The variations of the intracellular concentration of calcium ion ([Ca2+]i) are at the heart of intracellular signaling, and their imaging is therefore of enormous interest. However, passive [Ca2+]i imaging provides no control over these variations, meaning that a full exploration of the functional consequences of [Ca2+]i changes is difficult to attain. The tools designed so far to modify [Ca2+]i, even qualitatively, suffer drawbacks that undermine their widespread use. Here, we describe an electro-optical technique to quantitatively set [Ca2+]i, in real time and with sub-cellular resolution, using two-photon Ca2+ uncaging and dynamic-clamp. We experimentally demonstrate, on neurons from acute olfactory bulb slices of Long Evans rats, various capabilities of this technique previously difficult to achieve, such as the independent control of the membrane potential and [Ca2+]i variations, the functional knocking-in of user-defined virtual voltage-dependent Ca2+ channels, and the standardization of [Ca2+]i patterns across different cells. Our goal is to lay the groundwork for this technique and establish it as a new and versatile tool for the study of cell signaling.
Citation: Idoux E, Mertz J (2011) Control of Local Intracellular Calcium Concentration with Dynamic-Clamp Controlled 2-Photon Uncaging. PLoS ONE 6(12): e28685. https://doi.org/10.1371/journal.pone.0028685
Editor: Steven Barnes, Dalhousie University, Canada
Received: September 16, 2011; Accepted: November 13, 2011; Published: December 28, 2011
Copyright: © 2011 Idoux, Mertz. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The European Union International Outgoing Fellowship Marie Curie number 221146 (DynaCa-DA) both funded this project and supported EI (http://ec.europa.eu/research/fp7/understanding/marie-curieinbrief/europe-world_en.html). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Variations in intracellular Ca2+ concentration ([Ca2+]i) play a key role in governing the response of virtually all cell types . The importance of [Ca2+]i is particularly marked in neurons where it is instrumental in information processing, plasticity and neurotransmitter exocytosis .
While [Ca2+]i variations are now routinely imaged by one- or two-photon excited fluorescence , , few attempts have been made to design tools to actually control these variations. Examples of such attempts have relied on controlling extracellular Ca2+ concentration while increasing plasma membrane permeability , on using iontophoresis with a sharp electrode , or on uncaging chelated Ca2+ . Because neuronal physiology is critically sensitive to plasma membrane permeability, the first technique is poorly suited to neurons. The second technique has also been shown to modify neuronal electrophysiology  and has the disadvantage that iontophoresis creates a single entry point for Ca2+, in contrast to physiological calcium signals that enter via voltage-gated Ca2+ channels distributed throughout the membrane. Finally, in the third technique, single-photon ultraviolet pulses trigger the photolysis of a Ca2+chelator. Throughout the 90′s, Neher and his group intensively used this to investigate the cellular mechanisms involving Ca2+ at the synapse – and in chromaffin cells . However, this method leads to [Ca2+]i peaks and decays, which differ from the continuous variations experienced by cells under physiological conditions. Moreover, the excitation volume defined by the UV flash is spatially extended, making it difficult to perform quantitative Ca2+ release with high spatial resolution.
The present study describes a new technique that solves many of these problems. Specifically, we have developed a tool to control [Ca2+]i with sufficient temporal and spatial resolution to mimic naturally occurring intracellular Ca2+ signals. We demonstrate the capabilities of this tool by mimicking voltage-dependent Ca2+channels in neurons. The theory of our technique and its experimental design are discussed. We focus on three applications, namely the independent control of the membrane potential and [Ca2+]i variations, the functional knocking-in of user-defined virtual voltage-dependent Ca2+ channels, and the standardization of [Ca2+]i patterns across different cells.
Principles of dynamic two-photon uncaging
Our approach is two-pronged : we spatially confine the Ca2+ release from the chelator DM-nitrophen ,  using two-photon uncaging, and we control its dosage and dynamics in real time by dynamic-clamp . Together, these two techniques provide dynamic two-photon calcium control (DTC) and enable us to liberate ions with similar local concentration and temporal dynamics as would an endogenous voltage-gated Ca2+ channels (see Fig. 1a).
