Functional Inositol 1,4,5-Trisphosphate Receptors Assembled from Concatenated Homo- and Heteromeric Subunits*♦

Background: IP3R forms homo- and heterotetrameric channels. However, the impact of the specific composition of the heterotetrameric channel is undefined. Results: Concatenated IP3R dimers formed functional tetrameric channels. Heterotetrameric channels containing IP3R1 and IP3R2 had identical properties to IP3R2. Conclusion: Heterotetrameric IP3R do not behave as a blend of the constituent monomers. Significance: This study represents the first demonstration of the properties of heterotetrameric IP3R of unambiguously defined composition. Vertebrate genomes code for three subtypes of inositol 1,4,5-trisphosphate (IP3) receptors (IP3R1, -2, and -3). Individual IP3R monomers are assembled to form homo- and heterotetrameric channels that mediate Ca2+ release from intracellular stores. IP3R subtypes are regulated differentially by IP3, Ca2+, ATP, and various other cellular factors and events. IP3R subtypes are seldom expressed in isolation in individual cell types, and cells often express different complements of IP3R subtypes. When multiple subtypes of IP3R are co-expressed, the subunit composition of channels cannot be specifically defined. Thus, how the subunit composition of heterotetrameric IP3R channels contributes to shaping the spatio-temporal properties of IP3-mediated Ca2+ signals has been difficult to evaluate. To address this question, we created concatenated IP3R linked by short flexible linkers. Dimeric constructs were expressed in DT40–3KO cells, an IP3R null cell line. The dimeric proteins were localized to membranes, ran as intact dimeric proteins on SDS-PAGE, and migrated as an ∼1100-kDa band on blue native gels exactly as wild type IP3R. Importantly, IP3R channels formed from concatenated dimers were fully functional as indicated by agonist-induced Ca2+ release. Using single channel “on-nucleus” patch clamp, the channels assembled from homodimers were essentially indistinguishable from those formed by the wild type receptor. However, the activity of channels formed from concatenated IP3R1 and IP3R2 heterodimers was dominated by IP3R2 in terms of the characteristics of regulation by ATP. These studies provide the first insight into the regulation of heterotetrameric IP3R of defined composition. Importantly, the results indicate that the properties of these channels are not simply a blend of those of the constituent IP3R monomers.

Functional characterization of individual IP 3 R subtypes have generally employed cell types expressing predominantly or, more recently, exclusively one IP 3 R subtype (26,34,37,43,44). The activity of individual homotetramers has been interrogated using single channel electrophysiology or fluorescence-based experiments providing fundamental information regarding the binding of IP 3 , together with the gating and regulation of the activity of individual IP 3 R subtypes (19,25,35,45). However, because virtually all cells express a complement of at least two IP 3 R subtypes, it is predicted that most cells express a combination of homo-and heterotetrameric channels (13,14,18). Assuming IP 3 R subunit oligomerization to form tetramers is a random process and follows a binomial distribution, co-expression of multiple IP 3 R subtypes would be predicted to lead to the formation of homomeric as well as heteromeric channels with many possible subunit (homo and hetero) combinations. Thus, IP 3 R activity in cells likely reflects an undefined amalgamation of many possible combinations of IP 3 R. Given the markedly distinct subtype-specific properties of individual IP 3 R (8), it is not clear at present how individual IP 3 R subunits within a heterotetrameric channel contribute to the tetrameric channel's "character" or define its overall functional properties.
This report is a first step toward assessing the activity of defined heteromeric IP 3 R channels. We have generated cDNA constructs encoding concatenated IP 3 R subunits to co-opt the cells' biosynthetic machinery into creating a tetrameric IP 3 R channel with predetermined and defined subunit composition. We show that, as with the expression of monomeric constructs, dimeric IP 3 R constructs were localized to membranes, ran as intact dimeric proteins on SDS-PAGE, and migrated as tetrameric IP 3 Rs on native blue gels. Importantly, IP 3 R channels formed from concatenated homodimers were fully functional as indicated by agonist-induced Ca 2ϩ release. Furthermore, they exhibited properties identical to channels formed from corresponding monomeric proteins when monitored at the single channel level by "on-nucleus" patch clamp. Notably, the activity of channels formed from concatenated IP 3 R1 and IP 3 R2 heterodimers was dominated by IP 3 R2 in terms of modulation of activity by ATP. Importantly, we demonstrate that heterotetramers may not simply behave as a blend of constituent subunits and thus may add an additional level of flexibility to the control of Ca 2ϩ signals through IP 3 R channels. These studies detail a novel experimental platform to study regulation of heterotetrameric IP 3 R activity.

MATERIALS AND METHODS
Reagents-Restriction enzymes and DNA T4 ligase were from New England Biolabs. RPMI 1640 media, penicillin/streptomycin, G418 sulfate, ␤-mercaptoethanol, chicken serum, and all reagents used for Native Blue PAGE were obtained from Invitrogen. Fetal bovine serum was purchased from Gemini. CHAPS was from G Biosciences. Enhanced chemiluminescent substrate and Dylight TM 800CW secondary antibodies were from Thermo Scientific. Horseradish peroxidase-conjugated secondary antibodies and all reagents used for SDS-PAGE were from Bio-Rad. Fura2-AM was from TEFLABS. Protein A/Gagarose beads, rabbit polyclonal antibodies raised against residues 1894 -1973 of human IP 3 R1, rabbit polyclonal anti-glyceraldehyde dehydrogenase (GAPDH), and anti-SERCA were obtained from Santa Cruz Biotechnology. Two rabbit anti-IP 3 R2 antibodies raised against amino acids 320 -338 or against amino acids 2686 -2701 in mouse IP 3 R2 were generated by Pocono Rabbit Farms and Laboratories. Mouse monoclonal antibody against residues 22-230 of human IP 3 R3 was from BD Transduction Laboratories. All other chemicals were obtained from Sigma unless otherwise indicated.
