Chromis-1, A Ratiometric Fluorescent Probe Optimized for Two-Photon Microscopy Reveals Dynamic Changes in Labile Zn(II) in Differentiating Oligodendrocytes

Abstract

Despite the significant advantages of two-photon excitation microscopy (TPEM) over traditional confocal fluorescence microscopy in live-cell imaging applications, including reduced phototoxicity and photobleaching, increased depth penetration, and minimized autofluorescence, only few metal ion-selective fluorescent probes have been designed and optimized specifically for this technique. Building upon a donor-acceptor fluorophore architecture, we developed a membrane-permeant, Zn(II)-selective fluorescent probe, chromis-1, that exhibits a balanced two-photon cross section between its free and Zn(II)-bound form and responds with a large spectral shift suitable for emission-ratiometric imaging. With a K_d of 2.4 nM and wide dynamic range, the probe is well suited for visualizing temporal changes in buffered Zn(II) levels in live cells as demonstrated with mouse fibroblast cell cultures. Moreover, given the importance of zinc in the physiology and pathophysiology of the brain, we employed chromis-1 to monitor cytoplasmic concentrations of labile Zn(II) in oligodendrocytes, an important cellular constituent of the brain, at different stages of development in cell culture. These studies revealed a decrease in probe saturation upon differentiation to mature oligodendrocytes, implying significant changes to cellular zinc homeostasis during maturation with an overall reduction in cellular zinc availability. Optimized for TPEM, chromis-1 is especially well-suited for exploring the role of labile zinc pools in live cells under a broad range of physiological and pathological conditions.

for TOC only

Fluorescence microscopy ranks among the most widely utilized methods for the non-invasive observation of biological processes in live cells, tissues, and whole organisms. As most biological targets are non-fluorescent, this technique relies on the application of selective fluorescent labels and indicators;¹ however, photobleaching and phototoxicity associated with most fluorophores impose significant limitations, especially for prolonged imaging studies with live specimens. Two-photon excitation microscopy (TPEM), which is based on the simultaneous absorption of two photons with half the energy compared to traditional fluorescence imaging, overcomes many of these challenges with the added benefit of intrinsic 3D imaging capabilities, increased depth penetration, and reduced cellular background fluorescence.^2–3 Although common fluorescent tags can be employed for TPEM imaging, their brightness may be hampered due to low two-photon absorption (2PA) cross sections.⁴ The need for TPEM-optimized fluorophores is particularly evident in the area of metal-ion sensing with ratiometric indicators that rely on a chromatic shift for analyte detection.^5–6 Although tunable to a specific wavelength, Ti-sapphire femtosecond-pulsed lasers employed in commercial TPEM instruments cannot alternate between different excitation energies in rapid succession. Therefore, ratiometric TPEM imaging requires indicators that offer not only sufficient brightness, but also respond with a spectral shift of the emission profile,⁷ requirements that are currently not met by most ratiometric indicators.⁸

To address these challenges, we employed a fluorophore design strategy that yields both a large 2PA cross section, as well as a chromatic shift in the fluorescence emission upon metal binding.⁹ By electronically coupling the analyte binding site to the acceptor moiety of a push-pull fluorophore architecture rather than the donor moiety, coordination of the metal cation elicits an increased intramolecular charge transfer (ICT) upon photoexcitation, which in turn results in a more favorable 2PA cross section along with a red-shifted fluorescence emission. Building upon this approach, we present here a Zn(II)-selective indicator, chromis-1, that features a balanced 2PA cross section between the free and bound probe, combined with a large emission shift suitable for emission-ratiometric TPEM imaging.

In addition to rigorous characterization of the coordination chemistry and photophysical properties, we employed chromis-1 for monitoring dynamic changes of labile Zn(II) pools in live mouse fibroblasts and for assessing differences in the zinc availability in oligodendrocytes at different stages of development. These cells play a critical role in the central nervous system for insulating axons through the formation of a myelin sheath composed of lipids and proteins.¹⁰ As an essential trace element, zinc serves a diverse set of physiological functions in the brain. Cellular zinc pools are tightly controlled by a complex interplay between zinc transporters, influx and efflux pathways, as well as subcellular zinc storage sites.^11–12 Not surprisingly, zinc excess can have profound effects on the developing and mature nervous system. For example, excess zinc released from glutamatergic synapses is a cause of neuronal death during ischemia,¹³ and accumulation of zinc following optic nerve injury triggers retinal ganglion cell death.¹⁴ Similarly, surges in intracellular zinc have also been implicated in excitotoxic or nitrative insults to mature oligodendrocytes.^15–17 Despite the demonstrated importance of zinc in the central nervous system, studying homeostatic fluctuations in labile zinc pools that are estimated to be buffered in the picomolar range¹⁸ has been challenging due to a lack of suitable reagents to visualize and probe cellular zinc levels. Optimized for TPEM, chromis-1 addresses this need and should be well-suited for exploring the physiological role of labile zinc pools in the central nervous system.

Experimental Section

Synthesis of Chromis-1.

Synthetic procedures for compounds 1a-b and all intermediates are provided in the Supporting Information.

X-ray structure determination.

Single colorless plate-shaped crystals of the Zn(II) complex of chromis-1 acid 1b were grown as described in the Supporting Information. A suitable crystal (0.19×0.10×0.03 mm) was selected and mounted on a loop with paratone oil on a Bruker APEX-II CCD diffractometer. The crystal was cooled to T = 100(2) K during data collection. The structure was solved with the ShelXT-2014/4¹⁹ structure solution program using combined Patterson and dual-space recycling methods and by using Olex2²⁰ as the graphical interface. The crystal structure was refined with version 2014/7 of ShelXL²¹ using Least Squares minimization. Crystal structure data and structure refinement parameters are provided in Table S1 of the Supplementary Information.