A. Schematic of the optical setup for DTC. Uncaging was performed by a mode-locked Ti:sapph laser beam (734 nm), the power of which was controlled by an electronic-optic modulator (EOM, Conoptics M350-80). The laser focus was scanned in a closed, curvilinear path along the inner membrane surface of the patched neuron. The patch electrode delivers caged Ca2+ (Ca2+-laden DM-nitrophen, 1.51 mM) and fluo-4 (77 µM, fluorescent Ca2+ sensor). ΔF/F is monitored using a 491 nm DPSS laser (Cobolt Calypso) or a 488 nm argon laser (JDS Uniphase), co-aligned with the uncaging laser through dichroic mirror 1 (see Methods). B. Control case with no uncaging: the reference membrane potential V0(t) and fluorescence F0(t) are recorded. C. Application 1: V0(t), the membrane potential sequence recorded in B. , is used as the input command in RTXI to compute the laser power required for the desired Ca2+ uncaging, while the cell is maintained hyperpolarized to prevent endogenous [Ca2+]i variations. V1(t) and F1(t) are recorded by DTC. D. Application 2a: To test for residual endogenous Ca2+ influx, V0(t) is played back into the cell without any uncaging. Since the amplifier is in voltage-clamp mode, V2a(t) is recorded and should be identical to V0(t). E. Application 2b: Similarly, V0(t) is played back, but the uncaging is driven by the measured membrane potential of the neuron (V2b(t)).
Based on the optics in our setup, our two-photon excitation focal volume is approximately 0.5 fL (see Equations 9a and b). This small focal volume can provide targeted Ca2+ release, or it can be scanned within the soma, in the vicinity of the cell membrane. This second option was chosen for our purposes since it more closely mimics physiological trans-membrane Ca2+ entry and reduces the chance of a non-negligible local depletion of cage complex (cf. Discussion). Within the range of laser intensities used for our two-photon photolysis, the number of released Ca2+ ions is proportional to the square of the laser power (cf. Equations 7 and 8 and ref. ).
While two-photon Ca2+ flash uncaging in neurons has been demonstrated in the past, DTC provides analog uncaging that is tightly regulated. Since the number of Ca2+ ions entering through channels is directly proportional to the electrical current flowing through these channels, a dynamic-clamp plug-in determines, in real-time, the laser power required to release this amount of Ca2+. We begin by first programming a Hodgkin-Huxley conductance to model the voltage dependent gating properties of a population of user-defined Ca2+ channels. The dynamic-clamp then uses it to compute the current at the measured membrane potential. Instead of directly injecting this current into the cell via the patch electrode, as is done in the standard implementation of dynamic-clamp, we calculate the number of Ca2+ ions required to bear such a current and adjust our laser power accordingly. In this manner, the influx of Ca2+ is the same as that expected from the Ca2+ channels. Moreover, by sweeping our two-photon excitation along the inner surface of the cell membrane, we closely mimic the expected spatial distribution of the Ca2+ entry.
To verify that DTC can simulate endogenous [Ca2+]i variations, we monitored these with fluorescence imaging using fluo-4 as the Ca2+ indicator and a second laser (Fig. 1A). The similarity between endogenous and photolytic [Ca2+]i was evaluated by first recording spontaneous and/or induced membrane potential variations and their associated fluo-4 fluorescence (subsequently referred to as “F”). We then played back the membrane potential recordings to control Ca2+ uncaging. Finally, the cross-correlation between the endogenous and the photolytically-induced variations of ΔF/F was evaluated at zero time lag. A cross-correlation of 100% means both signals have exactly the same dynamics, whereas 0% means they are independent.
Experimental validation of dynamic two-photon uncaging
To demonstrate the capabilities of DTC, we concentrate here on three applications.
First, we demonstrate that DTC can effectively dissociate [Ca2+]i and membrane potential variations. To date, only Ca2+channels blockers have been effective at decoupling these by blocking the Ca2+ influx altogether for a particular sub-population of channels, effectively preventing the Ca2+ influx from depending on membrane potential . We show that DTC enables just the opposite: controlled variations of [Ca2+]i while the membrane potential is held constant. To begin, the membrane potential (V0(t)) and fluorescence variations (F0(t)) are recorded when the neuron is spiking (Fig. 1B and 2A2). These recordings serve as a reference. The neuron is then hyperpolarized below the activation threshold of L-type Ca2+ channels, causing these endogenous channels to close and the Ca2+-associated fluorescence signal to flatline (2A2, F1). As expected, F0 and F1 are not significantly correlated and XC0,1(0) is down to 8%±15% (n = 56). When DTC is activated (Fig. 1C and 2A3), the uncaging restores the [Ca2+]i variations with dynamics significantly correlated to the endogenous ones: XC0,1(0) = 88%±8% (n = 56). The cross-correlations with and without uncaging are also verified to be significantly different from each other (p<0.001). Thus, DTC effectively simulates the endogenous L-type Ca2+ channels even though the membrane potential is held below threshold.
