Functional properties of recombinant type I and type III inositol 1, 4,5-trisphosphate receptor isoforms expressed in COS-7 cells.

Inositol 1,4,5-trisphosphate receptors (IP(3)Rs) are ubiquitous intracellular Ca(2+) release channels whose functional characterization by transfection has proved difficult due to the background contribution of endogenous channels. In order to develop a functional assay to measure recombinant channels, we transiently transfected the rat type I IP(3)R into COS-7 cells. Saponin-permeabilized COS cells transfected with type I IP(3)R showed a 50% increase in inositol 1,4,5-trisphosphate (IP(3))-mediated Ca(2+) release at saturating [IP(3)] (10 micrometer) but no enhancement at subsaturating [IP(3)] (300 nm). However, cotransfection of the IP(3)R and human sarco/endoplasmic reticulum ATPase (SERCA)-2b ATPase cDNA resulted in 60 and 110% increases in Ca(2+) release at subsaturating and saturating doses of IP(3), respectively. IP(3) or adenophostin A failed to release (45)Ca(2+) from microsomal vesicles prepared from cells expressing either type I IP(3)R or SERCA cDNAs alone. However, microsomal vesicles prepared from cells doubly transfected with IP(3)R and SERCA cDNAs released 33.0 +/- 0.04% of the A23187-sensitive pool within 30 s of 1 micrometer adenophostin A addition. Similarly, the initial rate of (45)Ca(2+) influx into oxalate-loaded microsomal vesicles was inhibited by IP(3) only when the microsomes were prepared from COS cells doubly transfected with SERCA-2b and IP(3)R DNA. The absence of a functional contribution from endogenous IP(3)Rs has enabled the use of this assay to measure the Ca(2+) sensitivities of IP(3)-mediated (45)Ca(2+) fluxes through recombinant neuronal type I (SII(+)), peripheral type I (SII(-)), and type III IP(3)Rs. All three channels displayed a biphasic dependence upon [Ca(2+)](cyt). Introduction of mutations D2550A and D2550N in the putative pore-forming region of the type I IP(3)R inhibited IP(3)-mediated (45)Ca(2+) fluxes, whereas the conservative substitution D2550E was without effect. This assay therefore provides a useful tool for studying the regulatory properties of individual IP(3)R isoforms as well as for screening pore mutations prior to more detailed electrophysiological analyses.

Inositol 1,4,5-trisphosphate receptors (IP 3 Rs) 1 are large tetrameric Ca 2ϩ channels that are gated by the second messenger IP 3 (1). Full-length cDNAs for three separate isoforms encoded by different genes have been isolated, with each isoform exhibiting unique tissue-specific expression patterns (2)(3)(4)(5). IP 3 R isoforms can assemble as homo-or hetero-oligomers (6 -9). In addition, the type I isoform is alternatively spliced at three separate locations (SI, SII, and SIII), although the functional significance of these splice variants is not fully understood (1, 2, 10 -13). In neural tissues, the SII splice segment is retained, whereas in all nonneuronal cell types this segment is removed (11). It has been shown that alternative splicing at the SII location affects whether the protein is protein kinase A-phosphorylated at residue 1589 (peripheral SII(Ϫ)) or 1755 (neuronal SII(ϩ)) (11). Splicing at the SII location also affects the affinity of recombinant fusion proteins for Ca 2ϩ /calmodulin (14).
The presence of multiple IP 3 R isoforms and splice variants has made it difficult to study the unique properties of individual IP 3 R isoforms. Some cells contain predominantly one isoform, for example the type I (SII(ϩ)) IP 3 R in the cerebellum. The channel properties of this particular isoform have been extensively characterized in studies utilizing the cerebellar IP 3 R incorporated into planar lipid bilayers (15,16). An alternative approach has been to overexpress the recombinant IP 3 R in mammalian cells containing relatively lower densities of endogenous IP 3 Rs, thereby minimizing the chances of endogenous channel incorporation into bilayers when compared with the channels recorded from transfected cells (12,17). Batches of Xenopus laevis oocytes expressing low background IP 3 R channel activity have been employed to study the properties of recombinant III IP 3 Rs in isolated patch-clamped nuclei (18). Nevertheless, in most studies, measurement of recombinant IP 3 R channel Ca 2ϩ release activity in transfected cells by techniques such as 45 Ca 2ϩ flux, Ca 2ϩ imaging using ratiometric dyes, or Ca 2ϩ selective microelectrodes has been hampered by significant contributions from endogenous channels (19 -22). Furthermore, enhanced expression of recombinant IP 3 Rs has not necessarily resulted in large changes in the Ca 2ϩ signal. 2 Mutagenesis studies of other voltage-gated and ligand-gated channels have proved invaluable in determining the mechanisms of gating, permeation, selectivity, and regulation and for predicting channel structure (reviewed in Ref. 23). To carry out such studies in IP 3 Rs it is first necessary to develop a simple functional assay for measuring recombinant IP 3 R channel ac-tivity while minimizing the functional contribution of endogenous channels.