Liposome Preparation.

A solution of 32.3 mg 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) and 8.2 mg 1,2-dimyristoyl-sn-glycero-3-phospho-(1’-rac-glycerol) (DMPG) in 4:1 dichloromethane-methanol (20 mL) was evaporated under reduced pressure followed by drying for 2 hours in high vacuum. The resulting film was hydrated with aqueous buffer (29.8 mL, 20 mM PIPES, 0.1 M KCl, pH 7.0), sonicated for 5 minutes, and left to equilibrate for 5 hours. After dilution to a concentration of 500 μM, the hydrated lipids were passed 21 times through an extruder (Liposofast, Avestin) using a polycarbonate membrane with 200 nm pore size. Liposome particle sizes were determined based upon the Tunable Resistance Pulse Sensing (TRPS) principle using the Izon qNano particle analyzer (Izon Science, Ltd., Burnside, New Zealand) to afford an average diameter of 122 nm.²² Freshly extruded lipids were used immediately.

Photophysical Characterization.

All buffers and stock solutions were prepared with 18.2 MΩ•cm Milli-Q water and filtered through 0.2 μm membrane filters to remove interfering particles or fibers. UV-Vis absorption measurements were acquired at 25°C using a Cary Bio50 spectrophotometer (Varian) with constant temperature accessory. Fluorescence measurements were performed with a PTI fluorimeter equipped with a 75 W xenon arc lamp excitation source and a photomultiplier detection system. Fluorescence spectra were corrected for the spectral response of the detection system and the spectral irradiance of the excitation source (via a calibrated photodiode). Spectra were acquired using quartz cuvettes with 10 cm path length for the OD measurements of quantum yield determinations and 1 cm path length for all other measurements.

Fluorescence quantum yields were determined using quinine sulfate as a reference (Φ_f = 0.55 in 1.0 N H₂SO₄)²³ with excitation at 365 nm by a 4-point measurement over an OD range of 0.1–0.4 (l = 10 cm) in pH 7.0 buffer (10 mM PIPES, 0.1 M KCl, 25°C). In the case of lipophilic chromis-1 ester (1a), measurements were performed by diluting the compound from a 3 mM DMSO stock solution to a final concentration of 2 μM in pH 7.0 buffer containing 100 μM 4:1 DMPC:DMPG liposomes.²² To sequester adventitious Zn(II), quantum yields of free chromis-1 ester and acid were determined in the presence of 20 μM EDTA.

Metal Ion Selectivity.

A 5 μM solution of chromis-1 acid 1b was prepared in pH 7.0 buffer (10 mM PIPES, 0.1 M KCl), which had been treated with Chelex (1% w/v, Biorad) overnight. Metal ions were supplied from aqueous stock solutions (1 M for Mg(II) and Ca(II), 1 mM for all other metal ions) of the corresponding sulfate salts (Mn(II), Fe(II), Cu(II), and Zn(II)) or nitrate salts (Mg(II), Ca(II), Co(II), and Ni(II)). Fluorescence spectra were acquired in the absence and upon addition of 0.8 molar equivalents of the respective metal ions (excitation at 358 nm). Responses to Mg(II) and Ca(II) were measured each at a concentration of 2 mM. In addition, fluorescence spectra were acquired after saturating the remaining 0.2 molar equivalents of the probe by supplementation with 1 μM Zn(II). All ratios R = BP2/BP1 were calculated based on the integrated fluorescence between 510-570 nm (BP2) and 440-495 nm (BP1).

Stability Constants.

Chromis-1 acid 1b (20 μM) in pH 7.0 buffer (10 mM PIPES, 0.1 M KCl, pH 7.0, 25°C) was saturated with 1.0 molar eq. of Zn(II) supplied from a 6 mM stock solution of ZnSO₄·7H₂O in deionized H₂O and then titrated with EGTA up to a total concentration of 0.5 mM. After addition of each aliquot, the solution was equilibrated for 10 minutes and a UV-vis spectrum was acquired over the range of 250–500 nm. The spectral data were analyzed by non-linear least squares fitting using SPECFIT.²⁴ to yield an average apparent logK of 10.31 ± 0.13 (Figure S2).

To determine the Zn(II) stability constant of chromis-1 ester 1a, the compound was diluted from a 3 mM DMSO stock solution to a final concentration of 2 μM in pH 7.0 buffer (20 mM PIPES, 0.1 M KCl, pH 7.0, 25°C) containing 1.0 mM EGTA and 100 μM liposomes (4:1 mixture of DMPC and DMPG, vide infra). The solution (3.0 mL total volume) was equilibrated at 25°C with gentle stirring for a minimum of 10 minutes, and then titrated with Zn(II) to a final concentration of 1 mM. After addition of each aliquot of Zn(II), the solution was equilibrated for 5 minutes, and a fluorescence spectrum was acquired over the spectral window of 380–680 nm with excitation at 358 nm (Figure S3). The data were analyzed by non-linear least squares fitting over the entire spectral range using SPECFIT.

Protonation Constants.

