Calcium Flux Assays for Gq Receptors and the Non-Equilibrium Problem

Add agonist to a plate of cells expressing a G_q-coupled receptor and cytosolic free calcium climbs within seconds, peaks, and falls back — usually most of the way — before the read is a minute old. In most such reads the pharmacological parameters come off an event that finishes before ligand and receptor have equilibrated. Charlton and Vauquelin made the consequence explicit: modelling the calcium transient in a kinetic mode allows the influence of ligand kinetics, receptor reserve and read time to be explored, and their analysis showed that observed ligand pharmacology at very early time points can be quite different from that determined after longer incubations, even resulting in reversal of agonist potency orders that may be misinterpreted as agonist biased signalling ^[1].

This guide is about what a calcium mobilisation number is made of: which parts describe the ligand, which describe the indicator, and which describe the way you forced the receptor onto the phospholipase C arm.

What the well is actually reporting

The canonical chain is textbook and is stated here without a primary citation: agonist-occupied receptor promotes nucleotide exchange on Gα_q, Gα_q activates phospholipase C beta, PLCβ cleaves PIP₂ into inositol 1,4,5-trisphosphate and diacylglycerol, and IP₃ opens its receptors on the endoplasmic reticulum so that store calcium enters the cytosol.

The plate reader does not see that chain. It sees an added reporter that binds calcium with its own affinity, its own association and dissociation rates, and its own capacity to compete with the cell for free calcium ions. Each property inserts a distortion between the biology and the trace; the third is the one most often forgotten. Neher and Augustine, dialysing Fura-2 into bovine chromaffin cells, reported that this caused concentration-dependent changes in the depolarization-induced Ca²⁺ signal, decreasing its magnitude and slowing its recovery time course; two independent estimates in that work suggested that 98–99% of the Ca²⁺ entering the cell is normally taken up by a fast endogenous buffer whose low capacity, affinity and mobility make it possible for relatively small amounts of exogenous buffers such as Fura-2 to exert a significant influence on the measured signal ^[2]. That was patch-pipette dialysis with depolarization-evoked entry, not IP₃-driven store release in a microplate, so the numbers do not transfer; the principle — the indicator is a buffer, and buffers change what they measure — does.

Indicator chemistry sets the ceiling on the answer

Single-wavelength dyes

Fluo-4 is a common single-wavelength choice for plate work. Gee and colleagues developed it for quantifying cellular Ca²⁺ in the 100 nM to 1 µM range, reporting that its greater absorption near 488 nm gives substantially brighter emission than fluo-3 under argon-ion excitation with a standard fluorescein filter set, and that in vitro it showed high fluorescence emission, a high rate of cell permeation, and a large dynamic range for reporting calcium around a K_d of 345 nM. They also made lower-affinity relatives for studies needing quantification of higher calcium concentrations, and noted that the higher emission intensity lets the dye be used at lower intracellular concentrations, which they described as a less invasive practice ^[3].

Two consequences follow. First, the signal is a single intensity, so it carries dye loading, cell number per well, optical path and photobleaching along with the calcium; ratioing the trace to its own pre-injection baseline removes the static part of that but not the drift. Second, affinity is a ceiling. Charlton and Vauquelin point out that above roughly ten-fold the dye K_d the dye saturates, so in cells that generate large calcium changes a high-affinity dye may saturate and underestimate the magnitude of the signal ^[1]. A saturating dye compresses E_max, flattening exactly the difference between a full and a partial agonist that you were trying to measure; if your best agonists all top out at the same plateau, test a lower-affinity indicator before concluding they are equally efficacious.

Ratiometric dyes

Fura-2 came from the family Grynkiewicz, Poenie and Tsien synthesised by combining an 8-coordinate tetracarboxylate chelating site with stilbene chromophores. Against their predecessor quin2 the new dyes offered up to 30-fold brighter fluorescence, major changes in wavelength and not just intensity on calcium binding, slightly lower affinities for calcium, and considerably improved selectivity for Ca²⁺ over other divalent cations ^[4]. The wavelength shift is what buys the ratio: dividing emission collected at two excitation wavelengths cancels terms scaling with dye concentration and cell thickness, so a poorly loaded well and a well-loaded one stay comparable. The cost is instrumentation: near-UV excitation and dual-wavelength switching, which is why high-throughput plates usually run single-wavelength dyes and absorb the normalisation problem instead.

Aequorin

The photoprotein route removes the dye-loading step by expressing the reporter, but not loading altogether: the apoprotein still has to be reconstituted with a cofactor supplied to the cells. Stables and colleagues transiently expressed apoaequorin in CHO cells, reconstituted it with coelenterazine, and obtained a large, concentration-dependent agonist-mediated luminescent response on cotransfection with endothelin ET_A, angiotensin, thyrotropin-releasing hormone and neurokinin NK₁ receptors. Substantially greater luminescence was obtained with mitochondrially targeted apoaequorin than with the cytoplasmic form ^[5]. Hold onto that targeting detail: a mitochondrially targeted photoprotein reports the mitochondrial compartment, so the trace is compartment-specific rather than a cytosolic concentration.

