2.1.

Plasmid Construction, Protein Expression, and Purification

Three residues in circularly permuted (cp) 173Venus were replaced using the approach previously described by Sawano and Miyawaki,¹¹ to obtain (S220E, S31D, Q33E)-cp173Venus. The sequence encoding this Venus variant was amplified using primers containing XhoI (forward) and EcoRI (reverse) restriction sites, and the open reading frame (ORF) was cloned into ${\mathrm{pRSET}}_{\mathrm{B}}$ vector (Invitrogen, Carlsbad, California). ORFs of reference FPs were amplified using primers containing BamHI (forward) and XhoI (reverse) restriction sites and ligated in the vector containing the (S220E, S31D, Q33E)-cp173Venus between the XhoI restriction site and an upstream BamHI restriction site [Fig. 1(b)]. In the final indicator (see below), spontaneous mutations—G4S [enhanced green fluorescent protein (EGFP) 177S], V89I (EGFP 18V), and Q142M (EGFP 72Q)—were found in the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus moiety, so additional indicator variants were generated: (1) with reversed spontaneous mutations, (2) a variant holding all the acidic residues present in CatchER (F52E, T54E in the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus, in addition to those shown in Table 1),¹² and (3) the reversed mutant holding only the five negatively charged residues of CatchER in the original cp173Venus moiety. All these mutations were introduced following the approach proposed by Sawano and Miyawaki¹¹ and the subsequent construction process was identical to that described above.

Fig. 1

Working principle and schematic representation of MagIC. ${\mathrm{Mg}}^{2+}$ concentration changes are reported by the fluorescence intensity of ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive cp173Venus variant, which is then compared to the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-insensitive (constant) reference fluorescence intensity of mCherry. (a) Working principle and schematic representation of spectral changes of the ratiometric non-Förster resonance energy transfer (FRET)-based indicator composed of ${\mathrm{Mg}}^{2+}$-sensitive and reference moieties. ${\mathrm{Mg}}^{2+}$ increase is indicated by the arrow. (b) Structure of the indicator gene featuring mutations introduced in cp173Venus.

Table 1

Amino acid replacements in the cp173Venus variant used in MagIC and corresponding mutations in EGFP-based CatchER.12

Protein expression was carried out using the JM109 (DE3) Escherichia coli strain. Culturing was performed in $1\times$ lysogeny broth medium at 23°C for 60 h with continuous shaking at 150 rpm. The bacterial pellet obtained after centrifugation was resuspended in phosphate-buffered saline (PBS) and mechanically lysed. The cleared lysate was purified using a Ni-NTA column (Qiagen, Hilden, Germany) and then gel-filtered in a PD-10 column (GE-Healthcare, Uppsala, Sweden) for replacement of the buffer with 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.4.

For expression in mammalian cells, the indicator’s ORF was subcloned into the pcDNA3.2 vector (Life Technologies). For organelle targeting, subcloning was performed into modified pcDNA3.2 vectors containing either the human ornithine transcarbamylase signal sequence (for mitochondrial targeting)¹³ or the calreticulin signal sequence/KDEL motif [for endoplasmic reticulum (ER) targeting].

2.2.

Seminative Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis of Purified Protein Samples or HeLa Cells Lysate

The purpose of seminative sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS PAGE) was to conduct the separation of the proteins contained in the sample, preserving, at the same time, the ability of FPs to emit light when excited. For this purpose, the procedure for SDS PAGE did not include the step of heat treatment of the sample to avoid the denaturation of all contained proteins. Seminative SDS PAGE was carried either for the purified protein sample of MagIC or for the lysate of HeLa cells expressing MagIC in the cytosol. The protein sample was prepared by dilution with sample buffer containing 2% SDS, 62.5 mM Tris–HCl (pH 6.8), 5% (v/v) $\beta$-mercaptoethanol, 10% glycerol (v/v) and bromophenol blue. The lysate of HeLa expressing cytosolic MagIC was prepared as follows: cells grown on a 10-cm plastic plate 24 h after transfection (see Sec. 2.5) were washed with 4 mL ice cold PBS, and then detached by scrapping and resuspended in 2 mL ice-cold PBS for washing. After a centrifugation step ($1000\times \mathrm{g}$, 10 min), the supernatant was removed and cells were resuspended in $100\mu \mathrm{L}$ of the sample buffer described above and applied directly to the polyacrylamide gel. The separation (a variant of this procedure is provided in Ref. 14) was performed using 12% separating polyacrylamide gel in the presence of 10% SDS.

2.3.

Spectral Characterization of the Mg²⁺/Ca²⁺-Sensitive Venus Fusion Constructions with Reference FPs

For the indicator variants (containing different reference FPs), characterization of the ${\mathrm{Mg}}^{2+}$-sensitivity of the absorption, excitation, and emission spectra was conducted. For this, protein samples diluted in 20 mM HEPES-KOH buffer (pH 7.2) were analyzed in a V-630 Bio spectrophotometer (JASCO; for absorption measurements), in a F-2500 fluorescence spectrophotometer (Hitachi, Japan; for excitation spectra acquisition) or in F7000 fluorescence spectrophotometer (Hitachi; for emission spectra acquisition) either in the absence or presence of 5 mM ${\mathrm{Mg}}^{2+}$.