A combination glass electrode with double junction was calibrated based on Gran’s method by titrating a 5 mM solution of HCl in 0.1 M KCl with a standardized solution of 0.1 M KOH at 25°C. The titration data were analyzed with the GLEE software package to derive the corresponding electrode potential E° and slope s relating the measured emf with the hydronium ion concentration p[H] or –log[H₃O⁺].²⁵ The sequential protonation constants of chromis-1 acid 1b were determined based on spectrophotometric titrations at 25°C. To this end, a solution of 1b (22.7 μM) was prepared in a mixed buffer (1 mM PIPES/PIPBS, 0.1 M KCl, pH 7.1) and a total of 50 UV-vis spectra were acquired in a quartz cuvette (1 cm path length) over a p[H] range from 7.1 to 1.0. The titration solution was gradually acidified by addition of 1–10 μL aliquots of aqueous HCl supplied from 0.1, 1, or 6 M stock solutions. The combined UV-vis traces were analyzed by non-linear least squares fitting over the entire spectral range (300–500 nm) using the SPECFIT software package.²⁴ Titration data, deconvoluted spectra for each protonation state, and a species distribution diagram is provided in the Supporting Information (Figure S5).

pH-Dependent Response.

Two batches of 2 mM 4:1 DMPC:DMPG liposomes were prepared according to the procedure described above in pH 5.0 buffer (10 mL of 10 mM PIPES, 0.1 M KCl) and pH 7.0 buffer (10 mL of 10 mM PIPES, 0.1 M KCl), respectively. Freshly extruded liposomes were diluted to 100 μM final concentration in the respective buffer solutions and equilibrated in a quartz cuvette for 15 minutes at 25°C. Chromis-1 ester was added from a 3 mM DMSO stock solution to a final concentration of 2 μM. After an additional 10-minute equilibration period, a fluorescence spectrum was acquired from 380–700 nm with excitation at 358 nm. Spectra of the Zn(II)-saturated probe at pH 5 and 7 were acquired after addition of an equimolar amount of ZnSO₄•7H₂O supplied from a 3 mM stock solution in deionized H₂O (Figure S4).

Cell Culture and Reagents.

NIH 3T3 mouse fibroblasts were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Lonza) supplemented with 10% bovine serum (Gemini), penicillin/streptomycin (50 IU/mL), 200 μM L-glutamine, and 100 μM sodium pyruvate at 37°C under an atmosphere of humidified air containing 5% CO₂.

For oligodendrocyte cultures, mixed glia cultures were first prepared from postnatal day 2 (P2) rat forebrains and grown for 10–17 days as previously described.²⁶ In brief, developing OLs were separated from microglia and astrocytes using selective detachment and then plated onto coverslips. Cells were maintained in the presence of platelet-derived growth factor (PDGF) and basic fibroblast growth factor (FGF) for 8 days prior to imaging. To produce mature OLs, PDGF and FGF were replaced by triiodothyronine (T3) and ciliary neurotrophic factor (CNTF) and cultured for an additional week. Media was changed every other day with 2x growth factors. Health of the cells was determined by visual inspection using phase contrast microscopy.

Two-photon microscopy.

For imaging experiments, cells were grown to 70% confluency on MatTek glass bottom culture dishes pre-coated with poly-L-lysine using phenol red-free DMEM (10% bovine serum, 50 IU/mL penicillin/streptomycin, 200 μM L-glutamine, and 100 μM sodium pyruvate). After replacing the growth medium with serum-free DMEM containing chromis-1 ester (10 μM), cells were incubated for 15 minutes at 37°C (5% CO₂ atmosphere), and the incubation solution was again replaced with pre-warmed full media. Cells were imaged in an atmospheric chamber at 37°C, 8% humidity, and 5% CO₂ using a Zeiss LSM confocal NLO 710 microscope equipped with a femtosecond pulsed Ti:sapphire laser. Scanning fluorescence micrographs were acquired with excitation at 720 nm, and the integrated emission intensity was simultaneously collected through two band pass filters between 425–462 nm and 478–540 nm. Reagents for probing the response of chromis-1, including pyrithione, ZnSO₄, N,N,N’,N’-tetrakis(2-pyridylmethyl)ethylenediamine (TPEN), 2,2′-dithiodipyridine (DTDP), and bafilomycin A₁ were directly added from DMSO stock solutions to the dish mounted on the microscope stage. Ratiometric image analysis was performed with ImageJ²⁷ as outlined in the Supporting Information.

For colocalization with cellular marker proteins, cells were grown to 30–40% confluency on MatTek glass bottom as described above and transfected with 2 μg of either pDsRed2-mito or pEYFP-ER plasmids using Turbofect (4 μL in 200 μL DMEM, Thermo Scientific). After 24 hours, the growth medium was replaced with DMEM-free medium containing chromis-1 ester. Cells were incubated for 15 minutes, and the medium was replaced with pre-warmed full media for acquiring fluorescence micrographs as described above. Fluorescence emission of both chromis-1 and the respective fluorescent organelle markers were collected simultaneously with the laser and filter settings switching on a line-by-line basis while scanning (chromis-1 ester: λ_ex 720 nm, λ_em BP 400–498 nm, DsRed2-mito: λ_ex 562 nm, λ_em BP 570 – 700 nm, EYFP-ER: λ_ex 514 nm, λ_em BP 525–700 nm). For colocalization with lipid droplets, cells were co-stained with 10 μM chromis-1 and 2 μM BODIPY 493/503 (4,4-difluoro-1,3,5,7,8-pentamethyl-4-bora-3a,4a-diaza-s-indacene, Thermo Fisher, λ_ex 488 nm, λ_em BP 525–675 nm), or 50 nM LysoTracker Red (Thermo Fisher, λ_ex = 514 nm, λ_em 525 – 675 nm).

Chromis-1 Imaging and Analysis with Oligodendrocytes.

