforms in specific cell types has been inferred from mRNA analysis (10,22,23), the quantitation of relative levels of immunoreactive protein and the subcellular localization of multiple isoforms has been analyzed in very few cell types. In PC12 cells, an IP 3 R antibody that does not discriminate between isoforms has been shown to detect a doublet of polypeptides of which only the upper band is the type-I IP 3 R (22). The Jurkat T-cell has been shown to contain all three isoforms, mast cells contain both type-I and type-II, and only the type-II isoform can be detected in macrophages (17). WB rat liver epithelial cells contain high levels of both type-I and type-III IP 3 R isoforms with no detectable type-II isoform (data not shown). Isolated hepatocytes are different in that they have a much lower total expression of receptors and that they contain the type-I and type-II IP 3 R as the predominant isoforms. A quantitation of the relative abundance of the three isoforms in several cell lines and tissues has recently been published (24). In a megakaryocytic cell line, fluorescence confocal microscopy 2 C. Lin, G. Hajnoczky, S. K., Joseph, T. G. Schneider, and A. P. Thomas, manuscript in preparation. was used to show that the type-I and type-II receptors are localized differently (17). This is clearly different from our results in WB cells where no gross differences in localization of type-I and type-III isoforms were noted from subcellular fractionation or immunofluorescence measurements. Both approaches also showed that the localization of the IP 3 R isoforms was distinct from the distribution of ribophorin-I, a rough ER marker. This supports the conclusion that in some cell types the IP 3 R may be localized in a subcompartment of the ER (25)(26)(27).
The present study demonstrates that antibodies specific to one IP 3 R isoform can co-immunoprecipitate additional isoforms. The same finding was also noted by Wojcikiewicz in a recent study (24). Co-immunoprecipitation of different voltagegated K ϩ channels (28,29) has been used as evidence for the formation of heteromultimeric complexes in vivo. However, the validity of these conclusions relies on the absolute specificity of the antibodies that are used in this experimental approach. The isoform specificity of the IP 3 R anti-peptide antibodies used in the present studies are based on the following observations: (a) The antibodies are raised against the nonconserved COOHterminal regions and do not cross-react with multiple isoforms on immunoblots. Thus, the type-III polyclonal antibody does not react with the type-I IP 3 R (from cerebellum or WB cells), and the type-I polyclonal does not recognize the type-III IP 3 R. (b) The ability of an isoform-specific antibody to immunopre- FIG. 5. Comparison of the distribution of type-I and type-III IP 3 R receptors in WB cell membranes. WB cell homogenates were fractionated on 50 -80% sucrose density gradients as described under "Materials and Methods." 70 l of each fraction was treated with SDS sample buffer and electrophoresed on a 7% gel. After transfer, the nitrocellulose sheet was cut below the prestained myosin marker (214 kDa), and the upper part of the sheet was consecutively immunoblotted with type-I and type-III IP 3 R Ab. Similarly, the lower part of the nitrocellose sheet was probed consecutively with antibodies against ribophorin-I, lamin-B, and Na ϩ /K ϩ -ATPase.
FIG. 6. Indirect immunofluorescence of IP 3 R in WB cells. WB cells were grown on glass coverslips and processed for immunofluorescence as described under "Materials and Methods." All panels were photographed at the same magnification through a ϫ40 lens. A, staining with type-I antibody; B, staining with type-III antibody; C, staining with anti-ribophorin antibody; D, staining with type-III antibody preblocked with 100 g/ml of type-III COOH-terminal peptide.
cipitate that isoform and co-immunoprecipitate other isoforms is blocked by the COOH-terminal cognate peptide of that antibody but not by any other peptides. These results imply that co-immunoprecipitation occurs as a result of the association of receptor isoforms and not as a result of cross-reactivity. (c) The type-III IP 3 R is absent from hepatocyte lysates. Immunoprecipitation with the type-III antibody does not bring down any type-I or type-II IP 3 R, as would be expected if the observed antibody interactions are specific. (d) Finally, the experimental observations on co-immunoprecipitation and subcellular localization have been validated with two different antibodies for each isoform (data not shown).
We have interpreted our results to indicate that a pool of IP 3 receptors in WB cells exists as heterotetramers. The conclusion that the heterocomplexes are heterotetramers rather than large aggregates of receptor is based on our demonstration that the co-immunoprecipitating fractions of receptor have a migration position on sucrose density gradients that is consistent with that expected of a heterotetramer. Such experiments also revealed that freshly prepared WB cell lysates contained low amounts of disassembled receptors, including a pool of free monomers. The same observation was made in our previous studies when the lysates were fractionated by gel filtration (14). It is therefore possible that the heterotetramers assembled after lysis of the cells and do not occur in vivo. However, the proportion of 35 S-labeled type-I and type-III receptor immunoprecipitated by type-I antibody was not altered by lysing the cells in radioimmune precipitation buffer (which contained 0.1% SDS, 1% sodium deoxycholate, and 1% Triton X-100) or by pretreating the intact cells for 1 h with 100 g/ml of the cleavable cross-linking agent dithiobis(succinimidylpropionate) (data not shown). Because the amount of co-immunoprecipitating receptor cannot be changed by these experimental manipulations, we conclude that heteroligomerization of IP 3 R pre-exists in the native WB cell membranes.