Absorption spectra were acquired in the 450 to 600 nm range.

Excitation spectra for each of the constructions were acquired separately for the Venus variant (monitored wavelength corresponding to its emission maximum, 530 nm) and the corresponding reference FP (monitored at the respective emission maximum of the reference FP). These monitored wavelengths were, specifically, 610 nm for mCherry, 633 nm for mKate2, 563 nm for mKOκ, and 565 nm for the nonphotoconverted PSmOrange. After these separate measurements (for Venus and the reference FP) were conducted in the absence of ${\mathrm{Mg}}^{2+}$, 5 mM ${\mathrm{Mg}}^{2+}$ were added into the same cuvette and the measurements were repeated.

Similarly, emission measurements of the PSmOrange-[${\mathrm{Mg}}^{2+}$-sensitive cp173Venus], mKOκ-[${\mathrm{Mg}}^{2+}$-sensitive cp173Venus] and mKate2-[${\mathrm{Mg}}^{2+}$-sensitive cp173Venus] constructions were conducted. In this case, the Venus variant fluorescence and the emission of the reference FP were acquired using separate excitation. In all constructions, the ${\mathrm{Mg}}^{2+}$-sensitive Venus was excited with 516-nm light (bandwidth 5 nm). Emissions of the corresponding reference FPs were obtained by excitation with the following wavelengths: 548 nm for PSmOrange, 551 nm for mKOκ, and 588 nm for mKate2. In detail, characterization of the ${\mathrm{Mg}}^{2+}$-sensitivity of the emission spectrum of mCherry-[${\mathrm{Mg}}^{2+}$-sensitive cp173Venus] construction (MagIC) is provided in Sec. 2.4.

2.4.

In Vitro Analysis of MagIC

Affinity and selectivity were determined by separate excitation of the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus variant (excitation 516 nm) and mCherry (excitation 587 nm) in an F-7000 fluorescence spectrophotometer (Hitachi).

The indicator affinity was determined using serial dilutions of ${\mathrm{MgCl}}_2$ in a background of 100 mM KCl and 10 mM MOPS (pH 7.4, KOH). Selectivity (as the ability to displace ${\mathrm{Mg}}^{2+}$ at physiological concentrations from the created binding site) was assayed in solutions with 100 mM KCl and 1 mM ${\mathrm{MgCl}}_2$, 10 mM MOPS (50 mM for the case of polyamines) (pH 7.4) and the ion of interest and then compared with a similar solution without that ion. pH sensitivity of the indicator was analyzed in solutions containing either 10 mM MES–HEPES (pH range 5.5 to 8.0) or Tris (pH 8.3 to 9.0) in a background of 100 mM KCl and 1 mM ${\mathrm{MgCl}}_2$.

2.5.

Cell Culture and Transfection

HeLa cells were cultured at 37°C in Dulbecco’s modified Eagle’s medium (DMEM, Sigma) supplemented with 10% heat-inactivated fetal bovine serum (FBS, Biowest) in a 5% ${\mathrm{CO}}_2$ atmosphere.

Pituitary cells of the GH3 line (ATCC CCL-82.1) were maintained in DMEM/F12 + GlutaMAX (Gibco, Life Technologies) supplemented with 2.5% FBS (Biowest) and 15% heat inactivated horse serum (Gibco, Life Technologies) in 5% ${\mathrm{CO}}_2$ atmosphere.

Cell transfection was carried as in the “${\mathrm{Ca}}^{2+}$-phosphate” method¹⁵ or with Lipofectamine 2000 (Life Technologies) following the manufacturer’s protocol. Simultaneous expression of different indicators was achieved by cotransfection with a $1:1$ mixture of plasmids containing the indicators.

2.6.

General Imaging Conditions

Cell imaging was performed 24 to 48 h after transfection using a Nikon Ti Eclipse confocal microscope equipped with $60\times$ oil immersion lens (NA: 1.4) and a Nikon A1 imaging system. Experiments, if not otherwise specified, were conducted at room temperature (around 25°C).

2.7.

Imaging with Indicators in Cells

In the present work, three genetically encoded indicators were used: MagIC (or the negative control mCherry-cp173Venus), B-GECO1—a genetically encoded ${\mathrm{Ca}}^{2+}$ indicator [$K_d({\mathrm{Ca}}^{2+})=164\mathrm{nM}$],¹⁶ and pHluorin—a genetically encoded pH indicator ($\mathrm{p}{K}_a$ around 7.0).¹⁷

In experiments with cells expressing MagIC or the negative control mCherry-cp173Venus, the Venus variants were excited with a 488-nm laser and their emissions were collected in the 500 to 550 nm range. mCherry excitation was performed with a 561 nm laser and its emission was collected in the 570 to 620 nm range. When MagIC was coexpressed with B-GECO1,¹⁶ a genetically encoded ${\mathrm{Ca}}^{2+}$ indicator with excitation and emission maxima at 378 and 446 nm, respectively, this last was excited with a 403.5-nm laser, and emission was read in the 425 to 475 nm range. Channels were acquired sequentially.

Excitation of ratiometric pHluorin was performed with 403.5 and 488 nm lasers and the corresponding emissions were read in the 500 to 550 nm range.

2.8.