On the day of imaging, media containing chromis-1 ester (10 μM) was added without growth factors and incubated at 37°C for 30 minutes. Medium containing chromis-1 was removed, cells were washed with Hanks Buffered Saline Solution (HBSS) containing calcium and magnesium and then transferred to 60mm dishes containing HBSS. Imaging was performed using a Zeiss LSM710 confocal microscope equipped for 2-photon microscopy. Zinc-pyrithione, and TPEN were pipetted onto the coverslips using concentrated stock solutions. Ratiometric image analysis was performed in ImageJ as above. Statistical differences were calculated using prism software.

Results and Discussion

Synthesis.

Following a convergent synthetic approach, chromis-1 was assembled from three building blocks, the Fmoc-protected amino acid 3, α-amino ketone 4, and pyridine derivative 9 (Figure 1A). While the latter two were synthesized based on modified literature procedures,^{9, 28} acid 3 was accessible in two steps from inexpensive isonicotinic acid N-oxide. Alkylation with dimethylsulfate followed by vicarious nucleophilic substitution with NaCN yielded 2-cyanoisonicotinic acid 2, which was converted to Fmoc-protected 3 in a one-pot procedure with 85% overall yield. The bridging thiazole unit of the fluorophore π-system was constructed by EDCI-mediated condensation of 3 and 4 followed by tandem thionation-cyclization of coupling intermediate 5 with Lawesson’s reagent. Subsequent Fmoc-deprotection yielded intermediate 6b with 65% overall yield. By using ethanethiolate as the base in lieu of piperidine that is commonly employed for Fmoc-removal, amine 6b could be isolated in analytically pure form through simple acid-base extraction. Similarly, intermediates 5 and 6a were isolated and crystallized in high yield without the need for chromatographic purification. The synthesis was concluded by nucleophilic substitution of amine 6b with 9 to afford crystalline chromis-1 ester 1a, a portion of which was hydrolyzed to give chromis-1 acid 1b as the corresponding water-soluble disodium salt.

Figure 1.

(A) Synthetic scheme for the assembly of chromis-1 (1). (B) ORTEP representation and atom numbering scheme for the crystal structure of the Zn(II)-complex of chromis-1 acid 1b. Ellipsoids shown represent 50% probability. Hydrogen atoms and counter ions have been omitted for clarity. Detailed crystallographic data are provided in the Supporting Information (Table S1-S3).

X-ray structure determination.

The Zn(II) complex of chromis-1 acid was crystallized from a mixture of aqueous KCl (0.1 M), DMF, and MeOH to produce plate-like single crystals suitable for X-ray crystallographic analysis. The unit cell of the triclinic crystals contained two conformers with notable structural differences. A list of selected bond angles and distances is provided in the supporting information (Table S2, SI). As illustrated in Figure 1B for one of the two conformers, the Zn(II)-complex adopts a trigonal-bipyramidal coordination geometry with chloride in the axial position. Compared to previously published structures of the unsubstituted C3-symmetric [tris(picolyl)amine-Zn(II)Cl]⁺ complex,^29–31 the equatorial Zn-N bonds are elongated by an average of 0.03 ± 0.01 Å. The slightly weaker interaction with the pyridine ligands can be attributed to the electron-withdrawing character of the thiazole and carboxylic acid substituents. In contrast, there are no significant differences for the axial Zn-Cl and Zn-N bonds, which adopt in all structures a uniform bond length of 2.26 ± 0.01 Å. In both conformers, the fluorophore architecture exhibits some deviation from planarity as evident from the dihedral angles of the corresponding aryl-aryl bonds. The largest difference is observed for the anisole ring, which is rotated out of plane by 35 degrees relative to the thiazole, however, in only one of the two conformers. The near-planar dihedral angle of 14.7 degrees for the second conformer suggests that the distortion is associated with a soft potential and likely due to packing forces.

Steady-state absorption and fluorescence properties.

Molar ratio titrations of chromis-1 acid (1b) with Zn(II) in neutral aqueous buffer (pH 7.0, 10 mM PIPES 0.1 M KCl, 25°C) revealed strong chromatic shifts in both the absorption and emission spectra together with sharp saturation at equimolar metal and ligand concentrations (Figure 2A and B). Pertinent photophysical data, including quantum yields, are compiled in Table 1. Consistent with Zn(II)-coordination to the pyridyl acceptor moiety of the fluorophore, the absorption maximum at 349 nm (ε ~ 20,500 M^–1 cm^–1) is red-shifted by 17 nm upon saturation with Zn(II). The appearance of clean isosbestic points at 285 and 353 nm combined with equimolar saturation are indicative of a well-defined solution equilibrium involving only free probe and a Zn(II)-complex with 1:1 metal-ligand stoichiometry. Similarly, the fluorescence emission maximum of the probe is red-shifted from 483 to 520 nm upon Zn(II)-binding. By exciting the probe at the isosbestic point, the observed emission intensities are a direct reflection of the quantum yield, which increases more than 2-fold from 0.32 to 0.71 upon Zn(II)-coordination (Figure 2B). Overall, the Zn(II)-induced fluorescence changes are well suited for emission-ratiometric measurements. For example, the ratio of the integrated fluorescence intensities between 510–570 nm and 440–495 nm increases from 0.7 to 2.6 upon saturation with Zn(II), corresponding to a large dynamic range R_max/R_min of 3.7.

Figure 2.

