An Inconvenient Truth: Calcium Sensors Are Calcium Buffers

Abstract

Recent advances in Ca²⁺ imaging have given neuroscientists a tool to follow the activity of large numbers of individual neurons simultaneously in vivo in the brains of animals as they are presented with sensory stimulation, respond to environmental challenges, and engage in behaviors. The Ca²⁺ sensors used to transduce changes in cellular Ca²⁺ into changes in fluorescence must bind Ca²⁺ to produce a signal. By binding Ca²⁺, these sensors can act as buffers, often reducing the magnitude of a Ca²⁺ change several fold and producing a proportional slowing of the rates of change. Ca²⁺ probes can thus distort the patterns of activity they are intended to study and modify ongoing Ca²⁺ signaling functions. Recognizing these factors will enhance the use of in vivo Ca²⁺ imaging in the investigation of neural circuit function.

The impact of Ca²⁺ probes on Ca²⁺ signals

Any molecule that binds Ca²⁺ can influence its dynamics. All fluorescent Ca²⁺ sensors currently in use, including both synthetic and genetically-encoded probes, report the presence of Ca²⁺ through a binding induced change in fluorescence. If the sensor binds to an appreciable fraction of cellular Ca²⁺ then it will act as a buffer to limit the magnitude of changes in Ca²⁺ concentration and reduce their speed. This action will depend critically on the affinity and concentration of the sensor. If a sensor’s buffering is strong, it will provide a distorted picture of neuronal activity and perturb a neuron’s Ca²⁺ signaling. Although the buffering actions of Ca²⁺ probes are well recognized [1–3], they are often overlooked and underestimated. In fact, the buffering action of a probe can dominate observed Ca²⁺ signals. In this Opinion we revisit some early evidence for Ca²⁺ buffering by probes, and review the well-established framework for the quantitative analysis of buffering strength. We then discuss the magnitudes of these effects and their impact on experiments, concluding with suggestions for estimating the impact of Ca²⁺ probes on Ca²⁺ imaging experiments.

The buffering action of a Ca²⁺ probe is illustrated in Fig. 1 taken from an early fluorometric Ca²⁺ imaging experiment [4]. The Ca²⁺-sensitive fluorescent dye fura-2 was loaded into a chromaffin cell through a patch pipette (Fig. 1A). The loading was monitored by observing the fluorescence elicited by 360 nm light, a wavelength that excites both Ca²⁺-bound and free fura-2 roughly equally, so this signal does not depend on Ca²⁺. This trace shows that the fura-2 from the patch pipette fills the cell in about 5 minutes. At the plateau in fluorescence, the concentration of fura-2 in the cell reached that in the patch pipette, 400µM.As the loading progressed, voltage pulses of a set amplitude and duration were applied repeatedly to open Ca²⁺ channels and raise intracellular free Ca²⁺ ([Ca²⁺]_i).These Ca²⁺ rises were signaled by transient reductions in fluorescence elicited by 390 nm light, a wavelength that preferentially excites the free species of fura-2. The [Ca²⁺]_i computed from the ratio of the two spectral channels illustrates two important consequences of buffering by the Ca²⁺ sensor: the voltage-induced increase in [Ca²⁺]_i declines from nearly 0.5 µM early in the loading process to about 50nM at the plateau; and the decay time increases from a few seconds to nearly half a minute. Note that the final voltage pulse in the sequence was 500msec instead of 50msec to demonstrate that a large [Ca²⁺]_i transient could still be evoked but with a slow decay. This experiment demonstrates convincingly that the sensor strongly buffers intracellular Ca2+. This study also reversed the buffering effect by establishing a new patch clamp recording with a lower fura-2 concentration (Fig. 1B). The [Ca²⁺]_i transients then reverted to their larger and aster character seen early in the loading process. Similar probe loading sequences were reported in septal neurons [5] and proximal dendrites of neocortical pyramidal cells [6].

Figure 1. Ca²⁺ buffering by the Ca²⁺ sensor fura-2 in chromaffin cells.

A. In a patch clamped chromaffin cell, the fluorescence of the dye is excited with two different wavelengths nearly simultaneously by rapid switching between excitation filters. The fluorescence at the Ca²⁺-independent wavelength (360 nm, top trace) gives the time course of fura-2 loading into the cell. Voltage pulses of 50msec from −70 mV to 10mV were applied at irregular intervals during the loading, and dips in the Ca -sensitive wavelength (390 nm, middle trace) reported Ca rises. The final fluorescence change was elicited by a longer voltage pulse of 500 msec. The intracellular free Ca²⁺ concentration, [Ca²⁺]_i, computed from the ratio of fluorescence elicited by the two wavelengths (bottom trace), illustrates the decrease in amplitude of the [Ca²⁺]_i rise and increase in decay time as [fura-2] in the cell rises to a final concentration of 400 µM. B. After removing the patch electrode containing 400 µM fura-2, a second recording was made from the same cell using a patch pipette containing 50 µM fura-2. The decline in fluorescence at 360 nm tracked the reduction in fura-2 concentration. As the dye concentration fell, the Ca²⁺ rises evoked by 20msec pulses became larger and the decays became more rapid. Extracellular Ca²⁺ was 5mM. Reproduced with permission from Ne her and Augustine, 1992 [4].