In the present study, we show that this can be achieved by co-expressing the sarco/endoplasmic reticulum ATPase (SERCA) together with recombinant IP 3 Rs in COS-7 cells. Microsomes prepared from cells expressing only IP 3 Rs or SERCA pumps exclusively were not responsive to IP 3 , whereas those co-expressing both proteins were sensitive to IP 3 in a 45 Ca 2ϩ flux assay. This suggests that endogenous IP 3 Rs and SERCA pumps segregate to different vesicle populations during microsome preparation, and overexpressing both recombinant proteins allows co-segregation into the same vesicle pool, enabling Ca 2ϩ accumulated in these stores to be released by IP 3 . Using this assay, we have measured the Ca 2ϩ sensitivities of recombinant type I (SII(ϩ)), type I (SII(Ϫ)), and the type III IP 3 Rs in the absence of a functional contribution from the endogenous channel population. We also show that a point mutation in the putative pore-forming region eliminates channel function.

EXPERIMENTAL PROCEDURES
Materials-Taq DNA polymerase, EXPAND long template DNA polymerase mixture, shrimp alkaline phosphatase, T4 DNA ligase, ATP, dNTPs, and protease inhibitor mixture were from Roche Molecular Biochemicals. Pfu polymerase was purchased from Stratagene (Madison, WI). Oligonucleotides were synthesized by Life Technologies, Inc. TransIT-LT1 cationic lipid transfection reagent was from Pan Vera Corp. (Madison, WI). Protogel stabilized acrylamide solution was from National Diagnostics (Atlanta, GA). Horseradish peroxidase-conjugated donkey anti-rabbit was from Amersham Pharmacia Biotech. Horseradish peroxidase-conjugated goat anti-mouse was purchased from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA). SuperSignal and Ultra SuperSignal chemiluminescent substrate was obtained from Pierce. myo-Inositol 1,4,5-trisphosphate (IP 3 ) was purchased from Calbiochem. Adenophostin A was the kind gift of Kazuhiko Tanzawa (Sankyo Co., Ltd., Tokyo, Japan). Saponin was purchased from Sigma. A23187 was from Molecular Probes, Inc. (Eugene, OR). 45 Ca 2ϩ was purchased from Amersham Pharmacia Biotech. All other chemicals and reagents were purchased from Fisher at the highest quality available.

Expression Constructs
Koz-I/SII(ϩ)-The cDNA encoding the rat IP 3 R type I SI(Ϫ), SIII(ϩ), SII(ϩ) splice variant in pCMV3 was the kind gift of Dr. Thomas Sü dhof (University of Texas Southwestern Medical Center). To ensure high levels of expression, the 5Ј-untranslated region was removed, and a Kozak sequence (24) was engineered immediately preceding the start codon using polymerase chain reaction as described previously (17). Sequences were confirmed by automated dye terminator cycle sequencing (Applied Biosystems model 377, Nucleic Acid Facility, Thomas Jefferson University). This plasmid is referred to as Koz-I/SII(ϩ), and we refer to the protein expressed from this plasmid as type I SII(ϩ).
Koz-I/SII(Ϫ)-The generation of the type I construct lacking the sequence corresponding to amino acids 1693-1732 in pCMV3 has been described elsewhere (14). This plasmid was cut with BamHI, which cleaves the type I cDNA at base pairs 4126 and 6158, which flank the SII splice region. The 1912-base pair cassette was then ligated into Koz-I/SII(ϩ), which had been cut with BamHI, and the 12.6 kilobase fragment was purified away from the 2032-base pair fragment. All cloning junctions and the absence of the SII region were confirmed by sequencing. This plasmid is referred to as Koz-I/SII(Ϫ), and we refer to the protein expressed from this plasmid as type I SII(Ϫ).
Koz-III-The cDNA for the entire coding sequence of the rat type III IP 3 R isoform in pCB6ϩ was the kind gift of Dr. Graeme Bell (University of Chicago). The strategy employed is analogous to that used to generate Koz-I/SII(ϩ). A 7082-base pair fragment was excised from the type III cDNA using AflII and NotI. This was subcloned into the AflII/NotI sites in the multiple cloning site of pCDNA3.1. The 5Ј 1574 bases of the coding sequence were amplified with two primers, allowing removal of the 5Ј-untranslated region and insertion of a Kozak sequence. Primer K3-F (5Ј-AATCTTAAGGCCACCATGAATGAAATGTCCAGCTTT-3Ј) bound to bases 1-21 of the translated sequence and an AflII site (underlined) and a Kozak sequence (boldface type) were engineered 5Ј of the coding sequence. Primer K3-R (5Ј-AAAGATCTGCTTAAGTATGTT-3Ј) anneals to bases 1564 -1584 in the cDNA and encompassed the region flanking the AflII site in the coding sequence. The 1574-base pair polymerase chain reaction product was cleaved by AflII and subcloned into the plasmid containing the 7082-base pair 3Ј IP 3 R-III fragment. The resultant plasmid was sequenced over the entire amplified portion and is referred to as Koz-III.