Plasmid Construction-All PCR steps were carried out using Pfu Ultra II Hotstart 2ϫ Master Mix (Agilent). Primers used in this study were synthesized by Integrated DNA Technologies and are listed in supplemental Table 1. To subclone two IP 3 R1 subunits, designated as head and tail, in a tandem fashion, IP 3 R1 cDNA in pcDNA3.1(ϩ) vector was modified as follows. To make the head subunit, NcoI sites in the IP 3 R1-coding sequence were silently mutated (primers 1-4), and a new NcoI site and Kozak sequence were introduced at the start codon (primer 5). To enable such insertion, a glycine codon (GGC) was inserted after the initiation methionine. The head subunit was further modified so that the stop codon was deleted, and a nucleotide sequence coding for the first half of the linker was inserted immediately after the coding sequence of IP 3 R1 followed by an AgeI site (primer 6). The tail subunit was made by introducing an AgeI site followed by a nucleotide sequence coding for the second half of the linker inserted immediately before the start codon (primer 7), and a blunt end restriction site (HpaI) was inserted immediately after the stop codon (primer 8). Plasmid DNAs were then digested accordingly, and fragments were gel-extracted and directly ligated between the two arms of pJAZZ mamm linear vector based on coliphage N15 (Lucigen Middleton, WI). The resultant construct codes for one open reading frame consisting of two IP 3 R1 subunits connected with a 14-amino acid linker (QLNQLQTGQLNQLQ).
The trimeric IP 3 R1 concatemer was made similarly by linking three IP 3 R1 subunits, designated I, II, and III, in a tandem Nto C-terminal fashion. In brief, to make subunit I, the head subunit used for the dimer construction was modified so that it has an NheI site instead of AgeI at the end of the coding sequence (primer 9). Subunit II was made by introducing an NheI site followed by a nucleotide sequence coding for the second half of the linker inserted immediately before the start codon (primer 10). The stop codon was deleted, and a nucleo-tide sequence coding for the first half of the linker was inserted immediately after the coding sequence of IP 3 R1 followed by an AgeI site (primer 11). Subunit III was as the tail subunit used for the dimeric construct. Subunit cDNAs were digested accordingly; fragments were directly ligated into pJAZZ mamm linear vector. The resultant construct codes for one open reading frame consisted of three IP 3 R1 subunits with a 14-amino acid linker (QLNQLQLAQLNQLQ) separating subunits I and II, and a 14-amino acid linker (QLNQLQTGQLNQLQ) separating subunits II and III.
The tetrameric IP 3 R1 construct was created by generating four IP 3 R subunits (I-IV). Subunit I of the tetramer was identical to subunit I of the trimeric construct. To make tetramer subunit II, the trimeric subunit II was modified by introducing a NotI restriction site instead of an AgeI site at the end of the coding sequence (primer 12). Subunit III was made by introducing a NotI restriction site followed by a nucleotide sequence coding for the second half of the linker sequence prior to the first codon (primer 13). The stop codon was deleted, and a nucleotide sequence coding for the first half of the linker was inserted at the end of the coding sequence, followed by an AgeI restriction site (primer 14). Subunit IV is the same as the tail construct used to generate dimers. Subunits I-IV were digested accordingly and subcloned into pJAZZ-mamm. The resulting conjoined cDNAs form one reading frame encoding four IP 3 R1 subunits with linker QLNQLQLAQLNQLQ, separating subunits I and II, linker QLNQLQAAAQLNQLQ separating subunits II and III, and linker QLNQLQTGQLNQLQ separating subunits III and IV. IP 3 R2 head and tail subunits were generated in a similar fashion. Briefly, to make the head subunit, IP 3 R2 cDNA in pcDNA3.1(ϩ) vector was modified so that all NcoI sites in the coding region were silently mutated (primers [15][16][17][18]. A glycine codon (GGC) was inserted after the initiation methionine to create a new NcoI site and a Kozak sequence in the context of the start codon (primer 19). The stop codon was deleted, and a nucleotide sequence coding for the first half of the linker was inserted immediately after the coding sequence of IP 3 R2 followed by an AgeI site (primer 20). Finally, IP 3 R2 tail subunit was made by introducing an AgeI site followed by a nucleotide sequence coding for the second half of the linker inserted immediately before the start codon (primer 21), and a blunt end restriction site (SnaBI) was inserted immediately after the stop codon (primer 22). All constructs were confirmed by sequencing.
Cell Transfection and Culture Conditions-DT40 -3KO cells, a chicken B lymphocyte line with targeted deletion of the three endogenous IP 3 R isoforms (44), were grown in RPMI 1640 media supplemented with 1% chicken serum, 10% fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin at 39°C with 5% CO 2 and were subcultured every 3 days. DT40 -3KO transfection was performed as described before (46). Briefly, 5 million cells were pelleted, washed once with phosphate-buffered saline and electroporated with 5-10 g of DNA using Amaxa cell nucleofector kit T (Lonza Laboratories). Cells were allowed to recover for 24 h and then passaged into media containing 2 mg/ml G418 in 96-well plates. After 10 -14 days, clones expressing the desired constructs were identified by immunoblot analyses.