Quantifying buffering actions

The buffering actions of fluorescent Ca²⁺ sensors came to light in studies of the role of Ca²⁺ in exocytosis [4] and excitation-contraction coupling [7]. This work played an essential conceptual role in elucidating how Ca²⁺ serves as a signal in these processes, and led to the development of quantitative tools for understanding the impact of buffers on Ca²⁺ fluorescent imaging. The amount of buffering by a Ca²⁺-binding molecule, B, under a particular set of experimental conditions can be expressed as the buffering capacity, k[1, 4, 7].

$$K=\frac{\Delta \left[CaB\right]}{\Delta \left[C{a}^{2+}\right]}=\frac{K_d\left[B\right]}{{\left(K_d+\left[C{a}^{2+}\right]\right)}^2}$$ (1)

K is the ratio of the change in bound Ca²⁺ to the change in free Ca²⁺, and K_d is the dissociation constant of B. The binding sites of endogenous Ca²⁺ buffers saturate as Ca²⁺ rises [8–11], so k decreases as [Ca²⁺] rises. But κ provides a useful index of the impact of Ca²⁺ increments at a given ambient concentration. The experiment illustrated in Fig. 1A used a high concentration of fura-2 (400 µM) in order to demonstrate clearly the buffering action of the dye. For this concentration fura-2 contributes a κ of 889 (with K_d = 200nM and a nominal resting [Ca²⁺]_i of 100nM). Thus, considering only fura-2, the sensor binds 99.9% of entering Ca²⁺ at rest, leaving only about 0.1% as free ion in solution.

Many proteins and metabolites also bind Ca²⁺, and these endogenous buffers work together with a sensor in an additive fashion to shape Ca²⁺ transients. Taking the endogenous and sensor buffering capacities as e and K_s, respectively, it can be shown that when a Ca²⁺ addition produces a change in total Ca²⁺ of Δ[Ca²⁺]_Total, free Ca²⁺ changes by [4]

$$\Delta {\left[C{a}^{2+}\right]}_i=\Delta {\left[C{a}^{2+}\right]}_{\text{Total}}/\left(1+{\kappa }_s+{\kappa }_e\right)$$ (2)

The two buffers prolong the observed decay time, τ_obs, by the same factor

$${\tau }_{obs}={\tau }_0\left(1+{\kappa }_s+{\kappa }_e\right)$$ (3)

where τ₀ is the decay time in the absence of any buffer. The effects of the endogenous buffers on a Ca²⁺ signal, through ke, are biological, while the effects of the sensor, through K_s, are artifacts. Given this additive action of buffers, we can anticipate that a Ca²⁺ sensor will dominate Ca²⁺ dynamics as K_s surpasses K_e.

Implications for Ca²⁺ imaging experiments

The above reasoning indicates that one can assess the impact of the Ca²⁺ buffering action of a probe by comparing ks and k_e. k_e values have a range of roughly 50–500 [1]. A value of 100 is fairly typical, although recent work in one particular cell type gave a value of only 20 [12]. According to Eq. 1, with [Ca²⁺]_i = 100 nM fura-2 will have a ks = 20 at a concentration of only 450 nM. Most imaging systems could not detect the fluorescence of a dye at such a low concentration. Even a more typical k value of 100 corresponds to [fura-2] = 2.25 µM, which is on a Ca²⁺ signal, through k_e, are biological, while the effects of the sensor, through ks, are artifacts. Given this additive action of buffers, we can anticipate that a Ca²⁺ sensor will dominate Ca²⁺ dynamics as κs surpasses k_e The above reasoning indicates that one can assess the impact of the Ca²⁺ buffering action of a probe by comparing ks and k_e. k_e values have a range of roughly 50–500 [1]. value of 100 is fairly typical, although recent work in one particular cell type gave a value of only 20 [12]. According to Eq. 1, with [Ca²⁺]_i = 100nM fura-2 will have a ks = 20 at a concentration of only 450nM. Most imaging systems could not detect the fluorescence of a dye at such a low till usually too low to use. Thus, if there is enough sensor to image Ca²⁺, then there is almost certainly enough to produce major distortions of Ca²⁺ signals. In hippocampal axons filled with known concentrations by patch loading, extrapolating to zero sensor yielded action potential-induced Ca²⁺ rises that were 20-fold higher and decays that were 20-fold faster than those seen in fura-2-AM loaded axons [10]. In the absence of a sensor, Ca²⁺ transients decay in tens of milliseconds in spines, thin dendritic shafts [13], and nerve terminals [10]; in less than 100msec in thick dendrites adjacent to the soma [6]; and perhaps hundreds of milliseconds in cell bodies [5, 14]. Thus, the decay times of seconds often seen in Ca²⁺ imaging experiments in vivo suggest that probe buffering is a dominant force. The strong buffering action of the probe in typical experiments can have serious consequences. For instance, by dominating Ca²⁺ dynamics a probe could alter the timing of responses in sequential activations of different cells in a circuit Slowing down Ca²⁺ changes could mask timing differences and create the appearance of synchrony.