D2550E,D2550N,D2550A-Aspartic acid 2550 (rat) was mutated to glutamic acid, aspartate, and alanine using the QuickChange point mutation kit (Stratagene, La Jolla, CA). Briefly, a cassette encompassing base pairs 7001-9466 in the C-terminal portion of the receptor was excised from Koz-I/SII(ϩ) using BstBI and XbaI and subcloned into pBLUESCRIPT (Stratagene). Forward primers were designed following manufacturer's recommendations as follows: D2550EF, 5Ј-GGCGGAG-TAGGAGAGGTGCTCAGGAAG-3Ј; D2550NF, 5Ј-GGCGGAGTAGGA-AATGTGCTCAGGAAG-3Ј; D2550AF, 5Ј-GGCGGAGTAGGAGCTGTG-CTCAGGAAG-3Ј (codon changes are shown in boldface type). Reverse primers were the complementary sequence of the forward primers. Polymerase chain reaction cycling conditions were performed as per the manufacturer's instructions, and following the polymerase chain reactions, parental template was digested using DpnI. 1 to 5 l of the resulting mix was transformed directly into Escherichia coli. strain DH5␣, and positive colonies were screened by BstEII/XbaI digestion prior to automated sequencing to confirm mutations. The cassette was then excised and subcloned back into Koz-I/SII(ϩ) and reconfirmed by automated sequencing.
Human SERCA-2b-The cDNA encoding the human isoform of the SERCA-2b Ca 2ϩ -ATPase in pcDNA 3.1 was the kind gift of Dr. Jonathan Lytton (University of Calgary, Alberta, Canada) and Dr. David H. MacLennan (University of Toronto, Ontario, Canada).

Cell Culture and Transfection
COS-7 SV40-transformed African Green monkey kidney fibroblasts (ATCC CRL 1651) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Life Technologies, Inc.) in a 37°C humidified incubator in an atmosphere of 5% CO 2 . DNA for transfection was purified on a CsCl gradient as described elsewhere (25). Cells were plated at a density of 1.5 ϫ 10 6 cells/75-cm 2 flask and transfected in serum-free medium as described previously (14). Cells were exposed to DNA-lipid complexes for 5 h before replacing the medium with Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Cells were processed after 48 h as described below. In all cases, the total amount of each expression construct DNA was 14 g/flask. In those instances where one only cDNA was being transfected, pcDNA 3.1 vector DNA was added to adjust the total DNA concentration to 28 g.

SDS-Polyacrylamide Gel Electrophoresis and Immunoblotting
Cells were solubilized in 150 mM NaCl, 50 mM Tris-HCl, pH 7.8, 1% (w/v) Triton X-100, 1 mM EDTA, 0.5 mM phenylmethanesulfonyl fluoride and 1ϫ Complete protease inhibitor mixture (Roche Molecular Biochemicals) for 5 min on ice and spun at 12,000 ϫ g, and 20 g of the resultant supernatant was run out on a 5% SDS-polyacrylamide gel and subsequently electrotransferred to nitrocellulose membranes (Bio-Rad). Nitrocellulose sheets were then probed with the antibodies listed below and developed with chemiluminescent substrates (Pierce). In those cases where a single blot was probed sequentially with more than one antibody, the nitrocellulose was stripped at 60°C for 30 min in stripping buffer (2% SDS, 100 mM ␤-mercaptoethanol, 62.5 mM Tris-HCl) before probing with the next antibody.

Antibodies
Type I IP 3 R-specific polyclonal antibody raised against amino acids 2731-2749 has been described previously (26). Type II isoform-specific polyclonal antibody was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Type III isoform-specific monoclonal antibody was purchased from Transduction Laboratories (Lexington, KY). Monoclonal antibody to SERCA-2 ATPase (2a, 2b, and species-cross-reactive) was purchased from Affinity Bioreagents (Golden, CO). 37°C with 200 g of protein using a Ca 2ϩ -selective minielectrode to measure changes in [Ca 2ϩ ] bath . Each experiment was initiated by adding 1 mM MgATP followed by 50 g/ml of saponin. Filling of intracellular Ca 2ϩ stores was allowed to proceed until steady state was attained (15 min). Cells were then challenged with varying doses of IP 3 . At the end of each experiment, a calibrating dose of 0.4 nmol of Ca 2ϩ was added.