Preparation of Tissue and Cell Lysates and SDS-PAGE Analyses-Mice were euthanized following the University of Rochester's guidelines, and the indicated tissues were immediately harvested, sonicated, and vortexed vigorously in Triton X-100 lysis buffer (50 mM Tris base, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, pH 8) supplemented with protease inhibitors (Roche Applied Science). Lysates were then incubated for 30 min on ice and pre-cleared by centrifugation at 16,000 ϫ g for 10 min at 4°C, and cleared lysates were collected. DT40 cell lysates were prepared by harvesting cells by centrifugation, and the cell pellet was washed once with ice-cold phosphate-buffered saline (PBS) and solubilized in lysis buffer (100 mM NaCl, 10 mM Tris-HCl, 1% Triton X-100 (v/v), 0.5% sodium deoxycholate (w/v), 10% glycerol, 1 mM EGTA, 1 mM EDTA, 1 mM NaF, 20 mM Na 4 P 2 O 7 , 2 mM Na 3 VO 4 ) containing a mixture of protease inhibitors. After 30 min on ice, cell lysates were centrifuged at 16,000 ϫ g for 10 min at 4°C. Protein concentrations in cleared lysates were determined using D c protein assay kit (Bio-Rad). Proteins were resolved on SDS-polyacrylamide gels, transferred to a nitrocellulose membrane, processed for immunoblotting with the indicated primary antibodies and corresponding secondary antibodies, and subsequently visualized with an Odyssey infrared imaging system (LICOR Biosciences).
Co-immunoprecipitation-Cleared lysates were incubated with the indicated antibodies for 3-5 h followed by the addition of 30 l of protein G slurry for an additional 2 h. Immunocomplexes were then washed three times with ice-cold lysis buffer, resuspended in 2ϫ gel loading buffer, and processed for immunoblots. For immunodepletion experiments, lysates were prepared from mouse pancreas and were immunoprecipitated first with anti-IP 3 R2 followed by anti-IP 3 R1 to sequentially deplete IP 3 R2-and IP 3 R1-containing complexes. Equivalent supernatants, before and after immunodepletion, were fractionated on SDS-PAGE and probed with the indicated antibodies.
Subcellular Fractionation-Cells were harvested and washed once with ice-cold PBS, and cell pellets were resuspended in homogenization buffer (0.32 M sucrose, 20 mM HEPES, 0.5 mM EGTA, 5 mM NaN 3 at pH 7.4) containing protease inhibitor mixture and homogenized using a glass homogenizer with a Teflon pestle. Homogenates were centrifuged at 1000 ϫ g for 10 min at 4°C to remove nuclei and any unbroken cells. Supernatants were then subjected to ultracentrifugation at 100,000 ϫ g for 1 h at 4°C. The supernatants (cytosolic fraction) and pellets (membrane fraction) were collected, and protein concentrations were determined. Equivalent amounts of proteins were resolved on SDS-polyacrylamide gel and processed for immunoblots with the indicated antibodies.
Native Blue PAGE Analysis-DT40 -3KO cells expressing various constructs were harvested by centrifugation and washed two times in ice-cold PBS. Cells were lysed in 100 l of lysis buffer (40 mM NaCl, 25 mM HEPES, 10 mM CHAPS, 1 mM EDTA at pH 7.4) supplemented with protease inhibitors. After 20 min on ice, cell lysates were cleared by centrifugation at 16,000 ϫ g for 10 min at 4°C. Afterward, 75 l of cleared lysates were mixed with 25 l of 4ϫ sample buffer, and 5 l of 5% G-250 sample additive. Samples were then fractionated on 3-12% NativePAGE TM Novex at 150 V for 1 h using dark cathode buffer and then at 250 V for 1.5 h using light cathode buffer. Molecular weight markers are based on the mobility of Native-Mark TM unstained protein standard (Invitrogen), ranging from 1236 kDa (IgM hexamer) to 20 kDa (soybean trypsin inhibitor). Separated proteins were then electroblotted onto polyvinylidene difluoride (PVDF) membrane and probed with either anti-IP 3 R1 or anti-IP 3 R2 antibodies followed by horseradish peroxidase-conjugated secondary antibodies. Protein bands were detected using enhanced chemiluminescent substrate.
Gel Filtration Chromatography-Cells were harvested, washed three times with ice-cold PBS, and lysed in CHAPS lysis buffer (10 mM CHAPS, 25 mM HEPES, 120 mM NaCl, 1 mM EDTA, 1 mM DTT at pH 8) containing protease inhibitors. Lysates were incubated on ice for 30 min and centrifuged at 16,000 ϫ g for 15 min at 4°C. Cleared lysates were transferred into polyallomer Beckman centrifuge tubes and centrifuged at 130,000 ϫ g for 15 min at 4°C. Supernatants were transferred into fresh cold polyallomer centrifuge tubes, and samples were recentrifuged at 130,000 ϫ g for 15 min at 4°C. 50 l of cleared supernatants was removed and mixed with 2ϫ sample buffer as an input control. The remaining cleared supernatants were immediately chromatographed through a HiPrep 16/60 Sephacryl S-400 HR column (GE Healthcare). The eluate (45 ml) was fractionated (1 ml/fraction), and portions of each fraction were separated by SDS-PAGE and processed by immunoblotting.
Cytosolic Ca 2ϩ Measurement in Intact Cells-Single cell Ca 2ϩ imaging was performed as described previously (46). DT40 -3KO cells expressing IP 3 R constructs were washed once with imaging buffer (137 mM NaCl, 0.56 mM MgCl 2 , 4.7 mM KCl, 10 mM HEPES, 5.5 mM glucose, 1.26 mM Ca 2ϩ , 1 mM Na 2 HPO 4 at pH 7.4) and incubated with 2 M Fura2-AM on a glass coverslip for 20 min at room temperature to allow for fluorescent dye loading and attachment of cells to the coverslip forming the base of the chamber. Loaded cells were perfused with imaging buffer and stimulated with agonist at the time points indicated. Image acquisition was performed using TILLVision software. Experiments were repeated at least four times, and data are presented as 340/380 ratios. All ratio changes were determined to be within the linear dynamic range of the dye and are therefore directly related to changes in [Ca 2ϩ ] i .