Spectral characteristics of chromis-1 acid (1b, 5 μM) in response to divalent metal cations in aqueous buffer (pH 7.0, 10 mM PIPES, 0.1 M KCl, 25°C). (A) UV-vis absorption spectral changes upon saturation with Zn(II). (B) Fluorescence emission response upon saturation with Zn(II) (excitation at 358 nm). Inset: Fluorescence intensity at 520 nm as a function of molar equivalents of Zn(II). (C) Two-photon absorption cross section of the free (blue trace) and Zn(II)-bound (red trace) form of 1b. (D) Emission-ratiometric response towards selected divalent metal cations at 80% fractional saturation in absence (black bars) or presence (white bars) of Zn(II). The emission ratio was calculated based on the integrated fluorescence intensity between 510–570 nm and 440–495 nm. Excitation: 358 nm.

TABLE 1.

One and Two-photon photophysical properties of chromis-1 ester (1a) and acid (1b) in aqueous buffer at pH 7.0.^a

	1a b,c	[1a-Zn(II)] b	1b c	[1b-Zn(II)]
abs λmax (nm) d	351	378	349	366
ε(104M−1cm−1)e	2.20	2.86	2.05	2.41
2PA λmax (nm) f	n.d.	n.d.	720	740
δmax (GM)g	n.d.	n.d.	19	25
em λmax (nm)h	438	495	483	520
ΦF i	0.26	0.56	0.32	0.71

^a10 mM PIPES, 0.1 M KCl, 25°C.

^bbuffer supplemented with 100 μM liposomes (4:1 DMPC:DMPG).

^csupplemented with 20 μM EDTA to sequester adventitious Zn(II).

^dlowest-energy band of the one-photon absorption spectrum.

^emolar extinction coefficient at λ_max.

^fmaximum two-photon absorption.

^gtwo-photon absorption cross section.

^hmaximum fluorescence emission.

ⁱfluorescence quantum yield, referenced to quinine sulfate (Φ_F = 0.55).²³

Analogous to the UV-vis spectral properties, the two-photon absorption (2PA) cross section revealed a red-shifted maximum, which increased from 720 to 740 nm upon Zn(II)-coordination (Figure 2C). Consistent with greater excited-state polarization,³² the 2PA cross section of the Zn(II)-bound form is significantly increased compared to the free form (Figure 2C). Nevertheless, in the range between 710–740 nm the probe offers balanced cross sections, allowing for excitation of the free and Zn(II)-bound probe with similar efficiency. In contrast to Zn(II), all other biologically relevant divalent metal ions showed either no change in the emission response or complete fluorescence quenching (Figure S1). As both the fluorescent probe and its Zn(II)-complex are brightly fluorescent, the quenched emission does not contribute to the ratiometric readout, thus binding of these metal ions results only in a masking effect by competing for Zn(II)-coordination. For this reason, we evaluated the selectivity of the ratio-response in the presence of 0.8 molar equivalents of the interfering metal ion (Figure 2D, black bars). Under these conditions, the emission ratio of the remaining free probe was unaffected by any of the quenching metal ions, while addition of Zn(II) still produced the emission ratio response of the probe-Zn(II) complex (Figure 2D, white bars) as observed in the molar ratio titration (Figure 2B).

Binding affinity.

The linear emission increase together with sharp saturation at the equivalence point for the molar ratio titration shown in Figure 2B indicates tight Zn(II) binding with near unity fractional saturation throughout the entire titration range. Because the determination of reliable stability constants demands conditions where complex formation is incomplete, ideally within a 10–90% fractional saturation window,³³ we performed spectrophotometric titrations in the presence of EGTA as a competing ligand with matching affinity (apparent K_d = 4.0 nM at pH 7.0, calculated from published pK_a’s and logβ values).³⁴ Regardless of whether the preformed Zn(II)-complex of chromis-1 acid 1b was titrated with EGTA, or whether Zn(II) was added to a mixture of the probe and EGTA, least-squares fitting of the spectrophotometric data yielded a uniform apparent stability constant of logK = 10.31 ± 0.13 at pH 7.0 and 0.1 M KCl ionic background (25°C), corresponding to an apparent dissociation constant of 49 ± 13 pM (Figure S2). To gauge the effect of pH changes on the Zn(II) affinity, we also determined the protonation constants of chromis-1 acid 1b through combined potentiometric and spectrophotometric titrations (Figure S5). Concluding from the derived experimental pK_a values of 5.5, 3.5, and 1.2, chromis-1 acid is almost exclusively present as the fully deprotonated dianionic species at neutral pH, and thus the apparent stability constant is not strongly affected by variations in cellular pH between 6 to 8. For example, the apparent Zn(II) affinity decreases only from logK of 10.3 to 10.2 upon acidification from pH 8 to 6.

Because the anionic dicarboxylate salt 1b is not membrane permeable, cellular imaging studies were performed with diester 1a which is readily internalized and partitions within intracellular membranes (vide infra). To determine the apparent Zn(II) stability constant of 1a under conditions that mimic the cellular environment, fluorimetric titrations were carried out in the presence of liposomes as model membranes, composed of a 4 : 1 ratio of zwitterionic dimyristoyl phosphatidylcholine (DMPC) and anionic dimyristoyl phosphatidylglycerol (DMPG). Upon dilution from a DMSO stock solution into pH 7.0 buffer containing 100 μM liposomes and 1.0 mM EGTA as competing ligand, diester 1a was titrated up to a total concentration of 1 mM Zn(II). Non-linear least-squares fitting of the fluorescence titration data over the entire spectral range yielded an average apparent logK of 8.62 ± 0.07, corresponding to a K_d of 2.4 ±0.4 nM (Figure S3). Consistent with a stronger electron withdrawing character of the neutral ester groups in 1a compared to the anionic carboxylate substituents in 1b, chromis-1 ester exhibits a Zn(II) affinity that is nearly two orders of magnitude lower. For the same reason, the ester form is also expected to exhibit lower protonation constants, thus further attenuating the effect of pH changes on the apparent Zn(II) affinity.