Preparation of Microsomal Vesicles
75-cm 2 flasks were transfected as described above and washed with phosphate-buffered saline. Cells were released from the flask by incubation with 7 ml of PBE (5 mM EDTA, 0.5% bovine serum albumin in phosphate-buffered saline) for 10 min at room temperature. Cells were washed again with phosphate-buffered saline and swollen in hypotonic solution (10 mM Tris, pH 7.5, 0.05 mM MgCl 2 ) for 10 min on ice. Phenylmethanesulfonyl fluoride (0.5 mM) was added, and the cells were homogenized by 30 strokes in a tightly fitting Dounce homogenizer. Cell disruption was confirmed by trypan blue exclusion. Cells were then brought to isotonicity by dilution in 0.5 M sucrose, 6 mM ␤-mercaptoethanol, 40 M CaCl 2 , 300 mM KCl, 10 mM Tris, pH 7.5. Homogenates were spun at 5000 ϫ g to remove debris, and the supernatant was brought to 0.6 M KCl. The supernatant was spun at 100,000 ϫ g, and the pellet was brought up in 0.25 M sucrose, 0.15 M KCl, 3 mM ␤-mercaptoethanol, 10 mM Tris, pH 7.5 and either used fresh or stored at Ϫ80°C in 20% glycerol until use.

Ca 2ϩ Uptake Measurements
Microsomes were incubated in uptake buffer (release buffer supplemented with 5 mM potassium oxalate, pH 7.2) at 30°C, and 2-g aliquots were removed at various time points and processed as described above. In those experiments in which response to 1 M IP 3 was being measured, IP 3 was added just prior to the microsome addition, and triplicate samples were removed at 5, 10, and 15 min when uptake was still linear (see Fig. 6). In those cases where the sensitivity of release to [Ca 2ϩ ] free was being measured, buffers were calibrated on a Ca 2ϩ minielectrode to the desired [Ca 2ϩ ] free .

Data Analysis
All graphing and statistical calculations were performed using Sig-maPlot (Jandel Corp.). In the 45 Ca 2ϩ uptake experiments, the linear regression of the time points through 5, 10, and 15 min was used to calculate the initial rate of uptake. The initial rate of uptake in the presence of IP 3 (r IP3 ) was expressed as a percentage of the control rate (r). In the Ca 2ϩ sensitivity experiments, the effect of IP 3 at each Ca 2ϩ concentration was expressed as 1 Ϫ (r IP3 /r). Ca 2ϩ sensitivity curves were fit as described previously (27).

Expression of Recombinant IP 3 R Isoforms in COS-7 Cells-
To ensure high levels of IP 3 R expression, each construct had its 5Ј-untranslated sequence removed and a Kozak sequence inserted 5Ј of the start codon (17). The levels of expressed type I SII(Ϫ) and SII(ϩ) isoforms were equivalent to endogenous levels of expression found in cerebellum (Fig. 1, compare lanes 3-7  with lane 9). COS cells contain endogenous type II and type III IP 3 Rs but no detectable type I IP 3 R (5). Importantly, overexpression of recombinant IP 3 R isoforms was not associated with up-regulation of endogenous type II or type III IP 3 R proteins (for example, compare lane 3 in the first three panels). Our previous studies demonstrated that recombinant IP 3 R chan-nels expressed in COS-7 cells form tetramers, bind IP 3 , localize to the endoplasmic reticulum, and do not form heterotetramers with the endogenous receptor population (28).
Ca 2ϩ Release in Saponin-permeabilized Cells-To measure the functional consequences of IP 3 R overexpression in COS-7 cells, we saponin-permeabilized cells transfected with various constructs and allowed the intracellular Ca 2ϩ stores to fill to steady state by the addition of MgATP. We then measured the Ca 2ϩ release responses to 150 nM, 300 nM, and 10 M IP 3 by measuring changes in bath [Ca 2ϩ ] using a Ca 2ϩ -selective minielectrode. Representative traces of the responses to IP 3 are shown in Fig. 2 (the uptake portion has been omitted for clarity). COS-7 cells overexpressing type I SII(ϩ) channels showed no enhancement of Ca 2ϩ release at subsaturating doses of IP 3 , but a 50% increase in Ca 2ϩ release was observed in response to a saturating (10 M) dose of IP 3 (1.34 Ϯ 0.09 nmol/mg protein released in control versus 2.0 Ϯ 0.1 nmol/mg protein released in IP 3 R-transfected, p Ͻ 0.001, Fig. 3). Since these cells were expressing large amounts of IP 3 R protein (Fig. 1), we hypothesized that the rather modest enhancement of Ca 2ϩ release indicated that either the channels were not folded properly or that a co-factor was necessary for proper functional expression. In an attempt to achieve a more robust enhancement of Ca 2ϩ release in IP 3 R-transfected cells, we co-transfected IP 3 R cDNA in combination with cDNAs encoding several other proteins. Co-transfecting the molecular chaperones calnexin and calreticulin, which associate with nascent