Patch Clamp Experiments-Single IP 3 R channel potassium currents (I k ) were measured in the on-nucleus configuration of the patch clamp technique using PClamp 9 and an Axopatch 200B amplifier (Molecular Devices, Sunnydale, CA) as described previously in detail (26). Briefly, the pipette solution contained 140 mM KCl, 10 mM HEPES, with the indicated concentrations of IP 3 , ATP, 1,2-bis(2-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid, and free Ca 2ϩ . Free [Ca 2ϩ ] was calculated using Max Chelator freeware and verified fluorometrically. Traces were consecutive 3-s sweeps recorded at Ϫ100 mV, sampled at 20 kHz, and filtered at 5 kHz. A minimum of 15 s of recordings was considered for data analyses. Current/voltage relationships were generated by obtaining multiple sweeps at the indicated holding potential. Pipette resistances were typically 20 megohms and seal resistances were Ͼ5 giga-ohms.
Data Analysis-Single channel openings were detected by half-threshold crossing criteria using the event detection protocol in Clampfit9. We assumed that the number of channels in any particular nucleus was represented by the maximum number of discrete stacked events observed during the experiment. Even at low P o , stacking events were evident (data not shown). Only patches with one apparent channel were considered for analyses. Probability of opening (P o ), unitary current (I k ), open and closed times, and burst analyses were calculated using Clampfit9 and Origin 6 software (Origin Lab, Northampton, MA). All-points current amplitude histograms were generated from the current records and fitted with a normal Gaussian probability distribution function. The coefficient of determination (R 2 ) for every fit was Ͼ0.95. The P o was calculated using the multimodal distribution for the open and closed current levels. Channel dwell time constants () for the open and closed states were determined from exponential fits of all-points histograms of open and closed times. The threshold for an open event was set at 50% of the maximum open current, and events shorter than 0.1 ms were ignored. A "burst" was defined as a period of channel opening following a period of no channel activity, which was greater than five times the mean closed time within a burst. The slope conductances were determined from the linear fits of the current-voltage relationships with Equation 1, where g is unitary conductance; I k is unitary current; V is voltage, and V k is the reversal potential. Two tailed heteroscedastic t tests with p values Ͻ0.05 were considered to have statistical significance.

RESULTS AND DISCUSSION
Differential Expression and the Formation of Homo-and Heteromeric IP 3 R Complexes in Native Tissues-Three IP 3 R genes are essentially ubiquitously expressed, and individual cell types often express different combinations of IP 3 R subtypes (10,31). We therefore initially sought to document IP 3 R subtype expression in native tissues. Fig. 1A shows that all tissues examined express distinct quotas of IP 3 R subtypes. Consistent with previous studies, although IP 3 R1 predominates in the central nervous system, IP 3 R2 and IP 3 R3 are enriched in peripheral tissues, in particular exocrine secretory glands (31,47,48). The functional impact of the co-expression of multiple IP 3 R sub-types in an individual cell may simply reflect redundancy or, alternatively, a means of fine-tuning the Ca 2ϩ signal to accomplish a particular physiological specialization (41,42,49). Added complexity has been suggested by several studies, performed mainly using cultured cell lines, demonstrating that IP 3 R subtypes form both homo-and heterotetrameric channels, although the relative levels of the assembled channel have not generally been established (14,16,18). To verify that heterotetrameric channels are assembled in native tissues, co-immunoprecipitation experiments were performed with lysates from the pancreas, parotid, and spleen. The specificity and efficiency of these antibodies used for co-immunoprecipitation experiments have been established previously (46,50). As shown in Fig. 1B, when immune complexes captured with anti-IP 3 R2 antisera were probed with anti-IP 3 R1, a band corresponding to IP 3 R1 was detected. These data indicate that IP 3 R2 oligomerizes with IP 3 R1 in these tissues. Likewise, when immune complexes captured following incubation with anti-IP 3 R3 antibodies were immunoblotted with anti-IP 3 R1 or -IP 3 R2, bands corresponding to IP 3 R1 and IP 3 R2, respectively, were detected. These results suggest that IP 3 R3 is assembled in complexes with IP 3 R1 and IP 3 R2. Although we cannot formally exclude that inter-molecular complexes of homotetrameric channels exist, these data are consistent with previous crosslinking and immunoprecipitation data reporting the formation of heterotetrameric IP 3 R (51). We also performed immunoprecipitation designed to deplete IP 3 R2 and IP 3 R1 sequentially from pancreatic lysates. Anti-IP 3 R1 and anti-IP 3 R2 antisera efficiently removed the targeted IP 3 R from the lysate (Fig. 1C).
Immunoblotting the depleted lysates revealed that the majority of IP 3 R3 exists in heteromeric complexes, mostly with IP 3 R2, as might be expected in tissue where IP 3 R2 and IP 3 R3 constitute ϳ90% of total IP 3 R (Fig. 1, C and D) (31). Specifically, only 11.3% Ϯ 15 of IP 3 R3 immunoreactivity was not captured with anti-IP 3 R2 or anti-IP 3 R1 antisera, suggesting that only a small IP 3 R3 subpopulation exists as homomeric channels. Thus, both homo-and heterotetrameric IP 3 R channels are assembled in native tissues. Although it is outside the scope of this study, a remaining question is whether all three IP 3 R subtypes exist in an individual tetrameric channel.