Image Analysis

Image analysis was performed with ImageJ and Metamorph (Molecular Devices, United States). MagIC ratio images were obtained dividing the Venus channel by the mCherry one after background subtraction.

2.9.

In Situ Mg²⁺ Sensitivity

${\mathrm{Mg}}^{2+}$ sensitivity of the indicator expressed in cells was assayed as follows: cells expressing MagIC or mCherry–cpVenus, used as a negative control, were washed with ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-free Hanks balanced salt solution [HBSS(−); Sigma, containing, in mM: KCl, 5.33; ${\mathrm{KH}}_2{\mathrm{PO}}_4$, 0.44; NaCl, 138; ${\mathrm{NaHCO}}_3$, 4.0; ${\mathrm{Na}}_2{\mathrm{HPO}}_4$, 0.3; glucose, 5.6; pH in the 7.2 to 7.6 range] and then the cellular membrane was permeabilized with $20\mu \mathrm{g}/\mathrm{mL}$ digitonin in HBSS(−). To this, 10 mM ${\mathrm{Mg}}^{2+}$ in HBSS(−) was added, followed by the application of a saturating (20 mM) concentration of EDTA in HBSS(−).

2.10.

Hyperosmotic Shock of HeLa Cells

HeLa cells cultured in glass-bottom dishes for 48 h after transfection were imaged in their original medium (DMEM + 10% FBS) at 5-s interval for 10 min at room temperature. Application of the hyperosmotic shock was performed by removing the cell medium and simultaneously, a new medium with either ${\mathrm{MgCl}}_2$, KCl, or d-sorbitol (Wako Chemicals, Japan) was added to obtain the equimolar final concentrations of 50 mM ${\mathrm{MgCl}}_2$, 75 mM KCl, or 150 mM d-sorbitol in the imaged dish.

3.1.

Indicator Design

${\mathrm{Mg}}^{2+}$-sensitivity of the indicator was achieved following a strategy similar to that one used for the development of the low affinity ${\mathrm{Ca}}^{2+}$ indicator CatchER.¹² In the present study, three of the five negatively charged residues (corresponding to 147E, 202D, and 204E in CatchER) were sequentially introduced into cp 173 Venus,^18,19 becoming 220E, 31D, and 33E, respectively (in cp173Venus numeration, Table 1). Additional spontaneous mutations G4S (EGFP 177S), V89I (EGFP 18V), and Q142M (EGFP 72Q) were found in the brightest colonies expressing the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive cpVenus variant. Of these, the last two were preserved in the final indicator construction. In addition, the following variants of the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive cp173Venus were created: (1) with the five negative residues of CatchER, (2) with the five negative residues and reversed spontaneous mutations, and (3) with the three negative residues and reversed spontaneous mutations. However, the resulting fluorescence of all these variants in bacterial cells was dimmer (not shown). The chosen indicator variant (corresponding to the brightest colonies) displayed a variation in the fluorescence intensity as a function of the applied ${\mathrm{Mg}}^{2+}$ or ${\mathrm{Ca}}^{2+}$ concentration and constituted the intensiometric (${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive) part of the indicator.

Unlike other single-FP based indicators (e.g., Camgaroos,²⁰ Pericams,²¹ GECOs,¹⁶ or G-CaMPs^22,23), in our case, the fluorescence intensity variation upon ion binding was not provoked by large intramolecular conformational changes (see details in Ref. 12). This feature led us to introduce a second FP as a constant reference, the fluorescence intensity of which did not depend on ${\mathrm{Mg}}^{2+}$ or ${\mathrm{Ca}}^{2+}$ levels, thus allowing ratiometric imaging of changes in the concentration of these ions [Fig. 1(a)]. The resulting indicator required irradiation at two separate wavelengths: one to obtain the intensiometric signal from the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus variant, and the other to obtain the reference signal used to perform the signals’ ratioing.

3.2.

Selection of the Reference FP

To avoid undesirable resonance energy transfer between the two proteins in the indicator, FPs with excitation peaks at wavelengths longer than the emission maximum of the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus variant (530 nm) were selected as reference. As mentioned above, the reference fluorescence intensity should by itself be insensitive to variations in ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$ concentrations. In addition to this requirement, the reference fluorescence intensity should be large enough to reduce errors related to the stochastic character of the photon emission process, which can result in errors in the final ratio image.

From the tested FPs (Table 2), the closest one to meet all the above requirements was mCherry.²⁴ An initial approach to use photoconverted PSmOrange (excitation/emission maxima $636/662\mathrm{nm}$)²⁵ resulted in almost complete absence of the Venus fluorescence due to photobleaching by the intense $475/28\mathrm{nm}$ light used for photoconversion, without significant changes observed in the optical properties of PSmOrange (not shown). On the other hand, the nonphotoconverted PSmOrange (excitation/emission maxima $548/565\mathrm{nm}$) proved the viability of the chosen approach for in vitro ratiometric imaging using a reference fluorescence signal. PSmOrange emission did not interfere with the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus fluorescence and could be detected separately, though this result is impractical for imaging due to low fluorescence intensity [Fig. 2(c)] and requirement of somewhat complicated filter sets in the microscopy system. In the case of other reference FPs, mKOκ,²⁶ and mKate2,²⁷ either Venus fluorescence was lost due to almost complete FRET [mKOκ; Fig. 2(c)] or the reference fluorescence was too dim [mKate2; Fig. 2(c)]. Moreover, the analysis of the excitation, absorption and emission spectra of PSmOrange, mKOκ, and mKate2 (in fusions with ${\mathrm{Mg}}^{2+}$-sensitive cpVenus) revealed their sensitivity to addition of 5 mM ${\mathrm{Mg}}^{2+}$ (Fig. 2) and only mCherry spectra remained unchanged. Thus, we confirmed that mCherry fluorescence intensity is stable enough to serve as a constant reference for the indicator.