Ratiometric imaging of cellular zinc.

To evaluate the Zn(II)-dependent ratiometric response of chromis-1 within the chemical complexity of a live cell, we performed a series of perfusion experiments with NIH 3T3 mouse fibroblasts as a model system (Figure 3). Live adherent cells were incubated with 10 μM chromis-1 ester 1a in serum-free medium for 15 minutes at 37˚C and imaged by two-photon excitation at 720 nm using a Zeiss LSM 710 Confocal/NLO microscope. Fluorescence emission was collected with a polychromatic detector between 425–462 nm and 478–540 nm, and the corresponding ratio-images were derived based on the integrated photon counts on a pixel-per-pixel basis (Figure 3A, right). As evident from the intensity images (Figure 3A, middle), chromis-1 ester readily entered cells to yield bright fluorescence throughout the cytoplasm but was excluded from cell nuclei. Although the fluorescence intensity distribution appears uneven, including distinct punctate localizations, the corresponding ratio-image indicates a rather even fractional saturation with an average ratio R of 0.60 ± 0.16. Upon exposure to a mixture of ZnSO₄ and pyrithione, a membrane-permeable ionophore for Zn(II),³⁵ the intensity ratio rapidly increased by more than 4-fold to R = 2.53 ± 0.56, indicating saturation of the probe with Zn(II) (Figure 3B). Addition of the membrane-permeant high-affinity chelator TPEN (N,N,N’,N’-tetrakis(2-pyridylmethyl)ethylenediamine, apparent K_d = 9.3 fM at pH 7.0)³⁴ elicited a ratio response towards a slightly lower R value of 0.54 ± 0.13 compared to the initial conditions. A similar R value of 0.52 ± 0.14 was obtained when TPEN was directly added to cells grown in basal medium without prior exposure to the Zn(II)-pyrithione mixture (Figure S7). A plot of the intensity ratio vs. time revealed a rapid dynamics for Zn(II)-binding and release, both of which occurred within a few seconds (Figure 3B, right). Using the ratios in the presence of TPEN and Zn(II)-pyrithione as the limiting R_min and R_max values, respectively, 3T3 cells grown under basal conditions display a fractional saturation around 3–6%. Based on the measured K_d of 2.4 nM of chromis-1 ester, we estimated an average buffered Zn(II) concentration in the range of 50–100 pM. This value agrees well with previous reports on the buffered concentration of cytosolic free Zn(II), ranging between 0.1–1 nM in a broad range of cell types.^36–38 Although chromis-1 ester could be subject to hydrolysis by non-specific esterases, the subcellular distribution with a pronounced association with neutral lipids (vide infra) and nuclear exclusion indicates that the probe remains intact in its charge-neutral ester form.

Figure 3.

Ratiometric imaging of labile Zn(II) pools in live NIH 3T3 mouse fibroblasts with chromis-1 ester (1a) by TPEM (excitation at 720 nm). (A) Left: Phase contrast image (DIC) and fluorescence intensity images acquired with 425–462 nm (BP1) and 478–540 nm (BP2) bandpass filters, respectively. Right: Intensity ratio image with R = BP2/BP1. (B) Left: Ratio images (BP2/BP1) at selected time points prior and after addition of 50 μM ZnSO₄ and 5 μM pyrithione (ZnPyr). Further addition of 100 μM TPEN at 13 min resulted in a reversal of the intensity ratio (shown at 20 min). Right: Time course of the average intensity ratio change for the ROI indicated with a white circle in the left panel. The asterisks indicate time points for the respective ratio images shown to the left. (C) Left: Ratio images prior (at 5 min) and after addition of 100 μM DTDP (at 45 min). This treatment resulted first in a ratio increase peaking around 40 minutes followed by a gradual decrease back to the initial level (shown at 115 min). Right: Time course of the average intensity ratio change for the ROI indicated with a white circle. The asterisks indicate time marks for the respective ratio images to the left. Scale bars: 40 μm.

To explore whether chromis-1 can be utilized to follow dynamic fluctuations of endogenous Zn(II) pools, cells were exposed to 2,2′-dithiodipyridine (DTDP), an oxidizing agent that is expected to induce dissociation of Zn(II) ions from cellular sulfhydryl binding sites as present, for example, in metallothionein.³⁹ Consistent with the initial perfusion experiment (Figure 3B, left), cells grown under basal conditions and treated with chromis-1 ester produced similarly low intensity ratios around 0.5 (Figure 3C, left), which gradually increased upon addition of 100 μM DTDP, reaching a maximum after 40 min with an average around 1.3 (Figure 3C, middle). The dramatic ratio increase was followed by a gradual decrease without addition of an exogenous reagent, thus indicating re-equilibration of cytoplasmic Zn(II) levels back to basal conditions with an intensity ratio around 0.5.

Subcellular localization of chromis-1.