Establishing Stable Cell Lines Expressing Homo-or Heterodimer IP 3 R Constructs-Given the expression of a poorly defined complement of tetrameric IP 3 R in native cells, it has been impossible to dissect out the contribution of heteromeric channels to the particular characteristics of intracellular Ca 2ϩ signals in cells expressing multiple IP 3 R subtypes. Therefore, to study how the subunit composition affects the function of the IP 3 R channel activity, we designed a strategy to create heterotetrameric channels with predetermined and thus defined stoichiometry by generating multimers of concatenated monomer IP 3 R subunits. This approach has been used successfully with many other receptors and ion channels, including potassium channels, Orai channels, and glutamate receptors (52)(53)(54). The multimeric constructs are typically expressed from a single open reading frame, which is derived from tandem cDNAs that encode one or more subunits, each connected to the next by a flexible linker. A barrier to applying these general techniques to IP 3 R is the large size of cDNA encoding the IP 3 R monomer FIGURE 1. Differential expression and the formation of homo-and heteromeric IP 3 R complexes in native tissues. A, lysates from the indicated mouse tissues were prepared, and equivalent amounts of proteins were resolved on SDS-polyacrylamide gels, transferred to a nitrocellulose, and processed for immunoblots with the indicated antibodies. Lysates from DT40 -3KO cells (3KO) expressing IP 3 R1 (R1), IP 3 R2 (R2), or IP 3 R3 (R3) were used as controls. B, lysates from the indicated mouse tissues were prepared as in A. IP 3 Rs were immunoprecipitated with subtype-specific antibodies. Immunoprecipitates were processed for immunoblots with the indicated antibodies. C, lysates from mouse pancreas were prepared and subjected to sequential immunoprecipitation first with anti-IP 3 R2 followed by anti-IP 3 R1 to deplete R2 (middle lane) and both R1 and R2 (right lane) as described under "Materials and Methods." Equivalent supernatants, before and after immunodepletion (Id), were fractionated on SDS-PAGE and probed with the indicated antibodies. Shown is a representative experiment. D, histograms comparing R3 immunoreactivities after immunodepletion of R2 (middle column) or R1 and R2 (right column). R3 immunoreactivities were quantified and expressed as percentage of immunoreactivity before immunodepletion. Data are presented as mean Ϯ S.E.
(ϳ8.5 kb of cDNA per monomer, coding for ϳ2700 amino acids). We expected that concatenated IP 3 R tetramers would likely be unstable, cumbersome to manipulate, and doubtless difficult to create with the currently available expression plasmids. Alternatively, we reasoned that concatenated IP 3 R dimers might utilize the biosynthetic machinery of the cells and assemble into tetrameric complexes.
To establish proof of principle, we created homodimeric IP 3 R1 (R1R1), homodimeric IP 3 R2 (R2R2), and two reciprocal heterodimeric constructs based on IP 3 R1 and IP 3 R2, designated R1R2 and R2R1. The two IP 3 R subunits were covalently linked by a 14-amino acid linker and cloned into a propriety plasmid designed for large fragments of DNA ( Fig. 2A). The monomeric and dimeric constructs were stably transfected and expressed in DT40 -3KO. Anti-IP 3 R1 antibody detected a single band migrating at ϳ260 kDa, corresponding to monomeric IP 3 R1 and an ϳ500-kDa band in cells expressing R1R1, R1R2, and R2R1, corresponding to the expected molecular weight of a dimeric IP 3 R (Fig. 2B). As expected, no immunoreactivity was detected with anti-IP 3 R1 antisera in lanes corresponding to 3KO, IP 3 R2, or R2R2 lysates. Similarly, anti-IP 3 R2 antisera detected monomeric IP 3 R2 at ϳ260 kDa and bands migrating at ϳ500 kDa in lysates from cells expressing R1R2 and R2R1 and R2R2 but not in lysates from cells expressing 3KO, IP 3 R1, or R1R1 lanes (Fig. 2C). No evidence of lower molecular weight species or degradation products in cells expressing dimeric constructs was apparent. Despite the potential issues described above, concatemers containing 3 or 4 monomers were generated using similar techniques. Fig. 2D shows that concatenated R1R1R1 trimer and R1R1R1R1 tetramer are clearly expressed, albeit at apparently lower levels than the other constructs. In total, these data show that various combinations of dimeric IP 3 R subunits, together with higher order concatenated structures, can be stably expressed in DT40 -3KO cells.
Membrane Localization and Tetrameric Assembly-As transmembrane proteins, it is reported that IP 3 Rs are cotranslationally targeted first to membranes of the endoplasmic reticulum (ER) where they are either retained as permanent ER resident proteins or are trafficked to other organelles, such as the Golgi apparatus or plasma membrane (55). To determine whether concatenated dimers were properly targeted and retained in membranes, subcellular fractionation was performed using cells expressing IP 3 R1 and R1R1 as a representative of dimeric IP 3 R constructs. Immunoreactivity corresponding to IP 3 R1 and R1R1 was detected in the total cell homogenates and in membrane fractions but not in cytosolic fractions (Fig. 3). Immunoblotting using anti-SERCA antisera, as a marker of ER membrane proteins, and anti-GAPDH antisera, as a cytoplasmic marker, confirmed efficient separation of subcellular fractions. Thus, in similar fashion to IP 3 R monomeric subunits, dimeric IP 3 Rs were indeed targeted and retained in cellular membranes.
We next determined whether dimeric IP 3 R constructs were assembled to form oligomeric structures. Native blue gels have

. Construction of concatenated IP 3 Rs and generation of stable DT40 -3KO cell lines.