Ca²⁺ probes can also disrupt Ca²⁺-dependent cellular signaling and thus alter cellular function. Injecting chelators into the squid giant synaptic terminal reduced release [15], so one should expect Ca²⁺ probes to have a similar action. GCaMP6m expression in the calyx of Held synaptic terminal decreased synaptic vesicle release and slowed short-term depression [16]. Transgenic mice expressing GCaMP6 had a higher incidence of epileptiform events [17], probably because lower rises in intracellular Ca²⁺ reduced the activation of Ca²⁺-activated K⁺ channels.

There has been a surge of studies using Ca²⁺ imaging to monitor electrical activity in vivo during behavior [e.g. 18–21]. In many of these studies, there seems to be little indication that the buffering action of the Ca²⁺ sensor is fully recognized. The Ca²⁺ sensor is introduced either by bolus injection of the AM-ester form of the dye [22], or by expression of a protein that serves as a genetically-encoded Ca²⁺ sensor [23, 24]. With both methods the cytosolic concentration of the sensor is almost impossible to control. But by the logic outlined above, if one can see its fluorescence then the sensor is almost certainly having an impact on cellular Ca² dynamics. In fact, because of the problem of tissue autofluorescence, in order to image Ca²⁺ in vivo the cytosolic concentration of the probe must be quite high to see over background. One might argue that the fluorescence of a Ca²⁺ sensor still reports the electrical activity during behavior, and that the distortions are only relevant to mechanistic studies of Ca²⁺ dynamics. However, differences in the signal amplitudes between cells may simply reflect different levels of probe expression. Brighter cells might appear less active because their higher sensor concentrations suppress Ca²⁺ changes. Furthermore, two cells with the same level of sensor but different k_e values would also appear different in terms of Ca²⁺ signaling because the probe would buffer a greater fraction of entering Ca²⁺ in the low-k_e cell compared to the high-k_e cell.

Concluding remarks and future perspectives

Researchers will be able to use the techniques of in vivo Ca²⁺ imaging more effectively if they recognize the artifacts caused by Ca²⁺ sensors. The direct determination of k_e and ks is not practical in many kinds of experiments. Nevertheless, researchers can take some measures to address the buffering action of a Ca²⁺ sensor. Low affinity dyes are weaker buffers, but also produce smaller fractional fluorescence changes for physiological changes in Ca²⁺. Calibration of the optical system against test samples of probe, along with assessment of tissue autofluorescence will permit probe brightness within a voxel to be converted to obtain a rough measure of probe concentration. The fluorescence of beads implanted in tissue could be compared to the fluorescence of beads in a transparent medium to assess the impact of tissue light scattering on excitation and light collection. In principle, with an estimate of sensor concentration one could use Eq. 1 to calculate K_s from the K_d and a nominal resting free [Ca²⁺]i. This can be difficult, though, because free [Ca²⁺]_i is often unknown, probe affinities differ between in vitro and in vivo environments, and the fraction of cytosolic volume accessible to the probe is not precisely known. The accumulated impact of these factors could be quite large. Another approach to evaluating the impact of a Ca²⁺ sensor is based on decay times. The much longer decay times in virtually all the in vivo reports indicate that the sensor is buffering. One can estimate the decay time of an evoked Ca²⁺ rise by patch clamping a neuron with a known probe concentration. For a rapidly equilibrating dye, the extrapolation to zero dye concentration will give the biologically relevant decay time for a Ca²⁺ rise, τobs = τ0(1 +k_e) from Eq. 3. he prolongation of this decay time under less well defined loading conditions would then provide an indication of the impact of the sensor, including its increment to k from k_d. As in vivo Ca²⁺ imaging advances, attention to these issues will enhance the power of a technique that is offering unique and remarkable insights into the function of the nervous system.

Outstanding Questions.

• How do Ca²⁺ sensors alter the patterns of observed neural activity?

• How much do Ca²⁺ sensors alter Ca²⁺ signaling functions of a neuron during sensory input processing and the generation of behavior?

• Can technological advances mitigate the artifacts due to Ca²⁺ buffering by Ca²⁺ sensors?

Highlights.

• Although the buffering actions of Ca²⁺ sensors are widely known, their magnitude and impact are often not fully appreciated. This buffering action will reduce the amplitude of a change in intracellular Ca²⁺ and prolong the decay.

• The magnitude of this buffering action is reflected in a sensor’s buffering capacity, which depends on the concentration and affinity of the sensor The buffering capacity of a sensor exceeds the buffering capacity of endogenous Ca²⁺ buffers (Ca²⁺-binding proteins and metabolites) under the conditions that pertain during most Ca²⁺ imaging experiments. As a result, the Ca²⁺ sensor will have a dominant effect on Ca²⁺ signals.

• Recognizing the impact of the Ca²⁺ sensors used for Ca²⁺ imaging of intracellular Ca²⁺ will allow investigators to use these tools more effectively.