The non-uniform staining pattern of chromis-1 ester in 3T3 cells suggests specific localization to subcellular structures. To elucidate potential associations with cellular organelles, we utilized selective fluorescent protein markers and live cell stains to perform co-localization studies. To this end, 3T3 cells were either transfected with plasmids encoding for mitochondrial (pDsRed2-mito) or endoplasmic reticulum (pEYFP-ER) organelle markers or co-stained with LysoTracker Red (LTR) or BODIPY 493/50 for visualizing acidic compartments or neutral lipids, respectively.⁴⁰ As shown in Figure 4A, the fluorescence staining pattern of chromis-1 ester (green) revealed minimal spatial correlation with the subcellular distribution of LysoTracker Red or the mitochondrial and ER marker proteins (red); however, significant overlap with the BODIPY lipid marker was observed (last column). Consistent with its lipophilic properties, chromis-1 ester 1a preferentially associates with intracellular membranes and other lipid-rich regions.

Figure 4:

(A) Colocalization analysis of chromis-1 ester (1a) in NIH 3T3 mouse fibroblasts using selected cellular markers. Top row: Fluorescence micrograph of cells that express either dsRed2-mito (1^st column, λ_ex = 561 nm, λ_em 570 – 700 nm) or EYFP-ER (2^nd column, λ_ex = 514 nm, λ_em 525 – 675 nm) as mitochondrial or ER markers, respectively, or that are co-stained with LysoTracker Red (3^rd column, λ_ex = 514 nm, λ_em 525 – 675 nm) or the lipid droplet marker BODIPY 493/503 (4^th column, λ_ex = 488 nm, λ_em 525 – 675 nm). Center row: Two-photon fluorescence micrograph of the same cells co-stained with chromis-1 ester (λ_ex = 720 nm, λ_em 400 – 498 nm). Bottom row: Red-green false-color overlay images revealing spatial overlap between chromis-1 ester and the respective fluorescent cellular markers in yellow hues. (B) Emission ratiometric response of chromis-1 towards intracellular pH changes. Top panel: Live NIH 3T3 cells were co-incubated with LysoTracker Red (LTR, 50 nM) and chromis-1 ester (2 μM) and fluorescence images were acquired simultaneously to assess changes in LTR fluorescence (top row, λ_ex = 514 nm, λ_em 525 – 675 nm) and chromis-1 emission ratio (λ_ex = 720 nm, ratio = BP2/BP1 BP1 = 425–462 nm, BP2 = 478–540 nm band pass emission filters) upon addition of bafilomycin A₁ (900 nM) at 14 min. Bottom: Plot for the time-dependent changes of the average emission intensity of LTR (right scale, red trace) and intensity ratio of chromis-1 fluorescence (left scale, blue trace) for the ROI indicated with a white circle. The asterisk marks the time points for above micrographs, and the arrow shows the time point of bafilomycin A₁ addition. Scale bars: 20 μm.

Sensitivity towards cellular pH fluctuations.

To test the ratiometric response of chromis-1 ester towards interference from changes in cellular pH, live fibroblasts cells were exposed to 900 nM bafilomycin A₁, a selective inhibitor of vacuolar-type H⁺-ATPases.^41–42 In order to dynamically track the effect of bafilomycin A₁, cells were incubated with LysoTracker Red in addition to chromis-1 ester. Functionalized with a weak base, LysoTracker Red acts as an acidotropic dye that is rendered membrane impermeable upon protonation and thus selectively accumulates within acidic compartments such as lysosomes. As shown in Figure 4B, the fluorescence intensity of LysoTracker Red was immediately attenuated upon addition of bafilomycin A₁, thus indicating rapid pH-equilibration of the stained acidic compartments and surrounding cytosol. In contrast, negligible changes were detected for the ratiometric response of chromis-1 ester. This result is consistent with in vitro studies using liposomes as model membranes, where only small changes of the fluorescence emission were observed upon acidification from pH 7 to pH 5 (Figure S4).

Probing labile Zn(II) pools in developing oligodendrocytes.

Zinc is abundant in the central nervous system but tightly regulated, with the majority of Zn(II) bound to proteins or sequestered in intracellular organelles. At the same time, a fraction of total cellular Zn(II) is kinetically labile and has been increasingly recognized to play important signaling roles in the brain.¹¹ To test whether chromis-1 is suitable for visualizing alterations of this labile pool in a system relevant to zinc regulation within the central nervous system, we assessed differences in intracellular free zinc concentrations in oligodendrocytes at different stages of development. The development of oligodendrocytes occurs following a well-characterized lineage that can be identified using antibodies against cell surface antigens. Representative phase contrast images of developing and mature oligodendrocytes and fluorescence immunohistochemistry images obtained using the O4 antigen (O4) as a marker of developing oligodendrocytes, myelin basic protein (MBP) as a marker of mature oligodendrocytes and DAPI as nuclear counter stain are shown in Figure 5A and 5B. As expected, there was virtually no MBP expression in developing oligodendrocytes, but robust expression in mature oligodendrocytes. Similar cultures were used for all subsequent imaging experiments. Cells were loaded with chromis-1 ester (10 μM) in serum-free medium and then imaged within 60 minutes. Baseline intensity ratios were determined in both developing and mature oligodendrocytes. As shown in Figure 5C, the Zn(II)-saturation of chromis-1 was decreased in mature versus developing oligodendrocytes as judged from the different intensity ratios of 0.59 ± 0.01 (n = 54 cells from 3 different cultures) vs 0.70 ± 0.01 (n = 103 cells from 3 different cultures), respectively ( p<0.0001).

Figure 5.

(A) Phase contrast microscopy images of developing oligodendrocytes at DIV9 and mature oligodendrocytes at DIV16. (B) Immunocytochemistry for O4 antigen (O4, green), myelin basic protein (MBP, red) and DAPI (blue) in developing and mature oligodendrocytes. (C) Representative chromis-1 ratio images for developing and mature oligodendrocytes at DIV9 and DIV16 respectively. (D) Scatterplot for chromis-1 ratios of individual developing (red) and mature (blue) oligodendrocytes. Mean ± SEM shown with black solid lines and indicated a decrease in ratio in mature oligodendrocytes compared to developing oligodendrocytes. * P < 0.0001 (t-test developing vs mature oligodendrocytes). Broken line indicates average ratio after application of TPEN.

