Using Concatenated Subunits to Investigate the Functional Consequences of Heterotetrameric Inositol 1,4,5-Trisphosphate Receptors

Abstract

Inositol 1,4,5-trisphosphate receptors (IP₃R) are a family of ubiquitous, ER localized, tetrameric Ca²⁺ release channels. There are 3 subtypes of the IP₃Rs (R1, R2, R3), encoded by 3 distinct genes, that share ~60-70% sequence identity. The diversity of Ca²⁺ signals generated by IP₃Rs is thought to be largely the result of differential tissue expression, intracellular localization and subtype-specific regulation of the three subtypes by various cellular factors, most significantly InsP₃, Ca²⁺ and ATP. However, largely unexplored is the notion of additional signal diversity arising from the assembly of both homo and heterotetrameric InsP₃Rs. In the present article, we review the biochemical and functional evidence supporting the existence of homo and heterotetrameric populations of InsP₃Rs. In addition, we consider a strategy that utilizes genetically concatenated InsP₃Rs to study the functional characteristics of heterotetramers with unequivocally defined composition. This approach reveals that the overall properties of IP₃R are not necessarily simply a blend of the constituent monomers but that specific subtypes appear to dominate the overall characteristics of the tetramer. It is envisioned that the ability to generate tetramers with defined wild type and mutant subunits will be useful in probing fundamental questions relating to IP₃R structure and function.

Introduction

Dynamic changes in intracellular Ca²⁺ control a vast array of cellular processes, including muscle contraction, secretion of fluid and protein, gene transcription, metabolism and cell fate [1-5]. Although multiple Ca²⁺ dependent processes often operate simultaneously within the same cell, the activation of a specific process is accomplished with precision and fidelity, such that individual cellular events can be controlled appropriately to meet the cell’s need. It is widely believed that this precise control is the result of intricate control over the spatial and temporal characteristics of the Ca²⁺ signal. This, in turn, occurs because of the localization, abundance and regulation of the Ca²⁺ handling machinery. These molecules, responsible for both the increase in Ca²⁺ (release and influx mechanisms) and terminating the signal (pumps and transporters and buffers) have been termed the “Ca²⁺ signaling toolkit” by Michael Berridge and colleagues [1]. A fundamental constituent of the “toolkit” responsible for shaping the temporal and spatial characteristics of Ca²⁺ release is the inositol 1,4,5-trisphosphate receptor (IP₃R).

Architecture and function of IP₃R

In non-excitable cells, stimulation with hormones, neurotransmitters and growth factors result in the production of inositol 1,4,5-trisphosphate (IP₃), activation of endoplasmic reticulum (ER) localized IP₃R and the release of intracellular stored Ca²⁺ [1,6,7]. IP₃Rs are encoded by 3 distinct genes (ITPR1, ITPR2, ITPR3) leading to the generation of ~300 kDa monomeric proteins (R1, R2, R3 and splice variants). The IP3R monomers oligomerize co-translationally to assemble into ~1100 kDa tetrameric channels [8-11]. The 3 isoforms share ~60-70% sequence homology and are conventionally divided into 3 key functional domains. The extreme amino terminus (NT) serves as the ligand binding domain (LBD) while the carboxyl terminal (CT) 6 transmembrane region contributes to the oligomerization and localization of the protein to the ER and to formation of the ion conducting pore between transmembrane helices 5 and 6. These two conserved regions are flanked by a large, but weakly conserved intermediary regulatory domain which, together with the poorly conserved extreme CT tail, contains numerous putative sites for regulation by different modulators, notably Ca²⁺, protein kinase A (PKA) and adenosine triphosphate (ATP) [9]. The variation in primary sequence homology between the 3 isoforms results in each subtype exhibiting distinct IP₃R binding affinities and modulatory properties [12,13]. The differences in these properties and their contribution to the distinct Ca²⁺ release profiles of the 3 isoforms have been extensively investigated [9]. For instance, R2 has a ~ 3x and ~ 12x greater affinity for IP₃ than R1 and R3, respectively [12]. Similarly, R2 is more sensitive to regulation by ATP than R1 or R3 [14,15]. Additionally, although PKA phosphorylation of R1 and R2 potentiates Ca²⁺ release and single channel activity, the phosphorylation occurs at very different residues [16-19]. Conversely, although a substrate for PKA, the activity of R3 does not appear to be regulated by PKA phosphorylation [20].