A. Application 1, demonstrating the uncoupling of [Ca2+]i and membrane potential variations. Blue, control: F0(t)/V0(t); black, hyperpolarization without uncaging: F1(t)/V1(t); magenta, hyperpolarization with uncaging: F1(t)/V1(t). A1 Membrane potential variations. A2 Ca2+-induced fluorescence variations (F0(t) and F1(t)) are not significantly correlated. A3 Same, but with uncaging. B. Application 2, demonstrating Ca2+ channel function restoration. B1 Membrane potential variations measured and played back -- the three superimposed traces are identical. B2 Fluorescence traces for control (blue F0(t)) and without uncaging (black, F2a(t)) -- these were not significantly correlated. B3 Same, but with uncaging (magenta F2b(t)). C Application 3: standardization of [Ca2+]i variations across neurons. F0(t) and V0(t) were recorded in one cell while the uncaging (F3(t)) was performed in another that was kept hyperpolarized (V3(t)). The same parameters for the conductance model were used for C1-4 and endogenous dynamics were reliably reproduced in all cases (see Table 2).
In a second application, the endogenous Ca2+ channels are blocked pharmacologically and DTC is used to functionally reinstate the [Ca2+]i variations. To do this, we first record a reference, as before (Fig. 1B). Then, 40 µM of verapamil or nifedipine are added to the bath for 15 minutes, thus blocking L-type Ca2+ channels . The reference membrane potential variations are then played back as the voltage command in voltage clamp (Fig. 1D), while the resultant fluorescence is monitored. Without DTC, the endogenous fluorescence variations are greatly reduced or even completely canceled, as expected (Fig. 2B2, XC0,2a(0) = 32%±20%, n = 17). With DTC, the [Ca2+]i variations are restored to their original level (Fig. 1E and 2B3 XC0,2b(0) = 86%±13%, n = 17, significantly higher than playback without uncaging, p<0.001). That is, DTC effectively re-introduces “virtual” L-type Ca2+ channels in the soma to replace those that were pharmacologically blocked. Interestingly, uncaging while the Ca2+ channels are blocked leads to similar results as uncaging under hyperpolarization but without blockers (i.e. experiment 1), as shown by the high cross-correlation for this subset (XC0,1(0) = 91%±4%, n = 17, p = 0.243).
In a third application, DTC is used to standardize the [Ca2+]i variations between different cells. The procedure is the same as for application 1 (Fig. 1C), except that the membrane potential variations recorded as a reference in one neuron (neuron0) are played back in a different neuron (neuron1) that is hyperpolarized. Because the cytoplasms of both neurons are dialyzed by the content of the patch pipette, the Ca2+ buffers inside both neurons are also quantitatively and qualitatively similar: the playback of the membrane potential of neuron0 thus induces the same [Ca2+]i variations in both neuron0 and neuron1 (Fig. 2C1 XC0,3(0) = 93%±6%, n = 108).
Finally, to verify the robustness of the L-type Ca2+ conductance model and its ability to reproduce arbitrary [Ca2+]i variations, the same model parameters were used to reproduce up to four different sequences in a given neuron. The measured [Ca2+]i variations all reached similar levels (e.g. Fig. 2c1 to 2c4 XC0,3(0) = 91%±5% n = 26, 89%±6% n = 26, 87%±9% n = 13 and 91%±3% n = 2, Friedman test, p = 0.194). Reproducibility was further confirmed with back to back sequences, differences being on average 0%±2%, n = 401, p = 0.491 (see Methods).
In summary, DTC provides a dynamic control of [Ca2+]i with high enough spatio-temporal resolution to simulate the cellular variations of [Ca2+]i caused by endogenous channels. It relies on the regulated release of Ca2+ by uncaging with two-photon laser, whose power is modulated by a dynamic-clamp plug-in. To our knowledge, such precise control of [Ca2+]i dynamics has never been achieved before. DTC therefore provides a new and versatile tool for the study of cellular function, with capabilities that include the dissociation of membrane potential from [Ca2+]i variations and the standardization of [Ca2+]i patterns through the virtual knock-in of calcium-specific, user-defined conductances.
Advantages of the technique
DTC is based on two-photon uncaging, which has several advantages: two-photon uncaging both defines a 3D spatially confined Ca2+ release volume  as well as increases penetration depth in tissue and minimizes photodamage inflicted on non-target cells. Moreover, it offers a wealth of possibilities: in application 2 presented above, DTC not only functionally reinstates pharmacologically blocked channels but also provides full control over both the temporal dynamics of the [Ca2+]i variations, and their spatial location. Whereas blockers act indiscriminately in both the soma and dendrites, DTC can be applied more selectively by targeting the laser focus to either or both, enabling the user to gauge the effect of subcellular location of Ca2+ influx. Thus, DTC provides the tool to test hypotheses related to key properties of Ca2+ signaling, such as the functional importance of subcellular domains, or the impact of slowing or speeding the rate of Ca2+ influx.