Table 2

Fluorescent proteins (FPs) tested as reference to obtain a ratiometric low affinity Mg2+/Ca2+-sensitive indicator.

Fig. 2

Selection of an appropriate ${\mathrm{Mg}}^{2+}$-insensitive reference fluorescent protein (FP). Shown are absorption, excitation, and emission spectra of the reference FPs (in fusion with the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus variant) in absence (dashed lines) or presence (solid lines) of ${\mathrm{Mg}}^{2+}$ (5 mM). (a) Absorption spectra (blue lines) of constructions normalized to ${\mathrm{Mg}}^{2+}$-sensitive Venus absorption peak at 516 nm in the presence of 5 mM ${\mathrm{Mg}}^{2+}$. The right peak in the spectra corresponds to the absorption maximum of the specified reference FP. Only the mCherry absorption peak remains unchanged after addition of 5 mM ${\mathrm{Mg}}^{2+}$. (b) Excitation spectra of the constructions normalized to the excitation maxima of the ${\mathrm{Mg}}^{2+}$-sensitive Venus in absence of ${\mathrm{Mg}}^{2+}$. The excitation of the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus in all constructions (black lines) was monitored at its emission maximum (530 nm). The monitored wavelengths for the reference FPs (in red lines) were—PSmOrange, 565 nm; mKOκ, 563 nm; mKate2, 633 nm; mCherry, 610 nm—as displayed in the corresponding legends. As in the case of absorption spectrum, the excitation peak of mCherry is unaffected by ${\mathrm{Mg}}^{2+}$ addition. Note the decrease of the excitation peak of mKOκ after ${\mathrm{Mg}}^{2+}$ addition due to FRET from the Venus variant to mKOκ. (c) Emission spectra obtained by separate excitation of either the ${\mathrm{Mg}}^{2+}$-sensitive Venus variant (green lines) or the reference FP (magenta lines) within the fusion constructions. Emission of the Venus variant in all cases was obtained by excitation with 516-nm light. The excitation wavelengths for the references were—548 nm, for PSmOrange; 551 nm, for mKOκ; 588 nm, for mKate2. Note the almost complete absence of the Venus emission peak in the mKOκ-[${\mathrm{Mg}}^{2+}$-sensitive cp173Venus] construction due to FRET. Traces are the average of three independent measurements.

The fusion of mCherry with the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive moiety described in Sec. 3.1 constituted the final indicator, termed as MagIC [Fig. 1(b)].

We should mention that during our work we found that the non-FRET ratiometric imaging approach referred to here has already been applied in several previous studies,^28,29 which also used mCherry as a reference FP. In these studies, however, the performance of other reference FPs was not analyzed.

3.3.

In Vitro Properties of MagIC

As the reported ${\mathrm{Mg}}^{2+}$ concentration in MagIC relies on the ratio of fluorescence signals from two FPs, a preferential expression of one of them could potentially alter the results of measurements. That is why after purification, the protein samples were tested by seminative SDS PAGE to estimate their purity and the indicator integrity. For this purpose, the samples with SDS were subjected to PAGE without preliminary heating. Under these conditions, FPs retain their fluorescence and can be easily visualized on excitation using appropriate light. Our results show that more than 90% of the observed fluorescent band intensity corresponds to the correct construction in the purified protein sample [Fig. 3(a)].

Fig. 3

In vitro properties of MagIC. (a) Seminative SDS PAGE of a purified sample of MagIC; fluorescence acquired with a conventional digital camera through an orange filter (top) and gel stained with Coomassie Brilliant Blue (bottom). (b) and (c) Emission spectra of MagIC as a function of the applied concentrations of divalent cations. Following the principle shown in Fig. 1, the fluorescence intensity of the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus variant increases with an increasing concentration of either ${\mathrm{Mg}}^{2+}$ [in (b)] or ${\mathrm{Ca}}^{2+}$ [in (c)]. The fluorescence spectra of mCherry (in red), taken separately, are not affected by addition of either ${\mathrm{Mg}}^{2+}$ or ${\mathrm{Ca}}^{2+}$. The zoomed inset in (b) shows a fragment of the mCherry spectra obtained at the different tested ${\mathrm{Mg}}^{2+}$ concentrations. (d) Calibration curve obtained by ratioing the fluorescence intensities of the two components of the indicator, obtained at room temperature (25°C). The dynamic range of the indicator is 1.5. (e) Selectivity of MagIC is shown as the change in the ratio induced by the application of inorganic cations, polyamines or ATP in 1 mM ${\mathrm{Mg}}^{2+}$ background. Respective ions or compounds were added at the specified concentrations and then the ratio difference was calculated by subtracting the ratio at 1 mM ${\mathrm{Mg}}^{2+}$. The $530/610$ ratio at 1 mM ${\mathrm{Mg}}^{2+}$ is 4.7 (25°C). Put, putrescine; Spd, spermidine; Sp, spermine. (f) pH dependency of the ($I_{530}/I_{610}$) ratio in a background of 100 mM KCl and 1 mM ${\mathrm{MgCl}}_2$. The $\mathrm{p}{K}_a$ of the indicator determined under these conditions is 7.5. (g) Relaxation constant $k_{\mathrm{obs}}([{\mathrm{Mg}}^{2+}])$ for MagIC as a function of the applied ${\mathrm{Mg}}^{2+}$ concentration. Rate constants, $k_{\text{on}}$ and $k_{\text{off}}$, were determined by fitting the equation $k_{\mathrm{obs}}=k_{\text{on}}[{\mathrm{Ca}}^{2+}]+k_{\text{off}}$. Their values at 31°C are presented.