Evidence for heterotetramer formation

An additional layer of complexity that would be predicted to have a major influence over IP₃R function, is the impact of heterotetrameric channel assembly. The initial studies identifying the IP₃R subtypes reported that all 3 isoforms were commonly expressed at the message or protein level often in the same cells and tissues [21-25]. Subsequently, It was demonstrated that IP₃R are probably ubiquitously expressed and all cells investigated expressed at least 2 isoforms [26]. Interestingly, it was noted that, while R1 is predominantly localized to neuronal cells and tissues, R2 and R3 are more frequently, though not exclusively, localized to peripheral tissues. These patterns of ubiquitous, yet differential expression have been confirmed in numerous subsequent studies [22,27-29]. Consistent with the existence of heterooligomer formation, co-localization and immunostaining studies also indicate that the different isoforms are often distributed similarly in subcellular compartments. For example, R1, R2 and R3 are all localized to the apical pole of pancreatic acinar cells on ER close to, or immediately below, the luminal membrane [30]. A similar localization of the three isoforms was reported in submandibular acinar cells [31]. In addition, a partial co-localization is evident in hepatocytes, as R2 specifically localizes to the pericanalicular regions at the apical portion of hepatocytes, while R1 is uniformly localized throughout the cytoplasmic ER [32,33].

The first strong evidence for hetero-oligomerization came from co-immunoprecipitation studies. For example, R1 and R2 were capable of associating with each other in rat liver lysate, and similar interactions between all 3 isoforms were documented in CHO-K1 cells [27]. In addition, heteromers of R1 and R2 were shown to assemble in AR42J cells, while similar associations between R1 and R3 were seen in RINm5F cells. Importantly, mixing lysates of different cell lines revealed that co-immunoprecipitation was not an artifact of post lysis dissociation and reassembly, but rather the consequence of isoform association in a heteromeric macromolecular complex [34]. Studies have also begun to address the biosynthetic “rules” governing heterotetramer formation. In COS-7 cells, heterologously expressed tagged IP₃R were found not associate with native IP₃R. Presumably, this lack of association occurs because the likelihood that endogenous IP₃R is translated coincidently with ectopically expressed protein is low. Further, the degree of heterooligomer formation depended on the relative expression of the two recombinant isoforms and was shown to not follow a strict binomial distribution. Interestingly, these studies demonstrated that heterooligomers took longer to synthesize than homooligomers [35]. Lastly, cells co-transfected with cDNA encoding dominant negative mutations of the ligand binding domain of R3 and channel pore of R1 were able to produce functional channels, providing additional compelling evidence for heterotetramer formation [36].

While these data provide strong evidence that heterotetramers are formed and further provide some biochemical “rules” for their generation, they do not provide any information regarding the relative abundance of homo vs. heterotetrameric assemblies. To investigate this important issue, our group has performed sequential immunodepletion of IP₃R isoforms from mouse pancreatic lysates. R2 and R3 isoforms predominate in the exocrine pancreas, with R1 only accounting for ~3% of IP3Rs. To remove both homo and heterotetrameric assemblies containing R1, lysates were first incubated with α-IP3RR1 antisera (experimental strategy shown in Fig 1A). This indeed resulted in total depletion of R1 and a reduction in immunoreactivity associated with R2 and R3. These data are of course, entirely consistent with the presence of heterotetrameric IP₃R in pancreatic acinar cells. Similarly, subsequent incubation of this lysate with α-IP3RR2, totally removed R2 expressing tetramers, leaving only R3 homomeric IP₃R. Notably, these experiments revealed that only ~11% of R3 exists as a homotetramer in the pancreas (Fig 1B/C) [28]. While these data support the notion that heterotetrameric IP₃R receptors are the predominant species in exocrine cells, further similar experiments are needed to establish that this is a common occurrence in cells expressing multiple IP₃R subtypes.

FIG 1. Immunodepletion experiments supporting the existence of heteromeric IP₃Rs in vivo.

(A) Schematic representation of strategy for immunodepletion experiment.

(B) Mouse pancreas lysates were subject to sequential immunoprecipitation, first with α-InsP₃R2 (left blot, middle lane) followed by α-InsP₃R1 (left blot, right lane) to maximally deplete R2 and R1, respectively (right blot). Equivalent supernatants, before (Input) and after immunodepletion (Post-ID), were separated using SDS-PAGE and probed with the indicated antibodies.

(C) A histogram comparing R3 immunoreactivity post R2 and post R2 + R1 immunodepletion. Quantitative densitometry revealed ~11% of R3 exists as homotetramers. Adapted from [28] with permission.