Choice of the Ca2+ cage and of the Ca2+ fluorescent dye
While several cages have been reported in the literature , only two of them are currently commercially available: DM-nitrophen and NP-EGTA. Because DM-nitrophen is EDTA based while NP-EGTA is EGTA based, their chelating properties are slightly different. The EGTA core of NP-EGTA is highly selective at binding Ca2+ over Mg2+ (KdCa 80 nM vs. KdMg 9 mM), unlike DM-nitrophen (5 nM vs. 25 µM). Therefore, with physiological concentrations of Mg2+ in the pipette (∼0.5 to 1 mM ), Ca2+-laden DM-nitrophen exchanges Ca2+ for Mg2+, leading to an increase in the resting [Ca2+]i from 100 nM to 5 µM and a mixture of Ca2+ and Mg2+ uncaging . These problems can be avoided with the use of NP-EGTA or, as was proposed before  and done here, by removing Mg2+ from the internal pipette solution. Then, DM-nitrophen offers numerous advantages over NP-EGTA: it has a higher affinity for Ca2+ when intact (5 vs. 80 nM) and a lower affinity when photolyzed (3 vs. 1 mM), and it exhibits a higher two-photon cross-section (0.01 vs. 0.001 GM). The lower affinity allows DM-nitrophen to be loaded above 95% for [Ca2+]i ≈100 nM, while under the same conditions (same [Ca2+]i and same amount of free Ca2+ buffer), NP-EGTA could only be loaded slightly above 50% . Moreover, NP-EGTA would be restricted to very low concentration (around 200 µM) to allow signal monitoring with 100 µM fluo-4. The combined effect of low concentration of caged NP-EGTA and lower two-photon cross-section means that NP-EGTA would require more than 10 times more optical power (536 mW instead of 47 mW to attain 1000 pA Ca2+ current), which would be difficult to achieve with our current setup, in addition to being potentially detrimental to the preparation. A possible solution is to increase both NP-EGTA and fluo-4 concentrations, though this may be ill-advised for two reasons. First, the NP-EGTA would still be only half loaded, which may cause the Ca2+ to be released only transiently rather than continuously, as freshly released Ca2+ is likely to be quickly recaptured by non-photolyzed neighboring empty cages . An increase in NP-EGTA and fluo-4 concentrations would also increase the concentration of free Ca2+ buffer, which might distort the endogenous Ca2+ signal , . For example, to lower the laser power from 536 to 107 mW would entail about a 3.3 mM increase in free Ca2+ buffers (free NP-EGTA + free fluo-4). Thus, we conclude that whenever Mg2+ is not crucial for the study in question (the case here), then DM-nitrophen seems to be a better choice than NP-EGTA for the application of DTC.
Local depletion is yet another risk to be aware of when uncaging Ca2+. As noted above, only 47 mW of laser power is required to attain a 1000 pA Ca2+-current when using DM-nitrophen, while a typical current in mitral cells is usually less than 150 pA . To reproduce such a typical current between sequential dynamic-clamp updates would locally consume about 10% of the caged Ca2+ pool in a static release volume, i.e. if the laser were kept parked at one spot. Because the laser is instead scanned along the inner membrane surface, the swept release volume is about 90 times larger than the focal volume (about 40 fL for a 15 µm diameter cell). Therefore, the local depletion is about 0.1% and can be readily replenished by diffusion from non-illuminated neighboring regions, as experimentally verified when comparing back to back sequences.
As noted above, our Ca2+ reporter is fluo-4. This was chosen because it provides several advantages compared to other available Ca2+ reporters. First, with its absorption peak at 494 nm , it is almost optimally excited by our 491 nm CW laser, while not excited by the two-photon laser (734 nm). It is furthermore compatible with DM-nitrophen since the excitation wavelength of nm also prevents unwanted uncaging since it does not photolyze DM-nitrophen . Second, its Kd (345 nM) makes fluo-4 one of the best choices to monitor both single elementary events (such as the [Ca2+]i increase after an action potential) and larger trends in Ca2+ variations , . Third, because we use a cell-impermeant version of fluo-4, it is likely to be homogeneously distributed throughout the cytoplasm, with little inhomogeneity due to compartmentalization .