First, we tested in vitro the affinity of our indicator to ${\mathrm{Mg}}^{2+}$. The emission spectrum of MagIC displays a rising 610-nm peak with an increase in the applied ${\mathrm{Mg}}^{2+}$ concentration when the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$ sensitive Venus variant is excited [Fig. 3(b)]. The intensity of this peak is higher than the intensity of separately excited mCherry, suggesting that it is provoked both by excitation of mCherry and the emission of the Venus variant at wavelengths longer than 580 nm. Such undesired FRET forced us to acquire the reference signal by a separate excitation, not only in vitro but also during imaging.

To determine the indicator affinity for ${\mathrm{Mg}}^{2+}$, solutions with ${\mathrm{Mg}}^{2+}$ concentrations ranging from 0 to 25 mM were used. The calculated affinity constant for this ion in the presence of 100 mM KCl at room temperature was $K_d({\mathrm{Mg}}^{2+})=5.1\mathrm{mM}$ (Hill constant $n=1.1$) [Fig. 3(d)] with saturation around 20 mM ${\mathrm{Mg}}^{2+}$.

The designed ${\mathrm{Mg}}^{2+}$-chelating site does not exclude the possibility of other divalent cations binding, so the selectivity of MagIC for several of these cations was tested. As shown in Fig. 3(c), the indicator also displays affinity for ${\mathrm{Ca}}^{2+}$ in the millimolar range [$K_d({\mathrm{Ca}}^{2+})=4.8\mathrm{mM}$, $n=0.8$]. This suggests that, similar to many other fluorescent dyes used for cytosolic ${\mathrm{Mg}}^{2+}$ monitoring, our indicator is also a low affinity (though of millimolar order) ${\mathrm{Ca}}^{2+}$ indicator.

The selectivity of MagIC in 1 mM ${\mathrm{Mg}}^{2+}$ was analyzed against ${\mathrm{Ca}}^{2+}$, ${\mathrm{Mn}}^{2+}$ or ${\mathrm{Ni}}^{2+}$ at a concentration of 0.1 mM each. In the presence of these concentrations, much larger than their normal levels in the eukaryotic cell cytosol, the (Venus/mCherry) ratio of MagIC remained almost unaffected. ${\mathrm{Co}}^{2+}$, ${\mathrm{Cu}}^{2+}$, ${\mathrm{Fe}}^{2+}$, and ${\mathrm{Fe}}^{3+}$ at micromolar concentrations also did not show any noticeable effect on the ratio value observed for ${\mathrm{Mg}}^{2+}$ alone [Fig. 3(e)]. The ratio value, however, was increased by the application of 1 mM spermine and spermidine, polyamines, which are known to compete with ${\mathrm{Mg}}^{2+}$ for binding sites in nucleic acids [Fig. 3(e)]. Interestingly, application of 1 mM ATP provoked a strong decrease in the (Venus/mCherry) ratio, indicating competition with MagIC for ${\mathrm{Mg}}^{2+}$ by this high affinity ${\mathrm{Mg}}^{2+}$ chelator [$K_d({\mathrm{Mg}}^{2+})=50\mu \mathrm{M}$],³⁰ suggesting at the same time that our indicator does not sense Mg-ATP. The $\mathrm{p}K\mathrm{a}$ of MagIC, as in the case of CatchER,¹² the $\mathrm{p}K\mathrm{a}$ of MagIC is 7.5 [Fig. 3(f)], being the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus the main pH-sensitive component. Analysis of the kinetics of ${\mathrm{Mg}}^{2+}$-binding by MagIC using the stopped flow method revealed values of $k_{\text{on}}=11859{\mathrm{M}}^{-1}{\mathrm{s}}^{-1}$ and $k_{\text{off}}=83.956{\mathrm{s}}^{-1}$ [at 31°C; Fig. 3(g)].

3.4.

MagIC can be Selectively Targeted to Different Intracellular Compartments

MagIC is a genetically encoded indicator; so, in a next step, we analyzed its intracellular distribution when expressed in cells and the possibility of its targeting to specific intracellular locations. Expression of MagIC in HeLa cells without any particular targeting sequence resulted in mixed cytosolic and nuclear localization, a unique feature of our ${\mathrm{Mg}}^{2+}$ indicator [Fig. 4(a)]. MagIC-transfected cells were easily identified both by the bright yellow–green fluorescence of ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus and by the red fluorescence of mCherry.