Functional implications of heterotetrameric IP₃R

These studies clearly suggest that most cells and tissues express multiple IP₃R isoforms that likely oligomerize into heterotetrameric complexes. Nevertheless, the functional consequences of heterotetramer formation have largely been overlooked. This is especially important given that each isoform exhibits markedly different regulatory and functional properties [9]. For example, it is not clear if the activity of heterotetramers simply reflects the integrated “blended” activity of the constituent monomers or alternatively if a particular monomer dominates the overall IP₃R properties. Indeed, several studies exploring the regulatory properties of IP₃Rs in native tissues have suggested that the overall characteristics of regulation of IP₃R activity are not consistent with distinct populations of homotetrameric channels and may reflect unique properties of heterotetramers. For example, the IP₃ binding affinity of R2 is ~3 fold greater than R1 and ~11 fold greater for R3 (14 nM vs. 50 nM vs. 163 nM; R2 vs R1 vs R3, respectively) [12]. Nevertheless, it has been reported that membrane fractions isolated from rat liver exhibited a Kd of 1.7 nM, obviously significantly different from that of R1 or R2 in isolation [37]. Additionally, experiments from our group have shown that the IP₃ binding affinity to parotid acinar and pancreatic acinar cell membranes revealed Kds of 6.3 nM and 6.1 nM, respectively [38]. These Kd values are again significantly different from those reported for R2 and R3, in isolation, possibly pointing towards a degree of functional interaction between the isoforms.

Notably, studies investigating IP₃R regulation by ATP have shown that the regulatory properties appear to be “dominated” by a specific isoform. Studies in DT40 3KO cells (an IP₃R null background) stably expressing individual mammalian isoforms, reported R2 to have a ~3 fold higher affinity for ATP than R1 and ~10 fold higher than R3, respectively (40 μM vs. 150 μM vs. 400 μM) [14,15]. In addition, while R1 and R3 appear to require ATP at all [IP₃] to achieve maximal release rates, the activity of R2 is only modulated by ATP at sub-maximal [IP₃]. Given this information, it is interesting to note that despite approximately equal expression of R2 and R3, Ca²⁺ release assays performed on mouse pancreatic acini revealed an EC₅₀ of 38 μM and activity was not modulated at saturating [IP₃]; properties essentially identical to R2 in isolation. In contrast, pancreatic acini isolated from R2 KO mice revealed an EC₅₀ of 450 μM, indistinguishable from R3 in isolation [39]. Similar EC₅₀ values were seen in AR42J (10 μM) and RinM5F cells (430 μM), which predominantly express R2 and R3, respectively [26]. Moreover, transient transfection of RinM5F cells with cDNA encoding murine R2 resulted in a ~10 fold increase in sensitivity to ATP. Together, these data point towards R2 dictating the ATP sensitivity of IP₃Rs [39]. In hindsight, these findings are entirely consistent with studies using DT40 cells expressing native chicken IP₃Rs and engineered cells expressing a defined complement of IP₃R by targeted knockdown of either one, two or three genes. [13]. Ca²⁺ release assays revealed that R2 dominated the ATP sensitivity, followed by R1 and subsequently R3 (R2>R3>R1). In stark contrast, [IP₃] concentration vs response relationships revealed a ‘blended’ sensitivity, further contributing to the notion that the different modes of IP₃R regulation are differentially influenced by the various isoforms.

Generating heterotetrameric IP₃R with defined composition

The major limitation to these studies conducted in cultured cells and isolated tissues is the inability to predict the exact proportion of each isoform in a heterotetramer, or indeed the percentage of homotetramers in the IP₃R population. Similar issues arise using any approach utilizing ectopic co-expression of cDNA constructs encoding multiple IP₃R subtypes. To circumvent this, our group recently described a strategy whereby the functional characteristics of heterotetramers with unequivocally defined composition can be explored [28]. This approach is based on engineering concatenated IP₃R whereby the C-terminus of the initial monomer is connected to the N-terminus of a subsequent monomer by a short flexible glutamine rich linker. In theory, this approach can be extended to construct cDNA encoding the tetrameric protein. This paradigm has been successfully utilized in the past to study the functional characteristics of a variety of multi-subunit channels including Orai and various K⁺ channel family members [40-44]. The extremely large size of the cDNA encoding IP₃R, however, made constructing tetrameric IP₃R concatamers initially a rather daunting undertaking. We therefore decided to construct cDNA encoding dimeric InP₃R in a novel vector designed to manipulate large DNA strands (see cartoon of pJAZZ_mamm in Fig 2A). The guiding rationale for this approach being that, upon expression, the biosynthetic machinery of the cell might dimerize the concatenated proteins to form tetrameric channels of defined subunit composition. Briefly, linear cDNA constructs encoding mammalian IP₃R encoding all possible combinations of concatenated homodimeric and heterodimeric IP₃Rs were successfully generated and stably expressed in DT40 3KO cells (Fig 2B and [28]). Remarkably, these proteins were localized to ER membranes and exclusively assembled into tetrameric receptors in an identical fashion to channels assembled from monomeric constructs (Fig 2C). Functionally, competitive IP₃R binding assays demonstrated that the dimeric proteins bound IP₃R comparably to the monomers [28]. Importantly, cell surface receptor stimulation and Ca²⁺ release assays revealed that all channels assembled from concatenated receptors formed fully functional IP₃R. A closer investigation into the single channel properties using ‘on-nuclear’ patch clamp revealed that the homodimeric R1R1 and R2R2 exhibited open probabilities and gating characteristics similar to the R1 and R2 monomeric proteins, thereby indicating that they are functionally indistinguishable [28].