A, schematic representation of pJAZZmamm showing its main structural features as follows: left arm contains the plasmid origin of replication, a transcriptional terminator, CMV promoter, and NcoI site; right arm, cut with SwaI to result in a blunt end, contains a transcriptional terminator and bacterial and mammalian antibiotic resistance genes. The inset contains a tandem construct of IP 3 R1 head cDNA digested with NcoI and AgeI and an IP 3 R1 tail cDNA digested with AgeI and HpaI (blunt end). B, DT40 -3KO cells stably expressing various IP 3 R constructs were harvested and lysed, and ϳ60 g of lysate proteins were fractionated on 4% gels and processed in immunoblots with either anti-IP 3 R1 antibody (B) or anti-IP 3 R2 antibody (C). Only 15 g of lysates were loaded for R1 and R2 lanes because R1 and R2 are very much overexpressed. D, cell lysates were prepared from DT40 -3KO cells expressing either R1 or increasingly longer concatenated constructs as follows: R1R1, R1R1R1, or R1R1R1R1. Proteins were fractionated on 3-5% gradient gels, transferred to nitrocellulose, and immunoblotted with anti-IP 3 R1. Less lysate proteins (15 g) were loaded in the R1 lane relative to other lanes (60 g). Asterisks denote nonspecific bands.

Functional Concatenated IP 3 R
OCTOBER 11, 2013 • VOLUME 288 • NUMBER 41 been widely used to assess the size and nature of native protein complexes (56). Anti-IP 3 R1 antisera detected a band migrating at ϳ1100 kDa on native gels with lysates isolated from InsP 3 R1transfected cells (Fig. 4A). This band corresponds to native IP 3 R1 tetrameric channels, as shown previously (46). Similarly, immunoreactivity of an ϳ1100-kDa species was detected with anti-IP 3 R2 antisera and corresponds to the molecular mass predicted to be formed by IP 3 R2 tetrameric channels (Fig. 4B). Importantly, anti-IP 3 R1 antisera detected a band corresponding to tetrameric IP 3 R in lanes containing R1R1, R1R2, and R2R1 but not in lanes with R2R2. Similarly, anti-IP 3 R2 antisera detected a band in lanes with dimeric constructs containing IP 3 R2 (Fig. 4B). No evidence of individual dimeric proteins or high molecular weight aggregation was observed. Thus, dimeric IP 3 Rs, like native monomeric IP 3 R subunits, were efficiently assembled into tetrameric complexes.
To provide further evidence that dimeric constructs assemble into complexes equivalent to oligomerization of native IP 3 R monomers, and also to investigate if higher order molecular weight assemblies were evident, lysates containing IP 3 R1 and R1R1 were fractionated by gel filtration. 45 fractions were collected to cover the entire elution spectrum, and alternative fractions were resolved on an SDS-polyacrylamide gel followed by immunoblotting with anti-IP 3 R1 antisera. Fig. 4C shows that the profile of immunoreactive bands from both of the IP 3 R1and R1R1-expressing cells was essentially indistinguishable, with a distinct peak eluting between 2000 and 670 kDa (Fig. 4C). Little immunoreactivity was evident in the void volume suggesting that higher order constructs larger than a tetramer, which potentially might be formed by inappropriate assembly/ folding of the dimers, do not constitute a prominent species. The very faint protein bands smaller than 670 kDa are likely artifacts as a result of sample preparation and may represent partially assembled channels as shown previously (57). Taken together, these data indicate that intact dimeric IP 3 Rs are synthesized and oligomerized into protein complexes identical to that resulting from expression of monomeric IP 3 R constructs and are efficiently assembled into tetrameric channels. IP 3 -mediated Ca 2ϩ Release-We next sought to determine whether tetrameric channels formed from dimeric IP 3 R constructs were functional in terms of IP 3 -induced Ca 2ϩ release. DT40 -3KO expressing the indicated dimeric IP 3 R constructs were loaded with the Ca 2ϩ -sensitive fluorescent probe Fura2 AM and stimulated with 500 nM trypsin to activate the protease-activated receptor (PAR) leading to IP 3 formation (43). Intracellular [Ca 2ϩ ] i signals were monitored by digital imaging. Cells that responded with a change in fluorescence of Ͼ0.1 were included in the analysis, and over 80% of the cells met this criterion (Fig. 5B). As shown before, DT40 -3KO cells lacking IP 3 R expression did not exhibit [Ca 2ϩ ] i signals in response to PAR activation (Fig. 5A) (44). Remarkably, all dimeric constructs responded robustly to a comparable extent following PAR activation (Fig. 5

, A/C).
Tetrameric channels formed from R1R2 heterodimers must contain equal numbers of IP 3 R1 and IP 3 R2 subunits. Because we cannot assess whether like subunits are situated adjacent to each other or diagonally opposed, or whether this positioning impacts channel activity, we created the reciprocal dimer R2R1. Notably, this dimer gave rise to tetrameric channels with robust IP 3 -induced Ca 2ϩ releasing activity, identical to that of R1R2. Stimulation of cells expressing trimeric constructs gave rise to Ca 2ϩ signals meeting our inclusion criteria in only a very small minority of cells (2% of 244 cells in six experimental runs and 4% of 693 cells in eight experimental runs, representative of 2 separate stable lines; Fig. 5B). These data indicate that trimeric constructs are largely nonfunctional and again suggest that the formation of higher order constructs capable of mediating IP 3induced gating is not favored following expression of concatenated IP 3 R constructs. Strikingly, stimulation of cells expressing R1R1R1R1 also resulted in a marked Ca 2ϩ release (Fig. 5,  A/C). However, the number of responding cells and the magnitude of the Ca 2ϩ signal was reduced when compared with the  Lysates from DT40 -3KO cells stably expressing various IP 3 R constructs were prepared using CHAPS lysis buffer. Lysates were then resolved on 3-12% NativePAGE TM Novex. Proteins were immunoblotted with anti-IP 3 R1 (A) or anti-IP 3 R2 (B). Representative experiments are shown. C, cell lysates were prepared from R1 and R1R1 cells and subjected to ultracentrifugation. Cleared supernatants were injected through a prepacked HiPrep 16/60 Sephacryl S-400 HR gel filtration column. Fractions of eluates were collected (1 ml/sample) and were separated on SDS-PAGE and processed by immunoblotting. IP 3 R1 immunoreactivities were quantified and expressed as percentage of the peak immunoreactivity.
other constructs, perhaps reflecting the relatively low expression level of this protein (Fig. 5B). In turn, the low expression level may be a function of the immense size of the protein encoded, which may provide a challenge to the transcriptional and translational machinery of the cell.