Limitations of the technique
Since DTC uncaging is light-based, its spatial resolution cannot be better than the diffraction limit. The water-immersion objective used here has a numerical aperture of 0.95, which is close to the maximum commercially available while remaining compatible with patch clamp (i.e. with a working distance over 2 mm). With our setup (see Table 1), the release volume is a prolate spheroid with equatorial and axial radii of 0.4 and 1.4 µm respectively. The size of our focal volume is thus on the order of the femtoliter. This is clearly much larger than the volume of a Ca2+ spark or sparklet from a single channel , , meaning that DTC cannot replicate the spatial confinement of a single channel Ca2+ influx. Nevertheless, when considering the Ca2+ influx produced by a local population of channels (the channels distributed in the plasma membrane of the soma in our case), their averaged signals over the time scale of interest can be faithfully reproduced, as experimentally demonstrated here. Working with a population of channels allows us to use the deterministic Hodgkin-Huxley formalism to model this global conductance. If needed, a random component can also be introduced to mimic population stochastic response .
How fast can our DTC be pushed? In this set of experiments, we use a refresh rate of 10 kHz for the dynamic-clamp. This is about one order of magnitude faster than the fastest time constant of the conductance we mimic (cf. Table 2) and is thus sufficient for our applications. Faster refresh rates could easily be achieved with our computer, however one must bear in mind that the photolysis mechanism itself cannot be sped up: the maximum release rate for DM-nitrophen is about 40 kHz, whereas it reaches about 70 kHz for NP-EGTA .
Finaly, we emphasize here that DTC should not be regarded as a “calcium-clamp” for two reasons. First, a proper clamp would involve a feedback mechanism to maintain [Ca2+]i at a set value. To mimic voltage-dependent Ca2+ channels, DTC does not rely on feedback, and instead is based only on a feed-forward mechanism, as illustrated in figure 1C. That is, the actual value of [Ca2+]i is never used to compute how much Ca2+ is to be released. Second, a proper clamp would involve a way to actively modulate the Ca2+ concentration both upward and downward. In our case, DTC only achieves the upward modulation: the downward modulation is ensured by the cell Ca2+ clearance mechanisms, as is the case when Ca2+ enters the cell through voltage dependent Ca2+ channels. While in principle it may be possible to introduce an active Ca2+ scavenger in addition to our Ca2+ chelator, the only photoactivatable scavenger commercially available is, to our knowledge, Diazo-2 (available from Invitrogen ), whose activation spectrum overlaps that of both DM-nitrophen and NP-EGTA, thus precluding an independent control. For these reasons, DTC cannot be regarded as a bona-fide “calcium-clamp”.
Extension of the tool to other Ca2+ sources and to more caged compounds
It should be noted that DTC can, in principle, be applied to quantitatively control the release of any caged compound with adequate two-photon cross-section. For example, another mechanism through which Ca2+ enters the cytoplasm is CICR (Ca2+-induced Ca2+ release) and DTC can be programmed to mimic this as well, provided an appropriate CICR model is implemented. In this case, the RTXI input would not be the membrane potential but rather a measure of the actual [Ca2+]i , for example obtained by fluorescence monitoring with a Ca2+-sensor. With the appropriate Ca2+ cage, such an implementation of DTC would be all-optical and allow simultaneous multicellular control. Finally, DTC need not be restricted to caged Ca2+. For example, caged neurotransmitters can be used to reproduce localized synaptic release with controlled dynamics. Alternatively, caged-second messengers such as inositol trisphosphate can be used to study second messenger dynamics in sub-cellular compartments.
In short, DTC is a general technique with many potential ramifications. Our goal in this study is to lay the groundwork for this technique, with the hope of establishing it as a new tool for the study of cellular dynamics. Specifically, we have confined our proof of concept to the modulation of Ca2+ in neurons, where we show that the combination of a dynamic-clamp with two-photon uncaging enables the release of caged compounds with an unprecedented degree of control and spatiotemporal resolution. This technique is robust and easy to implement with standard two-photon microscopes, making it attractive as a general tool for the study chemical signaling in cells.
Materials and Methods
The following protocol was approved by the Boston University Institutional Animal Care and Use Committee (IACUC Protocols # 07–014 and 10–021) and everything was done to minimize the number of animals and their suffering. We prepared parasagittal olfactory bulb slices (350 µm) from weaned sub-adult Long-Evans rats (age: P21 to P41, median P32). Briefly, after a deep, isofluorane-induced anesthesia, confirmed by absence of reaction to toe pinching, the animal was decapitated and the head submerged in ice-cold sucrose aCSF (see composition below). The olfactory bulbs were removed, sliced with a Vibratome 1500 (formerly sold by Vibratome, now Leica), and placed in regular aCSF bubbled with 95% O2-5% CO2 to attain a pH of 7.3–7.4. Slices were kept at room temperature until individually transferred into the recording chamber where experiments were carried out. One neuron was patched per slice, with a borosilicate glass pipette pulled on a Sutter P97 puller from BF120-69-10 capillaries (3–6 MΩ).