Fig. 4

In situ properties of MagIC. (a) Expression pattern of MagIC in HeLa cells without targeting sequences (top; cytosolic and nuclear localization), targeted to mitochondria (middle) and to the endoplasmic reticulum (bottom). (b) Seminative SDS PAGE of nonheated lysate of HeLa cells expressing cytosolic MagIC. A single fluorescent band is observed (top), corresponding to the indicator. At the bottom is shown the overlay of the fluorescence image with other cellular proteins stained with Coomassie Brilliant Blue (CBB). In (c) and (d), ${\mathrm{Mg}}^{2+}$ sensitivity of MagIC was assayed with ${\mathrm{Mg}}^{2+}/\mathrm{EDTA}$ in digitonin-permeabilized HeLa cells. Experiments were conducted at 21°C. (c) (Venus/mCherry) ratio images in each of the applied conditions. (d) Time course of the change in the ratio in cells from (c). (e) The negative control mCherry–cpVenus does not show ${\mathrm{Mg}}^{2+}$-dependent ratio changes. (f) and (g) MagIC is insensitive to cytosolic ${\mathrm{Ca}}^{2+}$ increase. MagIC was coexpressed with the ${\mathrm{Ca}}^{2+}$ indicator B-GECO1 in pituitary cells, in which spontaneous ${\mathrm{Ca}}^{2+}$ spikes are observed which do not provoke visible associated changes in the cytosolic ${\mathrm{Mg}}^{2+}$. Imaging carried out at 37°C in Dulbecco’s modified Eagle’s medium-F12. (f) Representative images of the MagIC ratio (proportional to cytosolic ${\mathrm{Mg}}^{2+}$) and the corresponding B-GECO1 fluorescence intensity (proportional to cytosolic ${\mathrm{Ca}}^{2+}$) taken with 1 min interval. Corresponding images taken with 5-s interval during 10 min are shown in Video 1. The movie shows the calculated (Venus/mCherry) ratio for MagIC and the B-GECO1 fluorescence intensity for a representative group of cells. Color bars show the corresponding lookup tables (LUTs) for both images. (Venus/mCherry) ratio LUT range is 0 (violet) to 2 (red). For B-GECO1’s channel LUT, the black color corresponds to low fluorescence intensity, and the light blue in the opposite extreme—to a higher fluorescence intensity. (g) Traces of MagIC ratio and B-GECO1 fluorescence intensity for representative cells shown in (f). Shown are representative experiments from 3 (${\mathrm{Mg}}^{2+}$-calibration) or 6 (${\mathrm{Ca}}^{2+}/{\mathrm{Mg}}^{2+}$ monitoring in GH3 cells) trials. Scale bars, $10\mu \mathrm{m}$. Video 1 (MOV, 6.96 MB) [URL: http://dx.doi.org/10.1117/1.JBO.20.10.101203.1].

Expression of MagIC in fusion with the targeting sequence of human ornithine transcarbamylase resulted in mitochondrial localization of the sensor [Fig. 4(a)], being it’s both components highly fluorescent at this organelle. On the other hand, MagIC targeted to ER displayed significantly reduced fluorescence of the ${\mathrm{Mg}}^{2+}/{\mathrm{Ca}}^{2+}$-sensitive Venus [Fig. 4(a)]. This effect was not indicative of low level of divalent cations in the ER, as the ER-targeted negative control mCherry–cpVenus displayed even lower yellow–green fluorescence when imaged under similar conditions (not shown).

A seminative SDS PAGE of the nonboiled lysate of HeLa cells expressing cytosolic MagIC displayed a single fluorescent band, indicating the expression of FPs predominantly within the full indicator construction [Fig. 4(b)].

3.5.

In Situ ${\mathsf{Mg}}^{\mathsf{2}+}$ Sensitivity and Calibration of MagIC

Having determined the in vitro properties of MagIC, we next analyzed its ${\mathrm{Mg}}^{2+}$-sensitivity in cells. For this purpose, we initially incubated cells in solutions with increasing ${\mathrm{Mg}}^{2+}$ concentrations in the presence of ionomycin—a divalent cation ionophore. Under these conditions, MagIC displayed an increasing ratio with the increase in the extracellular ${\mathrm{Mg}}^{2+}$ concentration. However, in each case, the equilibration of the system took a relatively long time, during which the cell shape was altered, probably as a response to cytosolic ${\mathrm{Mg}}^{2+}$ elevation (not shown).

In addition to these difficulties, we also found concerns in the literature regarding the use of ionophores to equilibrate ${\mathrm{Mg}}^{2+}$ concentration in cells (see in Ref. 5). To overcome this, we decided to analyze the in-cell ${\mathrm{Mg}}^{2+}$ sensitivity of MagIC using a different strategy. In the new scheme, HeLa cells expressing MagIC were treated with $20\mu \mathrm{g}/\mathrm{mL}$ digitonin at the beginning of the experiment to permeabilize the cell membrane. Next, 10 mM ${\mathrm{Mg}}^{2+}$ was applied followed by the application of 20 mM EDTA. In these conditions, MagIC reported an increase in the (Venus/mCherry) ratio when ${\mathrm{Mg}}^{2+}$ was added and showed a pronounced decrease in the ratio value upon EDTA addition [Figs. 4(c) and 4(d)]. At the same time, no change in the (Venus/mCherry) ratio value was observed in the case of cells expressing mCherry–cpVenus, which was used as a negative control [Fig. 4(e)], indicating that the ratio changes observed in the case of MagIC corresponding to changes in the free ${\mathrm{Mg}}^{2+}$ concentration.