Constructs were also generated to encode an array of other dimeric IP 3 R proteins, including R1R3, R3R1, R2R3, R3R2, and R3R3. When stably expressed in DT40 -3KO cells, these receptors also uniformly supported robust agonist-induced Ca 2ϩ release (data not shown). Taken together, concatenated IP 3 R dimers and tetramers are assembled into functional, tetrameric IP 3 -gated channels.
These results may have some important structural implications. Current models accounting for IP 3 R gating suggest that IP 3 binding alters the relationship between the suppressor domain and the IP 3 binding core in the N terminus, which in turn triggers a conformational change leading to channel opening in the C terminus (20,58,59). Notably, linking the C and N termini of adjacent subunits by the 14-amino acid linker appears not to sterically hinder nor inappropriately constrain this process. Gating may be accomplished in channels formed from IP 3 R dimers because the linker is sufficiently flexible to accommodate this conformational change and is appropriately sized to not cause a major structural disruption in any interaction between N and C termini.
Subunit Composition Influences the Characteristics of Cytosolic Ca 2ϩ Signals-Previous studies have indicated that intracellular Ca 2ϩ fluctuations are generated in a subtype-specific manner (24,41,60). Although the mechanism underlying Ca 2ϩ oscillations is not clearly understood, the general consensus is that Ca 2ϩ oscillations may reflect the dynamic interplay between IP 3 binding, regulation by cytosolic and luminal Ca 2ϩ , as well as the availability of various modulators of IP 3 R (41). Given isoform-specific regulation, the coordinated effect of these factors would obviously be influenced by the complement of IP 3 R subtypes expressed in a particular cell (60). For example, studies using DT40 cells genetically manipulated to express one IP 3 R subtype in isolation have indicated that IP 3 R1 generates largely monophasic Ca 2ϩ peaks with some low amplitude infrequent fluctuations. In contrast, IP 3 R2 supports robust long lasting oscillations (41). Therefore, to investigate how subtype composition of the tetrameric channel contributes to the generation of oscillatory Ca 2ϩ signals, DT40 -3KO cells expressing various IP 3 R constructs were incubated with submaximal concentrations of anti-IgM, which leads to stimulation of B-cell receptors and IP 3 formation (44). Consistent with previous studies, DT40 -3KO expressing mammalian IP 3 R1 exhibited irregular low amplitude Ca 2ϩ responses that generally were not maintained (Fig. 6A) (61). Likewise, cells expressing R1R1  showed low amplitude, infrequent Ca 2ϩ peaks in response to B-cell receptor stimulation (Fig. 6B). However, DT40 -3KO cells expressing IP 3 R2 displayed robust high amplitude and more stable Ca 2ϩ oscillations (Fig. 6, F; quantitation in G/H), corroborating previous reports (24,41). Furthermore, R2R2 supported Ca 2ϩ oscillations very similar to those generated by IP 3 R2, suggesting that channels assembled from dimeric R2R2 form authentic "IP 3 R2 oscillatory units" (Fig. 6, C/H). Notably, expression of R1R2 as well as R2R1 led to vigorous oscillatory patterns qualitatively characteristic of IP 3 R2-mediated behavior (compare C and D with E or F, quantitated in Fig. 6H). In total, these data suggest that the presence of constituent IP 3 R2 subunits may impart dominant regulatory characteristics such that oscillatory behavior is observed.
Single Channel Properties of R1R1 Dimers-Imaging experiments performed on intact cells described above clearly show that IP 3 R channels formed by dimerization of two linked subunits are functional and mediate strong Ca 2ϩ release activities in response to PAR and B-cell receptor activation. To study the properties of concatenated IP 3 R in more detail, we next examined the electrophysiological properties of the channels assembled in cells expressing dimeric R1R1 using the on-nucleus configuration of the patch clamp technique. Initial experiments were performed with a pipette solution containing 200 nM Ca 2ϩ , 5 mM ATP, and either 1 or 10 M IP 3 . The former conditions result in sub-maximal IP 3 R1 activity, whereas the latter leads to the maximum achievable P o of IP 3 R1 expressed in this system (Fig. 7A) (25,26). The activity of channels assembled from monomeric IP 3 R1 or R1R1 dimers was essentially indistinguishable under these conditions in terms of the P o achieved (pooled data in Fig. 7, A and B), single channel conductance (Fig. 7C), and the characteristic gating of the channels to result in "bursts" of activity (Figs. 7A and 9, A and B) (26). These data provide direct evidence that R1R1 dimers form channels that are functionally competent and exhibit core conductance and IP 3 -stimulated gating properties, which are very similar, if not identical, to native channels.