Both the room temperature incubation and subsequent recording at 31°C used artificial cerebro-spinal fluid (aCSF) composed of (in mM): 120 NaCl, 25 NaHCO3, 1 NaH2PO4, 2.5 KCl, 10 glucose, 2 MgCl2, 1 CaCl2. The sucrose aCSF used for the dissection and slicing is the same, but with the 120 mM NaCl replaced by 240 mM sucrose and the MgCl2 increased to 3 mM to compensate for the removal of CaCl2. Both solutions have the same osmolality (∼295 mosm/kg, Wescor Vapro 5520) and pH (7.3–7.4). L-type calcium channels  were blocked by addition in the aCSF of 40 µM of verapamil or nifedipine while the cell was depolarized so that it spiked for 15 minutes. The solution used to fill the pipette was composed of (in mM) 135 K-gluconate, 5 KCl, 7.5 HEPES, 2.5 NaHEPES, 4 K2ATP, 0.47 NaxGTP, 1.56 CaCl2, 1.585 DM-nitrophen and 0.1 K5fluo-4. Therefore, the concentrations of loaded DM-nitrophen, free Ca2+, and free fluo-4 are, respectively 1.51 mM, 104 nM and 77 µM.
All chemicals were obtained from Sigma-Aldrich, with the exception of DM-nitrophen and fluo-4, obtained from Invitrogen.
Stimulation methods: playback
To ensure that the Ca2+ fluorescence traces could be compared between runs and conditions, the membrane potential variations had to be kept identical between different traces. Therefore, our custom-made stimulation and recording program (Matlab and Labview) included a playback mode. This mode outputs previously-recorded sequences of membrane potential variations through a National Instruments card (PCI-6221 with a BNC-2110) and sends them to either the amplifier or dynamic-clamp setup (see Fig. 1).
Our upright two-photon microscope was custom-built and described elsewhere . The rest of the setup is described in Fig. 1A. The ΔF/F was calculated from the fluorescence signal (F) recorded by a PMT (Hamamatsu H7422PA-40).
To compare the dynamics of two fluorescent traces of N points (F0 and F1), we evaluated their cross-correlation  at zero lag, XC0,1 (τ = 0). If µ0 and σ0 are respectively the mean and standard deviation of F0 (similarly for F1), then XC0,1 (τ) is given by
Furthermore, two traces are significantly correlated with a 5% threshold , only if
Unless stated otherwise, we used only significantly correlated traces. To test for reproducibility between sequences, some sequences (n = 401) were repeated back to back, separated by a 1s interval. Since the differences were not statistically significant (p = 0.49), only the highest of the two cross-correlation coefficients was kept for subsequent analysis.
Dynamic-clamp and model for uncaging dynamics
Experiments were performed using the Real-Time eXperiment Interface (RTXI; www.rtxi.org), connected through a National Instruments card PCI-6040E and a plug-in with a model of the L-type Ca2+ conductance. This samples the membrane potential V(t) and calculates the value of the conductance every 100 µs (10 kHz refresh rate). The corresponding current, namely ICa(V(t)), is then computed based on V(t) and a locally applied Ohm's law. Since an ionic current corresponds to a number of charges flowing through a surface per unit of time, we thus determine the number of calcium ions (ΔNCa) required to produce such a current, for a given time step.
The time step (ΔtDC = 100 µs) is set to be the sampling period of the voltage by the dynamic-clamp software, which is much faster than typical time constants involved in the dynamics of the Ca2+conductances (see Table 2). The surface in question is that of the plasma membrane of the soma, however it could be chosen elsewhere, depending on the experiment (e.g. it could be set to correspond to a portion of dendrite, subcellular compartment, etc.).
The above provides us with the first relationship between the current and the number of Ca2+ ions:
where ce is the elementary charge and V(t) the membrane potential.
The laser in our system is operated at an intensity IL far below the onset of pulse saturation , . The rate of release of Ca2+ions (RCa(IL,x,y,z)) is therefore linearly proportional to the actual caged Ca2+available ([CCa2+]0), at any point in space (x,y,z) . That is,
where α, the photolysis rate, is defined as
where σ2 is the two-photon photolysis cross-section and we have adopted a 3D Gaussian model for the focal volume, with lateral and axial waists wr and wz respectively. We further adopt the notation. In writing Eq. 2, we have made the assumption that [CCa2+]0 is roughly a constant, i.e. not depleted by the uncaging process itself. This assumption is justified in the discussion and experimentally confirmed.
The principle of DTC is to match ΔNCa,uncaging and ΔNCa,conductance during the interval ΔtDC. The total number of Ca2+ ions (ΔNCa) liberated in the release volume (Vr) can be expressed as
where NA, Avogadro's number, converts the result of the volume integral from moles into number of ions.