3.6.

MagIC is Insensitive to Cytosolic ${\mathsf{Ca}}^{2+}$ Elevations

As MagIC shows sensitivity to ${\mathrm{Ca}}^{2+}$ (though in the millimolar range), we next tested the possibility that its ratio value could be affected by elevations in the cytosolic ${\mathrm{Ca}}^{2+}$ levels. For this purpose, we coexpressed MagIC with the genetically encoded ${\mathrm{Ca}}^{2+}$ indicator B-GECO1¹⁶ in pituitary cells, which are known to generate spontaneous ${\mathrm{Ca}}^{2+}$ transients. Simultaneous monitoring of cytosolic ${\mathrm{Mg}}^{2+}$ and ${\mathrm{Ca}}^{2+}$ dynamics in this cell line at 37°C in DMEM-F12 first showed that MagIC is insensitive to the cytosolic ${\mathrm{Ca}}^{2+}$ increase and indicated at the same time that the cytosolic ${\mathrm{Mg}}^{2+}$ dynamics was not affected by the generation of these ${\mathrm{Ca}}^{2+}$ spikes, at least during the course of imaging [Figs. 4(f) and 4(g)].

3.7.

Hyperosmotic Shock Induces MagIC Ratio Increase in HeLa Cells

With the experiments presented above, we could confirm the ${\mathrm{Mg}}^{2+}$ sensitivity of MagIC expressed in cells, however, the most demanding part of the work was to search for an appropriate stimulus that could induce visible changes in the cytosolic ${\mathrm{Mg}}^{2+}$ level. This task was further hindered by the fact that cytosolic free ${\mathrm{Mg}}^{2+}$ concentration seems to be tightly regulated even in situations when the stimulus induces a large ${\mathrm{Mg}}^{2+}$ flux through the plasma membrane.⁶

We initially focused on nordihydroguaretic acid (NDGA) a polyphenolic compound abundant in the creosote bush (Larrea tridentata), and which was reported to enhance the ${\mathrm{Na}}^{+}$-dependent ${\mathrm{Mg}}^{2+}$ efflux mediated by the plasma membrane-located ${\mathrm{Na}}^{+}/{\mathrm{Mg}}^{2+}$ exchanger,³¹ and is, in addition, a potent inhibitor of the TRPM7—one of the main cellular ${\mathrm{Mg}}^{2+}$-uptake systems.³² Application of NDGA ($50\mu \mathrm{M}$) to cells expressing MagIC provoked an immediate and strong decrease in the (Venus/mCherry) ratio that achieved its minimum within 5 min and was thereafter maintained constant at this low level during the course of imaging [Figs. 5(a) and 5(b)]. It appeared, however, that acidification of the cytosol was largely responsible for this effect, as the negative control mCherry–cpVenus also displayed a (Venus/mCherry) ratio decrease with a very similar time course [Figs. 5(c) and (d)]. Interestingly, simultaneous application of NDGA (enhancer of ${\mathrm{Na}}^{+}/{\mathrm{Mg}}^{2+}$ exchange)³¹ and $100\mu \mathrm{M}$ imipramine (inhibitor of the corresponding exchanger)^31,33 slightly delayed the onset of the MagIC ratio value decrease and reduced it, then finally induced the formation of large membrane blebs within a few minutes of application [Fig. 5(e)]—an effect which was not observed when either of these chemicals was applied alone [Fig. 5(f)].

Fig. 5

Effect of nordihydroguaretic acid (NDGA) application on the (Venus/mCherry) ratio of MagIC expressed in HeLa cells. (a and b) Effect of NDGA alone on the ratio value of MagIC expressed in HeLa cells. Application of NDGA in HBSS(−) induced an almost immediate decrease of the (Venus/mCherry) ratio. (a, top) Merged (Venus + mCherry) images of cells before (left) or after (right) application of $50\mu \mathrm{M}$ NDGA. (a, bottom) Pseudocolored images of the same cells displaying the ratio change induced by NDGA application. (b) Time courses of the ratio decrease for the cells shown in (a). (c and d) The negative control, mCherry–cpVenus displays a similar (Venus/mCherry) ratio decrease, suggesting that this effect is produced by the cytosol acidification. (e and f) Addition of NDGA together with the inhibitor of the ${\mathrm{Na}}^{+}/{\mathrm{Mg}}^{2+}$ exchanger imipramine ($100\mu \mathrm{M}$) induces blebs formation (e, right), an effect which is not observed when imipramine is added alone (f). Scale bars, $10\mu \mathrm{m}$.