IP 3 Rs (29), was without any functional consequences (data not shown). The immunophillin FKBP12 has been shown to associate with both ryanodine receptors and IP 3 Rs (30,31). Overexpression of FKBP12 was also without effect in our assay system (data not shown). We next tested the hypothesis that the recombinant IP 3 R channels may localize to intracellular compartments, which lacked the ability to accumulate Ca 2ϩ . Co-transfection of the human 2b isoform of the sarco/endoplasmic reticulum ATPase (SERCA-2b) sensitized the cells to a subsaturating dose of IP 3 (0.76 Ϯ 0.04 nmol/mg protein in co-transfected . Lanes 2-7 were co-transfected with human SERCA-2b ATPase. WB is a rat liver epithelial cell line used as a control for type III expression, and CER is cerebellum prepared from rat. After a 48-h transfection, cells were lysed in a Triton X-100containing buffer, and 20 g of protein was run on 5% SDS-polyacrylamide gel electrophoresis. Blots were sequentially probed with isoformspecific IP 3 R antibodies and SERCA antibody as described under "Experimental Procedures." Due to the very high concentration of antibody necessary to detect type II IP 3 Rs, we observed nonspecific crossreactivity with the type III isoform (lane 2, type II blot), although they could be distinguished by the higher mobility of the type II isoform (indicated by an arrow). versus 0.48 Ϯ 0.06 nmol/mg protein in SERCA-2b alone at 300 nM IP 3 ; see Fig. 3). The size of the total releasable pool measured with 10 M IP 3 nearly doubled from 1.62 Ϯ 0.05 nmol/mg protein in SERCA-2b-expressing cells to 2.84 Ϯ 0.06 in cells expressing both type I SII(ϩ) and SERCA-2b (Fig. 3). IP 3induced Ca 2ϩ release from cells expressing SERCA-2b alone was not significantly different from that observed in cells transfected with vector alone (Figs. 2 and 3). Enhanced Ca 2ϩ release was probably not due to a direct physical association between the IP 3 R and SERCA-2b as determined by co-precipitation assays (data not shown). Furthermore, SERCA-2b expression did not increase the number of functional IP 3 binding sites (Fig. 9B). 45 Ca 2ϩ Release from Microsomal Vesicles-Microsomal vesicles prepared from many cell types can accumulate 45 Ca 2ϩ in the presence of MgATP but are unresponsive to IP 3 unless GTP and polyethylene glycol are added (32)(33)(34). It is thought that IP 3 Rs and SERCA pumps segregate to different vesicle populations during microsome preparation, and the addition of GTP and polyethylene glycol promotes fusion of the vesicles restoring the sensitivity to IP 3 . We hypothesized that overexpression of IP 3 Rs and SERCA pumps in combination would result in co-segregation of both proteins during vesicle preparation, resulting in vesicles responsive to IP 3 . To test this hypothesis, the time course of 45 Ca 2ϩ uptake by microsomal vesicles was de-termined using preparations in which SERCA-2b was expressed alone or in conjunction with the type I SII(ϩ) IP 3 R (cells not expressing SERCA-2b did not accumulate significant levels of 45 Ca 2ϩ in our assay system; data not shown). 45 Ca 2ϩ accumulation achieved equilibrium by 40 min, and the addition at 60 min of the potent nonhydrolyzable IP 3 R agonist AdA (1 M) elicited a 33.0 Ϯ 0.04% reduction in the size of the A23187releasable pool within 30 s (Fig. 4). No reduction in the size of the A23187-releasable pool was observed upon the AdA addition to microsomes expressing SERCA-2b alone (n ϭ 10). Similar results were obtained when 10 M IP 3 was used instead of AdA (data not shown). 45 Ca 2ϩ Uptake in the Presence of Potassium Oxalate-The addition of potassium oxalate (KOx) when measuring SERCA Ca 2ϩ uptake activity allows an almost 50-fold increase in 45 Ca 2ϩ uptake, due to the fact that KOx acts as an intravesicular Ca 2ϩ sink (compare Figs. 4 and 5). SERCA-dependent uptake in the presence of KOx is linear during the first 15 min, and the slope of this line is a direct measure of the initial rate of SERCA-dependent 45 Ca 2ϩ uptake (Fig. 6) (35). The addition of IP 3 reliably reduced the initial rate of 45 Ca 2ϩ uptake, which provided an indirect measure of IP 3 R-mediated Ca 2ϩ release (36,37). The IP 3 -induced decrease in SERCA activity was fully reversible by the addition of the IP 3 R antagonist heparin, further suggesting that the effects of IP 3 are due to the gating of IP 3 R channels (Fig. 6). 45 Ca 2ϩ uptake in microsomal vesicles prepared from cells transfected with SERCA alone or in combination with IP 3 R cDNAs demonstrated that there was no IP 3dependent reduction in the initial rate of uptake unless both recombinant IP3R and SERCA pumps were present (Fig. 7). 