Genetically encoded indicators

Nakai and colleagues built G-CaMP from a single GFP and characterised the properties that matter for kinetic work: an apparent K_d for Ca²⁺ of 235 nM, association kinetics that were faster at higher calcium with time constants decreasing from 230 ms at 0.2 µM to 2.5 ms at 1 µM, and dissociation kinetics of roughly 200 ms that were independent of calcium concentration; in HEK-293 cells and mouse myotubes expressing the probe, large fluorescent changes followed compound application or electrical stimulation ^[6]. A calcium-independent off-rate of that size sets a floor on the decay you can resolve: any parameter taken from the falling phase of a GCaMP trace — time to half decay, the tail of an integral — is partly the sensor relaxing and partly the cell clearing calcium, and the assay cannot separate them.

Loading, leak, and the price of blocking it

Acetoxymethyl ester loading followed by cleavage by cytosolic esterases to trap the charged free acid is standard practice and is uncited here. What is not standard is assuming the trapped dye stays put. Di Virgilio, Steinberg and Silverstein found that in mouse peritoneal macrophages, J774 cells, PC12 cells and N2A cells, Fura-2 sequestration and secretion are mediated by organic anion transport systems and are blocked by the inhibitors probenecid and sulfinpyrazone, adding that under appropriate conditions these agents have little effect on calcium transients and may facilitate the use of Fura-2 in a variety of cell types ^[7]. Note the hedges: under appropriate conditions, and may facilitate. Neither the block nor its innocence is established for a cell background the authors did not test.

The reason to check is that the blocker has pharmacology of its own. Bang and colleagues showed by electrophysiology and calcium imaging that probenecid activates TRPV2 in TRPV2-expressing HEK293T cells, while TRPV1, TRPV3, TRPV4, TRPM8 and TRPA1 failed to respond to it in the same heterologous expression system, which they read as probenecid being a specific activator for TRPV2 ^[8]. That is a heterologous TRP panel, not a claim about every cell line, but the instruction it generates is cheap: add the blocker alone to a set of wells and watch the baseline before attributing any of it to your receptor.

Peak, area, and the read window

Most laboratories quantify the transient by its peak. The alternative is to integrate it. Charlton and Vauquelin report that in many systems the concentration-response curves that result are similar whether peak or area under the curve is used as the measure of response ^[1]. Take that in both directions: the peak-versus-AUC choice is not automatically what is distorting your EC₅₀, and a divergence between the two in your own hands is therefore worth chasing rather than expecting, because it localises the difference to the falling phase that the peak discards.

What does move the number reliably is the read window. The same analysis showed that pharmacology at very early time points can differ substantially from pharmacology after longer incubations, up to reversing potency order, and that estimates of antagonist affinity — whether obtained by Schild analysis or from inhibition curves — are similarly affected by hemi-equilibrium conditions ^[1]. An EC₅₀ from a calcium plate is therefore inseparable from its sampling interval and its integration limits, and a table that omits them is not reporting a reproducible measurement.

Klein Herenbrink and colleagues set out the wider version: kinetic context, as determined by ligand-binding kinetics and the temporal pattern of receptor-signalling processes, can have a profound influence on the apparent bias of a series of agonists and can even lead to reversals in the direction of bias ^[9]. Their system was the dopamine D₂ receptor with a defined agonist series, so the specific reversals are theirs; the transferable conclusion is that a difference between a calcium potency and a potency from a slower readout is not on its own evidence about receptor conformation.

Forcing the coupling

Much calcium screening is done on receptors that do not natively couple to G_q. Two constructs make that possible, and both change the pharmacology they let you read.

Offermanns and Simon cotransfected receptors and G-protein alpha subunits into COS-7 cells and found that β₂-adrenergic and muscarinic M₂ receptors gave very small stimulation of endogenous phospholipase C when expressed alone or with Gα_q, Gα₁₁ or Gα₁₄, but that coexpression with Gα₁₅ or Gα₁₆ produced a severalfold increase in inositol phosphate production that varied with time and with agonist concentration. Several other receptors previously shown to couple to G_i and G_s family members behaved the same way, and stimulation via receptors natively coupled to phospholipase C was itself enhanced by cotransfection of Gα₁₅ and Gα₁₆ ^[10]. That last clause is the one to plan around: adding a promiscuous alpha subunit to a receptor that already couples to G_q does not merely enable the readout, it raises the gain on a pathway already present.