In Ref. 10, the authors noted that their genetically encoded ${\mathrm{Mg}}^{2+}$ indicator, MagFRET, did not report ${[{\mathrm{Mg}}^{2+}]}_{\mathrm{cyt}}$ changes under several conditions which were previously reported to provoke ${[{\mathrm{Mg}}^{2+}]}_{\mathrm{cyt}}$ increase. These conditions were (as cited in Ref. 10): (1) application of elevated ${\mathrm{Mg}}^{2+}$ in the medium,³⁴ (2) application of ${\mathrm{Li}}^{+}$ as ${\mathrm{Mg}}^{2+}$-competitor,³⁵ and (3) perfusion with solutions containing variable ${\mathrm{Na}}^{+}$ concentration,^36,37 each applied to different cell types. We decided to analyze which kind of response is reproduced in HeLa cells expressing MagIC in one of these conditions, namely, stimulation with high extracellular ${\mathrm{Mg}}^{2+}$.

In our experiments, treatment of HeLa cells expressing MagIC with 50 mM ${\mathrm{MgCl}}_2$ induced a significant MagIC ratio increase [Figs. 6(a) and 6(b)]. This result is similar to that previously observed with the ${\mathrm{Mg}}^{2+}$-selective dye KMG-104 in HEK293T cells.³⁴ The applied ${\mathrm{Mg}}^{2+}$ concentration (50 mM) is, however, significantly higher than the ${\mathrm{Mg}}^{2+}$ level present in physiological fluids (around 1 mM, reported elsewhere) and provoked visible shrinking of the treated cells. So, to analyze whether the MagIC ratio increase was a specific response to the high extracellular ${\mathrm{Mg}}^{2+}$ or to the associated hyperosmotic shock, we replaced ${\mathrm{MgCl}}_2$ by an equimolar amount of KCl or the carbohydrate d-sorbitol. Application to cells of either 75 mM KCl [Figs. 6(c) and 6(d)] or 150 mM d-sorbitol [Figs. 6(e) and 6(f)] in the medium induced a MagIC ratio increase similar to that provoked by the elevated extracellular ${\mathrm{Mg}}^{2+}$, indicating that the presence of this last response is not indispensable for the observed effect. Part of this response seems, however, to depend on a highly probable change in the cytosolic ionic strength due to cellular water loss, since the application of these osmolytes to cells expressing the negative control mCherry–cpVenus also induced a slight (Venus/mCherry) ratio change, though in the opposite direction [Figs. 6(g)–6(m)].

Fig. 6

Hyperosmotic shock induces cytosolic MagIC ratio increase. HeLa cells expressing either MagIC. (a–f) or mCherry–cpVenus (g–l) were treated with either 50 mM ${\mathrm{MgCl}}_2$, 75 mM KCl, or 150 mM d-sorbitol applied to the original medium. (a) and (b) Similar to results reported in previous studies,³⁴ 50 mM ${\mathrm{MgCl}}_2$ stimulation induces an apparent cytosolic ${\mathrm{Mg}}^{2+}$ increase which is reported by MagIC. However, application of an equimolar amount of KCl (c and d) or d-sorbitol (e and f) in the medium induces a similar apparent ${\mathrm{Mg}}^{2+}$ increase, suggesting that it is provoked by the resulting high osmotic potential of the medium, and not by the extracellular ${\mathrm{Mg}}^{2+}$ itself. Part of this response, however, seems to be associated to nonspecific changes in the intracellular milieu provoked by hyperosmotic stresses, as ratio changes were also observed in the ratio reported by the ${\mathrm{Mg}}^{2+}$-insensitive negative control mCherry–cpVenus (g–l). Application of hyperosmotic media (with either ${\mathrm{MgCl}}_2$, KCl or d-sorbitol) in this case also induce (Venus/mCherry) ratio changes, though oppositely directed to those reported by MagIC. Shown are images of cells in the medium before (left) and after the addition of the indicated osmolyte (right). Traces correspond to the measured intensities of the Venus variants (green) and mCherry (red) fluorescence intensities, and the calculated (Venus/mCherry) ratios for the shown cells. Experiments were repeated at least three times under each condition. Scale bars, $10\mu \mathrm{m}$.

To confirm that the changes observed with MagIC or the negative control mCherry–cpVenus were not associated to changes in the cytosolic pH, to which MagIC should be sensitive according to the measurements presented above [Fig. 3(f)], we performed monitoring of the cytosolic pH using a genetically encoded pH indicator, ratiometric pHluorin.¹⁷ This indicator has two absorption maxima around 410 and 470 nm and a single emission peak. The ($410/470$) emission ratio of pHluorin is directly proportional to the pH in its environment, so we used it for monitoring cytosolic pH under hyperosmotic stimulation (Fig. 7). We performed imaging of cells expressing pHluorin with excitation by 403.5 and 488 nm lasers lines. In this case, addition to cells of medium supplemented with 50 mM ${\mathrm{MgCl}}_2$ did not induce obvious changes in the cytosolic pH, indicating that the (Venus/mCherry) ratio increase reported by MagIC has a different origin.

Fig. 7

Hyperosmotic shock does not affect cytosolic pH. HeLa cells expressing ratiometric pHluorin in the cytosol were stimulated by medium supplemented with 50 mM ${\mathrm{MgCl}}_2$, as in a similar experiment in cells expressing MagIC [Figs. 6(a) and 6(b)]. The emissions produced by excitation with either 403.5 or 488 nm lights display an increase in fluorescence intensity similar to that observed for the ${\mathrm{Mg}}^{2+}$-sensitive cp173Venus and mCherry in MagIC. However, the ($403.5/488$) ratio value does not show evidence of cytosolic pH change induced by this stimulation.