Since is directly proportional to ICa, the current calculated by the dynamic-clamp, we then control the intensity of the laser beam to reproduce the Ca2+ variations in the cell in real time. The mean of the squared intensity is linked to the square of the mean intensity (), through the “two-photon advantage factor” resulting from the pulsed nature of the laser beam used in the experiments , :
where f is the laser repetition rate, tp is the pulse duration and gp(2) is the second-order temporal coherence factor.
An electro-optic modulator (EOM) controls the laser power, which, for wavelength λ, is given by
since the area of the beam at the focal plane is . From Equation 7, we observe that the only time varying parameter is ICa(V(t)) (see conductance model below), the rest being just constant scaling factors.
Finally, the input voltage range of the EOM (0 to Vmax) is adjusted to provide a linear control of the optical power (0 to Pmax). The voltage output of RTXI then becomes
where η is he power loss throughout the system, which was evaluated to be 75% for our setup. Specifically, the 326 mW power exiting the laser was reduced to 82 mW at the focal plane because of losses mostly due to the EOM and the dichroic mirrors in our setup (cf. Fig.1 1: DMLP567 from ThorLabs, 2: Custom-made by Chroma 540DCRB, reflectance band: 540±40, transmittance 94–95% at 488/491 nm and 88–92% in the 734±4 nm band).
wr and wz, the radial and axial waists of the release volume, were calculated using :
with NA being the objective numerical aperture and n the index of refraction of water.
See Table 1 for parameter values.
To program the plug-in that models the L-type calcium conductance, we adopted an approach similar to one used previously , using standard Hodgkin-Huxley equations , with parameters adapted from Bhalla and Bower . The equations are summarized as
where ECa2+ is the reversal potential of Ca2+ and g the conductance, expressed as
with a(V(t)) being the activation gate, i(V(t)) the inactivation gate, and gmax the maximal value of the conductance. The gates a and i are modeled as typical Hodgkin-Huxley gates:
See Table 2 for parameters.
To determine whether the differences observed in XC0,1(0) between conditions were statistically significant, we used the Wilcoxon signed rank test, unless stated otherwise. This non-parametric test was chosen since some sets were not normally distributed (as tested with a Shapiro-Wilk test). However, the medians and means were close enough to report the results as mean ± S.D as usual.
The authors would like to thank J.C. Bettencourt, R.J. Lin for their help with RTXI and K.P. Lillis for his help with the home-built microscope.
Conceived and designed the experiments: EI. Performed the experiments: EI. Analyzed the data: EI. Contributed reagents/materials/analysis tools: EI. Wrote the paper: EI JM. Built the microscope: JM.
- 1. Nowycky MC, Thomas AP (2002) Intracellular calcium signaling. J Cell Sci 115: 3715–3716.
- 2. Benarroch EE (2010) Neuronal voltage-gated calcium channels: brief overview of their function and clinical implications in neurology. Neurology 74: 1310–1315.
- 3. MacLean JN, Yuste R (2009) Imaging Action Potentials with Calcium Indicators. Cold Spring Harb Protoc. 2009. pdb.prot5316 p. https://doi.org/10.1101/pdb.prot5316
- 4. Adams SR (2010) How calcium indicators work. Cold Spring Harb Protoc. 2010. pdb.top70. pdb.top70 p. https://doi.org/10.1101/pdb.top70
- 5. Dolmetsch RE, Xu K, Lewis RS (1998) Calcium oscillations increase the efficiency and specificity of gene expression. Nature 392: 933–6.
- 6. Belan PV, Kostyuk PG, Snitsarev VA, Tepikin AV (1993) Calcium clamp in single nerve cells. Cell Calcium 14: 419–25.
- 7. Zucker RS (1993) The calcium concentration clamp: spikes and reversible pulses using the photolabile chelator DM-nitrophen. Cell Calcium 14: 87–100.
- 8. Li WC, Soffe SR, Roberts A (2004) A direct comparison of whole cell patch and sharp electrodes by simultaneous recording from single spinal neurons in frog tadpoles. J Neurophysiol 92: 380–6.
- 9. Sørensen JB (2004) Formation, stabilisation and fusion of the readily releasable pool of secretory vesicles. Pflugers Arch 448: 347–362.
- 10. Zucker RS (1993) Calcium and transmitter release. J Physiol. Paris 87: 25–36.
- 11. Neher E, Sakaba T (2008) Multiple roles of calcium ions in the regulation of neurotransmitter release. Neuron 59: 861–872.
- 12. Neher E (2006) A comparison between exocytic control mechanisms in adrenal chromaffin cells and a glutamatergic synapse. Pflugers Arch 453: 261–268.