3 Rs is a biphasic dependence of channel activity on cytoplasmic Ca 2ϩ concentration (38,39). We used the flux assay shown in Fig. 6 to measure the Ca 2ϩ sensitivities of the two type I splice variants and the type III receptor isoforms. Flux assays were performed in oxalate-containing buffers, which had been calibrated to the indicated [Ca 2ϩ ] free as described under "Experimental Procedures" in the presence or absence of 1.0 M IP 3 . All three isoforms demonstrated a biphasic dependence upon [Ca 2ϩ ] free (Fig. 8). The Ca 2ϩ dependence of the recombinant type I SII(ϩ) channel is similar to that previously reported for the endogenous channels of cerebellum (39). As summarized in Table I, calculated values of half-maximal activation by Ca 2ϩ (K act ) and inhibition (K inh ) were 0.014 and 1.4 M [Ca 2ϩ ] free , respectively, for the type I SII(ϩ) splice variant. The Hill coefficient for activation was 1, whereas for inactivation it was 2.5. The Ca 2ϩ sensitivity of the SII(Ϫ) variant was shifted to the left with respect to the SII(ϩ) channel (Fig. 8, Table I). K act and K inh for the SII(Ϫ) splice variant were 0.0025 and 0.27 M, respectively. The activation curve for the SII(Ϫ) receptor showed greater cooperativity with an H act value of 1.8 providing a good fit to the data (Table I). Conversely, Ca 2ϩ dependent inhibition was less cooperative, with an H inh of 1.0 ( Table I) receptor was similar to that of the type I SII(ϩ) receptor, with a K act of 0.033 and a K inh of 1.0 M. The Hill coefficient for activation was also similar to the SII(ϩ) isoform, while exhibiting less cooperativity in Ca 2ϩ -dependent inactivation (H act ϭ 1.1, H inh ϭ 1.8; Table I).

Ca 2ϩ Sensitivities of Recombinant Type I SII(Ϫ), Type I SII(ϩ), and Type III IP 3 R Isoforms-A fundamental regulatory property of IP
Effect of Mutations at Aspartic Acid Residue 2550 -A series of experiments were undertaken to determine the feasibility of utilizing this assay system to screen for the Ca 2ϩ release activity of expressed IP 3 Rs with engineered point mutations. We hypothesized that the aspartic acid residue at position 2550 may be important for Ca 2ϩ permeation through the channel. This residue is adjacent to a sequence in the IP 3 R that is analogous to the selectivity filter in potassium channels (underlined in Fig. 9) (40). We made three point mutations in which the aspartic acid was mutated to alanine (D2550A), aparagine (D2550N), and glutamic acid (D2550E). These mutations did not affect the ability of the channel to express at high levels ( Fig. 1) or form oligomers, as determined by their ability to co-precipitate with an epitope-tagged type III IP 3 R (data not shown). Furthermore, all three mutants bound IP 3 at levels comparable with wild type (Fig. 9B). Mutating aspartic acid to either asparagine or alanine eliminated responsiveness of 45 Ca 2ϩ fluxes to IP 3 (Fig. 9A). However, the conservative mutation to glutamate preserved IP 3 -induced Ca 2ϩ release activity (Fig. 9A). Taken together, these data suggest that a negatively charged residue at position 2550 is necessary for Ca 2ϩ ion permeation through the IP 3 R channel.  (Control, circles), type I SII(ϩ) (Type I, squares), or SERCA 2b (SERCA, diamonds) or co-transfected with type I SII(ϩ) and SERCA 2b (Type I ϩ SERCA, triangles). Uptake was initiated by adding microsomes to release buffer containing 5 mM KOx. 2-g samples were removed every 20 min in triplicate. In most cases, error bars are smaller than the symbols. Uptake in SERCA-expressing microsomes was linear over the first 15 min (see Fig. 6).
FIG. 6. Effect of IP 3 on the initial rate of SERCA uptake in co-transfected microsomes. Microsomes prepared from cells coexpressing type I SII(ϩ) IP 3 R and SERCA were assayed for SERCA activity in the presence of no IP 3 , 1 M IP 3 , 1 M IP 3 plus 100 g/ml heparin, or 1 M A23187. All additions were made to the uptake buffer containing KOx before microsome addition. Time points (in triplicate) were taken every 5 min during the linear portion of uptake. The initial rate of SERCA uptake was calculated from the linear regression through the data points (not shown). Error bars are smaller than the symbols. 45 Ca 2؉ uptake in microsomes that are not expressing recombinant IP 3 R isoforms. SERCA activity was measured in the presence or absence of 1 M IP 3 in microsomes containing recombinant SERCA either by itself or in combination with the type I SII(ϩ) or type III IP 3 R. Uptake was initiated by microsome addition to release buffer supplemented with 5 mM KOx, and 2-g samples were removed in triplicate at 10 min. *, p Ͻ 0.001.