The chimeric route is more surgical. Conklin and colleagues replaced C-terminal amino acids of α_q with the corresponding residues of α_i2, producing α_q/α_i2 chimaeras that could mediate stimulation of phospholipase C by receptors otherwise coupled exclusively to G_i, and found that a minimum of three α_i2 amino acids, including a glycine three residues from the C terminus, sufficed to switch receptor specificity ^[11]. Chavkin and colleagues used this pairing in practice, cotransfecting either the chimeric Gα_qi5 or the universal Gα₁₆ and quantifying agonist-evoked calcium mobilization as an additional measure of potency and efficacy at a cloned human receptor in HEK-293 cells ^[12].

Kostenis reviewed the approach critically, noting that the promiscuous nature of Gα₁₆ is one route to a universal assay but that a truly critical look at it is still required, and setting out its strengths, weaknesses, problems and pitfalls ^[13]. The boundary this draws is sharp. In a Gα₁₆ or Gα_qi5 well, an EC₅₀ licenses two statements: the ligand activates that receptor, and the activated receptor can drive that alpha subunit. It licenses nothing about whether the receptor mobilises calcium in a cell expressing it at native density, and nothing about the potency you would measure through the receptor's own transducer.

Receptor reserve, and why the clone owns part of your EC₅₀

Black and Leff's operational model describes agonism with three parameters — an agonist-receptor dissociation constant K_A, the total receptor concentration [R₀], and a parameter K_E defining the transduction of agonist-receptor complex into effect — and identifies the ratio [R₀]/K_E, the transducer ratio τ, as a logical definition for the efficacy of an agonist in a system ^[14]. The phrase in a system is doing the work. τ contains the receptor number, so a clone selected for bright signal has a larger τ than a clone selected for physiological density, and the same ligand returns a smaller EC₅₀ in the first than in the second with no change in its affinity.

Two things follow from the model's arithmetic rather than from any single experiment. A weak-efficacy ligand in a high-τ system can still reach the system maximum and read as a full agonist, so apparent efficacy in an overexpressing calcium line is a statement about the line. And the leftward displacement of EC₅₀ from K_A grows with τ, so the gap between a functional potency and a binding affinity is not, by itself, an anomaly to be explained away.

Chavkin and colleagues took the practical route, stating that because issues of receptor reserve complicate estimates of agonist efficacy and potency they also measured potassium conductance through G protein-gated K⁺ channels coexpressed in Xenopus oocytes, a system in which receptor reserve is minimal ^[12]. The transferable move is the cross-check, not that particular orthogonal assay: bound the reserve by rerunning the ligand set in a lower-expressing clone or at lower transient DNA load, or by titrating functional receptors down with an irreversible antagonist and refitting. Either way you are asking whether the EC₅₀ belongs to the ligand or to the transfection.

Controls that decide what the trace means

Because the readout is a downstream convergence point, a great many things produce a calcium spike. The controls below are procedural rather than findings, and are uncited for that reason, with one exception.

• The parental line carrying no receptor of interest, run with identical additions. Heterologous hosts are not blank: Atwood and colleagues profiled non-chemosensory GPCRs and over a hundred GPCR signalling-related gene products by microarray in HEK293, AtT20, BV2 and N18, observing that although often treated as blank slates these lines nevertheless endogenously express GPCRs and related signalling proteins, and that the outcome of a given study can be profoundly influenced by this largely unknown complement ^[15]. Those are mRNA levels in four named lines, not protein, not function and not your clone.
• Compound alone into wells with no cells, to separate autofluorescence and compound quench from a biological response.
• The anion-transport blocker alone, as above.
• A receptor-independent positive control, so that a failed plate can be distinguished from an inactive ligand.
• The same additions in nominally calcium-free assay buffer, which is the standard way to ask whether the transient depends on entry across the plasma membrane rather than on store release.

Reading a calcium number beside a cAMP number

The companion guide on cAMP accumulation assays describes that readout as a running integral over the whole incubation, terminated by lysis, with a phosphodiesterase inhibitor holding degradation off throughout. A calcium transient is the structural opposite: a rise and a fall sampled at sub-second to few-second resolution, complete on a timescale at which the cAMP plate has not yet been read. The two differ in amplification, in integration, and in how far from equilibrium they are when the number is taken.

The rule that falls out is narrow. A calcium EC₅₀ supports rank ordering of ligands within one plate, one clone, one indicator and one read window; it supports a yes-or-no statement about whether a ligand drives the G_q/PLC arm in that system; and it supports antagonist estimates only when the hemi-equilibrium caveat travels with them ^[1]. It does not support a claim that one ligand is biased because its calcium potency and its cAMP potency differ, and it does not support carrying a potency from a Gα₁₆ plate into any statement about the receptor's native signalling.