- 13. Kaplan JH, Ellis-Davies GC (1988) Photolabile chelators for the rapid photorelease of divalent cations. Proc. Natl. Acad. Sci. U.S.A 85: 6571–6575.
- 14. Ellis-Davies GCR (2008) Neurobiology with Caged Calcium. Chemical Reviews 108: 1603–1613.
- 15. Lin RJ, Bettencourt J, White JA, Christini DJ, Butera RJ (2010) Real-time Experiment Interface for biological control applications. Conf Proc IEEE Eng Med Biol Soc 1: 4160–4163.
- 16. Brown EB, Shear JB, Adams SR, Tsien RY, Webb WW (1999) Photolysis of caged calcium in femtoliter volumes using two-photon excitation. Biophys J 76: 489–499.
- 17. Catterall WA, Perez-Reyes E, Snutch TP, Striessnig J (2005) International Union of Pharmacology. XLVIII. Nomenclature and structure-function relationships of voltage-gated calcium channels. Pharmacol. Rev 57: 411–425.
- 18. Wang X, McKenzie JS, Kemm RE (1996) Whole cell calcium currents in acutely isolated olfactory bulb output neurons of the rat. J Neurophysiol 75: 1138–1151.
- 19. Mertz J (1998) Molecular photodynamics involved in multi-photon excitation fluorescence microscopy. The European Physical Journal D - Atomic, Molecular and Optical Physics 3: 53–66.
- 20. Li-Smerin Y, Levitan ES, Johnson JW (2001) Free intracellular Mg2+ concentration and inhibition of NMDA responses in cultured rat neurons. The Journal of Physiology 533: 729–743.
- 21. Ellis-Davies GCR (2006) DM-nitrophen AM is caged magnesium. Cell Calcium 39: 471–473.
- 22. Neher E, Zucker RS (1993) Multiple calcium-dependent processes related to secretion in bovine chromaffin cells. Neuron 10: 21–30.
- 23. McCray JA, Fidler-Lim N, Ellis-Davies GC, Kaplan JH (1992) Rate of release of Ca2+ following laser photolysis of the DM-nitrophen-Ca2+ complex. Biochemistry 31: 8856–8861.
- 24. Neher E (2005) Some Quantitative Aspect of Calcium Fluorimetry (Chapter 30). In: Yuste R, Konnerth A, editors. Imaging in Neuroscience and Development: A laboratory manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press.
- 25. Higley MJ, Sabatini BL (2008) Calcium signaling in dendrites and spines: practical and functional considerations. Neuron 59: 902–913.
- 26. Paredes RM, Etzler JC, Watts LT, Zheng W, Lechleiter JD (2008) Chemical calcium indicators. Methods 46: 143–151.
- 27. Thomas D, Tovey SC, Collins TJ, Bootman MD, Berridge MJ, et al. (2000) A comparison of fluorescent Ca2+ indicator properties and their use in measuring elementary and global Ca2+ signals. Cell Calcium 28: 213–223.
- 28. Cheng H, Lederer WJ (2008) Calcium Sparks. Physiological Reviews 88: 1491–1545.
- 29. Baylor SM (2005) Calcium sparks in skeletal muscle fibers. Cell Calcium 37: 513–530.
- 30. Dorval AD, White JA (2005) Channel noise is essential for perithreshold oscillations in entorhinal stellate neurons. J Neurosci 25: 10025–8.
- 31. Adams SR, Kao JPY, Tsien RY (1989) Biologically useful chelators that take up calcium(2+) upon illumination. Journal of the American Chemical Society 111: 7957–7968.
- 32. Lillis KP, Eng A, White JA, Mertz J (2008) Two-photon imaging of spatially extended neuronal network dynamics with high temporal resolution. J Neurosci Methods 172: 178–84.
- 33. Netoff TI, Schiff SJ (2002) Decreased Neuronal Synchronization during Experimental Seizures. J Neurosci 22: 7297–7307.
- 34. Xu C, Webb WW (1997) Multiphoton excitation of molecular Fluorophores and Non Linear Microscopy. Topics in Fluorescence Spectroscopy: Nonlinear and two-photon-induced fluorescence. Springer. 564 p.
- 35. Idoux E, Eugène D, Chambaz A, Magnani C, White JA, et al. (2008) Control of neuronal persistent activity by voltage-dependent dendritic properties. J. Neurophysiol 100: 1278–1286.
- 36. Hodgkin AL, Huxley AF (1952) A quantitative description of membrane current and its application to conduction and excitation in nerve. J Physiol. (Lond.) 117: 500–544.
- 37. Bhalla US, Bower JM (1993) Exploring parameter space in detailed single neuron models: simulations of the mitral and granule cells of the olfactory bulb. J Neurophysiol 69: 1948–1965.