To determine the limiting emission intensity ratios R_min and R_max, we conducted experiments analogous to those with mouse fibroblast cells by applying either TPEN (100 μM) or Zn(II)-pyrithione (50 μM ZnCl₂ and 5 μM pyrithione), respectively. Addition of TPEN to the medium resulted in an immediate decrease of the emission ratio to 0.49 ± 0.008 in developing oligodendrocytes (n = 46 cells from 3 different cultures) and 0.51 ± 0.003 in mature oligodendrocytes (n = 28 cells from 3 different cultures). These values are in close agreement with R_min determined in mouse fibroblast cells and indicate a robust probe response regardless of cell type. Moreover, the reduction to a uniform R_min starting from different cell-type dependent baseline ratios is consistent with quantitative depletion of labile Zn(II) from the probe. Similarly, saturation of chromis-1 upon incubation with exogenous Zn(II)-pyrithione yielded an R_max value of 2.76 ± 0.03 in developing oligodendrocytes (n = 21 cells from 2 different cultures) and 2.47 ± 0.04 in mature oligodendrocytes (n = 31 cells from 3 different cultures), again closely mirroring the response observed in mouse fibroblast cells. Altogether, these experiments demonstrate a robust and reproducible emission-ratiometric response of chromis-1, regardless of the specific biological system. Based on the limiting R_min and R_max values, we calculated a fractional saturation of 8–9% in developing and 3.5–6.5% in mature oligodendrocytes, thus translating into estimated Zn(II)-levels of 135–152 pM and 65–95 pM, respectively.

The decrease in Zn(II) levels as oligodendrocytes mature may play a role in the differentiation of these cells. Several zinc-finger transcription factors have been identified that are critical for proper oligodendrocyte differentiation.^43–46 Modulation of intracellular zinc concentrations alters nuclear interactions of at least one of these transcription factors, Ying Yang-1,⁴⁷ highlighting a mechanism by which changes in intracellular zinc status may promote epigenetic modifications. A second mechanism by which intracellular free zinc concentrations may influence differentiation is by the regulation of enzyme function, as zinc acts as a co-factor for several hundred enzymes.⁴⁸ A classic example of this type of regulation is the calcium-calmodulin pathway where the activity of proteins is directly modulated by alterations in intracellular calcium concentrations, leading to well characterized signaling cascades.⁴⁹ Similar zinc signaling pathways are increasingly being identified. As a single example, the ERK signaling pathway is essential for oligodendrocyte differentiation⁵⁰ and the conversion from phosphorylated ERK to dephosphorylated ERK is directly regulated by intracellular zinc concentrations.⁵¹ In addition, proliferating cells require higher zinc concentrations to promote DNA synthesis and regulate cell cycle progression.⁵² The change in zinc status may reflect the transition from proliferating to differentiating phenotype. Finally, zinc ions also stabilize association of myelin basic protein with myelin membranes,⁵³ so the expression of MBP may contribute to the reduction in free cytosolic zinc.

Conclusions

Optimized for ratiometric two-photon excitation microscopy, the fluorescent probe chromis-1 features a balanced brightness and large emission shift upon saturation with Zn(II). By integrating the metal-binding site within the framework of a donor-acceptor substituted fluorophore platform, the emission shift was achieved without compromising the two-photon cross section. Because background noise is amplified in the ratio image compared to the raw signal of each channel, a balanced brightness between the free and Zn(II)-bound form as offered by chromis-1 is a critical prerequisite for robust quantitative ratiometric image analysis. Although a large R_max/R_min ratio might appear advantageous, the apparent increase in dynamic range would be offset by the increased noise due to the limited dynamic range of the detector system.

In situ measurements in three different cell systems established uniform R_min and R_max values from which we were able to estimate the corresponding levels of cytosolic Zn(II). As demonstrated with differentiating oligodendrocytes, ratiometric two-photon microscopy is especially well suited for visualizing subtle changes of physiologically relevant Zn(II) levels. While cytosolic Zn(II) in developing oligodendrocytes ranged between 135–152 pM, mature MBP cells revealed nearly two-fold lower levels in the 65–95 pM range. These values reside well within the concentration window reported for cell systems that were analyzed based on genetically encoded ratiometric fluorescence probes. For example, the level of cytosolic Zn(II) was estimated to be 80 pM in HeLa cells,³⁶ 270 pM in HEK293 cells,³⁷ or 400 pM in INS-1 pancreatic β cells.³⁷ It is important to note that ratiometric image analysis informs only on the fractional saturation of the probe, not the actual buffered Zn(II) levels. The latter must be derived on the basis of independent probe-calibration procedures, which are typically carried out in vitro. Due to variations in calibration conditions, such as temperature, pH, or ionic background, and the discrepancies of reported protonation constants and Zn(II) affinities of the employed competitor ligands and metal ion buffers, the derived cellular Zn(II) levels should be considered as estimates. Nevertheless, ratiometric imaging represents a powerful approach to visualize relative changes in dynamic Zn(II)-availability as demonstrated across the oligodendrocyte lineage. With the ability to visualize physiologically relevant changes in mobile Zn(II) levels, chromis-1 is well poised to untangle specific molecular pathways that regulate zinc homeostasis and zinc signaling in a broad range of biological systems.