DISCUSSION
In the present study, we have described a system for measurements of the Ca 2ϩ release activity of recombinant IP 3 R channels in the absence of a contribution from the endogenous channel population, which can be used for determining structure-function relationships of IP 3 R channels. A key feature of this assay system is the necessity to transfect SERCA pumps together with IP 3 Rs in order to observe IP 3 responses in microsomal vesicles. We interpret this result to indicate that under these circumstances the recombinant IP 3 Rs and SERCA pumps co-segregate into the same compartments. An implication of this result is that recombinant and endogenous IP 3 Rs may reside in different subcellular compartments of the endoplasmic reticulum. Although we have not directly addressed the subcellular localization of IP 3 Rs and SERCA pumps, our studies lend support to the view that IP 3 Rs are not uniformly distributed in intracellular membranes (reviewed in Ref. 41).
Various strategies have been employed to reduce the contribution of endogenous channels in functional studies of recombinant IP 3 R channel activity. In one approach, IP 3 R channels were overexpressed in cell lines that have a low density of endogenous IP 3 R channels. Microsomes or proteoliposomes enriched in IP 3 Rs were prepared from these cells and incorporated into planar lipid bilayers (12,17). An alternative nuclear patch clamp approach employed IP 3 R mRNA injection into Xenopus oocytes in which low levels of endogenous IP 3 R channel activity were detected (18). The advantage of the assay system described in this study is that it is relatively rapid and simple and allows for the measurement of recombinant IP 3 R Ca 2ϩ release activity in native endoplasmic reticulum mem-branes. In contrast to single-channel analyses, it measures global Ca 2ϩ flux, which is important when considering how a population of channels responds to stimulation with IP 3 . A potential disadvantage of this approach is that it does not permit detailed biophysical analyses of channel properties. In all expression systems used to date, a possible limitation is the likelihood that transfected receptors may form heterotetramers with the endogenous channels, which could complicate interpretation of the data. However, we have shown that recombinant IP 3 R channels do not associate significantly with the endogenous IP 3 R population in the COS cells used in the present study (28). We therefore consider the methodology described in the this study well suited for investigating the regulatory properties of recombinant IP 3 Rs, as well as for preliminary screening of the functionality of IP 3 R mutations prior to more laborious electrophysiological analyses. FIG. 8. Ca 2؉ dependence of type I SII(؉), type I SII(؊), and type III IP 3 Rs. Uptake in microsomes prepared from cells expressing type I (SII(ϩ)), type I (SII(Ϫ)), or type III IP 3 R in conjunction with SERCA was assayed exactly as described in Fig. 6. The effect of IP 3 on inhibiting flux at each [Ca 2ϩ ] was expressed as 1 Ϫ (r IP3 /r) and is plotted on the ordinate. Assays were performed at Ca 2ϩ concentrations of 1.9 nM, 6.5 nM, 55 nM, 200 nM, 1.0 M, and 3.0 M. The data were fit to a biphasic equation assuming independent Ca 2ϩ -activating and -inhibiting sites (27). Parameters for fitting the equation are listed in Table I. SERCA-2b activity was monophasically dependent upon [Ca 2ϩ ] free with a K act of 0.06 M and a Hill coefficient of 2 (data not shown).

FIG. 9. Effect of amino acid substitutions at aspartic acid 2550.
Above A is shown the aligned sequences of the three IP 3 R isoforms and type I RyR flanking amino acid 2550 (in boldface type) of the rat type I IP 3 R sequence. Amino acids 2545-2549 (rat type I) are similar to the selectivity filter TVGYGD of potassium-selective voltage-gated channels (solid bar). A shows the effect on 45 Ca 2ϩ flux when aspartic acid 2550 is substituted for glutamic acid (D2550E), asparagine (D2550N), and alanine (D2550A). The inhibition of the rate of 45 Ca 2ϩ uptake in the presence of 1 M IP 3 is represented as a percentage of the control rate of uptake. *, not significantly different from control uptake (p Ͼ 0.1). B shows the IP 3 binding capacity of the recombinant type I SII(ϩ) IP 3 R (in the presence and absence of SERCA-2b), D2550E, D2550N, and D2550A. There was negligible binding in vector-transfected cells (Mock). IP 3 binding to cell lysates was done exactly as described previously (57). Rat IP 3 R-I, IP 3 R-II, and IP 3 R-III and rabbit RyR-I sequences were obtained from GenBank TM accession numbers J05510, X61677, L06096, and X15209, respectively. A fundamental regulatory property of IP 3 R channels is their biphasic dependence upon cytoplasmic Ca 2ϩ concentration (38,39). This property is presumed to be one of the underlying mechanisms by which IP 3 R channels are capable of generating and propagating Ca 2ϩ waves (42). We show here that the recombinant type I SII(ϩ), type I SII(Ϫ), and type III IP 3 Rs are all biphasically regulated by [Ca 2ϩ ] cyt . The type I SII(ϩ) receptor displays a Ca 2ϩ sensitivity similar to that found previously (39). The behavior of the type I SII(Ϫ) splice variant was analyzed for the first time in the present study. The type I SII(Ϫ) splice variant was activated by lower Ca 2ϩ concentrations and with greater cooperativity when compared with type III or type I SII(ϩ). This splice variant was also more sensitive to Ca 2ϩ -dependent inhibition. Therefore, alternative splicing of the type I transcript appears to have significant effects on channel regulation by Ca 2ϩ . It has been proposed that deletion of the SII region creates an additional binding site for calmodulin (14). It is therefore possible that an additional calmodulin binding site is responsible for shifting the Ca 2ϩ sensitivity to the left. Further characterization of the effects of calmodulin on IP 3 Rs is necessary to validate these conclusions.