A random association of two different monomers can generate five possible combinations of tetramers. The proportion of homotetrameric and heterotetrameric receptor would be dependent on the relative amounts of the type-I and type-III isoforms. The ratio of type-I and type-III isoforms in WB cells cannot be easily estimated by comparing immunoblots because of the difficulties of standardizing and comparing the reactivities of two different antibodies. An alternative is to quantitate the 35 S-labeled type-I and type-III isoforms immunoprecipitated by their respective antibodies from the lysate of 35 Slabeled cells (cf. Fig. 1A). A type-I/type-III ratio of 2.7 Ϯ 0.2 (n ϭ 3) is obtained by this method. Assuming an entirely random association, the relative proportions of the five tetrameric forms can be calculated from the expansion of the polynomial (a ϩ b) 4 ϭ 1, where a and b are the relative concentrations of the two different monomers (30). Taking a as the FIG. 7. Analysis of WB cell lysates on sucrose density gradients. 0.5-ml WB cell lysates (4 -5 mg protein/ml) were loaded on to 11-ml 5-20% sucrose density gradients prepared in a buffer containing 150 mM NaCl, 50 mM Tris (pH 7.8), 1 mM EDTA, and 0.25% Triton X-100. The tubes were centrifuged in a SW 41 Ti rotor at 100,000 ϫ g for 16 h. 34 fractions were collected from the top of each tube. Aliquots of alternate fractions were run out directly on SDS-PAGE for immunoblotting or were immunoprecipitated with the indicated antibody. Immunoblotting of gradients containing 0.5 ml of cerebellum microsome extract (1 mg protein/ml) or denatured cerebellum extract (treated for 15 min at 4°C with 0.2% SDS) were used to localize the position of type-I heterotetramer and monomer, respectively. Separate gradients containing 1 mg of ␤-amylase (200 kDa) or thyroglobulin (669 kDa) as size markers were processed, fractionated, and assayed for protein.
Only the data from WB cell extract fractionation is shown in A. B and C show the densitometric quantitation of the data in A and includes the peak localization of the size standards used in the experiment. q, type-I; E, type-III.
FIG. 8. Co-immunoprecipitation of IP 3 R isoforms in hepatocyte lysates. Isolated hepatocytes were solubilized as described previously (37). 1-ml aliquots of hepatocyte lysate (14 mg of protein/ml) were immunoprecipitated for 4 h at 4°C with 5 g of monoclonal antibodies to type-I, -II, or -III IP 3 R. Where indicated, the COOH-terminal peptides of the type-I and type-III IP 3 R were present at 50 g/ml. The immunoprecipitates were collected on protein A-Sepharose, run out on SDS-PAGE, and immunoblotted with either type-I (A) or type-II (B) monoclonal antibodies.
type-I and b as the type-III isoform and the relative concentrations of a ϭ 0.73 and b ϭ 0.27, the calculated proportions of homotetrameric type-I and type-III receptors would be 28 and 5% of the total receptor concentration with the remaining 67% as heterotetramers. Experimentally, the exact proportion of heterotetrameric type-I or type-III is difficult to estimate, in part due to the presence of nontetrameric forms in the lysate. However, it is clear from our data that heterotetramers do not represent 67% of either isoform (cf. Figs. 1 and 3). This suggests that they do not arise by random mixing of monomers. Lower amounts of heterotetramers may arise if not all combinations of monomers are equally favored or if the ability to form heterotetramers is restricted to a localized pool of receptors.
There are precedents for the combinatorial mixing of subunits of channel proteins derived from separate gene products. The formation of heteromultimeric members of the Shaker subfamily of K ϩ channels has been viewed as allowing considerable heterogeneity in channel gating and regulatory properties (28,29,31). The same could be true for the IP 3 receptor channel, because each monomer is believed to contribute transmembrane segments to the pore-forming domain of the tetrameric channel (2). A limiting concentration of IP 3 has been shown to open only a population of IP 3 R ion channels with further increases resulting in activation of additional channels (32,33). One interpretation of this "quantal" or "increment detection" behavior of the channel is that it arises from heterogeneity in sensitivity to IP 3 (34 -37). Recently, it has been demonstrated that bacterially expressed ligand-binding domains of the type-I and type-III IP 3 R have affinities for IP 3 that differ by an order of magnitude (22). A mathematical model incorporating the presence of high and low affinity IP 3 binding sites randomly associated as homo-and heterotetramers has been used to describe the kinetics of Ca 2ϩ release from rat basophilic leukemia cells (30). Heteroligomeric IP 3 R ion channels would be expected to have a wide range of sensitivity to IP 3 and potentially could have different channel gating characteristics. The inability to phosphorylate the type-III IP 3 R by protein kinase A also points to important differences in the regulation of IP 3 R isoforms. Heteroligomerization of isoforms may thus provide an additional mechanism that generates considerable diversity in the regulation of the IP 3 R Ca 2ϩ ion channel.