Fig 2. Generation of concatenated IP₃R constructs and their stable expression DT40-3KO IP₃R null cells.

(A) InsP₃R ‘head’ and ‘tail’ subunits were subcloned into pJAZZ_mamm, a linear vector plasmid developed from N15 coliphage (Lucigen).

(B) Representative western blots showing that concatenated dimer constructs were stably transfected and expressed into DT40-3KO cells.

(C) Representative blue native gels demonstrating that DT40-3KO cells express monomeric or dimeric constructs are assembled to form tetramers. Adapted from [28] with permission.

The generation of tetrameric IP₃R from defined heterodimers now allows us to begin to address a longstanding question as to how the distinct subunit composition of a heterotetramer contributes to the functionality and regulation of the receptor. In DT40 cells expressing a single isoform, B cell receptor stimulation with sub maximal α-IgM resulted in each IP₃R subtype exhibiting Ca²⁺ signals with distinct temporal “signatures” [13]. The characteristics of these Ca²⁺ signals are believed to depend on the characteristic IP₃R binding and integration of the distinct regulatory properties of each isoform. For example, while R1 was shown to generate monophasic Ca²⁺ transients, R2 was shown to produce long lasting oscillatory Ca²⁺ signals. Similar “signatures” were observed by our group in DT40 3KO cells stably expressing mammalian R1 and R2 monomer and importantly, concatenated dimer constructs [28]. Interestingly, cells expressing R1R2 heterodimers, unequivocally forming a tetramer with equal numbers of R1 and R2 monomers, generated robust oscillatory Ca²⁺ signals, qualitatively and quantitatively similar to those generated by R2 alone [28] (Fig 3A-F). R2 dominance was also observed in cells expressing R2R3 heterodimers (data not shown). Further investigation into IP₃R regulation revealed similar dominant characteristics of R2. As noted, regulation of R1 and R2 by ATP has distinct defining characteristics, which include specific gating kinetics observed at the single-channel level. Increasing [ATP] prolongs the duration of R1 bursts, while ATP increased the number of bursts without affecting duration for R2. As shown in Fig 3 G-I), ATP does not regulate the activity of R1R2 heterodimers at saturating [IP₃] and mimics the gating characteristics formed by R2 monomers and concatenated dimers [28], clearly demonstrating that two R2 subunits within the heterotetramer leads to R2 dominant properties.

Fig 3. Activity of concatenated heterodimers is dominated by R2.

(A-F) Activation of B cell Receptor in DT40-3KO cells stably expressing R1 and R2 containing monomeric and dimeric InsP₃Rs results in characteristic Ca²⁺ signals. R2 properties appear to dominate.

(G-I) Representative single channel K⁺ current recordings of R1R1, R2R2 and R1R2 dimers, measured using on-nucleus patch clamp, at 1 or 10 μM InsP₃ in the presence of 200 nM free Ca²⁺ and either 100 μM or 5 mM ATP. R2 properties appear to dominate. Adapted from [28] with permission.

Clearly, the use of concatenated IP₃R constructs is a powerful tool to study the function and regulation of IP₃R of defined composition. In fact, our group has successfully generated a membrane localized, functional homotetrameric R1R1R1R1 concatamer [28]. The generation of this tetramer provides proof of principle that any possible combination of heterotetrameric IP₃R can potentially be generated and studied. This may be especially important in studying the impact of incorporation of defined numbers of disease associated IP₃R mutations to mimic heterozygous expression. Furthermore, given that all known modulatory motifs are present in each monomer and it is not known how many monomers need to be “engaged” to influence activity, the ability to mutate residues in defined numbers of monomers should allow us to establish the stoichiometry required for channel regulation. A priority is to definitely answer fundamental questions such as the number of IP₃ molecules required to bind to gate the IP₃R.