When compared with type I SII(ϩ), the type III isoform was least sensitive to Ca 2ϩ activation and more sensitive to Ca 2ϩ dependent inhibition (Table I). Recent nuclear patch clamp studies have also shown a biphasic Ca 2ϩ dependence of recombinant type III IP 3 R channel in Xenopus oocytes (18). These observations are in contrast to other reports, which have demonstrated that both type II and type III IP 3 R channels are not biphasically regulated by Ca 2ϩ when measured in planar lipid bilayers (43,44). These studies utilized tissues that are specifically enriched in either the type II receptor (ventricular cardiac myocytes) or type III receptors (RIN-m5F cell line) (43,44). Hagar et al. (44) suggested that those cells that have predominately type II or type III IP 3 Rs would not support Ca 2ϩ oscillations due to a lack of Ca 2ϩ -dependent inhibition. However, studies using targeted IP 3 R knockouts in DT40 cells have shown that those cells that are expressing only type II IP 3 Rs can support oscillations (45). Furthermore, Swatton et al. (46) showed that IP 3 -mediated Ca 2ϩ release in permeabilized RIN-m5F cells is biphasically regulated by Ca 2ϩ . Therefore, the lack of Ca 2ϩ -dependent inhibition of the type II and type III receptor observed in some studies may have resulted from the loss of accessory proteins after incorporation into planar lipid bilayers, as has been demonstrated with the type I receptor from cerebellum (47). Swatton et al. (46) have suggested that Ca 2ϩdependent inhibition of type III IP 3 Rs in RIN-m5F cells might result from hetero-oligomer formation with the low levels of endogenous type I receptor. We consider this an unlikely explanation to account for the Ca 2ϩ sensitivity of recombinant type III IP 3 Rs in our system, since we have no evidence for the occurrence of hetero-oligomerization under our experimental conditions (28). Hetero-oligomerization was similarly ruled out in studies of the type III channel in nuclear patch clamp studies (18). Hence, we conclude that a general property of the type III IP 3 R is a biphasic dependence upon cytoplasmic Ca 2ϩ concentration.
The success of our approach for studying the functionality of recombinant IP 3 R isoforms suggested that it will be useful for future structure-function studies of mutant IP 3 R channels. We explored this possibility in initial experiments to determine which residues are important for IP 3 R channel permeation. The transmembrane topology of IP 3 R channels is similar to voltage-gated potassium channels with six transmembranespanning segments and a pore-forming region between segments 5 and 6 (48 -50). The crystal structure of a potassium channel from Streptomyces lividans (KcsA) has highlighted the importance of the highly conserved sequence GYG within the pore-forming region (40). The carbonyl oxygens of these three amino acids function as a selectivity filter by precisely coordinating potassium ions (40). The IP 3 R is not as ion-selective as K ϩ channels (51,52), and it is therefore likely that there will be differences between the selectivity filters of the two channels. However, this region of IP 3 Rs, RyRs, and K ϩ channels does show a high degree of homology. In particular, the aspartate residue immediately following the GYG sequence in potassium channels is highly conserved, being GVGD in IP 3 Rs and GIGD in RyRs (Fig. 9). Based on the crystal structure of KcsA, the aspartate 2550 in the IP 3 R would be positioned in the luminal compartment of the endoplasmic reticulum at the entrance to the selectivity filter. This would have the effect of concentrating Ca 2ϩ ions at the mouth of the channel. The present study shows that mutation of aspartate 2550 to uncharged asparagine or alanine inhibits IP 3 -stimulated 45 Ca 2ϩ fluxes. However, preserving a negative charge at this position by replacing aspartate with glutamate retained channel function. We conclude that a negative charge at position 2550 is required for channel function. This is a result that is consistent with analogous mutations in Shaker (53,54) and Kv2.1 (55) potassium channels, where a negative charge is absolutely required at this position. Furthermore, this residue has been shown to be important for ion selectivity in potassium channels (55) and has been proposed to increase K ϩ ion occupancy in the channel pore (54). Nevertheless, the corresponding mutation in cardiac RyRs (D4829A) failed to block caffeine-mediated Ca 2ϩ release, although ryanodine binding was inhibited (56). Although future mutagenesis studies will be required to clarify these differences and to investigate the mechanism of ion permeation by IP 3 R channels, the results described in the present study demonstrate the utility of this system in furthering such efforts.