Regulation of Properties of IP 3 R Assembled from Homodimers and Heterodimers-The characteristics of anti-IgM-evoked [Ca 2ϩ ] i signals in cells expressing various heterodimeric IP 3 Rs are consistent with the presence of IP 3 R2 subunits within the channel leading to prominent oscillatory signals (Fig. 6, C and  D). This behavior is a prominent property of homotetrameric IP 3 R2 (41). We therefore next investigated in more mechanistic detail whether the activity of IP 3 R formed from heterodimers reflected a blend of properties of the constituent monomers or, alternatively, whether the characteristics of one subtype dominated the overall pattern of regulation. As a paradigm to investigate this concept, we chose to monitor the regulation of single IP 3 R channel activity in response to adenine nucleotides. This form of regulation is very well defined, and importantly, prominent subtype-specific characteristics have been documented (24 -26, 35 (25,26,35). In addition, the increase in activity occurs by subtype-specific effects on the kinetics of channel gating (26).
We examined the effects of increasing [ATP] levels on the activity of channels formed from homo-and heterodimeric IP 3 R1 and IP 3 R2 and compared this information with previously published data from IP 3 R channels assembled from monomeric constructs. In nuclei prepared from cells expressing R1R1 dimers, increasing [ATP] greatly increased the overall channel P o , by markedly enhancing the time the channel spends . These data are essentially indistinguishable from previous data obtained from monomeric IP 3 R1 (Fig. 8D). In contrast, increasing the [ATP] only augmented the channel activity formed by R2R2 dimers in the presence of low but not high [IP 3 ] in a manner entirely consistent with the effect on homotetrameric IP 3 R2 ( Fig. 8B and pooled data in D). Together, these data again indicate that the channel formed from R1R1 or R2R2 dimers faithfully behaves in a manner representative of its respective native homotetrameric IP 3 R.
Remarkably, ATP regulation of activity of channels formed from R1R2 dimers, reflecting heterotetramers with equal numbers of IP 3 R1 and IP 3 R2 subunits, displayed properties similar to channels formed from either IP 3 R2 monomers or R2R2 dimers. Specifically, ATP only enhanced activity at low [IP 3 ] ( Fig. 8C and pooled data in D). Furthermore, previous reports have established that IP 3 R exhibits a form of "modal gating," such that the agonist-induced increase in Ca 2ϩ flux is accomplished at the single channel level by entering a "bursting mode." Within the bursts, the mean open and closed times of the channel are constant, and, therefore the overall increase in channel P o is achieved solely by the channel spending more time in this bursting state. Notably, increasing the [ATP] enhances bursting activity by distinct and characteristic bio-physical mechanisms in IP 3 R1 and IP 3 R2. Specifically, an increase in [ATP] increases the length of the IP 3 R1 bursts, whereas the number of bursts (of a relatively constant period) are increased for IP 3 R2 (Fig. 8, A and B, compare traces Ϯ ATP at low [IP 3 ]) (26). These particular subtype-specific kinetic characteristics were conserved for R1R1 and R2R2 dimeric channels. Analysis of the bursting activity of R1R1 confirmed that the ATP-induced increase in bursting activity occurred as a function of a reciprocal increase in the time constant () describing the burst duration and a corresponding decrease in the describing the burst interval (Fig. 9, A and B). In contrast, the increase in R2R2 occurs solely by a marked decrease in the describing the burst interval, thus resulting in multiple repetitive bursts of a constant time period (Fig. 9C). Finally, identical effects to only alter the for the inter-burst intervals were observed for R1R2 dimeric channels (Fig. 9E). Thus, the dominant effects of IP 3 R2 subunits were also conserved in the specific gating kinetics evident in R1R2 dimeric channels.
Conclusion-Previously, it has been difficult to assess the contribution of heterotetrameric IP 3 R channels to Ca 2ϩ signaling, primarily because of the lack of a cell system in which the composition of heteromeric channels can be defined and then monitored. Using concatenated dimeric IP 3 R subunits, we have demonstrated that dimeric proteins are assembled to reconstitute functional tetrameric IP 3 R channels. Ca 2ϩ imaging exper- FIGURE 8. Single channel recording of homo-and heteromeric channels. Single channel recordings were conducted as in Fig. 7. A, single channel recordings under the indicated conditions, representing "high" and "low" [ATP] from DT40 -3KO cells stably expressing R1R1 dimeric construct. The modulation of activity by ATP is characteristic of the R1. B, recordings under the indicated conditions from DT40 -3KO cells stably expressing R2R2 dimeric construct. The modulation of activity by ATP is characteristic of R2. C, recordings under the indicated conditions from DT40 -3KO cells stably expressing R1R2 dimeric construct. The modulation of activity by ATP is characteristic of the R2. D, pooled data for the indicated conditions. Data from R1 and R2 are modified from Ref. 26.

Functional Concatenated IP 3 R
OCTOBER 11, 2013 • VOLUME 288 • NUMBER 41 iments conducted on intact cells showed that IP 3 R2, when present in heteromeric channels, imparts IP 3 R2-like cytosolic Ca 2ϩ signals. In addition, the regulation of defined heterotetrameric IP 3 R activity by ATP was largely determined by the presence of IP 3 R2 subunits. These studies suggest strongly that heterotetramers may not simply behave as a blend of constituent subunits, and the expression of heterotetramers may provide cells with an additional level of fine-tuning of IP 3 R-mediated Ca 2ϩ signals.
To our knowledge, the findings presented in this study represent the first demonstration in situ of functional heterotetrameric IP 3 R channels of unequivocally defined composition. The concatenated IP 3 R approach described here can be applied to assess in detail the functional properties of tetrameric IP 3 R channels with defined subunit stoichiometry and predetermined arrangement or positioning. We envision that this approach can be used to address many long standing, important questions regarding IP 3 R regulation by various ligands and modulators. In particular, because the sites of regulation are present in each monomer of the assembled tetrameric channel, the ability to manipulate the amino acid sequence of an individual monomer within the tetramer will facilitate an understanding of the stoichiometry of regulatory input necessary to influence IP 3 R channel activity.